onc brain

Single-institution retrospective review of consecutive pts treated 1966-1997. 468 pts, 531 lesions; median f/u 5.8 yr (range 2-24).

Biopsy-proven epithelial skin cancer: 389 BCC, 142 SCC; 364 previously untreated, 167 recurrent. Median age 73 (BCC) and 72 (SCC).

Superficial x-rays 60%, electrons 19%, electrons + superficial 20%, megavoltage photons <2%. Volume was tumor plus microscopic extension, with an additional 1 cm margin depending on size; lead eye shields when near the eye, and bolus use evolved over time.

Local control (relapse-free from RT start), cosmesis (telangiectasia, pigment change, fibrosis) and complications.

Overall LC 89%; untreated 93% vs recurrent 80%. On multivariate analysis, fraction size (p=0.01), lesion diameter and histology (p<0.01) predicted local failure; modality did not (p=0.345).

Complications 5.8% overall (13% for lesions >5 cm), mainly soft-tissue necrosis: 9% of SCC vs 2% of BCC. Fair/poor cosmesis was 4% with superficial RT vs 12% electrons and 19% combination.

The same institution's 1986 series (Lovett et al.) found electrons worse on multivariate analysis. Here that association disappears, which the authors attribute to wider electron margins in later years. The <2 Gy/fx failure signal matches Silva et al. in pinna lesions.

Treatment spans three decades with evolving bolus and margin policy. The cosmesis data are missing for 15-25% of untreated lesions, and the SCC dose cells are small, so the absent SCC dose response may reflect low power.

The data support RT as definitive therapy for early BCC/SCC and put the main lever on fraction size and lesion size, not beam type. Recurrent SCC carries the poorest control and the most cosmetic cost, so failure at first treatment is expensive.

Retrospective single-institution series spanning three decades; modality chosen by lesion size, not randomized, so dose and modality associations are confounded. No surgical comparator.

Retrospective single-centre series, 434 tumours in 333 pts treated January 2001 to December 2006. Analysis per tumour; LRF and MFS as one minus cumulative incidence with death as competing risk.

Biopsy-proven BCC (n=332) or SCC (n=102), primary RT (n=386) or postop for residual disease (n=48). Median age 74.8y BCC, 78.3y SCC; 59.4% face, 22.4% elsewhere on head.

Electrons, 54 Gy/18 fx of 3 Gy (n=159) or 44 Gy/10 fx of 4.4 Gy (n=275), four fx per week. 4-12 MeV by tumour depth, tumour + 1 cm margin, copper foil to offset skin sparing.

No significant LRF difference between schedules in BCC (p=0.67) or SCC (p=0.46). T stage was the only factor associated with local recurrence (p=0.036); gender, age, histology, indication, site, dose and lesion number were not.

Toxicity not reported beyond cosmesis. In a 14-pt pilot, pt-scored good/fair cosmesis was 75% (54 Gy) and 100% (44 Gy), observer-scored 87% and 83%, no significant difference.

The authors cite RT local control of 87-97% for BCC and 56-91% for SCC in prior series, placing these rates at or above that range. They describe them as comparable to surgical series.

Frail pts were moved to 44 Gy before it became standard, which may bias toward fewer recurrences being seen (competing deaths) rather than toward better control. SCC CIs are wide (54 Gy lower bound 79.1%), so equivalence for SCC is not shown.

The authors argue follow-up was adequate because recurrences came early (median 10.4 mo BCC, 3.3 mo SCC). They conclude 44 Gy/10 fx is optimal for T1-T2 disease. The data support non-inferior-looking control at small size, not a formal non-inferiority claim.

Single-centre retrospective; schedule assigned by era and frailty, older 44 Gy cohort with shorter follow-up; univariate only, 15 events total. Cosmesis from 14 pts.

Single-centre analysis of a prospectively maintained database, RT delivered 2007-2011. Recurrence was assessed clinically. Univariate then backwards logistic regression, per tumour, with no correction for multiple lesions per pt.

389 pts, 448 BCC/SCC lesions, median age 78 (41-99). 67.9% <2 cm; 9% in immunosuppressed pts. Primary 53.3%, recurrent 29.7%, postoperative 16.7%.

Orthovoltage 100-300 kV, electrons 6-18 MeV with bolus, or 1-2 field photons (IMRT selectively). Most common schedules 50 Gy/20 fx and 40 Gy/10 fx. GTV by clinical mark-up plus a 1-2 cm margin; CT planning for deep tumours.

LC 84.2% at median f/u 18.4 mo; median time to recurrence 11.4 mo. Independent predictors were age, size ≥2 cm, immunosuppression and modality; EQD2 and indication (primary vs recurrent vs postop) were not.

LC sits below the ranges the authors cite (BCC 87%-98%, SCC 56%-97%). The van Hezewijk series used higher-EQD2 schedules (44/10 vs 54/18) with 3y LC 96.1% and 96.9%, but the populations are not comparable.

Logistic regression treats a time-to-event outcome as binary, so pts with 0-mo follow-up count as controlled. The 3-category modality variable cannot separate technique from tumour depth. Recurrence was not split into in-field vs marginal.

The independent risk factors are patient and tumour features, not dose, which points to selection and target coverage rather than prescription as the lever. The modality OR reflects case mix per the authors and should not drive technique choice.

Single-centre retrospective analysis; logistic model ignores censoring, modality effect confounded by indication, and 18.4-mo median f/u undercounts late BCC recurrence.

Prospective single-centre observational cohort following STROBE. Pts were resected between 1990 and 2001, follow-up closed Dec 2005, median f/u 43 mo. Cox models used forward selection, with a 200/415 learning-validation split for internal validation only.

615 assessable white pts with primary cutaneous SCC at any body site, median age 73, 5% immunosuppressed. Excluded: SCC in situ, keratoacanthoma, collision tumours, and recurrent SCC treated elsewhere.

No pt received adjuvant RT. Every resection reached tumour-free margins by 3D-histology, with re-excision as needed, so outcomes reflect surgery alone.

The source states the primary endpoints as time to metastasis and time to local recurrence, each measured from diagnosis of the primary. Seven candidate predictors were chosen a priori.

26 (4%) metastases and 20 (3%) local recurrences. No tumour ≤2.0 mm metastasized. The thinnest tumour that metastasized was 2.8 mm thick.

Clayman and colleagues reported LR in 12% (15 of 130) of primary SCC. Margin-controlled resection here gave 3% (20 of 615). The authors cite an Australian series of 266 pts with nodal disease that also supports tumour thickness over horizontal size.

Follow-up after the early period relied partly on mailed questionnaires to treating physicians. New field-cancerization SCC can mimic local recurrence. University referral may have enriched for poorly differentiated tumours.

Thickness, immunosuppression and desmoplasia outperformed Broders' grade and horizontal size as prognostic factors. The authors propose thickness-based T1-T3 cutoffs (≤2.0, 2.1-6.0, >6.0 mm) with "d" and "i" modifiers. The study does not test whether any treatment changes the high-risk course.

Single-centre prognostic cohort, model selected by forward selection with events per predictor below 5. Validated only internally; no external cohort.

Retrospective single-institution series of 671 lesions treated 1972-2007. Mean follow-up 38.016 mo (range 1-351). Actuarial cure rate by life-table method; chi-squared for dose and size associations.

Basal and squamous cell carcinomas of the skin overlying the nasal cartilages. Histology split of the full cohort not given in the source excerpt. Relapses included 10 lesions already recurrent after prior non-radiological treatment.

CRT in 668 lesions: 55-60 kV, HVD 2-12 mm, 5 Gy/fx twice weekly, total 30-75 Gy (median 55 Gy). Margin 0.5-1 cm, 1.5 cm for morpheaform BCC. Septal cartilage shielded with 1 mm Pb nostril inserts at the alae.

CR 98.80% at 1 mo; 5-yr cure 88.09%, 10-yr 86.51%. Recurrence 38 lesions (13 central, 25 marginal), onset 3-96 mo. No relationship between relapse and dose or size.

Cosmesis good or acceptable in 96.84%; 16 lesions not acceptable. Authors report no complications or sequelae, and flag the 7 newly not-acceptable cases for continued follow-up.

No contemporaneous comparator. The authors cite randomised facial BCC data comparing RT with surgery: cosmesis was similar on nasal skin but control favoured surgery, in tumors under 4 cm treated with mixed RT techniques. Mohs surgery, the usual comparator for nasal BCC today, is not addressed.

Toxicity is clinician-rated cosmesis without a named scale, and a mean follow-up near 3 years is short for late cartilage effects. The null dose-relapse finding is likely confounded by dose being chosen by lesion size.

The data support kV RT as an effective, cartilage-sparing option for nasal skin carcinoma, where tight adherence of skin to cartilage complicates excision. They do not establish superiority over surgery, and the marginal-failure pattern raises the question of whether margins were adequate.

Single-institution retrospective series, 1972-2007, no surgical comparator; the authors' claim of superiority over other modalities rests on cross-literature comparison this design cannot support.

Retrospective single-institution series, 114 pts and 134 lesions treated Nov 2006 to Dec 2011. Analysis per lesion; median follow-up 33 mo.

92 BCC and 42 SCC, all clinically N0. 77.3% definitive, 22.7% adjuvant after excision with positive or close margins.

Ir-192 HDR plesiotherapy, 3 Gy/fx, 3 fx/wk; mean 48.79 Gy (45-54) for BCC and 50.55 Gy (45-57) for SCC, prescribed 3-5 mm below the surface with a 0.5-1 cm margin. Leipzig 2 or 3 cm applicators for flat lesions ≤2 cm (75.4%); customized molds for raised or larger lesions.

Overall 3y/5y DFS 95.12%/93.36%, CR 98%. Recurrent lesions did worse than primary (85.9% vs 95.6% at 5y, p=0.003); applicator and definitive vs adjuvant did not separate DFS.

Acute RTOG G3 40.3%, G4 2.2%; toxicity higher with definitive intent (OR 2.928, p=0.013) and borderline with molds (p=0.067). Late G0-1 95.2%; cosmesis excellent/good 82%.

Authors place results alongside the Kohler Leipzig series (91% complete remission) and Guix's mold series (5y DFS 98%). They cite one randomized surgery vs RT trial as the only head-to-head data, with no numbers given in source.

Molds treated larger lesions, so mold vs Leipzig DFS and toxicity differences cannot be separated from size. Reported recurrence counts do not add up, and the acute toxicity figure differs between abstract and results.

Supports surface HDR as a definitive alternative to excision for small primary NMSC. It does not settle optimal fraction size, and the high acute G3 rate argues for weighing applicator and dose per fraction for larger lesions.

Retrospective single-institution series, no comparator arm; applicator comparison confounded by lesion size and location; DFS per lesion with 33-mo median f/u.

Phase 3, randomized 1:1, N=231 (113 vs 118), accrued July 2021 to March 2023. Median follow-up 36.8 vs 35.6 mo at the October 1, 2025 cutoff.

cT3-4 or N+ rectal adenocarcinoma, tumor ≤10 cm from the anal verge. MMR/MSI composition of the enrolled cohort is not reported in source.

Experimental: SCRT 25 Gy/5f. Control: 50.4 Gy/28f with concurrent capecitabine 825 mg/m2 BID. Target volumes are not reported in source.

Both arms: 2 neoadjuvant + 6 adjuvant cycles of CAPOX. Experimental adds camrelizumab 200 mg d1 given alongside CAPOX for up to 17 cycles.

Primary: BIRC-assessed pCR rate (result not in source). Key secondary endpoints with prespecified α: investigator-assessed EFS and OS.

G≥3 TRAE over the whole treatment period 51.3% (59/115) vs 48.2% (54/112). The RT-attributable split is not reported.

RAPIDO set SCRT followed by consolidation chemo as a TNT option, and PRODIGE 23 did the same for an LCCRT-based approach. UNION is the first phase 3 to layer a PD-1 inhibitor onto SCRT in LARC, building on earlier SCRT + IO phase 2 signals such as TORCH.

The safety population (115 vs 112) does not match randomization (113 vs 118), so some pts received the other arm's treatment. Neoadjuvant CAPOX was only 2 cycles in the control arm, lighter than full TNT backbones. OS and surgical or organ-preservation outcomes are absent.

The EFS effect is large and was tested under alpha control, but it comes from a two-variable comparison. It supports the whole package and does not establish what SCRT or camrelizumab contributes individually.

Randomized phase 3, but the primary pCR result is not in source and EFS is secondary. SCRT and camrelizumab change together, so the benefit cannot be attributed to either.

Phase III, randomised 1:1, 25 centres in 9 countries, accrual 2018-2024. Powered (80%, alpha 10% two-sided) for a 10-point MDADI difference; 54 evaluable needed, 84 planned for 35% attrition, 86 randomised.

T1/T2 N0-1 OPSCC/SGSCC or T1 N0 HPSCC suitable for either TOS or DO-IMRT. Median age 64, 86% OPSCC, 49% HPV-positive; arms balanced.

Dysphagia-optimized IMRT (DO-IMRT). Dose, fractionation and target volumes not reported in source. 7/44 (16%) allocated to DO-IMRT declined treatment.

Primary: MDADI total score at 12 months (ANCOVA). Secondary: MDADI at 4.5, 6 and 9 months, PSS-HN, oncological outcomes, toxicity.

12-mo MDADI adjusted difference +3.9 points (90%CI -2.7 to +10.5; p=0.326), consistent across sensitivity analyses. Arm-level secondary outcomes are in the table.

G≥3 treatment-related AEs 21% in each arm. TOS: hemorrhage 5%, fistula 2%. DO-IMRT vs TOS: mucositis 39% vs 10%, dry mouth 49% vs 14%, dysgeusia 46% vs 17%.

ORATOR randomised TORS against RT in OPSCC and also found comparable function with different toxicity profiles. Best-of extends the comparison to supraglottic and hypopharyngeal primaries and to a phase III frame, but still with a small N.

The EFS HR CI (0.77-3.50) is wide and follow-up reaches only 2 years. A 49% HPV-positive mix limits transfer to HPV-positive de-escalation. RT technique is unreported, so the DO-IMRT arm cannot be audited.

With swallowing function apparently similar, toxicity profile and patient preference become the deciding factors. The numerically lower CR and EFS in the DO-IMRT arm (98% vs 82%; 84% vs 72%) are not significant and may partly reflect pts who declined RT, so they need longer follow-up before they carry weight.

Phase III, but differential 12-mo MDADI evaluability (86% vs 64%) and 16% refusal of DO-IMRT; CI does not exclude the 10-point margin, so equivalence is not shown.

30-question electronic survey sent to ~400 American Brachytherapy Society clinician members. 16 fully completed surveys analyzed; one incomplete survey excluded. Nearly all US respondents, one from Spain.

Radiation oncologists using skin surface brachytherapy for cSCC and BCC. Practice settings spanned academic, hospital-based and private; aggregate experience ranged 8 to 1800 cases.

Hypofractionation preferred; clinical-scenario doses ranged 30 Gy in 5 fractions to 64 Gy in 32 fractions (EQD2 40 Gy-65 Gy). Median max prescription depth 5 mm; shielded applicators most common, custom molds for large or curved areas. Respondents avoided daily schedules when hypofractionating.

Margin on gross disease 3-15mm, median 5mm. Max skin dose 125-150% (median 135%) of prescription; CTV coverage 90-100% (median 95%). No consensus on dosimetric constraints.

Eight respondents described severe acute and late toxicity on the lower extremity, especially pretibial skin. Three referred non-healing in-field ulcers for hyperbaric oxygen.

The cited prior literature (Guix custom-mold HDR, Paravati electronic brachytherapy, van Hezewijk hypofractionated electrons) reports single-series outcomes. This survey adds no outcomes; it documents how widely dose, margin and constraints vary across practices using the technique.

Several key tables carry ambiguous row-value pairing in the source, and the scenario table (Table 4) was not captured. Toxicity is anecdotal recall with no denominators.

The useful output is the spread itself: margins, skin-dose ceilings and the lesion-size threshold for gentler fractionation all vary several-fold. Prospective trials and expert consensus are what the authors call for, and this survey cannot substitute for either.

Practice survey of 16 self-selected respondents; no efficacy or toxicity endpoints measured, so the authors' 'convenient and safe' conclusion is not testable from this design.

A treatment-planning study with no patients. Plato BPS 14.2.4 plans were built on Nucletron radiographic cylinders (tip space ~0.5 cm) and CT/MR-compatible cylinders (tip space 0.6-0.8 cm). Dose uniformity at each depth was measured as the mean dose (MD) and SD across dose points spaced evenly around the dome and along the cylinder.

Cylinders were 2.0, 2.5, 3.0, 3.5 and 4.0 cm in diameter, with active lengths of 3 to 8 cm (upper third to the whole vagina). Prescription was to the surface (P0) or to 0.5-cm depth (P5), with or without optimization points on the curved end (NEP). The worked illustration used 7 Gy × 3 fractions.

Primary measures: MD and SD at the surface and at 0.5-cm depth. Dose fall-off was the ratio of MD at 0.75 cm to MD at 0.5 cm.

Surface prescription gave surface SD 6% vs 18% with P5 (2-cm cylinder, 0.8-cm tip), with almost the same fall-off beyond 0.5 cm. NEP plans produced hot spots >150% and cold spots <60%. A shorter tip space raised dosimetric uncertainty >30%.

The paper frames its result against the ABS 2000 guidance (dome optimization points, report doses at both surface and 0.5 cm) and an ABS survey in which >90% prescribe to 0.5 cm. It also cites Sorbe's randomized 2.5 vs 5.0 Gy × 6 trial, in which higher dose increased atrophy, bleeding and shortening.

The M regression was fit on only two lengths × two diameters, using one vendor's applicators and TPS. The claim that a lower surface dose reduces atrophy and shortening is extrapolated from other trials, not measured.

A 0.5-cm depth prescription delivers its intended depth dose only when dome points are included, and the surface dose it gives varies with cylinder size and tip space. Prescribing to the surface with a geometry-scaled dose fixes the surface number. Whether that changes toxicity remains open.

Planning-system dosimetry with no patients or outcomes. 'Surface prescription is optimal' rests on mucosal dose uniformity, a surrogate never tied to toxicity or vaginal control here.

Retrospective NCDB cohort, dx 2010-2013, analyzed 2018-2019. Univariate + multivariate Cox for OS; points assigned to features in proportion to effect size, then the composite was selected on C-index, AIC and clinical judgement.

Upfront surgery + adequate neck dissection (LN yield ≥15), non-metastatic T1-4 N0-3 M0 (AJCC 7th). HPV+ OPSCC n=2302 vs HPV-unassociated HNSCC n=13 022; OPSCC without HPV testing and surgery + chemo alone excluded.

5-y OS (KM, log-rank) and discrimination by C-index. Multivariate models adjusted for age, sex, race, comorbidity, insurance, adjuvant RT and/or CRT, and hospital type.

In HPV+ OPSCC only T stage, LVI, number of LN mets and ENE tracked with mortality; margins, LN size, level IV-V and bilateral nodes did not. Composite C-index 0.67 (3 tiers) vs 0.58 traditional.

The traditional tiers being tested come from NCCN, ASCO, ECOG 3311 and PATHOS definitions, the de-escalation trials that use pathology to allocate adjuvant RT vs CRT. ICON-S refined clinical staging for HPV+ disease but was built on definitive CRT with only 2% surgical pts, so it could not weigh pathologic features.

Points include margins despite no multivariate association, a judgement call that inflates their weight. The source states the traditional high-risk share inconsistently (46.1% and 45.1%), and HPV testing method is not described.

Traditional HNSCC pathologic tiers are weakly prognostic in HPV+ OPSCC, and nodal burden plus T4 matter more than ENE or margins. Whether the reclassified majority can safely receive less adjuvant therapy is untested here and belongs to the next deintensification trials.

Registry retrospective, OS-only, model scored in its own derivation cohort without external validation; cannot show reclassified pts tolerate less adjuvant RT/CRT.

Secondary dose-volume analysis of the HRT arm of the multicenter NRG/RTOG 0415 trial, N=521. Cohort split 70%/30% into training and validation with preserved toxicity rates; univariate and multivariate logistic regression on bootstrapping.

Low-risk prostate cancer treated on the HRT arm. Median follow-up 5.9 yr.

70 Gy at 2.5 Gy/fraction, 3DCRT/IMRT N=108/413. Doses converted to EQD2 with rectum α/β 3 Gy and bladder 6 Gy; trial rectal constraint EQD2 74 Gy ≤15%.

Late ≥G2 GI and GU toxicity, plus proctitis and pollakiuria (the symptoms with rates ≥10%). Generalizability required validation AUC within training AUC ± SD, p≤0.05, and pHL >0.05.

Rectal D5% was the dominant GI predictor; with or without acute GI toxicity it generalized (AUCvalidation 0.64, 0.65). Modeled late GI falls from 20% to 10% at D5% ≤62 Gy. No GU dose-response was found, including at α/β 0.8 Gy.

PROFIT, CHHiP, HYPRO and RTOG 0415 established moderate HRT with similar biochemical control to CFRT but a tendency to more late toxicity. The late GI rate here was 19% vs 2% in a cited high-dose CFRT IMRT series planned with a strict high-dose rectal constraint.

Validation is an internal random split of the same cohort, not an external dataset. The IMRT-heavy mix (79% vs 21%) left a technique interaction untestable, and constraint adherence was incomplete, so part of the signal may reflect plan quality rather than tolerance.

Results point to small rectal hotspots as the modifiable driver of excess late GI toxicity after moderate HRT, with authors also advocating rectal spacers. GU toxicity likely needs a substructure (not whole-bladder) approach.

Post-hoc NTCP modeling within one arm of a phase III; modest AUC 0.64-0.65, internal split validation only, no prospective test of the proposed D5% constraint.

Single-institution retrospective cohort at MSKCC, mastectomies January 2006 to December 2011. Median follow-up 5.6y. LRR defined as biopsy-proven ipsilateral chest wall or axillary, internal mammary or supraclavicular recurrence.

T1-T2N0 invasive breast cancer with ≥1 high-risk feature (age <40y, multifocal/multicentric, LVI, medial/central location, high nuclear grade). N0 meant no tumor in the SLNs, including ITCs and micrometastases. Neoadjuvant chemo, bilateral and recurrent disease excluded; 15 PMRT pts excluded, leaving 657.

98% had SLNB alone. 86% received adjuvant systemic therapy: 41% chemo plus endocrine, 14% chemo alone, 32% endocrine alone. 70% of HER2+ pts got trastuzumab.

No PMRT in the analyzed cohort. Only 15 of 672 (2%) received PMRT over the study period, so no RT comparator exists.

LRR 4.7% (31 pts), local and regional each 2.6%. Tumor size HR 1.70 (95% CI 1.26-2.29); systemic therapy HR 0.40 (95% CI 0.17-0.92). Number of risk factors not associated with LRR (P=.54).

EORTC 22922 and MA.20 showed DFS benefit from regional nodal RT in populations that included high-risk N0 disease; per the source, N0 made up 10% of MA.20 and 44% of EORTC. The regional recurrence here without RT (2.6%) sits below the EORTC 22922 control arm (4.2%). SUPREMO, which includes high-risk N0 pts, is the randomized test of this question.

Only 32 pts had 4-5 features (Fig 1), so the null risk-count finding is underpowered where it matters most. Surgery was done by high-volume breast surgeons, and axilla-only failures after SLNB suggest a false-negative contribution not addressed by PMRT decisions alone.

Authors attribute the low LRR to contemporary systemic therapy and argue conventional features do not by themselves mandate PMRT. The data establish baseline risk without RT, not the size of any PMRT benefit, so the question for T2 tumors carrying several features stays open.

Single-institution retrospective series with no PMRT comparator; describes baseline LRR risk, cannot estimate PMRT benefit. Small ≥4-feature subgroup limits the null risk-factor finding.

Retrospective translational cohort within phase 3 SWOG S8814, which enrolled 1989-1995. RT use was not randomized: it was expected after BCS and left to physician discretion after mastectomy. Reported under REMARK guidelines.

Postmenopausal, ER/PR+, node-positive women from the tamoxifen-only and CAF-T arms who had RS assayed. Final N=316: BCS + RT or mastectomy without RT. Pts who received PMRT were excluded, so the mastectomy cohort is the group whose physicians judged RT unnecessary.

Tamoxifen 20 mg daily for 5 y, alone or after 6 cycles of CAF (CAF-T). The concurrent CAF + tamoxifen arm was excluded.

Primary: LRR (breast, chest wall, axillary, infraclavicular, supraclavicular or IMN), censored at distant recurrence or death. The primary analysis used continuous RS; the secondary analysis compared RS <18 vs ≥18.

The continuous-RS primary analysis was not significant (P=.11). The categorical split and the subset tables carry the signal.

Direction agrees with NSABP B-28, where RS also independently predicted LRR. Absolute LRR at 1-3 nodes was higher here than in B-28, which treated every pt with chemo and allowed no PMRT. MA.20 and EORTC 22922 show a nodal RT DFS benefit even where LRR is low, so LRR alone may underestimate what RT buys.

Treatment era: CAF and tamoxifen only, with no taxanes or AIs, both of which may lower baseline LRR. LRR was censored at distant recurrence, which can undercount local failures. The ≥4-node mastectomy subset had no n reported.

RS looks prognostic for LRR beyond nodal count, but it cannot yet select pN1 pts for PMRT omission. That question belongs to TailorRT (MA.39), which randomizes only low-RS pN1 pts. Nodal burden still dominates the risk at ≥4 nodes.

Retrospective correlative of a chemo trial with nonrandomized RT. Prespecified continuous-RS analysis null (P=.11); the cut at 18 was confirmed post hoc; nodal subsets exploratory with few events.

Retrospective population-based cohort from the BCCA Breast Cancer Outcomes Unit database. N=821 of 1117 identified, after excluding 296 who received RT. Cox multivariable modeling, then a classification model built from the significant factors. Median follow-up 7.7y.

pT1-T2, 1-3 positive axillary nodes, diagnosed 1989-1997, all treated with modified radical mastectomy and axillary dissection. Excluded pT3-T4, ≥4 N+, M1 and unknown pTN. Median age 62.

Adjuvant systemic therapy in 94% per BCCA guidelines: chemo alone 21.7%, hormonal alone 57.1%, both 15.1%. Chemo was AC, CMF or FAC; hormonal therapy was mostly tamoxifen.

No pt received locoregional RT; the study defines baseline no-PMRT risk to inform selection, not RT efficacy. Authors anchor benefit estimates on an RT relative LRR reduction of approximately two-thirds.

Primary: isolated LRR (chest wall, axillary, infraclavicular, IMN as first site) and LRR±simultaneous distant recurrence. Secondary: BCSS and OS.

Overall 10y LRR±SDR 15.9%. Each of age <45, T2, grade 3, ER-negative, medial location, >1 N+ or >25% N+ alone conferred >20% 10y LRR on univariate analysis. Age was the first model split (29.3% vs 13.7%, p<0.0001).

No-PMRT LRR here (13-16% at 10y) sits well below the ~30% in DBCG 82b/82c and nearer the BCCA randomized trial (16% at 10y) and systemic-therapy trial cohorts (12-20%). Authors attribute the gap to axillary staging, systemic therapy and selection differences.

Excluding the 296 PMRT recipients likely removed higher-risk pts, biasing baseline risk downward. The >25% nodes positive ratio depends on ALND node yield, and its ILRR HR CI (0.99-4.05) touches 1. The ER status row in Table 1 shows a transposition in the source.

Authors propose <10% 10y LRR as a low-risk group that may be spared PMRT, >30% as justifying it, and 15-20% as the threshold for discussion. The model supports considering PMRT in all women <45 and recommending it when >25% of nodes are positive.

Retrospective single-province no-RT cohort; excluded PMRT recipients; data-driven breakpoints, no validation set; pre-taxane, ALND-era systemic therapy limits transfer to modern pts.

Retrospective analysis of two MD Anderson cohorts used to build (932 pts, 1997 to 2003) and validate (804 pts, 2004 to 2005) CPS+EG; 39 excluded, N=1697. Median f/u 51 mo.

All pts received neoadjuvant anthracycline and/or taxane chemotherapy; median age 50. Neoadjuvant trastuzumab and preoperative RT excluded; 20 pts missing ER or grade also excluded.

Local therapy: BCT 656 (39%), mastectomy alone 297 (17%), mastectomy + PMRT 744 (44%). Institutional PMRT indications were clinical stage III, ypN+, or select cT1/T2N1 with high-risk features. Dose, fractionation and fields not reported in source.

LRR = invasive cancer in ipsilateral breast, chest wall or regional nodes, counted regardless of distant timing. Objectives: CPS+EG vs stage for LRR stratification, and whether a score threshold could identify pts safely omitting PMRT.

5-yr freedom from LRR 71% to 99% by CPS+EG vs 86% to 97% by clinical or pathologic stage. In the PMRT-indicated cohort with CPS+EG ≤2, local therapy type made no difference (P=0.73).

NSABP B-18/B-27 (cited) reported 10-yr LRR of 10.3% with BCT and 12.3% with mastectomy but gave no PMRT, so cannot address it. The MD Anderson Prognostic Index omitted ER and grade. NSABP B-51 later randomized regional nodal RT in cN+ → ypN0, a different population from the ypN+ pts here.

Prose direction for the CPS+EG ≥3 subgroup (mastectomy alone favored) contradicts the MVA; overall LRR is reported as both 6.5% and 6.1%. Median f/u of 51 mo is short for ER-positive disease, the biology driving the favorable scores.

The stratification claim is sound and adds biology to anatomic staging for LRR. The omission claim is a null comparison in a small, selected subgroup with no equivalence margin, so it frames a trial question rather than settling one.

Retrospective single-institution cohort; the PMRT-omission claim rests on a non-randomized subgroup with confounding by indication and only 72 mastectomy-alone pts.