JG: Welcome, Tom. You've already had an important career in biopharmaceutical development, and I know you're soon to start a new role that we would like to hear about.

TA: Thanks Joanne. It's a pleasure to be here and to talk with you. I have had an interesting career; started in academia and spent some time in industry, as well as on the vendor side with data and analytics.

Most recently I was with Johnson & Johnson as head of their healthcare informatics group for their medical device and diagnostics business, and I'm looking forward to joining HCL Technologies. So, I've played a number of different roles on different sides, but have always been focused around "How do we best use data and analytics to help drive clinical and business decision-making?"

One of the things that we have found is that we're changing healthcare practice patterns. It's really important to understand what's going on today, not what might have happened five or ten years ago which would be the basis of the published report.

And so, frequently what we've done is we've analyzed real-world data to understand the frequency and severity of events that would be the subject of the clinical trial itself to really help design and power the study and ensure that there's an adequate sample size.

JG: I always find this intriguing; how do you know if you need three samples or 3000 samples?

TA: Well it actually fundamentally comes down to statistics, and the critical assumptions are: "What's the existing rate today?" and then "What is the percentage decrease in the event outcome that you're expecting?" and based on that you can then design a study that has an 80% power of showing 95% certainty that you're getting or achieving the clinical difference that you're expecting to see.

JG: Earlier you mentioned the use of medical practice information to look at clinical trial recruitment patterns and approaches.

TA: One of the major problems with clinical trials today is the lack of recruitment. Often it takes a lot longer to recruit the patients, and as a result of that the clinical trials drag out longer than anticipated and they end up being more expensive.  And so one of the things that we started to do is try to use information about physician practice patterns to understand which physicians are likely to have patients that would qualify for the clinical trial.

It's important to know that among patients who are qualified for the trial, frequently it's less than five percent of patients who consent to agree to the clinical study, and so you really need to find practices that have a lot of patients in order to ensure that you can get an adequate number of patients enrolled in the study.

One of the things that companies are really concerned about are the cost of clinical trials, and in fact your most expensive site is a site that only enrolls one patient, in the sense that you've got all the fees associated with the physician, you've got all the fees associated with the site monitoring, and yet you're only getting a single patient.

It's much more efficient from the perspective of conducting the clinical trial to get multiple patients at the same site. So, can we use the information about physician practices to identify those physicians that are likely to have large numbers of patients who would qualify, and hopefully enroll in the study?

JG: Our upcoming conference, Best Practices in Outsourcing VI: Preclinical and Early Clinical will focus on that translational period, so essentially from the final toxicology studies into first in humans. Are there any particularly intriguing aspects of that moment that are different from other parts of the drug development process?

TA: Well I think some of the key challenges at that point are really to identify and understand who your key opinion leaders are, because those initial clinical studies typically involve only a small number of patients and so some of the technologies that we talked about earlier, in terms of identifying practices with large numbers, probably don't apply in that phase. It's more in phase two and three, where you're starting to need hundreds or thousands of patients, that it's critical to identify the physicians that would be able to participate in the study.

JG: Clearly, in first in human studies, the opinion leaders are going to be key, and can you talk a bit about the impact those choices can have on later studies?

TA: Well the results that you get in your preliminary studies are really going to help shape the design of your stage two and three studies, and so it's really important that you get good clean data, but I think one of the challenges that people get caught up in in the development process is they tend to then lean very heavily on those initial key opinion leaders to identify other physicians, and they may not necessarily have the best contacts or the best information about who would make additional clinical investigators.

JG: Anything else we should be sure that we talk about?

TA: Well I think one of the thoughts that occurred to me is that different vendors are going to have different capabilities for understanding practice patterns, both here in the United States and abroad, and so a critical factor if you're looking for a vendor to be a developmental partner might be to what extent they have access to information that can help you to characterize and plan future clinical trials in that area, assuming your early studies are successful.

JG: Thank you, Tom- we’ll look forward to more of your perspectives on the impact of bioinformatics thinking and practice.