REPLACEMENT HEART VALVE RANDOMIZED TRIALS ARE "INAPPROPRIATE," HIMA TELLS FDA; USE OF OBJECTIVE PERFORMANCE CRITERIA URGED AS ALTERNATIVE

Replacement heart valve clinical trials conducted under strict randomization would create "selection biases" that could be difficult to resolve in data analysis, the Health Industry Manufacturers Association says in a response to FDA's draft premarket application guidance for replacement heart valves that was submitted to the agency Feb. 26. The draft recommends use of randomized, controlled clinical studies to support PMAs for replacement valves ("The Gray Sheet" March 1, I&W-2). HIMA asserts that while randomized trials are appropriate for the clinical study of drugs and certain devices, they are "inappropriate for permanent implants, such as heart valves, that require major surgery." The association points out that surgeons have an obligation to use the device that is best suited to the patient's needs. However, that obligation would force surgeons to "violate the treatment assignment" of a randomized study, HIMA maintains. The comments also cite ethical difficulties surgeons would face in telling patients that they "do not know which of two valves is better," when they "do, in fact, have an opinion." In addition, the task force's comments note that many patients might refuse to participate in a randomized trial, mistrusting both the impersonal decision-making approach and any surgeon unwilling to recommend the best valve for their needs. Thus, HIMA continues, "many patients would be lost to the study, and significant selection bias would remain." HIMA's comments were prepared by representatives from Baxter Edwards CVS division, Carbomedics, Medtronic, St. Jude Medical and Sorin Biochemical. A cover letter enclosed with the response acknowledges that the guidance will undergo further review by FDA's circulatory system devices panel and revision by FDA before being implemented. HIMA mentions that it was "very pleased to hear" that Alan Andersen, director of the device center's office of device evaluation, stated at a Feb. 11 dental products advisory committee meeting that "FDA would not apply new clinical trial requirements to device studies currently under way" ("The Gray Sheet" Feb. 15, I&W-6). The letter says that HIMA's membership will continue to follow the requirements stated in the May 1990 premarket approval application guidance until the revised guidance becomes effective. HIMA also maintains that the study sample size necessary to conduct a randomized trial is problematic. For example, the group cites the thromboembolism rate, which is suggested by the guidance as a complication appropriate for assessment. Assessing thromboembolism would require a 15,000 patient sample in a properly randomized trial, HIMA estimates. This number, representing more than one third of all valve implants performed in the U.S. each year, demonstrates "a magnitude far in excess of anything that sponsors or clinicians could support," according to HIMA. The task force also says that extending enrollment over many years as a method to obtain the large sample is not viable because of the difficulty of result analysis due to changes in etiology of valve disease, operating techniques and postoperative care. The group asserts that the time delay thus created to bring the product to market is "unacceptable to the medical community and the public." As an alternative to randomized controlled clinicals, the HIMA document proposes a parameter estimation study, using "hypothesis testing to compare the results of an observational study against a set of predetermined Objective Performance Criteria." OPCs, they assert, can be developed for a range of endpoint values, such as valve-related complications and their expected rates of occurrence, as well as hemodynamic and hemolytic data. In the proposed study design, OPCs for a given clinical trial would be identified, validated and justified by a sponsor in the investigational device exemption. The OPCs, determined through data on specific event rates gathered from recently published literature, pre-clinicals and related studies, also would be used as the basis for sample size calculation. According to HIMA, the OPC proposal for heart valve study design "rectifies the deficits of the draft guidance, while maintaining a high standard of clinical science." HIMA adds that the approach is appropriate because it is: "well-controlled;" uses a study design acceptable to both patients and clinicians; incorporates "the wealth of published data" currently available on valves; permits "quantification of safety and efficacy"; permits meaningful statistical analysis within a "reasonable" time period; and uses a valid statistical model for calculating sample size.