Thoracic / Lung
One randomized phase 2 stopped early for toxicity, with dosimetry unremarkable and the harm signal tracking IO exposure rather than plan quality.
Consolidative TRT + Atezolizumab Maintenance in ES-SCLC NCT04462276
ForES-SCLC with ≥SD after carbo-etoposide-atezolizumab induction, unselected
6.7 vs 13.4 mo
HR 1.55 (95% CI 0.90-2.69), P = .34; primary end point not met
TL;DRPrimary EP missed: mOS 6.7 vs 13.4mo (HR 1.55, P=.34) with consolidative 30Gy/10fx; trial halted for fatal SAEs.
The failure is toxicity, not tumor control: PFS was identical (2.4 vs 2.6 mo) while fatal AEs hit 19.4% vs 3.0%, and TRT carried an AE HR of 2.47 (1.15-5.32). Dose was modest (30 Gy/10 fx, postinduction volumes, below OAR thresholds), so de-escalating the plan is not the obvious fix; baseline DLCO SB and radiation-induced lymphopenia are the selection levers.
In unselected ES-SCLC responding to chemoimmunotherapy, this argues against offering consolidative thoracic RT during atezolizumab maintenance off-trial, particularly with low baseline DLCO; it says nothing about limited-stage disease or thoracic RT given without concurrent IO maintenance.
The dose was already conservative (30 Gy/10 fx, postinduction volumes, below OAR thresholds) and no dosimetric parameter tracked with serious events, so de-escalating the plan is not an obvious mitigation. Concurrent vs sequential timing did not change AE risk either. Selection (baseline DLCO SB 45.3 vs 56.7 in fatal-AE pts) and lymphocyte-sparing planning are the live levers.
Arm B's mOS of 13.4 mo and 56.6% 1-yr OS beat the IMpower133 atezolizumab benchmark, so maintenance alone was performing normally and the detriment is attributable to the added modality. TRT carried an AE HR of 2.47 (1.15-5.32) while longer atezolizumab exposure did not, which supports continuing maintenance unmodified rather than adding thoracic RT off-trial.
12 details 5 trials watching
Multicenter open-label phase 2 randomized trial (TREASURE, AIO-TRK-0320) at 20 sites in Germany and Austria, accrual September 2020 to August 2022, follow-up to September 2024, database lock April 2025, post hoc OS update April 2026. Planned 104 randomized for 80% power on a 20% absolute 12-month OS improvement; halted at 68 on SMC recommendation.
ES-SCLC with at least stable disease after induction carboplatin-etoposide-atezolizumab. 34 per arm, mean age 63.3 vs 65.6 y, 63.2% male, ECOG 0-1, thoracic PR in 82.4% overall. Randomization stratified by brain metastases, induction response, and prophylactic cranial irradiation.
30 Gy in 10 fractions to the postinduction thoracic primary and involved lymph node volume. Doses and volumes sat within or below typical clinical ranges and below established organ-at-risk thresholds, and concurrent versus sequential delivery relative to atezolizumab made no difference to AE occurrence.
Primary: overall survival from randomization, by stratified log-rank and multivariable Cox in the ITT population. Secondary: PFS and frequency plus severity of AEs and SAEs. Sensitivity analyses in per-protocol and an adjusted ITT excluding fatal AEs.
SAEs 61.3% vs 18.2% (P < .001) and fatal AEs 19.4% vs 3.0% (P = .04), dominated by infection and respiratory events. Grade 3 or greater trAE rate in arm A (26%) exceeded published benchmarks for atezolizumab monotherapy and for consolidative TRT without immunotherapy, which is the argument for an interaction between the two modalities rather than either alone.
Arm B tracked or beat the IMpower133 atezolizumab benchmark (13.4 mo and 56.6% 1-yr vs 12.3 mo and 51.7%), so the control arm was not underperforming; arm A fell well below it. Prior prospective single-arm series of TRT added to chemoimmunotherapy reported no toxicity increase, while randomized PACIFIC-2 and CheckMate-73L in NSCLC both showed more fatal infections in the concurrent thoracic RT plus IO arms.
Early termination at 68 of 104 makes every efficacy estimate exploratory, and the OS confidence interval (0.90-2.69) crosses 1. Causal attribution of the deaths was contested: investigators called 4 of the arm A fatalities unrelated, the SMC reclassified 3 of those as possibly or probably related. Risk-factor analyses (DLCO SB, GTV, dosimetry) are small-N and post hoc.
The PFS/OS dissociation is the load-bearing observation. Identical PFS argues the RT did nothing systemically, so the OS gap most plausibly reflects treatment-related deaths rather than faster progression, a reading the adjusted-ITT sensitivity analysis complicates by remaining consistent after excluding fatal AEs.
| Endpoint | Arm A (+TRT) | Arm B | P |
|---|---|---|---|
| Any toxic effects | 30 (96.8%) | 25 (75.8%) | .02 |
| SAEs | 19 (61.3%) | 6 (18.2%) | <.001 |
| trAEs | 71.0% | 30.3% | .001 |
| trSAEs | 29.0% | 6.1% | .01 |
| Fatal AEs | 6 (19.4%) | 1 (3.0%) | .04 |
CONSORT flow
Randomized, prespecified OS primary, halted early for fatal SAEs; result contests the single-arm safety data that encouraged consolidative TRT in the IO era.
- Can DLCO or lymphocyte kinetics select pts who tolerate consolidative TRT active Thymus Dosimetric and Morphologic Predictors of Radiation-Induced Lymphopenia in Stage III NSCLCn=450 · primary completion 2027-12 · thymus dose + morphology as RT lymphopenia predictors
- Does lymphocyte-sparing planning mitigate the infection signal n=55 · primary completion 2023-05 · SBRT planning optimized to cut lymphocyte depletionn=212 · primary completion 2026-11 · randomises lymphocyte-sparing vs conventional thoracic RT
- Is the late 2-year OS crossover real or small-numbers noise recruiting Phase II Trial of Consolidative Thoracic Radiotherapy for ES-SCLC After Standard Care of Chemo-immunotherapy Phase NAn=104 · primary completion 2025-09 · consolidative TRT after chemo-IO, PD-L1 maintenancerecruiting Association of Thoraco-mediastinal Radiotherapy With Maintenance Immunotherapy Treatment With Atezolizumab Phase 2n=37 · primary completion 2026-12 · consolidative TRT + atezolizumab maintenance in ES-SCLC