onc brain

About · curated by Nick Boehling, MD · @nb2276

    2026-06-01 ASCO Annual Meeting 2026

    digest generated 2026-07-19

    DeLLphi-304: tarlatamab CNS PFS 6.5 vs 4.2mo (HR 0.40) in SCLC brain mets, first DLL3xCD3 intracranial signal but post hoc.
    TL;DR
    Thoracic and lower-GI led, both pitting molecular tools against RT-adjacent decisions. Tarlatamab's post-hoc CNS signal in SCLC (HR 0.40) is the first DLL3xCD3 intracranial read, historically WBRT/SRS/PCI territory. ctDNA in post-TNT rectal NOM misses 59% of local regrowths, augmenting but not replacing endoscopy/MRI.

    GI Lower

    ctDNA surveillance in organ-preservation rectal cancer: strong for distant mets, weak for the local regrowth watch-and-wait actually hinges on.

    Confirmatory

    ctDNA and Local Regrowth/Distant Mets in Nonoperative Management of Stage II-III Rectal Cancer

    ForMSS stage II-III rectal adenoca in cCR/nCR on watch-and-wait after TNT

    TL;DRctDNA sensitivity 41% for local regrowth vs 74% for distant mets; positive result predicts both but can't replace endoscopy/MRI surveillance in rectal NOM.

    Why it mattersRadiation oncology

    The asymmetry is the actionable read: ctDNA caught only 41% of local regrowths vs 74% of distant mets, so for post-TNT organ-preservation surveillance it flags distant-met risk but can't replace endoscopy/MRI. ctDNA+ at regrowth also tracked more advanced ypT3-4 salvage pathology (75% vs 21%).

    ctDNA and Local Regrowth/Distant Mets in Nonoperative Management of Stage II-III Rectal Cancer
    7 details 1 trial watching
    Design

    Prospective single-center cohort (MD Anderson INTERCEPT program), N=110, 2020-2024. Serial tumor-informed ctDNA (Signatera Exome), 669 samples; median follow-up 25 mo (IQR 18-37).

    Population & inclusion

    MSS stage II-III rectal adenocarcinoma in cCR (69) or nCR (41) after NAT, managed non-operatively. Median age 56; cT3 in 72 and cN1-2 in 60 of 110; NAT almost entirely TNT (104/110).

    Endpoints

    Regrowth-free and distant-metastasis-free survival by longitudinal and first-post-NAT ctDNA; per-sample diagnostic accuracy (±90d of each draw); salvage-surgery pathology by ctDNA status. No registered primary endpoint (observational).

    Results

    Local regrowth in 23 (21%), distant mets in 12 (11%). Ever-positive ctDNA tracked worse regrowth-free and distant-met-free survival (log-rank p=0.0002 and p<0.0001). Per-sample sensitivity 41% local regrowth vs 74% distant mets.

    EndpointSensitivitySpecificityAccuracy
    Local regrowth41% (12/29)94% (480/509)91% (492/538)
    Distant metastasis74% (31/42)97% (611/627)96% (642/669)
    vs leading data

    Consistent with the emerging ctDNA-as-prognostic signal in rectal NOM, but quantifies a sensitivity ceiling (41%) that limits its use for local-regrowth surveillance.

    Applies to
    MSS stage II-III rectal adenocarcinoma in cCR/nCR pursuing organ preservation after TNT
    Does not represent MSI-H tumors, operative-intent pts, or short-course-RT / chemoRT-only NOM pathways.
    Limitations

    Single-center, N=110 with only 29 regrowth- and 42 distant-met-associated samples driving diagnostic estimates. Assay-specific (Signatera tumor-informed); no comparison vs imaging-alone surveillance.

    Rationale

    Single-center prospective cohort (N=110), no interventional comparator; prognostic association consistent with emerging ctDNA-in-NOM data and reaffirms imaging/endoscopy surveillance can't be replaced.

    Monday clinic

    Applies to the MSS stage II-III rectal pt in cCR/nCR pursuing watch-and-wait after TNT, in whom a negative ctDNA still can't exclude local regrowth; it does not extend to MSI-H tumors or pts not pursuing organ preservation.

    Open questions
    • Does ctDNA-guided surveillance improve organ-preservation outcomes vs imaging alone
      recruiting ctDNA Monitoring to Predict the Efficacy of TNT for Rectal Cancer
      n=100 · primary completion 2025-12 · candidate match
    • Can serial ctDNA kinetics raise local regrowth sensitivity above 41%
    • Generalizability beyond single-center tumor-informed Signatera cohort

    Sourced from @NiuSanford

    📚 Sources · 🐦 1 tweet

    Curative Retrospective Confirmatory ASCO Annual Meeting 2026

    Thoracic / Lung

    A systemic bispecific posts an intracranial signal in SCLC, RT's historic domain for CNS control.

    Caveats dominate

    DeLLphi-304

    ForRelapsed SCLC with brain mets, >70% prior CNS-directed therapy

    TL;DRIntracranial: time-to-CNS-progression HR 0.54 ITT; in brain-mets pts CNS PFS 6.5 vs 4.2mo (HR 0.40), CNS CR 15% vs 5% for tarlatamab.

    Why it mattersRadiation oncology

    The RT read is intracranial control without new radiation: brain-mets CNS PFS 6.5 vs 4.2mo (HR 0.40) and CNS CR 15% vs 5%. But >70% had prior CNS-directed therapy, so this is salvage intracranial activity, not a defer-brain-RT signal in RT-naive disease. Informs sequencing systemic ahead of repeat cranial RT.

    DeLLphi-304
    ArmMedian CNS PFSHR (95% CI)
    Tarlatamab (n=254)NE (13.7-NE)0.54 (0.39-0.75)
    Chemotherapy (n=255)7.2 mo (5.6-NE)n/a
    +2 more figures
    DeLLphi-304
    ArmMedian CNS PFSHR (95% CI)
    Tarlatamab (n=67)6.5 mo (4.3-13.7)0.40 (0.24-0.66)
    Chemotherapy (n=56)4.2 mo (2.9-5.5)n/a
    DeLLphi-304
    CNS endpointTarlatamab (n=67)Chemotherapy (n=56)
    Complete response10 (14.9%)3 (5.4%)
    Disease control rate52 (77.6%)40 (71.4%)
    Median duration CNS disease control8.2 mo5.2 mo
    5 details 2 trials watching
    Design

    Post hoc intracranial analysis of DeLLphi-304, a phase 3 RCT of tarlatamab vs chemotherapy in relapsed SCLC. CNS endpoints were not prespecified primary. Data cutoff Jan 29 2025; median follow-up 11.4 mo (tarlatamab), 11.5 mo (chemo).

    Population & inclusion

    Relapsed SCLC. ITT CNS cohort n=254 tarlatamab vs 255 chemo; brain-mets subgroup n=67 vs 56. >70% of brain-mets pts had prior CNS-directed therapy.

    Endpoints

    Time to CNS progression or death (ITT, RECIST per investigator); brain-mets CNS PFS by mRANO-BM (BICR); best CNS response, CNS disease control rate, and response durations.

    Results

    ITT time to CNS progression HR 0.54 (0.39-0.75), median NE vs 7.2 mo. Brain-mets CNS PFS and response detail in figures.

    vs leading data

    First DLL3xCD3 bispecific intracranial signal in SCLC; CNS control historically relied on WBRT/SRS and prophylactic cranial irradiation, not a systemic agent.

    Applies to
    relapsed SCLC with brain mets, mostly previously treated for CNS disease
    Does not represent RT-naive or symptomatic brain mets needing upfront local therapy.
    Limitations

    Post hoc, unstratified brain-mets HRs, small subgroup N. >70% prior CNS therapy confounds de novo intracranial activity. No new safety data in source.

    Rationale

    Post hoc CNS analysis; brain-mets HRs unstratified with small N (67 vs 56), and >70% had prior CNS-directed therapy, confounding de novo intracranial activity.

    Monday clinic

    For relapsed SCLC with previously-treated brain mets, this supports systemic intracranial control as an option before committing to repeat cranial RT; it does not inform RT-naive or symptomatic brain mets, where upfront local therapy still applies.

    Open questions

    Sourced from @StephenVLiu

    📚 Sources · 🐦 1 tweet

    Systemic Palliative Phase 3 RCT Caveats dominate ASCO Annual Meeting 2026