Several accomplishments have been achieved in triple-negative breast cancer (TNBC) research over the last year. The phase III IMpassion130 trial comparing chemotherapy plus atezolizumab versus chemotherapy plus placebo brought breast cancer into the immunotherapy era. Nevertheless, despite encouraging results being obtained in this trial, many open questions remain. A positive overall survival outcome was achieved only in PD-L1+ TNBC patients, suggesting a need to enrich the patient population more likely to benefit from an immunotherapeutic approach. Moreover, it remains unknown whether single-agent immunotherapy might be a good option for some patients. In this context, the discovery and implementation of novel and appropriate biomarkers are required. Focusing on the early onset of TNBC, neoadjuvant trials could represent excellent in vivo platforms to test immunotherapy agents and their potential combinations, allowing the performance of translational studies for biomarker implementation and improved patient selection. The aim of our review is to present recent advances in TNBC treatment and to discuss open issues in order to better define potential future directions for immunotherapy in TNBC.

Triple-negative breast cancer (TNBC) is not a unique disease, encompassing multiple entities with marked histopathological, transcriptomic and genomic heterogeneity. Despite several efforts, transcriptomic and genomic classifications have remained merely theoretic and most of the patients are being treated with chemotherapy. Driver alterations in potentially targetable genes, including PIK3CA and AKT, have been identified across TNBC subtypes, prompting the implementation of biomarker-driven therapeutic approaches. However, biomarker-based treatments as well as immune checkpoint inhibitor-based immunotherapy have provided contrasting and limited results so far. Accordingly, a better characterization of the genomic and immune contexture underpinning TNBC, as well as the translation of the lessons learnt in the metastatic disease to the early setting would improve patients’ outcomes. The application of multi-omics technologies, biocomputational algorithms, assays for minimal residual disease monitoring and novel clinical trial designs are strongly warranted to pave the way toward personalized anticancer treatment for patients with TNBC.

Numerous pathology foundation models have been developed to extract clinically relevant information. There is currently limited literature independently evaluating these foundation models on external cohorts and clinically relevant tasks to uncover adjustments for future improvements. Here we benchmark 19 histopathology foundation models on 13 patient cohorts with 6,818 patients and 9,528 slides from lung, colorectal, gastric and breast cancers. The models were evaluated on weakly supervised tasks related to biomarkers, morphological properties and prognostic outcomes. We show that a vision-language foundation model, CONCH, yielded the highest overall performance when compared with vision-only foundation models, with Virchow2 as close second, although its superior performance was less pronounced in low-data scenarios and low-prevalence tasks. The experiments reveal that foundation models trained on distinct cohorts learn complementary features to predict the same label, and can be fused to outperform the current state of the art. An ensemble combining CONCH and Virchow2 predictions outperformed individual models in 55% of tasks, leveraging their complementary strengths in classification scenarios. Moreover, our findings suggest that data diversity outweighs data volume for foundation models.