TY - JOUR
AU - Packer, Milton
AB - A Novel Framework Emerges from the Dissonance Created by the MITRA-FR and COAPT Trials  Replication is a cornerstone principle of science. If a finding in science is valid, it should be replicable when an experiment is repeated, whether the study is performed in the laboratory or in the context of a large-scale clinical trial. Replication provides reassurance that the same answer emerges when the same question is asked at a different point in time. A perspective of the requirements for replication in clinical trials In clinical research, replication is particularly essential when trials are methodologically primitive. Many clinical trials are small, have imprecisely defined endpoints, are replete with missing data, and use unplanned statistical methods. When these conditions prevail, authors have enormous flexibility to forge a positive outcome if they are willing to make the requisite effort. Therefore, when the opportunities for analytical creativity abound, it is critical to confirm that a similar result can be achieved by researchers who apply rigorous standards and who (conceivably) might be less motivated to report a positive finding. Regardless of the quality of the original study, the likelihood of nonreproducibility is particularly great when the number of clinical events reported in a clinical trial is sparse.1 However, as clinical trial standards have advanced, it is now possible for investigators to design and execute trials that can yield consistently reliable results. We collectively understand the need to establish rules for the prespecification of endpoints and analyses and to minimize the biases and uncertainty that result from the missingness of data. When trials are able to base findings on a large number of clinically important events (i.e. >200–300 events) with few patients lost to follow-up, between-group differences become less fragile. Furthermore, when trials require the participation and coordination of hundreds (or thousands) of clinical sites, it is impossible for one researcher with less than perfect ethical, operational, or methodological standards to exert any meaningful influence. If unbiased principles are followed faithfully, replication of a trial may not only be unnecessary, but may be particularly challenging if the investigators report a persuasively favourable result on a cardiovascular event of unquestioned clinical importance, e.g. mortality. When a large-scale cardiovascular outcomes trial enrols 10 000 patients and finds a reduced risk of death, based on >1000 events with a statistically persuasive result and very few missing data, the results are so scientifically and morally compelling that it becomes unethical and unfeasible for the trial to be replicated in the same population. Under these circumstances, replication is typically sought in a distinct (but complementary) population afflicted with a disorder that is governed by similar pathophysiological mechanisms.1 Therefore, according to our current paradigm, the finding of a large-scale trial can be considered reliable and valid if the trial is large, well-designed and executed; collects a large number of events with little missing data; and demonstrates treatment effects that are clinically meaningful, analytically robust, and statistically persuasive. Such a framework has served us well over the past two decades. The collision of two trials of transcatheter mitral valve repair Given this context, what are we to do when two large-scale, definitively-designed clinical trials of the same intervention in apparently the same population report findings that differ strikingly in their conclusions? According to our accepted methodological framework, the results of both trials should be ‘reliable’. So, is our current conceptual scheme wrong? How can such dissonance happen? In 2018, two large-scale randomized trials reported the effects of transcatheter mitral valve repair in patients who had severe functional mitral regurgitation (MR) associated with heart failure due to left ventricular (LV) systolic dysfunction. Both trials randomly assigned patients to transcatheter mitral valve repair or to a control group that received only medical therapy; both studies were published in the same issue of the New England Journal of Medicine.2,3 In the 307 patients enrolled in the MITRA-FR trial (performed in France),2 mitral valve repair did not reduce the risk of death or the risk of hospitalization for heart failure after 1 year, and use of the device did not reduce LV volumes. In contrast, in the 614 patients enrolled in the COAPT trial (performed in North America),3 mitral valve repair reduced LV volumes after 1 year and decreased all-cause mortality as well as the risk of hospitalization for heart failure after 2 years. Can these discordant results be explained by methodological irregularities in one or both trials? The COAPT trial was twice as large and followed patients for twice as long as the MITRA-FR trial, but both trials reported results that were based on a meaningful number of events, with no apparent time-dependency. Although the COAPT trial made a specific effort to optimize medical therapy for heart failure prior to randomization, there is no evidence that medical treatment differed across the two trials, and if it did, it is not clear why differences in background drug treatment might explain the discordant findings. Finally, although some have speculated that procedural success was more common in COAPT than in MITRA-FR, > 90% of patients in both trials experienced adequate reduction of mitral regurgitant flow immediately following the procedure. The new concept of proportionate and disproportionate mitral regurgitation Given their methodological strengths, is it possible that both the MITRA-FR and COAPT trials yielded valid results? The two trials enrolled two distinctly different groups of patients (Table 1).4 The participants in MITRA-FR had striking LV dilatation but relatively modest degrees of MR. Prior to randomization, the mean LV end-diastolic volume (LVEDV) was 252 mL and the mean effective regurgitant orifice area (EROA) was 0.31 cm2. More than 50% of the patients had an EROA <0.3 cm2, and only 16% had an EROA ≥0.4 cm2. In contrast, the COAPT investigators primarily enrolled patients in whom the degree of MR was disproportionately great compared with the degree of LV chamber enlargement. Only 14% of the patients had an EROA <0.3 cm2, and 41% had an EROA ≥0.4 cm2; the mean EROA was 0.41 cm2, while the mean LVEDV was only 192 mL. Therefore, when the patients enrolled in COAPT are compared with those in MITRA-FR, the EROA was approximately 30% greater but their LV volumes were approximately 30% smaller. Table 1 Key features of the MITRA-FR and COAPT trials Mitra-FR (France) COAPT (North America) Study patients and duration of follow-up 307 patients followed for 1 year 614 patients followed for 2 years Criteria for severe MR EROA of 0.2–0.4 cm2 EROA of 0.3–0.4 cm2 Patients with marked LV dilatation Included by protocol Excluded by protocol Background medications for heart failure Appropriate Maximized before randomization EROA at baseline 0.31 cm2 (52% had EROA <0.3 cm2) 0.41 cm2 (41% had EROA ≥0.4 cm2) LVEDV at baseline 252 mL 192 mL 3+ or 4+ MR immediately post-procedure <10% <10% Change in LV volume after 1 year No change Decreased Hazard ratio (95% CI) for all-cause mortality 1.11 (0.69–1.77) (No benefit) 0.62 (0.46–0.82) (38% reduction in risk) Mitra-FR (France) COAPT (North America) Study patients and duration of follow-up 307 patients followed for 1 year 614 patients followed for 2 years Criteria for severe MR EROA of 0.2–0.4 cm2 EROA of 0.3–0.4 cm2 Patients with marked LV dilatation Included by protocol Excluded by protocol Background medications for heart failure Appropriate Maximized before randomization EROA at baseline 0.31 cm2 (52% had EROA <0.3 cm2) 0.41 cm2 (41% had EROA ≥0.4 cm2) LVEDV at baseline 252 mL 192 mL 3+ or 4+ MR immediately post-procedure <10% <10% Change in LV volume after 1 year No change Decreased Hazard ratio (95% CI) for all-cause mortality 1.11 (0.69–1.77) (No benefit) 0.62 (0.46–0.82) (38% reduction in risk) Table 1 Key features of the MITRA-FR and COAPT trials Mitra-FR (France) COAPT (North America) Study patients and duration of follow-up 307 patients followed for 1 year 614 patients followed for 2 years Criteria for severe MR EROA of 0.2–0.4 cm2 EROA of 0.3–0.4 cm2 Patients with marked LV dilatation Included by protocol Excluded by protocol Background medications for heart failure Appropriate Maximized before randomization EROA at baseline 0.31 cm2 (52% had EROA <0.3 cm2) 0.41 cm2 (41% had EROA ≥0.4 cm2) LVEDV at baseline 252 mL 192 mL 3+ or 4+ MR immediately post-procedure <10% <10% Change in LV volume after 1 year No change Decreased Hazard ratio (95% CI) for all-cause mortality 1.11 (0.69–1.77) (No benefit) 0.62 (0.46–0.82) (38% reduction in risk) Mitra-FR (France) COAPT (North America) Study patients and duration of follow-up 307 patients followed for 1 year 614 patients followed for 2 years Criteria for severe MR EROA of 0.2–0.4 cm2 EROA of 0.3–0.4 cm2 Patients with marked LV dilatation Included by protocol Excluded by protocol Background medications for heart failure Appropriate Maximized before randomization EROA at baseline 0.31 cm2 (52% had EROA <0.3 cm2) 0.41 cm2 (41% had EROA ≥0.4 cm2) LVEDV at baseline 252 mL 192 mL 3+ or 4+ MR immediately post-procedure <10% <10% Change in LV volume after 1 year No change Decreased Hazard ratio (95% CI) for all-cause mortality 1.11 (0.69–1.77) (No benefit) 0.62 (0.46–0.82) (38% reduction in risk) Why does this matter? In most patients with functional MR, enlargement of the LV is the primary determinant of regurgitant flow (see Figure 1). When the LV is markedly enlarged, moderate or severe degrees of MR can be expected entirely as a result of LV dilatation—in the absence of any incremental deficiency of mitral valve leaflet coaptation. These patients—whose MR is proportionate to LV volume—were enrolled in the MITRA-FR trial.4 Because the driving lesion is LV dilatation, such individuals should respond favourably to medical treatments that act to reduce LV chamber size,5 and they did not benefit from mitral valve repair in the MITRA-FR trial. In contrast, in some patients, uncoordinated contraction of the mitral valve supporting structures leads to degrees of regurgitation that are much greater than can be explained by LV dilatation. These patients—who have MR that is disproportionate to the degree of LV enlargement—were enrolled in the COAPT trial.4 Because LV size is no longer the sole determinant of MR, such patients would not be expected to adequately improve following medical treatments that reduce LV volumes; instead, in COAPT, they benefited from a procedure that acted to directly decrease regurgitant flow. Figure 1 View largeDownload slide Relation (defined by the Gorlin hydraulic formula) between effective regurgitant orifice area (EROA) and LV end-diastolic volume, assuming LV ejection fraction of 30% and regurgitant fraction of 50%. The dark blue line depicts severe MR that is proportional to LVEDV. The gray area approximates the uncertainty due to variabiility in the transmitral pressure gradient and in the measurement of EROA and LVEDV. The pink area depicts severe MR that is disproportionate to the degree of LV dilatation; the green area indicates mild-to-moderate MR. The average patient enrolled in the COAPT and MITRA-FR trials is shown. Adapted from JACC Cardiovasc Imaging. 2018 Dec 6. pii: S1936-878X(18)31017-9. Figure 1 View largeDownload slide Relation (defined by the Gorlin hydraulic formula) between effective regurgitant orifice area (EROA) and LV end-diastolic volume, assuming LV ejection fraction of 30% and regurgitant fraction of 50%. The dark blue line depicts severe MR that is proportional to LVEDV. The gray area approximates the uncertainty due to variabiility in the transmitral pressure gradient and in the measurement of EROA and LVEDV. The pink area depicts severe MR that is disproportionate to the degree of LV dilatation; the green area indicates mild-to-moderate MR. The average patient enrolled in the COAPT and MITRA-FR trials is shown. Adapted from JACC Cardiovasc Imaging. 2018 Dec 6. pii: S1936-878X(18)31017-9. The relation (as defined by the Gorlin hydraulic formula) between EROA and LVEDV is depicted in the Figure. assuming a LV ejection fraction of 30% and a regurgitant fraction of 50%. The blue line depicts the relationship when MR is severe but proportional to the LVEDV. The grey area approximates the uncertainty due to the hemodynamic state of the patient and imprecision in measurements. The pink area depicts severe MR that disproportionate to the degree of LV dilatation, while the green area indicates nonsevere MR. The typical patient enrolled in the MITRA-FR and COAPT trials is shown. Adapted from Grayburn et al, 2018. It seems likely that the marked LV enlargement—rather than the expertise of the operators—explains the lower rates of procedural success during long-term follow-up in the two trials. In COAPT, the proportion of patients with severe residual MR was 5% immediately following the procedure and remained at 5% after 1 year. In contrast, in MITRA-FR, the proportion of patients with severe residual MR was 9% after device placement and increased to 17% after 1 year. Interestingly, it is not possible to distinguish patients who have proportionate or disproportionate MR by examining their overall rates for death and hospitalization for heart failure. The annualized rates for these two major adverse clinical outcomes were similar in the medically-treated groups in the MITRA-FR and COAPT trials. However, the rate of these events was primarily related to LV dysfunction in MITRA-FR but to disproportionate MR in the COAPT trial. Lessons learned from the collision of two cardiovascular trials In the modern era of clinical trials, we do not expect large-scale, definitively designed and executed studies to yield misleading results if they are meticulously analysed in an unbiased manner. Fortunately, the discordant results of the MITRA-FR and COAPT trials do not undermine this conventional wisdom. Instead, the reporting of discordant results from the two studies has led us to rediscover a conceptual framework in which the results of both trials are valid. With respect to MITRA-FR and COAPT, the combined results of the trials tell us that patients with functional MR do not have a homogenous disorder. By providing complementary information, the trials (when considered together) indicate that we must now make a clear distinction between MR that is proportionate or disproportionate to the degree of LV enlargement. Optimal medical therapy seems to be the best choice for patients with proportionate MR, whereas mitral valve repair is highly desirable in those with disproportionate MR. Such a coherent understanding could not have been achieved if the MITRA-FR and COAPT trials had studied the same patients and yielded the same results. Integration of the results of MITRA-FR and COAPT allows us to optimally individualize the management of the heterogeneous group of patients who present with functional MR, heart failure, and LV dysfunction. Conflict of interest: none declared. References References are available as supplementary material at European Heart Journal online. Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2019. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - When Cardiovascular Trials Collide
JF - European Heart Journal
DO - 10.1093/eurheartj/ehy908
DA - 2019-02-07
UR - https://www.deepdyve.com/lp/oxford-university-press/when-cardiovascular-trials-collide-yk2Ts2Sr07
SP - 501
EP - 504
VL - 40
IS - 6
DP - DeepDyve
ER -