Abstract 5189
Background
The Royal Marsden Hospital score (RMHs) (albumin <35 g/L, lactate dehydrogenase [LDH]>upper limit of normal [ULN], and >two sites of metastases [met]) is a validated prognostic index for Ph1 pt selection. Recently, a lung immune prognostic index (LIPI) (derived neutrophil/(leukocytes minus neutrophils) ratio [dNLR]>3, and LDH>ULN) proved to be useful for identifying pts with different outcomes under ICI. We aimed to improve pt selection for ICI Ph1 trials by developing a composite VIO that included all clinical-laboratory (CL) variables linked with worse median Overall Survival (mOS).
Methods
Retrospective analysis of pts treated with ICI at VHIO Ph1 Unit from Jan’12 to Oct’17. VIO includes four CL factors previously described (albumin<35g/L, LDH>ULN, >two sites of met, dNLR>3) and a fifth variable (liver met) as per univariate Cox modeling. The following VIO clusters were defined based on Kaplan Meier OS estimates: low risk (0 and 1), intermediate risk (2 and 3) and high risk (4 and 5).
Results
In total, 174 out of 214 pts (81%) treated with ICI (antiPD1/PDL1 ICI in 93%, combination regimens in 53%) had complete CL data for modeling. Most common tumor types were melanoma (22%) and lung (14%). Overall, best response was PD 47%, SD 38%, PR 12%, CR 2% and mOS 9.8 (95% CI 7.3-12.7) months (m). Concordance index of OS models including LIPI, RMHs or VIO scores were 0.62, 0.66 and 0.69, respectively. Estimated mOS in low risk (40.2% of all pts), intermediate risk (50.3%) and high risk (9.2%) were 22.0 m (10.5.4-33.4), 6.7 m (4.1-9.3) and 3.8 m (2.5-5.1), respectively (log rank test, p < 0.001). PD as best response was higher in high risk VIO group (81%) as compared to intermediate (50%) and low risk (34%, Chi-square p = 0.002). 6m OS rates were 85% (77%-94%), 55% (44%-67%) and 30% (14%-65%) in low, intermediate and high risk VIO groups (log rank test, p < 0.001).
Conclusions
Our results suggest that the VIO is a better predictor of OS on ICI in Ph1 trials as compared to existing prognostic scores. The VIO is a helpful tool for identifying Ph1 candidates unlikely to benefit from ICI and with higher chances of death within 6 m of trial recruitment.
Clinical trial identification
Legal entity responsible for the study
VHIO.
Funding
Has not received any funding.
Editorial Acknowledgement
Disclosure
A. Oaknin: Advisory boards: Roche, AstraZeneca, PharmaMar, Clovis Oncology, Tesaro; Travel or accommodation support: Roche, AstraZeneca, PharmaMar. J. Tabernero: Advisory Boards: Bayer, Boehringer Ingelheim, Genentech/Roche, Lilly, MSD, Merck Serono, Merrimack, Novartis, Peptomyc, Roche, Sanofi, Symphogen, Taiho. All other authors have declared no conflicts of interest.