As anyone that has ever read a novel or seen a movie about time travel can tell you, moving through time is not without its risks. Remember when Michael J. Fox started to dissolve in Back to the Future? Messing with the past is not something to be taken lightly.
Which brings me to my point: the perils of retrospective clinical trials. It's hard to resist the temptation of clinical samples collected for some drug trial. After all, they're just sitting there in the freezer, with reams of clinical data tied to them. What could be easier? Well, before you go signing those clinical gems out of the freezer, there are several questions you should ask yourself. And you can bet they are questions that the FDA will ask you as well.
- Are these samples stable? This is by far the most common and toughest question that FDA asks. Yes RNA, DNA, and even proteins stored under the right conditions can be stable for a long time. But how long? And under what conditions? How do you know? The scientific literature can help to support your claim, but you'll also need some stability data of your own. If you're planning on using retrospective samples, you'll want to check first with the FDA to find out what they might be worried about. They have recently seen a lot of banked samples turn up in applications, and they have some definite opinions. Oh, and make sure you have access to the storage data for the clinical samples you're interested in, so you can confirm that the planned storage and the actual data match up.
- Do you have access to all the samples and data? Sometimes there are restrictions on how much of the clinical data you can access on these samples. It's important to make sure up front that you can get your hands on all of the data you'll need to support your clinical validation. And it's a good idea to audit some of the data early on, so that you can feel comfortable that the documentation is accurate and complete. Remember, just because it's part of a drug trial doesn't mean that all the data is there and going to meet your needs.
- Is the retrospective trial design relevant? This is a biggie, and worth some due diligence on your part. Just because you've found samples from some awesome cohort doesn't mean the specific trial design for a new drug meets your target outcomes. Do they have samples for each time point? Is there a complete set of supporting clinical data for each sample? Most importantly, does that arm of the trial match the demographics of your target population? Any of these can leave you with significant gaps in your clinical validation.
- Is there a paper trail? Just because these samples were part of a clinical trial doesn't mean they are perfect. Any clinical trials coordinator will tell you that even the best planned studies can have data problems. Auditing the trial records is one step in the process, but you also need to look at the chain of custody after the study was completed. This includes a review of the facility and the procedures they have in place for storage, sample security, etc. Samples can get moved, data can get filed, and all of this can cause problems when it comes time to test the samples or analyze the data.
- Do you need IRB or center approval? Chances are you do, especially if you want access to the clinical outcome data. That means you may need to have IRB approval of your protocols prior to the samples being released. And in some cases this can take months to acheive. Nothing can kill a timeline faster, so it is in your best interest to make sure you have the okay to use the samples before you get too far into development.
- Why does the FDA care about this anyway? It's always a good idea to spend some time thinking like a FDA reviewer. Take sample stability for example. If you don't know that the clinical samples are slightly degraded, and you develop and test your algorithm on those samples, it's possible that when the test hits the market you could have problems, since the values you expected to see are now much higher than what you saw in clinical trials. Does that change your false positive rate? What about the negatives? The further your clinical trial samples get from your target population, the more nervous they will get and the more data they will need to be convinced the risk is minimal. Remember, they really prefer prospectively samples colected under a sponsor-specific protocol, so you should be prepared to have some study that has samples collected and tested in "real-time".
Finally, if you are considering using retrospective samples in a clinical trial, it's a good idea to have a pre-IDE meeting with the FDA to talk about it. These meetings can be very productive, as you can get some feedback on the design of your trials before you get too far into the project. It may be painful to prosepctively collect samples, but it's a lot worse if you have to do it after your submission.