Cheng, Davy*; Diegeler, Anno†; Allen, Keith‡; Weisel, Richard§; Lutter, Georg∥; Sartori, Michele¶; Asai, Tohru**; Aaberge, Lars††; Horvath, Keith‡‡; Martin, Janet*§§
Chronic symptomatic angina represents a significant clinical and economic burden worldwide. Complete revascularization is commonly not feasible for patients with severe angina caused by coronary anatomy that is not amenable to conventional revascularization through coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI) and/or concomitant patient risk factors that preclude conventional revascularization. Since medical management alone has already failed these patients, the options for reducing symptoms and improving quality of life remain limited or nonexistent. Typically, these patients with end-stage coronary artery disease have been referred to as “no-option” patients.
Transmyocardial laser revascularization (TMR) has recently been introduced for the treatment of refractory angina in patients with coronary disease not amenable to standard revascularization.1–3 Although the mechanism of action of TMR for coronary revascularization has eluded description, a number of theories have been put forward including possible sustained channel patency, neovascularization, denervation, placebo, or some combination of these.4 Despite this lack of understanding, a number of clinical studies have demonstrated that TMR reduces symptom severity in patients with chronic angina when compared with medical therapy alone.3 More recently, TMR has been studied as an adjunct to conventional CABG to determine whether it may play a role in improving outcomes in patients with coronary morphology that is only partially amenable to surgical revascularization. Although a number of randomized clinical trials have been published over the past decade, most of these trials have had insufficient power to measure differences in clinical and economic outcomes with precision. Synthesis of the existing trials through appropriate statistical methods using meta-analysis may improve the power to detect clinically important differences between TMR and conventional methods of revascularization.
Presently, no comprehensive and methodologically rigorous meta-analysis of TMR for chronic severe angina exists. One meta-analysis3 and one early systematic review5 of TMR have previously been published. However, these reviews did not follow the recommendations currently set out for rigorous meta-analysis and systematic review,6 and the most recent trials were not included. Additionally, a number of key clinical and resource-related outcomes were not addressed. Therefore, a comprehensive current meta-analysis is needed to determine the benefits of TMR to patients (improved clinical outcomes), practitioners (improved quality of care), and providers (improved cost-effectiveness).
This comprehensive, systematic review with meta-analysis sought to determine whether:
I. Transmyocardial laser revascularization improves clinical and resource outcomes compared with conventional maximal medical treatment (MMT) in chronic angina patients with coronary artery and myocardial morphology deemed not amenable to revascularization by conventional revascularization (CABG or PCI).
II. TMR adjunctive to CABG improves clinical and resource outcomes compared with CABG alone in patients with chronic angina with coronary morphology deemed only partially amenable to revascularization by conventional revascularization.
To address these overall objectives, the following three categories of outcomes will be addressed:
Severity of angina symptoms, reported as a change in New York Heart Association (NYHA) or Canadian Cardiovascular Society (CCS) class achieved. Secondary end points were defined as improvement in NYHA or CCS class by at least 2 classes, perioperative pain, patient satisfaction or preference, and quality of life (QOL).
B. Perioperative and Long-term Complications
Myocardial infarction (as defined by the study authors), stroke or neurologic event, need for repeat intervention (CABG or PCI), surgical reexploration, transfusions, renal failure, or a composite of all perioperative complications. Changes in results of imaging studies such as perfusion imaging wall motion studies, ejection fraction, exercise treadmill stress testing (ETT), and hemodynamic assessment will also be considered.
C. Resource Utilization
Duration of procedure, ICU length of stay, total hospital length of stay, emergency department visits, doctor visits, rehospitalizations, and costs.
This meta-analysis of randomized trials was performed in accordance with state of the art methodologic recommendations6,7 and according to a protocol that prespecified outcomes, search strategies, inclusion criteria, and statistical analyses.
Definition of End Points
The primary end point of interest was reduction in severity of angina symptoms reported as mean angina class at end of study (NYHA or CCS angina scores). Secondary end points were defined as improvement in NYHA or CCS class by ≥2 classes, continued severe angina, continued angina of any severity, results of ETT, myocardial infarction (as defined by the authors), stroke or neurologic event, need for repeat intervention (CABG or PCI), surgical reexploration, transfusions, renal failure, perioperative complications, QOL, patient preference or satisfaction, duration of procedure, ICU length of stay, total hospital length of stay, emergency department visits, physician visits, readmissions, and costs. Each end point was assessed for the perioperative period and for the longer term. In addition to clinical end points, the results of imaging studies to assess perfusion and ventricular function were examined, including perfusion imaging wall motion studies.
A comprehensive literature search of MEDLINE, EMBASE, and Cochrane CENTRAL, using keywords and variants of “transmyocardial,” “TMR,” “TMLR,” “coronary and revascularization,” “laser,” “excimer,” “holmium,” and “CO2” was performed from the earliest available date to March 2006. The most recent 12 months of relevant surgical and anesthesia journals were hand-searched, and databases of conference abstracts were reviewed electronically (AATS, STS, ISMICS). Experts were contacted to solicit unpublished studies.
Eligible trials had to be randomized or nonrandomized controlled trials in adults with chronic refractory angina not amenable to conventional revascularization alone or only partially amenable to conventional revascularization with CABG or PCI. Published and unpublished trials were included, in any language. Specifically, trials comparing one of (1) TMR versus MMT for patients not eligible for conventional CABG and (2) TMR + CABG versus CABG alone in patients with morphology only partially amenable to conventional CABG surgery. Trials of percutaneous laser revascularization were excluded. Noncomparative studies (ie, before-after studies, case series) were not included in the statistical meta-analysis.
Two authors independently extracted the following data points: baseline demographics including number of patients, inclusion/exclusion criteria for patient entry to study, age, sex, smoking status, ejection fraction, diabetes, hypertension, hyperlipidemia, perioperative and long-term complications, procedure time, ventilation time, length of stay, costs, and all relevant clinical outcomes. Discrepancies were resolved by consensus.
Odds ratios and their 95% confidence intervals (OR, 95% CI) were calculated for discrete data. Weighted mean differences (WMD, 95% CI) or standardized mean difference (95% CI) were calculated for continuous data, as appropriate. Heterogeneity was explored through the Q-statistic and by calculating the I2. Summary odds ratios and WMDs were calculated by using the fixed-effects model when statistical heterogeneity was not found (ie, Q-test P value >0.10 and I2 ≤50%). The random-effects model was used when statistical heterogeneity was found (ie, Q-test P value <0.10 or I2 >50%). Statistical significance for overall effect was defined as P < 0.05 or a 95% confidence intervals that excluded the value 1.00 for odds ratios and 0.00 for WMDs.
Subgroup analysis was planned a priori for CO2 laser, holmium:YAG, and XeCl laser energy. In addition, sensitivity analysis was planned for elderly patients (>70 years), patients with left ventricular dysfunction (LVD), and publication status (unpublished versus published data). Post hoc subanalysis was conducted for trials with the majority of patients having class III angina at study entry versus trials with predominantly class IV angina at entry. Publication bias was explored though visual inspection of funnel plots.
Of more than 1300 studies screened, 53 were identified as potentially relevant and were retrieved for review. Of these, 9 randomized, controlled trials (1294 patients) and 3 nonrandomized trials (938,452 patients) met the inclusion criteria. Six of the 12 randomized trials compared TMR versus MMT, including 967 patients.8–13 Secondary results and/or longer term follow-up of these randomized trials were reported in subsequent papers after the index publication, and all relevant outcomes considered in these papers were included in the systematic review.4,14–22 Three of the 15 randomized trials, reported in 4 papers, compared TMR + CABG versus CABG alone in a total of 327 patients.23–26 Only 3 nonrandomized trials were identified, and each of these compared TMR + CABG versus CABG alone (938,452 patients).27–29 Trials with overlapping or duplicate data were excluded,30 and one trial of TMR plus sympathectomy was excluded.31 Figures 1 and 2 outline the trial selection for randomized and nonrandomized trials, respectively.
Tables 1 and 2 outline the characteristics of the included randomized and nonrandomized trials, respectively. The trials were conducted in Europe and the United States between 1995 and 2003. All included studies were published in English. No unpublished studies were identified. Publication bias was not clearly evident after visual inspection of funnel plots; however, the statistical test for publication bias was underpowered given the small number of included trials.
In TMR versus MMT studies, approximately 11% of patients were lost to follow-up in the studies at 12-month follow-up. At longer-term follow-up (3 to 5 years), about 24% of patients were lost to follow-up. For TMR + CABG versus CABG studies, approximately 3% were excluded or withdrawn during surgery, and 17% were lost to longer-term follow-up (4 to 5 years). Whenever possible, data were analyzed by intention-to-treat. Some studies allowed patients to cross over from MMT to TMR if they failed MMT (ie, developed unstable angina).9,11 Outcomes for these patients were generally reported and analyzed in the MMT arm to which they were originally randomly assigned. In TMR + CABG trials, patients originally randomly assigned to CABG alone but who actually received TMR + CABG were analyzed by intention-to-treat when possible; however, in some trials, the study authors did not report the data for these patients separately.
Table 3 outlines the aggregate baseline characteristics for all patients in the included trials. It is important to note that most trials explicitly excluded patients with unstable angina or with ejection fraction <25% to 35%. In some trials, patients with advanced age (>75 years) were excluded. In randomized studies of TMR versus MMT, baseline characteristics did not differ between groups with the exception of hyperlipidemia. Approximately 7% more patients in the MMT group had hyperlipidemia at baseline compared with TMR. Table 4 outlines the baseline characteristics of TMR + CABG versus CABG trials for randomized trials (Table 4 [a]) and for randomized plus nonrandomized trials combined (Table 4 [b]). Although baseline characteristics did not statistically differ between groups for randomized trials, there were some numeric differences that should be noted: An absolute difference of 12% more patients with diabetes were included in the adjunctive TMR + CABG group, and an absolute increase of 21% of patients in the TRM + CABG groups reported a history of CABG or PCI before study entry. When nonrandomized trials and randomized trials were considered together, there were a number of baseline imbalances that were noted to be of potential concern, including a greater number of patients with acute myocardial infarction (AMI) history, stroke history, diabetes, hyperlipidemia, peripheral vascular disease (PVD), and renal failure for the TMR + CABG group. These baseline imbalances in key prognostic characteristics may be evidence of selection bias, wherein patients who are sicker and have worse prognosis were preferentially selected to be treated with TMR + CABG, perhaps because of the perception that they would benefit most from adjunctive TMR. Given this imbalance resulting from nonrandomized trials, the results of randomized trials will be the major focus of this analysis. When nonrandomized trials are included, this will be made explicit to the reader to allow for interpretation accordingly. Overall, the baseline characteristics of TMR + CABG versus CABG trials appear to favor the CABG arm. Therefore, the results of this analysis probably will represent conservative estimates of the clinical impact of adjunctive TMR. Nonetheless, given that few nonrandomized trials reported outcomes that were of interest in this meta-analysis, the majority of results in this analysis included only randomized data by default.
Important differences in subgroup analyses by laser type or by angina severity at study entry were not found for any clinical or resource-related outcomes. However, the subgroups were underpowered due to the small number of studies available (ie, 3 randomized, controlled trials used CO2 laser; 2 randomized, controlled trials used Ho:YAG laser; 1 study used XeCl laser). Similarly, comparisons between trials based on baseline severity of angina should be interpreted with caution because of the small number of trials in this analysis. Sensitivity analysis for elderly patients, left ventricular function, and publication status was not possible because of inadequate data.
Heterogeneity between trials was found for angina class improvement at 1 year, severe angina at 1 year, QOL at 1 year, MACE at 1 year, myocardial infarction at 1 year, ETT time at 3 months, and readmissions at 1 year for TMR versus MMT. For adjunctive TMR versus CABG, heterogeneity was noted for ICU length of stay. All-cause mortality rates at 30 days showed heterogeneity only when both randomized and nonrandomized studies were combined.
A number of outcomes of interest, including surgical reexploration, transfusions, renal failure, total perioperative complications, patient satisfaction, duration of procedure, emergency department visits, and physician visits, were not adequately reported for meta-analysis.
Clinical Outcomes for TMR Versus MMT
Figure 3 and Table 5 display the results, individually and in aggregate, for all studies that reported on relevant outcomes for TMR versus MMT. The mean angina class was significantly reduced at 3 months [WMD, –1.60; 95% CI, –2.00 to –1.20], 6 months [WMD, –2.00; 95% CI, –2.42 to –1.58], and 12 months [WMD, –1.80; 95% CI, –2.36 to –1.24]. Compared with MMT, TMR resulted in significantly greater number of patients achieving ≥2 class improvement in angina class at 3 months [11.80; 95% CI, 7.57 to 18.30], 6 months [OR, 7.62; 95% CI, 4.79 to 12.11], 1 year [OR, 7.91; 95% CI, 3.33 to 18.80, P < 0.0001], and at 3 to 5 years' follow-up [OR, 9.20; 95% CI, 3.52 to 24.07, P < 0.0001]. The estimated number needed to treat with TMR to result in an additional patient achieving ≥2 class improvement compared with MMT was between 2 to 3 at each time point (3 months, 6 months, 1 year, and 5 years). Patients with continued severe angina (class III or IV) at 1 year was significantly reduced at 3 months [OR, 0.009; 95% CI, 0.001 to 0.074], 1 year [OR, 0.06; 95% CI, 0.01 to 0.27], and 5 years [OR, 0.25; 95% CI, 0.08 to 0.83], for a number needed to treat of approximately 2 for each time point.
Exercise function was reported in few trials. Change in ETT time was significantly improved at 12 months for TMR versus MMT [WMD, 69.00 seconds; 95% CI, 27.14 to 110.86 seconds] and was not adequately reported at earlier time points for meta-analysis. The mean time on ETT did not differ at any time point. Patients with angina that led to limitation of exercise time on ETT was significantly decreased at 3 months [OR, 0.40; 95% CI, 0.26 to 0.60], 6 months [OR, 0.51; 95% CI, 0.27 to 0.95], and 1 year [OR, 0.35; 95% CI, 0.23 to 0.54].
All-cause mortality rate was not significantly different at 30 days [OR, 2.08; 95% CI, 0.85 to 5.09], 3 months [0.74; 95% CI, 0.26 to 2.13], 6 months, [0.72; 95% CI, 0.32 to 1.54], 1 year [OR, 1.08; 95% CI, 0.60 to 1.95], or 3 to 5 years [OR, 0.65; 95% CI, 0.41 to 1.05]. Major adverse coronary events (MACE) were significantly reduced with TMR compared with MMT at 1 year [OR, 0.23; 95% CI,0.15 to 0.34]. No significant difference was found for stroke and myocardial infarction. Heart failure (defined as the need for new prescription of a diuretic or doubling of a preexisting diuretic at 12 months in one trial,10 and hospitalization for heart failure at 3 to 5 years in another trial16) was significantly increased at 1 to 5 years [OR, 2.65; 95% CI, 1.45 to 4.85]. Mean left ventricular ejection fraction was reported in only one study at 3 months and was significantly lower for TMR versus MMT [WMD, –5.30; 95% CI, –10.31 to –0.29]. At 1 year, mean ventricular ejection fraction was not significantly different between TMR and MMT [WMD, –2.44; 95% CI, –5.28 to 0.40].
Inadequate information was provided regarding the results of imaging studies, and the data were not amenable to combination through meta-analysis. Although one trial showed improved perfusion with TMR versus MMT,11 this trial was confounded by small sample size and relatively large loss to follow-up. Four additional randomized trials reporting the results of perfusion imaging studies did not show significant improvement with TMR versus MMT, and a number of these trials also had large loss to follow-up.
Hospital readmissions within 1 year were significantly reduced [OR, 0.27; 95% CI, 0.17 to 0.40], and need for reintervention for ischemia was significantly reduced for TMR versus MMT at 5 years [OR, 0.43; 95% CI, 0.25 to 0.72]. Hospital and ICU length of stay data were not provided in the randomized trials of TMR versus MMT. One study reported costs and cost-utility and found that TMR required an additional cost of £8901 per patient at 1 year, and the estimated incremental cost-effectiveness of TMR versus MMT was £228,000/QALY, which is beyond the threshold typically considered cost-effective in the United Kingdom.
Quality of life scores for TMR versus MMT at 1 year were reported in four randomized trials (Table 6). One trial reported SF-36 scores at 12 months,11 one trial reported EuroQOL scores at 12 months,13 and two trials reported SAQ scores at 12 months,11,13 and two trials reported summary QOL scores at 12 months.9,13 Summary QOL scores were significantly reduced at 1 year (Table 6). The physical limitation domain of the SAQ score was significantly improved at 1 year. However, improvements in other domains of the SAQ score were not statistically significant. SF-36 scores at 1 year showed significant improvement in the physical component score. EuroQOL at 1 year showed significant improvement in two of five domains.
TMR + CABG Versus CABG Alone
Table 7 and Figure 4 outline the results for TMR + CABG versus CABG alone. Nonrandomized studies contributed data only for the following outcomes: QOL scores and mortality. The forest plots present randomized and nonrandomized data separately for QOL and mortality.
Angina class improved similarly in both groups and was not significantly different at 12 months for TMR + CABG versus CABG alone [WMD, 0.01; 95% CI, –0.20 to 0.23]. At 5 years, there was significantly greater angina improvement in angina class with TMR + CABG versus CABG alone; however, the absolute difference was small [WMD, –0.21; 95% CI, –0.39 to –0.03]. Patients with continuing severe angina (class III or IV) at 1 year [OR, 0.45; 95% CI, 0.16 to 1.23] and at 4 to 5 years [OR, 0.30; 95% CI, 0.007 to 13.28] were not significantly different between groups but were reported in only two randomized trials. The proportion of patients achieving ≥2 angina class improvement was not reported in the studies of TMR + CABG versus CABG.
In randomized trials, all-cause mortality at 30 days was significantly decreased with TMR + CABG versus CABG alone [OR, 0.27; 95% CI, 0.10 to 0.77]; however, survival at 1 year [OR, 0.66; 95% CI, 0.27 to 1.10] and at 4 to 5 years [OR, 0.91; 95% CI, 0.48 to 1.72] was not significantly different in randomized trials. Nonrandomized trials, on the other hand, suggested decreased survival for TMR + CABG versus CABG at 30 days; however, this was largely weighted by the results from the STS database, which showed that higher-risk patients with more diffuse coronary artery disease were selectively chosen for TMR + CABG, and this mortality difference did not remain after adjustment for confounding prognostic factors.28 In addition, in the STS database, about 46% of patients treated with TMR + CABG had unstable angina at baseline, which is in contrast to the randomized, controlled trials wherein patients with unstable angina were generally excluded. In the STS database, patients with unstable angina had higher mortality rates than those without.28 These factors probably explain the heterogeneity in mortality rates observed between randomized and nonrandomized trials.
MACE was significantly reduced at 30 days [OR, 0.31; 95% CI, 0.10 to 0.99] but not at 1 year [OR, 0.70; 95% CI, 0.36 to 1.37] or 4 years [OR, 0.27; 95% CI, 0.07 to 1.14]. Other morbidities, including AMI, stroke, cardiac arrest, tamponade, heart failure, and renal failure were not statistically significantly different between groups (Table 7).
Exercise treadmill testing time was reported in only one randomized trial and did not differ significantly between TMR + CABG and CABG at 12 months [WMD, 0.08 minutes; –0.16 to 0.32]. Change in ETT time from baseline to 12 to 18 months was reported in two randomized trials and was significantly improved for TMR + CABG [WMD, 88.05 seconds; 95% CI 52.79 to 123.32 seconds]. Similar improvement in ETT time was shown at 6 months. At 5 years, the change in ETT was not significantly different between TMR + CABG versus CABG alone.
Cardiopulmonary bypass time, ventilation time, and ICU length of stay did not differ between groups. Hospital length of stay was significantly increased with TMR + CABG versus CABG [WMD, 1.55 days; 95% CI, 0.96 to 2.13 days]. Need for reintervention, reoperation, excessive bleeding, and readmissions did not significantly differ for TMR + CABG versus CABG (Table 7).
Quality of life was reported in only one nonrandomized trial. Although most scores were improved with TMR + CABG at 1 year, none of the domains reached statistical significance (Table 8).
This systematic review provides a synthesis of the current evidence comparing TMR with conventional management of patients with refractory angina. In summary, this meta-analysis suggests that TMR provides advantages for selected clinical and resource-related outcomes in patients with stable angina caused by coronary morphology that is not amenable to conventional medical or surgical management.
Clinical Outcomes for TMR Versus MMT
Transmyocardial laser revascularization significantly improves angina score (from baseline NYHA/CCS class of 3 to 4, down to NYHA/CCS class of 1 to 2 at 3 months, 6 months, and 12 months) and significantly increases the proportion of patients achieving at least 2 angina class improvement as early as 30 days and up to 3 to 5 years. The number of patients with continued severe angina (class III or IV) was significantly decreased. For each of these end points, the reduction in risk was statistically significant and clinically relevant.
Assuming that these results are applicable to patients of similar characteristics outside of the clinical trial setting, this meta-analysis predicts that for every 1000 patients who undergo TMR instead of receiving only MMT, there would be on average 345 more patients with ≥2 class improvement in angina scores at 1 year, 476 fewer patients with severe angina at 1 year, 357 fewer with MACE at 1 year, and 178 fewer patients requiring reintervention at 5 years. In addition, exercise functionality and selected domains of quality of life may be improved for patients receiving TMR instead of MMT. The outcomes did not appear to be related to type of laser modality and severity of angina at entry, although the small number of trials available precluded adequate power to rule out important differences in the subgroup analyses.
Clinical Outcomes for TMR + CABG Versus CABG Alone
Compared with CABG alone, adjunctive TMR reduced MACE at 30 days, improved survival at 30 days, and improved exercise tolerance at 6 months and 12 to 18 months. However, hospital length of stay was significantly increased with TMR + CABG (1.6 days). Extrapolating these results to a population of patients with baseline characteristics similar to the randomized trials in this meta-analysis suggests that for every 1000 patients treated with adjunctive TMR instead of CABG alone, there would be approximately 60 fewer MACE and 74 fewer deaths at 30 days, but at a cost of increased hospital length of stay.
Strengths and Limitations
Although this meta-analysis represents a comprehensive analysis of currently available randomized and nonrandomized evidence, it is likely that new evidence will emerge to better inform the longer-term outcomes, and this newer evidence will need to be incorporated over time. This is particularly important to note, since there exist few trials of TMR to date and additional trials (particularly of TMR + CABG versus CABG) may eventually change the results of this meta-analysis. Since there was consistency among trials of TMR versus MMT for symptom reduction, it is less likely that conclusions regarding symptom reduction will change as more trials become available. On the other hand, given the very few trials of TMR + CABG available to date, additional trials of adjunctive TMR could change the conclusions for its impact on clinical outcomes.
In addition, it is also important to note that the results of this analysis are based on the techniques and skills of the surgeons who administered TMR within the randomized trials, and the results described herein are no guarantee for outcomes produced by individual surgeons or assistants who may have less technical competence or experience than those represented within the trials. It was not clear in the included trials whether surgeons and technicians applying laser therapy were early in their experience or whether they had significant experience before conducting the trial. Since TMR is a relatively new technology, it probably is safe to suggest that the results of this meta-analysis provide a more conservative estimate of the potential benefit than might be possible once significant time has passed to allow operators to gain further experience with the differing laser technologies and techniques.
Another significant limitation of the results of this meta-analysis is that some of the prognostically significant characteristics of the patients differed at baseline in the TMR + CABG versus CABG analysis. Even when the studies were limited to randomized trials, the imbalance was noticeable and tended to have fewer sick patients in the CABG alone group at baseline. Therefore, the results of this analysis probably represent a conservative estimate of the potential efficacy of TMR + CABG, given that patients with worse prognostic factors at baseline weighed more heavily on the TMR + CABG side. Given this baseline imbalance, any benefits that were found for TMR + CABG may be considered a likely underestimate. Further balanced, randomized trials should be conducted to better define the efficacy of TMR + CABG. For TMR versus MMT, there were more patients with hyperlipidemia in the MMT group at baseline. Whether this imbalance confounded outcomes measurement remains unclear, and the results should be interpreted accordingly. In particular, the additional 7% of patients who had hyperlipidemia in the MMT group may have been more likely to develop angina, reinterventions, MACE, and readmissions. Nonetheless, the absolute difference between TMR and MMT groups for these outcomes generally exceeded 7% by severalfold, and correction for this baseline imbalance would not change the conclusions.
Randomized trials generally excluded patients with low ejection fraction, advanced age (>75 years), and recent myocardial infarction. All randomized trials, except for Frazier et al,11 excluded patients with unstable angina. Whether these results are generalizable to these excluded populations remains speculative.
Perioperative mortality rates observed within the randomized trials were generally low and comparable across studies. The low mortality rates probably were due to the strict selection criteria, especially the exclusion of patients with low ejection fraction, unstable angina, and advanced age. Although survival differences did not reach significance, the direction of effect for TMR versus MMT changed over the course of time. Mortality rate at 30 days showed a trend toward increase with TMR [OR, 2.08; 95% CI, 0.85 to 5.09, P = 0.11], whereas mortality rate at 3 to 5 years showed a trend toward reduction [OR, 0.65; 95% CI, 0.41 to 1.05]. Given these data, the possibility that there may exist a true increase in early mortality rate cannot be ruled out at this time. However, if the difference does exist, it is not large (3.8% mortality rate for TMR versus 1.8% mortality rate for MMT, for an overall absolute difference of 2.0%), and the overall benefit is in favor of TMR rather than MMT over the longer term. Therefore, if there is an early perioperative excess risk of mortality, the 3- to 5-year data suggest that this risk is outweighed by improved overall survival over the longer term. On the other hand, for TMR + CABG, there appeared to be an early survival benefit by adding TMR to CABG; however, this result should be interpreted with caution, as it is due primarily to an increase in mortality rate observed in the CABG-alone arm from one study23 and could be the play of chance alone. Further randomized trials are needed to better define the time course of survival for TMR versus conventional management.
Exercise tolerance was reported in most randomized trials up to 12 months; however, some trials reported the total time on ETT, whereas others reported the change from baseline in time on ETT. Change in ETT time was significantly improved by 69 seconds at 12 months for TMR versus MMT, whereas the mean time on ETT did not differ at any time point. For TMR + CABG versus CABG alone, ETT was improved 80 to 150 seconds at 6 months to 18 months. The clinical relevance of a 69- to 150-second improvement on ETT might be questioned. Nevertheless, the fact that there was a 50% to 65% reduced risk of experiencing angina that limited time on ETT at 30 days, 6 months, and 12 months suggests that the gains in exercise tolerance might be clinically relevant. An important caveat in interpreting the results of exercise tolerance over the longer term is that there was increasing and unbalanced loss to follow-up of patients over time that may affect the validity of the estimates. Furthermore, these trials did not generally involve blinded assessment of exercise tolerance. In addition, in two of the trials, patients were allowed to cross over from MMT to TMR.
Heart failure was measured in only two trials of TMR versus MMT and was variously defined. Combined analysis of these trials suggested that heart failure was significantly increased at 1 to 5 years [number needed to harm = 6]. Mean left ventricular ejection fraction was significantly reduced at 3 months and not at 12 months. Interpretation of these results is difficult because heart failure and ventricular function were reported in few trials, and the trials were confounded by large loss to follow-up. It has been previously demonstrated that TMR may increase the risk of postoperative heart failure, and the purported mechanism includes inflammatory reaction, edema, or reversible myocardial injury.29,32 Regardless of the mechanism, the potential for TMR-induced heart failure remains a concern that should be studied further so that the risk-benefit ratio of TMR can be better delineated. Overall, the results at this time suggest that there may be a small increase in risk of decreased left ventricular function within the first year. Whether this risk attenuates over time or whether the results have been confounded by incomplete follow-up remains to be addressed. Measures of heart failure were not assessed in trials of adjunctive TMR versus CABG.
Laser Technique and Subgroup Analyses
The paucity of available comparative trials did not allow for adequate subanalyses by laser type. Although there are purported benefits for each modality, the clinical relevance of these remains unknown. For example, the CO2 and Ho:YAG lasers deliver thermal energy to create channels in the myocardium, whereas the XeCl laser ablates tissue by dissociating molecular bonds. Each type of laser resulted in reduced angina symptoms in the included randomized trials of TMR versus MMT. However, direct comparisons between lasers have not yet been undertaken in clinical trials. Therefore, conclusions about which type of laser provides superior outcomes remain unestablished at this time.
Laser revascularization can be performed through open heart surgery or percutaneously by a catheter-based technique. This analysis focused only on surgical TMR and excluded percutaneous transmyocardial laser revascularizations. Therefore, conclusions regarding the relative efficacy of differing routes of applying laser revascularization cannot be drawn from this analysis. At least four randomized trials of PMR have been published, and the impact on outcomes has been inconsistent.33–36
Number of Channels
The optimal number and spacing of channels has yet to be determined. In the trials included in this meta-analysis, the mean number of channels varied from 18 to 47, and the intended concentration of channels throughout the target myocardium was not always reported. Further research will be required to delineate the relation between clinical outcomes and concentration of laser channels.
Perfusion was typically assessed with the use of 201SPECT or 99SPECT imaging. Of the six included randomized trials evaluating TMR versus MMT, only one showed a significant improvement in myocardial perfusion,11 and all of the trials were confounded by loss to follow-up before patients underwent imaging studies. The reasons for the observed heterogeneity in perfusion remains unknown and is consistent with the variation in observations in previous nonrandomized clinical investigations, some of which report demonstrable increased perfusion in ischemic myocardium after TMR and others of which report no change.29 One randomized trial and one nonrandomized trial of adjunctive TMR evaluated perfusion, and both found no difference.26,29
Considered in aggregate, the results of this systematic review and meta-analysis contribute little to our understanding of the mechanism of action for TMR and its impact on perfusion. The increased exercise tolerance that was found in this meta-analysis does not prove that the myocardium is better perfused after TMR since increased exercise tolerance can be related to multiple underlying mechanisms along with altered patient perception of angina.
The impact of TMR on resource-related outcomes was difficult to assess in this meta-analysis because of the paucity of trials reporting relevant resource and economic outcomes. The need for readmission was significantly reduced at 1 year, and reintervention for ischemia was significantly reduced at 5 years [OR, 0.43; 95% CI, 0.25 to 0.72] for TMR versus MMT. For adjunctive TMR, the impact on readmissions and reinterventions was nonsignificant. Hospital and ICU length of stay data was not provided in the randomized trials of TMR versus MMT; however, the total length of stay was increased significantly by over 1.5 days for adjunctive TMR. Nevertheless, this latter result was derived from only two randomized trials, and the generalizability to other institutions with contemporary protocols for early discharge criteria remains unclear. Campbell et al21 completed a formal cost-utility analysis of patients included in the randomized trial of TMR versus MMT by Schofield et al12 The results of the economic analysis estimated that TMR cost an additional £8901 versus MMT at 1 year and resulted in a QALY improvement of +0.039, with an estimated overall incremental £228,000/QALY. As a result, Campbell and colleagues concluded that TMR was an inefficient use of UK resources, and this conclusion was robust across varied assumptions of costs and effects.
A number of important methodologic limitations of the randomized trials included in this meta-analysis need to be highlighted, including cross-overs, loss to follow-up, and lack of blinding. In two trials, patients were allowed to cross over from the MMT group to receive TMR if they failed maximal medical therapy. In most trials, for various reasons, some patients were excluded or lost to follow-up (about 11% at 1-year follow-up for MMT trials). Although the cross-overs may have resulted in more conservative estimates of effect, the overall impact of loss to follow-up and exclusions on outcomes estimates is unknown. Furthermore, although blinded, sham-controlled surgical trials are difficult to perform, the lack of blinding does raise concern regarding a possible placebo effect.37,38
A Consensus Conference and Statement on TMR
A consensus conference was facilitated by the International Society for Minimally Invasive Cardiac Surgery (ISMICS) to clarify the role of TMR in patients with angina refractory to conventional medical or surgical management.39 After consideration of this systematic review, the ISMICS Consensus Conference suggested that in stable patients with refractory severe angina not amenable to conventional revascularization, TMR can be recommended instead of MMT to improve sustained angina relief [class I, level A evidence], reduce MACE and improve exercise performance [class I, level A evidence], and reduce readmissions and reinterventions [class IIa, level B evidence]. In patients with diffuse coronary artery disease who cannot be completely revascularized by CABG alone, adjunctive TMR + CABG can be recommended to improve long-term angina relief [class IIa, level A evidence], reduce 30-day mortality rates and MACE [class IIa, level A/B evidence], and to improve 1-year exercise performance [class IIa, level A evidence]. The consensus panel stated that TMR represents a viable alternative to conventional maximal medical treatment for patients with refractory stable angina not amenable to conventional surgical revascularization [class I, level B] and that adjunctive TMR + CABG can be considered a part of the therapeutic armamentarium in chronic stable angina patients with coronary morphology deemed only partially amenable to revascularization by conventional surgical revascularization [class IIa, level B].
An earlier published practice guideline by the Society of Thoracic Surgeons suggested that transmyocardial laser revascularization may be acceptable as sole therapy for a subset of patients with refractory angina and as an adjunct to CABG surgery for a subset of patients with angina who cannot be completely revascularized surgically.40
Despite the proven success of conventional revascularization for patients with coronary artery disease via PCI or CABG, a significant number of patients with diffuse disease cannot be successfully revascularized and many have refractory and debilitating angina symptoms despite maximal conventional management strategies (ie, drug therapies for patients not amenable to surgical revascularization or CABG with incomplete revascularization for patients only partially amenable to surgical bypass).
For patients with stable class III or IV angina who are not candidates for CABG due to diffuse morphology, transmyocardial laser revascularization can significantly reduce angina symptoms, and possibly improve exercise tolerance, compared with maximal medical treatment alone. Reinterventions and readmissions may also be reduced; however, the short-term risk of heart failure may be slightly increased.
For patients who have coronary artery morphology only partially amenable to revascularization through CABG, adjunctive TMR may also reduce MACE and improve exercise tolerance and survival over the short term. The role of adjunctive TMR remains unclear for angina symptom reduction because of the limited number of randomized, controlled trials.
© 2006 Lippincott Williams & Wilkins, Inc.