We wish to express our concerns about the study “Efficacy of Spironolactone on Survival in Dogs with Naturally Occurring Mitral Regurgitation Caused by Myxomatous Mitral Valve Disease” (Bernay, et al. JVIM 2010;24(2):331–341). As summarized by the authors, spironolactone has theoretical value for management of canine heart failure (HF) due to mitral regurgitation (MR). In this study, the outcomes with both spironolactone therapy and placebo (along with an ACE-inhibitor ± furosemide ± digoxin) were superior to previously published HF trials (such as QUEST) that evaluated diuretics (including spironolactone), ACE-inhibitors, and pimobendan.1 These findings surprised us; accordingly, we asked if other factors might explain this apparent benefit. We trust that the statistical analysis is expert and agree that differences are evident between groups (with wide confidence limits for benefits of spironolactone). However, we also believe the clinical relevance of these findings is diminished by the study design, event rate, patient withdrawals, and patient categorization with regard to HF.
The study design is odd as it includes two separate “lead-in” trials of different durations based on initial administration of furosemide at entry. Some dogs were then entered into a long-term study for a total of 3 combined studies. We draw attention to the atypically low event rates for an HF study (Figs 2 and 3), where only 39 out of 212 dogs reached the combined endpoints of cardiac death, euthanasia because of HF (euphemistically labeled MR in the paper), or severe worsening of “HF” over a 14–15-month span. We were also concerned by the high withdrawal rates, where 94 of 212 dogs (44%) failed to complete the study for reasons unrelated to the defined endpoints.
The key issue for us is whether the dogs were appropriately characterized with regard to HF. Although the term “HF” is not specified as an endpoint, the findings of “cardiac-related death, euthanasia because of MR, or severe worsening of MR” clearly implicate HF, not MR, as the cause for major events. Additionally, the dogs were classified as ISACHC class II (mild-to-moderate) or III (severe) HF, meaning mild-to-severe cardiogenic pulmonary edema should have been evident. Thus, we believe the study conclusions are predicated on a clinical and imaging diagnosis of HF caused by MR. However, clinical findings defining HF were imprecise with none of the signs used specific for HF. We would have expected entry criteria of increased respiratory rate at rest, moderate-to-severe left atrial enlargement, and caudodorsal pulmonary infiltrates typical of HF. Only mild cardiomegaly was required for study enrollment (VHS > 10.5) and radiographic evidence of pulmonary edema was not required. We fear that these inaccurate criteria may have been confounded by the different levels of experience in veterinarians enrolling cases (some were specialists; others practiced at first opinion clinics). The majority were placed in ISACHC class II by these investigators, a category in which dogs with MR often exhibit clinical signs (eg, cough) due to bronchial compression and primary airway diseases, conditions that frequently mimic HF. Also, furosemide is typically a critical background drug for HF trials, but in the 3-month lead in trial it was withheld for 5 days. We contend that dogs with cardiogenic pulmonary edema generally can't survive for long without furosemide; yet over half of the dogs were enrolled without it. This would have created an ethical dilemma if patients were actually symptomatic from HF and strongly suggests these dogs were never in HF. Unfortunately, we do not know the precise number of dogs receiving furosemide during the study, nor the average daily dosage for each group. Our contention that a large percentage of study dogs were not in HF is also supported by the low mortality rates and the regression hazard ratio showing decreased survival probability when furosemide was needed for management (ie, these were the patients that were actually in HF). We believe that had the entry data been independently reviewed by veterinary cardiologists to assure the dogs were actually in HF, that many subjects would have been excluded from this trial.
We also carry reservations about the low number of major events producing flat Kaplan-Meier slopes compared with past trials. While a 55% reduction in the risk of cardiac morbidity-mortality seems impressive, we believe it may relate to the small number of events superimposed on study groups that are heterogeneous relative to HF. The authors admit these dogs were not as seriously affected as those of other drug trials. This prompts the question: if spironolactone did work, at what stage of disease was benefit observed? Based on the vagaries of these data, we do not believe this question has been answered for any stage of the disease.
Finally, we are concerned that this is yet another pharmaceutical study conducted with the intent of bringing a drug to market, but leaving many questions in its wake. We are not suggesting a company-centered study is inferior; however, we do believe in the value of independent perspective for study design, investigator recruitment, raw data assessment, statistical analysis, and manuscript preparation. We appreciate that pharmaceutical companies exist to develop and market drugs; that clinical investigators want to study new treatments and publish those results; and that public funding for veterinary trials is minimal. In this environment, we consider the best clinical studies trials as those involving the sponsoring company in partnership with independent clinical experts, statisticians, and clinical monitors (when drugs are earmarked for regulatory review). We believe such collaborations can better address issues of clinical importance, minimize ambiguities related to study conclusions, and reduce potential conflicts of interest.2 Perhaps if the independent veterinary cardiologist and statistical consultant who coauthored this paper had been actively involved at the earliest stage of this study, it would be more conclusive.
Thus, we respectfully dispute that a survival benefit of spironolactone in canine MR is proven. Our arguments are centered on (1) patient categorization, (2) definition of HF, (3) heterogeneity of MR severity and HF within groups, and (4) the low event rates relative to the high number of case withdrawals. We are uncertain which dogs if any did benefit from spironolactone. Thus, regardless of the current EU regulatory approval, we consider use of spironolactone in MR empirical. We also plead for better collaboration between pharmaceutical companies and independent investigators at every stage of clinical trials. We believe this should be a focus for regulatory agencies and our scientific journals. While we would be happy to be corrected—that spironolactone greatly improves outcomes in dogs with HF due to MR—we believe that conclusion would require a different study.