The rate of our progress in lung cancer and other settings in medicine reaches a bottleneck in the slow rate at which clinical trials are completed. Nevertheless, only about 3% of patients with cancer in the US participate in clinical trials, and the number is even a little lower for people with lung cancer. There are many reasons for this: many patients don't have access to clinical trials without traveling significant distances, and others may object to participating in research (misleadingly sensational magazine covers showing a person in a cage, with a headline titled "Are you a guinea pig?" make me cringe), but another factor is that many patients are simply not eligible for our clinical trials. Some recent research looked at this question, important for its relevance for the rate of our progress and the generalizability of clinical trial results in a narrow subset to a much broader population of real world patients.

Dr. Lou Fehrenbacher is a medical oncologist in the large Kaiser Permanente network of Northern California (KPNC) who also conducts a significant amount of important clinical research there. He and collagues there took advantage of the breadth of this network to capture data on 326 consecutive patients diagnosed with advanced NSCLC just in the first quarter of 2005, comparing them to 196 patients who were enrolled on any of the randomized clinical trials that KP participated in over the preceding 10 years. They then presented the results on the differences between these two populations at the ASCO 2009 meeting.

We can be assured that the KPNC lung cancer population looked pretty much like what we'd expect from a general lung cancer population in North America, with 50% of the patients under 70, 18% over 80, and evenly split between men and women, the histology breakdown typical for NSCLC today, and 10% never-smokers. But when you look at the KPNC patients who were enrolled on NSCLC randomized clinical trials, it's clearly a disproportionate subset: 38% under 60 (more than twice the proportion (18%) of their overall NSCLC population), 74% under 70, and only 15% over 75 (less than half the proportion (34%) in the broader KPNC NSCLC population. I don't think this is surprising, especially since we know that 75-80% of the patients in nearly all of our recent large trials in NSCLC are under 70, and the median age on these trials is typically 61-62, nearly a decade younger than the median age of patients newly diagnosed with lung cancer in the US today. Also not surprising is the fact that patients enrolled on clinical trials had a lower extent of other medical problems (as measured by the Charlson Comorbidity Index).

What's more striking is the proportion of their patients who would have been eligible for typical current phase III trials in advanced NSCLC. Using the SWOG 9509 trial and the ECOG 4599 trial as benchmarks for general trials without or with, respectively, an anti-angiogenic agent, they found that only 29.5% of their NSCLC population would have been eligible for a chemotherapy trial, and a mere 19.2% would have been eligible for a trial of chemo with an anti-angiogenic therapy. The ineligibility is for many reasons, most commonly poor performance status (41.7% for both chemo and chemo/anti-angiogenic therapy trials), but also issues like concurrent other cancers, brain metastases, alcohol and drug issues, home oxygen requirements, etc., and with cardiac disease, use of blood thinners, and squamous NSCLC subtype being added restrictions for anti-angiogenic therapy trials.

These numbers include patients of all ages, and if we look only at who would be eligible among patients under 70 (for instance, if clinicians today were to factor in that results on the ECOG 4599 trial only showed a benefit for patients under 70, while older patients experienced significantly more toxicity and no improvement in survival), the proportion would be only 17.4% for a chemotherapy trial and 10.7% for a chemo/anti-angionenic agent trial.

If anything, these results overestimate the proportion of patients who would be eligible, because there may well be other ineligibility factors beyond the handful the investigators looked for in the charts and electronic records of these patients. For instance, while informed consent forms can be translated into languages other than English, it takes many weeks to get one translated and then approved by the required regulatory bodies. Most physicians and patients will likely opt for a treatment plan that can be enacted now if it takes 3-4 weeks to pursue a clinical trial.

It's debatable whether oncologists and patients should self-select against elderly patients enrolling on clinical trials of aggressive therapies, since many have demonstrated comparable results in the "fit elderly". My preference in general is to have broader eligibility and try to do the post-hoc analyses to look at different subsets, whether based on clinical or molecular features.

Even including older patients, we've clearly got a challenge in trying to move the field forward when only 3 in 10 patients could participate in our important trials and the results of these trials are being applied to the much broader population. Ideally, we sould develop trials suitable for patient populations more representative of the real world, and in the meantime we need to take care in determining how to apply a standard for which only a minority of people in our own clinics would have been eligible by the criteria that established them.