onc brain

Secondary analysis of the ProtecT RCT. Cribriform status was not a randomization characteristic; biopsy slides from 712 of 1,643 randomized pts were centrally reviewed, with centralization still ongoing.

PSA-detected clinically localized prostate cancer. ~13% cribriform-positive on biopsy, 87% cribriform-negative. Gleason re-graded to 2019 ISUP from the original 2005 criteria, which did not change the result.

The radical-treatment options were surgery or EBRT with 3-6mo neoadjuvant ADT. The RT+ADT arm carried a higher cribriform prevalence, attributed to chance since cribriform was not randomized.

Primary outcome was metastases at 15yr median follow-up, analyzed by both intention-to-treat and per-protocol (accounting for crossover between assigned and received treatment).

In cribriform-negative pts (87%), early radical Rx gave no significant 15yr metastasis reduction vs active monitoring, in ITT and per-protocol. Cribriform-negative GG2 matched GG1 metastasis risk on multivariable Cox.

The cribriform-positive group carrying the clinical message is small (~13% of the 712 reviewed), and its metastasis effect size is not reported in source. The 20yr ProtecT follow-up (census just reached) is not yet available.

Post-hoc analysis of a non-randomized histologic feature; only 712 of 1,643 randomized pts had slides reviewed. Cribriform-positive effect size not reported in source.

ASO editorial (Naoum, Taghian) critically evaluating SUPREMO. SUPREMO randomized post-mastectomy pts to chest-wall irradiation vs observation; RNI was prohibited, so it tested chest-wall RT alone, not chest-wall plus nodal PMRT.

Only 25% truly node-negative; the majority had N1 (1-3 positive nodes). 65% hormone-receptor-positive, 10% triple-negative. Post-neoadjuvant pts are out of scope.

Chest-wall irradiation alone. Supraclavicular coverage reached only 12% of the PMRT arm (n=97), internal mammary RT under 2% both arms, and regional nodal irradiation was otherwise prohibited.

No DFS or OS benefit from chest-wall irradiation; the only positive signal is reduced local recurrence (see table).

Danish and British Columbia trials, plus MA-20 and EORTC 22922, showed regional nodal RT improves LRR, DFS, and breast-cancer mortality in 1-3 node+ disease, even after ALND. SUPREMO's chest-wall-only field never tested this.

Node-negative dilution (25%) may mask node-positive benefit; incomplete nodal field; TNBC (10%) underpowered with a paradoxical forest-plot direction; only 10-year follow-up for a 65% HR-positive cohort prone to late recurrence.

The editorial's read: SUPREMO's null headline should not license omitting PMRT in node-positive pts, because it never tested chest-wall plus nodal RT. Even its truncated field improved node-positive local recurrence, consistent with the Danish, British Columbia, and EBCTCG data.

SUPREMO tested chest-wall RT alone, prohibited RNI, and diluted with 25% node-negative pts, so its null DFS/OS can't establish PMRT omission for node-positive disease.

Individual patient-level meta-analysis pooling 7 institutions, N=248 (59 DE-RT, 189 SD-RT). Retrospective, non-randomized; compared by Kaplan-Meier, Cox multivariable, and IPTW propensity analysis.

High-risk meningioma: 75.8% WHO grade 2, 41.5% recurrent (grade 3 the remainder), 75.2% subtotal resection.

DE-RT defined as biologically effective dose ≥79.2 Gy (equivalent 66 Gy/33 fx); SD-RT comparator conventionally fractionated 59.4 Gy/33 fx or 60 Gy/30 fx. Mixed photon/carbon vs photon-alone DE-RT showed no PFS difference (81.3% vs 92.0% at 3y, P=.34).

Primary: progression-free survival, DE-RT vs SD-RT. Also overall survival and CNS radionecrosis.

OS not improved despite the PFS gain: 5-yr OS 83.8% vs 68.4% (P=.056 UVA), non-significant on MVA (HR 0.66, P=.27) and IPTW (HR 0.77, P=.42).

Any-grade radionecrosis higher with DE-RT (33.9% vs 13.2%, P=.001) but grade 3+ similar (5.1% vs 3.2%); most RN was low-grade.

Standard-dose postoperative meningioma RT (RTOG-0539 high-risk 60 Gy, EORTC-22042 60 Gy) sits at/below this cohort's SD-RT arm; the dose-response signal here motivates the ongoing randomized escalation question.

Retrospective non-randomized pooling: DE-RT allocation confounded, IPTW mitigates but cannot fully adjust. PFS gain without OS benefit; prior-RT and grade imbalance across arms.

Retrospective non-randomized IPD pooling; DE-RT allocation confounded despite IPTW. PFS-only gain, no OS benefit. Signal supports escalation but needs randomized confirmation.

Randomized, apalutamide + ADT vs ADT alone. Results just posted (NEJM online); full ASCO26 presentation pending, so this is a preview read.

Most distant metastases were PSMA PET-detected: 53.0% in the apalutamide group vs 60.7% in the control group. Primary MFS effect size not reported in source.

MFS events were predominantly PET-detected, raising lead-time / detection-bias concerns about the endpoint. The 50-vs-60 BCR/100 figures circulating are an illustrative hypothetical, not PROTEUS primary data.

Curator frames it as 'homerun vs largest negative' pending full data. A companion thought experiment argues PSMA PET vs conventional detection differences could account for much of the apparent metastasis delta.

Preview NEJM data; primary effect size not in source. Most MFS events PSMA PET-detected (53.0% vs 60.7%), raising lead-time/detection bias that undermines the MFS read.

Territory-wide real-world cohort (OCEANUS, Hong Kong CDARS, >90% population coverage); NSCLC diagnosed 2010-2021 who received iRT. Overlap-weighting propensity score primary, IPTW sensitivity; landmark-based OS with weighted Kaplan-Meier and Cox (restricted mean survival time where PH violated).

335 of 3522 ICI-treated pts received RT: 155 newly-dx advanced, 180 refractory. Median age 64 (34-90), 73.7% male. Refractory analysis required survival ≥90 days (landmark).

RT dose, fractionation, modality, and target volume not reported in source. The variable studied is timing of RT relative to ICI (sequential vs concurrent), not technique.

Both comparisons favored the sequential / ICI-maintenance arm (magnitudes in table). Chemotherapy was associated with longer OS in newly-dx pts only; not significant in refractory disease.

Randomized data on iRT sequencing in advanced NSCLC are limited (stated by authors). RT-before-ICI direction is consistent with the PACIFIC consolidation paradigm, but PACIFIC treated curative-intent unresectable stage III with concurrent chemoRT, a different population.

Observational: sequential vs concurrent not randomized, residual confounding by indication despite weighting. Small subgroups (155 newly-dx). Refractory comparison non-significant (P=.20).

Observational real-world cohort; sequential-vs-concurrent not randomized, confounding by indication despite propensity weighting. Small subgroups (155 newly-dx). Authors label it hypothesis-generating.

Post-hoc predictor analysis of the pivotal single-arm phase 2 celiac plexus SRS trial (NCT03323489, Lancet Oncol 2024). 90 evaluable pts; univariate/multivariate logistic regression with 500-iteration bootstrap internal validation.

Pancreatic-cancer pts with intractable retroperitoneal (celiac) cancer pain treated on the phase 2 SRS trial. Age and BMI were the prior-reported predictors; median baseline values not in source.

Celiac plexus radiosurgery, target D12-L2. Dose and fractionation not reported in source (specified in the parent Lancet Oncol trial), so transferability of the technique can't be judged here.

Pain response = ≥2-point reduction in average BPI-SF pain from baseline to 3 weeks. Score performance judged by optimism-corrected AUC.

Two independent predictors (severe baseline pain, no prior neurotoxic chemo) build a 0-2 score with a monotonic response gradient (see tables). Discrimination modest, AUC 0.714 corrected.

Celiac plexus SRS is a novel non-invasive palliative option recently added to NCCN guidelines. This analysis converts the parent trial's flat response into an actionable selection score and argues for earlier use, before neurotoxic chemotherapy.

Post-hoc, single-arm data; predictors are associative. Internal bootstrap validation only (optimism-corrected AUC 0.714), external validation required. Subjective 3-week pain endpoint; score not prospectively tested.

Post-hoc predictor modeling of a single-arm phase 2; internal bootstrap validation only, external validation pending. Associations, not a prospectively validated selection tool.

Secondary analysis of SWOG S1007 (RxPONDER), a phase 3 RCT of chemo omission by 21-gene recurrence score. RNI receipt was prospectively recorded, not randomized. Median follow-up 6.1y; survival landmarked at 1y.

HR+/HER2−, Oncotype RS ≤25, 1-3 positive nodes. 4871 had RT forms; 81% received RT, and 59% of those with target data received RNI (targeting ≥ supraclavicular region).

RNI defined as targeting at least the supraclavicular region. The 59/41 split in RNI use reflects genuine practice equipoise in favorable N1 disease.

LRR was low across every locoregional approach; only mastectomy without RT reached 1.7%. IDFS did not differ by RNI in either menopausal group (see tables).

Consistent with MA.20 / EORTC 22922, where RNI's benefit was concentrated in higher-risk node-positive disease. TAILOR RT (MA.39) is randomizing RNI in exactly this RS-low N1 population.

RNI comparison is non-randomized: confounding by indication (higher-risk pts selected for RNI) can mask a true RNI effect. Landmarking at 1y and diverse-setting RT reporting add noise.

RNI vs no-RNI comparison is non-randomized within RxPONDER; confounding by indication limits the null IDFS and low-LRR read. Randomized RNI trial (TAILOR RT) pending.

Retrospective, IRB-approved, 2 centers within one institution, Jan 2017 to Dec 2023. N=175 (89 proton, 86 photon). CC estimated by Kaplan-Meier; Cox proportional hazards for HRs, logistic regression to verify. Median follow-up 42 mo (proton), 47 mo (photon).

Breast cancer pts with subpectoral 2-stage tissue expander/implant (TE/I) or direct-to-implant (DTI) reconstruction receiving PMRT. Median age 49 (24-78); 63% Hispanic. All TE/I pts had the expander irradiated. Groups imbalanced on laterality (P<.001) and reconstruction type (P<.001).

PBS proton vs IMRT photon PMRT. Dose and fractionation not reported in source. In TE/I, the tissue expander itself was the irradiated target.

Proton vs photon CC: univariate HR 2.3 (1.26-4.30, P=.007), attenuating to HR 1.76 (0.93-3.32, P=.083, ns) on multivariable. DTI vs TE/I: HR 3.0 (1.7-5.5, P<.001), the dominant driver. Per-group 2yr CC rates in the table.

Prior proton PMRT evidence centered on improved cardiac and pulmonary dosimetry; head-to-head capsular-contracture data proton vs photon were scarce. This is the largest proton reconstruction cohort reported to date.

Retrospective, single-institution, N=175, with groups imbalanced on reconstruction type and laterality. The proton effect lost significance after adjustment (P=.083), so residual confounding (reconstruction type driving both proton selection and CC) cannot be excluded.

The modifiable variable is reconstruction, not beam: DTI carries roughly 3x the CC hazard, and proton stacked on DTI reaches 50% 2yr CC. Where proton is chosen for heart or lung sparing, staged TE/I may blunt the CC penalty; a prospective comparison is needed to confirm.

Retrospective single-institution cohort; groups imbalanced on reconstruction type and laterality; proton effect lost significance on multivariable (HR 1.76, P=.083). Signal, not confirmation.

Ancillary exploratory analysis pooling the non-RT arm of NRG/RTOG 9804 with the good-risk cohort of ECOG-ACRIN E5194. N=878 (317 + 561), median follow-up 14.85 yr.

Good-risk DCIS: low or intermediate grade, ≤2.5 cm, margins ≥3 mm, lumpectomy without RT. Median age 59. Tamoxifen users skewed toward negative re-excision and pathologic size ≤5 mm.

Tamoxifen optional and non-randomized, used by 43.1% overall (65.6% in 9804, 30.3% in E5194).

No RT in either cohort by design; this characterizes the RT-omitted good-risk population, not an RT comparison.

IBR overall, invasive IBR, DCIS-IBR, contralateral breast event, OS. Fine-Gray competing-risk models, univariate plus multivariable.

NSABP B-24 randomized tamoxifen after lumpectomy plus RT and cut breast events; this extends the tamoxifen signal to the RT-omitted good-risk setting, though non-randomized.

Tamoxifen not randomized; users differed on prognostic factors (re-excision, size), so residual confounding is likely. Exploratory combined dataset, not a prespecified endpoint.

Non-randomized optional tamoxifen compared within a post-hoc combined dataset; users differed on prognostic factors. Consistent with randomized NSABP B-24 signal but confounded here.

Post-hoc analysis of the randomised OPERA trial, N=141 evaluable. Tests an earlier response timepoint (W14, ~1 mo after NAT end) against the protocol W24 triad, scoring clinical tumor response (CTRE) by DRE + rectoscopy as cCR / nCR / PR.

OPERA randomised the boost after neoadjuvant chemoradiotherapy (EBRT boost vs contact X-ray brachytherapy boost); these slides don't label which arm is which. The higher early good-response rate in one arm did not translate to a 5yr organ-preservation difference (see table).

Post-hoc, non-randomised comparison of response groups; W14 selection not prospectively validated against the protocol W24 assessment. Single-trial, modest N; 5yr OP differences underpowered (p=0.24).

Authors' RT-relevant read: nCR reflects radiation change, not residual tumor, and matches cCR for organ preservation (77% vs 81%), so an nCR should not trigger salvage TME. Early DRE + rectoscopy can move W&W selection up to W14 with MRI confirmation.

Post-hoc analysis of an alternate (W14 vs protocol W24) timepoint; response-group comparison non-randomised and not prospectively validated for W&W selection.