John Cramer, MD

Sharing the results of ReACT 1.0 in @NatureComms -- the first study to use HPV ctDNA to guide CRT de-escalation in higher risk HPV+ OPC. ctDNA metrics may improve risk stratification. Grateful to our coauthors. @Naveris_inc @DanaFarberNews @jdschoenfeld1 nature.com/articles/s4146…

Clear trend: more #preference #signals in medical residency application→ less overapplication. As signals increase, apps/applicant decline—reducing cost, noise, and inefficiency. Better alignment + less admin burden—is it for everbody? #OtoMatch #MedEd #GME #Match2026

4/4 OS is immature — but the CPS ≥10 HR is trending toward significance, and the Qian Markov model projects a 2.61 undiscounted LY gain (11.82 vs 9.21 yr life expectancy). Dismissing EFS as uninterpretable risks underselling a plausible forthcoming OS benefit.

1/4 Awan et al. (JCO) argue KEYNOTE-689 OS data are too immature to justify perioperative pembrolizumab as SOC and advocate restricting it to CPS ≥10 patients at low surgical risk. ascopubs.org/doi/10.1200/JC…

5/ The advance: NPC risk moves from a diffuse "polygenic + viral + environmental" model to a biologically interpretable host-pathogen interaction with discrete, targetable elements — with direct implications for screening, vaccines, and T cell therapy.

4/ This reframes 15 years of NPC GWAS data. HLA-A*11:01's protection isn't intrinsic — it's contingent on which EBV strain you carry. The host genotype and viral genotype only make sense together.

1/ EBV infects >95% of adults worldwide. Yet ~50% of all nasopharyngeal carcinoma cases arise from ~2% of the global population in southern China. A new Nature paper finally explains why. nature.com/articles/s4158…

Come for the critically important data on protons for oropharyngeal cancer, stay for the beautifully written paper. TORPEdO is out, and it is fascinating and instructive. I will warn you upfront that this is a long thread, but there is a lot here to discuss! thelancet.com/journals/lance…

PFS ≠ OS. ORR ≠ OS. No validated surrogate for OS exists in R/M HNSCC. LEAP-010: if lenvatinib had been submitted for accelerated approval on ORR or PFS, we'd be in trouble. OS must be the primary endpoint in HNSCC IO trials.

LEAP-010 results are in: lenvatinib + pembrolizumab improved ORR (+20%) and PFS (HR 0.64) vs pembro alone in R/M HNSCC — but OS was numerically worse (HR 1.15). A cautionary tale about surrogate endpoints. ascopubs.org/doi/10.1200/JC…

🚨 JAMA flags admin and access delays as safety hazards in care. In #headandneckcancer, this can derail trials & immunotherapy starts, likely worsening outcomes. Insightful viewpoint by Drs. Hassid and Kaafarani #JAMA #OncologySafety #Oncology #PatientSafety @MDAndersonnews

7y in the making!!! As I always spoke about this, the field of #NasopharyngealCarcinoma leads the way in driving individualised treatment in #HeadNeckCancer When we published our initial idea of stratifying patients by their early treatment response using #EBVDNA back then in 2018, we decided to test our idea formally with a trial, and kudos to a very talented #biostats and willing collaborators, we came up with EP-STAR Very glad to see it finally published today in @Nature 🍸 nature.com/articles/s4158…

In 2024 we showed that Modifier 22 doesn't actually pay more for complex cases owing to higher denial rates. The @AmCollSurgeons GSCRC committee has been working on this issue ever since... see our summary of what we have tried to do and what's next: jamanetwork.com/journals/jamas…

My take: The study’s strongest message isn’t “surrogate endpoint solved.” It’s “a 24-month checkpoint powerfully separates trajectories.” Use it to detect signal earlier—not to declare victory or stop measuring late harms.

Surrogate endpoints can speed answers—but they can also blind us. In HPV+ OPSCC, does being progression-free 24 months after RT mean we’re basically “done,” or are we missing the real story: late toxicity and non-cancer mortality? pubmed.ncbi.nlm.nih.gov/41643696/