Incorporation of artificial sensors in implantable pacing systems revolutionized the technology of cardiac pacing and its clinical practice [1-6]. Initially, artificial sensors were viewed exclusively as a means to modify the pacing rate in response to changing levels of physical exertion. In that rate-adaptive role, sensor-based pacemakers have been well received by physicians and patients Figure 1, Table 1. More recently, the role of artificial sensors expanded to include automatic adjustment of certain programmable pacemaker settings, such as the interval between atrial and ventricular pacing stimuli (that is, the atrioventricular interval). In addition, sensors are also being used to inform the pacing system of the development of certain arrhythmias, particularly atrial flutter or fibrillation. This latter capability permits those pacemakers that use the patient's atrium to set the pacing rate (atrial tracking) to change their mode of operation to one in which the pacing rate is determined by the sensor. This mode-switching feature prevents the undesirable high-rate pacing that can occur if the pacemaker continues to track the atrial rate during a pathologic atrial tachycardia. Soon, sensors will allow pacemakers to adjust many other operating characteristics automatically (such as the level of energy output, electrical polarity of the intracardiac electrodes, pacing mode, and so on) [7], thereby optimizing efficiency of operation while minimizing the need for specialized medical follow-up.

View larger version (52K): [in this window] [in a new window]   Figure 1. Changes in pacing mode use in North America in the past decade (see [3] ). The ordinate indicates the percentage of pacemakers of a given pacing mode (see Table 1 for definitions) that were implanted during the year indicated on the abscissa. There has been a trend toward more frequent use of dual-chamber pacing (DDD and DDDR modes) and a greater reliance on devices with rate-adaptive capability (DDD, DDDR, and VVIR modes). Single-chamber atrial pacing (AAI and AAIR modes) is used infrequently (fewer than 1% implants annually) and therefore is not plotted on this graphic.

In the past decade, physicians have gained considerable clinical experience with rate-adaptive pacing systems that incorporate a single artificial sensor [8-16]. Although this experience has been largely favorable, no single artificial sensor is suitable for all potential pacing system applications [6, 17]. Consequently, pacemakers using combinations of artificial sensors that can operate in a complementary manner are being developed and clinically evaluated (Table 2). This review surveys the role of artificial sensors in cardiac pacing systems and examines the rationale for and the clinical status of pacemakers that use multiple artificial sensors.

Role of Sensors in Cardiac Pacing Systems

Rationale for Using Sensors To Adjust Pacing Rate

Although prevention of symptoms such as syncope or dizziness resulting from severe bradycardia is the primary goal of pacemaker therapy, careful selection of the pacing system and its mode operation also provide the opportunity to improve exercise tolerance [8-13, 18], to diminish the risk for development of atrial fibrillation [19-24], and to decrease overall mortality rates [18, 21-26]. In terms of exercise tolerance, appropriate heart rate responsiveness contributes the most to increased cardiac output during exertion. This applies to healthy persons and to most patients with pacemakers, whether they are vigorously active or must simply manage the physical demands associated with activities of daily living (such as climbing stairs, walking, carrying groceries, and doing the laundry) [1, 4, 8-18]. For example, studies by Karlof [27], Fananapazir and coworkers [28], Ausubel and colleagues [29], and Ryden and associates [30] clearly show that exercise capacity, stroke volume changes with exercise, and maximum oxygen consumption (a measure of the body's ability to provide essential fuel for vigorous exertion) depend primarily on heart rate change in most patients. In addition, Ryden and associates [30] also showed that over a wide range of atrioventricular intervals, there was no difference between maximum oxygen uptake in patients with pacemakers during rate-adaptive, single- and dual-chamber pacing. Furthermore, in terms of the heart's own energy requirements during exercise, Nordlander and colleagues [31] found that increasing heart rate, such as occurs with rate-adaptive pacing, does not necessarily adversely affect myocardial oxygen consumption compared with fixed-rate pacing. In fact, provision of an appropriate heart rate change during exercise may prevent ventricular dilation and reduce the need for premature encroachment on cardiac compensatory responses.

In principle, providing heart rate responsiveness in many patients with pacemakers can be accomplished using the patient's own native atrial or sinus node rate to determine the appropriate moment-to-moment ventricular paced rate (that is, atrial tracking, such as in the VD or DDD pacing modes shown in Table 1. Indeed, if a healthy sinus node is present, such an approach not only provides the most physiologic heart rate response but also maintains a relatively normal relation between atrial and ventricular contractions. However, symptomatic disturbances of sinoatrial function [that is, the sick sinus syndrome] are common in patients with pacemakers [1, 19-24]. In fact, they are the primary indication for a pacemaker in more than 50% of patients who have them in Western countries [18]. Consequently, tracking the native atrial rate often does not provide the optimal physiologic heart rate response. Consequently, other techniques, such as the incorporation of artificial sensors into implantable pacing systems, are essential in many patients with pacemakers to ensure reliable heart rate responsiveness [1-6, 18].

Current Sensor Systems

The sensing of atrial electrical potentials by an intra-atrial electrode with subsequent adjustment of ventricular pacing rate (that is, atrial tracking) was the first attempt to develop a "sensor-triggered" rate-adaptive pacemaker [32, 33]. In that case, the sensor is the patient's own atrial rate. However, for the reasons previously noted, tracking to adjust the pacing rate may not be suitable in many patients.

The earliest efforts to develop artificial sensors to adjust the rate of implantable cardiac pacemakers were reported in the mid-1960s [34]. However, not until the mid-1970s did Cammilli and coworkers [35], using central venous pH sensing, provide the first implantable device. Theoretically, reduced central venous pH associated with exercise would signal the pacemaker to increase its rate. The resulting additional cardiac output would then not only tend to normalize the pH but ultimately would also cause the pacemaker to slow down again (a "closed-loop" sensor system). Unfortunately, the pH sensor was not particularly stable in the long term, and the relation between pH and heart rate was complex. For the most part, researchers have abandoned the pH sensor concept.

The first commercially successful sensor-based pacing system used a specialized ceramic element with piezoelectric characteristics (that is, they produce an electrical potential when physically deformed, such as by vibration) bonded to the inside surface of the pacemaker shield (Activitrax; Medtronic Inc., Minneapolis, Minnesota) [3, 4, 10-12, 36-39]. With this device and its more recent derivatives, the metal of the pacemaker shield may be thought of as a drum head. The ceramic activity sensor emits electrical signals (the piezoelectric effect) approximately in proportion to the vibrations transmitted through the pacemaker from the patient's working muscle and skeleton. These piezoelectric signals are an indirect but useful reflection of the patient's physical activity (thus, the widely used term activity sensor) and can be used to change the pacing rate appropriately [11, 12, 36-41].

The rapid clinical acceptance of the first activity-based, rate-adaptive pacing system encouraged development of similar pacemakers. Some manufacturers adopted the activity detection concept, either by using the drum head design described previously (such as Synchrony devices, Siemens-Pacesetter Systems, Sylmar, California) or an accelerometer modification (such as Relay; Intermedics Inc., Freeport, Texas). Other manufacturers devised unique rate-adaptive sensor solutions [42-52]. For example, measurement of the interval from the pacing stimulus to the peak of the T wave (the so-called stim-T interval, an estimate of the QT interval for paced heart beats), has proved effective (for example, Rhythmyx; Vitatron Medical b.v., Dieren, the Netherlands) [43]. This concept is based on the fact that the stim-T interval (like the QT interval on the standard electrocardiogram) is modified by changes in circulating or locally released catecholamines. Thus, for a given heart rate, a change in myocardial catecholamine exposure (for example, in association with physical exertion) will modify the stim-T interval and provide a marker for altered levels of physical activity. Similarly, estimation of exercise-induced change in minute ventilation by transthoracic plethysmography (such as Meta; Telectronics Inc., Sydney, Australia), or in central venous temperature using a thermistor on the pacing lead (such as Kelvin; Cook Pacemakers Inc., Leechburg, Pennsylvania) have yielded useful rate-adaptive pacing techniques [14, 44-47]. Additional promising concepts for rate-adaptive sensors include monitoring exercise-induced changes in central venous oxygen saturation using an intravascular oxygen sensor, and measuring exercise-related changes in right ventricular pressure characteristics (using a small pressure sensor built into the ventricular pacing lead) [50-52]. Conversely, some other interesting sensor proposals, such as rate adaptation based on changes of right ventricular stroke volume or pre-ejection interval or using measurement of the area under the paced ventricular depolarization signal ("evoked-response" sensor) [48, 53], have been slow to emerge as clinically useful systems.

Evolving Sensor Applications

In the past few years, sensors in cardiac pacemakers have shown a versatility that far surpasses their original rate-adaptive role [6, 7, 18]. For example, several pulse generators permit independent programming of atrial tracking and sensor-indicated upper pacing rates. Thus, the physician can limit the maximum native atrial rate that the device will track and simultaneously permit the sensor to achieve as high a heart rate as is deemed clinically appropriate for the patient during exercise. This feature is particularly useful in persons susceptible to intermittent atrial tachycardias (particularly atrial fibrillation) and is designed to preclude undesirable tracking of excessively rapid, nonphysiologic atrial rates if a paroxysmal atrial tachyarrhythmia occurs.

Another sensor innovation is cross-checking, in which the sensor confirms that the sensed native atrial rate is indeed physiologically appropriate. If the latter is not confirmed, the device relies more closely on the sensor-indicated rate to protect the patient against inadvertent tracking of abnormal atrial tachycardias. Another example of artificial sensor versatility is automatic capture detection. With pacemakers that use a stim-T interval detector, a paced depolarization sensor, or a right ventricular pressure detector, the sensor can automatically confirm whether the heart muscle has been "captured" by the pacing stimulus (that is, capture detection). This feature not only offers the opportunity for the device to automatically increase energy output if capture has failed but also raises the possibility of improving device longevity by reducing energy output when it is safe to do so. Finally, using automatic sensor-based adjustment of pacemaker timing settings (atrioventricular interval, refractory periods, for example) increasingly enables pacemakers to adjust to changing physiologic circumstances.

Clinical Studies of Single-Sensor, Rate-adaptive Pacing

The usefulness of single artificial sensor, rate-adaptive pacing to improve exercise tolerance in patients with pacemakers is documented in several studies [1, 7-14, 27-30, 54-58]. Early reports assessing sensor-based pacing focused on a comparison of rate-adaptive, single-chamber ventricular pacing (VVIR mode) with fixed-rate, ventricular pacing (VVI mode) because the latter was and, in many parts of the world, remains the standard therapy. For example, even the earliest activity-based pacing system resulted in a 35% increment of treadmill exercise duration, peak oxygen consumption, and anaerobic threshold in typical patients with pacemakers [11]. Briefly, patients could do more work and they did so with lower levels of perceived exertion [11]. Furthermore, this improvement in exercise tolerance was often maintained, as confirmed by follow-up cardiopulmonary exercise testing [54]. Meanwhile, others showed that provision of rate adaptation was advantageous in terms of diminishing the increase in heart size, which often occurs as a necessary compensatory response in patients exposed to long-term, fixed-rate pacing [55].

Although sensor-based, rate-adaptive pacing initially focused attention on the importance of an adequate chronotropic response, the merits of maintaining a normal atrioventricular contraction sequence (atrioventricular synchrony) during rate-adaptive pacing resurfaced in the late 1980s when dual-chamber pacemakers with both atrial tracking and sensor-triggered pacing capability (DDDR mode shown in Table 1 became available. In general, although exercise capacities in DDDR and VVIR pacing modes are often similar [16, 56], a crucial difference between these two pacing modes is the maintenance of atrioventricular synchrony to prevent atrial tachycardias (particularly atrial fibrillation), to reduce susceptibility to congestive heart failure [18-26], and to provide a measurable improvement in quality of life [57, 58].

Recently, there has been special interest in identifying clinically relevant differences between dual-chamber pacemakers that rely solely on tracking the native atrium to adjust pacing rate (VD and DDD pacing modes) and those that also offer sensor-based rate-adaptation (VDR and DDDR pacing modes). Pacemakers that rely on atrial tracking alone for rate adaptation are appropriate for those patients in whom intrinsic sinoatrial function is normal and unaffected by concomitant drug treatment. On the other hand, DDDR devices (which use sensors and atrial sensing and tracking) are more appropriate in patients with evident sinoatrial disease or in persons in whom concomitant drug therapy (such as ß-adrenergic blockers, calcium channel blockers, certain antiarrhythmic drugs, and some antidepressant agents [59]) is likely to affect native atrial rate responsiveness.

Basic Principles of Sensor Combinations

Rationale for Multiple Artificial Sensors

In persons in good health, control of important cardiovascular variables such as heart rate is multifactorial [59]. Therefore, it should not be surprising that attempts to replicate physiologic heart rate responsiveness in cardiac pacemakers will similarly require more than one sensor.

Artificial sensor combinations designed to optimize pacemaker rate adaptation should be selected in such a manner that the resulting pacing system 1) responds only when circumstances warrant a heart rate change with a high level of immunity from inappropriate signals, 2) initiates a prompt and appropriate heart rate response when called on, 3) maintains a rate proportional to workload throughout exertion, and 4) provides adequate heart rate support during the patient's recovery or "cooling-off" period after exercise is complete [60]. In the future, it may be possible to devise sensor combinations that also can initiate heart rate adjustments in response to nonphysical stresses, such as emotional upset or the anticipation of exercise. Finally, certain sensor combinations (pressure and oxygen saturation sensors, for example) may prove useful to monitor treatment progress in patients with congestive heart failure or chronic pulmonary diseases and also to provide appropriate chronotropic responsiveness.

In addition to adjusting heart rate, artificial sensors in cardiac pacing systems have many other potential applications. However, no optimal universal sensor exists. Consequently, using sensors that have different attributes but that work in a complementary manner offers distinct advantages. For example, a potentially important sensor application is to ensure that the pacemaker stimulus indeed captures the heart muscle (capture detection), particularly in patients who are dependent on their pacemaker for every heart beat. In this situation, a sensor is needed that searches for a T wave (that is, stim-T detector), a right ventricular pressure pulse, or a paced depolarization response after each pacing stimulus. Failure to elicit a paced beat would automatically trigger the pacemaker to increase its energy output on future cycles until heart muscle capture was restored. Unfortunately, sensors operating alone may not provide optimal heart rate responsiveness. Thus, more than one sensor is needed to offer the most desirable outcome.

Potential Limitations of Multiple-Sensor Systems

In principle, artificial sensor combinations, such as some of the examples noted previously, will improve the physiologic precision and extend the capabilities of current sensor-based pacing systems. However, complicated sensor interactions may occur. For example, although the minute ventilation sensor is relatively effective in establishing pacing rates that parallel metabolic requirements during sustained exertion, its plethysmographic method is susceptible to interference by upper arm and chest movement [61]. Thus, in certain circumstances, matching minute ventilation with a piezoelectric activity sensor may not yield two entirely independent assessments of an appropriate sensor-indicated rate (that is, both sensors may, to some extent, respond to physical movement). A similar interaction may be expected with other plethysmographic sensors (such as pre-ejection interval measurement and respiratory rate detection) when paired with vibratory or accelerometer sensors.

Apart from the possibility of complex interactions when multiple artificial sensors are used together, other limitations to such systems are important. First, increased battery drain may be inevitable with many sensor combinations. On the other hand, the potential for sensor systems to adjust pacing output energy appropriately (such as by periodic assessment of pacing threshold, as described previously) may permit a substantial net energy savings. This beneficial outcome is very likely because pacemakers are often programmed to unnecessarily high outputs. Second, designing a trouble-free algorithm to assign sensor priority among competing sensor inputs may prove difficult. Third, until pacemakers operate entirely automatically, the complexity of programming and follow-up may dampen many physicians' enthusiasm for taking full advantage of multisensor devices. This may be particularly true for physicians who do not make cardiac pacing an important part of their practice. Finally, increased costs compared with current pulse generators (especially for systems requiring unique lead designs such as the oxygen saturation and right ventricular pressure sensors) may decrease the appeal of multiple-sensor-based pacemakers. On the other hand, many multiple-sensor systems do not require expensive additional technology (such as those incorporating an activity sensor with either a minute ventilation sensor or stim-T detector). Furthermore, important cost savings may accrue if pulse generator longevity is enhanced and the frequency of clinic visits is reduced by automatic programming. The latter is particularly relevant because at the present time most implanted pacemakers are not appropriately reprogrammed during follow-up to take advantage of existing energy-saving capabilities (optimizing stimulus voltage and pulse width may increase useful device longevity by 25% or more, for example).

Benefits of Multiple Artificial Sensor Pacing Systems

The potential benefits of pacemakers using multiple complementary artificial sensors are numerous. Nonetheless, demonstrating an even more physiologic heart rate response with these systems than is possible with single-sensor systems is essential.

Maximal Cardiopulmonary Exercise Testing

In general, symptom-limited maximal exercise testing (treadmill or bicycle ergometer) has been the standard means to evaluate physiologic response during exertion [4, 5, 8, 9, 11-13, 27-30]. By extension, this approach has also been favored to assess chronotropic competence in humans [62, 63] and the effectiveness of current rate-adaptive pacemakers [11-14, 30, 56, 63]. During such testing, assessment of peak oxygen consumption provides an index of the maximum amount of work the cardiopulmonary system can sustain in a given person. Physical training can increase this value but only to a limited degree [63]. The anaerobic threshold indicates the point during sustained exertion when anaerobic metabolism is required to maintain the workload.

Typically, peak observed oxygen consumption (peak VO₂) and oxygen consumption at anaerobic threshold (AT-VO₂) have been preferred to exercise duration for assessing the physiologic benefits of cardiac pacing systems; both are relatively immune to investigator bias. However, even these sophisticated measurements may not be sufficiently sensitive to detect certain subtle but potentially important differences among pacing systems. For instance, peak VO₂ may not differ substantially between optimally programmed rate-adaptive pacemakers, although peak VO₂ may be achieved promptly in one case and slowly in another. Oxygen consumption at anaerobic threshold may be more sensitive, but this is still unproven.

The utility of maximal exercise testing to evaluate the physiologic appropriateness of rate-adaptive pacing system operation has become an increasing concern. Apart from arguments related to the nature (that is, treadmill rather than bicycle exercise), expense, and inconvenience of such testing, its clinical relevance may be questioned because most patients with pacemakers only rarely approach maximal levels of exertion. The same concerns apply to other techniques that validate a physiologic pacing response, such as assessment of heart rate:VO₂ and minute ventilation (VE):VO₂ relations [64-66]. Consequently, testing techniques that focus on modeling the more typical transient workload activities engaged in by most persons with pacemakers (for example, climbing stairs or carrying parcels) have become increasingly important.

Oxygen Kinetics Analysis during Submaximal Subanaerobic Exercise Testing

Analysis of oxygen uptake kinetics during periods of subanaerobic physical exercise may help to define the physiologic benefits of artificial sensor-based pacing systems in an environment that reasonably models daily activity (exertion at levels <5 metabolic equivalents) [67-71]. Specifically, such testing is directed toward assessing metabolic response at the initiation of and recovery from a brief period of submaximal constant workload (similar to the kind experienced by most patients every day). Measurement of the initial oxygen_deficit accrued at the onset of exertion and the time taken to approach an oxygen_consumption equilibrium (mean response time) during exercise provides indices that depict the efficiency with which the cardiopulmonary system, and specifically the chronotropic response of the pacemaker, adjust to changing physical demands (Figure 2). Comparable measurements may be made during postexercise recovery to determine whether the transition from work to rest occurs in an appropriately physiologic manner. Values in healthy, age-matched and sex-matched control participants provide the ultimate gold standard.

View larger version (28K): [in this window] [in a new window]   Figure 2. Oxygen kinetics analysis during abrupt-onset, fixed- workload (subanaerobic) exercise may be useful to distinguish subtle differences in sensor-system operation or programming. This type of exercise protocol mimics the transient activities of daily living and therefore may be useful to assess the clinical utility of new pacing systems and their programming. Ultimate peak oxygen consumption levels (VO2, mL/min, abscissa) areidentical in the two panels. However, the times taken to achieve these responses differ. In panel A, VO2 equilibrium is delayed because of a relatively slow chronotropic response. Panel B depicts a more physiologically desirable outcome, with a smaller oxygen deficit and a shorter mean response time (MRT). The region of oxygen deficit at onset of exercise is noted by the shaded area in each panel.

Standardized Activities-of-Daily-Living Protocol

Studies in healthy participants suggest that persons similar in age to patients with pacemakers have frequent heart rate excursions exceeding 100 beats per minute every day [63]. Optimally, physiologic cardiac pacing would produce similar heart rate profiles. However, because the establishment of normal values for all circumstances is unrealistic, assessment of device operation during standardized activities of daily living was proposed. Recently, we analyzed heart rate behavior in 20 healthy persons (34 to 62 years old) having repeated sets of seven representative activities of daily living activities, such as walking, walking with a load, descending stairs, stair climbing, showering (mimicked), vacuuming, and household chores [72]. Each event lasted 1.5 to 2 minutes, with an intervening rest period. The findings provided target heart rate ranges useful for determining whether pacemaker programming is appropriate for comparable daily activities. Further, using this technique, preliminary findings from our laboratory using a pacemaker that combines both activity and minute ventilation sensors (Legend Plus; Medtronic, Inc.) confirmed that conventional techniques to optimize device programming (treadmill exercise, walking in place) failed to account for broad differences in the nature of various common activities. Based on this experience, it seems likely that activities-of-daily-living testing may be essential to confirm the physiologic behavior of sophisticated new sensor systems and to document the incremental clinical benefit compared with current pacemakers [73].

Clinical Experience with Pacing Systems Incorporating Multiple Artificial Sensors

Initial attempts to use multiple artificial sensors for cardiac pacing applications produced several experimental pacing systems [74-78]. For example, Alt and colleagues [75] coupled temporary right ventricular temperature sensing with a piezoelectric accelerometer positioned on the chest wall in an external investigational system in which the algorithm was controlled by a personal computer [75]. The operation was designed so that the initial rate response would be governed by the accelerometer because it was the more rapidly responding of the two sensors. However, if the characteristic right ventricular (central body core) temperature changes associated with exercise did not subsequently materialize, thereby "confirming" the occurrence of physical exertion, the pacing rate returned to baseline. Thus, not only were two artificial sensors used but a form of sensor cross-checking was also incorporated. At about the same time, an even more complex experimental multisensor system was devised and investigated by Stangl and associates [76]. In this case, the investigators used a catheter with multiple-sensor capabilities (stroke volume, central venous oxygen saturation and temperature, right atrial and ventricular pressure, rate of right ventricular pressure change, and respiratory rate). The system, although too complex for clinical application, provided useful insight into the relative merits of these indices as sensors. Briefly, the oxygen saturation sensor responded more quickly than did any of the other sensors in this system and remained proportional to workload throughout low-level exercise. Temperature sensing responded more slowly than oxygen saturation but nonetheless was workload proportional, particularly at levels exceeding 75 watts.

Based on investigations such as these, the current strategy for multiple-sensor pacing systems has been to pair a relatively rapidly responding sensor (typically a piezoelectric "activity" sensor) with a slower but more workload-proportional detector, such as a stim-T interval or minute ventilation sensor [77, 79-81] (Figure 3). A few such systems are commercially available, and others are being investigated clinically. As of mid-1994, none have been approved by the Food and Drug Administration for applications other than clinical investigational use in the United States.

View larger version (18K): [in this window] [in a new window]   Figure 3. Two sensors, one relatively fast responder and one slower but more workload proportional, can work together to elicit a more physiologic heart rate response than either would if used alone. The ordinate indicates pacing rate. The abscissa shows the duration of exercise. The individual sensor responses and the combined response are shown.

Stim-T Interval and Piezoelectric Activity Sensor Combination

Topaz (Vitatron Medical) was the first commercially marketed, dual artificial sensor, rate-adaptive pulse generator. This single-chamber device and its more recent dual-chamber version (Diamond) use two sensors that are clinically well established: a conventional piezoelectric vibration-type activity sensor and a stim-T detector. The physician programs conventional lower and upper pacing rate settings, the sensitivity of the unit for detecting T waves, and the activity thresholds (an index of the aggressiveness with which the rate-adaptive capability will operate). In addition, the physician also weights (based on clinical judgment for each patient) each of the sensors, with three choices (that is, sensor "blending") available: QT <Act, QT = Act, or QT > Act. After implantation, the device gradually and automatically self-regulates its operation within these guidelines.

Apart from the blending feature, Topaz and Diamond also use the concept of sensor cross-checking. The cross-checking feature operates such that the device tends to default to the stim-T sensor if an intersensor discrepancy is detected. Thus, an abrupt increase in physical activity detected by the piezoelectric vibration sensor, but not subsequently confirmed by the stim-T sensor, will reverse any initial pacing rate increase (that is, a suspected false-positive response). On the other hand, if stim-T measurement indicates a rate increase but the piezoelectric sensor does not, then a limited heart rate increase will be permitted based on the assumption that a nonexercise stress (such as emotional upset or pain) has been experienced. Sustained activity, detected by both sensors, is handled as determined by the physician's blending choice, as described previously.

Clinical trials of Topaz were done in Europe between August 1991 and January 1992. Findings suggest that rate profiles provided by the sensor combination reasonably mimic healthy sinus node function. Extraneous signals that might cause a false-positive sensor response appear to be distinguished satisfactorily. In regard to device programming, Rickards [81] summarized results of a multicenter experience. Briefly, adequate stim-T sensing was achieved in 77 of 79 (97%) patients, and appropriate activity sensor operation was confirmed in all (100%). At 1 month follow-up, QT = Act blending predominated (41% of patients [32 of 79]), with QT < Act required in 24% of patients (19 of 79) when the stim-T response was too slow.

Minute Ventilation and Piezoelectric Activity Sensor Combination

Legend Plus is a single-chamber, rate-adaptive pulse generator that can estimate the patient's minute ventilation (by transthoracic plethysmography) and can also detect conventional piezoelectric vibratory activity. As currently configured, the device can be programmed to operate using either the minute ventilation or activity sensors alone, or with both sensors operating in combination. Generally, when both sensors are functioning, the higher sensor-indicated rate predominates. Although there is no blending feature, there are programmable separate upper rate limits for each sensor, as well as individual sensitivity settings. Consequently, the physician has considerable flexibility in manipulating sensor interaction.

Recently, in a preliminary report derived from the multicenter Legend Plus clinical investigation in North America [71], we assessed various sensor combinations in 14 chronotropically incompetent patients who had maximal cardiopulmonary exercise testing and a fixed subanaerobic workload protocol during pacing in each of three sensor configurations. Maximum paced rate and peak observed oxygen consumption did not differ substantially for the three sensor configurations (that is, activity alone, minute ventilation alone, combined). On the other hand, both the oxygen_deficit accrued by patients at the beginning of exercise (541 ±299 mL with the activity sensor "on" compared with 737 ±381 mL with the minute ventilation sensor alone) and time taken to reach oxygen consumption equilibrium (mean response time) were substantially reduced when the activity sensor was active (Figure 2). The latter observations reflect important differences in sensor responses to abrupt-onset, sustained physical activity, and they highlight the value of teaming a rapidly responding sensor with a more workload-proportional sensor.

Studies of activities of daily living have also been used to assess Legend Plus programming. In our experience, this technique has proved useful to identify false-positive sensor responses (apparently related to upper body movement) that may not be detected on treadmill exercise. Further, after optimizing programming for both sensors, the activities-of-daily-living technique has been helpful to show appropriate heart rate responses during submaximal workloads compared with previously published normal control values [73].

Minute Ventilation and Paced Depolarization Integral Sensor Combination

Sentri (Telectronics Inc.) uses a minute ventilation sensor in conjunction with measurement of the paced depolarization integral (see the previous discussion of individual sensors). In its current configuration, each sensor determines what it "believes" to be an appropriate rate (the sensor-indicated rate). For pacing rates slower than 100 beats per minute, the higher of the two sensor-indicated rates is chosen, whereas for pacing rates faster than 100 beats per minute, the minute ventilation sensor predominates. This device is still being evaluated, and clinical data are not yet readily available.

Venous Oxygen Saturation and Piezoelectric Activity Sensor Combination

The combination of an intravascular right ventricular oxygen sensor with a piezoelectric vibration sensor is in the initial stages of investigation [82]. This complementary combination matches a highly physiologic, workload-proportional sensor (oxygen saturation) with the more rapidly responding activity system. Only a few such devices (Oxyelite; Medtronic Inc.) have been implanted, and clinical outcomes are not yet reported.

Conclusions

Rate-adaptive cardiac pacemakers using single artificial sensors are now a well-established part of pacing therapy. Their acceptance by physicians is clear from the steady increase in the frequency of their use in many countries [83-85], and their utility is supported by numerous clinical studies. Further, we now know that the potential utility of sensor systems extends beyond their original rate-adaptive function. However, no single sensor has proved optimal for all potential pacing system applications. Consequently, development of pacing systems using multiple sensors working in concert is the center of considerable clinical study. By coordinating activities of multiple sensors with diverse attributes and operating characteristics, the fully self-monitoring and self-adjusting pacemaker may evolve.

Presented in part at the scientific sessions of the Asian Pacific Society of Pacing and Electrophysiology, Tokyo, Japan, August 1993.