FDA REVIEWS THE USE OF PFS IN CANCER TRIALS 

July 25, 2012 — A faster, less expensive method of verifying progression-free survival data in cancer drug clinical trials is being considered by the US Food and Drug Administration (FDA).

But the potential change in methodology is seen as a threat to quality data by a number of experts.

The matter was debated at an unusual meeting of the FDA Oncologic Drugs Advisory Committee (ODAC).

The meeting was "unique in that we are not asking the committee's advice on a specific drug product," said Rajeshwari Sridhara, PhD, director of the division of biostatistics V at the Center for Drug Evaluation and Research at the FDA. "The purpose of this meeting is to have a wide-ranging discussion on how best to assess and mitigate potential bias in the determination of progression-free survival of nonhematologic malignancies."

Currently, the FDA requires an independent review of all progression assessments in a clinical trial. An alternative, an independent audit of investigator assessments in a prespecified subgroup of patients, was discussed at the meeting.

An independent audit could streamline the conduct of clinical trials, according to the FDA, and avoid missing data when no additional protocol-specified progression assessments are mandated.

Hematologic malignancies were specifically excluded from this discussion because of mitigating factors that could influence the assessment of progression-free survival.

As the merits of this type of independent audit were considered, a lively discussion ensued, opinions were exchanged, and questions were asked by committee members and other participants. A vote was not taken, nor was any consensus opinion reached.

But Ralph D'Agostino, PhD, chair of the Department of Mathematics and Statistics at Boston University in Massachusetts, questioned the partial audit process in a basic way. "When you have a complete audit, you don't have to worry about it. But when you have an incomplete audit, how do you do the stratification?"

Progression-Free Survival as an End Point

Dr. Sridhara emphasized that the discussion was not about whether progression-free survival is an appropriate efficacy endpoint or what magnitude of progression-free survival benefit would lead to marketing approval. Instead, she explained, the meeting would focus on the relation between progression-free survival and independent reviews of related data in clinical trials.

Progression-free survival is increasingly being used in clinical trials as a primary end point to evaluate the effectiveness of oncologic agents. One of the key concerns raised with the use of this end point is the potential for systematic bias based on local evaluations performed at the site. Therefore, when progression-free survival is used as a primary end point, the FDA has typically required a review of scans by an independent radiologic review committee (IRC) because local evaluation or investigator assessment could be biased, and complete case progression-free survival determined by the IRC is used instead of progression-free survival determined by an investigator.

Progression-free survival is an important primary end point for regulatory approval, said presenter Lori Dodd, PhD, from the National Institutes of Health. "The use of progression-free survival is an area of active debate," she noted. "In general, it is not a measure of clinical benefit. It does not directly measure how a patient functions, feels, or behaves, nor has it been shown to be a surrogate end point for overall survival."

In many randomized trials that use progression-free survival as a primary end point, imaging scans are subjected to a blinded independent central review to confirm the results observed at the site. As a result, disagreement between the local and central evaluation, with respect to either the censoring status of the patient or the timing of the progression, sometimes occurs.

To examine the role of the IRC, the FDA — in collaboration with groups representing drug information associations, the Pharmaceutical Research and Manufacturers of America, and the National Cancer Institute — conducted a workshop in October 2009 to evaluate the discrepancy in progression-free survival determinations by investigators and IRCs, explained Dr. Sridhara during her presentation.

According to background information supplied by the FDA, there is general agreement between IRCs and investigators with respect to relative treatment effects based on progression-free survival assessments. But what has been commonly observed is an "inherent measurement error" in the reading of radiographic scans and disagreements between readers at the patient level.

Opinions and Issues

The FDA noted that regulatory considerations are based on the relative treatment effect at the population level, and acknowledged that progression-free survival might not be the appropriate primary efficacy end point for specific indications.

However, the ODAC was asked to assume that progression-free survival is an appropriate primary end point for a specific indication. Given the information provided by audit strategies using random samples, the variability in radiographic measurement, and logistical considerations, the ODAC was asked to discuss whether the current practice of a complete case review by the IRC of all patients should be replaced by an IRC audit using a random sample, and to discuss situations where an IRC audit using a random sample might not be appropriate.

Committee members voiced a number of opinions and raised potential issues and problems related to random audits. For example, David Harrington, PhD, from the Harvard School of Public Health in Boston, a temporary voting member, noted that he is "not ready yet to say that there is an approach that dominates here." Instead, he urged the FDA and others to try to learn more about the variability of clinical trials.

Another member found this to be a "unique effort" on how to assess responses, but pointed out that imaging techniques and quality can greatly vary, even within the same facility. "I think we need to raise the bar and make it more robust," he said.

Brent Logan, PhD, from the Medical College of Wisconsin in Milwaukee, explained that "a full IRC analysis will not contribute new or independent information beyond the investigator assessment." He noted that, in general, an audit approach that is appropriately conservative could be used, but "that the real benefit is cost savings."

Another member pointed out that there is more than just cost involved, and that there needs to be "clearly defined criteria for the audit."

In the background information, the FDA noted that if "a random sample IRC audit method [is implemented], further considerations need to be addressed, including whether the audit should be conducted only for 'positive by [investigator] assessment' trials, and whether more than 1 independent reviewer is necessary" to perform the review.

Other potential concerns include the uniform training of site radiologists, scan procurement and storage, and quality control, all of which will require further discussion.