Residual Tumor Bed Biology Is Now the Real Readout in Resectable NSCLC

Neoadjuvant chemoimmunotherapy has changed what surgeons and pathologists pull apart after resection in non-small-cell lung cancer. The specimen is no longer just a record of how much tumor remains. It is a map of how malignant cells, immune infiltrates, fibroblasts, extracellular matrix, and vascular repair have been rearranged by treatment pressure.

Residual viable tumor percentage still matters. It remains the most established measure of treatment effect in the resection bed, and it anchors pathological complete response and major pathological response. Yet two tumors can share the same residual viable fraction and look biologically unrelated once their architecture is examined. One may contain tumor-contacting effector T cells, mature tertiary lymphoid structures, and antigen-presenting myeloid cells. Another may trap immune cells behind a fibroblast-rich barrier, leaving residual malignant nests relatively insulated from immune attack.

The review argues that this spatial problem is the next biomarker frontier. It also makes a hard distinction between post-treatment evidence and the much larger body of work on untreated tumors or pretreatment predictors. That separation is essential, because a state that predicts response before therapy begins is not the same as a state that persists in the resection bed after therapy has already worked, failed, or partially worked.

What the Post-Treatment Resection Bed Reveals

Several recurring patterns emerged across the reviewed studies. Mature tertiary lymphoid structures appeared repeatedly, along with coordinated B-cell and T-cell programs, selected antigen-presenting myeloid states, and evidence that T cells can physically reach tumor after combined checkpoint blockade and chemotherapy. These signals suggest that effective treatment does more than shrink tumor volume; it can reorganize the tissue into immune-active neighborhoods that resemble localized sites of antigen presentation and adaptive immune coordination.

One sentence captures the central challenge: cell abundance alone is not enough.

A high count of immune cells does not guarantee that those cells are functionally engaged, spatially positioned to contact residual tumor, or organized into structures that support productive antitumor immunity. The review repeatedly shows why this matters. TLS density was not consistently associated with pathological complete response across all datasets, and several proposed spatial metrics have not been fixed in advance or validated independently. In other words, a promising image can still fail as a biomarker if its definition shifts from cohort to cohort.

Fibroblasts are especially important in this context. Interferon-responsive fibroblast programs, extracellular-matrix barriers, and context-dependent myeloid states were all flagged as plausible resistance mechanisms. Malignant-cell plasticity adds another layer of complexity, because the tumor is not a static target; it can alter phenotype in response to the inflammatory and cytotoxic environment created by treatment. A residual tumor bed can therefore contain both evidence of immune activation and the machinery that blunts it.

Why Spatial Biomarkers Still Need to Earn Their Place

The strongest post-treatment datasets in this review are still small, selective, or assay-limited. Some spatial analyses relied on only a handful of treated specimens. Others used selected regions, retrospective thresholds, or mixed assay denominators that do not translate cleanly into patient-level clinical practice. Even where a spatial feature tracks with response or survival, the question remains whether it adds information beyond the basics: residual viable tumor, nodal status, and circulating tumor DNA.

A particularly important practical issue is the structural absence of certain metrics in pathological complete response. Tumor-centered distance measures and contact-based variables cannot be applied where no viable tumor remains in the sampled compartment. That is not a nuisance detail. It is a biological and statistical constraint that changes how endpoint definitions must be built. Any biomarker framework that ignores this problem risks inflating apparent signal in non-pCR cases while producing undefined or non-comparable values in the very patients whose complete eradication is most clinically meaningful.

The review proposes a pathology-anchored solution. Multiregion sampling should be anchored to the resection bed itself, with patient-level spatial analysis rather than region-by-region inflation. Assays should be simplified in stages, so discovery platforms can identify candidate structures while smaller, more practical panels are tested for reproducibility. Definitions must be prespecified and fixed before validation, especially for TLS density, maturity, and neighborhood metrics that are not interchangeable with one another.

BAG3-Positive Fibroblasts, TLS Maturation, and the Question of Persistence

Pretreatment BAG3-positive fibroblast neighborhoods are discussed as a related hypothesis, but their persistence after treatment remains unproven. This is a useful reminder that treatment can erase, remodel, or amplify a pre-existing stromal pattern. A pretreatment spatial signature is not automatically a post-treatment biomarker.

The same caution applies to TLS biology. Mature TLSs and coordinated B-cell/T-cell programs recur across several studies, and some cohorts link TLS features to response or disease-free survival. Yet the evidence is not uniform. In one multi-platform dataset, B-cell states tracked pathological complete response while TLS density itself was not significant. Elsewhere, TLS-positive cohorts were selected in ways that limit generalizability. The signal is real enough to warrant continued study, but not yet stable enough to guide therapy.

This is where the review is most useful clinically. It does not equate biological plausibility with clinical readiness. It places spatial immunology inside the harder framework of residual disease, where pathology already tells us how much tumor is left and ctDNA may offer a complementary systemic signal. Spatial biomarkers will only matter if they clarify why two patients with similar residual burden have different outcomes, or if they identify a subgroup whose resection bed still contains actionable immune suppression after neoadjuvant therapy.

The review closes on a restrained but important point: the incremental value of these spatial immune-stromal markers beyond pathological burden and circulating tumor DNA still has to be proven in external validation before biomarker-guided treatment can be tested.


Reporting by Dr. Revathy Mohan V.R., MBBS for Medicine World Foundation, with additional data provided by Frontiers in Immunology.
PRIMARY CLINICAL EVIDENCE & REGISTRY DATA