Skip to content

Estimation of recurrence risk in patients with colon cancer relies primarily on the TNM classification, yet this system remains imperfect for guiding clinical decisions. Since the early twentieth century, tumor immune infiltration has nonetheless been suspected to have a favorable influence on prognosis. In the absence of a robust quantification method, however, this observation had never been incorporated into cancer classification systems. This study aimed to validate, on an international scale, the prognostic value of the consensus Immunoscore, an assay based on the enumeration of tumor-infiltrating T lymphocytes.

An international consortium bringing together fourteen centers across thirteen countries, under the auspices of the Society for Immunotherapy of Cancer, evaluated this assay in patients with TNM stage I to III colon cancer. From paraffin sections of the tumor and the invasive margin, the densities of total CD3+ T lymphocytes and CD8+ cytotoxic T lymphocytes were quantified by digital pathology. Each patient's Immunoscore was derived from the mean of four density percentiles, with patients randomly allocated to a training set, an internal validation set, and an external validation set. The primary endpoint was time to recurrence, analyzed using stratified multivariable Cox models. Statistical analysis was entrusted to an independent, external group.

Of 3,539 specimens processed, 2,681 were retained after quality control. The assay proved highly reproducible across observers and centers (r=0.97 for both the tumor and the invasive margin; p<0.0001). In the training set, patients with a high Immunoscore had the lowest five-year risk of recurrence (8%, versus 19% for an intermediate score and 32% for a low score; hazard ratio of 0.20 for high versus low; p<0.0001), a finding confirmed in both validation sets. This prognostic value proved independent of age, sex, T and N stage, microsatellite instability, and established prognostic factors. Among the 1,434 stage II patients, the difference remained significant. Of all clinical parameters, the Immunoscore made the greatest relative contribution to risk (47%), ahead of the TNM classification (28%). The authors emphasize that, unlike the subjective assessment of tumor-infiltrating lymphocytes, the Immunoscore provides an objective and reproducible quantification of defined T-cell subpopulations. They acknowledge as a limitation the heterogeneity of a population drawn from thirteen countries and treated in routine practice, an inherent feature of the choice of a nonrandomized study design. These findings support the incorporation of the consensus Immunoscore as a new component of a TNM-Immune classification of cancer.