Martin Sjöström

Background: This study evaluates DL using GEX and preoperatively available clinical data (PreopClinic) to predict SLNM, and explores their potential for guiding axillary surgery and prognostic assessment. Methods: We retrospectively included 6,836 clinically node-negative T1-T2 patients with invasive breast cancer who underwent primary surgery from the SCAN-B. Three DL models—a multilayer perceptron, a pathway-informed sparse neural network, and a transformer—were developed using the development set (n=4,625) and evaluated against XGBoost in the independent test set (n=2,211). Results: The Transformer outperformed other methods for GEX modeling and minimized prior gene selection. In the independent test set, the combined Pre-opClinic+GEX model significantly improved SLNM prediction compared to Pre-opClinic alone (ROC AUC 0.693 vs 0.596, P < 0.001) and identified low-risk patients who might avoid unnecessary SLNB (reduction rate 27.2% at 92.1% sen-sitivity). However, the combined model did not significantly outperform GEX alone. While GEX provided the dominant predictive signal, PreopClinic contributed complementary information with modest numerical gains in clinical utility. Across-subtype training outperformed within-subtype training, particularly in TNBC, where the combined model achieved AUC 0.734 (95% CI: 0.644-0.837). The derived SLNM predictor also provided prognostic information beyond the estab-lished prognostic factors. Although the models were developed primarily using surgical specimen–derived GEX, paired biopsy and surgical-specimen analyses (n=116) demonstrated substantial concordance of transcriptomic patterns and nodal predictions. Conclusion: These findings highlight the Transformer’s robustness against noise and effectiveness in capturing informative transcriptomic features for SLNM pre-diction. The agreement observed between paired biopsy and surgical specimens supports the feasibility of future biopsy-based preoperative applications.

Purpose:PrOstate cancer Radiation Therapy Outcomes Score (PORTOS) is a 24-gene radiation response signature developed and validated in multiple randomized trials in prostate cancer. Given its basis in general radiation and DNA damage response pathways, we sought to evaluate whether PORTOS could also predict radiotherapy (RT) benefit in breast cancer. Experimental Design:PORTOS scores were calculated from gene expression profiles of 765 tumor samples from SweBCG91RT, a randomized trial of adjuvant whole-breast RT versus observation following breast-conserving surgery in node-negative stages I to IIA breast cancer. The primary endpoint was 10-year cumulative incidence (CI) of any recurrence, which closely mirrored the prostate cancer endpoints PORTOS was validated on. Patient-reported breast pain and tumor-infiltrating leukocytes (TIL) were also analyzed. Results:A significant interaction (P = 0.049) between PORTOS and RT was observed for 10-year CI of any recurrence. Patients in the top 75% of PORTOS scores derived substantial benefit from RT [subdistribution hazard ratio (SHR) = 0.46, P < 0.001] compared with those in the bottom 25% (SHR = 0.73, P = 0.24). Higher PORTOS scores were also associated with increased risk of RT-related breast pain, though the interaction P value was not significant (interaction P = 0.07). PORTOS was weakly anticorrelated with tumor size, histologic grade, and TILs and provided orthogonal information to previously validated signatures ARTIC and POLAR. Conclusions:PORTOS predicts benefit from RT for any recurrence in early-stage breast cancer, mirroring its performance in prostate cancer. It is the first radiation response biomarker to demonstrate predictive value for both efficacy and toxicity across randomized trials in multiple tumor types.

Purpose: PrOstate cancer Radiation Therapy Outcomes Score (PORTOS) is a 24-gene radiation response signature developed and validated in multiple randomized trials in prostate cancer. Given its basis in general radiation and DNA damage response pathways, we sought to evaluate whether PORTOS could also predict radiotherapy (RT) benefit in breast cancer. Experimental Design: PORTOS scores were calculated from gene expression profiles of 765 tumor samples from SweBCG91RT, a randomized trial of adjuvant whole-breast RT versus observation following breast-conserving surgery in node-negative stages I to IIA breast cancer. The primary endpoint was 10-year cumulative incidence (CI) of any recurrence, which closely mirrored the prostate cancer endpoints PORTOS was validated on. Patient-reported breast pain and tumor-infiltrating leukocytes (TIL) were also analyzed. Results: A significant interaction (P = 0.049) between PORTOS and RT was observed for 10-year CI of any recurrence. Patients in the top 75% of PORTOS scores derived substantial benefit from RT [subdistribution hazard ratio (SHR) = 0.46, P < 0.001] compared with those in the bottom 25% (SHR = 0.73, P = 0.24). Higher PORTOS scores were also associated with increased risk of RT-related breast pain, though the interaction P value was not significant (interaction P = 0.07). PORTOS was weakly anticorrelated with tumor size, histologic grade, and TILs and provided orthogonal information to previously validated signatures ARTIC and POLAR. Conclusions: PORTOS predicts benefit from RT for any recurrence in early-stage breast cancer, mirroring its performance in prostate cancer. It is the first radiation response biomarker to demonstrate predictive value for both efficacy and toxicity across randomized trials in multiple tumor types.

Supplementary Figure 1 shows sgRNA validation and MHCI protein expression on tumor cell lines.

Table S1. Clinical characteristics of the 177LuPSMA-treated cohort

Supplemental Table 1: Median, Inter-Quartile Range (IQR), and 25th and 75th percentiles for Figure 2. Supplemental Table 2: Median, Inter-Quartile Range (IQR), and 25th and 75th percentiles for Figure 3. Supplemental Table 3: Proportions for Figure 4. Supplemental Table 4: Proportions for Figure 5.

A list of individuals in the SU2C/PCF West Coast Prostate Cancer Dream Team

Figure S10. Sample cDNA concentration and tumor fraction by CTC phenotype in the survival analysis subset.

Figure S1. Calibration of ESTIMATE tumor content prediction.