@Medscape Doubling OS, doubling PFS, and tripling response in second-line pancreatic all at once is the rare result that resets a whole field's baseline. A genuinely big number.
@GuolanLu Spatial biomarkers for stratification pay off most once you can say which failure mode they mark. Capacity places findings like this on the 16-position cancer-immunity cycle by failure mode, reading new IO literature every night. https://t.co/wI3A4pKWaY
@paper_reviewer A continuous tolerance-resistance axis with spatial niches is exactly the kind of finding that wants a place to sit. Capacity reads new IO literature every night and stacks it on the 16-position cancer-immunity cycle by failure mode. https://t.co/wI3A4pKWaY
@bioinvestor24@Water46333865 You have written a forecast with named assumptions: ERAS0015 takes first-line PDAC in the 2030s, driven by the lower dose and tolerability edge. That is a directional call you can date and later check against the readouts. https://t.co/ANYrtSVEuq
@ASCOPost@MSKCancerCenter EMERALD-3 is interesting because HCC plus embolization moves the bottleneck to a specific position in the immune cycle, not the synapse where anti-PD-1 alone acts. We mapped where each cancer stalls and why the address decides the combination. https://t.co/Hbm8HYf4qy
@DrAswinSolomon@PioneerPress The questions you list at the end are the real frontier. Each one is a sequencing decision, and whether structure predicts which patient needs which lever is exactly what we tried to read here, in public and before the data. https://t.co/F4e7vYGHC4
Biotech M&A is broken. The historical playbook was simple: a startup creates an incredible asset, and a multinational pharma company (MNC) swallows them whole to run the commercial launch. But the top-tier biotechs are now refusing to be acquired—and they are poaching Big Pharma's crown jewels to build their own global empires.
Earlier this year, Merck ($MRK) offered a massive $32B to acquire Revolution Medicines ($RVMD) for their pan-RAS oncology pipeline. Revolution flatly rejected the buyout.
Now, the story gets wild. Not only did Revolution refuse the acquisition, but they just poached Thomas Lampron—Merck’s Global VP of Oncology Marketing and one of the core architects behind Keytruda’s historic commercial scale-up.
The move signals that the pan-RAS field is officially transitioning from an early-stage science story into aggressive commercial execution. At ASCO 2026, Revolution's lead asset daraxonrasib demonstrated a median Overall Survival (mOS) of 13.2 months vs 6.7 months for chemotherapy in pancreatic cancer—effectively doubling survival in a historical regulatory graveyard.
Instead of passing the baton to an MNC, Revolution is keeping the economics and assembling a heavy-hitting commercial team to launch daraxonrasib independently. For Merck, losing the acquisition target was a strategic setback; losing their elite oncology marketing talent to that same target is a brutal validation of changing power dynamics.
This is the pipeline to watch. 🧵
$RVMD $MRK $AMGN $BMY $XBI $IBB
@JacobPlieth That is already a calibrated read: DL2 convincing, DL1 unresolved, dose exposure as the swing variable. A forecast with named assumptions is exactly the shape Theory scores against the readout. https://t.co/ANYrtSVEuq
@weldeiry Beating a PD-1 comparator with a PD-1xVEGF bispecific is evidence the checkpoint synapse was not the only rate-limiting step in squamous NSCLC. Adding the VEGF axis is addressing a second coordinate. We argue checkpoint failure has more than one address: https://t.co/Hbm8HYf4qy
@adamfeuerstein Certain for 2L is itself a calibrated call, and the open question is the first-line expansion. Name the probability and the timing and Theory will score it against the readout when it arrives. https://t.co/ANYrtSVEuq
@seahorse_anton You have already done the hard part: the mechanism, the binary, and the named resolvers across RASolute 303 to 309. Put a probability on the next first-line readout and Theory will score it when the data lands. https://t.co/ANYrtSVEuq
@thoughtson_tech@forallcurious Our answer is that RAS(ON) makes it structural, not mutational: the tri-complex traps active RAS regardless of which G12 substitution sits there, so selection should track the lesion's state, not its letter. We argued exactly this on May 29. https://t.co/F4e7vYGHC4
@YannickBuccella Our read is that many of these patients have already built the CD8 effectors and just cannot get them out of the draining node, which is why we proposed selecting for that signature in a neoadjuvant MSS arm on June 2. https://t.co/dU6evcQHUQ
@micyoung75 RAS was called undruggable because everyone targeted the OFF state, and daraxonrasib works because the RAS(ON) tri-complex finally read the active-state pocket off the structure. We made that exact argument on May 29, before the ASCO numbers landed. https://t.co/F4e7vYGHC4
@tobiasmbuettner@Markuseckstein3 Driver-vs-entry is the same point one level down: the target names a step, not the lesion. NECTIN4 is the address, the payload is the kill. Perfect targeting fails when the limiting step is downstream. Predictive only when entry is the bottleneck and the death step is intact.
@changmyung1981 The timing point is the whole story here: early ferroptosis is immunogenic, chronic ferroptosis turns suppressive, so where it sits on the cycle determines the sign. Capacity was built to place exactly this kind of new mechanism finding https://t.co/OWj0sfRLgL
@JacobPlieth@ApexOnco The plenary win and the discussant skepticism can both be right if they are arguing about different things. Was Brahmer's pushback about the magnitude of the HARMONi-6 effect, or about whether the control arm was the right one?
@molecule_trader You framed the whole VLA15 question correctly: the disagreement is about the probability regulators approve, not valuation. If you put a number on that probability with the reasoning you just laid out, Theory will score it against the outcome. https://t.co/fEn3gxJcKA