The prognostic value of patient reported outcomes (PROs) and clinical/demographic variables in the CASPIAN study.

Abstract

8516 Background: PROs have been shown to predict outcomes in non-small cell lung cancer and, using the FACT-G PRO instrument, in extensive stage small cell lung cancer (ES-SCLC). This study explored the relationships between EORTC QLQ-C30 domains and overall survival (OS) and progression free survival (PFS) in the phase III ES-SCLC CASPIAN trial, with and without adjustments for treatment (tx) and baseline clinical and demographic variables. Methods: CASPIAN had 3 treatment arms: durvalumab plus platinum–etoposide; durvalumab plus tremelimumab plus platinum–etoposide; and platinum–etoposide alone. It included the EORTC QLQ-C30, a PRO instrument consisting of 5 functioning scales (physical, role, emotional, cognitive, social), a global health status/quality of life (QoL) scale, 3 symptom scales (fatigue, nausea/vomiting, pain), 5 single item symptom assessments (dyspnea, insomnia, appetite loss, constipation, diarrhea) and a single financial difficulties item. Higher functional and QoL scores indicate better health status/function (anticipated positive OS/PFS associations) while higher symptom and financial difficulty scores represent a greater burden (anticipated negative OS/PFS associations). Baseline PRO scores were pooled across the 3 CASPIAN arms to obtain the most PRO information and used in 3 Cox proportional hazard models for OS and PFS: (A) scores only, (B) scores with a tx category for each CASPIAN arm, and (C) scores with tx categories and other baseline covariates (sex, age [ < 65, ≥65], ECOG performance status [0, > = 1], smoker [Y/N], CNS metastasis [Y/N], race [Asian, non-Asian], stage [III, IV], region [Asia, Europe, North and South America]). Models B and C were stratified by the type of platinum received. Backwards selection found the set of variables most associated with OS and PFS. There was no correction for multiplicity and nominal P-values < 0.05 were used to assess importance. Results: The outcomes and models associated with more favorable baseline PRO scale/item scores were: (i) better OS and PFS across all 3 Cox models [physical, role, social, QoL, fatigue, pain]; (ii) better OS [emotional (models A,C), nausea/vomiting (model C), dyspnea (model C)], and (iii) better PFS [emotional (model C), financial difficulties (models A,B,C)]. Among the 90 estimated Cox hazard ratios, 82 had the anticipated directions for OS and PFS with 54/90 of the estimates having P < 0.05. In each Cox model, more favorable baseline diarrhea scores were associated with reduced OS. All 15 baseline PRO score domains for OS and 10/15 for PFS were important in the stepwise selection process. Conclusions: The results largely indicate that baseline PROs had prognostic value in the CASPIAN trial across different specifications, some including tx and baseline clinical and demographic values. A better understanding of the relationship between tx and PROs may aid tx decisions. Clinical trial information: NCT03043872 .