Immunotherapy has changed clinical practise. Patient enrolment in immunotherapy phase I trials is still crucial. Therefore, new tests have been developed (GRIM score and LIPI score), challenging the classical ones designed for targeted therapy such as Royal Marsden Score (RMS). Although they include different variables, here we perform a comparison, aimed at identifying if there are differences in terms of their predictive power of survival.
Variables from patients treated in our institution (2012-2019) were collected retrospectively and used to calculate these different scores: RMS (number of metastasis, LDH, albumin), GRIM score (LDH, Albumin, NLR (Neutrophil/Lymphocyte ratio)>6) and LIPI score (LDH, NLR>3). Cox regression model was performed for predicting OS (time between the day of the first dose and date of death or lost to follow up). C-Harrell index and its confidence interval were calculated for each score for measuring the power of prediction. Differences between c-Harrell indexes were calculated. A p value < 0.05 was considered statistically significant. All calculations were performed with STATA v16.0.
In terms of predicting overall survival we found that a higher punctuation in RMS (high vs low HR = 2.14 95% CI 1.47-3.10), in Grim score (high vs low HR = 2.14; 95% CI 1.51-3.02) and LIPI score (high vs low HR = 1.83; 95% CI 1.42-2.37) were all statistically significant. In terms of power of prediction we found that LIPI score presented a c-Harrel index slightly higher (0.63; 95% CI 0.59-0.68) than GRIM score (0.62; 95% CI 0.57-0.66) and RMS (0.62; 95% CI 0.59-0.66). Nevertheless, differences between c-index of each score were not statistically significant.
Selecting patients for participating in phase I immunotherapy trials is still critical. Although in this study we have seen a slightly higher predictive power for LIPI score in comparison to GRIM score or RMS, differences were not statistically significant. Nevertheless, and assuming the most parsimonious model, LIPI score seemed to be an easy and accurate model for predicting survival.
The authors.
Has not received any funding.
L. Resano, A. Sánchez: Full/Part-time employment: Roche. All other authors have declared no conflicts of interest.