469P - The clinical utility of a change in the Royal Marsden Hospital prognostic score (rmhps) in patients (pts) prior to phase I clinical trial therapy

Date 01 October 2012
Event ESMO Congress 2012
Session Poster presentation III
Topics Drug Development
Presenter Caroline Michie
Authors C.O. Michie1, M. Ong1, J. Mateo1, A.M. Young1, D. Olmos Hidalgo2, J.E. Ang1, L.R. Molife1, U. Banerji1, J.S. De Bono3, S.B. Kaye1
  • 1Drug Development Unit, Royal Marsden Hospital, SM2 5PT - Sutton/UK
  • 2Ddu/prostate Unit, Royal Marsden HospitalNHS Foundation Trust, GB-SM2 5PT - Sutton/UK
  • 3Royal Marsden Hospital, SM2 5PT - Sutton/UK

Abstract

Background

Previously we developed and validated the RMHPS to aid patient selection for phase I clinical trials1. However, its sensitivity in predicting short-term outcome is limited. Here we assessed whether, prior to trial entry, a worsening of RMHPS during screening improves specificity.

Methods

Pts treated in phase I trials between Sep 09-11 at RMH Drug Development Unit were identified from a prospectively-collected database. RMHPS was calculated at 2 distinct timepoints; ≤1 wk prior to start of therapy and at an earlier timepoint ≤ 8 wks previously. RMHPS of 0-1 and 2-3 were denoted low and high respectively. Pts were divided into 4 groups based on the sequential RMHPS: low-low; low-high; high-low; and high-high. Median overall survival (OS) and time to study withdrawal were estimated by Kaplan-Meier method and prognostic variables compared by multivariate Cox regression analysis.

Results

One hundred and thirty one pts were identified. The most common tumours were colorectal, ovary and breast. Median age and lines of prior chemotherapy was 57y and 2 respectively. 99% of pts had ECOG PS 0/1 at baseline while 89% had distant metastases. Pts were treated on study for a median of 70 days(d), with 74.8% discontinuing for progression, 8.4% for toxicity and 12.2% for other reasons. Study withdrawal <30d was 17.6%. The proportion of pts in RMHPS groups were: low-low (62.6%); low-high (9.2%); high-low (6.1%); and high-high (22.1%). Pts in the low-high group had the worst median OS compared with the low-low, high-low and high-high groups (145 vs 358, 291 and 253d respectively, p < 0.0001). In multivariate analysis, the new RMHPS low-high category conferred significantly worse prognosis (HR 6.53, 95% CI 3.09-13.79) compared to other groups. Deterioration from low-high was 95.3% specific with a positive likelihood ratio of 6.47 for predicting 30-day trial discontinuation.

Conclusions

Deterioration in RMHPS ≤ 8 weeks prior to phase I trial treatment was highly predictive of early trial withdrawal and death, with a median time on study of 28 days. This simple, dual timepoint assessment potentially increases the clinical utility of the validated RMH prognostic model, identifying a specific population who withdraw early. We therefore recommend a 2nd RMHPS is repeated in pts of uncertain suitability for phase I trials. 1 Olmos, JCO 2012

Disclosure

All authors have declared no conflicts of interest.