Submission

Software to transform PI learning

In the 80% of cancer clinical trials that complete successfully lie the answers to closing the patient accrual gap. Software can help.

The immediate barrier that we are addressing is that 8% of current clinical trials terminate because of insufficient accrual. This translates to $120M a year of research that doesn't advance our scientific progress on treating cancer.

This "accrual gap" exists for many reasons: lack of patient and physician awareness, inadequate support and infrastructure, and negative perceptions, to name a few. Cancer clinical trials are very diverse, and each new clinical trial is unique, so an effective solution will have to tackle these 2nd level obstacles in an "individualized" way.

The first link details the numbers behind the current 8% rate of cancer clinical trials failing to complete because of insufficient accrual. The second link describes a research study that shows that cancer clinical trials are diverse in their characteristics. The third link has numbers on the expected dollar cost to run a clinical trial.

http://www.medscape.com/viewarticle/819930

http://www.ncbi.nlm.nih.gov/pubmed/23699837

http://blogs.wsj.com/venturecapital/2009/09/21/the-hidden-costs-of-running-a-biotech-company/

Troy Shu Clinical Trials HeroX Solution.pdf

Potential obstacles exist around user acquisition, competition, and market size. Getting already strapped for time principal investigators to use our product can be a challenge. Competition, e.g. from CROs, is another risk. The potential market size also needs to be large enough to sustain the costs of developing such a solution.

We believe that if our solution solves a big pain point for principal investigators, then they will be willing to use our product and provide feedback. We don't think competition from CROs is a big risk because patient recruitment is not their core competency and they usually will be slower to move. The potential market size is large--non-cancer clinical trials is included--though serving and really understanding a small niche to start will allow us to grow without spreading ourselves too thin.

We think that a more useful metric is the percentage of cancer clinical trials that terminate because of insufficient patient accrual, and it is desired that this percentage approach zero. This "failure percentage" is currently at 8%: it is not unreasonable to say that it can be halved to 4% in 3 or 4 years. In absolute numbers, this is helping about 30 more cancer clinical trials complete every year, at current volume. We also hope that scientific progress is accelerated and that more clinical trials are conducted every year, which implies that a lower reduction in the "failure percentage" may still be respectable. Quantifying the impact directly attribute to our solution will be difficult--confounding variables like cost of a clinical trial, other changing characteristics of clinical trials, demographics trends will have to be controlled in a multiple regression. But we can get a rough picture about the impact we have by getting feedback from our users, or for example keeping track of how many completed clinical trials used our solution.

This response refers to our previously suggested metric, the percentage of cancer clinical trials that terminate because of insufficient patient accrual. Maintaining or increasing a reduction in this percentage depends on how well we acquire customers: the more principal investigators we have using our product, the more we can reduce the percentage of trials that terminate early because of low accrual. Luckily, as mentioned in the attached PDF, our digital information based solution has an inherent positive reinforcement loop: as the product helps PIs run clinical trials with sufficient accrual, those PIs contribute more information about their successful trials to the product’s database, which makes the product even more valuable and allows it to help even more PIs run successful trials.

There don't seem to be any immediately apparent regulations that apply to this solution, since it does not touch patient data and is not directly communicating with the patient.

We think one reason our proposed solution hasn't been suggested or tried before is the anchoring effects of the "3% of cancer patients participate in a clinical trial" and "85% of cancer patients aren't aware" statistics. This can lead to efforts of increasing patient and/or physician awareness, which may not be high ROI because clinical trials are so different and may require different recruiting techniques. As an example, here (https://goo.gl/drKIfl) is a spreadsheet model showing that at the current levels of cancer clinical trials that take place very year, even if all trials were given 100% accrual rates (excluding the trials that terminated for other reasons like lack of funding or toxicity), the number of cancer patients that participate in a clinical trial would only be able to be increased by less than 10%.

Another reason our software-based solution hasn't been tried before is that only recently has the structured data on clinicaltrials.gov reached a sufficiently large size that insights can be learned from it. CROs also do not focus exclusively on patient recruitment, they have many other services that help institutions and corporations run clinical trials effectively, so provide manual and potentially inefficient solutions to the accrual problem.