Questions Raised About FDA Approval Studies
A survey of 83 clinical studies supporting 78 PMA panel-track supplements approved by FDA between 4/19/06 and 10/9/15 found that only 45% were randomized clinical trials and 30% were blinded. Researchers writing in the Journal of the American Medical Association say that of the 150 primary end points in the studies, 81% were surrogates and 38% were compared with controls. “These findings suggest that the quality of the studies and data evaluated to support approval by FDA of modifications of high-risk devices should be improved,” the authors say.
Meanwhile, another group of researchers reported in the journal that they looked at the characteristics and findings of preapproval and mandated postapproval trials of drugs granted FDA accelerated approval between 2009 and 2013, and found that for 22 drugs given accelerated approval for 24 indications, clinical trials conducted before and after approval had similar design characteristics such as lack of blinding, randomization, and comparator groups. “Of the required postapproval confirmatory studies, half were completed a minimum of three years after approval, and of these, most showed some benefit but relied on surrogate measures rather than clinical outcomes,” the authors say.
The PMA supplement study found that the majority of clinical studies supporting approval of original PMA devices are non-randomized, are unblinded, and often use surrogate endpoints that are not compared with active controls. Changes to PMA devices come through supplements, the researchers say, and devices often have hundreds of supplements, meaning that many high-risk devices differ substantially from the originally approved device.
Of the six PMA supplement pathways, only panel-track supplements always require clinical data. Given the increasing use of PMA supplements, the study says, the authors characterized the strength of evidence of the studies used in FDA-approved panel-track supplements during the past decade.
“Although randomization and blinding are widely accepted as prerequisites for high-quality clinical studies,” the article says, “they were used infrequently in studies to support device modifications. This means lower quality data often supported changes in high-risk devices that are modifications to previously approved devices.”
In the drug study, researchers concluded that among the 22 drugs granted accelerated approval for 24 indications, efficacy was often confirmed in postapproval trials a minimum of three years after approval, although confirmatory trials and preapproval trials had similar design elements, including reliance on surrogate measures as outcomes. “The clinical trial evidence for therapeutic agents granted accelerated approval by FDA between 2009 and 2013 shows that 14 of 24 indications for these drugs entered the market on the basis of single-intervention group studies that enrolled a median of 132 patients, which some investigators would consider a small number,” the authors say. Half of required confirmatory studies were completed a minimum of three years after the approved drug was on the market. The quality and quantity of postmarketing studies required by FDA to confirm clinical benefit varied widely across indications.”
The article reports that drugs granted accelerated approval receive market authorization on the basis of fewer studies, smaller patient populations, shorter follow-up, and less established surrogate measures than drugs approved via the traditional pathway. In such cases, it says, postapproval confirmation of clinical benefit is essential. But such postapproval studies can progress very slowly.
In a commentary on the articles, former FDA commissioner Robert Califf says they raise significant issues that should be considered by FDA and other stakeholders such as providers and payers. He calls for a sweeping overhaul of the current clinical trial system, including creation of a federated (rather than fully centralized) national system that generates fundamental information such as identification of drug dispensing and device identifiers and that measures clinical outcomes continuously. “These data should be linked to practices within organized health systems to enable use for multiple purposes, including conducting observational studies and pragmatic trials,” Califf writes. “Because the same information can be used and reused for different purposes, the data quality should improve and costs should decline over time as data are curated and prospective research systems are reused. Ongoing evaluation of these new systems will be needed to ensure they are as effective as hoped.”