The FDA and the Person With Cancer: Give PROs a Chance

In February 2006, the FDA released a draft Guidance for Industry on Patient-Reported Outcome Measures (see www.fda.gov/cder/guidance/index.htm). The document was offered for public comment through at least early April. It will have a major impact on the field of measuring patient-reported outcomes (PROs) in clinical trials. Although it carefully states that the use of the word "should" means that something is recommended but not required, industry-based readers of such documents usually adhere: Some will literally interpret and overinterpret the guidance, and others will push the limits to work around them.

The document also clearly states that this guidance does not address the use of PROs for instruments beyond evaluation of claims made about a drug or medical product in its labeling. This bears repeating: the guidance does not address use of PROs outside of the regulatory setting. It should not become guidance for all clinical trials research and certainly not for general clinical research.

FDA officials have often stressed that they are bound by law to ensure that approved label claims be supported by the submitted data. By statute and by stated intention in the guidance, "It's all about the label." The FDA interpretation of this charge, as reflected in section IV of the guidance, is at times so literal that the only conclusion one can draw is that absolutely no currently available questionnaire used in clinical trials would pass muster.

Based on the tone and emphasis of the guidance, PRO discussions with industry are likely to be focused on the exact wording of the label claim and the presumed concepts tied to those words. There will be intense scrutiny and focus on label terminology and the evidence that a "concept" being measured is true to its terminology.

All concepts are subjective conveniences, shared by some and never all good and thoughtful people. No questionnaire can capture everything that is important to any one patient no less all patients with a given condition ... at best we create finite sets of questions to approximate the patient's life...and they can do a very good job at measuring what matters most. In a guidance-driven review of a well-studied and validated questionnaire, the emphasis will be on the words in the label: Every word will require back up and proof that it is truly what is being measured. Because this is never provable, any submission is always rejectable.

If a questionnaire comes to the agency that was developed with extensive patient input, this matters little unless the label claim can be supported. Our colleagues at the FDA feel very strongly about that label and protecting it. Who can blame them, as the pharmaceutical industry has done its share of exploiting loopholes.

This places the future of PROs on the dance floor between the FDA and the pharmaceutical industry. In this unfortunate dance, the patient is lost. How do we really make this process "all about the patient." I believe it is by carefully looking at the scrutiny our federal regulators place on the label terminology and the dangerous exercise of sizing up questionnaires based on their compliance to hypothetical concepts about which three scientists are likely to have four opinions.

We can simplify this by remembering the patient and keeping him or her at center stage. Measurement science and critical peer review can support many PRO end points if our regulatory colleagues can find a way to relax the hegemony of the label. The most important victim of overemphasis on the label and ensuing emphasis on questionnaire content validity is, ironically, the patient, whose voice may ultimately be silenced by the well-intentioned effort. Instead of focusing on concepts and labels, let's focus on the more challenging and consequential issues of study design, hypothesis testing, missing data, and the meaningfulness of a therapeutic benefit to patients in their world, not ours.

Unfortunately in today's regulatory review environment, it can not be about the patient and the label at the same time. If it's all about the patient, then the label has to listen...if it's all about the label, then the patient has to think like labels. Unfortunately, we are far too diverse a species to think one way, to answer questions in the same way, and to be perfectly reliable in our responses. Even still, these humble questionnaires developed to measure PROs in oncology and elsewhere, work pretty well, and better than most other data submitted and approved for label claims. Let's really level the playing field—and "give PROs a chance."

—David Cella, PhD