Pharmacovigilance and adverse events

June 29, 2007

I first started writing this post when L’Affaire Avandia initially came to light, but decided to put it off until the incessant coverage had died down. Of course, days became weeks, but better late than never. I will not focus on the specifics of the Avandia situation, but rather on the concept of pharmacovigilance and the role of post-marketing analysis, at least as far as I understand the problem. Listening to a lot of talks on adverse event reporting at the DIA conference also definitely helped crystallize some of these thoughts.

In the recent NEJM paper, Steve Nissen and Kathy Wolksi essentially carried out a meta analysis, where they combined results from multiple independent studies (42 clinical trials) to arrive at their conclusions. That got me thinking. What constitutes appropriate pharmacovigilance and how do we minimize ADRs and make sure that a drug is safe?

The coming years are going to see several changes in clinical trial design and hopefully the approach to drug development itself. While this is not going to happen overnight, those of us who are in the software and informatics business need to understand and appreciate where the industry is headed. In the future, not only will there be data available from randomized clinical trials, and from post-market follow up, but also increasing amounts of molecular data will be included in the results. That means significant amounts of data complexity, diversity and, frankly, noise. In Steve Nissen’s own words on the limitations of meta-analysis (via In the Pipeline)

Our study has important limitations. We pooled the results of a group of trials that were not originally intended to explore cardiovascular outcomes. Most trials did not centrally adjudicate cardiovascular outcomes, and the definitions of myocardial infarction were not available. Many of these trials were small and short-term, resulting in few adverse cardiovascular events or deaths. Accordingly, the confidence intervals for the odds ratios for myocardial infarction and death from cardiovascular causes are wide, resulting in considerable uncertainty about the magnitude of the observed hazard. Furthermore, we did not have access to original source data for any of these trials. Thus, we based the analysis on available data from publicly disclosed summaries of events. The lack of availability of source data did not allow the use of more statistically powerful time-to-event analysis. A meta-analysis is always considered less convincing than a large prospective trial designed to assess the outcome of interest.

At DIA drug safety and pharma reputation were repeatedly listed as the top two issues facing the industry. Companies have started doing a lot of post-marketing analyses, but the success of such studies is still unproven. Companies like Prosanos and Phase Forward are developing applications that perform real-time quantitative analysis and signal detection to identify post-marketing issues, primarily by tracking adverse event databases. People are developing semantic web tools that do much the same. Soon, we will have to worry about the post-marketing success rate of companion diagnostics and how they correlate with successful treatment. In a perfect world, post-marketing intelligence would feed back into the discovery process so we could learn what kind of molecular signatures, phenotypes, etc were common across adverse events if any such pattern exists.

The entire industry is working hard to develop methods that try and make sure that a drug is indeed safe. As far as I can tell, meta-analysis during the trial phase should feed into continued post-marketing analysis along with signal detection of adverse effects. Some of the algorithms being developed are quite advanced, a rich area of research with a lot of direct patient benefit. But we are only at the tip of the iceberg. Our data are only going to get more complex once pharmacogenomics moves into the mainstream of drug development. The coming years are going to be very interesting.

Further reading:
In the pipeline
Respectful Insolence

Disclosure: Many former colleagues work at Prosanos

Technorati Tags: Healthcare, Adverse Event Reporting, Pharmacovigilance