BackgroundPrognostic information is essential for decision-making in breast cancer management. In recent years, trials and clinical practice have emphasized genomic prognostication tools, despite clinicopathological methods being more affordable and accessible. PREDICT v3 is one such tool with promising results across cohorts. Advances in machine learning (ML), transfer learning, and ensemble methods provide opportunities to enhance these approaches, especially where missing data and model assumptions differ across diverse populations. ObjectiveThis study evaluates the potential to improve survival prognostication in breast cancer. More precisely, we compare de novo ML, transfer learning from the pretrained prognostication model PREDICT v3, and a stacked ensemble approach. MethodsData from the MA.27 trial (NCT00066573) were used for model training, with external validation on data from the Tamoxifen Exemestane Adjuvant Multinational trial (NCT00279448 and NCT00032136) and a US Surveillance, Epidemiology, and End Results cohort. Transfer learning was applied by re-estimating the parameters of the pretrained prognostic tool PREDICT v3. De novo ML included random survival forests and extreme gradient boosting, and the ensemble was implemented using weighted linear stacking of model predictions. Internal and external validation was assessed in terms of the integrated calibration index and discrimination. Shapley Additive Explanations values were used to explain model predictions and decision-curve analysis to facilitate the interpretation of performance differences. ResultsTransfer learning, de novo random survival forest, and the stacked ensemble improved calibration in MA.27 over the pretrained model (integrated calibration index reduced from 0.042 in PREDICT v3 to ≤0.007) while discrimination remained comparable (AUROC increased from 0.738 in PREDICT v3 to 0.744-0.799). In decision-curve analysis, these approaches demonstrated consistently positive net benefit across clinically relevant thresholds, while PREDICT v3 lost net benefit beyond 7.5% predicted risk. Invalid PREDICT v3 predictions were observed in 23.8% to 25.8% of MA.27 individuals due to missing information. In contrast, ML models and the stacked ensemble predicted survival despite missing data. Across all models, patient age, nodal status, pathological grading, and tumor size had the highest Shapley Additive Explanations values, indicating their importance for survival prognostication. External validation in the US Surveillance, Epidemiology, and End Results cohort confirmed the benefits of transfer learning, RSF, and ensemble in terms of calibration while maintaining discrimination at comparable levels. In contrast, generalizability was limited in the Tamoxifen Exemestane Adjuvant Multinational trial, a cohort with a substantially different distribution of clinicopathological characteristics. ConclusionsThis study demonstrates that transfer learning, de novo RSF, and a stacked ensemble can improve prognostication compared with the pretrained PREDICT v3, particularly in the presence of missing or uncertain inputs. Transportability may be limited in cohorts with different clinicopathological profiles, requiring local validation before clinical deployment. Ultimately, better survival estimation can provide more meaningful guidance in breast cancer care. Trial RegistrationClinicalTrials.gov NCT00066573; https://clinicaltrials.gov/study/NCT00066573, NCT00279448; https://clinicaltrials.gov/study/NCT00279448, NCT00032136; https://clinicaltrials.gov/study/NCT00032136

PURPOSE:The aim of PREDICT was to confirm clinical validity and the potential for clinical utility of serial circulating tumor cell (CTC) enumeration in patients with metastatic breast cancer, focusing on its prognostic value in different breast cancer subtypes and clinical settings. EXPERIMENTAL DESIGN:In total, 4,436 individual patient-level data with CTC results from both baseline and one follow-up (CellSearch; Menarini Silicon Biosystems) were analyzed to evaluate the association between CTC detection and overall survival (OS) in the full patient cohort and separately for tumor and treatment types. RESULTS:Using the cutoff ≥1 CTC for CTC positivity, 913 (20.6%) patients had 0 CTCs at both time points (neg/neg) and 325 (7.3%) and 1,189 (26.8%) patients converted from CTC negative to CTC positive (neg/pos) or vice versa (pos/neg), whereas 2,009 (45.3%) patients had at least one CTC at both time points (pos/pos). The median OS for the neg/neg, neg/pos, pos/neg, and pos/pos group was 45.6, 26.1, 32.3, and 17.3 months, respectively (P < 0.0001, global log-rank test). CTC responders (pos/neg) showed a lower risk of death compared with CTC nonresponders (pos/pos; HR, 0.48; 95% confidence interval, 0.44-0.53). Similar results were obtained in subgroup analyses according to hormone receptor and HER2 subtype, treatment type, and with a ≥5 CTC cutoff for CTC positivity. CONCLUSIONS:Follow-up CTC assessments strongly predict OS independently from tumor subtype and treatment. New randomized trials to define the clinical utility of CTC monitoring for risk stratification and as an early response marker in metastatic breast cancer are urgently needed.

Prognostic information is essential for decision-making in breast cancer management. Recently trials have predominantly focused on genomic prognostication tools, even though clinicopathological prognostication is less costly and more widely accessible. Machine learning (ML), transfer learning and ensemble integration offer opportunities to build robust prognostication frameworks. We evaluate this potential to improve survival prognostication in breast cancer by comparing de-novo ML, transfer learning from a pre-trained prognostic tool and ensemble integration. Data from the MA.27 trial was used for model training, with external validation on the TEAM trial and a SEER cohort. Transfer learning was applied by fine-tuning the pre-trained prognostic tool PREDICT v3, de-novo ML included Random Survival Forests and Extreme Gradient Boosting, and ensemble integration was realized through a weighted sum of model predictions. Transfer learning, de-novo RSF, and ensemble integration improved calibration in MA.27 over the pre-trained model (ICI reduced from 0.042 in PREDICT v3 to <=0.007) while discrimination remained comparable (AUC increased from 0.738 in PREDICT v3 to 0.744-0.799). Invalid PREDICT v3 predictions were observed in 23.8-25.8

Effective targeting of somatic cancer mutations to enhance the efficacy of cancer immunotherapy requires an individualized approach. Autogene cevumeran is a uridine messenger RNA lipoplex-based individualized neoantigen-specific immunotherapy designed from tumor-specific somatic mutation data obtained from tumor tissue of each individual patient to stimulate T cell responses against up to 20 neoantigens. This ongoing phase 1 study evaluated autogene cevumeran as monotherapy (n = 30) and in combination with atezolizumab (n = 183) in pretreated patients with advanced solid tumors. The primary objective was safety and tolerability; exploratory objectives included evaluation of pharmacokinetics, pharmacodynamics, preliminary antitumor activity and immunogenicity. Non-prespecified interim analysis showed that autogene cevumeran was well tolerated and elicited poly-epitopic neoantigen-specific responses, encompassing CD4+ and/or CD8+ T cells, in 71 NCT03289962 . In this phase 1 trial, patients with locally advanced or metastatic solid tumors were treated with the individualized mRNA neoantigen-specific immunotherapy (iNeST) autogene cevumeran alone or in combination with the anti-PD-L1 agent atezolizumab, showing long-lasting neoantigen-specific immune responses and preliminary clinical activity, supporting further development of this therapeutic approach.

The Breast Cancer Index (BCI) was previously shown to identify ~20% of postmenopausal patients with early stage, hormone receptor positive (HR+), node negative (N0) breast cancer with minimal (<5%) risk of 10-year distant recurrence (DR) even without receiving adjuvant endocrine therapy (ET). This prospective-retrospective study further validated the BCI minimal risk classification in postmenopausal patients with early-stage, HR + HER2– N0 breast cancer from the Netherlands Cancer Registry (NCR) and the Tamoxifen and Exemestane Adjuvant Multinational (TEAM, NCT00279448, NCT00032136) randomized trial who received 5 years of primary adjuvant ET. BCI classified approximately 15% of patients as minimal risk. In the NCR cohort (n = 1264 out of 15,053 HR+ patients in the registry), risks of DR in the minimal, low, intermediate, and high groups were 4.8%, 3.3%, 8.0%, and 12.4%, respectively (P < 0.001). In the TEAM cohort (n = 978 out of 3544 in the BCI study), DR risks were 3.8%, 8.3%, 12.6% and 22.7% (P < 0.001). In multivariate analyses, BCI risk scores provided independent information over standard prognostic factors (P < 0.001). This study confirmed the ability of the adjusted BCI model to identify postmenopausal women with HR + HER2– N0 breast cancer who are at minimal risk of DR and may consider de-escalating adjuvant ET.

Monitoring tumors and detecting biomarkers through blood analysis remains challenging. Large oncosomes (LOs) are tumor-derived extracellular vesicles that offer a promising avenue for understanding tumor biology and biomarker status, potentially surpassing circulating tumor cells (CTCs) in accuracy. Validated LO detection methods are required to unlock LOs analytical potential. We evaluated two LO detection methods using known CTC platforms, the FDA-approved CellSearch, and RareCyte to compare their performance. In addition, we assessed the potential of LOs for HER2 detection in metastatic breast cancer (MBC) patients. LO counts were analyzed in parallel blood samples from 96 MBC patients using either the CellSearch/ACCESS tdEV analysis package, or the RareCyte method with in-house LO analysis. Overall survival (OS) was defined as the time from baseline blood draw to death or last follow-up. HER2 scores from primary tumor biopsies were retrieved from patient charts. HER2 overexpression in LOs and CTCs was determined using the RarePlex HER2/ER CTC panel kit (n=9). Correlations between LO and CTC counts were assessed using Spearman’s Rho test, and comparability of LO counts between platforms was evaluated with the Wilcoxon Signed Ranks test. Cutoffs for favorable vs. unfavorable LO counts were determined using an online tool. Survival curves for these groups were compared using the log-rank test. Correlation of HER2+ analytes with tissue biopsy HER2 scores was also analyzed using Spearman’s Rho test. A strong positive correlation was observed between LO counts from the two platforms (r = 0.49, p < 0.001). LO counts also correlated with their respective CTC counts (CellSearch: r = 0.88, p < 0.001; RareCyte: r = 0.93, p = 0.001). Higher LO counts were associated with worse survival outcomes (CellSearch: r = 0.35, p < 0.001; RareCyte: r = 0.39, p = 0.006). Patients with unfavorable LO counts had significantly shorter survival compared to patients with favourable LO counts (CellSearch: HR = 3.5, p < 0.001; RareCyte: HR = 3.6, p = 0.02). RareCyte detected significantly more LOs than CellSearch (median LO count: 132 vs. 41, p = 0.01). HER2 signal was detected in 5/9 patients with CTCs and in 8/9 patients with LOs. Notably, the proportion of HER2+ LOs positively correlated with tissue biopsy HER2 score (r=0.799, p=0.01). In contrast, the proportion of HER2+ CTCs did not correlate with tissue biopsy. While RareCyte captured more LOs than CellSearch, LO counts were highly concordant across platforms, validating both methods. The association of LO counts with overall survival suggests that LOs reflect tumor progression, and may complement CTC-based monitoring. Furthermore, the correlation between LO HER2 status and tumor biopsy HER2 scores underscores the potential of LO analysis for biomarker monitoring in metastatic breast cancer. Eszter Papp, Pieter Mestdagh, Luc Y. Dirix, Peter Vermeulen, Mark Kockx. Comparable detection of large tumor-derived vesicles (oncosomes) by known CTC platforms: Potential new analyte for biomarker monitoring [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1968.

The antibody–drug conjugate trastuzumab deruxtecan has proven to be not only efficient in patients with HER2+ breast cancers (BC), but also in those patients with so-called HER2-low BC. HER2-low tumors are well described in the general BC population, but not in patients with invasive lobular carcinoma (ILC). Here, we aimed at analyzing the association of HER2-low with clinicopathological features and survival outcomes in patients with early-stage pure ILC. A multicentric retrospective cohort of patients diagnosed with stage I-III estrogen receptor positive (ER+) HER2 negative (HER2-) ILC between 01/01/2000 and 12/31/2020 was assembled. HER2- disease was categorized further by immunohistochemical (IHC) score into HER2 0, HER2 1+ and HER2 2+ following time appropriate ASCO/CAP guidelines from 2007 onward and by local guidelines prior to 2007. The association of HER2-low (HER2 1+ and 2+) with clinicopathological variables was assessed using multinomial logistic regression. Survival analyses were performed to evaluate the association of HER2-low with disease-free (DFS), distant recurrence-free (DRFS) and overall survival (OS). The data of 2098 patients with ER+ HER2- ILC was collected of which 1103 (52.6

PURPOSE:To test whether dose-escalated single fraction (SF) stereotactic body radiotherapy (SBRT) of 20 Gy to painful bone metastases is superior to conventional SF three-dimensional (3D) conformal radiotherapy (RT) to a standard dose of 8 Gy in achieving complete pain response (CR). METHODS:A single-blind, randomized, controlled, phase III trial (ROBOMET) included 126 patients with up to three painful bone metastases, randomly assigned between April 2019 and October 2022 at multiple centers in Belgium. Inclusion criteria were uncomplicated painful bone metastases (worst pain score ≥2 on a 0-10 pain scale) arising from a solid tumor. Treatment consisted of either a single SBRT fraction of 20 Gy or a single conventional RT fraction of 8 Gy. The primary end point was the proportion of patients with a CR 1 month after RT scored according to the International Consensus on Palliative Radiotherapy Endpoints analyzed as per an intention-to-treat principle. RESULTS:After 1 month, 16 of 63 (25% [95% CI, 15 to 38]) patients treated with conventional RT achieved CR versus 23 of 63 (37% [95% CI, 25 to 50]) treated with SBRT (P = .25). After 3 months, 15 of 63 (24% [95% CI, 14 to 36]) patients achieved CR after conventional RT versus 21 of 63 (33% [95% CI, 22 to 46]) after SBRT (P = .32). Among patients evaluable after 3 months and treated per protocol, the SBRT group had more complete responders (21/39, 54% [95% CI, 37 to 70]) than the conventional RT group (15/48, 31% [95% CI, 19 to 46]; P = .048). CONCLUSION:SBRT failed to demonstrate improved CR rates after 1 month.

PURPOSE:Few studies have compared the performance of gene-expression profiling tests (e.g. Oncotype-Dx) to clinico-pathologic risk calculators (e.g. PREDICT 2.1, INFLUENCE 2.0, and CTS5) or tools that combine both (e.g. RSClin) in patients with early breast cancer (EBC). A large trial dataset was used to evaluate the prognostic performance of different tests based on patient outcomes. METHODS:The TEAM pathology cohort accrued samples from 4736 postmenopausal hormone positive women with EBC, treated with either exemestane or tamoxifen followed by exemestane. Oncotype-Dx-trained risk scores were previously generated by gene-expression profiling. Patient data was used to calculate various recurrence scores. Analysis was restricted to the N0/N1 population and prognostic ability of selected risk tools was assessed using Cox regression analysis and Harrell's C-statistic. RESULTS:Results were available for 2065 patients. There was low correlation between PREDICT 2.1 (r = -0.12), INFLUENCE 2.0 (r = 0.20), CTS5 (r = 0.16) with Oncotype-Dx-trained results. In N0 patients, RSClin had improved prognostic ability (C-statistic = 0.66) on DMFS compared to PREDICT 2.1 (0.60), INFLUENCE 2.0 (0.57), CTS-5 (0.62), and Oncotype-Dx (0.63). CONCLUSION:Combining molecular and clinico-pathologic factors enhances prognostic information. However, the impact of this on actual patient management requires further prospective validation. The trial is registered with clinicaltrials.gov NCT00279448 and NCT00032136; with Netherlands Trial Register, number NTR 267; and the Ethics Commission Trial, number 27/2001.

INTRODUCTION:Inflammatory Breast Cancer (IBC) is an aggressive presentation of BC present in 1-5 % of all patients with BC. While obesity has been consistently associated with worse prognosis in patients with BC, it is understudied in patients with IBC. PATIENTS AND METHODS:We retrospectively evaluated the association of body mass index (BMI) at diagnosis with clinicopathological characteristics, pathological complete response (pCR) to chemotherapy and survival in a multicentric cohort of patients with IBC treated with pre-operative chemotherapy. RESULTS:Of the 542 patients, 6 were underweight (1.1 %, excluded in further analysis), 190 were normal-weight (35.1 %), 187 had overweight (34.5 %), and 159 had obesity (29.3 %). Of the 536 included patients, 463 had non-metastatic and 73 metastatic IBC at diagnosis. Higher BMI was associated with older age at diagnosis, increased stromal tumor infiltrating lymphocytes (sTIL), and particularly in the ER-/HER2+ subgroup, a greater likelihood of metastasis at diagnosis. Tumor emboli were less frequently detected in peritumoral samples of patients with obesity as compared to patients with normal weight. Among non-metastatic patients, those with obesity in the ER-/HER2+ and ER+/HER2-subgroups showed numerically, but not statistically significantly, lower rates of pCR following neoadjuvant chemotherapy than normal-weight patients (38.5 % vs 42.5 %, and 6.6 % vs 16.7 %, respectively). BMI was not associated with any survival endpoint. CONCLUSION:We observed a limited association of BMI with clinicopathological variables and pCR in patients with IBC without evidence of its relationship with survival outcomes. Future studies should investigate the biological impact of obesity on IBC and its tumor microenvironment.

Immunotherapeutic interventions leading to increased T-cell responses in cancer.

Supplementary Figure 4 shows an overview of tumor emboli assessment.

Supplementary Figure 7 shows subgroup analyses of the association of pCR with clinicopathological and treatment variables.

Association of pCR with DFS. A, Kaplan–Meier curves of DFS according to pCR. B, Forest plots showing the association of pCR and standard clinicopathologic and treatment variables with DFS quantified by Cox regression.

Supplementary Figure 9 shows potential non-linear association of RCB score with sTIL.

PK concentration to time profiles for 6-mg and 30-mg efti

Abstract Inflammatory breast cancer (IBC) is a rare (1%–5%), aggressive form of breast cancer, accounting for approximately 10% of breast cancer mortality. In the localized setting, standard of care is neoadjuvant chemotherapy (NACT) ± anti-HER2 therapy, followed by surgery. Here we investigated associations between clinicopathologic variables, stromal tumor-infiltrating lymphocytes (sTIL), and pathologic complete response (pCR), and the prognostic value of pCR. We included 494 localized patients with IBC treated with NACT from October 1996 to October 2021 in eight European hospitals. Standard clinicopathologic variables were collected and central pathologic review was performed, including sTIL. Associations were assessed using Firth logistic regression models. Cox regressions were used to evaluate the role of pCR and residual cancer burden (RCB) on disease-free survival (DFS), distant recurrence-free survival (DRFS), and overall survival (OS). Distribution according to receptor status was as follows: 26.4% estrogen receptor negative (ER−)/HER2−; 22.0% ER−/HER2+; 37.4% ER+/HER2−, and 14.1% ER+/HER2+. Overall pCR rate was 26.3%, being highest in the HER2+ groups (45.9% for ER−/HER2+ and 42.9% for ER+/HER2+). sTILs were low (median: 5.3%), being highest in the ER−/HER2− group (median: 10%). High tumor grade, ER negativity, HER2 positivity, higher sTILs, and taxane-based NACT were significantly associated with pCR. pCR was associated with improved DFS, DRFS, and OS in multivariable analyses. RCB score in patients not achieving pCR was independently associated with survival. In conclusion, sTILs were low in IBC, but were predictive of pCR. Both pCR and RCB have an independent prognostic role in IBC treated with NACT. Significance: IBC is a rare, but very aggressive type of breast cancer. The prognostic role of pCR after systemic therapy and the predictive value of sTILs for pCR are well established in the general breast cancer population; however, only limited information is available in IBC. We assembled the largest retrospective IBC series so far and demonstrated that sTIL is predictive of pCR. We emphasize that reaching pCR remains of utmost importance in IBC.

Supplemental Table 2. Distribution of patients as defined by BCI prognostic (BCI/BCIN+) and BCI predictive (BCI (H/I)).