Assessment of FDA Approval for New High-risk Therapeutic Devices Not Meeting Pivotal Study Primary End Points, 2016-2020

Abstract

This cross-sectional study explores the frequency of, and rationale for, US Food and Drug Administration approval of high-risk therapeutic medical devices not meeting pivotal study primary end points from 2016 through 2020.

The US Food and Drug Administration (FDA) regulates high-risk medical devices through the premarket approval (PMA) pathway. To obtain FDA approval, manufacturers typically must conduct a pivotal premarket study providing evidence of device safety and effectiveness.¹ The FDA assesses device risks and benefits as measured in pivotal studies by primary end points, such as the magnitude of clinical symptom improvement or occurrence of adverse events.¹ However, the FDA recently approved a temporary cardiac support device for use during high-risk percutaneous coronary interventions despite early termination of its pivotal study owing to futility.² In this cross-sectional study, we sought to systematically determine the frequency of, and rationale for, FDA approval of high-risk therapeutic medical devices not meeting pivotal study primary end points.

Methods

Using the FDA PMA database,³ we conducted a retrospective cross-sectional analysis of new high-risk therapeutic medical devices approved between January 1, 2016, and December 31, 2020 (eFigure in the Supplement). In accordance with previously described methods,⁴ we reviewed FDA summary documents to identify all pivotal studies evaluating these devices and their primary end points. We then determined which studies did not meet at least 1 prespecified primary end point.

For each device not meeting a pivotal study primary end point, we extracted device, pivotal study, and primary end point characteristics (eMethods in the Supplement). We also determined the FDA-provided rationale for approval and whether a postmarket study was mandated as a condition of approval.³

All search results were current as of March 7, 2021. All data were summarized with descriptive statistics using Excel for Mac, version 16.49 (Microsoft). This study used public, nonidentifiable data that did not constitute human participants research (45 CFR 46.102) and was not submitted for institutional review board review.

Results

Between January 1, 2016, and December 31, 2020, the FDA approved 107 new high-risk therapeutic medical devices, of which 14 (13.1%) did not meet at least 1 pivotal study primary end point. Among these devices, 10 were cardiovascular, 4 were life sustaining, and 11 were implantable (Table 1).

Table 1. Characteristics of FDA-Approved High-risk Therapeutic Medical Devices Supported by a Pivotal Study Not Meeting a Primary End Point, 2016-2020.

Device characteristic	No. (%)
Total	14
Implantablea
Yes	11 (79)
No	3 (21)
Life sustaininga
Yes	4 (29)
No	10 (71)
Specialtya
Cardiovascular	10 (71)
Gastroenterology/urology	3 (21)
Neurology	1 (7)
FDA advisory panel convened
Yes	2 (14)
No	12 (86)
FDA approval rationaleb
Post hoc effectiveness analysis	11 (79)
Nonsignificant numerical benefit	4 (29)
Success of other primary end points	11 (79)
Benefit suggested by other outcomes	11 (79)
Premarket confirmatory study	2 (14)
FDA-required postapproval study
Yes	11 (79)c
No	3 (21)

Abbreviation: FDA, US Food and Drug Administration.

^aCategorization per FDA product code designation.

^bCategorizations not mutually exclusive.

^cIncludes 2 devices for which all postapproval studies were complete and 9 devices for which at least 1 postapproval study remained in progress.

In 15 pivotal studies, the 14 devices did not meet 16 of the 36 (44.4%) primary end points; 3 devices did not meet all of the primary end points, 10 devices did not meet an effectiveness end point, and 2 did not meet a composite safety-effectiveness end point. Seven devices were assessed against a comparator, and 7 were assessed against other criteria (eg, performance goals). Three devices did not meet noninferiority end points, and 6 did not meet surrogate end points. The FDA frequently cited post hoc analysis, success of other primary end points, and positive trends in secondary or nonprespecified outcome measures as rationales for approval (Table 1). For 11 devices, the agency required at least 1 postapproval study.

The 14 devices are described in Table 2. Among the salient examples are 3 aortic valve prostheses associated with elevated bleeding rates compared with objective performance criteria and 2 donor lung perfusion systems that did not meet survival end points.

Table 2. Descriptions of the 14 New High-risk Therapeutic Medical Devices Supported by a Pivotal Study Not Meeting a Primary End Point^a.

Device name	FDA-approved indication	No. of primary end points	Unmet primary end points, No. (%)	Description of unmet primary end point(s)	FDA approval rationale
AspireAssist	Assist in weight reduction of patients with obesity for whom nonsurgical therapy has previously failed	2	1 (50)	52-wk Responder rate ≥50% as defined by excess weight loss ≥25%	Near miss at lower bound of CI Success of other primary end point (excess weight loss ≥10% relative to control at 52 wk)
Cheatham Platinum stent system	Treatment of native and/or recurrent coarctation of the aorta	7b	1 (14)	12-mo Mean reduction in systolic blood pressure (delta)	Success of second pivotal study Success of all other primary end points Favorable post hoc analysis demonstrating obliteration of preprocedure pressure gradient
Intuity Elite valve system	Replacement of native or prosthetic aortic heart valves	4	2 (50)	Bleeding and paravalvular leak rates compared with OPC	Determination that bleeding events were not device related Determination that leak rates were due to valve undersizing Manufacturer analysis suggesting longer procedural times (statistically nonsignificant) in subgroup owing to surgical complexity and small sample size
				Average cardiopulmonary bypass time and aortic cross clamp times in 1 (of 6) subgroups as compared with registry-derived benchmarks
Tryton side branch stent	Improving the side branch luminal diameter of de novo native coronary artery bifurcation lesions	1	1 (100)	9-mo Composite end point of cardiac death, target vessel MI, and target vessel revascularization	Favorable noninferiority post hoc subgroup analysis Successful single-arm confirmatory study
Pericardial aortic bioprosthesis and Inspiris Resilia aortic valve	Replacement of native or prosthetic aortic heart valves	4	1 (25)	All and major bleeding rates compared with OPC	Determination that bleeding rates were not device related Success of other primary end points (eg, New York Heart Association functional status classification)
Organ Care System lung system	Preservation of donor lungs for bilateral lung transplantation	2	1 (50)	Composite of noninferior 30-d posttransplantation survival and 72-h freedom from primary graft dysfunction	Success of amended primary effectiveness end point in per-protocol population Success of primary safety end point (lung graft–related serious AE) Comparable 2-y survival rate to standard of care
Lifestream balloon expandable covered stent	Treatment of atherosclerotic lesions in common and external iliac arteries	1	1 (100)	30-d Composite safety and effectiveness end point measuring major AE	Uncertainty regarding potential inflation of restenosis rates owing to Doppler ultrasound use Favorable post hoc analysis of 12-mo major AE rates
Astron Pulsar stent system	Improve luminal diameter in patients with symptomatic de novo, restenotic, or occlusive lesions located in the superficial femoral or proximal popliteal arteries	2	1 (50)	395-d Postprocedure primary patency rate compared with prespecified performance goal	Near miss at lower bound of CI Uncertainty owing to inclusion of long lesions without corresponding adjustment of performance goal Success of primary safety end point and positive signals from secondary effectiveness end points (eg, related to quality of life)
Lutonix drug coated balloon percutaneous transluminal angioplasty catheter	Treatment of stenotic lesions in dysfunctional native arteriovenous dialysis fistulae	2	1 (50)	6-mo Target lesion primary patency	Statistically nonsignificant superiority for primary effectiveness end point Success of primary safety end point and positive signals from secondary effectiveness end points (eg, reduced reintervention rate)
Avalus bioprosthesis	Replacement of diseased, damaged, or malfunctioning native or prosthetic aortic valves	3	1 (33)	All and major hemorrhage rates compared with OPC	Determination that hemorrhage rates were not device related Success of other primary end points (eg, New York Heart Association functional status classification)
XPS with Steen Solution perfusate	Perfusion of initially unacceptable excised donor lungs during which time the ex vivo function of the lungs can be reassessed for transplantation	2	2 (100)	12-mo Noninferior mortality	Uncertainty owing to selection bias during study enrollment Expansion of donor pool owing to use of previously unacceptable lungs Favorable survival rates compared with real-world registry data
				72-h Noninferior incidence of primary graft dysfunction
Optimizer smart system	Treatment of heart failure in patients who are symptomatic despite medical management and who are not candidates for resynchronization therapy	2	1 (50)	6-mo Responder rate as measured by ventilatory anaerobic threshold	Favorable post hoc subgroup analysis Successful confirmatory study Success of primary safety end point and secondary effectiveness end points (eg, related to quality of life)
Treo abdominal stent-graft system	Endovascular treatment of patients with infrarenal abdominal aortic and aorto-iliac aneurysms	2	1 (50)	1-y Postimplantation treatment success relative to performance goal	Near miss at lower bound of CI Uncertainty owing to increased stringency of performance goal Success of primary safety end point and secondary effectiveness end points (eg, stent leakage)
ReActiv8 implantable neurostimulation system	Management of intractable chronic low back pain in patients for whom medical management has failed and who are not candidates for spinal surgery	2	1 (50)	120-d Responder rate for low back pain improvement as measured by average visual analog scale	Favorable post hoc analysis Success of primary safety end point and secondary effectiveness end points (eg, quality of life)

Abbreviations: AE, adverse event; CI, confidence interval; FDA, US Food and Drug Administration; MI, myocardial infarction; OPC, objective performance criteria.

^aIndications, primary end points, and FDA rationale summarized for concision and clarity.

^bFDA approval of device supported by 2 pivotal clinical studies.

Discussion

Among 107 new high-risk therapeutic medical devices that received FDA approval between January 2016 and December 2020, 14 were supported by a pivotal study that did not meet 1 or more primary end points. The rationale for approval frequently included favorable post hoc analyses and success as measured using other end points. The FDA required postapproval studies for 11 devices.

Study limitations include that the results may not be generalizable to diagnostic devices and that we could not determine the percentage of high-risk therapeutic medical devices not meeting pivotal study primary end points that the FDA approved. Information on rejected or withdrawn PMA applications is nonpublic.

The FDA’s approval of devices not meeting pivotal study primary end points reflects the complex nature of regulatory risk-benefit assessment, which takes into account scientific uncertainty, clinical factors, and therapeutic alternatives.⁵ For example, the FDA approved an ex vivo pulmonary perfusion system that did not meet coprimary end points for survival and graft dysfunction in part because the device permitted transplantation of previously unacceptable lungs⁶ (Table 2). Such complexity underscores the need for the timely postmarket evidence to help patients and clinicians make informed decisions about the use of high-risk therapeutic devices.

Supplement.

eMethods

eFigure. Study Cohort of New High Risk Therapeutic Medical Devices Receiving FDA Premarket Approval between January 1, 2016 and December 31, 2020

eReferences