Is genetic evidence worth a 40 percent cost cut?
The short answer is “yes”, and it comes from 2 totally different sources…
It’s hard to get reliable information on data science in R&D, and one way I look for signal, is when multiple people allude to the same thing. At #ddpeast26 in June, two speakers gave talks in different tracks, at different times, on what looked like completely different subjects. Neither of them heard the other's talk, but they landed on the same finding from opposite ends.
Shivani Nanda (Zifo) opened with the observation that our field has an abundance problem, not a scarcity problem. A decade of multiomic expansion, and roughly 90% of clinical programs still fail. Her framing was that data is not the bottleneck. Evidence quality is. And the evidence that actually moves the number is causal human genetics: targets with human genetic support succeed at roughly 2x the rate.
A track over, Sun-Gou Ji from Bridgebio was walking through how they decide whether a rare disease program is financially feasible (source). Rare disease forces the question in a way other indications don't. If the population is small enough, no amount of good science makes the arithmetic work, so they build the arithmetic explicitly: Monte Carlo Net Present Value (NPV) models per program, updated as assumptions change rather than built once and filed.
A few key numbers:
A dollar spent preclinically needs about $2.50 in post-approval revenue just to cover the time value of money. Risk-adjust it for stage-gate failure and it's closer to $18. Which is why an early R&D dollar is the most expensive dollar you will ever spend. Then he ran the equivalence, and this is the number I keep coming back to. Genetic support on a target has the same effect on NPV as cutting costs by 40% across the entire clinical lifecycle
Put the two talks side by side and you get something neither contains on its own. Shivani's 2x is a finding about how often programs succeed. Sun-Gou fed a number like that into a financial model and got out what it's worth: the same effect on NPV as cutting costs 40% across the clinical lifecycle. One is the probability. The other is the price of the probability. Most people reading this spend their careers making the first argument to people who only respond to the second.
We are going to revisit this topic in a panel, and in a talk from Sun-Gou’s colleague at #ddpwest26 in October.
Best,
Ilya Captain
DataDrivenPharma Founder
DDP West is October 15 and 16 in South San Francisco.
Early bird closes July 31 (This Friday!). Grab your passes here before they go up!
