Why releasing the brakes
works for only some
For a century the goal was to boost immunity. It kept failing. The answer was the other way round — take your foot off the brake.
That idea won a Nobel Prize and produced responses lasting years in some patients. Yet in most cancers only a minority respond. These 20 lessons follow why.
Not just in words: you open 10,071 TCGA tumours, draw the checkpoint correlation network, and score four predictive markers against lymphocyte infiltration measured from H&E slides. Then you compute synergy from a mouse 2×2 — and work out why no p-value can be attached to it.
How you'll learn
Not a simulation. Checkpoint gene expression across 10,071 real TCGA tumours from cBioPortal, joined with TMB and H&E lymphocyte infiltration to give 4,541 fully annotated samples.
Scoring an expression-derived marker on the same expression data is circular. So the ground truth comes from H&E microscopy instead — an entirely different measurement from RNA-seq.
In lesson 15 the mouse 2×2 shows synergy at 6.9× the additive expectation. But with one mouse per arm the smallest attainable p-value is 0.25 — regardless of effect size. You count that out yourself.
Clinical response rates never enter a calculation cell. They are not our measurements — and putting someone else's numbers beside your own makes it impossible for a reader to tell which is which.