Hepatitis C-Avoiding Missteps When Evaluating New HCV Trial Data

Source- CCO
*Free registration required

Avoiding Missteps When Evaluating New HCV Trial Data

Graham R. Foster FRCP, PhD – 10/30/2012 More from this author

For the last 20 years, I have evaluated treatments for chronic hepatitis C by studying results from randomized, controlled clinical trials. I hunt for the “intent-to-treat” (ITT) sustained virologic response (SVR) rates and use these to compare the different regimens. Using this approach, my practice has matured from a fixed-dose, fixed-term regimen for everyone to nuanced therapy based on genotype and early virologic response. Latterly, I have introduced the protease inhibitors for patients with genotype 1 HCV.

ITT vs Per Protocol SVR
I believe that SVR is the most important outcome from therapy, as we know that patients who achieve SVR almost never relapse and have long-term benefits. The ITT analysis is important as it assesses adverse effects and withdrawals as well as the efficacy of the drug. Take, for example, a hypothetical drug that has terrible adverse effects and kills 9 out of 10 patients but cures the other 1. By ITT analysis, its problems are obvious (10% of patients achieve an SVR) but by “per protocol” analysis—where only patients who complete therapy are evaluated—the drug has an efficacy of 100%. Clearly, the per protocol analysis is far from ideal.

Rethinking ITT
As we move to an era of interferon-free potent antivirals, I am beginning to question the value of the ITT analysis. At the forthcoming AASLD meeting, we will see new data on combinations of direct-acting antivirals. I expect exciting data with very high cure rates. I suspect that it will be difficult to compare the various drugs as, for many combinations, the final ITT SVR will be heavily influenced by patient withdrawals. Take, for example, a hypothetical drug where every patient is cured but 10% of the patients fail to attend for their final blood test; by ITT analysis, the drug has a success rate of 90%. Contrast this with a second hypothetical drug that cures 95% of patients but where only reliable patients are enrolled and are then hunted down by an army of private detectives. If the investigators identify all the patients, then this inferior drug has an SVR of 95% and appears the superior molecule.

These issues will become less important as we move into large clinical trials where the numbers of patients are sufficient to dilute out the effects of poor attendance. However, in the current era of small phase II studies, these issues will play a major role.

As a result, I now wonder if we should shift our attention from ITT to per protocol analysis when evaluating early-phase studies. But the dangers of a per protocol analysis are that tolerability issues may be masked—a drug with adverse effects where patients withdraw from therapy can hide those deficiencies behind a non-ITT analysis.

Reading Between the Lines
On balance, I believe that the ITT SVR analysis is still the best way to compare drugs but I will be looking very carefully at the dropout rates and I will be studying with great care the withdrawn for adverse event and the failed to attend rates in studies presented at AASLD. Only when I have considered all 3 metrics will I reach my decision on the promise of the new regimen. For our patients struggling to cope with the deluge of new data, invariably spun into the most promising strands, it will be important to draw their attention to the many factors that determine the outcome of a clinical trial.

Your Thoughts
Tell me your thoughts. Do you put more value in ITT or per protocol analysis when interpreting and comparing results from small clinical trials of direct-acting antivirals? How do you help your patients wade through the data and interpret the results?

Graham R. Foster, FRCP, PhD, is Professor of Hepatology, The Liver Unit, and Consultant Hepatologist, Queen Marys University of London, London, United Kingdom.