Role of derived neutrophil lymphocyte ratio as a biomarker in advanced colorectal cancer

• Published in: The Lancet (British edition) Vol. 383; p. S96
• Main Authors: Seligmann, Jenny, Dr, Hall, Peter, PhD, Wilson, Harriet, Richman, Susan, PhD, Barrett, Jenny, Prof, Seymour, Matthew, Prof, Quirke, Philip, Prof
• Format: Journal Article
• Language: English
• Published: London Elsevier Ltd 01.02.2014; Elsevier Limited
• Subjects: Colorectal carcinoma; Internal Medicine; Lymphocytes; Multivariate analysis
• ISSN: 0140-6736; 1474-547X
• DOI: 10.1016/S0140-6736(14)60359-7

Abstract Background The discovery and validation of prognostic and predictive biomarkers in advanced colorectal cancer is crucial for the personalisation of therapy. Inflammation-based prognostic scores have been developed, including the neutrophil lymphocyte ratio (NLR), and the derived NLR (dNLR) calculated using a differential white cell count. The NLR has been assessed in several cancers; in a retrospective series of advanced colorectal cancer, higher NLR was associated with reduced survival and poorer outcomes with chemotherapy. We aimed to validate the usefulness of the dNLR as a prognostic and a predictive biomarker to anti-epidermal growth factor receptor (EGFR) agents in advanced colorectal cancer. Methods We studied patients from a phase 3, open-label, randomised controlled trial that assessed the addition of panitumumab to irinotecan in pretreated advanced colorectal cancer (PICCOLO trial). The prognostic analysis included all patients for whom dNLR data were available; a subset with KRAS wild-type tumours randomised to standard irinotecan with or without panitumumab was also analysed for predictive value. The dNLR was calculated from the prerandomisation blood count and categorised as high (≥2) or low (<2). Prognostic and predictive values of the dNLR were assessed with Cox's proportional hazard and logistic regression, adjusting for known prognostic factors in advanced colorectal cancer. Findings 1159 patients were included in the prognostic analysis, and 444 in the predictive analysis. In multivariate analysis, high dNLR was independently associated with reduced overall survival compared with low dNLR (hazard ratio 1·59, 95% CI 1·32–1·78; p<0·001), the only other significant factor being poor performance status (p<0·001). High dNLR was also associated with reduced progression free survival (1·29, 1·13–1·47, p<0·001). In the predictive analysis, dNLR was not predictive of overall benefit with the addition of panitumumab to irinotecan. The main trial results showed that the benefit of panitumumab is confined to patients with no mutations in EGFR downstream pathways ( KRAS, BRAF, NRAS , and PIK3CA ), and in the subset of 323 patients in this all-wild-type category panitumumab had a more marked effect on tumour response rate in patients with low dNLR (odds ratio 7·4, p<0·001) than in those with high dNLR (3·7, p<0·001); however, interaction testing was not significant. Interpretation The dNLR seems to offer the clinician useful prognostic information. Since this information can easily be calculated from routine clinical data, its integration into routine patient assessment should be considered after further validation. We did not demonstrate significant predictive value for anti-EGFR therapy, but our finding of higher response with panitumumab in patients with a low dNLR is hypothesis generating. Further investigation in datasets from randomised controlled trials and exploratory translational work are warranted. Funding Cancer Research UK, Yorkshire Cancer Research.