Neoadjuvant immunotherapy continues to rewrite our understanding of anticancer immunity

Neoadjuvant immunotherapy studies have rapidly evolved from small investigator-led window-of-opportunity studies to standard of care paradigms across various tumour types. By treating patients before their surgery, we gain not only an opportunity to improve cure rates but also access to living tumour ecosystems responding – or not – to immunotherapy. Neoadjuvant clinical trials are now continuing to unravel principles that may guide the next generation of immunotherapy-based regimens.

Triple-negative breast cancer (TNBC) offers a compelling example. Although neoadjuvant chemo-immunotherapy (pembrolizumab, in combination with carboplatin and paclitaxel, followed by an anthracycline-based therapy) yields pathological complete responses in many patients, relapse risk remains high for those with residual disease (Clin Cancer Res. 2025;31:3916–3921). Study data presented at the ESMO Immuno-Oncology Congress 2025 (London, 10–12 December) showed that looking beyond traditional residual cancer burden (RCB) scores provides a better understanding of treatment efficacy (Abstract 328MO). Using single-cell spatial transcriptomics and proteomic profiling of post-therapy specimens, researchers identified a population of lipid-associated macrophages whose abundance strongly correlates with relapse risk. Low levels of these macrophages were consistently associated with higher RCB scores, while their protein signatures were detectable systemically as blood biomarkers. Therefore, an ‘immune-informed RCB’ may refine the prognostic value of conventional pathology.

Patients with mismatch repair-deficient (MMRd) cancer often display remarkable responses to immunotherapy. These effects are often attributed to reinvigoration of exhausted T cells. A phase I neoadjuvant trial of patients with MMRd endometrial cancer treated with immune checkpoint inhibitors revealed that B cells may play a critical role (Abstract 329MO). By using single-cell RNA sequencing and spatial transcriptomics, it was found that a highly organised, antibody-producing B-cell compartment, often embedded within tertiary lymphoid structures, displays dynamic changes following treatment. These findings raise new possibilities for harnessing B-cell biology as both biomarker and therapeutic target.

Finally, in patients with early-stage (IA2–IIIA) non-small cell lung cancer (n=33), the INNWOP01 study showed that two cycles of pembrolizumab plus lenvatinib induced a major pathological response rate of 33% (LBA1). Multi-omics data deconvolution revealed a pattern of tumour microenvironment remodelling in responders, characterised by expansion of cytotoxic and exhausted T cells, distinct shifts in macrophage and neutrophil subsets, and metabolic reprogramming of malignant epithelial cells.

Taken together, these studies reaffirm that by integrating spatial profiling, single-cell technologies and blood-based correlates, we can dissect why immunotherapy works in some tumours but not others. These translational research studies are critical for identifying immune states potentially predictive of cure, and for guiding a more rational design of combination strategies in different settings.