At the European Lung Cancer Congress 2026, the Phase III LATIFY trial reinforced a hard truth in drug development: even the most compelling biological hypotheses do not always translate into meaningful clinical outcomes for patients.
The trial evaluated the addition of the ATR inhibitor ceralasertib to PD-L1 blockade with durvalumab, compared to docetaxel, in patients with advanced non-small cell lung cancer (NSCLC) who had progressed after immunotherapy and chemotherapy. In a setting where treatment options remain limited, the outcome was clear -- the combination did not improve survival or disease control, despite demonstrating a more favorable safety profile.
As someone who has spent years working alongside pharmaceutical companies, I know that this type of setback isn’t an oncology problem, it’s a drug development problem.
Across therapeutic areas, we continue to see late-stage clinical failures despite strong early data. In neuroscience, multiple Alzheimer’s disease programs, including therapies targeting amyloid and tau, have failed to meet primary endpoints after years of investment. In immunology, several IL-23 inhibitors and an anti-ST2 monoclonal antibody have fallen short in clinical trials for inflammatory diseases. And in cardiometabolic disease, promising mechanisms targeting obesity, NASH, and cardiovascular risk have not consistently met their clinical trial goals. The pattern is consistent: biological plausibility alone does not translate to clinical benefit.
While the LATIFY trial is just one of the most recent examples, more than 5,000 NSCLC clinical trials have been initiated since 2019 and yet patients who progress after immunotherapy still face limited treatment options. What does that tell those of us who’ve dedicated themselves to helping bring more, safer, and more effective medicines to patients?
We need to do better.
The gap between biological rationale and clinical success remains one of the most persistent and costly problems in our industry. When trials fail, the impact is far-reaching: researchers lose years of work, companies absorb substantial costs, and most importantly, patients are left waiting for better solutions. The core issue is not a lack of innovation but a lack of early, reliable signals about what will and won’t work.
This is exactly the challenge we are addressing at QuantHealth. We developed an AI model to predict lung cancer patient outcomes across time and used it to simulate the LATIFY trial results eight months before the readout. Our model predicted outcomes closely aligned with what was ultimately observed.
Our model predicted the comparative treatment effect observed in the trial, both in magnitude and statistical significance:
For median overall survival (OS):
For median progression-free survival (PFS):
While small differences are expected between published and simulated data, the real value lies in what can be understood earlier. Clinical trial simulations can surface likely efficacy outcomes, identify responsive subpopulations, evaluate alternative trial designs, and test different combination strategies -- among other adjustments -- before they reach patients. This shifts drug development from a reactive “wait and see approach” to proactive decision-making.
The future of drug development will not be defined by better science alone: it will defined by better decisions made earlier. The companies that win won’t just generate better science they’ll predict it and make better decisions, earlier.
At QuantHealth we do this by combining AI with deep clinical expertise. It’s how we’ve established ourselves as the most advanced clinical trial simulation company, with 600+ clinical trials simulated, achieving up to 90% accuracy across 30 therapeutic areas, including oncology, respiratory, immunology, inflammation, cardiovascular and cardiometabolic diseases.
To see how clinical trial simulations can support your drug development process, email info@quanthealth.ai.