Prognosis assessment of central nervous system metastases from digestive cancers: discrepancies between validated scores

Considering the therapeutic options, from surgery to palliative radiotherapy, establishing prognosis is a major issue of decision-making in central nervous system metastasis (CNSM) management. Several prognostic scores have been established. Given the distinguished disease history in CNSM from digestive cancers as compared with breast or lung cancers, some scores were lately specifically designed using data from digestive cancers. This study aims to assess the accuracy of the prognostic scores available for brain metastases in digestive cancers and establish the robust prognostic factors. This retrospective, single-center study enrolled all patients with digestive cancer with CNSM. Each patient was assessed for the following prognostic scores: RPA (Recursive partitioning analysis), GPA (Graded Prognostic Assessment), GI-GPA (Graded Prognostic Assessment for Gastrointestinal Cancer), and Pietrantonio nomogram. The study included 222 patients, 156 with colorectal cancers and 66 with upper gastrointestinal cancer. The mean age was 63.3 years old; 64.0% were men, and 85.8% had a stage 3–4 tumor at initial diagnosis. Among them, 88.8% of patients had extra-cerebral metastases. CNSM were symptomatic in 95% of cases, leading to diagnosis, and in 50.4% of cases, there were multiple CNSM. ECOG-PS was 1–2 in 66.7% of patients. Median OS at brain disease diagnosis was 3.7 months (range: 0.0-60.4). There was no significant difference in OS according to the primary tumour site: 3.9 months (range: 0.1–60.3) for colorectal cancers and 2.9 months (range: 0.3–56.4) for upper gastrointestinal tumors ( p = 0.9). In multivariable analysis, performance status ( p < 0.0001) was an independent prognostic factor in both primary tumors. A discordance of more than three months between predicted OS and observed OS was found in 54.8% of patients with the GI-GPA score, 52% with the Pietrantonio nomogram, 42.7% with the GPA score, and 31.5% with the RPA. Central nervous system metastases in digestive cancers are of poor prognosis. ECOG-PS is an independent prognostic factor in that setting. Although non designed specifically for digestive cancers, RPA is a performant prognostic score in that situation. Not applicable.