Cleveland Clinic Journal of Medicine

A mbulatory electrocardiography (ECG) began in 1949 when Norman “Jeff” Holter developed a monitor that could wirelessly transmit electrophysiologic data.¹ His original device used vacuum tubes, weighed 85 pounds, and had to be carried in a backpack. Furthermore, it could send a signal a distance of only 1 block.²

At the time, it was uncertain if this technology would have any clinical utility. However, in 1952, Holter published the first tracing of abnormal cardiac electrical activity in a patient who had suffered a posterior myocardial infarction.³ By the 1960s, Holter monitoring systems were in full production and use.⁴

Since then, advances in technology have led to small, lightweight devices that enable clinicians to evaluate patients for arrhythmias in a real-world context for extended times, often with the ability to respond in real time.

Many ambulatory devices are available, and choosing the optimal one requires an understanding of which features they have and which are the most appropriate for the specific clinical context. This article reviews the features, indications, advantages, and disadvantages of current devices, and their best use in clinical practice.

INDICATIONS FOR AMBULATORY ECG MONITORING

Diagnosis

The most common diagnostic role of monitoring is to correlate unexplained symptoms, including palpitations, presyncope, and syncope, with a transient cardiac arrhythmia. Monitoring can be considered successful if findings on ECG identify risks for serious arrhythmia and either correlate symptoms with those findings or demonstrate no arrhythmia when symptoms occur.

A range of arrhythmias can cause symptoms. Some, such as premature atrial contractions and premature ventricular contractions, may be benign in many clinical contexts. Others, such as atrial fibrillation, are more serious, and some, such as third-degree heart block and ventricular tachycardia, can be lethal.

Arrhythmia symptoms can vary in frequency and cause differing degrees of debility. The patient’s symptoms, family history, and baseline ECG findings can suggest a more serious or a less serious underlying rhythm. These factors are important when determining which device is most appropriate.

Ambulatory ECG can also be useful in looking for a cause of cryptogenic stroke, ie, an ischemic stroke with an unexplained cause, even after a thorough initial workup. Paroxysmal atrial fibrillation is a frequent cause of cryptogenic stroke, and because it is transient, short-term inpatient telemetry may not be sufficient to detect it. Extended cardiac monitoring, lasting weeks or even months, is often needed for clinicians to make this diagnosis and initiate appropriate secondary prevention.

Prognosis: Identifying patients at risk

In a patient with known structural or electrical heart disease, ambulatory ECG can be used to stratify risk. This is particularly true in evaluating conditions associated with sudden cardiac death.

For example, hypertrophic cardiomyopathy and arrhythmogenic right ventricular dysplasia or cardiomyopathy are 2 cardiomyopathies that can manifest clinically with ventricular arrhythmias and sudden cardiac death. Ambulatory ECG can detect premature ventricular contractions and ventricular tachycardia and identify their frequency, duration, and anatomic origin. This information is useful in assessing risk of sudden cardiac death and determining the need for an implantable cardioverter-defibrillator.

Similarly, Wolff-Parkinson-White syndrome, involving rapid conduction through an accessory pathway, is associated with increased risk of ventricular fibrillation and sudden cardiac death. Ambulatory ECG monitoring can identify patients who have electrical features that portend the development of ventricular fibrillation.

Also associated with sudden cardiac death are the inherited channelopathies, a heterogeneous group of primary arrhythmic disorders without accompanying structural pathology. Ambulatory ECG monitoring can detect transient electrical changes and nonsustained ventricular arrhythmias that would indicate the patient is at high risk of these disorders.

Assessing arrhythmia treatment

Arrhythmia monitoring using an ambulatory ECG device can also provide data to assess the efficacy of treatment under several circumstances.

The “pill-in-the-pocket” approach to treating atrial fibrillation, for example, involves self-administering a single dose of an antiarrhythmic drug when symptoms occur. Patients with infrequent but bothersome episodes can use an ambulatory ECG device to detect when they are having atrial fibrillation, take their prescribed drug, and see whether it terminates the arrhythmia, all without going to the hospital.

Ambulatory ECG also is useful for assessing pharmacologic or ablative therapy in patients with atrial fibrillation or ventricular tachycardia. Monitoring for several weeks can help clinicians assess the burden of atrial fibrillation when using a rhythm-control strategy; assessing the ventricular rate in real-world situations is useful to determine the success of a rate-control strategy. Shortly after ablation of either atrial fibrillation or ventricular tachycardia, ECG home monitoring for 24 to 48 hours can detect asymptomatic recurrence and treatment failure.

Some antiarrhythmic drugs can prolong the QT interval. Ambulatory ECG devices that feature real-time monitoring can be used during drug initiation, enabling the clinician to monitor the QT interval without admitting the patient to the hospital.

Ultimately, ambulatory ECG monitoring is most commonly used to evaluate symptoms. Because arrhythmias and specific symptoms are unpredictable and transient, extended monitoring in a real-world setting allows for a more comprehensive evaluation than a standard 10-second ECG recording.

Continuous external monitoring: The Holter monitor

Recording is typically done continuously for 24 to 48 hours, although some newer devices can record for longer. Patients can press a button to note when they are experiencing symptoms, allowing for potential correlation with ECG abnormalities. The data are stored on a flash drive that can be uploaded for analysis after recording is complete.

What is its best use? Given its relatively short duration of monitoring, the Holter device is typically used to evaluate symptoms that occur daily or nearly daily. An advantage of the Holter monitor is its ability to record continuously, without requiring the patient to interact with the device. This feature provides “full disclosure,” which is the ability to see arrhythmia data from the entire recording period.

These features make Holter monitoring useful to identify suspected frequently occurring silent arrhythmias or to assess the overall arrhythmia burden. A typical Holter report can contain information on the heart rate (maximum, minimum, and average), ectopic beats, and tachy- and bradyarrhythmias, as well as representative samples.

The Holter device is familiar to most practitioners and remains an effective choice for ambulatory ECG monitoring. However, its use has largely been replaced by newer devices that overcome the Holter’s drawbacks, particularly its short duration of monitoring and the need for postmonitoring analysis. Additionally, although newer Holter devices are more ergonomic, some patients find the wires and gel electrodes uncomfortable or inconvenient.

Intermittent monitoring: Event recorders

Unlike the continuous monitors, intermittent recording devices (also called event recorders), capture and store tracings only during an event.

Intermittent recording monitors are of 2 general types: post-event recorders and loop recorders. These devices can extend the overall duration of observation, which can be especially useful for those whose symptoms and arrhythmias are infrequent.

Post-event recorders are small and self-contained, not requiring electrodes (Figure 1). The device is carried by the patient but not worn continuously. When the patient experiences symptoms, he or she places the device against the chest and presses a button to begin recording. These tracings are stored on the device and can be transmitted by telephone to a data center for analysis. Although post-event recorders allow for monitoring periods typically up to 30 days, they are limited by requiring the patient to act to record an event.

What is its best use? These devices are best used in patients who have infrequent symptoms and are at low risk. Transient or debilitating symptoms, including syncope, can limit the possibility of capturing an event.

Intermittent monitoring: Loop recorders

Loop recorders monitor continuously but record only intermittently. The name refers to the device’s looping memory: ie, to extend how long it can be used and make the most of its limited storage, the device records over previously captured data, saving only the most important data. The device saves the data whenever it detects an abnormal rhythm or the patient experiences symptoms and pushes a button. Data are recorded for a specified time before and after the activation, typically 30 seconds.

Loop recorders come in 2 types: external and implantable.

External loop recorders

External loop recorders look like Holter monitors (Figure 1), but they have the advantage of a much longer observation period—typically up to 1 month. The newest devices have even greater storage capacity and can provide “backward” memory, saving data that were captured just before the patient pushed the button.

In studies of patients with palpitations, presyncope, or syncope, external loop recorders had greater diagnostic yield than traditional 24-hour Holter monitors.^7,8 This finding was supported by a clinical trial that found 30-day monitoring with an external loop recorder led to a 5-fold increase in detecting atrial fibrillation in patients with cryptogenic stroke.⁹

Disadvantages of external loop recorders are limited memory storage, a considerable reliance on patient activation of the device, and wires and electrodes that need to be worn continuously.

What is their best use? External loop recorders are most effective when used to detect an arrhythmia or to correlate infrequent symptoms with an arrhythmia. They are most appropriately used in patients whose symptoms occur more often than every 4 weeks. They are less useful in assessing very infrequent symptoms, overall arrhythmia burden, or responsiveness to therapy.¹⁰

Implantable loop recorders are small devices that contain a pair of sensing electrodes housed within an outer shell (Figure 1). They are implanted subcutaneously, usually in the left parasternal region, using local anesthesia. The subcutaneous location eliminates many of the drawbacks of the skin-electrode interface of external loop recorders.

Similar to the external loop recorder, this device monitors continuously and can be activated to record either by the patient by pressing a button on a separate device, or automatically when an arrhythmia is detected using a preprogrammed algorithm.

In contrast to external devices, many internal loop recorders have a battery life and monitoring capability of up to 3 years. This extended monitoring period has been shown to increase the likelihood of diagnosing syncope or infrequent palpitations.^11,12 Given that paroxysmal atrial fibrillation can be sporadic and reveal itself months after a stroke, internal loop recorders may also have a role in evaluating cryptogenic stroke.^13,14

The most important drawbacks of internal loop recorders are the surgical procedure for insertion, their limited memory storage, and high upfront cost.¹⁵ Furthermore, even though they allow for extended monitoring, there may be diminishing returns for prolonged observation.

What is their best use? For patients with palpitations, intermittent event monitoring has been shown to be cost-effective for the first 2 weeks, but after 3 weeks, the cost per diagnosis increases dramatically.¹⁶ As a result, internal loop recorders are reserved primarily for scenarios in which prolonged external monitoring has not revealed a source of arrhythmia despite a high degree of suspicion.

Mobile cardiac telemetry

Mobile cardiac telemetry builds on other ECG monitoring systems by adding real-time communication and technician evaluation.

Physically, these devices resemble either hand-held event records, with a single-channel sensing unit embedded in the case, or a traditional Holter monitor, with 3 channels, wires, and electrodes  (Figure 1).

The sensor wirelessly communicates with a nearby portable monitor, which continuously observes and analyzes the patient’s heart rhythm. When an abnormal rhythm is detected or when the patient marks the presence of symptoms, data are recorded and sent in real time via a cellular network to a monitoring center; the newest monitors can send data via any Wi-Fi system. The rhythm is then either evaluated by a trained technician or relayed to a physician. If necessary, the patient can be contacted immediately.

Mobile cardiac telemetry is typically used for up to 30 days, which  allows for evaluation of less-frequent symptoms. As a result, it may have a higher diagnostic yield for palpitations, syncope, and presyncope than the 24-hour Holter monitor.¹⁷

Further, perhaps because mobile cardiac telemetry relies less on stored information and requires less patient-device interaction than external loop recorders, it is more effective at symptom evaluation.¹⁸

Mobile cardiac telemetry also has a diagnostic role in evaluating patients with cryptogenic stroke. This is based on studies showing it has a high rate of atrial fibrillation detection in this patient population and is more effective at determining overall atrial fibrillation burden than loop recorders.^18,19

What is its best use? The key advantage of mobile cardiac telemetry is its ability to make rhythm assessments and communicate with technicians in real time. This allows high-risk patients to be immediately alerted to a life-threatening arrhythmia. It also gives providers an opportunity to initiate anticoagulation or titrate antiarrhythmic therapy in the outpatient setting without a delay in obtaining information. This intensive monitoring, however, requires significant manpower, which translates to higher cost, averaging 3 times that of other standard external monitors.¹⁵

Patch monitors

These ultraportable devices are a relatively unobtrusive and easy-to-use alternative for short-term ambulatory ECG monitoring. They monitor continuously with full disclosure, outpatient telemetry, and post-event recording features.

Patch monitors are small, leadless, wireless, and water-resistant (Figure 1). They are affixed to the left pectoral region with a waterproof adhesive and can be worn for 14 to 28 days. Recording is usually done continuously; however, these devices have an event marker button that can be pressed when the user experiences symptoms. They acquire a single channel of data, and each manufacturer has a proprietary algorithm for automated rhythm detection and analysis.²⁰

Several manufacturers produce ECG patch monitors. Two notable devices are the Zio patch (iRhythm Technologies, San Francisco, CA) and the Mobile Cardiac Outpatient Telemetry patch (BioTelemetry, Inc, Malvern, PA).

The Zio patch is a continuous external monitor with full disclosure. It is comparable to the Holter monitor, but has a longer recording period. After completing a 2-week monitoring period, the device is returned for comprehensive rhythm analysis. A typical Zio report contains information on atrial fibrillation burden, ectopic rhythm burden, symptom and rhythm correlation, heart rate trends, and relevant rhythm strips.

The Mobile Cardiac Outpatient Telemetry patch collects data continuously and communicates wirelessly by Bluetooth to send its ECG data to a monitoring center for evaluation.

A principal advantage of patch monitors—and a major selling point for manufacturers—is their low-profile, ergonomic, and patient-friendly design. Patients do not have to manage wires or batteries and are able to shower with their devices. Studies show that these features increase patient satisfaction and compliance, resulting in increased diagnostic yield.^21,22 Additionally, patch monitors have the advantage of a longer continuous monitoring period than traditional Holter devices (2 weeks vs 1 or 2 days), affording an opportunity to capture events that occur less frequently.

Validation studies have reinforced their efficacy and utility in clinical scenarios.^22,23 In large part because of the extended monitoring period, patch monitors have been shown to have greater diagnostic yield than the 24-hour Holter monitor in symptomatic patients undergoing workup for suspected arrhythmia.

The role of patch monitors in evaluating atrial fibrillation is also being established. For patients with cryptogenic stroke, patch monitors have shown better atrial fibrillation detection than the 24-hour Holter monitor.²⁴ Compared with traditional loop monitors, patch monitors have the added advantage of assessing total atrial fibrillation burden. Further, although screening for atrial fibrillation with a traditional 12-lead ECG monitor has not been shown to be effective, clinical studies have found that the patch monitor may be a useful screening tool for high-risk patients.^25,26

Nevertheless, patch monitors have drawbacks. They are not capable of long-term monitoring, owing to battery and adhesive limitations.²⁰ More important, they have  been able to offer only single-channel acquisition, which makes it more difficult to detect an arrhythmia that is characterized by a change in QRS axis or change in QRS width, or to distinguish an arrhythmia from an artifact. This appears to be changing, however, as several manufacturers have recently developed multilead ECG patch monitors or attachments and are attempting to merge this technology with fully capable remote telemetry.

Recent improvements in battery life, memory, detection algorithms, wireless transmission, cellular communication, and adhesives have enabled multiple features to be combined into a single device. Patch monitors, for example, are small devices that now offer full-disclosure recording, extended monitoring, and telemetry transmitting. Automated arrhythmia recognition that triggers recording is central to all modern devices, regardless of type.

As a result of these trends, the traditional features used to differentiate devices may become less applicable. The classic Holter monitor may become obsolete as its advantages (full disclosure, continuous recording) are being incorporated into smaller devices that can record longer. Similarly, external monitors that have the capacity for full disclosure and continuous recording are no longer loop recorders in that they do not record into a circular memory.

It may be preferable to describe all non-Holter devices as event monitors or ambulatory monitors, with the main distinguishing features being the ability to transmit data (telemetry), full disclosure vs patient- or arrhythmia-activated recording, and single-channel or multichannel recording (single-lead or 3-lead ECG).

The following are the main distinguishing features that should influence the choice of device for a given clinical context.

Real-time data evaluation provided by mobile telemetry makes this feature ideal to monitor patients with suspected high-risk arrhythmias and their response to antiarrhythmic therapy.

Full-disclosure recording is necessary to assess the overall burden of an arrhythmia, which is frequently important in making treatment decisions, risk-stratifying, and assessing response to therapy. In contrast, patient- or arrhythmia-activated devices are best used when the goal is simply to establish the presence of an arrhythmia.

Multichannel recording may be better than single-channel recording, as it is needed to determine the anatomic origin of an arrhythmia, as might be the case in risk-stratification in a patient with a ventricular tachycardia.

Long duration. The clinician must have a reasonable estimate of how often the symptoms or arrhythmia occur to determine which device will offer a monitoring duration sufficient to detect an arrhythmia.

NEWER TECHNOLOGIES

The newest ambulatory ECG devices build on the foundational concepts of the older ones. However, with miniaturized electronic circuits, Bluetooth, Wi-Fi, and smartphones, these new devices can capture ECG tracings and diagnose offending arrhythmias on more consumer-friendly devices.

Smartphones and smartwatches have become increasingly powerful. Some have the ability to capture, display, and record the cardiac waveform. One manufacturer to capitalize on these technologies, AliveCor (Mountain View, CA), has developed 2 products capable of generating a single-lead ECG recording using either a smartphone (KardiaMobile) or an Apple watch (KardiaBand).

KardiaMobile has a 2-electrode band that can be carried in a pocket or attached to the back of a smartphone (Figure 1). The user places 1 or 2 fingers from each hand on the electrodes, and the device sends an ultrasound signal that is picked up by the smartphone’s microphone. The signal is digitized to produce a 30-second ECG tracing on the phone’s screen. A proprietary algorithm analyzes the rhythm and generates a description of “normal” or “possible atrial fibrillation.” The ECG is then uploaded to a cloud-based storage system for later access or transmission. KardiaMobile is compatible with both iOS and Android devices.

The KardiaBand is a specialized Apple watch band that has an electrode embedded in it. The user places a thumb on the electrode for 30 seconds, and an ECG tracing is displayed on the watch screen.

The Kardia devices were developed (and advertised) predominantly to assess atrial fibrillation. Studies have validated the accuracy of their algorithm. One study showed that, compared with physician-interpreted ECGs, the algorithm had a 96.6% sensitivity and 94.1% specificity for detecting atrial fibrillation.²⁷ They have been found useful for detecting and evaluating atrial fibrillation in several clinical scenarios, including discharge monitoring in patients after ablation or cardiac surgery.^28,29 In a longer study of patients at risk of stroke, twice-weekly ECG screening using a Kardia device for 1 year was more likely to detect incident atrial fibrillation than routine care alone.³⁰

Also, the Kardia devices can effectively function as post-event recorders when activated by patients when they experience symptoms. In a small study of outpatients with palpitations and a prior nondiagnostic workup, the KardiaMobile device was found to be noninferior to external loop recorders for detecting arrhythmias.³¹ Additional studies are assessing Kardia’s utility in other scenarios, including the evaluation of ST-segment elevation myocardial infarction^32,33 and QT interval for patients receiving antiarrhythmic therapy.³⁴

Cardiio Inc. (Cambridge, MA) has developed technology to screen for atrial fibrillation using an app that requires no additional external hardware. Instead, the app uses a smartphone’s camera and flashlight to perform photo­plethysmography to detect pulsatile changes in blood volume and generate a waveform. Based on waveform variability, a proprietary algorithm attempts to determine whether the user is in atrial fibrillation. It does not produce an ECG tracing. Initial studies suggest it has good diagnostic accuracy and potential utility as a population-based screening tool,^35,36 but it has not been fully validated.

Recently, Apple entered the arena of ambulatory cardiac monitoring with the release of its fourth-generation watch (Apple Watch Series 4 model). This watch has built-in electrodes that can generate a single-lead ECG on the watch screen. Its algorithm can discriminate between atrial fibrillation and sinus rhythm, but it has not been assessed for its ability to evaluate other arrhythmias. Even though it has been “cleared” by the US Food and Drug Administration, it is approved only for informational use, not to make a medical diagnosis.

Integration of ambulatory ECG technology with smartphone and watch technology is an exciting new wearable option for arrhythmia detection. The patient-centered and controlled nature of these devices have the potential to help patients with palpitations or other symptoms determine if their cardiac rhythms are normal.

This technology, however, is still in its infancy and has many limitations. For example, even though these devices can function as post-event recorders, they depend on user-device interactions. Plus, they cannot yet perform continuous arrhythmia monitoring like modern loop recorders.

Additionally, automated analysis has largely been limited to distinguishing atrial fibrillation from normal sinus rhythm. It is uncertain how effective the devices may be in evaluating other arrhythmias. Single-lead ECG recordings, as discussed, have limited interpretability and value. And even though studies have shown utility in certain clinical scenarios, large-scale validation studies are lacking. This technology will likely continue to be developed and its clinical value improved; however, its clinical use requires careful consideration and collaborative physician-patient decision-making.

The development of smartphone and watch ECG technology has led to a rise in direct-to-consumer healthcare delivery. By devising technology that is appealing, useful, and affordable, companies can bypass the insurer and practitioner by targeting increasingly health-literate consumers. For many companies, there is great motivation to enter this healthcare space. Wearable devices are immensely popular and, as a result, generate substantial revenue. One analysis estimates that 1 in 10 Americans (nearly 30 million) owns a wearable, smart-technology device.³⁷

This direct-to-consumer approach has specific implications for cardiology and, more broadly, for healthcare overall. By directly selling to consumers, companies have an opportunity to reach many more people. The Apple Watch Series 4 has taken this a step further: by including this technology in the watch, consumers not necessarily seeking an ambulatory cardiac monitor will have one with a watch purchase. This could lead to increases in monitoring and could alert people to previously undiagnosed disorders.

For consumers, this technology can empower them to choose how and when to be monitored. Further, it gives them personal control of their healthcare data, and helps move the point of care out of hospitals and clinics and into the home.

But wearable medical technology and direct-to-consumer healthcare have risks. First, in the absence of appropriate regulation, patients have to distinguish between products that are well validated and those that are unproven. Consumers also may inappropriately use devices for indications or in scenarios for which the value is uncertain.

Also, there is potential for confusion and misunderstanding of results, including false-positive readings, which could lead to excessive and costly use of unnecessary diagnostic workups. Instead of providing peace of mind, these devices could cause greater worry. This may be especially true with the newest Apple watch, as this product will introduce ambulatory ECG to a younger and healthier segment of the population who are less likely to have true disease.

Further, these devices have algorithms that detect atrial fibrillation, but is it the same as that detected by traditional methods? Sometimes termed “subclinical” atrial fibrillation, it poses uncertainties: ie, Do patients need anticoagulation, pharmacologic therapy, and ablation? The optimal management of subclinical atrial fibrillation, as well as its similarities to and differences from atrial fibrillation diagnosed by traditional methods, are topics that need further study.

Wearable technology is still developing and will continue to do so. Medical practice will have to adapt to it.

FUTURE DIRECTIONS

Changes in technology have led to remarkable advances in the convenience and accuracy of ambulatory ECG monitoring. Ongoing research is expected to lead to even more improvements. Devices will become more ergonomic and technically capable, and they may expand monitoring to include other biologic parameters beyond ECG.

Comfort is important to ensure patient adherence. Newer, flexible electronics embedded in ultrathin materials can potentially improve the wearability of devices that require gel electrodes or adhesive patches.³⁸ Wireless technology may obviate the need for on-skin attachments. Future recording systems may be embedded into clothing or incorporated into wearable vests capable of wirelessly transmitting ECG signals to separate recording stations.³⁹

In addition to becoming smaller and more comfortable, future devices will be more technically capable, leading to a merging of technologies that will further blur the distinctions among devices. Eventually, the features of full disclosure, extended monitoring duration, and telemetric communication will all be present together. Perhaps more important is that ambulatory ECG devices may become fully capable biosensor monitors. These devices would have the potential to monitor respiratory frequency, peripheral oxygen saturation, potassium levels, and arterial pulse pressure.^39,40

A mbulatory electrocardiography (ECG) began in 1949 when Norman “Jeff” Holter developed a monitor that could wirelessly transmit electrophysiologic data.¹ His original device used vacuum tubes, weighed 85 pounds, and had to be carried in a backpack. Furthermore, it could send a signal a distance of only 1 block.²

At the time, it was uncertain if this technology would have any clinical utility. However, in 1952, Holter published the first tracing of abnormal cardiac electrical activity in a patient who had suffered a posterior myocardial infarction.³ By the 1960s, Holter monitoring systems were in full production and use.⁴

Since then, advances in technology have led to small, lightweight devices that enable clinicians to evaluate patients for arrhythmias in a real-world context for extended times, often with the ability to respond in real time.

Many ambulatory devices are available, and choosing the optimal one requires an understanding of which features they have and which are the most appropriate for the specific clinical context. This article reviews the features, indications, advantages, and disadvantages of current devices, and their best use in clinical practice.

INDICATIONS FOR AMBULATORY ECG MONITORING

Diagnosis

The most common diagnostic role of monitoring is to correlate unexplained symptoms, including palpitations, presyncope, and syncope, with a transient cardiac arrhythmia. Monitoring can be considered successful if findings on ECG identify risks for serious arrhythmia and either correlate symptoms with those findings or demonstrate no arrhythmia when symptoms occur.

A range of arrhythmias can cause symptoms. Some, such as premature atrial contractions and premature ventricular contractions, may be benign in many clinical contexts. Others, such as atrial fibrillation, are more serious, and some, such as third-degree heart block and ventricular tachycardia, can be lethal.

Arrhythmia symptoms can vary in frequency and cause differing degrees of debility. The patient’s symptoms, family history, and baseline ECG findings can suggest a more serious or a less serious underlying rhythm. These factors are important when determining which device is most appropriate.

Ambulatory ECG can also be useful in looking for a cause of cryptogenic stroke, ie, an ischemic stroke with an unexplained cause, even after a thorough initial workup. Paroxysmal atrial fibrillation is a frequent cause of cryptogenic stroke, and because it is transient, short-term inpatient telemetry may not be sufficient to detect it. Extended cardiac monitoring, lasting weeks or even months, is often needed for clinicians to make this diagnosis and initiate appropriate secondary prevention.

Prognosis: Identifying patients at risk

In a patient with known structural or electrical heart disease, ambulatory ECG can be used to stratify risk. This is particularly true in evaluating conditions associated with sudden cardiac death.

For example, hypertrophic cardiomyopathy and arrhythmogenic right ventricular dysplasia or cardiomyopathy are 2 cardiomyopathies that can manifest clinically with ventricular arrhythmias and sudden cardiac death. Ambulatory ECG can detect premature ventricular contractions and ventricular tachycardia and identify their frequency, duration, and anatomic origin. This information is useful in assessing risk of sudden cardiac death and determining the need for an implantable cardioverter-defibrillator.

Similarly, Wolff-Parkinson-White syndrome, involving rapid conduction through an accessory pathway, is associated with increased risk of ventricular fibrillation and sudden cardiac death. Ambulatory ECG monitoring can identify patients who have electrical features that portend the development of ventricular fibrillation.

Also associated with sudden cardiac death are the inherited channelopathies, a heterogeneous group of primary arrhythmic disorders without accompanying structural pathology. Ambulatory ECG monitoring can detect transient electrical changes and nonsustained ventricular arrhythmias that would indicate the patient is at high risk of these disorders.

Assessing arrhythmia treatment

Arrhythmia monitoring using an ambulatory ECG device can also provide data to assess the efficacy of treatment under several circumstances.

The “pill-in-the-pocket” approach to treating atrial fibrillation, for example, involves self-administering a single dose of an antiarrhythmic drug when symptoms occur. Patients with infrequent but bothersome episodes can use an ambulatory ECG device to detect when they are having atrial fibrillation, take their prescribed drug, and see whether it terminates the arrhythmia, all without going to the hospital.

Ambulatory ECG also is useful for assessing pharmacologic or ablative therapy in patients with atrial fibrillation or ventricular tachycardia. Monitoring for several weeks can help clinicians assess the burden of atrial fibrillation when using a rhythm-control strategy; assessing the ventricular rate in real-world situations is useful to determine the success of a rate-control strategy. Shortly after ablation of either atrial fibrillation or ventricular tachycardia, ECG home monitoring for 24 to 48 hours can detect asymptomatic recurrence and treatment failure.

Some antiarrhythmic drugs can prolong the QT interval. Ambulatory ECG devices that feature real-time monitoring can be used during drug initiation, enabling the clinician to monitor the QT interval without admitting the patient to the hospital.

Ultimately, ambulatory ECG monitoring is most commonly used to evaluate symptoms. Because arrhythmias and specific symptoms are unpredictable and transient, extended monitoring in a real-world setting allows for a more comprehensive evaluation than a standard 10-second ECG recording.

Continuous external monitoring: The Holter monitor

Recording is typically done continuously for 24 to 48 hours, although some newer devices can record for longer. Patients can press a button to note when they are experiencing symptoms, allowing for potential correlation with ECG abnormalities. The data are stored on a flash drive that can be uploaded for analysis after recording is complete.

What is its best use? Given its relatively short duration of monitoring, the Holter device is typically used to evaluate symptoms that occur daily or nearly daily. An advantage of the Holter monitor is its ability to record continuously, without requiring the patient to interact with the device. This feature provides “full disclosure,” which is the ability to see arrhythmia data from the entire recording period.

These features make Holter monitoring useful to identify suspected frequently occurring silent arrhythmias or to assess the overall arrhythmia burden. A typical Holter report can contain information on the heart rate (maximum, minimum, and average), ectopic beats, and tachy- and bradyarrhythmias, as well as representative samples.

The Holter device is familiar to most practitioners and remains an effective choice for ambulatory ECG monitoring. However, its use has largely been replaced by newer devices that overcome the Holter’s drawbacks, particularly its short duration of monitoring and the need for postmonitoring analysis. Additionally, although newer Holter devices are more ergonomic, some patients find the wires and gel electrodes uncomfortable or inconvenient.

Intermittent monitoring: Event recorders

Unlike the continuous monitors, intermittent recording devices (also called event recorders), capture and store tracings only during an event.

Intermittent recording monitors are of 2 general types: post-event recorders and loop recorders. These devices can extend the overall duration of observation, which can be especially useful for those whose symptoms and arrhythmias are infrequent.

Post-event recorders are small and self-contained, not requiring electrodes (Figure 1). The device is carried by the patient but not worn continuously. When the patient experiences symptoms, he or she places the device against the chest and presses a button to begin recording. These tracings are stored on the device and can be transmitted by telephone to a data center for analysis. Although post-event recorders allow for monitoring periods typically up to 30 days, they are limited by requiring the patient to act to record an event.

What is its best use? These devices are best used in patients who have infrequent symptoms and are at low risk. Transient or debilitating symptoms, including syncope, can limit the possibility of capturing an event.

Intermittent monitoring: Loop recorders

Loop recorders monitor continuously but record only intermittently. The name refers to the device’s looping memory: ie, to extend how long it can be used and make the most of its limited storage, the device records over previously captured data, saving only the most important data. The device saves the data whenever it detects an abnormal rhythm or the patient experiences symptoms and pushes a button. Data are recorded for a specified time before and after the activation, typically 30 seconds.

Loop recorders come in 2 types: external and implantable.

External loop recorders

External loop recorders look like Holter monitors (Figure 1), but they have the advantage of a much longer observation period—typically up to 1 month. The newest devices have even greater storage capacity and can provide “backward” memory, saving data that were captured just before the patient pushed the button.

In studies of patients with palpitations, presyncope, or syncope, external loop recorders had greater diagnostic yield than traditional 24-hour Holter monitors.^7,8 This finding was supported by a clinical trial that found 30-day monitoring with an external loop recorder led to a 5-fold increase in detecting atrial fibrillation in patients with cryptogenic stroke.⁹

Disadvantages of external loop recorders are limited memory storage, a considerable reliance on patient activation of the device, and wires and electrodes that need to be worn continuously.

What is their best use? External loop recorders are most effective when used to detect an arrhythmia or to correlate infrequent symptoms with an arrhythmia. They are most appropriately used in patients whose symptoms occur more often than every 4 weeks. They are less useful in assessing very infrequent symptoms, overall arrhythmia burden, or responsiveness to therapy.¹⁰

Implantable loop recorders are small devices that contain a pair of sensing electrodes housed within an outer shell (Figure 1). They are implanted subcutaneously, usually in the left parasternal region, using local anesthesia. The subcutaneous location eliminates many of the drawbacks of the skin-electrode interface of external loop recorders.

Similar to the external loop recorder, this device monitors continuously and can be activated to record either by the patient by pressing a button on a separate device, or automatically when an arrhythmia is detected using a preprogrammed algorithm.

In contrast to external devices, many internal loop recorders have a battery life and monitoring capability of up to 3 years. This extended monitoring period has been shown to increase the likelihood of diagnosing syncope or infrequent palpitations.^11,12 Given that paroxysmal atrial fibrillation can be sporadic and reveal itself months after a stroke, internal loop recorders may also have a role in evaluating cryptogenic stroke.^13,14

The most important drawbacks of internal loop recorders are the surgical procedure for insertion, their limited memory storage, and high upfront cost.¹⁵ Furthermore, even though they allow for extended monitoring, there may be diminishing returns for prolonged observation.

What is their best use? For patients with palpitations, intermittent event monitoring has been shown to be cost-effective for the first 2 weeks, but after 3 weeks, the cost per diagnosis increases dramatically.¹⁶ As a result, internal loop recorders are reserved primarily for scenarios in which prolonged external monitoring has not revealed a source of arrhythmia despite a high degree of suspicion.

Mobile cardiac telemetry

Mobile cardiac telemetry builds on other ECG monitoring systems by adding real-time communication and technician evaluation.

Physically, these devices resemble either hand-held event records, with a single-channel sensing unit embedded in the case, or a traditional Holter monitor, with 3 channels, wires, and electrodes  (Figure 1).

The sensor wirelessly communicates with a nearby portable monitor, which continuously observes and analyzes the patient’s heart rhythm. When an abnormal rhythm is detected or when the patient marks the presence of symptoms, data are recorded and sent in real time via a cellular network to a monitoring center; the newest monitors can send data via any Wi-Fi system. The rhythm is then either evaluated by a trained technician or relayed to a physician. If necessary, the patient can be contacted immediately.

Mobile cardiac telemetry is typically used for up to 30 days, which  allows for evaluation of less-frequent symptoms. As a result, it may have a higher diagnostic yield for palpitations, syncope, and presyncope than the 24-hour Holter monitor.¹⁷

Further, perhaps because mobile cardiac telemetry relies less on stored information and requires less patient-device interaction than external loop recorders, it is more effective at symptom evaluation.¹⁸

Mobile cardiac telemetry also has a diagnostic role in evaluating patients with cryptogenic stroke. This is based on studies showing it has a high rate of atrial fibrillation detection in this patient population and is more effective at determining overall atrial fibrillation burden than loop recorders.^18,19

What is its best use? The key advantage of mobile cardiac telemetry is its ability to make rhythm assessments and communicate with technicians in real time. This allows high-risk patients to be immediately alerted to a life-threatening arrhythmia. It also gives providers an opportunity to initiate anticoagulation or titrate antiarrhythmic therapy in the outpatient setting without a delay in obtaining information. This intensive monitoring, however, requires significant manpower, which translates to higher cost, averaging 3 times that of other standard external monitors.¹⁵

Patch monitors

These ultraportable devices are a relatively unobtrusive and easy-to-use alternative for short-term ambulatory ECG monitoring. They monitor continuously with full disclosure, outpatient telemetry, and post-event recording features.

Patch monitors are small, leadless, wireless, and water-resistant (Figure 1). They are affixed to the left pectoral region with a waterproof adhesive and can be worn for 14 to 28 days. Recording is usually done continuously; however, these devices have an event marker button that can be pressed when the user experiences symptoms. They acquire a single channel of data, and each manufacturer has a proprietary algorithm for automated rhythm detection and analysis.²⁰

Several manufacturers produce ECG patch monitors. Two notable devices are the Zio patch (iRhythm Technologies, San Francisco, CA) and the Mobile Cardiac Outpatient Telemetry patch (BioTelemetry, Inc, Malvern, PA).

The Zio patch is a continuous external monitor with full disclosure. It is comparable to the Holter monitor, but has a longer recording period. After completing a 2-week monitoring period, the device is returned for comprehensive rhythm analysis. A typical Zio report contains information on atrial fibrillation burden, ectopic rhythm burden, symptom and rhythm correlation, heart rate trends, and relevant rhythm strips.

The Mobile Cardiac Outpatient Telemetry patch collects data continuously and communicates wirelessly by Bluetooth to send its ECG data to a monitoring center for evaluation.

A principal advantage of patch monitors—and a major selling point for manufacturers—is their low-profile, ergonomic, and patient-friendly design. Patients do not have to manage wires or batteries and are able to shower with their devices. Studies show that these features increase patient satisfaction and compliance, resulting in increased diagnostic yield.^21,22 Additionally, patch monitors have the advantage of a longer continuous monitoring period than traditional Holter devices (2 weeks vs 1 or 2 days), affording an opportunity to capture events that occur less frequently.

Validation studies have reinforced their efficacy and utility in clinical scenarios.^22,23 In large part because of the extended monitoring period, patch monitors have been shown to have greater diagnostic yield than the 24-hour Holter monitor in symptomatic patients undergoing workup for suspected arrhythmia.

The role of patch monitors in evaluating atrial fibrillation is also being established. For patients with cryptogenic stroke, patch monitors have shown better atrial fibrillation detection than the 24-hour Holter monitor.²⁴ Compared with traditional loop monitors, patch monitors have the added advantage of assessing total atrial fibrillation burden. Further, although screening for atrial fibrillation with a traditional 12-lead ECG monitor has not been shown to be effective, clinical studies have found that the patch monitor may be a useful screening tool for high-risk patients.^25,26

Nevertheless, patch monitors have drawbacks. They are not capable of long-term monitoring, owing to battery and adhesive limitations.²⁰ More important, they have  been able to offer only single-channel acquisition, which makes it more difficult to detect an arrhythmia that is characterized by a change in QRS axis or change in QRS width, or to distinguish an arrhythmia from an artifact. This appears to be changing, however, as several manufacturers have recently developed multilead ECG patch monitors or attachments and are attempting to merge this technology with fully capable remote telemetry.

Recent improvements in battery life, memory, detection algorithms, wireless transmission, cellular communication, and adhesives have enabled multiple features to be combined into a single device. Patch monitors, for example, are small devices that now offer full-disclosure recording, extended monitoring, and telemetry transmitting. Automated arrhythmia recognition that triggers recording is central to all modern devices, regardless of type.

As a result of these trends, the traditional features used to differentiate devices may become less applicable. The classic Holter monitor may become obsolete as its advantages (full disclosure, continuous recording) are being incorporated into smaller devices that can record longer. Similarly, external monitors that have the capacity for full disclosure and continuous recording are no longer loop recorders in that they do not record into a circular memory.

It may be preferable to describe all non-Holter devices as event monitors or ambulatory monitors, with the main distinguishing features being the ability to transmit data (telemetry), full disclosure vs patient- or arrhythmia-activated recording, and single-channel or multichannel recording (single-lead or 3-lead ECG).

The following are the main distinguishing features that should influence the choice of device for a given clinical context.

Real-time data evaluation provided by mobile telemetry makes this feature ideal to monitor patients with suspected high-risk arrhythmias and their response to antiarrhythmic therapy.

Full-disclosure recording is necessary to assess the overall burden of an arrhythmia, which is frequently important in making treatment decisions, risk-stratifying, and assessing response to therapy. In contrast, patient- or arrhythmia-activated devices are best used when the goal is simply to establish the presence of an arrhythmia.

Multichannel recording may be better than single-channel recording, as it is needed to determine the anatomic origin of an arrhythmia, as might be the case in risk-stratification in a patient with a ventricular tachycardia.

Long duration. The clinician must have a reasonable estimate of how often the symptoms or arrhythmia occur to determine which device will offer a monitoring duration sufficient to detect an arrhythmia.

NEWER TECHNOLOGIES

The newest ambulatory ECG devices build on the foundational concepts of the older ones. However, with miniaturized electronic circuits, Bluetooth, Wi-Fi, and smartphones, these new devices can capture ECG tracings and diagnose offending arrhythmias on more consumer-friendly devices.

Smartphones and smartwatches have become increasingly powerful. Some have the ability to capture, display, and record the cardiac waveform. One manufacturer to capitalize on these technologies, AliveCor (Mountain View, CA), has developed 2 products capable of generating a single-lead ECG recording using either a smartphone (KardiaMobile) or an Apple watch (KardiaBand).

KardiaMobile has a 2-electrode band that can be carried in a pocket or attached to the back of a smartphone (Figure 1). The user places 1 or 2 fingers from each hand on the electrodes, and the device sends an ultrasound signal that is picked up by the smartphone’s microphone. The signal is digitized to produce a 30-second ECG tracing on the phone’s screen. A proprietary algorithm analyzes the rhythm and generates a description of “normal” or “possible atrial fibrillation.” The ECG is then uploaded to a cloud-based storage system for later access or transmission. KardiaMobile is compatible with both iOS and Android devices.

The KardiaBand is a specialized Apple watch band that has an electrode embedded in it. The user places a thumb on the electrode for 30 seconds, and an ECG tracing is displayed on the watch screen.

The Kardia devices were developed (and advertised) predominantly to assess atrial fibrillation. Studies have validated the accuracy of their algorithm. One study showed that, compared with physician-interpreted ECGs, the algorithm had a 96.6% sensitivity and 94.1% specificity for detecting atrial fibrillation.²⁷ They have been found useful for detecting and evaluating atrial fibrillation in several clinical scenarios, including discharge monitoring in patients after ablation or cardiac surgery.^28,29 In a longer study of patients at risk of stroke, twice-weekly ECG screening using a Kardia device for 1 year was more likely to detect incident atrial fibrillation than routine care alone.³⁰

Also, the Kardia devices can effectively function as post-event recorders when activated by patients when they experience symptoms. In a small study of outpatients with palpitations and a prior nondiagnostic workup, the KardiaMobile device was found to be noninferior to external loop recorders for detecting arrhythmias.³¹ Additional studies are assessing Kardia’s utility in other scenarios, including the evaluation of ST-segment elevation myocardial infarction^32,33 and QT interval for patients receiving antiarrhythmic therapy.³⁴

Cardiio Inc. (Cambridge, MA) has developed technology to screen for atrial fibrillation using an app that requires no additional external hardware. Instead, the app uses a smartphone’s camera and flashlight to perform photo­plethysmography to detect pulsatile changes in blood volume and generate a waveform. Based on waveform variability, a proprietary algorithm attempts to determine whether the user is in atrial fibrillation. It does not produce an ECG tracing. Initial studies suggest it has good diagnostic accuracy and potential utility as a population-based screening tool,^35,36 but it has not been fully validated.

Recently, Apple entered the arena of ambulatory cardiac monitoring with the release of its fourth-generation watch (Apple Watch Series 4 model). This watch has built-in electrodes that can generate a single-lead ECG on the watch screen. Its algorithm can discriminate between atrial fibrillation and sinus rhythm, but it has not been assessed for its ability to evaluate other arrhythmias. Even though it has been “cleared” by the US Food and Drug Administration, it is approved only for informational use, not to make a medical diagnosis.

Integration of ambulatory ECG technology with smartphone and watch technology is an exciting new wearable option for arrhythmia detection. The patient-centered and controlled nature of these devices have the potential to help patients with palpitations or other symptoms determine if their cardiac rhythms are normal.

This technology, however, is still in its infancy and has many limitations. For example, even though these devices can function as post-event recorders, they depend on user-device interactions. Plus, they cannot yet perform continuous arrhythmia monitoring like modern loop recorders.

Additionally, automated analysis has largely been limited to distinguishing atrial fibrillation from normal sinus rhythm. It is uncertain how effective the devices may be in evaluating other arrhythmias. Single-lead ECG recordings, as discussed, have limited interpretability and value. And even though studies have shown utility in certain clinical scenarios, large-scale validation studies are lacking. This technology will likely continue to be developed and its clinical value improved; however, its clinical use requires careful consideration and collaborative physician-patient decision-making.

The development of smartphone and watch ECG technology has led to a rise in direct-to-consumer healthcare delivery. By devising technology that is appealing, useful, and affordable, companies can bypass the insurer and practitioner by targeting increasingly health-literate consumers. For many companies, there is great motivation to enter this healthcare space. Wearable devices are immensely popular and, as a result, generate substantial revenue. One analysis estimates that 1 in 10 Americans (nearly 30 million) owns a wearable, smart-technology device.³⁷

This direct-to-consumer approach has specific implications for cardiology and, more broadly, for healthcare overall. By directly selling to consumers, companies have an opportunity to reach many more people. The Apple Watch Series 4 has taken this a step further: by including this technology in the watch, consumers not necessarily seeking an ambulatory cardiac monitor will have one with a watch purchase. This could lead to increases in monitoring and could alert people to previously undiagnosed disorders.

For consumers, this technology can empower them to choose how and when to be monitored. Further, it gives them personal control of their healthcare data, and helps move the point of care out of hospitals and clinics and into the home.

But wearable medical technology and direct-to-consumer healthcare have risks. First, in the absence of appropriate regulation, patients have to distinguish between products that are well validated and those that are unproven. Consumers also may inappropriately use devices for indications or in scenarios for which the value is uncertain.

Also, there is potential for confusion and misunderstanding of results, including false-positive readings, which could lead to excessive and costly use of unnecessary diagnostic workups. Instead of providing peace of mind, these devices could cause greater worry. This may be especially true with the newest Apple watch, as this product will introduce ambulatory ECG to a younger and healthier segment of the population who are less likely to have true disease.

Further, these devices have algorithms that detect atrial fibrillation, but is it the same as that detected by traditional methods? Sometimes termed “subclinical” atrial fibrillation, it poses uncertainties: ie, Do patients need anticoagulation, pharmacologic therapy, and ablation? The optimal management of subclinical atrial fibrillation, as well as its similarities to and differences from atrial fibrillation diagnosed by traditional methods, are topics that need further study.

Wearable technology is still developing and will continue to do so. Medical practice will have to adapt to it.

FUTURE DIRECTIONS

Changes in technology have led to remarkable advances in the convenience and accuracy of ambulatory ECG monitoring. Ongoing research is expected to lead to even more improvements. Devices will become more ergonomic and technically capable, and they may expand monitoring to include other biologic parameters beyond ECG.

Comfort is important to ensure patient adherence. Newer, flexible electronics embedded in ultrathin materials can potentially improve the wearability of devices that require gel electrodes or adhesive patches.³⁸ Wireless technology may obviate the need for on-skin attachments. Future recording systems may be embedded into clothing or incorporated into wearable vests capable of wirelessly transmitting ECG signals to separate recording stations.³⁹

In addition to becoming smaller and more comfortable, future devices will be more technically capable, leading to a merging of technologies that will further blur the distinctions among devices. Eventually, the features of full disclosure, extended monitoring duration, and telemetric communication will all be present together. Perhaps more important is that ambulatory ECG devices may become fully capable biosensor monitors. These devices would have the potential to monitor respiratory frequency, peripheral oxygen saturation, potassium levels, and arterial pulse pressure.^39,40

A mbulatory electrocardiography (ECG) began in 1949 when Norman “Jeff” Holter developed a monitor that could wirelessly transmit electrophysiologic data.¹ His original device used vacuum tubes, weighed 85 pounds, and had to be carried in a backpack. Furthermore, it could send a signal a distance of only 1 block.²

At the time, it was uncertain if this technology would have any clinical utility. However, in 1952, Holter published the first tracing of abnormal cardiac electrical activity in a patient who had suffered a posterior myocardial infarction.³ By the 1960s, Holter monitoring systems were in full production and use.⁴

Since then, advances in technology have led to small, lightweight devices that enable clinicians to evaluate patients for arrhythmias in a real-world context for extended times, often with the ability to respond in real time.

Many ambulatory devices are available, and choosing the optimal one requires an understanding of which features they have and which are the most appropriate for the specific clinical context. This article reviews the features, indications, advantages, and disadvantages of current devices, and their best use in clinical practice.

INDICATIONS FOR AMBULATORY ECG MONITORING

Diagnosis

The most common diagnostic role of monitoring is to correlate unexplained symptoms, including palpitations, presyncope, and syncope, with a transient cardiac arrhythmia. Monitoring can be considered successful if findings on ECG identify risks for serious arrhythmia and either correlate symptoms with those findings or demonstrate no arrhythmia when symptoms occur.

A range of arrhythmias can cause symptoms. Some, such as premature atrial contractions and premature ventricular contractions, may be benign in many clinical contexts. Others, such as atrial fibrillation, are more serious, and some, such as third-degree heart block and ventricular tachycardia, can be lethal.

Arrhythmia symptoms can vary in frequency and cause differing degrees of debility. The patient’s symptoms, family history, and baseline ECG findings can suggest a more serious or a less serious underlying rhythm. These factors are important when determining which device is most appropriate.

Ambulatory ECG can also be useful in looking for a cause of cryptogenic stroke, ie, an ischemic stroke with an unexplained cause, even after a thorough initial workup. Paroxysmal atrial fibrillation is a frequent cause of cryptogenic stroke, and because it is transient, short-term inpatient telemetry may not be sufficient to detect it. Extended cardiac monitoring, lasting weeks or even months, is often needed for clinicians to make this diagnosis and initiate appropriate secondary prevention.

Prognosis: Identifying patients at risk

In a patient with known structural or electrical heart disease, ambulatory ECG can be used to stratify risk. This is particularly true in evaluating conditions associated with sudden cardiac death.

For example, hypertrophic cardiomyopathy and arrhythmogenic right ventricular dysplasia or cardiomyopathy are 2 cardiomyopathies that can manifest clinically with ventricular arrhythmias and sudden cardiac death. Ambulatory ECG can detect premature ventricular contractions and ventricular tachycardia and identify their frequency, duration, and anatomic origin. This information is useful in assessing risk of sudden cardiac death and determining the need for an implantable cardioverter-defibrillator.

Similarly, Wolff-Parkinson-White syndrome, involving rapid conduction through an accessory pathway, is associated with increased risk of ventricular fibrillation and sudden cardiac death. Ambulatory ECG monitoring can identify patients who have electrical features that portend the development of ventricular fibrillation.

Also associated with sudden cardiac death are the inherited channelopathies, a heterogeneous group of primary arrhythmic disorders without accompanying structural pathology. Ambulatory ECG monitoring can detect transient electrical changes and nonsustained ventricular arrhythmias that would indicate the patient is at high risk of these disorders.

Assessing arrhythmia treatment

Arrhythmia monitoring using an ambulatory ECG device can also provide data to assess the efficacy of treatment under several circumstances.

The “pill-in-the-pocket” approach to treating atrial fibrillation, for example, involves self-administering a single dose of an antiarrhythmic drug when symptoms occur. Patients with infrequent but bothersome episodes can use an ambulatory ECG device to detect when they are having atrial fibrillation, take their prescribed drug, and see whether it terminates the arrhythmia, all without going to the hospital.

Ambulatory ECG also is useful for assessing pharmacologic or ablative therapy in patients with atrial fibrillation or ventricular tachycardia. Monitoring for several weeks can help clinicians assess the burden of atrial fibrillation when using a rhythm-control strategy; assessing the ventricular rate in real-world situations is useful to determine the success of a rate-control strategy. Shortly after ablation of either atrial fibrillation or ventricular tachycardia, ECG home monitoring for 24 to 48 hours can detect asymptomatic recurrence and treatment failure.

Some antiarrhythmic drugs can prolong the QT interval. Ambulatory ECG devices that feature real-time monitoring can be used during drug initiation, enabling the clinician to monitor the QT interval without admitting the patient to the hospital.

Ultimately, ambulatory ECG monitoring is most commonly used to evaluate symptoms. Because arrhythmias and specific symptoms are unpredictable and transient, extended monitoring in a real-world setting allows for a more comprehensive evaluation than a standard 10-second ECG recording.

Continuous external monitoring: The Holter monitor

Recording is typically done continuously for 24 to 48 hours, although some newer devices can record for longer. Patients can press a button to note when they are experiencing symptoms, allowing for potential correlation with ECG abnormalities. The data are stored on a flash drive that can be uploaded for analysis after recording is complete.

What is its best use? Given its relatively short duration of monitoring, the Holter device is typically used to evaluate symptoms that occur daily or nearly daily. An advantage of the Holter monitor is its ability to record continuously, without requiring the patient to interact with the device. This feature provides “full disclosure,” which is the ability to see arrhythmia data from the entire recording period.

These features make Holter monitoring useful to identify suspected frequently occurring silent arrhythmias or to assess the overall arrhythmia burden. A typical Holter report can contain information on the heart rate (maximum, minimum, and average), ectopic beats, and tachy- and bradyarrhythmias, as well as representative samples.

The Holter device is familiar to most practitioners and remains an effective choice for ambulatory ECG monitoring. However, its use has largely been replaced by newer devices that overcome the Holter’s drawbacks, particularly its short duration of monitoring and the need for postmonitoring analysis. Additionally, although newer Holter devices are more ergonomic, some patients find the wires and gel electrodes uncomfortable or inconvenient.

Intermittent monitoring: Event recorders

Unlike the continuous monitors, intermittent recording devices (also called event recorders), capture and store tracings only during an event.

Intermittent recording monitors are of 2 general types: post-event recorders and loop recorders. These devices can extend the overall duration of observation, which can be especially useful for those whose symptoms and arrhythmias are infrequent.

Post-event recorders are small and self-contained, not requiring electrodes (Figure 1). The device is carried by the patient but not worn continuously. When the patient experiences symptoms, he or she places the device against the chest and presses a button to begin recording. These tracings are stored on the device and can be transmitted by telephone to a data center for analysis. Although post-event recorders allow for monitoring periods typically up to 30 days, they are limited by requiring the patient to act to record an event.

What is its best use? These devices are best used in patients who have infrequent symptoms and are at low risk. Transient or debilitating symptoms, including syncope, can limit the possibility of capturing an event.

Intermittent monitoring: Loop recorders

Loop recorders monitor continuously but record only intermittently. The name refers to the device’s looping memory: ie, to extend how long it can be used and make the most of its limited storage, the device records over previously captured data, saving only the most important data. The device saves the data whenever it detects an abnormal rhythm or the patient experiences symptoms and pushes a button. Data are recorded for a specified time before and after the activation, typically 30 seconds.

Loop recorders come in 2 types: external and implantable.

External loop recorders

External loop recorders look like Holter monitors (Figure 1), but they have the advantage of a much longer observation period—typically up to 1 month. The newest devices have even greater storage capacity and can provide “backward” memory, saving data that were captured just before the patient pushed the button.

In studies of patients with palpitations, presyncope, or syncope, external loop recorders had greater diagnostic yield than traditional 24-hour Holter monitors.^7,8 This finding was supported by a clinical trial that found 30-day monitoring with an external loop recorder led to a 5-fold increase in detecting atrial fibrillation in patients with cryptogenic stroke.⁹

Disadvantages of external loop recorders are limited memory storage, a considerable reliance on patient activation of the device, and wires and electrodes that need to be worn continuously.

What is their best use? External loop recorders are most effective when used to detect an arrhythmia or to correlate infrequent symptoms with an arrhythmia. They are most appropriately used in patients whose symptoms occur more often than every 4 weeks. They are less useful in assessing very infrequent symptoms, overall arrhythmia burden, or responsiveness to therapy.¹⁰

Implantable loop recorders are small devices that contain a pair of sensing electrodes housed within an outer shell (Figure 1). They are implanted subcutaneously, usually in the left parasternal region, using local anesthesia. The subcutaneous location eliminates many of the drawbacks of the skin-electrode interface of external loop recorders.

Similar to the external loop recorder, this device monitors continuously and can be activated to record either by the patient by pressing a button on a separate device, or automatically when an arrhythmia is detected using a preprogrammed algorithm.

In contrast to external devices, many internal loop recorders have a battery life and monitoring capability of up to 3 years. This extended monitoring period has been shown to increase the likelihood of diagnosing syncope or infrequent palpitations.^11,12 Given that paroxysmal atrial fibrillation can be sporadic and reveal itself months after a stroke, internal loop recorders may also have a role in evaluating cryptogenic stroke.^13,14

The most important drawbacks of internal loop recorders are the surgical procedure for insertion, their limited memory storage, and high upfront cost.¹⁵ Furthermore, even though they allow for extended monitoring, there may be diminishing returns for prolonged observation.

What is their best use? For patients with palpitations, intermittent event monitoring has been shown to be cost-effective for the first 2 weeks, but after 3 weeks, the cost per diagnosis increases dramatically.¹⁶ As a result, internal loop recorders are reserved primarily for scenarios in which prolonged external monitoring has not revealed a source of arrhythmia despite a high degree of suspicion.

Mobile cardiac telemetry

Mobile cardiac telemetry builds on other ECG monitoring systems by adding real-time communication and technician evaluation.

Physically, these devices resemble either hand-held event records, with a single-channel sensing unit embedded in the case, or a traditional Holter monitor, with 3 channels, wires, and electrodes  (Figure 1).

The sensor wirelessly communicates with a nearby portable monitor, which continuously observes and analyzes the patient’s heart rhythm. When an abnormal rhythm is detected or when the patient marks the presence of symptoms, data are recorded and sent in real time via a cellular network to a monitoring center; the newest monitors can send data via any Wi-Fi system. The rhythm is then either evaluated by a trained technician or relayed to a physician. If necessary, the patient can be contacted immediately.

Mobile cardiac telemetry is typically used for up to 30 days, which  allows for evaluation of less-frequent symptoms. As a result, it may have a higher diagnostic yield for palpitations, syncope, and presyncope than the 24-hour Holter monitor.¹⁷

Further, perhaps because mobile cardiac telemetry relies less on stored information and requires less patient-device interaction than external loop recorders, it is more effective at symptom evaluation.¹⁸

Mobile cardiac telemetry also has a diagnostic role in evaluating patients with cryptogenic stroke. This is based on studies showing it has a high rate of atrial fibrillation detection in this patient population and is more effective at determining overall atrial fibrillation burden than loop recorders.^18,19

What is its best use? The key advantage of mobile cardiac telemetry is its ability to make rhythm assessments and communicate with technicians in real time. This allows high-risk patients to be immediately alerted to a life-threatening arrhythmia. It also gives providers an opportunity to initiate anticoagulation or titrate antiarrhythmic therapy in the outpatient setting without a delay in obtaining information. This intensive monitoring, however, requires significant manpower, which translates to higher cost, averaging 3 times that of other standard external monitors.¹⁵

Patch monitors

These ultraportable devices are a relatively unobtrusive and easy-to-use alternative for short-term ambulatory ECG monitoring. They monitor continuously with full disclosure, outpatient telemetry, and post-event recording features.

Patch monitors are small, leadless, wireless, and water-resistant (Figure 1). They are affixed to the left pectoral region with a waterproof adhesive and can be worn for 14 to 28 days. Recording is usually done continuously; however, these devices have an event marker button that can be pressed when the user experiences symptoms. They acquire a single channel of data, and each manufacturer has a proprietary algorithm for automated rhythm detection and analysis.²⁰

Several manufacturers produce ECG patch monitors. Two notable devices are the Zio patch (iRhythm Technologies, San Francisco, CA) and the Mobile Cardiac Outpatient Telemetry patch (BioTelemetry, Inc, Malvern, PA).

The Zio patch is a continuous external monitor with full disclosure. It is comparable to the Holter monitor, but has a longer recording period. After completing a 2-week monitoring period, the device is returned for comprehensive rhythm analysis. A typical Zio report contains information on atrial fibrillation burden, ectopic rhythm burden, symptom and rhythm correlation, heart rate trends, and relevant rhythm strips.

The Mobile Cardiac Outpatient Telemetry patch collects data continuously and communicates wirelessly by Bluetooth to send its ECG data to a monitoring center for evaluation.

A principal advantage of patch monitors—and a major selling point for manufacturers—is their low-profile, ergonomic, and patient-friendly design. Patients do not have to manage wires or batteries and are able to shower with their devices. Studies show that these features increase patient satisfaction and compliance, resulting in increased diagnostic yield.^21,22 Additionally, patch monitors have the advantage of a longer continuous monitoring period than traditional Holter devices (2 weeks vs 1 or 2 days), affording an opportunity to capture events that occur less frequently.

Validation studies have reinforced their efficacy and utility in clinical scenarios.^22,23 In large part because of the extended monitoring period, patch monitors have been shown to have greater diagnostic yield than the 24-hour Holter monitor in symptomatic patients undergoing workup for suspected arrhythmia.

The role of patch monitors in evaluating atrial fibrillation is also being established. For patients with cryptogenic stroke, patch monitors have shown better atrial fibrillation detection than the 24-hour Holter monitor.²⁴ Compared with traditional loop monitors, patch monitors have the added advantage of assessing total atrial fibrillation burden. Further, although screening for atrial fibrillation with a traditional 12-lead ECG monitor has not been shown to be effective, clinical studies have found that the patch monitor may be a useful screening tool for high-risk patients.^25,26

Nevertheless, patch monitors have drawbacks. They are not capable of long-term monitoring, owing to battery and adhesive limitations.²⁰ More important, they have  been able to offer only single-channel acquisition, which makes it more difficult to detect an arrhythmia that is characterized by a change in QRS axis or change in QRS width, or to distinguish an arrhythmia from an artifact. This appears to be changing, however, as several manufacturers have recently developed multilead ECG patch monitors or attachments and are attempting to merge this technology with fully capable remote telemetry.

Recent improvements in battery life, memory, detection algorithms, wireless transmission, cellular communication, and adhesives have enabled multiple features to be combined into a single device. Patch monitors, for example, are small devices that now offer full-disclosure recording, extended monitoring, and telemetry transmitting. Automated arrhythmia recognition that triggers recording is central to all modern devices, regardless of type.

As a result of these trends, the traditional features used to differentiate devices may become less applicable. The classic Holter monitor may become obsolete as its advantages (full disclosure, continuous recording) are being incorporated into smaller devices that can record longer. Similarly, external monitors that have the capacity for full disclosure and continuous recording are no longer loop recorders in that they do not record into a circular memory.

It may be preferable to describe all non-Holter devices as event monitors or ambulatory monitors, with the main distinguishing features being the ability to transmit data (telemetry), full disclosure vs patient- or arrhythmia-activated recording, and single-channel or multichannel recording (single-lead or 3-lead ECG).

The following are the main distinguishing features that should influence the choice of device for a given clinical context.

Real-time data evaluation provided by mobile telemetry makes this feature ideal to monitor patients with suspected high-risk arrhythmias and their response to antiarrhythmic therapy.

Full-disclosure recording is necessary to assess the overall burden of an arrhythmia, which is frequently important in making treatment decisions, risk-stratifying, and assessing response to therapy. In contrast, patient- or arrhythmia-activated devices are best used when the goal is simply to establish the presence of an arrhythmia.

Multichannel recording may be better than single-channel recording, as it is needed to determine the anatomic origin of an arrhythmia, as might be the case in risk-stratification in a patient with a ventricular tachycardia.

Long duration. The clinician must have a reasonable estimate of how often the symptoms or arrhythmia occur to determine which device will offer a monitoring duration sufficient to detect an arrhythmia.

NEWER TECHNOLOGIES

The newest ambulatory ECG devices build on the foundational concepts of the older ones. However, with miniaturized electronic circuits, Bluetooth, Wi-Fi, and smartphones, these new devices can capture ECG tracings and diagnose offending arrhythmias on more consumer-friendly devices.

Smartphones and smartwatches have become increasingly powerful. Some have the ability to capture, display, and record the cardiac waveform. One manufacturer to capitalize on these technologies, AliveCor (Mountain View, CA), has developed 2 products capable of generating a single-lead ECG recording using either a smartphone (KardiaMobile) or an Apple watch (KardiaBand).

KardiaMobile has a 2-electrode band that can be carried in a pocket or attached to the back of a smartphone (Figure 1). The user places 1 or 2 fingers from each hand on the electrodes, and the device sends an ultrasound signal that is picked up by the smartphone’s microphone. The signal is digitized to produce a 30-second ECG tracing on the phone’s screen. A proprietary algorithm analyzes the rhythm and generates a description of “normal” or “possible atrial fibrillation.” The ECG is then uploaded to a cloud-based storage system for later access or transmission. KardiaMobile is compatible with both iOS and Android devices.

The KardiaBand is a specialized Apple watch band that has an electrode embedded in it. The user places a thumb on the electrode for 30 seconds, and an ECG tracing is displayed on the watch screen.

The Kardia devices were developed (and advertised) predominantly to assess atrial fibrillation. Studies have validated the accuracy of their algorithm. One study showed that, compared with physician-interpreted ECGs, the algorithm had a 96.6% sensitivity and 94.1% specificity for detecting atrial fibrillation.²⁷ They have been found useful for detecting and evaluating atrial fibrillation in several clinical scenarios, including discharge monitoring in patients after ablation or cardiac surgery.^28,29 In a longer study of patients at risk of stroke, twice-weekly ECG screening using a Kardia device for 1 year was more likely to detect incident atrial fibrillation than routine care alone.³⁰

Also, the Kardia devices can effectively function as post-event recorders when activated by patients when they experience symptoms. In a small study of outpatients with palpitations and a prior nondiagnostic workup, the KardiaMobile device was found to be noninferior to external loop recorders for detecting arrhythmias.³¹ Additional studies are assessing Kardia’s utility in other scenarios, including the evaluation of ST-segment elevation myocardial infarction^32,33 and QT interval for patients receiving antiarrhythmic therapy.³⁴

Cardiio Inc. (Cambridge, MA) has developed technology to screen for atrial fibrillation using an app that requires no additional external hardware. Instead, the app uses a smartphone’s camera and flashlight to perform photo­plethysmography to detect pulsatile changes in blood volume and generate a waveform. Based on waveform variability, a proprietary algorithm attempts to determine whether the user is in atrial fibrillation. It does not produce an ECG tracing. Initial studies suggest it has good diagnostic accuracy and potential utility as a population-based screening tool,^35,36 but it has not been fully validated.

Recently, Apple entered the arena of ambulatory cardiac monitoring with the release of its fourth-generation watch (Apple Watch Series 4 model). This watch has built-in electrodes that can generate a single-lead ECG on the watch screen. Its algorithm can discriminate between atrial fibrillation and sinus rhythm, but it has not been assessed for its ability to evaluate other arrhythmias. Even though it has been “cleared” by the US Food and Drug Administration, it is approved only for informational use, not to make a medical diagnosis.

Integration of ambulatory ECG technology with smartphone and watch technology is an exciting new wearable option for arrhythmia detection. The patient-centered and controlled nature of these devices have the potential to help patients with palpitations or other symptoms determine if their cardiac rhythms are normal.

This technology, however, is still in its infancy and has many limitations. For example, even though these devices can function as post-event recorders, they depend on user-device interactions. Plus, they cannot yet perform continuous arrhythmia monitoring like modern loop recorders.

Additionally, automated analysis has largely been limited to distinguishing atrial fibrillation from normal sinus rhythm. It is uncertain how effective the devices may be in evaluating other arrhythmias. Single-lead ECG recordings, as discussed, have limited interpretability and value. And even though studies have shown utility in certain clinical scenarios, large-scale validation studies are lacking. This technology will likely continue to be developed and its clinical value improved; however, its clinical use requires careful consideration and collaborative physician-patient decision-making.

The development of smartphone and watch ECG technology has led to a rise in direct-to-consumer healthcare delivery. By devising technology that is appealing, useful, and affordable, companies can bypass the insurer and practitioner by targeting increasingly health-literate consumers. For many companies, there is great motivation to enter this healthcare space. Wearable devices are immensely popular and, as a result, generate substantial revenue. One analysis estimates that 1 in 10 Americans (nearly 30 million) owns a wearable, smart-technology device.³⁷

This direct-to-consumer approach has specific implications for cardiology and, more broadly, for healthcare overall. By directly selling to consumers, companies have an opportunity to reach many more people. The Apple Watch Series 4 has taken this a step further: by including this technology in the watch, consumers not necessarily seeking an ambulatory cardiac monitor will have one with a watch purchase. This could lead to increases in monitoring and could alert people to previously undiagnosed disorders.

For consumers, this technology can empower them to choose how and when to be monitored. Further, it gives them personal control of their healthcare data, and helps move the point of care out of hospitals and clinics and into the home.

But wearable medical technology and direct-to-consumer healthcare have risks. First, in the absence of appropriate regulation, patients have to distinguish between products that are well validated and those that are unproven. Consumers also may inappropriately use devices for indications or in scenarios for which the value is uncertain.

Also, there is potential for confusion and misunderstanding of results, including false-positive readings, which could lead to excessive and costly use of unnecessary diagnostic workups. Instead of providing peace of mind, these devices could cause greater worry. This may be especially true with the newest Apple watch, as this product will introduce ambulatory ECG to a younger and healthier segment of the population who are less likely to have true disease.

Further, these devices have algorithms that detect atrial fibrillation, but is it the same as that detected by traditional methods? Sometimes termed “subclinical” atrial fibrillation, it poses uncertainties: ie, Do patients need anticoagulation, pharmacologic therapy, and ablation? The optimal management of subclinical atrial fibrillation, as well as its similarities to and differences from atrial fibrillation diagnosed by traditional methods, are topics that need further study.

Wearable technology is still developing and will continue to do so. Medical practice will have to adapt to it.

FUTURE DIRECTIONS

Changes in technology have led to remarkable advances in the convenience and accuracy of ambulatory ECG monitoring. Ongoing research is expected to lead to even more improvements. Devices will become more ergonomic and technically capable, and they may expand monitoring to include other biologic parameters beyond ECG.

Comfort is important to ensure patient adherence. Newer, flexible electronics embedded in ultrathin materials can potentially improve the wearability of devices that require gel electrodes or adhesive patches.³⁸ Wireless technology may obviate the need for on-skin attachments. Future recording systems may be embedded into clothing or incorporated into wearable vests capable of wirelessly transmitting ECG signals to separate recording stations.³⁹

In addition to becoming smaller and more comfortable, future devices will be more technically capable, leading to a merging of technologies that will further blur the distinctions among devices. Eventually, the features of full disclosure, extended monitoring duration, and telemetric communication will all be present together. Perhaps more important is that ambulatory ECG devices may become fully capable biosensor monitors. These devices would have the potential to monitor respiratory frequency, peripheral oxygen saturation, potassium levels, and arterial pulse pressure.^39,40

Insights from basic and clinical research are changing the way we treat diabetes mellitus. In 2016, several key diabetes organizations, ie, the American Diabetes Association (ADA), the Juvenile Diabetes Research Foundation (JDRF), the European Association for the Study of Diabetes (EASD), and the American Association of Clinical Endocrinologists (AACE), called for bringing therapeutic approaches in line with our updated understanding of disease pathophysiology, replacing “one-size-fits-all” management with a tailored approach.¹ This message has since been reiterated.²

Here, we review advances in our understanding of diabetes and how these inform a new model of diabetes treatment.

BETA CELLS ARE KEY

High levels of glucose and lipids damage and eventually kill beta cells through mechanisms including that of oxidative stress, so that glucose control deteriorates over time. The same processes are active in the target-organ damage seen in diabetes.^3,4 These 2 insights—that the disease arises from combinatorial, nondiscrete pressures and that it proceeds through common processes of cell damage—leads us to a more unified understanding of the mechanism of diabetes, and may eventually replace current classifications of type 1, type 2, or latent autoimmune diabetes in adults, as well as nomenclature such as “microvascular” and “macrovascular” disease.³

FIRST-LINE LIFESTYLE INTERVENTIONS

Lifestyle interventions are the first-line therapy for elevated blood glucose. Achieving and maintaining a healthy body mass index is essential to help correct insulin resistance and minimize beta-cell dysfunction.

Lifestyle modifications for overweight or obese patients with diabetes mellitus include optimal caloric intake, decreased intake of simple carbohydrates, increased physical activity, and a 3% to 5% reduction in body weight.⁵ Weight-loss drugs may be indicated in obese patients. Normalization of lipids and hypertension should be an early goal.

RIGHT MEDICATIONS, RIGHT PATIENTS

While all of the drugs approved for treating diabetes lower glucose levels, some are more beneficial than others, possessing actions beyond their effect on plasma glucose levels, both good and bad.

The AACE guideline for use of various antidiabetic medications⁶ grades factors such as risks of hypoglycemia, ketoacidosis, weight gain, cardiovascular events, and renal, gastrointestinal, and bone concerns. This represents a much-needed first step toward guidance on selecting the right medications for the right patients. Risk factors (such as heart failure) and comorbidities (such as nonalcoholic fatty liver disease and nonalcoholic steatohepatitis) are among the considerations for choosing treatment.

Two principles

We propose 2 principles when choosing treatment:

Use “gentle” agents, ie, those that are least likely to exhaust beta cells or damage the organs involved in diabetes-related complications. Since the disease course depends on the health of the beta cells, give preference to agents that appear to best support beta cells—ie, agents that create the least oxidative stress or wear-and-tear—as will be outlined in this article.

Diabetes is associated with risks of cardiovascular disease, cardiac events, heart failure, and accelerated renal decompensation. Thus, it is equally important to prevent damage to the cardiovascular system, kidneys, and other tissues subject to damage through glucolipotoxicity.

Balancing glycemic control and risk

The hemoglobin A_1c level is the chief target of care and an important barometer of risk of diabetes-related complications. In 2018, the American College of Physicians (ACP) relaxed its target for hemoglobin A_1c from 7% to 8%.⁸ This move was apparently to give physicians greater “wiggle room” for achieving goals in hypoglycemia-prone patients. This, however, may take a toll.

Hypoglycemia is closely tied to cardiovascular disease. Even mild and asymptomatic hypoglycemia that goes undiagnosed and unnoticed by patients has been found to be associated with higher rates of all-cause mortality, prolonged QT interval, angina, arrhythmias, myocardial dysfunction, disturbances in autonomic balance, and sudden death.^9–11

However, the ADA, AACE, American Association of Diabetes Educators (AADE), and the Endocrine Society jointly issued a strong indictment of the ACP recommendation.¹² They argue that tight glucose control and its well-documented “legacy effects” on long-term outcomes should not be sacrificed.^12,13 Indeed, there is no need to abandon evidence-based best practices in care when at least 8 of the 11 classes of antidiabetes agents do not introduce the same level of risk for hypoglycemia.

Current guidelines argue for tight glucose control but generally stop short of discriminating or stratifying the mechanisms of action of the individual classes of drugs. These guidelines also do not stress targeting the particular pathways of hyperglycemia present in any given patient. However, the 2016 ADA joint statement acknowledges the need to “characterize the many paths to beta-cell dysfunction or demise and identify therapeutic approaches that best target each path.”¹

The sections below highlight some of the recent data on the profiles of most of the currently available agents.

Metformin: Still the first-line treatment

Current guidelines from the ACP, ADA, and AACE keep metformin¹⁴ as the backbone of treatment, although debate continues as to whether newer agents such as GLP-1 receptor agonists are superior for first-line therapy.

Pathways affected. Metformin improves insulin resistance in the liver, increases endogenous GLP-1 levels via the gut, and appears to modulate gut flora composition, which is increasingly suspected to contribute to dysmetabolism. 

Advantages, benefits. Metformin is easy to use and does not cause hypoglycemia. It was found to modestly reduce the number of cardiovascular events and deaths in a number of clinical outcome studies.^15–19

Disadvantages, adverse effects. In some patients, tolerability restricts the use of this drug at higher doses. The most common adverse effects of metformin are gastrointestinal symptoms (diarrhea, nausea, vomiting, flatulence); other risks include lactic acidosis in patients with impaired kidney function, heart failure, hypoxemia, alcoholism, cirrhosis, contrast exposure, sepsis, and shock.

GLP-1 receptor agonists

GLP-1 receptor agonists^20–25 are injectable medications approved for adults with type 2 diabetes. Exenatide and liraglutide lower hemoglobin A_1c by 1 to 1.5 absolute percentage points and reduce body weight; these effects persist over the long term.²⁶ Newer once-weekly GLP-1 receptor agonists (albiglutide,²⁰ dulaglutide,²¹ and semaglutide²⁵) have similar benefits. In 2019, new drug applications were submitted to the FDA for the first-in-kind oral GLP-1 receptor agonists, which would improve convenience and adherence and make this class even more attractive.

Pathways affected. GLP-1 receptor agonists address multiple pathways of hyperglycemia. They increase insulin production and release, promote weight loss, and reduce insulin resistance, glucagon secretion, and inflammation. They also increase amylin, help overcome GLP-1 resistance, slow gastric emptying, and favorably modify gut flora.²⁷

Advantages, benefits. The cardioprotective actions of GLP-1 receptor agonists include reducing inflammation and dysfunction in endothelial and myocardial cells; slowing atherosclerosis; reducing oxidative stress-induced injury and scavenging of reactive oxygen species in coronary endothelial, smooth muscle, and other cells; and enhancing endogenous antioxidant defenses.²⁷ GLP-1 receptor agonism has also been found to inhibit apoptosis in cardiomyocytes, as well as in beta cells.

Several large-scale studies have shown improved outcomes with GLP-1 receptor agonists. The Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial²⁶ found that liraglutide reduced major adverse cardiovascular events by 13% and myocardial infarctions by 22% in more than 9,000 adults with type 2 diabetes who were at high risk of major adverse cardiovascular events compared with placebo. Rates of microvascular outcomes were also reduced.

A retrospective database analysis of 39,275 patients with type 2 diabetes who were treated with exenatide reported a lower incidence of cardiovascular events than in patients not treated with exenatide.²⁸

However, no effect on cardiovascular outcomes was found with a third GLP-1 agent, lixisenatide, in a large-scale trial in high-risk patients with diabetes.²⁹

The most recently evaluated GLP-1 receptor agonist is semaglutide. The Trial to Evaluate Cardiovascular and Other Long-term Outcomes With Semaglutide in Subjects With Type 2 Diabetes (SUSTAIN-6) demonstrated a reduced risk of major adverse cardiovascular events.³⁰

Disadvantages, adverse effects. The most common adverse effects in this class include nausea, hypoglycemia, diarrhea, constipation, vomiting, headache, decreased appetite, dyspepsia, fatigue, dizziness, abdominal pain, and increased lipase. The nausea can be mitigated by advising patients to stop eating at first sensation of stomach fullness.

DPP-4 inhibitors

Dipeptidyl peptidase 4 (DPP-4) is a ubiquitous enzyme that rapidly degrades GLP-1 and other endogenous peptides.³¹ Saxagliptin,³² sitagliptin,³³ linagliptin,³⁴ and alogliptin³⁵ are approved for use in the United States, and vildagliptin36 is available in Europe.

Pathways affected. These agents modify 3 pathways of hyperglycemia: they increase insulin secretion, decrease glucagon levels, and help overcome GLP-1 resistance.

Advantages, benefits. DPP-4 inhibitors have been used safely and effectively in clinically challenging populations of patients with long-standing type 2 diabetes (> 10 years).

Disadvantages, adverse effects. As this class increases GLP-1 levels only 2- to 4-fold, their efficacy is more modest than that of GLP-1 receptor agonists (hemoglobin A_1c reductions of 0.5% to 1%; neutral effects on weight).³⁷

Outcome trials have largely been neutral. Saxagliptin has been associated with an increase in admissions for heart failure. There have been a very small but statistically significant number of drug-related cases of acute pancreatitis.³⁸

The most common adverse effects with this class include headache, nasopharyngitis, urinary tract infection, upper respiratory tract infection, and elevated liver enzymes.

Drugs of this class currently available in the United States are canagliflozin,³⁹ dapagliflozin,⁴⁰ empagliflozin,⁴¹ and ertugliflozin.⁴²

Pathways affected. SGLT2 inhibitors lower the glucose reabsorption threshold in the kidney so that more glucose is excreted in the urine; they also decrease insulin resistance in muscle, liver, and fat cells (via weight loss) and possibly preserve beta-cell function by reducing glucotoxicity. A nonrenal mechanism—delayed gut absorption reducing postprandial glucose excursion—has been proposed to contribute to the glucose-lowering effects of canagliflozin.⁴³

Advantages, benefits. These agents reduce hemoglobin A_1c by about 0.5% to 1.0% from a baseline of about 8%. Because their action is independent of insulin, they can be used at any stage of type 2 diabetes, even after insulin secretion has significantly waned. Additional potential advantages include weight loss (up to 3.5% of body mass index) and lowering of systolic blood pressure (2–4 mm Hg) and diastolic blood pressure (1–2 mm Hg).^39–42

Canagliflozin was shown in the Canagliflozin Cardiovascular Assessment Study (CANVAS)⁴⁴ to significantly reduce the overall risk of cardiovascular disease by 14% and risk of heart failure hospitalization by 33% while significantly slowing the progression of renal disease.

In the BI 10773 (Empagliflozin) Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME),⁴⁵ empagliflozin reduced heart failure hospitalizations by 35%, cardiovascular deaths by 38%, and all-cause mortality by about 32%. These benefits are thought to be due to less arterial stiffness, lower sympathetic tone, and decreased arrhythmias. Notably, these dramatic benefits accrued in only about 3 years with use of add-on therapy, even though the reduction in hemoglobin A_1c was modest (0.6%), suggesting that pleiotropic effects are at work.

Disadvantages, adverse effects. The most common adverse effects of this class include urinary tract infections, yeast infections, dehydration, and hypovolemic symptoms; these can often be prevented. A trend toward increased incidence of amputations in earlier studies was not borne out in a 2018 meta-analysis of 4 observational databases.⁴⁶

Thiazolidinediones

There are currently 2 approved thiazolidine­diones in the United States, pioglitazone⁴⁷ and rosiglitazone.⁴⁸ Only pioglitazone is in common use, as rosiglitazone is associated with safety issues.⁴⁹

Pathways affected. Pioglitazone reduces insulin resistance in muscle, liver, and adipose tissue.

Advantages, benefits. Decreased levels of low-density lipoprotein cholesterol and triglycerides and increased high-density lipoprotein cholesterol levels⁴⁹ could plausibly account for the cardiovascular benefits reported in the Prospective Pioglitazone Clinical Trial in Macrovascular Events.⁵⁰ Pioglitazone has also been found to improve insulin secretion, endothelial function, and diastolic dysfunction; reduce inflammation; decrease plasminogen activator inhibitor 1; reverse lipotoxicity; and help correct nonalcoholic fatty liver disease and steatohepatitis.

Pioglitazone has also been found to reduce plaque in carotid and coronary arteries⁵¹; improve outcomes in patients with heart failure and myocardial infarction compared with insulin-sensitizing drugs⁵²; and reduce stroke and myocardial infarction in patients with insulin resistance (but not diabetes) and a recent history of ischemic stroke or transient ischemic attack (in the Insulin Resistance Intervention After Stroke trial).⁵³ It may also help maintain beta-cell function; the Actos Now for the Prevention of Diabetes Study found that pioglitazone reduced the risk of conversion of impaired glucose tolerance to frank diabetes by 72%.⁵⁴

Disadvantages, adverse effects. The most common adverse effects seen with this class include weight gain and salt retention, swelling, edema,⁵⁵ and related cardiovascular consequences in certain patients. While this may be mitigatable with lifestyle changes or use in combination with a GLP-1 receptor agonist or SGLT2 inhibitor,⁵⁶ pioglitazone is contraindicated in patients with heart failure, hemodynamic instability, or hepatic dysfunction.

Concerns that pioglitazone might increase the risk of bladder cancer seem to have been put to rest when a study in nearly 200,000 patients found no statistically significant association,⁵⁷ but the warning remains in the US label.

Long-term use of this class of drugs has been associated with an increased risk of bone fractures,⁵⁸ which warrants a risk-benefit assessment in each patient.

Injected insulin: Less safe than thought

Recent research suggests that injected insulin has a less favorable safety profile than previously thought.^15–19,59 Studies of the long-term safety of insulin therapy have had inconsistent results but suggest that injected insulin is associated with poorer cardiovascular and renal outcomes (in some of the same studies that showed metformin or other agents to improve outcomes),^17–19 and the association was dose-dependent. Several studies attempted to cancel out the poorer outcomes by adjusting for hemoglobin A_1c levels, stage of disease,^{17–19,26,27} or severe hypoglycemic episodes.⁶⁰ However, it may be inappropriate to reduce the impact of these variables, as these may themselves be the mediators of any deleterious effects of exogenous insulin.

When exogenous insulin is introduced into the peripheral circulation it causes a state of persistent iatrogenic hyperinsulinemia, which leads to insulin resistance and also appears to compromise the cardiovascular system. In contrast, endogenous insulin is released into the portal system in tightly controlled amounts.^5,61 This suggests that the same insulin peptide may not be equivalently beneficial when introduced in an artificial manner.

Before starting insulin therapy, consider its side effects such as weight gain and hypoglycemia. Most (about 85%) episodes of hypoglycemia occur with basal-bolus insulin regimens.⁶² Moreover, iatrogenic hyperinsulinemia can damage the vascular system.^63,64

We recommend. Insulin therapy is used early in the course of the disease as a short-term intervention for glucolipotoxicity. However, this can be accomplished without attendant risks of hypoglycemia and weight gain by using agents such as SGLT2 inhibitors and incretins. When insulin therapy is necessary, using it as add-on therapy might be considered instead of drug-switching. We have found alternate pharmacologic approaches successful in avoiding or delaying bolus insulin therapy. And in some patients taking insulin, we have had success in progressively introducing a noninsulin agent and were ultimately able to eliminate insulin altogether.

Bromocriptine-QR (quick release)⁶⁵ is a short-acting dopamine agonist that mimics the morning dopamine surge in the suprachiasmatic nucleus—the biologic clock.

Pathways affected. Bromocriptine addresses part of the brain contribution to hyperglycemia, with resultant reductions in both peripheral insulin resistance and sympathetic tone. This reduces muscle, liver, and adipose insulin resistance. It is moderately effective in glucose-lowering, especially in patients with significant insulin resistance.⁶⁶

Advantages, benefits. A 1-year clinical trial reported that bromocriptine reduced cardiovascular adverse outcomes by 39%, and the composite end point of myocardial infarction, stroke, and cardiovascular death by 52% compared with placebo.⁶⁷

Disadvantages, adverse effects. The most common adverse effects are nausea, rhinitis, headache, asthenia, dizziness, constipation, and sinusitis.

Alpha-glucosidase inhibitors

Alpha-glucosidase inhibitors (acarbose,⁶⁸ miglitol⁶⁹) work by decreasing the rate of absorption of glucose from the gastrointestinal tract.

Advantages, benefits. These drugs decrease hemoglobin A_1c by 0.5% to 0.8%.70 They are weight-neutral and do not pose a risk of hypoglycemia. Clinical studies suggest that they may delay or prevent diabetes progression. They were also found to reduce cardiovascular events, acute myocardial infarction, and the onset of hypertension.⁶⁹

Disadvantages, adverse effects. Their use remains limited due to gastrointestinal adverse effects. They may be contraindicated in patients with inflammatory bowel disease, partial bowel obstruction, or severe renal or hepatic disease.

Pramlintide

Pramlintide⁷¹ is an injectable amylin analogue. It is used as monotherapy or in combination with a sulfonylurea, metformin, or insulin glargine.

Pathways affected. Pramlintide decreases appetite, reduces glucagon levels, and minimizes absorption of glucose in the gut.

Disadvantages, adverse effects. Common side effects include mild to moderate hypoglycemia and nausea. Nausea may help explain the ability of pramlintide to confer weight loss when used in combination with insulin.

Sulfonylureas and meglitinides

These classes are still widely used in the treatment of type 2 diabetes, although the AACE⁶ and ADA⁷² guidelines de-emphasize their use based on associated risks of hypoglycemia, weight gain, morbidity, mortality, and loss of effect over time.

Pathways affected. Sulfonylureas stimulate insulin secretion from beta cells.

Disadvantages, adverse effects. Sulfonylureas and glinides are associated with poorer outcomes than newer agents in clinical trials^{15–19,59,60} and may be generally less beta-cell friendly.⁷³ Their harmful effects are difficult to measure in vivo, but these drugs sometimes appear to be associated with more rapid beta-cell failure and progression to insulin dependence compared with newer ones. Several large-scale registry studies have found sulfonylureas and glinides to be associated with poorer outcomes (reviewed by Herman et al).⁷⁴

Adverse effects include asthenia, headache, dizziness, nausea, diarrhea, epigastric fullness, and heartburn. Although they are often selected based on their low cost, other factors may offset their cost-effectiveness, such as need for glucose monitoring and hospital charges due to sulfonylurea-induced hypoglycemia. Their utility is also limited by dependence on beta-cell function.

Colesevelam

Colesevelam⁷⁵ is a bile acid sequestrant and low-density lipoprotein cholesterol-reducing agent that has been approved for use in diabetes. The mode of action of colesevelam in this capacity is under investigation. Its effect on hemoglobin A_1c is modest. It is associated with gastrointestinal adverse effects, particularly constipation.

Ranolazine

Ranolazine⁷⁶ is an antianginal drug that also lowers glucose by increasing insulin release. It also possesses cardioprotective properties. In patients with diabetes and non-ST-segment elevation acute coronary syndromes, ranolazine reduced hemoglobin A_1c by 1.2% and appeared to be weight-neutral.⁷⁶ Ranolazine is under clinical development for use in diabetes. Adverse effects include dizziness, headache, constipation, and nausea.


Rational combinations of agents

The ideal strategy would use combinations of agents that mechanistically complement one another and address each path of hyperglycemia present in a patient. This approach should supplant the former approaches of adding-on agents only after treatment failure or sequentially trying first-, second-, and third-line treatments.

Examples of synergistic combinations include those that target fasting plasma glucose and postprandial glucose, reduce reliance on insulin with add-on therapies, or manage hyperglycemia in specific patient groups, such as renal-impaired patients.

Large-scale long-term clinical studies are needed to determine the safety, efficacy, and outcomes of various combinations and whether they confer additive benefits. Some studies have begun to explore possible combinations.

Combined metformin, pioglitazone, and exenatide was reported to delay progression of diabetes in early dysglycemia.^77,78 Notably, this combination addresses multiple mediating pathways of hyperglycemia (Table 1).

A GLP-1 receptor agonist with an SGLT2 inhibitor would be another intriguing combination, as the mechanisms of action of these 2 classes complement one another. In limited clinical trials—the DURATION-8 study (lasting 26 weeks),⁷⁹ the Canagliflozin Cardiovascular Assessment Study (18 weeks),⁸⁰ and a 24-week study in nondiabetic obese patients⁸¹—additive benefits were also seen in systolic blood pressure, body weight, and cardiac risk factors by adding an SGLT2 inhibitor to a GLP-1 receptor agonist, compared with either agent alone. In theory, these improvements might slow or reverse cardiorenal compromise. Lower doses of 1 or more may be possible, and the regimen could prove cost-effective and life-sparing should it slow the progression of the disease and the onset of its complications. A clinical study of this combination is under way (Ralph DeFronzo, personal communication, July 2018). Similarly, the combination of metformin, saxagliptin and dapagliflozin has been shown to be effective.⁸²

CONCLUSION

Care for diabetes mellitus can be particularly challenging for the primary care physician. The progressive nature of diabetes, with worsening hyperglycemia over the course of the disease, further complicates disease management.

Best practices for care nonetheless need to evolve with well-evidenced data, and without years of delay for “trickle-down” education from the specialties to primary care. We have arrived at a juncture to leverage therapies that address the 11 mediating pathways of hyperglycemia, optimally protect beta cells, minimize hypoglycemia, manage risk factors associated with diabetes, and improve diabetes-related outcomes.

Insights from basic and clinical research are changing the way we treat diabetes mellitus. In 2016, several key diabetes organizations, ie, the American Diabetes Association (ADA), the Juvenile Diabetes Research Foundation (JDRF), the European Association for the Study of Diabetes (EASD), and the American Association of Clinical Endocrinologists (AACE), called for bringing therapeutic approaches in line with our updated understanding of disease pathophysiology, replacing “one-size-fits-all” management with a tailored approach.¹ This message has since been reiterated.²

Here, we review advances in our understanding of diabetes and how these inform a new model of diabetes treatment.

BETA CELLS ARE KEY

High levels of glucose and lipids damage and eventually kill beta cells through mechanisms including that of oxidative stress, so that glucose control deteriorates over time. The same processes are active in the target-organ damage seen in diabetes.^3,4 These 2 insights—that the disease arises from combinatorial, nondiscrete pressures and that it proceeds through common processes of cell damage—leads us to a more unified understanding of the mechanism of diabetes, and may eventually replace current classifications of type 1, type 2, or latent autoimmune diabetes in adults, as well as nomenclature such as “microvascular” and “macrovascular” disease.³

FIRST-LINE LIFESTYLE INTERVENTIONS

Lifestyle interventions are the first-line therapy for elevated blood glucose. Achieving and maintaining a healthy body mass index is essential to help correct insulin resistance and minimize beta-cell dysfunction.

Lifestyle modifications for overweight or obese patients with diabetes mellitus include optimal caloric intake, decreased intake of simple carbohydrates, increased physical activity, and a 3% to 5% reduction in body weight.⁵ Weight-loss drugs may be indicated in obese patients. Normalization of lipids and hypertension should be an early goal.

RIGHT MEDICATIONS, RIGHT PATIENTS

While all of the drugs approved for treating diabetes lower glucose levels, some are more beneficial than others, possessing actions beyond their effect on plasma glucose levels, both good and bad.

The AACE guideline for use of various antidiabetic medications⁶ grades factors such as risks of hypoglycemia, ketoacidosis, weight gain, cardiovascular events, and renal, gastrointestinal, and bone concerns. This represents a much-needed first step toward guidance on selecting the right medications for the right patients. Risk factors (such as heart failure) and comorbidities (such as nonalcoholic fatty liver disease and nonalcoholic steatohepatitis) are among the considerations for choosing treatment.

Two principles

We propose 2 principles when choosing treatment:

Use “gentle” agents, ie, those that are least likely to exhaust beta cells or damage the organs involved in diabetes-related complications. Since the disease course depends on the health of the beta cells, give preference to agents that appear to best support beta cells—ie, agents that create the least oxidative stress or wear-and-tear—as will be outlined in this article.

Diabetes is associated with risks of cardiovascular disease, cardiac events, heart failure, and accelerated renal decompensation. Thus, it is equally important to prevent damage to the cardiovascular system, kidneys, and other tissues subject to damage through glucolipotoxicity.

Balancing glycemic control and risk

The hemoglobin A_1c level is the chief target of care and an important barometer of risk of diabetes-related complications. In 2018, the American College of Physicians (ACP) relaxed its target for hemoglobin A_1c from 7% to 8%.⁸ This move was apparently to give physicians greater “wiggle room” for achieving goals in hypoglycemia-prone patients. This, however, may take a toll.

Hypoglycemia is closely tied to cardiovascular disease. Even mild and asymptomatic hypoglycemia that goes undiagnosed and unnoticed by patients has been found to be associated with higher rates of all-cause mortality, prolonged QT interval, angina, arrhythmias, myocardial dysfunction, disturbances in autonomic balance, and sudden death.^9–11

However, the ADA, AACE, American Association of Diabetes Educators (AADE), and the Endocrine Society jointly issued a strong indictment of the ACP recommendation.¹² They argue that tight glucose control and its well-documented “legacy effects” on long-term outcomes should not be sacrificed.^12,13 Indeed, there is no need to abandon evidence-based best practices in care when at least 8 of the 11 classes of antidiabetes agents do not introduce the same level of risk for hypoglycemia.

Current guidelines argue for tight glucose control but generally stop short of discriminating or stratifying the mechanisms of action of the individual classes of drugs. These guidelines also do not stress targeting the particular pathways of hyperglycemia present in any given patient. However, the 2016 ADA joint statement acknowledges the need to “characterize the many paths to beta-cell dysfunction or demise and identify therapeutic approaches that best target each path.”¹

The sections below highlight some of the recent data on the profiles of most of the currently available agents.

Metformin: Still the first-line treatment

Current guidelines from the ACP, ADA, and AACE keep metformin¹⁴ as the backbone of treatment, although debate continues as to whether newer agents such as GLP-1 receptor agonists are superior for first-line therapy.

Pathways affected. Metformin improves insulin resistance in the liver, increases endogenous GLP-1 levels via the gut, and appears to modulate gut flora composition, which is increasingly suspected to contribute to dysmetabolism. 

Advantages, benefits. Metformin is easy to use and does not cause hypoglycemia. It was found to modestly reduce the number of cardiovascular events and deaths in a number of clinical outcome studies.^15–19

Disadvantages, adverse effects. In some patients, tolerability restricts the use of this drug at higher doses. The most common adverse effects of metformin are gastrointestinal symptoms (diarrhea, nausea, vomiting, flatulence); other risks include lactic acidosis in patients with impaired kidney function, heart failure, hypoxemia, alcoholism, cirrhosis, contrast exposure, sepsis, and shock.

GLP-1 receptor agonists

GLP-1 receptor agonists^20–25 are injectable medications approved for adults with type 2 diabetes. Exenatide and liraglutide lower hemoglobin A_1c by 1 to 1.5 absolute percentage points and reduce body weight; these effects persist over the long term.²⁶ Newer once-weekly GLP-1 receptor agonists (albiglutide,²⁰ dulaglutide,²¹ and semaglutide²⁵) have similar benefits. In 2019, new drug applications were submitted to the FDA for the first-in-kind oral GLP-1 receptor agonists, which would improve convenience and adherence and make this class even more attractive.

Pathways affected. GLP-1 receptor agonists address multiple pathways of hyperglycemia. They increase insulin production and release, promote weight loss, and reduce insulin resistance, glucagon secretion, and inflammation. They also increase amylin, help overcome GLP-1 resistance, slow gastric emptying, and favorably modify gut flora.²⁷

Advantages, benefits. The cardioprotective actions of GLP-1 receptor agonists include reducing inflammation and dysfunction in endothelial and myocardial cells; slowing atherosclerosis; reducing oxidative stress-induced injury and scavenging of reactive oxygen species in coronary endothelial, smooth muscle, and other cells; and enhancing endogenous antioxidant defenses.²⁷ GLP-1 receptor agonism has also been found to inhibit apoptosis in cardiomyocytes, as well as in beta cells.

Several large-scale studies have shown improved outcomes with GLP-1 receptor agonists. The Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial²⁶ found that liraglutide reduced major adverse cardiovascular events by 13% and myocardial infarctions by 22% in more than 9,000 adults with type 2 diabetes who were at high risk of major adverse cardiovascular events compared with placebo. Rates of microvascular outcomes were also reduced.

A retrospective database analysis of 39,275 patients with type 2 diabetes who were treated with exenatide reported a lower incidence of cardiovascular events than in patients not treated with exenatide.²⁸

However, no effect on cardiovascular outcomes was found with a third GLP-1 agent, lixisenatide, in a large-scale trial in high-risk patients with diabetes.²⁹

The most recently evaluated GLP-1 receptor agonist is semaglutide. The Trial to Evaluate Cardiovascular and Other Long-term Outcomes With Semaglutide in Subjects With Type 2 Diabetes (SUSTAIN-6) demonstrated a reduced risk of major adverse cardiovascular events.³⁰

Disadvantages, adverse effects. The most common adverse effects in this class include nausea, hypoglycemia, diarrhea, constipation, vomiting, headache, decreased appetite, dyspepsia, fatigue, dizziness, abdominal pain, and increased lipase. The nausea can be mitigated by advising patients to stop eating at first sensation of stomach fullness.

DPP-4 inhibitors

Dipeptidyl peptidase 4 (DPP-4) is a ubiquitous enzyme that rapidly degrades GLP-1 and other endogenous peptides.³¹ Saxagliptin,³² sitagliptin,³³ linagliptin,³⁴ and alogliptin³⁵ are approved for use in the United States, and vildagliptin36 is available in Europe.

Pathways affected. These agents modify 3 pathways of hyperglycemia: they increase insulin secretion, decrease glucagon levels, and help overcome GLP-1 resistance.

Advantages, benefits. DPP-4 inhibitors have been used safely and effectively in clinically challenging populations of patients with long-standing type 2 diabetes (> 10 years).

Disadvantages, adverse effects. As this class increases GLP-1 levels only 2- to 4-fold, their efficacy is more modest than that of GLP-1 receptor agonists (hemoglobin A_1c reductions of 0.5% to 1%; neutral effects on weight).³⁷

Outcome trials have largely been neutral. Saxagliptin has been associated with an increase in admissions for heart failure. There have been a very small but statistically significant number of drug-related cases of acute pancreatitis.³⁸

The most common adverse effects with this class include headache, nasopharyngitis, urinary tract infection, upper respiratory tract infection, and elevated liver enzymes.

Drugs of this class currently available in the United States are canagliflozin,³⁹ dapagliflozin,⁴⁰ empagliflozin,⁴¹ and ertugliflozin.⁴²

Pathways affected. SGLT2 inhibitors lower the glucose reabsorption threshold in the kidney so that more glucose is excreted in the urine; they also decrease insulin resistance in muscle, liver, and fat cells (via weight loss) and possibly preserve beta-cell function by reducing glucotoxicity. A nonrenal mechanism—delayed gut absorption reducing postprandial glucose excursion—has been proposed to contribute to the glucose-lowering effects of canagliflozin.⁴³

Advantages, benefits. These agents reduce hemoglobin A_1c by about 0.5% to 1.0% from a baseline of about 8%. Because their action is independent of insulin, they can be used at any stage of type 2 diabetes, even after insulin secretion has significantly waned. Additional potential advantages include weight loss (up to 3.5% of body mass index) and lowering of systolic blood pressure (2–4 mm Hg) and diastolic blood pressure (1–2 mm Hg).^39–42

Canagliflozin was shown in the Canagliflozin Cardiovascular Assessment Study (CANVAS)⁴⁴ to significantly reduce the overall risk of cardiovascular disease by 14% and risk of heart failure hospitalization by 33% while significantly slowing the progression of renal disease.

In the BI 10773 (Empagliflozin) Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME),⁴⁵ empagliflozin reduced heart failure hospitalizations by 35%, cardiovascular deaths by 38%, and all-cause mortality by about 32%. These benefits are thought to be due to less arterial stiffness, lower sympathetic tone, and decreased arrhythmias. Notably, these dramatic benefits accrued in only about 3 years with use of add-on therapy, even though the reduction in hemoglobin A_1c was modest (0.6%), suggesting that pleiotropic effects are at work.

Disadvantages, adverse effects. The most common adverse effects of this class include urinary tract infections, yeast infections, dehydration, and hypovolemic symptoms; these can often be prevented. A trend toward increased incidence of amputations in earlier studies was not borne out in a 2018 meta-analysis of 4 observational databases.⁴⁶

Thiazolidinediones

There are currently 2 approved thiazolidine­diones in the United States, pioglitazone⁴⁷ and rosiglitazone.⁴⁸ Only pioglitazone is in common use, as rosiglitazone is associated with safety issues.⁴⁹

Pathways affected. Pioglitazone reduces insulin resistance in muscle, liver, and adipose tissue.

Advantages, benefits. Decreased levels of low-density lipoprotein cholesterol and triglycerides and increased high-density lipoprotein cholesterol levels⁴⁹ could plausibly account for the cardiovascular benefits reported in the Prospective Pioglitazone Clinical Trial in Macrovascular Events.⁵⁰ Pioglitazone has also been found to improve insulin secretion, endothelial function, and diastolic dysfunction; reduce inflammation; decrease plasminogen activator inhibitor 1; reverse lipotoxicity; and help correct nonalcoholic fatty liver disease and steatohepatitis.

Pioglitazone has also been found to reduce plaque in carotid and coronary arteries⁵¹; improve outcomes in patients with heart failure and myocardial infarction compared with insulin-sensitizing drugs⁵²; and reduce stroke and myocardial infarction in patients with insulin resistance (but not diabetes) and a recent history of ischemic stroke or transient ischemic attack (in the Insulin Resistance Intervention After Stroke trial).⁵³ It may also help maintain beta-cell function; the Actos Now for the Prevention of Diabetes Study found that pioglitazone reduced the risk of conversion of impaired glucose tolerance to frank diabetes by 72%.⁵⁴

Disadvantages, adverse effects. The most common adverse effects seen with this class include weight gain and salt retention, swelling, edema,⁵⁵ and related cardiovascular consequences in certain patients. While this may be mitigatable with lifestyle changes or use in combination with a GLP-1 receptor agonist or SGLT2 inhibitor,⁵⁶ pioglitazone is contraindicated in patients with heart failure, hemodynamic instability, or hepatic dysfunction.

Concerns that pioglitazone might increase the risk of bladder cancer seem to have been put to rest when a study in nearly 200,000 patients found no statistically significant association,⁵⁷ but the warning remains in the US label.

Long-term use of this class of drugs has been associated with an increased risk of bone fractures,⁵⁸ which warrants a risk-benefit assessment in each patient.

Injected insulin: Less safe than thought

Recent research suggests that injected insulin has a less favorable safety profile than previously thought.^15–19,59 Studies of the long-term safety of insulin therapy have had inconsistent results but suggest that injected insulin is associated with poorer cardiovascular and renal outcomes (in some of the same studies that showed metformin or other agents to improve outcomes),^17–19 and the association was dose-dependent. Several studies attempted to cancel out the poorer outcomes by adjusting for hemoglobin A_1c levels, stage of disease,^{17–19,26,27} or severe hypoglycemic episodes.⁶⁰ However, it may be inappropriate to reduce the impact of these variables, as these may themselves be the mediators of any deleterious effects of exogenous insulin.

When exogenous insulin is introduced into the peripheral circulation it causes a state of persistent iatrogenic hyperinsulinemia, which leads to insulin resistance and also appears to compromise the cardiovascular system. In contrast, endogenous insulin is released into the portal system in tightly controlled amounts.^5,61 This suggests that the same insulin peptide may not be equivalently beneficial when introduced in an artificial manner.

Before starting insulin therapy, consider its side effects such as weight gain and hypoglycemia. Most (about 85%) episodes of hypoglycemia occur with basal-bolus insulin regimens.⁶² Moreover, iatrogenic hyperinsulinemia can damage the vascular system.^63,64

We recommend. Insulin therapy is used early in the course of the disease as a short-term intervention for glucolipotoxicity. However, this can be accomplished without attendant risks of hypoglycemia and weight gain by using agents such as SGLT2 inhibitors and incretins. When insulin therapy is necessary, using it as add-on therapy might be considered instead of drug-switching. We have found alternate pharmacologic approaches successful in avoiding or delaying bolus insulin therapy. And in some patients taking insulin, we have had success in progressively introducing a noninsulin agent and were ultimately able to eliminate insulin altogether.

Bromocriptine-QR (quick release)⁶⁵ is a short-acting dopamine agonist that mimics the morning dopamine surge in the suprachiasmatic nucleus—the biologic clock.

Pathways affected. Bromocriptine addresses part of the brain contribution to hyperglycemia, with resultant reductions in both peripheral insulin resistance and sympathetic tone. This reduces muscle, liver, and adipose insulin resistance. It is moderately effective in glucose-lowering, especially in patients with significant insulin resistance.⁶⁶

Advantages, benefits. A 1-year clinical trial reported that bromocriptine reduced cardiovascular adverse outcomes by 39%, and the composite end point of myocardial infarction, stroke, and cardiovascular death by 52% compared with placebo.⁶⁷

Disadvantages, adverse effects. The most common adverse effects are nausea, rhinitis, headache, asthenia, dizziness, constipation, and sinusitis.

Alpha-glucosidase inhibitors

Alpha-glucosidase inhibitors (acarbose,⁶⁸ miglitol⁶⁹) work by decreasing the rate of absorption of glucose from the gastrointestinal tract.

Advantages, benefits. These drugs decrease hemoglobin A_1c by 0.5% to 0.8%.70 They are weight-neutral and do not pose a risk of hypoglycemia. Clinical studies suggest that they may delay or prevent diabetes progression. They were also found to reduce cardiovascular events, acute myocardial infarction, and the onset of hypertension.⁶⁹

Disadvantages, adverse effects. Their use remains limited due to gastrointestinal adverse effects. They may be contraindicated in patients with inflammatory bowel disease, partial bowel obstruction, or severe renal or hepatic disease.

Pramlintide

Pramlintide⁷¹ is an injectable amylin analogue. It is used as monotherapy or in combination with a sulfonylurea, metformin, or insulin glargine.

Pathways affected. Pramlintide decreases appetite, reduces glucagon levels, and minimizes absorption of glucose in the gut.

Disadvantages, adverse effects. Common side effects include mild to moderate hypoglycemia and nausea. Nausea may help explain the ability of pramlintide to confer weight loss when used in combination with insulin.

Sulfonylureas and meglitinides

These classes are still widely used in the treatment of type 2 diabetes, although the AACE⁶ and ADA⁷² guidelines de-emphasize their use based on associated risks of hypoglycemia, weight gain, morbidity, mortality, and loss of effect over time.

Pathways affected. Sulfonylureas stimulate insulin secretion from beta cells.

Disadvantages, adverse effects. Sulfonylureas and glinides are associated with poorer outcomes than newer agents in clinical trials^{15–19,59,60} and may be generally less beta-cell friendly.⁷³ Their harmful effects are difficult to measure in vivo, but these drugs sometimes appear to be associated with more rapid beta-cell failure and progression to insulin dependence compared with newer ones. Several large-scale registry studies have found sulfonylureas and glinides to be associated with poorer outcomes (reviewed by Herman et al).⁷⁴

Adverse effects include asthenia, headache, dizziness, nausea, diarrhea, epigastric fullness, and heartburn. Although they are often selected based on their low cost, other factors may offset their cost-effectiveness, such as need for glucose monitoring and hospital charges due to sulfonylurea-induced hypoglycemia. Their utility is also limited by dependence on beta-cell function.

Colesevelam

Colesevelam⁷⁵ is a bile acid sequestrant and low-density lipoprotein cholesterol-reducing agent that has been approved for use in diabetes. The mode of action of colesevelam in this capacity is under investigation. Its effect on hemoglobin A_1c is modest. It is associated with gastrointestinal adverse effects, particularly constipation.

Ranolazine

Ranolazine⁷⁶ is an antianginal drug that also lowers glucose by increasing insulin release. It also possesses cardioprotective properties. In patients with diabetes and non-ST-segment elevation acute coronary syndromes, ranolazine reduced hemoglobin A_1c by 1.2% and appeared to be weight-neutral.⁷⁶ Ranolazine is under clinical development for use in diabetes. Adverse effects include dizziness, headache, constipation, and nausea.


Rational combinations of agents

The ideal strategy would use combinations of agents that mechanistically complement one another and address each path of hyperglycemia present in a patient. This approach should supplant the former approaches of adding-on agents only after treatment failure or sequentially trying first-, second-, and third-line treatments.

Examples of synergistic combinations include those that target fasting plasma glucose and postprandial glucose, reduce reliance on insulin with add-on therapies, or manage hyperglycemia in specific patient groups, such as renal-impaired patients.

Large-scale long-term clinical studies are needed to determine the safety, efficacy, and outcomes of various combinations and whether they confer additive benefits. Some studies have begun to explore possible combinations.

Combined metformin, pioglitazone, and exenatide was reported to delay progression of diabetes in early dysglycemia.^77,78 Notably, this combination addresses multiple mediating pathways of hyperglycemia (Table 1).

A GLP-1 receptor agonist with an SGLT2 inhibitor would be another intriguing combination, as the mechanisms of action of these 2 classes complement one another. In limited clinical trials—the DURATION-8 study (lasting 26 weeks),⁷⁹ the Canagliflozin Cardiovascular Assessment Study (18 weeks),⁸⁰ and a 24-week study in nondiabetic obese patients⁸¹—additive benefits were also seen in systolic blood pressure, body weight, and cardiac risk factors by adding an SGLT2 inhibitor to a GLP-1 receptor agonist, compared with either agent alone. In theory, these improvements might slow or reverse cardiorenal compromise. Lower doses of 1 or more may be possible, and the regimen could prove cost-effective and life-sparing should it slow the progression of the disease and the onset of its complications. A clinical study of this combination is under way (Ralph DeFronzo, personal communication, July 2018). Similarly, the combination of metformin, saxagliptin and dapagliflozin has been shown to be effective.⁸²

CONCLUSION

Care for diabetes mellitus can be particularly challenging for the primary care physician. The progressive nature of diabetes, with worsening hyperglycemia over the course of the disease, further complicates disease management.

Best practices for care nonetheless need to evolve with well-evidenced data, and without years of delay for “trickle-down” education from the specialties to primary care. We have arrived at a juncture to leverage therapies that address the 11 mediating pathways of hyperglycemia, optimally protect beta cells, minimize hypoglycemia, manage risk factors associated with diabetes, and improve diabetes-related outcomes.

Insights from basic and clinical research are changing the way we treat diabetes mellitus. In 2016, several key diabetes organizations, ie, the American Diabetes Association (ADA), the Juvenile Diabetes Research Foundation (JDRF), the European Association for the Study of Diabetes (EASD), and the American Association of Clinical Endocrinologists (AACE), called for bringing therapeutic approaches in line with our updated understanding of disease pathophysiology, replacing “one-size-fits-all” management with a tailored approach.¹ This message has since been reiterated.²

Here, we review advances in our understanding of diabetes and how these inform a new model of diabetes treatment.

BETA CELLS ARE KEY

High levels of glucose and lipids damage and eventually kill beta cells through mechanisms including that of oxidative stress, so that glucose control deteriorates over time. The same processes are active in the target-organ damage seen in diabetes.^3,4 These 2 insights—that the disease arises from combinatorial, nondiscrete pressures and that it proceeds through common processes of cell damage—leads us to a more unified understanding of the mechanism of diabetes, and may eventually replace current classifications of type 1, type 2, or latent autoimmune diabetes in adults, as well as nomenclature such as “microvascular” and “macrovascular” disease.³

FIRST-LINE LIFESTYLE INTERVENTIONS

Lifestyle interventions are the first-line therapy for elevated blood glucose. Achieving and maintaining a healthy body mass index is essential to help correct insulin resistance and minimize beta-cell dysfunction.

Lifestyle modifications for overweight or obese patients with diabetes mellitus include optimal caloric intake, decreased intake of simple carbohydrates, increased physical activity, and a 3% to 5% reduction in body weight.⁵ Weight-loss drugs may be indicated in obese patients. Normalization of lipids and hypertension should be an early goal.

RIGHT MEDICATIONS, RIGHT PATIENTS

While all of the drugs approved for treating diabetes lower glucose levels, some are more beneficial than others, possessing actions beyond their effect on plasma glucose levels, both good and bad.

The AACE guideline for use of various antidiabetic medications⁶ grades factors such as risks of hypoglycemia, ketoacidosis, weight gain, cardiovascular events, and renal, gastrointestinal, and bone concerns. This represents a much-needed first step toward guidance on selecting the right medications for the right patients. Risk factors (such as heart failure) and comorbidities (such as nonalcoholic fatty liver disease and nonalcoholic steatohepatitis) are among the considerations for choosing treatment.

Two principles

We propose 2 principles when choosing treatment:

Use “gentle” agents, ie, those that are least likely to exhaust beta cells or damage the organs involved in diabetes-related complications. Since the disease course depends on the health of the beta cells, give preference to agents that appear to best support beta cells—ie, agents that create the least oxidative stress or wear-and-tear—as will be outlined in this article.

Diabetes is associated with risks of cardiovascular disease, cardiac events, heart failure, and accelerated renal decompensation. Thus, it is equally important to prevent damage to the cardiovascular system, kidneys, and other tissues subject to damage through glucolipotoxicity.

Balancing glycemic control and risk

The hemoglobin A_1c level is the chief target of care and an important barometer of risk of diabetes-related complications. In 2018, the American College of Physicians (ACP) relaxed its target for hemoglobin A_1c from 7% to 8%.⁸ This move was apparently to give physicians greater “wiggle room” for achieving goals in hypoglycemia-prone patients. This, however, may take a toll.

Hypoglycemia is closely tied to cardiovascular disease. Even mild and asymptomatic hypoglycemia that goes undiagnosed and unnoticed by patients has been found to be associated with higher rates of all-cause mortality, prolonged QT interval, angina, arrhythmias, myocardial dysfunction, disturbances in autonomic balance, and sudden death.^9–11

However, the ADA, AACE, American Association of Diabetes Educators (AADE), and the Endocrine Society jointly issued a strong indictment of the ACP recommendation.¹² They argue that tight glucose control and its well-documented “legacy effects” on long-term outcomes should not be sacrificed.^12,13 Indeed, there is no need to abandon evidence-based best practices in care when at least 8 of the 11 classes of antidiabetes agents do not introduce the same level of risk for hypoglycemia.

Current guidelines argue for tight glucose control but generally stop short of discriminating or stratifying the mechanisms of action of the individual classes of drugs. These guidelines also do not stress targeting the particular pathways of hyperglycemia present in any given patient. However, the 2016 ADA joint statement acknowledges the need to “characterize the many paths to beta-cell dysfunction or demise and identify therapeutic approaches that best target each path.”¹

The sections below highlight some of the recent data on the profiles of most of the currently available agents.

Metformin: Still the first-line treatment

Current guidelines from the ACP, ADA, and AACE keep metformin¹⁴ as the backbone of treatment, although debate continues as to whether newer agents such as GLP-1 receptor agonists are superior for first-line therapy.

Pathways affected. Metformin improves insulin resistance in the liver, increases endogenous GLP-1 levels via the gut, and appears to modulate gut flora composition, which is increasingly suspected to contribute to dysmetabolism. 

Advantages, benefits. Metformin is easy to use and does not cause hypoglycemia. It was found to modestly reduce the number of cardiovascular events and deaths in a number of clinical outcome studies.^15–19

Disadvantages, adverse effects. In some patients, tolerability restricts the use of this drug at higher doses. The most common adverse effects of metformin are gastrointestinal symptoms (diarrhea, nausea, vomiting, flatulence); other risks include lactic acidosis in patients with impaired kidney function, heart failure, hypoxemia, alcoholism, cirrhosis, contrast exposure, sepsis, and shock.

GLP-1 receptor agonists

GLP-1 receptor agonists^20–25 are injectable medications approved for adults with type 2 diabetes. Exenatide and liraglutide lower hemoglobin A_1c by 1 to 1.5 absolute percentage points and reduce body weight; these effects persist over the long term.²⁶ Newer once-weekly GLP-1 receptor agonists (albiglutide,²⁰ dulaglutide,²¹ and semaglutide²⁵) have similar benefits. In 2019, new drug applications were submitted to the FDA for the first-in-kind oral GLP-1 receptor agonists, which would improve convenience and adherence and make this class even more attractive.

Pathways affected. GLP-1 receptor agonists address multiple pathways of hyperglycemia. They increase insulin production and release, promote weight loss, and reduce insulin resistance, glucagon secretion, and inflammation. They also increase amylin, help overcome GLP-1 resistance, slow gastric emptying, and favorably modify gut flora.²⁷

Advantages, benefits. The cardioprotective actions of GLP-1 receptor agonists include reducing inflammation and dysfunction in endothelial and myocardial cells; slowing atherosclerosis; reducing oxidative stress-induced injury and scavenging of reactive oxygen species in coronary endothelial, smooth muscle, and other cells; and enhancing endogenous antioxidant defenses.²⁷ GLP-1 receptor agonism has also been found to inhibit apoptosis in cardiomyocytes, as well as in beta cells.

Several large-scale studies have shown improved outcomes with GLP-1 receptor agonists. The Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results (LEADER) trial²⁶ found that liraglutide reduced major adverse cardiovascular events by 13% and myocardial infarctions by 22% in more than 9,000 adults with type 2 diabetes who were at high risk of major adverse cardiovascular events compared with placebo. Rates of microvascular outcomes were also reduced.

A retrospective database analysis of 39,275 patients with type 2 diabetes who were treated with exenatide reported a lower incidence of cardiovascular events than in patients not treated with exenatide.²⁸

However, no effect on cardiovascular outcomes was found with a third GLP-1 agent, lixisenatide, in a large-scale trial in high-risk patients with diabetes.²⁹

The most recently evaluated GLP-1 receptor agonist is semaglutide. The Trial to Evaluate Cardiovascular and Other Long-term Outcomes With Semaglutide in Subjects With Type 2 Diabetes (SUSTAIN-6) demonstrated a reduced risk of major adverse cardiovascular events.³⁰

Disadvantages, adverse effects. The most common adverse effects in this class include nausea, hypoglycemia, diarrhea, constipation, vomiting, headache, decreased appetite, dyspepsia, fatigue, dizziness, abdominal pain, and increased lipase. The nausea can be mitigated by advising patients to stop eating at first sensation of stomach fullness.

DPP-4 inhibitors

Dipeptidyl peptidase 4 (DPP-4) is a ubiquitous enzyme that rapidly degrades GLP-1 and other endogenous peptides.³¹ Saxagliptin,³² sitagliptin,³³ linagliptin,³⁴ and alogliptin³⁵ are approved for use in the United States, and vildagliptin36 is available in Europe.

Pathways affected. These agents modify 3 pathways of hyperglycemia: they increase insulin secretion, decrease glucagon levels, and help overcome GLP-1 resistance.

Advantages, benefits. DPP-4 inhibitors have been used safely and effectively in clinically challenging populations of patients with long-standing type 2 diabetes (> 10 years).

Disadvantages, adverse effects. As this class increases GLP-1 levels only 2- to 4-fold, their efficacy is more modest than that of GLP-1 receptor agonists (hemoglobin A_1c reductions of 0.5% to 1%; neutral effects on weight).³⁷

Outcome trials have largely been neutral. Saxagliptin has been associated with an increase in admissions for heart failure. There have been a very small but statistically significant number of drug-related cases of acute pancreatitis.³⁸

The most common adverse effects with this class include headache, nasopharyngitis, urinary tract infection, upper respiratory tract infection, and elevated liver enzymes.

Drugs of this class currently available in the United States are canagliflozin,³⁹ dapagliflozin,⁴⁰ empagliflozin,⁴¹ and ertugliflozin.⁴²

Pathways affected. SGLT2 inhibitors lower the glucose reabsorption threshold in the kidney so that more glucose is excreted in the urine; they also decrease insulin resistance in muscle, liver, and fat cells (via weight loss) and possibly preserve beta-cell function by reducing glucotoxicity. A nonrenal mechanism—delayed gut absorption reducing postprandial glucose excursion—has been proposed to contribute to the glucose-lowering effects of canagliflozin.⁴³

Advantages, benefits. These agents reduce hemoglobin A_1c by about 0.5% to 1.0% from a baseline of about 8%. Because their action is independent of insulin, they can be used at any stage of type 2 diabetes, even after insulin secretion has significantly waned. Additional potential advantages include weight loss (up to 3.5% of body mass index) and lowering of systolic blood pressure (2–4 mm Hg) and diastolic blood pressure (1–2 mm Hg).^39–42

Canagliflozin was shown in the Canagliflozin Cardiovascular Assessment Study (CANVAS)⁴⁴ to significantly reduce the overall risk of cardiovascular disease by 14% and risk of heart failure hospitalization by 33% while significantly slowing the progression of renal disease.

In the BI 10773 (Empagliflozin) Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients (EMPA-REG OUTCOME),⁴⁵ empagliflozin reduced heart failure hospitalizations by 35%, cardiovascular deaths by 38%, and all-cause mortality by about 32%. These benefits are thought to be due to less arterial stiffness, lower sympathetic tone, and decreased arrhythmias. Notably, these dramatic benefits accrued in only about 3 years with use of add-on therapy, even though the reduction in hemoglobin A_1c was modest (0.6%), suggesting that pleiotropic effects are at work.

Disadvantages, adverse effects. The most common adverse effects of this class include urinary tract infections, yeast infections, dehydration, and hypovolemic symptoms; these can often be prevented. A trend toward increased incidence of amputations in earlier studies was not borne out in a 2018 meta-analysis of 4 observational databases.⁴⁶

Thiazolidinediones

There are currently 2 approved thiazolidine­diones in the United States, pioglitazone⁴⁷ and rosiglitazone.⁴⁸ Only pioglitazone is in common use, as rosiglitazone is associated with safety issues.⁴⁹

Pathways affected. Pioglitazone reduces insulin resistance in muscle, liver, and adipose tissue.

Advantages, benefits. Decreased levels of low-density lipoprotein cholesterol and triglycerides and increased high-density lipoprotein cholesterol levels⁴⁹ could plausibly account for the cardiovascular benefits reported in the Prospective Pioglitazone Clinical Trial in Macrovascular Events.⁵⁰ Pioglitazone has also been found to improve insulin secretion, endothelial function, and diastolic dysfunction; reduce inflammation; decrease plasminogen activator inhibitor 1; reverse lipotoxicity; and help correct nonalcoholic fatty liver disease and steatohepatitis.

Pioglitazone has also been found to reduce plaque in carotid and coronary arteries⁵¹; improve outcomes in patients with heart failure and myocardial infarction compared with insulin-sensitizing drugs⁵²; and reduce stroke and myocardial infarction in patients with insulin resistance (but not diabetes) and a recent history of ischemic stroke or transient ischemic attack (in the Insulin Resistance Intervention After Stroke trial).⁵³ It may also help maintain beta-cell function; the Actos Now for the Prevention of Diabetes Study found that pioglitazone reduced the risk of conversion of impaired glucose tolerance to frank diabetes by 72%.⁵⁴

Disadvantages, adverse effects. The most common adverse effects seen with this class include weight gain and salt retention, swelling, edema,⁵⁵ and related cardiovascular consequences in certain patients. While this may be mitigatable with lifestyle changes or use in combination with a GLP-1 receptor agonist or SGLT2 inhibitor,⁵⁶ pioglitazone is contraindicated in patients with heart failure, hemodynamic instability, or hepatic dysfunction.

Concerns that pioglitazone might increase the risk of bladder cancer seem to have been put to rest when a study in nearly 200,000 patients found no statistically significant association,⁵⁷ but the warning remains in the US label.

Long-term use of this class of drugs has been associated with an increased risk of bone fractures,⁵⁸ which warrants a risk-benefit assessment in each patient.

Injected insulin: Less safe than thought

Recent research suggests that injected insulin has a less favorable safety profile than previously thought.^15–19,59 Studies of the long-term safety of insulin therapy have had inconsistent results but suggest that injected insulin is associated with poorer cardiovascular and renal outcomes (in some of the same studies that showed metformin or other agents to improve outcomes),^17–19 and the association was dose-dependent. Several studies attempted to cancel out the poorer outcomes by adjusting for hemoglobin A_1c levels, stage of disease,^{17–19,26,27} or severe hypoglycemic episodes.⁶⁰ However, it may be inappropriate to reduce the impact of these variables, as these may themselves be the mediators of any deleterious effects of exogenous insulin.

When exogenous insulin is introduced into the peripheral circulation it causes a state of persistent iatrogenic hyperinsulinemia, which leads to insulin resistance and also appears to compromise the cardiovascular system. In contrast, endogenous insulin is released into the portal system in tightly controlled amounts.^5,61 This suggests that the same insulin peptide may not be equivalently beneficial when introduced in an artificial manner.

Before starting insulin therapy, consider its side effects such as weight gain and hypoglycemia. Most (about 85%) episodes of hypoglycemia occur with basal-bolus insulin regimens.⁶² Moreover, iatrogenic hyperinsulinemia can damage the vascular system.^63,64

We recommend. Insulin therapy is used early in the course of the disease as a short-term intervention for glucolipotoxicity. However, this can be accomplished without attendant risks of hypoglycemia and weight gain by using agents such as SGLT2 inhibitors and incretins. When insulin therapy is necessary, using it as add-on therapy might be considered instead of drug-switching. We have found alternate pharmacologic approaches successful in avoiding or delaying bolus insulin therapy. And in some patients taking insulin, we have had success in progressively introducing a noninsulin agent and were ultimately able to eliminate insulin altogether.

Bromocriptine-QR (quick release)⁶⁵ is a short-acting dopamine agonist that mimics the morning dopamine surge in the suprachiasmatic nucleus—the biologic clock.

Pathways affected. Bromocriptine addresses part of the brain contribution to hyperglycemia, with resultant reductions in both peripheral insulin resistance and sympathetic tone. This reduces muscle, liver, and adipose insulin resistance. It is moderately effective in glucose-lowering, especially in patients with significant insulin resistance.⁶⁶

Advantages, benefits. A 1-year clinical trial reported that bromocriptine reduced cardiovascular adverse outcomes by 39%, and the composite end point of myocardial infarction, stroke, and cardiovascular death by 52% compared with placebo.⁶⁷

Disadvantages, adverse effects. The most common adverse effects are nausea, rhinitis, headache, asthenia, dizziness, constipation, and sinusitis.

Alpha-glucosidase inhibitors

Alpha-glucosidase inhibitors (acarbose,⁶⁸ miglitol⁶⁹) work by decreasing the rate of absorption of glucose from the gastrointestinal tract.

Advantages, benefits. These drugs decrease hemoglobin A_1c by 0.5% to 0.8%.70 They are weight-neutral and do not pose a risk of hypoglycemia. Clinical studies suggest that they may delay or prevent diabetes progression. They were also found to reduce cardiovascular events, acute myocardial infarction, and the onset of hypertension.⁶⁹

Disadvantages, adverse effects. Their use remains limited due to gastrointestinal adverse effects. They may be contraindicated in patients with inflammatory bowel disease, partial bowel obstruction, or severe renal or hepatic disease.

Pramlintide

Pramlintide⁷¹ is an injectable amylin analogue. It is used as monotherapy or in combination with a sulfonylurea, metformin, or insulin glargine.

Pathways affected. Pramlintide decreases appetite, reduces glucagon levels, and minimizes absorption of glucose in the gut.

Disadvantages, adverse effects. Common side effects include mild to moderate hypoglycemia and nausea. Nausea may help explain the ability of pramlintide to confer weight loss when used in combination with insulin.

Sulfonylureas and meglitinides

These classes are still widely used in the treatment of type 2 diabetes, although the AACE⁶ and ADA⁷² guidelines de-emphasize their use based on associated risks of hypoglycemia, weight gain, morbidity, mortality, and loss of effect over time.

Pathways affected. Sulfonylureas stimulate insulin secretion from beta cells.

Disadvantages, adverse effects. Sulfonylureas and glinides are associated with poorer outcomes than newer agents in clinical trials^{15–19,59,60} and may be generally less beta-cell friendly.⁷³ Their harmful effects are difficult to measure in vivo, but these drugs sometimes appear to be associated with more rapid beta-cell failure and progression to insulin dependence compared with newer ones. Several large-scale registry studies have found sulfonylureas and glinides to be associated with poorer outcomes (reviewed by Herman et al).⁷⁴

Adverse effects include asthenia, headache, dizziness, nausea, diarrhea, epigastric fullness, and heartburn. Although they are often selected based on their low cost, other factors may offset their cost-effectiveness, such as need for glucose monitoring and hospital charges due to sulfonylurea-induced hypoglycemia. Their utility is also limited by dependence on beta-cell function.

Colesevelam

Colesevelam⁷⁵ is a bile acid sequestrant and low-density lipoprotein cholesterol-reducing agent that has been approved for use in diabetes. The mode of action of colesevelam in this capacity is under investigation. Its effect on hemoglobin A_1c is modest. It is associated with gastrointestinal adverse effects, particularly constipation.

Ranolazine

Ranolazine⁷⁶ is an antianginal drug that also lowers glucose by increasing insulin release. It also possesses cardioprotective properties. In patients with diabetes and non-ST-segment elevation acute coronary syndromes, ranolazine reduced hemoglobin A_1c by 1.2% and appeared to be weight-neutral.⁷⁶ Ranolazine is under clinical development for use in diabetes. Adverse effects include dizziness, headache, constipation, and nausea.


Rational combinations of agents

The ideal strategy would use combinations of agents that mechanistically complement one another and address each path of hyperglycemia present in a patient. This approach should supplant the former approaches of adding-on agents only after treatment failure or sequentially trying first-, second-, and third-line treatments.

Examples of synergistic combinations include those that target fasting plasma glucose and postprandial glucose, reduce reliance on insulin with add-on therapies, or manage hyperglycemia in specific patient groups, such as renal-impaired patients.

Large-scale long-term clinical studies are needed to determine the safety, efficacy, and outcomes of various combinations and whether they confer additive benefits. Some studies have begun to explore possible combinations.

Combined metformin, pioglitazone, and exenatide was reported to delay progression of diabetes in early dysglycemia.^77,78 Notably, this combination addresses multiple mediating pathways of hyperglycemia (Table 1).

A GLP-1 receptor agonist with an SGLT2 inhibitor would be another intriguing combination, as the mechanisms of action of these 2 classes complement one another. In limited clinical trials—the DURATION-8 study (lasting 26 weeks),⁷⁹ the Canagliflozin Cardiovascular Assessment Study (18 weeks),⁸⁰ and a 24-week study in nondiabetic obese patients⁸¹—additive benefits were also seen in systolic blood pressure, body weight, and cardiac risk factors by adding an SGLT2 inhibitor to a GLP-1 receptor agonist, compared with either agent alone. In theory, these improvements might slow or reverse cardiorenal compromise. Lower doses of 1 or more may be possible, and the regimen could prove cost-effective and life-sparing should it slow the progression of the disease and the onset of its complications. A clinical study of this combination is under way (Ralph DeFronzo, personal communication, July 2018). Similarly, the combination of metformin, saxagliptin and dapagliflozin has been shown to be effective.⁸²

CONCLUSION

Care for diabetes mellitus can be particularly challenging for the primary care physician. The progressive nature of diabetes, with worsening hyperglycemia over the course of the disease, further complicates disease management.

Best practices for care nonetheless need to evolve with well-evidenced data, and without years of delay for “trickle-down” education from the specialties to primary care. We have arrived at a juncture to leverage therapies that address the 11 mediating pathways of hyperglycemia, optimally protect beta cells, minimize hypoglycemia, manage risk factors associated with diabetes, and improve diabetes-related outcomes.

The clinical update of giant cell arteritis (GCA) by Rinden et al in this issue of the Journal reminded me of just how much of our management of this disease has, for decades, been based on retrospective studies (we owe a lot to clinicians from the Mayo Clinic for their compiled observations) tempered by our own recalled experiences, which we may at times twist a bit to fit prevailing paradigms. Several prospective interventional studies, perhaps most importantly the Giant-Cell Arteritis Actemra (GIACTA) trial,¹ evaluated the ability of the interleukin 6 (IL-6) antagonist tocilizumab to supplant the protracted use of glucocorticoids in the treatment of GCA. But I learned much more from this trial, in the form of collected clinical tidbits, than just the bottom-line abstract conclusion that IL-6 antagonism is at least a promising approach in many patients with GCA.

As teachers, we tell students to read the entire published clinical trial report, not just the abstract and conclusions. Over the years, I have been impatient with those who violated this dictum, but I now often find myself among the ranks of those who would have been targets of my disapproval. Usually, the articles that I merely skim lie outside my subsubspecialty areas of interest, as time constraints make this abridged reading a necessity for survival, but that excuse does not diminish the self-recognition of my often less-than-complete understanding of the clinical condition being reported. Delving into the nuances of GIACTA truly emphasized that point.

The external validity of any trial rests on understanding the trial’s methods. In the case of GIACTA, there was much more to be learned and affirmed from the trial¹ than that 1 year of tocilizumab treatment met the primary end point of increasing the percent of patients achieving sustained remission at week 52 after a rapid 26-week tapering off of prednisone compared with placebo.

One treatment group in the GIACTA trial underwent an aggressive 6-month tapering of prednisone, while another underwent a more protracted tapering over 12 months (more in line with common practice). Patients tapered over 6 months also received either the IL-6 antagonist or placebo for the full year. The concept was that if IL-6 blockade is a correct approach, then it will maintain remission in more patients, and significantly reduce the total amount of steroid needed to control the disease, despite rapid, aggressive steroid tapering. This turned out to be correct, although more than 20% of the drug-treated patients still experienced a flare of GCA (vs 68% of the placebo-treated group).

Somewhat surprising was that almost 20% of the entered patients did not achieve an initial remission despite receiving high-dose prednisone. The traditional teaching is that if a patient diagnosed with GCA does not respond to high-dose steroids, the diagnosis should be questioned.

Another interesting facet of the study relates to the diagnosis. We are becoming more aware of the different GCA phenotypes, which include prominent polymyalgia rheumatica or constitutional features, “classic” GCA with cranial symptoms, and dominant large-vessel vasculitis (aortitis and major aortic branch disease). In GIACTA, even though imaging was not mandated, 37% of participants were enrolled based in part on imaging results (CT, MRI, angiography, or PET-CT), not on the results of temporal artery biopsy. This forces us to think more broadly about diagnosing and staging GCA, and to wonder if we should even modify our approach to other clinical challenges, including unexplained fever and wasting in older patients.

Another tidbit that came out of the study relates to the relationship between the acute-phase response and clinical flares. We already knew that a rise in the erythrocyte sedimentation rate is a nonspecific sign and does not equate with a flare. In this trial one-third of patients in the placebo group who had a flare had a normal sedimentation rate or C-reactive protein during the flare, and about one-third of patients in the placebo group were receiving more than 10 mg of prednisone. In preliminary reports of follow-up after  52 weeks of treatment,² patients who had achieved complete remission with the IL-6 antagonist and were off of prednisone were still not out of the woods; when the drug was discontinued, many flares continued to occur over the 2-year study extension. We have more to learn about what triggers and drives flares in this disease.

Thus, in addition to informing us of a successful “steroid-sparing” and rescue drug option for our patients with GCA, the details of this well-conducted trial both challenge and reaffirm some of our clinical impressions. Clearly, GCA must be defined for many patients as a very chronic disease, perhaps with occult vascular reservoirs, the biologic basis of which remains to be defined.

The clinical update of giant cell arteritis (GCA) by Rinden et al in this issue of the Journal reminded me of just how much of our management of this disease has, for decades, been based on retrospective studies (we owe a lot to clinicians from the Mayo Clinic for their compiled observations) tempered by our own recalled experiences, which we may at times twist a bit to fit prevailing paradigms. Several prospective interventional studies, perhaps most importantly the Giant-Cell Arteritis Actemra (GIACTA) trial,¹ evaluated the ability of the interleukin 6 (IL-6) antagonist tocilizumab to supplant the protracted use of glucocorticoids in the treatment of GCA. But I learned much more from this trial, in the form of collected clinical tidbits, than just the bottom-line abstract conclusion that IL-6 antagonism is at least a promising approach in many patients with GCA.

As teachers, we tell students to read the entire published clinical trial report, not just the abstract and conclusions. Over the years, I have been impatient with those who violated this dictum, but I now often find myself among the ranks of those who would have been targets of my disapproval. Usually, the articles that I merely skim lie outside my subsubspecialty areas of interest, as time constraints make this abridged reading a necessity for survival, but that excuse does not diminish the self-recognition of my often less-than-complete understanding of the clinical condition being reported. Delving into the nuances of GIACTA truly emphasized that point.

The external validity of any trial rests on understanding the trial’s methods. In the case of GIACTA, there was much more to be learned and affirmed from the trial¹ than that 1 year of tocilizumab treatment met the primary end point of increasing the percent of patients achieving sustained remission at week 52 after a rapid 26-week tapering off of prednisone compared with placebo.

One treatment group in the GIACTA trial underwent an aggressive 6-month tapering of prednisone, while another underwent a more protracted tapering over 12 months (more in line with common practice). Patients tapered over 6 months also received either the IL-6 antagonist or placebo for the full year. The concept was that if IL-6 blockade is a correct approach, then it will maintain remission in more patients, and significantly reduce the total amount of steroid needed to control the disease, despite rapid, aggressive steroid tapering. This turned out to be correct, although more than 20% of the drug-treated patients still experienced a flare of GCA (vs 68% of the placebo-treated group).

Somewhat surprising was that almost 20% of the entered patients did not achieve an initial remission despite receiving high-dose prednisone. The traditional teaching is that if a patient diagnosed with GCA does not respond to high-dose steroids, the diagnosis should be questioned.

Another interesting facet of the study relates to the diagnosis. We are becoming more aware of the different GCA phenotypes, which include prominent polymyalgia rheumatica or constitutional features, “classic” GCA with cranial symptoms, and dominant large-vessel vasculitis (aortitis and major aortic branch disease). In GIACTA, even though imaging was not mandated, 37% of participants were enrolled based in part on imaging results (CT, MRI, angiography, or PET-CT), not on the results of temporal artery biopsy. This forces us to think more broadly about diagnosing and staging GCA, and to wonder if we should even modify our approach to other clinical challenges, including unexplained fever and wasting in older patients.

Another tidbit that came out of the study relates to the relationship between the acute-phase response and clinical flares. We already knew that a rise in the erythrocyte sedimentation rate is a nonspecific sign and does not equate with a flare. In this trial one-third of patients in the placebo group who had a flare had a normal sedimentation rate or C-reactive protein during the flare, and about one-third of patients in the placebo group were receiving more than 10 mg of prednisone. In preliminary reports of follow-up after  52 weeks of treatment,² patients who had achieved complete remission with the IL-6 antagonist and were off of prednisone were still not out of the woods; when the drug was discontinued, many flares continued to occur over the 2-year study extension. We have more to learn about what triggers and drives flares in this disease.

Thus, in addition to informing us of a successful “steroid-sparing” and rescue drug option for our patients with GCA, the details of this well-conducted trial both challenge and reaffirm some of our clinical impressions. Clearly, GCA must be defined for many patients as a very chronic disease, perhaps with occult vascular reservoirs, the biologic basis of which remains to be defined.

The clinical update of giant cell arteritis (GCA) by Rinden et al in this issue of the Journal reminded me of just how much of our management of this disease has, for decades, been based on retrospective studies (we owe a lot to clinicians from the Mayo Clinic for their compiled observations) tempered by our own recalled experiences, which we may at times twist a bit to fit prevailing paradigms. Several prospective interventional studies, perhaps most importantly the Giant-Cell Arteritis Actemra (GIACTA) trial,¹ evaluated the ability of the interleukin 6 (IL-6) antagonist tocilizumab to supplant the protracted use of glucocorticoids in the treatment of GCA. But I learned much more from this trial, in the form of collected clinical tidbits, than just the bottom-line abstract conclusion that IL-6 antagonism is at least a promising approach in many patients with GCA.

As teachers, we tell students to read the entire published clinical trial report, not just the abstract and conclusions. Over the years, I have been impatient with those who violated this dictum, but I now often find myself among the ranks of those who would have been targets of my disapproval. Usually, the articles that I merely skim lie outside my subsubspecialty areas of interest, as time constraints make this abridged reading a necessity for survival, but that excuse does not diminish the self-recognition of my often less-than-complete understanding of the clinical condition being reported. Delving into the nuances of GIACTA truly emphasized that point.

The external validity of any trial rests on understanding the trial’s methods. In the case of GIACTA, there was much more to be learned and affirmed from the trial¹ than that 1 year of tocilizumab treatment met the primary end point of increasing the percent of patients achieving sustained remission at week 52 after a rapid 26-week tapering off of prednisone compared with placebo.

One treatment group in the GIACTA trial underwent an aggressive 6-month tapering of prednisone, while another underwent a more protracted tapering over 12 months (more in line with common practice). Patients tapered over 6 months also received either the IL-6 antagonist or placebo for the full year. The concept was that if IL-6 blockade is a correct approach, then it will maintain remission in more patients, and significantly reduce the total amount of steroid needed to control the disease, despite rapid, aggressive steroid tapering. This turned out to be correct, although more than 20% of the drug-treated patients still experienced a flare of GCA (vs 68% of the placebo-treated group).

Somewhat surprising was that almost 20% of the entered patients did not achieve an initial remission despite receiving high-dose prednisone. The traditional teaching is that if a patient diagnosed with GCA does not respond to high-dose steroids, the diagnosis should be questioned.

Another interesting facet of the study relates to the diagnosis. We are becoming more aware of the different GCA phenotypes, which include prominent polymyalgia rheumatica or constitutional features, “classic” GCA with cranial symptoms, and dominant large-vessel vasculitis (aortitis and major aortic branch disease). In GIACTA, even though imaging was not mandated, 37% of participants were enrolled based in part on imaging results (CT, MRI, angiography, or PET-CT), not on the results of temporal artery biopsy. This forces us to think more broadly about diagnosing and staging GCA, and to wonder if we should even modify our approach to other clinical challenges, including unexplained fever and wasting in older patients.

Another tidbit that came out of the study relates to the relationship between the acute-phase response and clinical flares. We already knew that a rise in the erythrocyte sedimentation rate is a nonspecific sign and does not equate with a flare. In this trial one-third of patients in the placebo group who had a flare had a normal sedimentation rate or C-reactive protein during the flare, and about one-third of patients in the placebo group were receiving more than 10 mg of prednisone. In preliminary reports of follow-up after  52 weeks of treatment,² patients who had achieved complete remission with the IL-6 antagonist and were off of prednisone were still not out of the woods; when the drug was discontinued, many flares continued to occur over the 2-year study extension. We have more to learn about what triggers and drives flares in this disease.

Thus, in addition to informing us of a successful “steroid-sparing” and rescue drug option for our patients with GCA, the details of this well-conducted trial both challenge and reaffirm some of our clinical impressions. Clearly, GCA must be defined for many patients as a very chronic disease, perhaps with occult vascular reservoirs, the biologic basis of which remains to be defined.

See Mangione and Aronowitz editorials

Mechanical pressure induced by friction of the elbows on the thighs may result in proliferation of the stratum corneum and the release of hemosiderin from erythrocytes, resulting in the skin changes seen in this patient, which because of the tripod positioning are known as “thinker’s sign,” a term coined in 1963 by Rothenberg¹ to describe findings in patients with chronic pulmonary disease and advanced respiratory insufficiency. It is also referred to as the Dahl sign, based on a report by Dahl of similar findings in patients with emphysema.²

See Mangione and Aronowitz editorials

Mechanical pressure induced by friction of the elbows on the thighs may result in proliferation of the stratum corneum and the release of hemosiderin from erythrocytes, resulting in the skin changes seen in this patient, which because of the tripod positioning are known as “thinker’s sign,” a term coined in 1963 by Rothenberg¹ to describe findings in patients with chronic pulmonary disease and advanced respiratory insufficiency. It is also referred to as the Dahl sign, based on a report by Dahl of similar findings in patients with emphysema.²

See Mangione and Aronowitz editorials

Mechanical pressure induced by friction of the elbows on the thighs may result in proliferation of the stratum corneum and the release of hemosiderin from erythrocytes, resulting in the skin changes seen in this patient, which because of the tripod positioning are known as “thinker’s sign,” a term coined in 1963 by Rothenberg¹ to describe findings in patients with chronic pulmonary disease and advanced respiratory insufficiency. It is also referred to as the Dahl sign, based on a report by Dahl of similar findings in patients with emphysema.²

PRIMARY HEART TUMORS ARE RARE