Context Is Everything: Tumour-Type-Dependent Prioritisation of Tumour-Intrinsic, Microenvironmental and Host...

Predicting who benefits from immune-checkpoint blockade remains unsolved, and the search for a single universal biomarker has repeatedly failed: tumour mutational burden has no threshold that transfers across tumour types, PD-L1 expression is inconsistent, and the determinants of response lie on three interacting axes - tumour, microenvironment and host - studied in isolation. We argue that the predictive weight of each determinant is not fixed but depends on tumour type, and we integrate the axes into a determinant-weight matrix: an evidence-graded, falsifiable synthesis ranking, for each histology, which factor most plausibly governs baseline response. It is a hypothesis-generating framework for prioritising which determinants to measure, not a validated decision algorithm. Four regimes recur: mutation-led tumours such as melanoma, non-small-cell lung, mismatch-repair-deficient and urothelial cancers; microenvironment-led tumours such as renal-cell carcinoma, where mutational burden is paradoxically misleading, together with sarcoma and head-and-neck cancer; host-sensitive contexts in which the gut microbiome and concomitant medications become rate-limiting; and cold tumours in which no single axis suffices. These weightings follow from the axes converging on one effector pathway, explaining why single-axis biomarkers are unstable and why integrative scores, though largely retrospective, outperform them. Each cell records the weight, evidence tier, context, predictive-versus-prognostic and validation status of the determinant, with an explicit rule where axes disagree. The framework covers baseline determinants at the point where first-line treatment is chosen, and sets immune-related toxicity aside. Presented as a scaffold rather than a decision rule, it reframes biomarker selection from a universal panel to a tumour-type-calibrated sequence.