Tumour-infiltrating CD8+ lymphocytes and colorectal cancer recurrence by tumour and nodal stage

Abstract

BackgroundIntratumoural T-cell infiltrate intensity cortes wrelaith clinical outcome in stage II/III colorectal cancer (CRC). We aimed to determine whether this association varies across this heterogeneous group.MethodsWe performed a pooled analysis of 1804 CRCs from the QUASAR2 and VICTOR trials. Intratumoural CD8+ and CD3+ densities were quantified by immunohistochemistry in tissue microarray (TMA) cores, and their association with clinical outcome analysed by Cox regression. We validated our results using publicly available gene expression data in a pooled analysis of 1375 CRCs from seven independent series.ResultsIn QUASAR2, intratumoural CD8+ was a stronger predictor of CRC recurrence than CD3+ and showed similar discriminative ability to both markers in combination. Pooled multivariable analysis of both trials showed increasing CD8+ density was associated with reduced recurrence risk independent of confounders including DNA mismatch repair deficiency, POLE mutation and chromosomal instability (multivariable hazard ratio [HR] for each two-fold increase\u2009=\u20090.92, 95%CI\u2009=\u20090.87–0.97, P\u2009=\u20093.6\u2009×\u200910−3). This association was not uniform across risk strata defined by tumour and nodal stage: absent in low-risk (pT3,N0) cases (HR\u2009=\u20091.03, 95%CI\u2009=\u20090.87–1.21, P\u2009=\u20090.75), modest in intermediate-risk (pT4,N0 or pT1-3,N1-2) cases (HR\u2009=\u20090.92, 95%CI\u2009=\u20090.86–1.0, P\u2009=\u20090.046) and strong in high-risk (pT4,N1-2) cases (HR\u2009=\u20090.87, 95%CI\u2009=\u20090.79–0.97, P\u2009=\u20099.4\u2009×\u200910−3); PINTERACTION\u2009=\u20090.090. Analysis of tumour CD8A expression in the independent validation cohort revealed similar variation in prognostic value across risk strata (PINTERACTION\u2009=\u20090.048).ConclusionsThe prognostic value of intratumoural CD8+ cell infiltration in stage II/III CRC varies across tumour and nodal risk strata.