Author + information
- Received January 5, 2005
- Revision received June 23, 2005
- Accepted June 27, 2005
- Published online November 15, 2005.
- Martin B. Leon, MD⁎ (, )
- Ran Kornowski, MD,
- William E. Downey, MD,
- Giora Weisz, MD,
- Donald S. Baim, MD,
- Robert O. Bonow, MD,
- Robert C. Hendel, MD,
- David J. Cohen, MD, MSc,
- Ernest Gervino, DSc,
- Roger Laham, MD,
- Nicholas J. Lembo, MD,
- Jeffrey W. Moses, MD and
- Richard E. Kuntz, MD, MSc
- ↵⁎Reprint requests and correspondence:
Dr. Martin B. Leon, Cardiovascular Research Foundation, Center for Interventional Vascular Therapy, Columbia University Medical Center, 161 Fort Washington Avenue, New York, New York 10032.
Objectives This study was a randomized, patient- and evaluator-blinded, placebo-controlled trial in patients treated using percutaneous myocardial laser revascularization.
Background Previous studies using similar therapies have been confounded by placebo bias.
Methods A total of 298 patients with severe angina were randomly assigned to receive low-dose or high-dose myocardial laser channels or no laser channels, blinded as a sham procedure. The primary end point was the change in exercise duration from baseline examination to six months.
Results The incidence of 30-day death, stroke, myocardial infarction, coronary revascularization, or left ventricular perforation occurred in two patients in the placebo, eight patients in the low-dose, and four patients in the high-dose groups (p = 0.12); 30-day myocardial infarction incidence was higher in patients receiving either low-dose or high-dose laser (nine patients) compared with placebo (no patients, p = 0.03). At six months, there were no differences in the change in exercise duration between those receiving a sham (28.0 s, n = 100), low-dose laser (33.2 s, n = 98), or high-dose laser (28.0 s, n = 98, p = 0.94) procedure. There were also no differences in the proportion of patients improving to better than Canadian Cardiovascular Society class III angina symptoms at six months. The follow-up visual summed stress single-photon-emission computed tomography scores were not significantly different from baseline in any group and were no different between groups. The modest improvement in angina symptoms assessed by the Seattle Angina Questionnaire also was not statistically different among the arms.
Conclusions Treatment with percutaneous myocardial laser revascularization provides no benefit beyond that of a similar sham procedure in patients blinded to their treatment status.
More than 100,000 patients each year have refractory or end-stage angina not amenable to percutaneous coronary intervention or coronary artery bypass graft surgery in the U.S. alone (1).
Surgical transmyocardial laser revascularization was developed to address this population of “no option” patients. Previous clinical trials have repeatedly shown improved angina and have also shown improved exercise tolerance in patients treated with surgical transmyocardial laser revascularization (2–19). A catheter-based laser revascularization approach was developed to provide equal clinical efficacy without the need for a thoracotomy or general anesthesia, thereby reducing procedural complications and perhaps costs.
The Biosense direct myocardial revascularization (DMR) holmium: yttrium-aluminum-garnet (Ho:YAG) laser system (Biosense-Webster, Diamond Bar, California) uses a magnet-based catheter tip location system to allow the precise placement of intra-myocardial laser channels (20–22). A phase I study of 77 patients treated with this system showed favorable results at six months, with a significant improvement in both angina symptoms and exercise duration (23).
The possibility that placebo effects could be responsible for the positive results of the previous unblinded surgical and percutaneous laser revascularization trials has been raised. Therefore, the present study was designed to evaluate the safety and efficacy of Biosense DMR using a patient-blinded and evaluator-blinded randomized trial design.
This was a phase II, multicenter, randomized study, with prospective comparison of two treatment groups (low-dose and high-dose) and a control group with blinded patient and end point evaluation. No crossover was allowed, and all patients received the intended therapy. The study was conducted at 14 U.S. centers with experience in obtaining Biosense LV electromechanical maps (see Appendix). The trial was approved for an investigational device exemption by the U.S. Food and Drug Administration, and all sites received approval from their local institutional review boards. Written informed consent was obtained from all patients. Patient enrollment was initiated in October 1998 and was completed in October 1999.
Patient eligibility criteria
Patients were eligible for inclusion in the study if they had a history of coronary artery disease with refractory angina (Canadian Cardiovascular Society [CCS] class III or IV), despite optimal medical therapy (including beta-adrenergic blockers, calcium antagonists, and nitrates). All patients were considered unacceptable candidates for percutaneous revascularization therapies or surgical revascularization procedures. All patients were able to complete a minimum of 2 min but not more than 12 min of a Beth Israel Modified Bruce Protocol exercise test, and had reversible ischemia during dual isotope perfusion imaging studies.
The principal exclusion criteria from the study included severe left ventricular dysfunction (ejection fraction <30%, assessed by echocardiography), recent myocardial infarction (within 30 days of treatment), Braunwald class IIIb unstable angina, chronic atrial fibrillation, prosthetic valve or significant aortic valve pathology, myocardial wall thickness <9 mm (by transthoracic echocardiography), left ventricular thrombus, and major life-threatening comorbidity.
Study end points
The pre-specified primary end point was change in exercise treadmill duration from baseline to six months. Pre-specified secondary end points included: 1) major adverse cardiac events (MACE) at 30 days (the composite of cardiac death, acute Q-wave and non–Q-wave myocardial infarction [MI], revascularization procedures for procedure-related complications or coronary ischemia, left ventricular [LV] perforation, and stroke); 2) the incidence of MACE at 6 and 12 months; 3) the change in CCS angina classification score from baseline to 6 and 12 months; 4) the change in health status and anginal symptoms using a semi-quantitative instrument incorporating the Medical Outcome Study, General Health Survey (SF-12), and Seattle Anginal Questionnaire from baseline to 6 and 12 months; 5) the change in global and regional radionuclide perfusion scores at rest, peak stress, and redistribution using semi-quantitative and quantitative methods from baseline to 6 months; and 6) changes in all exercise test parameters including duration, time to angina, and time to ST-segment changes from baseline to 6 and 12 months.
Myocardial perfusion imaging
At baseline and at six months after the randomized DMR procedure, all patients underwent dual-isotope single-photon-emission computed tomography (SPECT) myocardial perfusion imaging with 3 mCi 201Tl at rest and 25 mCi 99mTc-sestamibi for adenosine stress imaging according to previously published methodology (24). Sestamibi was administered at three minutes during the six-minute adenosine infusion (140 μg·kg−1·min−1). Stress imaging was performed beginning 60 min after 99mTc-sestamibi injection. These data were submitted to the central core laboratory for blinded uniform processing, interpretation, and comparative analysis. A 20-segment semiquantitative visual analysis was used as well as an automated quantitative analysis. The summed scores of rest, rest redistribution, and stress images were determined at six-month follow-up and compared with baseline values.
LV mapping and the randomized DMR procedure
All patients were heavily sedated, had blindfolds, and had earphones with music to minimize the potential for inadvertent identification of the treatment modality. The LV electromechanical mapping was performed according to previously described guidelines (20–23). Heparin was administered (70 U/kg) and supplemented as needed to maintain an activated clotting time of 200 to 250 s. The acquisition of an acceptable LV map (70 to 100 points) usually required approximately 30 min. Treatment zones, generally subtending the anatomic distribution of one of the three major epicardial coronary arteries (left anterior descending, left circumflex, and/or right coronary artery), were pre-specified using the combination of a recent coronary angiogram, the SPECT imaging results, and the diagnostic LV electromechanical map (23). Areas of previous infarction (Q waves on the electrocardiogram, myocardial wall thickness <9 mm, or voltage amplitudes of <5 mV) were carefully excluded as treatment zones. The DMR was performed in one or two designated treatment zones in each patient.
After completing the LV mapping procedure and deciding on the potential DMR treatment zones, the patients were randomized to placebo (mock procedure, 102 patients), low-dose (98 patients), or high-dose (98 patients) DMR treatment. If the patient was randomized to placebo, the laser (already in the room) was turned on but no further procedure was performed. If the patient was randomized to receive the Biosense DMR treatment, a laser catheter was introduced and advanced to the LV. The laser source was a pulsed Ho:YAG laser (Sharplan 2040; Ho:YAG Laser Systems, Lumenis, New York, New York). Laser channels were created using a tip-deflecting mapping and Ho:YAG laser catheter integrated with a 300 μm fiber (LaserStar; Biosense-Webster). A single laser pulse (2 J/pulse) was fired perpendicular to the endocardial surface with the catheter tip icon used to verify the location. Either 10 to 15 (low-dose) or 20 to 25 (high-dose) laser pulses were fired within the treatment zone with 5- to 10-mm spacing between each laser pulse. At the conclusion of the laser DMR procedure (usually lasting an additional 30 min), all patients were monitored (electrocardiogram and arterial pressures) in the catheterization laboratory for 15 min. Patient blinding to the procedural randomization code was preserved by continuous heavy sedation, sensory isolation (blindfold and music), and lack of patient familiarity with procedural details and duration (i.e., patients could not differentiate the LV mapping procedure alone from the LV mapping plus DMR procedure).
At the end of the procedure, blood samples for creatine kinase-MB were acquired from all patients every 8 h for 24 h after the procedure. In addition, 12-lead electrocardiograms were obtained at the end of the procedure and the morning after the procedure.
Follow-up assessments and data collection
All follow-up assessments and data were collected by study coordinators and research assistants blinded to treatment identity. Major adverse cardiovascular events were recorded in-hospital, at 30 days, and at 3, 6, and 12 months. Exercise tests and CCS angina class were obtained at 3, 6, and 12 months (core laboratory: Beth Israel-Deaconess Hospital, Boston, Massachusetts). Dual-isotope adenosine stress radioisotope perfusion imaging was repeated at six months (core laboratory: Northwestern University, Chicago, Illinois). Standardized quality of life assessments were obtained by independent research monitors at 6 and 12 months (core laboratory: Cardiovascular Data Analysis Center, Boston, Massachusetts).
Case-report forms were completed at each site, monitored by independent study monitors, and submitted to the data-coordinating center (Cardiovascular Data Analysis Center, Harvard Clinical Research Institute, Boston, Massachusetts). All events were classified by an independent clinical events committee that was unaware of each patient’s treatment assignment. Angiograms obtained before the procedure were submitted to an angiographic core laboratory (Cardiovascular Research Foundation, New York, New York) for independent assessment.
The sample size was based on 80% power to detect a 15% relative increase in exercise duration from an expected baseline of 7.4 ± 2 min (at α = 0.05). The 15% relative increase was chosen because it represents a difference of approximately 1 min, which was thought to be the minimal clinically relevant difference in exercise duration.
Continuous variables including exercise duration, change in exercise duration, Seattle Angina Questionnaire scores, and SPECT myocardial perfusion scores were compared by analysis of variance under the null hypothesis of no difference among the arms, with a post-test procedure for pairwise comparisons to be performed whenever the analysis of variance false-positive rate was <5%. The binary end points of MACE and ≥2 CCS class change were compared using exact methods. A secondary analysis of the primary end point was undertaken using multivariable linear regression to control for differences in selected baseline covariates. These selected covariates were chosen if they differed between groups at a significance level of p < 0.20 (prior MI, percutaneous coronary intervention, or coronary artery bypass graft) or they were thought to be potentially important clinical indicators of the change in exercise performance (baseline exercise treadmill test duration, ejection fraction, diabetes, and symptom presentation).
All statistical tests were two-sided, with a p value ≤ 0.05 considered significant. Analyses were performed with SAS software (version 8.2, SAS Institute Inc., Cary, North Carolina).
Two hundred ninety-eight patients were enrolled at 14 investigational sites; 98 were assigned to the low-dose DMR group, 98 patients to the high-dose DMR group, and 102 patients to the placebo group.
The mean patient age was 62.9 ± 10.1 years, 77% were male, and hypertension, diabetes mellitus, and hyperlipidemia were present in 73.5%, 43.9%, and 83.2% of patients, respectively. Prior myocardial infarction and prior coronary artery bypass graft operation were reported in 67.0% and 88.3% of patients, respectively. The mean left ventricular ejection fraction was 49.3 ± 12.0%. All baseline patient demographics were similar in the three patient groups.
Full electromechanical maps were obtained without complications in all but one patient, in whom the procedure was aborted after transient right side weakness and aphasia developed. The overall number of mapping points was 69 ± 19 (range, 1 to 136), and the LV mapping time before DMR was 33 ± 14 min (range, 6 to 85 min). In the treatment phase of the study, the low-dose patients received an average of 21 ± 8 laser channels in an average of 1.4 ± 0.5 treatment zones. The high-dose patients received an average of 34 ± 11 laser channels in an average of 1.5 ± 0.5 treatment zones.
In-hospital and 30-day clinical events
In-hospital procedure-related adverse events in the combined active treatment groups (low-dose or high-dose) included death in one patient (0.5%), Q-wave or non–Q-wave myocardial infarctions in seven patients (3.6%), strokes in two patients (1.0%), and LV perforation in one patient (0.5%). In the placebo patients, other than vascular complications in two patients (2.0%), there were no other adverse in-hospital clinical events. The primary safety end point (30-day MACE) occurred in 4.1% of the low-dose patients, in 8.2% of the high-dose patients, and in 2.0% of the placebo patients (p = 0.117). Myocardial infarction (Q-wave or non–Q-wave) occurred in nine of the DMR-treated patients and in none of the placebo patients (p = 0.026 at 30 days). The in-hospital and 30-day adverse events are detailed in Table 1.
Follow-up clinical events
Six and 12 months after the index procedure, there were no statistically significant differences in cumulative death, acute MI, or repeat revascularization among the three treatment groups (Table 2,Fig. 1).At 12 months, death or acute MI occurred in 9.2%, 7.2%, and 7.8% of patients in the high-dose, low-dose, and placebo treatment groups, respectively. Of 13 deaths (4 high-dose, 4 low-dose, 5 placebo), 9 were cardiac with similar distribution between groups. Death occurred at a mean of 195 days (range, 5 to 360 days) after the procedure. Importantly, there were no significant differences in repeat revascularization events among the three treatment groups at 6 or 12 months after the index procedure.
Changes in exercise parameters
The primary end point was similar for the three groups, and importantly, there was a similar significant improvement in exercise duration during the 6-month follow-up for both the active and placebo treatment patients. This significant improvement was maintained for all treatment groups during the 12-month assessments (Table 3,Fig. 2).Other exercise parameters evaluated (time to angina and time to ST-segment depression) improved significantly from baseline to 6 months and from baseline to 12 months, but the improvements were similar, without any consistent differences among the three groups at 6 and 12 months (Table 3).
Changes in angina severity
Significant improvement in CCS angina class was noted for all treatment groups at 6 months (to CCS class 2.0 ± 1.2, 1.9 ± 1.3, and 2.2 ± 1.2, in high-dose, low-dose, and placebo groups, respectively, p = 0.413), which was maintained during the 12-month assessment. Improvement of at least two CCS angina classes was noted in 41%, 48%, and 41% of the high-dose, low-dose, and placebo groups, respectively, at six months. Figure 3shows that the proportions of patients with CCS III to IV angina at 6 and 12 months were similar in all three groups.
Analysis of quality-of-life assessment instruments indicated significant improvements at 6 and 12 months for all of the treatment groups, but no differences among the three groups (Table 4).Specifically, angina frequency and stability, physical functioning, treatment satisfaction, disease perception, and current overall health were all improved, but without indications of a DMR treatment effect.
Radioisotope imaging studies
During the follow-up radioisotope imaging studies, quantitative scores of the mean ischemic areas per patient, at rest and during stress, showed no significant changes suggesting an anti-ischemic effect, and were similar among the three treatment groups (at rest, 7.5 ± 5.2, 8.7 ± 7.9, 7.3 ± 6.2, p = 0.390; during stress, 17.7 ± 8, 19.3 ± 9.5, 17.3 ± 7.6, p = 0.345 for high-dose, low-dose, and placebo patients, respectively).
The present study
This study was the first randomized, placebo-controlled, blinded assessment of DMR in patients with refractory angina. A comparison of placebo (sham) with two treatment groups (differing in the numbers of laser channels) using the Biosense DMR system showed no differences in exercise duration (primary end point), exercise time to the onset of chest pain, and exercise time to the appearance of ST-segment changes at 6 and 12 months. There were also no differences among the treatment groups in the improvement in angina frequency and in all other quality-of-life parameters. Moreover, there was an increased rate of adverse clinical events associated with laser treatment at 30 days. There were also no changes in myocardial perfusion, as manifested by rest and stress SPECT nuclear imaging studies done at 6 months.
Previous laser transmyocardial laser revascularization (TMR) clinical studies
The negative results from this blinded randomized trial are in striking contrast to an extensive literature purporting to show the clinical benefits of either surgical or catheter-based laser myocardial revascularization procedures in patients with end-stage symptomatic coronary ischemia. Using surgical TMR modalities to treat “no option” patients, there have been 11 observational studies with at least one-year clinical follow-up (2–12) and six non-blinded randomized studies (13–18) versus best medical therapy. The most consistent beneficial finding was a significant improvement in CCS angina class, which seemed to peak by six months, but sustained angina benefit at three and five years has also been reported (8,25). Exercise tests, performed in approximately half the studies, also showed improved exercise duration, but myocardial perfusion assessments were unsuccessful in showing improved regional or global myocardial perfusion after TMR therapy.
There have now been five randomized clinical trials (including the present study), involving 1,072 patients, examining the safety and efficacy of percutaneous laser myocardial revascularization (26–29). The first two studies (26,27) were non-blinded versus best medical therapy and showed improved angina symptoms and exercise duration. Importantly, the percutaneous TMR procedures were not without associated complications; in the present study, there were increased acute adverse events in the laser-treated patients, and in another study (27), there were five episodes (3% of the TMR patients) of myocardial perforation resulting in cardiac tamponade. In addition to the present study, there are two other percutaneous TMR studies that have used a blinded sham control arm to eliminate the influence of placebo effects on outcomes (28,29). In a study involving 141 patients with failed recanalization of chronic total occlusions (28), no differences were observed in follow-up angina symptoms and in exercise test parameters between the two groups. In a smaller study (82 patients) (29) using a true placebo sham procedure, there was an improvement in CCS angina scores associated with laser treatment. However, exercise test results and anti-anginal medication usage were similar between the two groups at 12 months.
Study result comparisons
Clearly, there are many discrepant findings when comparing blinded and non-blinded percutaneous TMR trials, and more generally, when comparing the ambiguous results of the percutaneous randomized trials with the more consistent positive outcomes from the surgical randomized (but non-blinded) TMR studies. Arguments forwarded to explain the differences between percutaneous and surgical TMR randomized clinical trials have focused on procedural details, such as differences in laser channel density (30,31) and laser channel depth (greater with surgical TMR) and more precise localization of the laser channels during surgery.
Rather than imposing procedural issues as the explanation for the disparate findings, we submit that a fundamental difference in clinical trial methodology is the likely causative factor. The patients studied in these trials had severe angina symptoms and had exhausted all forms of conventional therapy. They were highly motivated and desperate for a novel therapy to provide symptom relief, and they had high expectations for clinical benefit. It is recognized that the greater the subject’s stress, the greater the placebo effect (32,33), particularly when pain is the subjective end point (34), and it has been argued that devices and procedures have a larger placebo effect than do pills (35). There is an extensive literature on the dangers of non–placebo-controlled anti-anginal studies (36), and one can expect a 30% to 80% improvement in angina symptoms and a 90-s to 120-s improvement in exercise duration after placebo treatment alone. There are also previous examples of novel highly touted surgical therapies for angina relief, including internal mammary artery ligation in the 1950s, that were cast aside after rigorous placebo-controlled blinded studies were performed (37,38). For these very reasons, the present placebo-controlled blinded clinical trial was designed to eliminate both patient and investigator bias in the interpretation of outcome end points. There was a pronounced placebo effect in the present study resulting in a 30% improvement in exercise duration and angina symptoms that was sustained for 12 months and was identical to both laser treatment arms.
Mechanisms of laser TMR
Another disturbing component of the surgical and percutaneous TMR saga is confusion surrounding the pathophysiologic mechanism(s) to explain the observed clinical benefits. It seems clear that persistent open transendocardial laser channels with direct myocardial perfusion are rarely observed (39,40). Earlier theories of epicardial denervation to rationalize chest pain relief have been similarly discounted (41). The final and most plausible theory, local angiogenesis caused by injury, thrombosis, and inflammation, has been suggested in experimental histopathology studies (42,43) and by upregulation of angiogenic growth factors (44,45). Nevertheless, improved regional myocardial perfusion associated with angiogenesis after laser TMR has been much more difficult to show in animal models (46,47). This parallels the consistent negative findings when standard myocardial perfusion imaging assessments are included in either surgical or percutaneous laser TMR studies.
There are several limitations in the present study that should be considered: 1) Although we are unaware that catheter-based endocardial trauma has ever been associated with angina relief, the diagnostic LV mapping procedure performed in all patients may have induced sufficient endocardial trauma in the placebo sham patients to minimize the incremental effects of the laser treatment. 2) There may be more optimal lasing parameters (channel size and depth, injury zone, and so on) than those used in this study, which can be explored in the future to further improve clinical outcomes. 3) This is a single clinical study using a specific laser system, and the results, although concordant with one blinded percutaneous TMR study (28), are somewhat discordant with another blinded trial (29), indicating that additional large blinded studies with other laser TMR systems are warranted.
Conclusions and clinical implications
In this blinded randomized clinical trial, using the percutaneous Biosense laser DMR system in so-called “no option” patients with refractory angina, a substantial placebo effect was observed during sham therapy, and no incremental clinical benefits could be discerned after low-dose or high-dose laser treatment. In fact, the laser therapy was associated with more frequent acute MI (usually non–Q-wave) during the first 30 days. Based on these findings, we believe that all other clinical trials in this field should be viewed with caution and skepticism, unless proper attention is taken to account for placebo effects of the experimental laser therapies.
For the additional institutions and investigators that participated in the DIRECT trial, please see the online version of this article.
- Abbreviations and Acronyms
- Canadian Cardiovascular Society
- direct myocardial revascularization
- holmium: yttrium-aluminum-garnet
- left ventricle/ventricular
- major adverse cardiac events
- myocardial infarction
- single-photon emission computed tomography
- transmyocardial laser revascularization
- Received January 5, 2005.
- Revision received June 23, 2005.
- Accepted June 27, 2005.
- American College of Cardiology Foundation
- Burns S.M.,
- Sharples L.D.,
- Tait S.,
- Caine N.,
- Wallwork J.,
- Schofield P.M.
- Vincent J.G.,
- Bardos P.,
- Kruse J.,
- Maass D.
- Gassler N.,
- Stubbe H.-M.
- Nagle H.,
- Stubbe H.-M.,
- Nienaber C.,
- Rodiger W.
- Aaberge L.,
- Nordstrand K.,
- Dragsund M.,
- et al.
- Burkhoff D.,
- Schmidt S.,
- Schulman S.P.,
- et al.,
- ATLANTIC Investigators
- Saririan M.,
- Eisenberg M.J.
- Kornowski R.,
- Hong M.K.,
- Gepstein L.,
- et al.
- Kornowski R.,
- Baim D.S.,
- Moses J.W.,
- et al.
- Kornowski R.,
- Hong M.K.,
- Leon M.B.
- Stone G.W.,
- Teirstein P.S.,
- Rubenstein R.,
- et al.
- Domkowski P.W.,
- Biswas S.S.,
- Steenbergen C.,
- Lowe J.E.
- Yamamoto N.,
- Kohmoto T.,
- Gu A.,
- DeRosa C.,
- Smith C.R.,
- Burkhoff D.