One of the successful examples of incorporating patient-reported outcome (PRO) measures into an important clinical trial was in the NCI-Canada study BR.21 (abstract here). This study assigned patients to either tarceva or placebo in a 2:1 randomization to the active drug:

(Click on image to enlarge)

This study showed a 9% response rate and an improvement in median overall survival of two months with tarceva, which led to it’s US FDA approval and subsequent widespread use. While the fact that there were responses (although only 8%) and a survival benefit is very important, and probably the most important factor to many patients and oncologists, it’s important to ask whether this comes at the cost of significant side effects. Do patients need to trade quality of life for improved survival? A separate report on the BR.21 trial described PROs on this trial (abstract here).

In BR.21, patients were required to complete questionnaires that asked about a wide range of symptoms commonly seen in lung cancer, as well as global quality of life (QoL) and ability to function normally. Several questions focused on measured of cough, pain, and shortness of breath. The questionnaire was completed before starting treatment, every four weeks during treatment, four weeks after the end of treatment, and (for patients who came off of the study for reasons other than progression, such as side effects) every 12 weeks after ending the study, until progression. Completion rates were about 93% at the start, dipping down to the 80% range as the trial continued. Such a decline in returned responses is typical for QoL patient response assessments.

The symptom-based portion of the study assessment focused on pain, shortness of breath (SOB, also known as dyspnea), and cough. Importantly, the key measure was the time before progression of these symptoms rather than whether there was improvement in these symptoms. Why time to symptomatic worsening rathe than improvement in symptoms? Because you can’t have improvement if you don’t have the symptom, but everyone is a candidate for worsening of symptoms.

The results clearly demonstrated that the survival improvement with tarceva was also accompanied by a relative improvement in cancer symptoms. Specifically, while patients tended to have eventual worsening of symptoms at some point (as indicated by the downward slope of the curves below), recipients of tarceva had an average of a 1-2 month delay in their development of worsening of cough (4.9 vs. 3.7 months), SOB (2.9 vs. 4.7 months), and pain (2.8 vs. 1.9 months).

These differences were all statistically significant, and I would argue also clinically significant, even if we wish the results were better still.

In fact, patients on tarceva also often reported improvements in their symptoms. Not surprisingly, a higher proportion of patients on tarceva reported improvements than those receiving placebo (cough – 44% vs. 27%; SOB: 34% vs. 23%; pain: 42% vs. 28%). Global QoL was also reported as improved in more patients on tarceva: 35% vs. 26% (p < 0.0001, which means this is a very statistically significant difference despite the numbers not being extremely far apart). Personally, I’m pretty impressed with how well placebo fared, with about a quarter of the patients reporting improvement on a sugar pill, reinforcing why placebo-controlled trials are especially important when we assess PROs.

Finally, as with chemotherapy (see prior post), patients who had tumor shrinkage or stable disease on tarceva were also significantly more likely to have symptomatic improvement. For global QoL, physical functioning, cough, SOB, and pain, responding patients (complete or partial response) were the most likely to show symptomatic improvement, followed by patients with stable disease, and the lowest likelihood of symptomatic improvement in patients who have progression of disease:

But here is another example of how patients with stable disease receive benefit, along with a improved survival: just as survival is improved by 1-2 months among patients who show stable disease, 40-50% report symptomatic improvement in the absence of significant tumor shrinkage.

This work describes the beneficial effects of patient reported outcomes: the results corroborate the response and survival findings, and they show that patients can feel better and also live longer. These results are also factors that many patients generally care about: what is the chance that tarceva will make me feel better overall, and will my shortness of breath improve? But this study also includes some of the shortcomings: patients tend to return QoL questionnaires less consistently over time (and are the people who don’t respond the ones doing worst?), and there can be a very significant placebo effect. I’ll continue to cover some of the practical issues with PROs in the coming weeks.