History and Rationale for Different Pacing Modes

Cardiac pacing has been developed to treat syncope caused by bradycardia and asystole, initially in patients with permanent third-degree atrioventricular (AV) block. For this purpose, a single lead implanted in the ventricle and pacing at a prespecified rate was sufficient. This simple method of pacing (later coded “VOO 70”) was able to demonstrate a mortality reduction in patients with AV block despite that pacemaker implantation required open-chest surgery at that time.¹ Cardiac pacing in permanent third-degree AV block was found not only to prevent syncope, but also to improve symptoms, exercise capacity, and prognosis. It was therefore also used for other patients with bradycardia, such as from intermittent high-degree AV block and sinus node disease (SND).

With technical advancement, the limitations of fixed-rate ventricular pacing were overcome by the development of sensors to provide rate-adaptive (R) ventricular pacing. Rate-adaptive pacing was used to improve the hemodynamic benefit of ventricular single-chamber pacing by providing ventricular rates that take into account the metabolic needs of the patient. In patients with SND, however, ventricular single-chamber pacing in VVI and VVI(R) modes did not reduce mortality² and could even cause symptoms that were later termed pacemaker syndrome.³ To treat SND, atrial single-chamber pacing has been used, particularly in the Scandinavian countries,⁴ although many physicians remained skeptical about atrial pacing because it could not prevent asystole if AV block occurred.⁵ However, a number of advantages of atrial versus ventricular single-chamber pacing were observed, including the absence of pacemaker syndrome, better cardiac output and exercise capacity, resolution of mitral regurgitation that could be induced by ventricular pacing, and a lower incidence of atrial fibrillation.

Therefore, alternative pacing modes were desired, particularly for patients with SND. The development of dual-chamber pacemakers created hope that DDD pacing would restore and maintain normal AV synchrony and thus provide “physiologic” pacing. It was anticipated that this closer approximation of normal physiology would reduce patient morbidity and mortality and improve quality of life (QOL). After the introduction of dual-chamber pacing, numerous case reports and small studies showed how dual-chamber (physiologic) pacing resolves mitral regurgitation, is associated with lower plasma epinephrine levels, improves exercise tolerance, resolves VVI pacing–associated heart failure, and has other beneficial effects.

However, the theoretical advantages of dual-chamber pacing (maintenance of AV synchrony, preservation of sinus node control over heart rate in patients with AV block, maintenance of normal ventricular activation via His-Purkinje system in SND patients) required confirmation in prospective clinical studies given their higher cost, greater complexity, and potential for complications. This was the rationale for trials comparing pacing modes, as outlined in this chapter.

Currently, recommendations in international guidelines require confirmation by large, randomized controlled trials (RCTs). However, most RCTs have yielded unexpected results, particularly because the theoretical advantages of dual-chamber pacing have never been clearly demonstrated. All these trials have some results that require further explanation. From these experiences, physicians should consider a number of lessons in the daily practice of implanting and following pacemakers. This chapter presents only a selection of the most important or most representative studies comparing different pacing modes and modalities.

Comparison Studies

Ventricular Pacing with Versus Without Rate Response (VVI vs. VVIR)

Crossover studies in the 1980s and early 1990s randomized small numbers of patients to fixed-rate ventricular single-chamber pacing (VVI) versus rate-adaptive ventricular pacing (VVIR) with the use of different sensors^6–12 (Table 10-1). These studies unequivocally favor rate-adaptive pacing and showed an improvement in submaximal and maximal exercise in terms of objective data (e.g., exercise duration, maximal exercise capacity in watts or duration, VO₂ at anaerobic threshold) or subjective data (e.g., pacing mode preference, symptom scores/questionnaires) compared with fixed-rate VVI pacing. This advantage was still present in studies that included patients with SND or retrograde AV conduction. The most striking advantage of VVIR over VVI pacing was observed in patients with AV node ablation for atrial fibrillation.

TABLE 10-1 Studies Comparing VVI and VVIR Pacing*

However, these studies were small and not blinded. Therefore, a significant bias in favor of rate-adaptive pacing cannot be excluded. Many patients in these studies were younger than the average age of pacemaker patients. In elderly patients with a sedentary lifestyle and patients with severe heart disease that precludes any exercise above minimal levels (two large groups of pacemaker patients), there is likely no difference between VVI and VVIR pacing. In some patients, VVIR pacing resulted in an increase in angina pectoris. The improvement in exercise capacity in VVIR pacing is based on an increase in heart rate, which increases myocardial oxygen consumption with potentially disadvantageous consequences in ischemic heart disease. In addition, whenever a patient has an intrinsic ventricular rhythm at an adequate rate, this is preferred to avoid deleterious effects of ventricular desynchronization by right ventricular pacing (see later). Lastly, the use of rate-adaptive pacing requires conservative programming of parameters related to the pacing rate, because “overpacing” (i.e., pacing rate faster than needed) is more symptomatic than “underpacing.”

Ventricular Pacing with Rate Response Versus Dual-Chamber Pacing (VVIR vs. VDD/DDD/DDDR)

With the advent of dual-chamber pacing, a number of trials tested if rate-adaptive ventricular pacing (VVIR) with an artificial sensor driving the heart rate was comparable to dual-chamber pacing, in which the sinus node determines the heart rate. Investigators observed that sensors frequently reacted inappropriately and were therefore inferior to a chronotropically competent sinus node to increase the heart rate linearly according to metabolic needs during exercise.¹³ Most studies showed a superiority of dual-chamber pacing in terms of exercise capacity, symptoms, and hemodynamic parameters^13–20 (Table 10-2).

TABLE 10-2 Studies Comparing VVIR and DDD(R)/AAI(R)

Of note, when patient preference of pacing mode was assessed, there was a clear advantage for AV-synchronous pacing modes; this can be ascertained only in crossover trials. All the randomized large-scale studies comparing single-chamber ventricular and dual-chamber pacing had a parallel trial design and therefore lack this information (apart from crossover patients, who showed a clear preference for dual-chamber pacing mode in PASE and MOST).

Ventricular Single-Chamber Pacing Versus Atrial Synchronized Ventricular Pacing (VVI vs. VDD)

The question if dual-chamber pacing with ventricular pacing synchronized to atrial activity was superior to single-chamber ventricular pacing was evaluated early. In 16 patients with high-degree AV block, Kruse and Rydén²¹ found a significant increase in maximal exercise capacity in VDD compared to VVI pacing. Interestingly, the extent of improvement was the same for patients older and younger than 65 years. At comparable exercise levels, the atrial rate was significantly lower in VDD than VVI pacing. In a crossover design, 13 patients underwent invasive hemodynamic measurements after 3 months of VDD and VVI pacing.²² During VDD pacing, cardiac output was significantly higher (>30%), particularly during exercise, because of the heart rate increase in VDD and despite a stroke volume increase in VVI pacing. Arteriovenous oxygen difference was significantly higher during VVI pacing. Arterial blood lactate was significantly higher in VVI than VDD pacing during exercise. Heart size was significantly smaller with VDD pacing. In a questionnaire on symptoms and pacemaker choice, patients preferred the VDD mode.

These early observations favoring dual-chamber pacing significantly in all respects over VVI pacing were an important part of initiating the strong belief in AV synchronous dual-chamber pacing as the “physiologic” and preferred pacing mode.

Single-Lead VDD Pacing Versus Dual-Chamber DDD Pacing in Atrioventricular Block

The VDD pacing mode offers the advantage of using only one lead, typically tripolar or quadripolar, with two floating rings that sense atrial activity.²³ This pacing mode may be similar to DDD pacing with two leads in patients with AV block and normal sinus node function. However, no RCTs have been performed comparing VDD and DDD pacing. A retrospective study analyzed 1214 consecutive patients with pacemaker implantation for AV block (36.5% VVI; 32.9% DDD; 30.6% VDD).²⁴ At implantation, operation and fluoroscopic times were longer in DDD than in VDD and VVI systems. During follow-up of 5 years, complications requiring surgical interventions occurred 2.3 times more frequently for DDD systems than VDD systems, mainly because of a higher incidence of early atrial lead dysfunction.

Dual-Chamber Pacing in Chronotropic Incompetence or Paroxysmal Atrial Fibrillation: DDD(R) with Mode Switching

There are few studies prospectively comparing different pacing modes within dual-chamber pacing, such as triggered versus inhibited mode.

Sulke et al.²⁵ compared DDDR, DDD, DDIR, and VVIR pacing modes in 22 patients with high-degree AV block and chronotropic incompetence. In a randomized double-blind crossover design, patients were paced in each of these modes for 4 weeks. Patients were asked their preference, received three questionnaires for subjective assessment, and underwent treadmill testing, mental stress testing, testing of everyday activities (e.g., suitcase lifting, staircase climbing), and echocardiographic evaluation. Scores for subjective grading of well-being were highest for DDDR pacing, similar for DDD and DDIR modes, and lowest for VVIR pacing. DDDR pacing was the preferred mode in 59% of patients, VVIR mode was the least acceptable mode for 73% of patients. Treadmill testing was best in DDDR mode; nontracking modes underresponded to mental stress; and all rate-adaptive modes overreacted to staircase descent. Cardiac output was best in DDDR mode and non-tracking modes showed varying values in patients with AV block. This study suggested that rate-responsive dual-chamber tracking mode is superior to nontracking dual-chamber pacing, dual-chamber pacing without rate-response, and single-chamber ventricular pacing in patients with AV block and chronotropic incompetence, based on both objective and subjective measurements.

Another study compared different pacing modes in 48 patients with AV block and paroxysmal atrial tachyarrhythmias.²⁶ Patients received dual-chamber systems programmed to DDIR mode for 30 days. In a randomized double-blind crossover manner, devices were then programmed for 4 weeks each to DDDR with mode switching, DDDR with limitation of the upper tracking rate, and VVIR. Patients received transtelephonic self-activated devices to transmit electrocardiograms whenever they perceived symptoms. In addition, at the end of each 4-week study period, devices were interrogated; patients underwent 24-hour Holter recordings, received three symptom questionnaires, and at the end of the study were asked for their preferred pacing period. Patients with intolerable symptoms were allowed to cross over early to the next pacing mode. Different devices with fast- and slow-reacting mode-switching algorithms were used. In this study, one third of patients requested early crossover from VVIR to another mode because of intolerable symptoms, most likely pacemaker syndrome. In DDDR pacing without mode switching but limited upper tracking rate, 19% of patients requested early crossover (29% of patients with documented atrial tachyarrhythmias during this period). Only one patient with DDDR and mode switching requested early crossover because of inappropriate tracking of atrial tachyarrhythmia. Patients with fast-reacting mode-switching algorithms in particular preferred DDDR over VVIR mode. Symptom questionnaires favored DDDR with mode switching over DDDR with limited upper racking rates, DDIR, and VVIR pacing. The authors conclude that dual-chamber pacing with tracking and mode switching to a nontracking mode, whenever this automatic function is fast and reliable, is the preferred pacing mode for patients with AV block and paroxysmal atrial tachyarrhythmias (Fig. 10-1).

Figure 10-1 Patient preference of pacing mode.

A, Overall patient preference. B, Preference in patients without atrial tachyarrhythmias during the study period.

Dual-Chamber Pacing with Versus Without Rate Response (DDD vs. DDDR)

The need for a sensor in patients with chronotropic incompetence has received little attention after dual-chamber pacing became available and resolved this problem for patients with AV block and normal sinus function. Most studies subsequently compared different sensors.^27,28 Additionally, the definition of “chronotropic incompetence” varied widely.

An early study evaluated cardiopulmonary exercise testing with the DDDR versus DDD pacing modes in eight patients.²⁹ Unfortunately, this was a mix of patients, five with AV block and three with sinus node disease, and five with and three without chronotropic incompetence. Chronotropic incompetence was defined as a maximum heart rate at exercise before pacemaker implantation of less than 110 beats per minute (bpm), but not whether it was caused by AV block (resolved with DDD pacing alone) or sinus node dysfunction (requires rate-adaptive pacing). In a crossover study, patients were randomized to DDD or DDDR, each for 3 weeks, and performed an exercise test and completed a symptom evaluation form at the end of each study period. Rate-adaptive pacing was associated with a higher oxygen uptake at the anaerobic threshold and maximum exercise. However, exercise time and subjective symptoms were not different in the whole patient group, only in patients with chronotropic incompetence, and symptoms of exercise intolerance were better with DDDR pacing.

In a study in 63 patients with a DDD pacemaker, a symptom-limited exercise test was performed 1 month after implantation.³⁰ With chronotropic (sinus node) incompetence defined as a maximum heart rate of less than 60% of the age-predicted maximum or less than 100 bpm, 25 patients were then randomized to perform exercise testing in DDD or DDDR mode on two consecutive days. To minimize bias, patients and physicians were blinded with respect to the pacing mode. Paired analysis revealed a statistically significant improvement in the DDDR mode, with higher heart rates at maximum exercise (113 vs. 84 bpm), longer duration of total exercise and time to anaerobic threshold, and higher oxygen uptake at maximum exercise and anaerobic threshold.

Only one large RCT performed according to the principles of evidence-based medicine (EBM) specifically assessed the benefit of rate-adaptive pacing in the dual-chamber mode in patients with sinus rhythm.³¹ Unfortunately, the single-blind Advanced Elements of Pacing Randomized Controlled Trial (ADEPT) included a mix of patients with SND (~66%) and AV block. After an exercise test 1 month after implantation, 872 patients were enrolled. To be eligible, patients needed to complete at least stage 2 on the Chronotropic Assessment Exercise Protocol and to have a heart rate at peak exercise of 80% or greater of the age-predicted maximum. The mean age was 71 years, however, so patients with a maximum heart rate not exceeding 120 bpm at maximum exercise could be included. Patients were randomized to DDD versus DDDR pacing (dual-sensor system; maximum adaptive rate, 220 bpm minus age) with rate-adaptive AV delay shortening in a parallel study design. After 12 months, general QOL (measured by Short-Form 36 General Health Survey [SF-36], Ferrans and Powers Quality of Life Index) and cardiovascular functional status (Specific Activity Scale and Duke Activity Status Index) were assessed by questionnaires, together with the maximum exercise time at treadmill testing 6 months after randomization. ADEPT found no specific or general differences in scores of the activity and QOL questionnaires. The total exercise time was not different between DDD and DDDR pacing. Independent of discussions about the ideal sensor(s), severe limitations included the following:

• The definition of “chronotropic incompetence” was too broad to demonstrate an improvement by rate-adaptive pacing. Previous studies suggest that rate-adaptive pacing can improve exercise capacity only in patients with a maximum sinus rate less than 90 to 100 bpm.^32,33 Although patients with a maximum heart rate of 80% or less of the age-predicted maximum have by definition chronotropic incompetence, this is too mild to be improved by rate-adaptive pacing.

• Patients with AV block causing limitation of heart rate increase at maximum exercise (one third of patients enrolled in ADEPT) will substantially benefit from DDD pacing alone. If sinus node function is normal, no further improvement can reasonably be expected from the addition of rate-adaptive pacing.

• Because of the negative consequences of right ventricular pacing, programming the AV delay to 150 to 200 msec in patients with SND, particularly activating a rate-adaptive AV delay–shortening algorithm, may have severely biased the results and limited exercise capacity, particularly during DDDR pacing.

 Atrial Versus Ventricular Single-Chamber Pacing in Sinus Node Disease

In 1994, Andersen et al.³⁴ published the first RCT on pacing mode. A total of 225 patients with SND were randomized to receive a single-chamber pacemaker in the right atrium or ventricle. Of note, the study excluded patients with AV block of any degree (including first-degree AV block with PQ interval >220/260 msec), Wenckebach block at less than 100 bpm, atrial fibrillation (AF) more than 50% of the time (or with pauses >3 seconds or ventricular rates <40 bpm), or bifascicular bundle branch block. In total, 827 of 1052 screened patients were excluded from study participation, predominantly because of AV block (n = 552). Study outcome parameters were mortality, AF, thromboembolism (TE), and heart failure (HF). From retrospective studies, the authors expected AF in 30%, TE in 15%, and mortality in 15% of patients in the ventricular pacing group. The study intent was to demonstrate a 50% reduction in AF and HF and a 66% reduction in TE and mortality with atrial single-chamber pacing, which would have required 100 patients in each group. This rather optimistic prediction of the capabilities of physiologic pacing reflects the widespread belief in its superiority over ventricular pacing at the time of conception of this study.

During a follow-up of 3.3 years, all-cause mortality was similar (22% with ventricular vs. 19% with atrial pacing). There was no significant difference in the risk of AF (in the ECG of at least one follow-up visit in 23% of patients with ventricular vs. 14% with atrial pacing). However, TE was significantly more frequent with ventricular pacing (17% vs. 5%; P = .008). At the same time, ventricular pacing was associated with a significant increase in left atrial diameter compared to baseline and to patients randomized to atrial pacing.

Andersen et al.³⁵ extended the follow-up until the last patient completed 5 years in the assigned pacing mode. By that time, the first patients enrolled into the trial had completed an 8-year follow-up (mean, 5.5 years). This prolonged follow-up demonstrated a significantly lower mortality in atrial versus ventricular pacing (relative risk reduction [RRR], 34%), mainly from a reduction in cardiovascular mortality (Table 10-3). Similarly, atrial fibrillation was less likely in the atrial pacing group (RRR, 46% vs. ventricular pacing) as were TE events (RRR, 53%). Only 4 of 110 patients developed AV block during the study period (annual rate, 0.6%), two of whom had a right bundle branch block at baseline. The authors detected retrograde conduction in 63% of patients randomized to VVI pacing.

TABLE 10-3 Clinical Events in Atrial versus Ventricular Pacing for Sick Sinus Syndrome (Danish Study)*

* Results from long-term follow-up.

† P < .05 for lower incidence respectively; better value with AAIR.

From Andersen HR, Nielsen JC, Thomsen PE, et al: Long-term follow-up of patients from a randomised trial of atrial versus ventricular pacing for sick-sinus syndrome. Lancet 350:1210-1216, 1997.

 Ventricular Versus Dual-Chamber Pacing

LV, Left ventricular; NYHA, New York Heart Association.

Randomized Studies

In the early 1980s, double-blind crossover studies prospectively compared VVI to dual-chamber pacing. Perrins et al.⁶² randomized 13 patients with AV block and a VDD pacemaker to dual-chamber or VVI mode. They found a better exercise capacity and reduction of symptoms during the period with dual-chamber pacing, as documented by patients’ diaries and monthly symptom scores. Similarly, Kristensson et al.⁶³ randomized 44 patients to VVI or VDD for 3 weeks for each mode. They observed a statistically significant improvement of maximal exercise tolerance of 14% with AV synchronous pacing. Arterial lactate, respiratory rates, and perceived exertion ratings during submaximal exercise were higher on ventricular inhibited pacing. Symptom scores during the 3-week periods with VVI pacing were worse than with dual-chamber pacing. Most patients had a better functional class during AV synchronous pacing and preferred this pacing mode.

Importantly, these two studies attempted to reduce any bias by double-blinding patients and the physicians who performed exercise testing. Also, in-patient comparison was done between ventricular and dual-chamber pacing modes, not available in later trials, which used a parallel trial design to avoid carryover effects.

In one of the first prospective randomized, intention-to-treat trials on pacing mode selection, Mattioli et al.⁶⁴ compared ventricular (VVI or VVIR) with physiologic pacing (AAI, DDD, DDDR, VDD) in 210 patients (SND in 110, AV block in 100) with a median age of 77 years. Only six patients received an AAI pacemaker. Patients with previously known AF were excluded from this study. A total of 56 patients (27%) developed AF during follow-up. The incidence of AF was 10% at 1 year, 23% at 3 years, and 31% at 5 years. The investigators observed an increased incidence of chronic AF in patients receiving ventricular pacing (P < .05). No difference was observed in the incidence of chronic AF between patients paced for AV block and patients paced for SND. In the subgroup of SND patients, the incidence of chronic AF was significantly less in patients receiving a physiologic than in those receiving a ventricular pacemaker (see Table 10-7).

Mode Selection Trial in Sinus Node Dysfunction (MOST)

The group of investigators of PASE followed their observation that differences in outcome parameters between VVIR and DDDR were (if present at all) more pronounced in sinus node disease, performing a pacing mode comparison in SND patients.⁶⁷ A total of 2010 patients were randomized to dual-chamber (1014) or ventricular pacing (996) and followed for 33 months. The incidence of the primary endpoint (all-cause mortality or nonfatal stroke) did not differ significantly (21.5% in DDDR, 23.0% in VVIR). Adjusted analyses revealed that in patients assigned to DDDR pacing, the risk of AF was 23% lower (21% vs. 27%) and HF scores were better. Heart failure hospitalization was reduced by 27%, the composite endpoint death, stroke, or HF hospitalization was reduced by 15% (Table 10-9). Dual-chamber pacing resulted in significantly better QOL for six of the eight SF-36 subscales during the 4 years of follow-up. The authors concluded that in SND patients, dual-chamber pacing does not increase stroke-free survival compared to ventricular single-chamber pacing, but DDDR significantly reduces AF risk and HF symptoms and improves QOL.

TABLE 10-9 Clinical Events in Mode Selection Trial in Sinus Node Dysfunction (MOST)

Clinical Event	VVIR (n = 996)	DDDR (n = 1014)
Death	23.0%	21.5%
Stroke	4.9%	4.0%
Heart failure hospitalization	12.3%	10.2%*
Atrial fibrillation	27.1%	21.4%*

* P < .05 for lower incidence with DDDR.

Data from Lamas GA, Lee KL, Sweeney MO, et al: Ventricular pacing or dual-chamber pacing for sinus-node dysfunction. N Engl J Med 346:1854-1862, 2002.

Critical Appraisal

The MOST trial had the ambitious aim to assess clinical events (mortality, stroke), AF incidence, and symptoms (HF, pacemaker syndrome, QOL) in VVIR versus DDDR pacing for SND. While the latter two endpoints favored dual-chamber pacing, the first and most important endpoint (mortality) did not demonstrate a significant difference. Therefore, the MOST trial was frequently regarded as a negative trial, showing no significant benefit of dual-chamber compared to ventricular single-chamber pacing. In-depth analyses of mortality⁶⁸ and stroke risk⁶⁹ showed no advantage of dual-chamber pacing, and QOL was better in dual-chamber mode in several items, particularly related to physical function.⁷⁰ However, there were two important lessons from MOST: (1) pacemaker syndrome is frequent in SND and VVIR pacing,⁷¹ and (2) unnecessary right ventricular pacing in SND is detrimental.³⁸

A strong argument in favor of dual-chamber pacing is the observation of severe pacemaker syndrome in approximately 20% of SND patients treated by VVIR pacing.⁷¹ The investigators in MOST attempted to prevent the major weakness of PASE, where the crossover rate from single- to dual-chamber pacing of 26% was so high that it may have prevented any meaningful intention-to-treat analysis of the difference between the two pacing modes. In MOST, therefore, the diagnosis of pacemaker syndrome strictly required documentation of new symptoms (e.g., dyspnea, syncope) together with VA conduction or systolic BP reduction of 20 mm Hg or more during VVI pacing compared to intrinsic rhythm. Even in the presence of pacemaker syndrome, crossover of pacing mode had to be confirmed by the clinical coordinating center; other attempts to resolve pacemaker syndrome (e.g., reprogram pacing rate, deactivate sensor) were required; and only if these failed, the centers could ask the principal investigator to review the indication for crossover to dual-chamber mode.

Despite these rigid requirements, 20% of patients programmed to VVIR met criteria for pacemaker syndrome and required crossover from single- to dual-chamber. Interestingly, pacemaker syndrome could basically develop at any time after implantation: At 6 months, it was present and required reprogramming in 14% of patients with VVIR pacing, in 16% at 1 year, in 18% at 2 years, and in 20% at 4 years. Slow sinus rates but not the presence or absence of VA conduction before pacemaker implantation predicted the development of pacemaker syndrome. QOL in patients with pacemaker syndrome was significantly worse than in patients without this complication of VVIR pacing, and it improved significantly after crossover to dual-chamber mode. Therefore, pacemaker syndrome with 20% incidence can be regarded as the most frequent complication of ventricular single-chamber pacing in SND, with significant impact on patients’ symptoms and QOL (Fig. 10-2).

Figure 10-2 Pacemaker syndrome in MOST.

Percentage of patients meeting criterion 1 or 2 or both criteria in Mode Selection Trial in Sinus Node Dysfunction.

A second post hoc analysis of MOST detected one of the basic problems of studies on mode selection and changed the practice of cardiac pacing. Sweeney et al.³⁸ found that the percentage of right ventricular (RV) pacing was strongly associated with the risk of developing HF and AF. This refers to patients with SND and narrow QRS at baseline. Interestingly, this association was more striking for dual-chamber than single-chamber pacing. From these results, hypothetically, a “badly timed” (i.e., too early) ventricular stimulus may be more detrimental in dual-chamber pacing, where the pacing mode forces adverse AV timing to be present in every cycle, versus single-chamber ventricular pacing, where (besides pacemaker syndrome with VA conduction) only some cycles have by chance a detrimental AV relation, whereas in most cardiac cycles, atrium and ventricle are dissociated.

The potential adverse effects of unnecessary RV pacing and inappropriate AV timing are problems not only in PASE and MOST, but in all trials comparing ventricular single- and dual-chamber pacing and may partly explain the lack of a positive effect of dual-chamber pacing in these studies.

Meta-Analysis of Ventricular Versus Atrial-Based Pacing

After UKPACE, Healey et al.⁸⁵ performed a meta-analysis of the large trials on pacing mode selection (Danish study,^34,35 PASE,⁶⁵ CTOPP,⁷² MOST,⁶⁷ UKPACE⁷⁹). As expected, no advantage of dual-chamber pacing was found for mortality, for the combination of mortality and stroke, or for heart failure hospitalization (Figs. 10-3 to 10-8). The incidence of atrial fibrillation was lower in dual-chamber pacing. However, predominantly because of UKPACE results, a subanalysis showed a lower incidence of AF only for dual-chamber pacing in sinus node disease, but not in AV block. Even though this meta-analysis included the impressive number of 7231 patients, results cannot overcome the general limitations of the individual trials, such as selection bias and biased detection of AF when an atrial lead is present.

Figure 10-3 Pooled estimate of mortality in randomized trials of pacing mode.

Danish, Atrial versus ventricular pacing in sick sinus syndrome (Andersen et al.^34,35); CI, Confidence interval; CTOPP, canadian Trial of Physiologic Pacing; HR, hazard ratio; MOST, Mode of Selection Trial in Sinus Node Dysfunction; PASE, Pacemaker Selection in the Elderly; UKPACE, United Kingdom Pacing and Cardiovascular Events trial.

Figure 10-4 Pooled estimate of cardiovascular mortality or stroke in randomized trials of pacing mode.

Figure 10-5 Pooled estimate of atrial fibrillation in randomized trials of pacing mode.

Figure 10-6 Pooled estimate of hospitalization for heart failure in randomized trials of pacing mode.

Figure 10-7 Cardiovascular mortality or stroke, based on indication for pacing.

A, Sinus node disease. B, Pacing indications other than sinus node disease.

Figure 10-8 Atrial fibrillation based on indication for pacing.

A, Sinus node disease. B, Pacing indications other than sinus node disease.

 Atrial Single-Chamber Versus Dual-Chamber Pacing: Clinical Trials

In 2004, Masumoto et al.⁸⁶ performed a retrospective analysis of 196 patients with sinus node disease who were implanted with dual-chamber devices programmed to DDD (n = 101) or AAI (n = 95) mode. Consecutive patients implanted between 1979 and 2000 were included if they had no second- or third-degree AV block, a maximum PR interval of 220 msec, and no left bundle branch block or bifascicular block. During a mean follow-up of 8 years, eight AAI devices had to be switched to DDD because high-degree AV block developed (incidence, 1.1%/yr). At 10 years, there was no need to add a ventricular lead in 89% of patients with AAI pacing. Excluding generator exchanges, 18 patients with DDD versus 17 with AAI system underwent reoperation, predominantly for lead failure. All-cause mortality was similar in both groups (10-year survival: 88% in AAI, 93% in DDD), as was freedom from permanent AF (94% in AAI, 91% in DDD) (Table 10-12). The authors concluded that AAI pacing can achieve similar clinical benefits as DDD pacing in sinus node disease with relevant AV or infrahisian conduction disturbances.

Nielsen et al.⁸⁷ compared myocardial blood flow and left ventricular ejection fraction (LVEF) in 30 patients with SND randomized to AAI versus DDD pacing. After 22 months of pacing in the assigned mode, myocardial blood flow was assessed with positron emission tomography (PET) and LVEF measured with echocardiography. Patients with dual-chamber pacemakers were temporarily programmed to AAI to obtain an intraindividual comparison. Myocardial blood flow was similar in the groups with AAI and DDD pacing. However, blood flow significantly improved if the patient’s DDD pacing was temporarily programmed to AAI mode. LVEF was similar to baseline results after 22 months in AAI but had become significantly lower in patients with DDD pacing, from 61% to 56% (P = .01). However, if DDD pacing was temporarily changed to AAI, measurements of LVEF returned to baseline results. The authors concluded that dual-chamber compared with single-chamber AAI pacing has detrimental effects on myocardial blood flow and LVEF, which are still reversible after 22 months.

In a single-center pilot study, consecutive patients with sinus node disease were screened and enrolled if they had no bundle branch block (BBB), and the spontaneous PQ interval did not exceed 220 msec (patients ≤70 years) or 260 msec (patients >70 years).⁸⁸ Of 952 patients with a first pacemaker implantation between 1994 and 1999, 755 were excluded from the study, mainly for AV block (48%), permanent AF (10%), or BBB (5%), and only 2% of patients did not give consent. The 177 enrolled patients were randomized to receive an AAIR pacing system (n = 54) or a DDDR system programmed to a short AV delay (150 msec, DDDR-s, n = 60) or a long AV delay (300 msec, DDDR-l, n = 63). Echocardiographic (left atrial diameter, left ventricular function and size) and clinical (AF, stroke or arterial embolism, mortality, heart failure) outcomes were assessed in these three modes of pacing in SND. During the study, patients in the DDDR-s group received pacing for a mean of 90%. Patients in the DDDR-l group still received 17% of (a priori unnecessary) ventricular pacing. After a mean follow-up of 3 years, patients randomized to AAIR did not show a significant change in left atrial diameter, left ventricular diameter or LVEF shortening compared to baseline. In contrast, left atrial diameter was increased and left ventricular shortening fraction decreased significantly in the DDDR-s patients. In the DDDR-l group, the left atrial diameter and the left ventricular end-systolic and end-diastolic diameters increased significantly during the 3-year follow-up. Atrial fibrillation in the 12-lead ECG during at least one follow-up visit was significantly less common in AAIR (7.4%) than in DDDR-s (23.3%) or DDDR-l (17.5%).⁸⁹ During the study, 16 strokes occurred in 14 patients, but no peripheral embolism was observed. Clinical events, including all-cause mortality, cardiovascular mortality, development of CHF, and increased use of diuretics, did not differ among the three groups. However, progression of HF symptoms, although not significantly different among the groups, was seen, with an increase of at least one NYHA class in 31% of patients paced in AAIR mode versus 30% in DDDR-s and 46% in DDDR-l.

The study by Nielsen et al.⁸⁷ would have required 450 patients for an 80% power to detect clinical or echocardiographic differences between AAIR and DDDR (no differences between DDDR-s and DDDR-l were expected). Inclusion was stopped when a multicenter trial (DANPACE) randomizing SND patients to AAIR versus DDDR pacing was initiated. However, the results of this study from Aarhus suggest that in patients with SND, narrow QRS and normal AV conduction dual-chamber pacing compared to AAIR pacing may result in deleterious effects in terms of left atrial enlargement, left ventricular volumes and ejection fraction, and AF incidence. Surprisingly, even in these select patients, programming the AV delay to very long values was unable to prevent a significant proportion of unnecessary right ventricular pacing. The observed proportion of ventricular pacing of 17% had detrimental effects in terms of left atrial dilatation and incidence of AF similar to DDDR pacing with short AV delay.

In a study following these observations, Albertsen et al.⁹⁰ randomized 50 patients with sinus node disease to AAIR or DDDR (AV delay 220 msec paced and 200 msec sensed). Primary endpoint was dyssynchrony (evaluated by tissue-Doppler imaging with at least 1 segment with delayed longitudinal contraction) and LVEF (evaluated by 3D echocardiography) after 12 months compared to baseline. Physicians performing echocardiography were blinded with regard to pacing mode. Secondary endpoints referred to 6-minute walking test and NT-proBNP. During follow-up, patients with DDDR pacing received ventricular pacing for 66% of the time. Dyssynchrony slightly increased with DDDR pacing (mean of 1.3 delayed segments at baseline vs. 2.1 after 12 months) but not in AAIR pacing. Although LVEF decreased statistically significant with DDDR pacing (63% at baseline vs. 59% after 12 months; no change observed in AAIR pacing), there was no difference between AAIR and DDDR pacing after 12 months in NYHA class or NT-proBNP. These results confirm other observations of deteriorated left ventricular function with unnecessary right ventricular pacing, and suggest the development of left ventricular dyssynchrony as one cause. At the same time, echocardiographic changes from baseline did not translate into significant clinical changes in this small study.

The Danish multicenter randomized trial on single-lead atrial versus dual-chamber pacing in patients with sick sinus syndrome (DANPACE) was started in 1999.⁹¹ It compared AAIR and DDDR pacing in patients with SND but with no BBB or AV block (except for PR interval ≤220 msec for patients <70 years and ≤260 msec for those ≥70 years). This trial used hardware randomization. Primary endpoint was all-cause mortality; secondary endpoints included detection of atrial fibrillation during scheduled follow-up, progression to permanent AF, incidence of stroke, heart failure, and the need for reoperation. It planned to include 1900 patients followed for a mean of 5.5 years to detect an absolute 6% difference in all-cause mortality with a power of 80%. The trial was terminated after 1415 patients, implanted in 20 centers in Denmark, United Kingdom, and Canada, were followed for a mean of 5.5 years. No patient was lost to follow-up. Patients in the DDDR group had been programmed to an AV delay of ≤220 msec paced and ≤200 msec sensed with rate-adaptive shortening. This caused ventricular pacing for a mean of 65% of the time in DANPACE. All-cause mortality (30% in AAIR, 27% in DDDR) was not significantly different (P = 0.53) nor was there a difference in stroke, NYHA functional class, use of diuretics, or hospitalization for heart failure.

To the surprise of the investigators, the incidence of paroxysmal AF was significantly lower in patients paced in the DDDR mode (RRR, 21%; P = .024). This difference appeared already within 6 months after randomization and was still observed in a multivariate analysis. However, permanent AF was similar in both pacing modes. Of clinical significance is the observation of a doubling of the need for a reoperation during follow-up in AAIR pacing (hazard ratio = 2.0 with AAIR; 95% confidence interval = 1.54-2.61; P < .001), mainly for the need of a ventricular lead. Only one patient in the DDDR pacing group had to be upgraded to a biventricular system. The authors concluded that DDDR systems programmed to an AV delay of ≤220 msec instead of AAIR systems should be used in sinus node disease.

The incidence of AF related to pacing mode is different in DANPACE compared to results from a 2003 study by Nielsen et al.⁸⁸ In this pilot study in 177 patients, the incidence of AF was significantly lower with AAIR pacing (7.4%) compared to DDDR with long AV delay (17.5%) and DDDR with short AV delay (23.3%). In the SAVE PACe trial, patients had significantly less persistent AF if randomized to a dual-chamber mode allowing single nonconducted P waves not to be tracked, to minimize ventricular pacing in the dual-chamber mode, than patients with conventional DDDR pacing⁹² (see Role of Prevention of Unnecessary Right Ventricular Pacing). However, patients with SND and conventional dual-chamber pacing in DANPACE received RV pacing for only 65% of the time versus 99% in SAVE PACe.

In conclusion, several earlier studies comparing AAIR and DDDR pacing in patients with sinus node disease and narrow QRS complex suggested an advantage of the single-chamber atrial pacing mode that systematically rules out unnecessary ventricular pacing. This was strongly confirmed by post hoc analyses from MOST, in which unnecessary RV pacing was associated with worse clinical outcome. However, two randomized prospective trials (DANPACE and MVP trial; see later) did not confirm a benefit of AAIR pacing (DANPACE) or DDDR pacing with strict prevention of RV pacing in patients with an implantable cardioverter-defibrillator (ICD). Most likely, the development of first-degree AV block with long PR intervals or higher-degree AV block adversely affects hemodynamics if the ventricle is left unpaced. AV block may also enhance the development of AF, thus questioning the concept that RV pacing should always be systematically avoided for its associated ventricular dyssynchrony.

Therefore, at present, there are no firm recommendations on the ideal pacing mode in sinus node disease. AAI pacing does not appear as desirable as before DANPACE. Potentially, implantation of a dual-chamber device programmed to maintain AAI pacing as long as sufficient for proper AV synchrony may be the best option, avoiding reoperation whenever first-degree AV block or higher develops. The role of first-degree AV block induced by atrial pacing from sites far from the septum (e.g., right atrial appendage, lateral right atrial wall) needs to be further evaluated. Atrial pacing–induced AV desynchronization with first-degree paced AV block may be one mechanism limiting the potential benefits of atrial and dual-chamber pacing in patients with SND.

 Dual-Chamber Versus Ventricular Pacing in Defibrillator Patients

The possibility of physiologic pacing in patients with an indication for an ICD and dual-chamber pacing was highly desired before it was available.⁹³ However, the first studies on dual- versus single-chamber ICDs were disappointing, failing to show improved discrimination between ventricular and supraventricular tachycardia.^94–96 Some newer studies question if single-chamber ICDs are comparable to dual-chamber ICDs in prevention of inappropriate therapy.^97–99

However, one of the aims of dual-chamber pacing in ICD patients is the reduction in clinical events, particularly heart failure and HF progression caused by bradycardia, which frequently develops during treatment with β-blockers and amiodarone. Even an increased risk of ventricular tachyarrhythmia (e.g., from bradycardia-dependent arrhythmias) has been observed in ICDs that were programmed to a VVI 40-bpm backup pacing mode.¹⁰⁰ The Dual Chamber and VVI Implantable Defibrillator (DAVID) trial therefore evaluated if dual-chamber pacing reduced the incidence of HF hospitalization and all-cause mortality compared to backup ventricular pacing in patients with a conventional indication for ICD therapy and heart failure (LVEF ≤40%) but no standard indication for antibradycardia pacing.¹⁰¹ In 37 U.S. centers, 506 patients were enrolled and randomized to DDDR pacing at 70 bpm versus VVI backup pacing at 40 bpm (software randomization, DDDR systems implanted in all patients). Programming of other parameters, particularly the AV delay, was left to the individual investigator. The mean sinus rate at baseline was 72 bpm, the mean PR interval of patients in DAVID was 184 msec, and 27% of patients had a BBB. At 6 months, 86% of patients received β-blockers and 27%, amiodarone. The trial was prematurely stopped at a mean follow-up of 8 months because, unexpectedly, patients with dual-chamber ICDs had a significantly higher incidence of HF hospitalization (43 [23%] at 1-year follow-up) and death (23 [10%] at 1 year) than patients with single-chamber ICDs (30 and 15, equivalent to 1-year incidence of 13% and 6.5%). This difference reached statistical significance in favor of single-chamber ICDs (Table 10-13).

One major difference in patients with single-chamber ICDs was the significantly lower proportion of ventricular pacing. At the 6-month control, device counters documented that VVI 40-bpm back-up pacing was associated with a mean proportion of ventricular pacing of only 0.6% in these patients without bradycardia compared to 59.6% ventricular pacing in the DDDR mode. A post hoc analysis corroborated the hypothesis that right ventricular pacing was the cause of worse outcome in patients with dual-chamber ICDs.¹⁰² In 380 patients who had a 3-month follow-up and had not yet reached a study endpoint, a multivariate analysis demonstrated that the percent of cumulative RV pacing was correlated with death or HF hospitalization during further study. If patients with dual-chamber ICDs had RV pacing more than 40% of the time, they had a more than fivefold higher risk of HF hospitalization or death than patients with a dual-chamber ICD and RV pacing for 40% or less of the time (P = .02). Patients with DDDR pacing at 70 bpm and ≤40% RV pacing trended toward a better outcome than those with VVI 40-bpm backup pacing.

Another analysis assessed data in patients with “soft” indications for pacing, such as bradycardia, defined as sinus bradycardia of less than 60 bpm, and without a conventional indication for pacing.¹⁰³ Results similar to the overall analysis were observed in these patients. However, the two soft indications are fundamentally different; patients with first-degree AV block may have experienced specifically deleterious effects of dual-chamber pacing because they likely were continuously paced (and unpaced in VVI 40-bpm group), whereas patients with sinus bradycardia might have specifically benefited from dual-chamber pacing as long as a sufficiently long programmed AV delay prohibited unnecessary RV pacing. However, this was not reported in this post hoc analysis.

These data were partly replicated in the second Multicenter Automatic Defibrillator Implantation Trial (MADIT-II), although differences between pacing modes in a primary prevention ICD trial did not reach statistical significance.¹⁰⁴ In 404 patients who received a single-chamber ICD and 313 patients with a dual-chamber ICD for primary prevention of sudden cardiac death after myocardial infarction (LVEF ≤30%), the trend was toward an increase in hospitalization for heart failure in patients with dual-chamber devices. Because of the nonrandomized assignment to dual-chamber ICDs, however, patients who had received dual-chamber devices were older and had higher NYHA functional class, more frequent prior coronary artery bypass surgery, wider QRS complex, more frequent left BBB, atrial tachyarrhythmias, and higher nitrate urea levels.

The DAVID investigators performed a second study with VVI backup pacing at 40 bpm in patients with heart failure, an indication for ICD therapy, and no conventional indication for bradycardia pacing (DAVID-II).¹⁰⁵ This time the dual-chamber group was programmed to AAI 70 bpm and tested for “non-inferiority” compared to backup VVI pacing at 40 bpm. In 29 U.S. centers, 543 patients were enrolled and implanted with a dual-chamber ICD programmable to the VVI and AAI pacing modes. Of note, 57 patients from the first DAVID trial who did not reach a study endpoint were also included. During a mean follow-up of 2.7 years, no significant difference was seen in the primary combined endpoint of time to HF hospitalization or death (incidence 11%, 17%, and 25% at 1, 2, and 3 years, respectively). In a subanalysis, asymptomatic sinus bradycardia at baseline was associated with a lower incidence of adverse outcome (12% vs. 26%) in patients randomized to VVI 40-bpm back-up or AAIR 70-bpm pacing.¹⁰⁶

The question if dual-chamber ICDs have any potential to reduce adverse clinical events compared to single-chamber ICDs was assessed in the Dual Chamber and Atrial Tachyarrhythmias and Adverse Events Study (DATAS).¹⁰⁷ Patients with Class I indication for an ICD (according to ACC/AHA 1998 guidelines) were included if permanent atrial tachyarrhythmias and an indication for dual-chamber pacing were absent. A total of 334 patients were then randomized to single-chamber ICD, dual-chamber ICD, or “simulated single-chamber ICD” (i.e., dual-chamber ICD implanted but programmed to single-chamber mode for pacing and discrimination between supraventricular and ventricular tachyarrhythmia). With crossover of the latter two groups after 8 months, all patients were followed for 16 months. The primary outcome was a composite of all-cause-mortality, need for cardiovascular hospitalization or intervention, two or more inappropriate shocks, and sustained symptomatic atrial tachyarrhythmia for 48 hours or longer. This composite was expressed as a prespecified score. Any one of these endpoints occurred in 65 patients with dual-chamber ICD (58%), 82 patients with single-chamber ICD (74%), and 84 patients with a dual-chamber ICD programmed to single-chamber parameters (76%). Mortality was 4% in dual-chamber, 9% in single-chamber, and 10% in simulated single-chamber ICD patients. The composite endpoint score was 33% lower for patients with dual-chamber ICDs (Fig. 10-9).

Figure 10-9 Clinical events in DATAS.

All numbers represent percentages in Dual Chamber and Atrial Tachyarrhythmias Adverse Events Study. HF, heart failure.

In ICD patients, study results suggest that prevention of unnecessary right ventricular pacing plays a major role. Comparing single- and dual-chamber devices, trials showed a significant increase particularly in HF hospitalization with dual-chamber ICDs that was not caused by the dual-chamber mode itself but by RV pacing. This can lead to severe deterioration of ventricular function, most likely through desynchronization of right and left ventricular contraction, especially in preexisting HF. Of note, studies randomizing ICD patients to single- or dual-chamber pacing enrolled only patients without an indication for antibradycardia pacing. Results therefore do not apply to ICD patients with SND or AV block. Also, patients without structural heart disease where overdrive pacing may play a role (e.g., patients with long QT syndrome) are not represented in studies randomizing patients to single-chamber versus dual-chamber ICDs. Other, “soft” indications have not been systematically studied. Although patients with first-degree AV block seem to have a particularly high percentage of unnecessary RV pacing with implications for HF progression (see below), sinus bradycardia, which may contribute and aggravate HF in some patients, may have a different outcome if ventricular pacing is prevented. Newer studies and post hoc analyses avoiding unnecessary RV pacing and using dedicated programming for better discrimination between supraventricular and ventricular tachyarrhythmias suggest some advantage for dual-chamber ICDs over single-chamber ICDs with 40-bpm backup VVI pacing.

 Critical Appraisal of Pacing Mode Trials

The large randomized trials on pacing mode dramatically changed the perception of “physiologic pacing.” In studies of the late 1980s, the addition of an atrial lead to a ventricular lead was perceived as a great step forward in cardiac pacing. The dual-chamber pacing mode enabled ventricular pacing at a programmable time after atrial sensing (AV-synchronous pacing), which improved cardiac output significantly compared with “asynchronous” ventricular pacing in the VVI(R) mode, as well as reducing mitral regurgitation, right atrial pressure, and pulmonary capillary wedge pressure and resolving pacemaker syndrome in a significant number of patients. Retrospective studies confirmed these hemodynamic advantages of dual-chamber pacing by clinical data. However, these clinical data from retrospective and nonrandomized studies were questioned^108,109 when selection bias became evident.^61,110

It is disappointing and difficult to explain why the theoretical advantages and the positive results of early studies did not translate into positive results in large, randomized clinical trials. However, some findings deserve special mention.

Other Issues

In addition to interpretations of the findings in the large RCTs on pacing mode selection, the following issues should be considered when evaluating the clinical implications of these trials:

1 Experience of study centers with the more complex, dual-chamber systems (atrial lead dislodgement, atrial over/undersensing, programming issues).

2 Lack of delineation of patient subgroups, such as patients with permanent bradycardia and pacemaker dependence.

3 Insufficient instruments to quantify pacing-associated symptoms, quality of life, and exercise capacity, particularly in elderly patients with comorbidity.

4 Unequal diagnostic tools to detect atrial fibrillation, an endpoint in all trials. The atrial lead allows easy detection of AF with the atrial EGM, whereas in ventricular devices, detection of AF relies on the surface ECG and may be difficult in complete AV block.⁸¹

5 Too few trials with sufficient long-term follow-up.

The randomized trials on pacing mode selection demonstrate that mortality is similar with ventricular single- compared to dual-chamber pacing. This mainly refers to patients with atrioventricular block; for those with sinus node disease, it has not been convincingly shown that pacing improves survival compared to the natural course. It also became quite evident that dual-chamber pacing is associated with less AF than ventricular single-chamber pacing. However, the central question is whether dual-chamber pacing is associated with improved hemodynamics that translate into less heart failure, less symptoms, and better QOL and exercise capacity; this remains unanswered even after randomizing more than 7000 patients. Beyond criticism about how such “soft endpoints” should best be evaluated (e.g., SF-36 unable to detect subtle, pacing-related QOL differences; symptom scores not sufficiently validated and not used), there are some critical, unresolved issues.

First, randomized trials on pacing mode selection suffer from selection bias, particularly if hardware randomization is applied. From theoretical considerations and earlier studies, investigators have ethical concerns regarding whether pacemaker-dependent patients are at an increased risk for adverse events with ventricular single-chamber pacing. It is a strength of published trials (particularly CTOPP and UKPACE, which were hardware randomized) to keep track of why eligible patients were not enrolled into the trial: In CTOPP, only 16% of eligible but not enrolled patients declined consent, whereas in 56% the physician declined.⁷²

Second, in a blinded trial, patient preference would be a powerful argument in favor of or against a pacing mode.^25,26,62,63 Unfortunately, but logically, all large RCTs on pacing mode selection had a parallel trial design preventing intraindividual comparisons. Intraindividual comparisons could only be made in crossover patients (because of suspected pacemaker syndrome) from VVIR to DDDR pacing in PASE and MOST, showing a striking preference for DDDR pacing in these select patients.

Third, the choice of software versus hardware randomization has both advantages and problems. If a patient “software randomized” to VVIR expresses symptoms, it is easy to reprogram the device to DDDR and see if symptoms are resolved. This may lead to an overestimation of the incidence of “VVIR intolerance” (26-31% in PASE and MOST). On the other hand, in hardware randomization, crossover to dual-chamber pacing requires surgical intervention with potential complications, medically and legally. It is therefore evident that the incidence of “VVIR intolerance” is underestimated in hardware-randomized trials (3.1% in UKPACE, 4.3% in CTOPP) (Fig. 10-10).

Figure 10-10 Intolerance to VVI(R) in PASE, MOST, CTOPP, and UKPACE trials.

Blue columns, Studies with software randomization requiring only reprogramming to crossover to DDD(R). Red columns, Studies with hardware randomization requiring reoperation to crossover to DDDR.

In summary, despite the RCT results on pacing mode selection, the perception that dual-chamber pacing is superior to single-chamber ventricular pacing in sinus rhythm still governs current guidelines^111,112 and international physicians’ choices^75,113 of pacing mode selection. However, trials have shown that (1) the beneficial effects of dual-chamber pacing may have been overestimated (see rationale and sample size calculation of the Danish trial and UKPACE), (2) dual-chamber pacing by itself may not convey hemodynamic benefits if not optimized to the patient (lead position, AV delay, pacing rates), (3) dual-chamber pacing is more complex and therefore associated with more pitfalls and potential for complications (atrial lead dislodgement, pacemaker mediated tachycardia), and (4) dual-chamber pacing is not as “innocent” as previously thought because it may cause more poorly timed right ventricular pacing than VVI(R), with potentially deleterious effects.

 Role of Prevention of Unnecessary Right Ventricular Pacing

Surprising data from the DAVID trial showed that compared with VVI 40-bpm backup pacing, DDDR pacing at 70 bpm produced more, not less, heart failure hospitalization and a slight, statistically not significant increase risk of mortality in patients with an ICD and no bradycardia¹⁰¹ (see earlier). In search of an explanation, it became obvious that right ventricular pacing was present in dual-chamber pacing much more frequently than in ventricular backup pacing (56% vs. 3% of patients with paced ventricular rhythm after 6 months). Similar results have been reported (independent of the presence of a single- or dual-chamber ICD) for patients in MADIT-II, where a cumulative proportion of ventricular pacing of more than 50% was associated with an almost twofold increased risk of new or worsened heart failure.¹¹⁴

A number of studies on histology,¹¹⁵ hemodynamics,¹¹⁶ and echocardiography¹¹⁷ have demonstrated deleterious acute and long-term consequences of RV apical pacing, their full clinical impact emerged only after DAVID.

The MOST trial enrolled only patients with sinus node disease; apart from 11% of patients with additional second- or third-degree AV block,⁶⁷ patients who did not require ventricular pacing. Sweeney et al.³⁸ investigated the effect of unnecessary RV pacing in patients with SND. They selected 1339 patients from MOST who had a narrow QRS (<120 msec) at baseline before pacing was instituted. Taken from pacemaker memory functions, 707 patients randomized to DDDR received RV pacing at an average 90% of the time. In contrast, in patients randomized to VVIR, RV pacing was present only 58% of the time. In a Cox proportional hazard model, RV pacing (again, unnecessary in SND patients) was a strong predictor of HF hospitalization in DDDR mode (Table 10-14). For percentages of RV pacing in the AV-synchronous mode up to 40%, every 10% increase in ventricular pacing linearly increased the relative risk increase for HF hospitalization by 54%. RV pacing in the DDDR mode more than 40% of the time versus less than 40% was associated with a 2.6-fold increased risk of HF hospitalization. This association was much more striking for DDDR than for VVIR pacing, in which RV pacing up to 80% of the time had only a minor impact on HF development, whereas pacing more than 80% of the time resulted in a 2.5-fold increased risk of HF hospitalization.

Similar to the increased risk of heart failure hospitalization, unnecessary right ventricular pacing also increased the risk of atrial fibrillation.³⁸ Up to 80% to 85% of RV pacing in the DDDR or VVIR modes, the risk of AF linearly increased by approximately 1% for every percent increase in cumulative ventricular pacing. The authors concluded that unnecessary RV pacing in SND leads to ventricular desynchronization in asynchronous as well as in AV-synchronous RV pacing, and over a broad range of cumulative percentages of pacing, linearly increases the risk of development of HF and AF in patients without BBB.

The higher percentage of RV pacing in dual-chamber than in ventricular single-chamber pacing may be one explanation for the lack of superiority of this pacing mode in large trials. A post hoc analysis confirmed this for the DAVID trial.¹⁰² The authors dichotomized the group of patients with DDDR pacing at 70 bpm into those with 40% or less versus those with more than 40% RV pacing. In a Cox regression analysis, ventricular pacing 40% or less of the time was associated with slightly fewer clinical events than with VVI 40-bpm backup (P = .07). Patients with ventricular pacing more than 40% of the time in DDDR mode had a 5.3-fold increased risk of a study endpoint (death or HF hospitalization) than patients with 40% or less ventricular pacing.

Prevention of unnecessary RV pacing in patients with a dual-chamber pacemaker or ICD is difficult to achieve with prolonged settings for the AV delay alone. In a pilot study with 177 patients with SND and normal AV conduction, Nielsen et al.⁸⁸ observed ventricular pacing still for 17% of the time in DDDR mode with a fixed AV delay as long as 300 msec. Beyond being ineffective, programming AV delays to such long values creates new problems, such as limitations of the upper tracking rate or AV desynchronization arrhythmias.^118,119

A better way to prevent unnecessary RV pacing is therefore an automatic algorithm to provide AAIR pacing, with monitoring of AV conduction and an automatic switch to DDDR when AV block occurs beyond a single dropped beat. Two such algorithms developed for minimized ventricular pacing are MVP (Medtronic, Minneapolis) and SafeR (Sorin, Milano). For the SafeR algorithm, a reduction in RV pacing to less than 1% in patients without permanent AV conduction block (SND, paroxysmal AV block) was shown, together with safe switching to DDDR mode when necessary.^120–122 For the MVP algorithm, its superiority in prevention of unnecessary RV pacing versus an extended AV search hysteresis has been demonstrated (0.9% in first-degree AV block vs. 80.6% ventricular pacing),¹²³ maintaining 99% ventricular pacing in patients with permanent third-degree AV block. In SND patients, MVP consistently reduced the percentage of ventricular pacing to 1.4%,^124,125 with a relative reduction of ventricular pacing of 99% in SND and 60% in paroxysmal AV block.¹²⁵ Similar reductions in RV pacing have been achieved in ICD patients.¹²⁶

The effect of consequent prevention of RV pacing on clinical outcomes was tested in 1065 SND patients in the SAVE-PACe trial.⁹² All patients had normal AV conduction and QRS duration less than 120 msec. After randomization to conventional dual-chamber pacing versus dual-chamber pacing with the addition of algorithms that specifically promote intrinsic AV conduction, patients were followed for 1.7 years, when the trial was stopped. RV pacing had been present for a median of 99% of the time in patients with conventional pacing; the addition of algorithms to promote spontaneous AV conduction and to preserve a narrow QRS limited unnecessary ventricular pacing to a median of 9%. The incidence of persistent AF was significantly lower (7.9%) in patients with dedicated algorithms to promote intrinsic AV conduction compared to patients with conventional dual-chamber pacing (12.7%; RRR, 40%). The need for any intervention for AF (cardioversion, catheter ablation) was also significantly reduced by 38%. Mortality was not significantly different (4.9% with minimized ventricular pacing, 5.4% with conventional dual-chamber pacing).

The MVP algorithm was also tested in DAVID-like patients with an indication for ICD therapy.¹²⁷ In a non-inferiority design, the MVP trial enrolled 1030 patients with an ICD indication, sinus rhythm and no indication for pacing to dual-chamber/minimal pacing at 60 bpm versus VVI 40 bpm backup pacing. The primary endpoint was time to death, HF hospitalization, or HF-related urgent care. The trial was stopped after 2.4 years after an interim analysis suggested that it was highly unlikely that there would be a statistically significant difference with regard to the primary endpoint regardless of follow-up duration. In these patients with a mean LVEF of 35%, all-cause mortality was slightly but not significantly higher in the dual-chamber group (11% vs. 9%), mainly driven by a difference in noncardiac death (4.1% vs. 3.1%). Heart failure hospitalizations and related urgent care was necessary in 20.8% of patients with a dual-chamber device, versus 18.8% in patients randomized to VVI 40-bpm backup pacing. These differences were not statistically significantly different (P = 0.873). However, because of fewer events than expected, the confidence interval crossed the prespecified border for non-inferiority. Therefore, the MVP trial was unable to demonstrate non-inferiority of dual-chamber ICDs incorporating an algorithm for prevention of RV pacing compared to VVI backup pacing at 40 bpm. Of note (even though not prespecified), a striking difference in outcome was observed in a post hoc analysis for patients with a baseline PR interval ≥230 msec (mean, 260 msec). Although event rates in patients with PR <230 msec were virtually identical between single- and dual-chamber ICDs, patients with a PR ≥230 msec had a significantly higher event rate with dual-chamber pacing promoting intrinsic AV conduction.

Similarly, the Inhibition of Unnecessary RV Pacing with AVSH in ICDs (INTRINSIC) trial investigated if dual-chamber ICDs are “non-inferior” to single-chamber ICDs with VVI 40-bpm backup, if an AV hysteresis algorithm prevents unnecessary RV pacing. Patients with a standard indication for an ICD were enrolled. All patients whose AV search hysteresis algorithm achieved ventricular pacing less than 20% of the time at the 1-week control visit were randomized to receive VVI 40-bpm or DDDR 60-bpm pacing. Investigators had to amend the study protocol (prolong AV delay during activity of AV search hysteresis) because a larger-than-expected number of patients would otherwise have been excluded from randomization.¹²⁸ Thus, of 1530 patients enrolled, a total of 988 with ventricular pacing less than 20% at the 1-week visit were randomized.¹²⁹ During a mean follow-up of 10.4 months, the endpoints of death and HF hospitalization were more frequent in VVI backup 40 bpm (9.5%) than in DDDR pacing with AV search hysteresis (6.4%). Non-inferiority of DDDR with AV search hysteresis met a significance of P <.001, even with a trend for superiority of DDDR with AV search hysteresis (P = .07) with regard to primary endpoints. Also, all-cause mortality was nonsignificantly lower in the dual-chamber ICD arm (3.6% vs. 5.1%; P = 0.23). Of note, the event rate was higher in the observational arm (patients excluded from randomization because ventricular pacing was present >20% of the time despite use of AV search hysteresis algorithm), with a total of 12.5% events and an all-cause mortality of 5.5%.

In summary, it has been shown impressively that unnecessary right ventricular pacing promotes the development of heart failure and atrial fibrillation, most likely through ventricular desynchronization and hemodynamic deterioration. In ICD patients with severely impaired left ventricular function, this leads to an increase in HF hospitalization and mortality. Even in patients with sinus node disease and normal LV function, RV pacing is associated with increased HF hospitalization. The higher proportion of unnecessary RV pacing in dual-chamber pacing than in single-chamber ventricular pacing likely helps explain the disappointing results of mode selection, particularly in SND patients. However, prevention of ventricular pacing is not automatically the better option. Patients with a normal PR interval at baseline and patients with a low percentage (<20%) of ventricular pacing achieved by simple means (e.g., intermittent prolongation of AV delay) seem to be ideal candidates for dual-chamber pacing with dedicated algorithms that prevent RV pacing. In contrast, patients with first-degree AV block, particularly with a baseline PR interval greater than 230 msec (or >260 msec) and those with more than 20% ventricular pacing despite prolongation of the AV delay do not seem to benefit from prevention of ventricular pacing; in these patients, ventricular pacing may not be entirely “unnecessary.”

However, if prevention of ventricular pacing is not the better option in all patients, and if RV pacing is associated with deleterious hemodynamic and structural side effects, biventricular pacing or pacing from an alternative RV site, such as the midseptum, may be superior in patients who require ventricular pacing.

 Trials Evaluating Right Ventricular Versus Biventricular Pacing

A growing body of evidence indicates that RV pacing leads to new-onset heart failure in a significant proportion (~25%) of patients with high-grade AV block and normal left ventricular function over the long term (>5 years), typically in patients who are older at implantation, those with a particularly wide QRS, and those with coronary artery disease.¹³⁰ To date, however few prospective randomized controlled trials have compared biventricular (BiV) to conventional pacing for standard bradycardia indications.

In the left ventricular–based cardiac stimulation post–AV nodal ablation evaluation (PAVE) trial, Doshi et al.¹³¹ randomized 184 patients with permanent AF and AV node ablation to BiV versus RV pacing. After 6 months, LVEF was significantly better in patients with BiV pacing (46% vs. 41%; P = .03) as was the 6-minute walking distance (6MWD), but no significant difference in perceived QOL. Patients with LVEF less than 45% or NYHA Class II and III at baseline had greater improvement in 6MWD by BiV pacing compared to patients with normal LVEF or no HF symptoms. Although PAVE was an important step in the evaluation of cardiac resynchronization therapy (CRT) for standard pacing indications outside hemodynamic indications in HF patients, it was certainly too small with insufficient follow-up to investigate the full impact of CRT in patients without symptomatic HF.

Biventricular pacing was tested in patients with standard bradycardia indications and heart failure in the Homburg Biventricular Pacing Evaluation (HOBIPACE) study.¹³² In 30 patients with AV block and LVEF of 40% or less, BiV devices were implanted and randomized to RV versus BiV pacing for 3 months, then a 3-month crossover phase. In patients with and without AF, BiV pacing induced improvements in LVEF and left ventricular end-diastolic as well as end-systolic volumes, a reduction in NT-proBNP by 31%, an improved QOL score, improved exercise capacity, and improved maximum oxygen consumption and oxygen uptake at the anaerobic threshold during cardiopulmonary exercise testing. Although these were impressive results, 63% of the patients in HOBIPACE had a left BBB (mean QRS duration, 174 msec), meeting the conventional indication for CRT.

In the RD-CHF study, 44 permanently paced patients with continuous single-chamber RV pacing for a standard pacing indication (worsening CHF) who required a pacemaker generator change for battery depletion were upgraded to a BiV system and completed a randomized crossover study.¹³³ All patients had NYHA Class III or IV with LVEF of 35% or less and echocardiographic signs of interventricular or intra-LV dyssynchrony. The mean paced QRS duration at baseline was 200 msec. During crossover with pacing in RV and BiV mode for 3 months each, QRS duration and LV preejection delay were significantly shortened by BiV pacing, and NYHA functional class, 6MWD, and QOL were significantly improved. Although these are important data in a population with a pacemaker indication, there is almost 100% overlap with current indications for CRT.

Albertsen et al.¹³⁴ randomized 50 patients with AV block (permanent or paroxysmal) who were referred for a first pacemaker implantation to either RV dual-chamber or BiV pacing.¹³⁴ Only four patients had a left BBB, and mean LVEF was 60% at baseline. All patients received a pacemaker capable of BiV pacing with three lead connectors, but only BiV-randomized patients received an LV lead (implantation success, 100%). During follow-up of 12 months, LVEF was unchanged in BiV pacing, with a minimal but statistically significant decrease to 57% with RV pacing. Tissue Doppler imaging revealed more LV dyssynchrony after 12 months with RV than BiV pacing (1.8 vs. 0.8 segments with delayed longitudinal contraction; P = .02). NT-proBNP levels, which were not high at baseline (122 pmol/L in DDDR: 91 pmol/L in BiV) remained unchanged with dual-chamber pacing but were significantly lower after 12 months with BiV pacing. In summary, Albertsen et al.¹³⁴ were the first to investigate BiV pacing in a group of patients with AV block (unrelated to AV node ablation) requiring ventricular pacing, without an indication for CRT. Although only surrogate parameters for outcome were analyzed, the 12-month follow-up was sufficient to produce echocardiographic and serologic results that indicate a hemodynamic superiority of BiV pacing in patients with AV block and normal LV function at baseline.

To date, PACE represents the largest trial on biventricular pacing for standard pacing indications.¹³⁵ In 177 patients with a pacing indication (AV block in 102, SND in 71, other indication in 4 patients) and normal systolic LV function (defined as LVEF >45%), a BiV pacemaker was implanted. Another 16 patients were excluded from randomization after an unsuccessful attempt of implantation of the LV lead. After stratification according to the presence or absence of diastolic dysfunction, patients were randomized to receive DDDR pacing (60-140 bpm) or BiV pacing. In patients with SND, the AV delay was programmed 20 msec shorter than intrinsic AV conduction. After 12 months, the primary endpoints (LVEF and LV end-systolic volume) were assessed. Left ventricular function was much better preserved in BiV pacing (LVEF = 62% vs. 55%, P <.001; end-systolic volume = 28 vs. 36 mL, P <.001) in patients with and without diastolic dysfunction at baseline. A significant deterioration in LVEF to less than 45% occurred in one patient (1%) during BiV versus eight (9%) during RV pacing (P = .02). Six patients with RV pacing and five with BiV pacing were hospitalized for heart failure; one patient with RV pacing died during the study. There was no significant difference in 6MWD or QOL using the SF-36 questionnaire. Echocardiographic data suggest that BiV pacing can prevent pacing-induced ventricular remodeling and heart failure in patients without significant systolic ventricular dysfunction. However, these are only surrogate parameters. PACE must be criticized for failed implant procedures that do not appear in the analysis, and for including SND patients with presumably normal PR interval and QRS duration, who underwent “forced” RV or BiV pacing. At least for RV pacing, it is known that heart failure can be caused by unnecessary pacing in these patients. The results of PACE therefore must be interpreted with caution.

Several studies on biventricular pacing for conventional pacing indications are ongoing. The PREVENT-HF trial has completed enrollment and follow-up of 108 randomized patients with a standard Class I or IIa indication for cardiac pacing and a projected need for ventricular pacing of at least 80% to RV or BiV pacing.¹³⁶ After 12 months, LV end-diastolic volume is compared as the primary endpoint.

The Biventricular Pacing for Atrioventricular Block to Prevent Cardiac Desynchronization (BioPace) study enrolled 1800 patients in 2007 with a conventional indication for pacing, with a high probability that the ventricle will be predominantly paced.¹³⁷ Patients with all NYHA functional classes were eligible, with no restriction in LVEF. Patients with an indication for primary prophylaxis of sudden cardiac death were allowed to receive an ICD instead of a pacemaker. There was a hardware randomization to RV versus BiV devices. The study is endpoint driven; follow-up will be terminated when 635 endpoints (all-cause mortality) have occurred.

In contrast, BLOCK-HF is a double-blind, software-randomized trial enrolling patients with a Class I or IIa pacing indication for second-degree or third-degree AV block or other pacing indications (e.g., first-degree AV block with pseudo–pacemaker syndrome, Wenckebach block rates <100 bpm, PR >300 msec).¹³⁸ Patients must have LVEF of 50% or less (NYHA Class I-III) and are software-randomized to RV versus BiV pacing (ICD if an indication for primary prevention of sudden cardiac death is present). During a follow-up of at least 12 months, the primary endpoint in the projected 1636 enrolled patients will be a combination of all-cause mortality, HF-related urgent care, or a 15% or more increase in LV end-systolic volume index.

In summary, there is some evidence that biventricular pacing can prevent the deleterious effects of right ventricular pacing in a significant number of patients. However, the results of BioPace and BLOCK-HF should be evaluated before BiV pacing, with the additional expenditure of time, money, and effort at implantation and the additional hardware, as well as potential short-term and long-term complications, is applied in patients with a pacing indication involving an expected high percentage of ventricular pacing. In the meantime, routine placement of an LV lead for patients with AV block is not recommended by the pacing guidelines. Patients with an indication for CRT, according to REVERSE, MADIT-CRT, and RAFT (LVEF ≤40%; NYHA I-III if positive history of ≥Class II),^139–141 may represent candidates that can benefit from biventricular devices based on the assumption that they would otherwise have a (paced) broad left bundle branch block (LBBB).

 Cost-Effectiveness of Dual-Chamber Pacing

Analyses of cost-effectiveness in pacemakers and ICDs have limited applicability because of the varied and highly time-dependent costs of devices. This is particularly important in countries where health care providers and industry can negotiate a mixed calculation (i.e., single- and dual-chamber devices sold at same price). Costs also depend on the outpatient availability of pacemaker (and ICD) implantation and cost of the operation room and staff need for anesthesiologist, conscious sedation vs. general anesthesia in ICD implantation), which vary greatly according to routine practices and legal issues. Given this background, cost-effectiveness calculations from study data must be considered in the study’s context. Interestingly, even data from large randomized trials can be interpreted so differently that cost-effectiveness can range from “highly ineffective” to “highly effective” based on the same data.

Similar to diverging assumptions on costs, assumptions on “efficacy” also vary widely. Cost-effectiveness in terms of “cost per life-year gained” depends only on mortality and the studies included in the analysis. However, “cost per quality-of-life year” (QOL-year) depends on the reliability of the QOL instruments used and inclusion of associated effects (e.g., AF, pacemaker syndrome, stroke, HF, exercise capacity). This may be even more important in the evaluation of pacing modes, but it is much more difficult to assess in a statistical model and less robust.

In 1996, Sutton and Bourgeois¹⁴² published a cost-effectiveness analysis based on the (nonrandomized) studies available. They found a 10-year survival in sinus node disease with DDD pacing of 71%, versus only 57% with VVI pacing. The numbers for AV block were 61% with DDD and 51% with VVI pacing. For QOL, prevalence of heart failure was 60% lower with dual-chamber pacing; severe disability was present with VVI pacing in 36% of SND patients and 22% of AV block patients, but only for 8% and 3% of patients paced in dual-chamber mode. Assuming the implantation of a dual-chamber system costs 1.5 times more than a VVI system, this higher cost was offset by the higher cost of complications and side effects (stroke, disability, upgrade) during VVI pacing after 3 years. Assuming a battery life of 10 years, dual-chamber pacing was highly cost-effective in this analysis.

Cost-effectiveness was calculated for a subset of patients from the Canadian Trial of Physiologic Pacing.¹⁴³ CTOPP authors calculated additional costs of approximately 3000 Canadian dollars ($) for implantation and follow-up of a dual-chamber versus single-chamber device. With a (nonsignificant) relative mortality reduction of 9.4% with dual-chamber pacing, the authors calculated a gain of 0.01 life-years with dual-chamber pacing, with costs of approximately $300,000 per life-year saved. Only in the subgroup of patients with an intrinsic heart rate less than 60 bpm would the cost per life-year be acceptable, at $16,000. A key consideration in this analysis is the need for reoperation because of pacemaker syndrome. In CTOPP, this was necessary in only 4.3% of patients randomized to VVI during 5 years,⁷² versus 25% to 30% in PASE and MOST, where “reoperation” could be replaced by reprogramming the devices.^65,67

An analysis of the U.S. MOST trial found that dual-chamber pacing in patients with sinus node disease was cost-effective, with an average of $53,000 per life-year saved.¹⁴⁴ However, depending on the statistical model and method, and exclusion or inclusion of QOL data, results varied between $3200 and $334,000.

A cost-effectiveness analysis by the British National Health Service in 2005 (before UKPACE results) took into account the lower incidence of AF and pacemaker syndrome in dual-chamber pacing, particularly in patients with SND.¹⁴⁵ The higher incidence of complications in dual-chamber devices at implantation was also considered. The cost of the dual-chamber system, including all costs of complications and clinical events during a 5-year period, was assumed to be 7400 . The cost difference compared to single-chamber systems was small, about 700 during 5 years. The cost for dual-chamber versus single-chamber ventricular pacing over 5 years was therefore estimated to be 8500 per QOL-year saved in patients with AV block and 9500 in patients with SND. This cost was reduced to approximately 5500 over 10 years. However, these calculations would dramatically change, to approximately 30,000 per QOL-year saved, if a lower incidence of pacemaker syndrome, similar to the CTOPP, would have been used.

A “cost-efficacy analysis” has been performed retrospectively in patients with single-chamber versus dual-chamber ICDs using the Yale University Electrophysiology database.¹⁴⁶ It included 453 patients who had received a single- or dual-chamber ICD between June 1997 (when the first dual-chamber ICD received FDA approval) and July 2001. Three strategies of pacing mode selection were assessed: (1) implantation of a single-chamber ICD with upgrade whenever needed, (2) implantation of a dual-chamber ICD in all patients, and (3) electrophysiologic (EP) study and implantation of a dual-chamber ICD only in patients with results indicating abnormal sinus or AV nodal function. Interestingly, the targeted, EP-guided device selection was the most expensive strategy (>$41,000); implantation of single-chamber ICDs with need for upgrade in an estimated 20% of patients cost $39,000. Surprisingly, implantation of dual-chamber ICDs in all patients would have been the least expensive strategy, only $36,000. Subsequent sensitivity analyses demonstrated that even with need to upgrade single-chamber ICDs in only 10% or even 5%, implantation of a dual-chamber ICD remained less expensive as long as it was no more than $1500 over a single-chamber ICD. An EP-guided approach would be least expensive in patients who underwent EP testing for other reasons (e.g., inducibility of ventricular tachyarrhythmias).

A recent analysis with assumptions representative for current systems calculated cost-effectiveness in a statistical model.¹⁴⁷ The authors assumed that the implantation of dual-chamber devices instead of VVIR systems in 1000 patients would prevent AF in 36 and severe pacemaker syndrome in 168 patients, while requiring 13 additional hospitalizations from complications over 5 years. In this model, health benefits would need an incremental cost of 23 per patient and 0.09 QOL-years, equal to incremental costs of 260 per QOL-year gained. Similar to other analyses, upgrade procedures for pacemaker syndrome would have the greatest impact on costs.

In summary, the majority of calculations of cost-effectiveness show that dual-chamber pacing provides benefits at an acceptable price. However, these analyses are weak and depend on a large number of hazy assumptions.

 Current Guidelines on Pacing Mode Selection

The current American College of Cardiology/American Heart Association (ACC/AHA/HRS) guidelines on cardiac pacing reference the five pacing mode selection trials discussed (Danish, PASE, MOST, CTOPP, UKPACE).¹⁴⁸ A major limitation of these trials was a rate of crossover and dropout of assigned pacing mode greater than 37%. Whenever a low percentage of pacing is expected (e.g., “very infrequent sinus pauses or infrequent AV block”), maintenance of AV synchrony during pacing likely does not seem important, and ventricular single-chamber pacing may be used as an alternative to dual-chamber pacing. However, progression of bradycardia is difficult to predict. Therefore, some patients with a seemingly rare need for pacing may develop bradycardia with permanent need of pacing (e.g., caused by β-blockers and antiarrhythmic drugs) and may then require a reoperation for an upgrade to dual-chamber pacing.

The current European Society of Cardiology (ESC) guidelines on pacing recommend only AAIR pacing and DDDR, with dedicated algorithms to prevent unnecessary right ventricular pacing in sinus node disease.¹⁴⁹ Single-chamber ventricular pacing is discarded because of the higher risk of atrial fibrillation and the trend toward lower exercise capacity and more pacemaker syndrome in a Cochrane review.¹⁵⁰ In patients with AV block, VVIR is recommended only in those without sinus rhythm. Otherwise, VDD, DDD, and DDDR devices are recommended, depending on chronotropic sinus node function. VVIR is mentioned only as a Class IIb indication (i.e., majority of experts would not chose this pacing mode in patients with AV block and sinus rhythm). Alternative RV pacing sites, such as septal, hisian, parahisian, and biventricular, are mentioned. However, no recommendation is provided for these pacing sites in AV block with narrow QRS.

 Future Perspectives in Pacing Mode Evaluation

The studies assessing whether any pacing mode is preferable in different indications for permanent pacing have gathered a tremendous amount of information. Unfortunately, the results are frequently contradictory and difficult to interpret, even in many well-conducted and designed trials. It was the explicit aim of this chapter to outline the difficulties in interpretation of results of pacing mode selection trials.

On the other hand, these trials have made it clear that, although it is easy to demonstrate a benefit of pacing regardless of mode in patients with life-threatening bradycardia, it is more difficult to show added benefit for any specific pacing mode, particularly in patients with less severe bradycardia. However, data from randomized trials suggest that implanting a more complex pacemaker system alone does not necessarily translate into better outcome; the pacing system must also be optimized to the patient’s needs. Despite some contradictory results, programmed parameters and algorithms should prevent unnecessary RV pacing in the ventricular single-chamber and perhaps even more in dual-chamber modes. At present, it is not clear which patients most need prevention of RV pacing or what alternative exists for different pacing indications. For patients with usually unavoidable ventricular pacing, conventional dual-chamber pacing must be tested against BiV pacing also in the absence of heart failure and wide QRS. In patients with intermittent AV block and particularly for patients with first-degree AV block, the ideal pacing mode at present is not clear. In patients with intermittent AV block and narrow QRS, BiV pacing in periods with AV block alternating with facilitation of intrinsic conduction in periods of normal AV nodal function may be superior to conventional dual-chamber pacing, particularly in patients with LV dysfunction and clinical HF. In patients without LV dysfunction or HF, it remains unclear what the best pacing mode currently is. In patients with AV block, there seems to be a turning point (intrinsic PR interval >260 msec) after which prevention of ventricular pacing and spontaneous AV conduction may be worse than RV pacing. This turning point may differ in patient with and without HF.

The diagnostic benefit of atrial leads has not been systematically evaluated. Information on the atrial rhythm may be particularly useful in ICDs, to discriminate between supraventricular and ventricular tachyarrhythmias, but potentially even more to detect asymptomatic, paroxysmal atrial fibrillation, particularly in patients with a high CHADS₂ score to establish anticoagulation.

Future trials on pacing mode selection should also take into account optimized programming of devices. Activation of useful algorithms (mode switching, prevention of unnecessary ventricular pacing), deactivation of useless algorithms (AV delay prolongation in patients with permanent AV block), optimization of algorithms (rate response), and optimization of parameters (atrial sensitivity to avoid AF undersensing; sensed and paced AV delay) may play a role. Similarly, the role of atrial lead position (septal, Bachmann bundle) and ECG-guided ventricular lead placement may further improve results of different pacing modes and should be investigated.