E-poster Presentation 2014 World Cancer Congress

A Palliative Performance Scale Based Prediction Model for Survival in Patients Undergoing Palliative Radiotherapy (#955)

Fuqiang Wang 1 , Eric Pang 1 , Whee Sze Ong 1 , Daniel Quah 1
  1. National Cancer Centre Singapore, Singapore

Background: The palliative performance scale (PPS) is a well validated prognostic tool for survival in terminally ill patients. Previously Chow et al proposed a number of risk factors (NRF) prognostic model (primary cancer site, site of metastases and Karnofsky Performance Scale (KPS)) based on patients attending palliative radiotherapy clinic.

Aim: We aim to develop a PPS based prediction model as an alternative to the NRF model. 

Methods: This is a prospective cohort study of patients scheduled to receive palliative radiotherapy at the National Cancer Centre Singapore between August and December 2013. Overall survival (OS)  was measured from the date of starting radiotherapy till date of death. Cox proportional hazard regression models assessed  factors associated with OS. A prognostic score based on number of risk factors present was developed and compared against the NRF  using Harrell’s concordance index (c-index), D statistic of Royston and Sauerbrei (D-stat) and likelihood ratio (LR) analysis.

Results: 288 patients were enrolled in the study. Significant factors of OS include serum albumin, haemoglobin, white cell count, inpatient status at start of radiotherapy and PPS. 5-factor (Model A) and 3-factor (Model B based on albumin, inpatient status and PPS) prediction models were developed with good discrimination (c-indices >0.7) and calibration ability. When comparing Model A and B with NRF, Model B had the highest discrimination (NRF: c-index=0.592, D-stat=0.732, log-rank p=0.001; Model A: c-index=0.722, D-stat=1.694, p<0.001; Model B: c-index=0.729, D-stat=1.785, p<0.001). Model A and B were superior in predicting OS than the NRF on pairwise comparison based on LR analysis.

Conclusions: A simple 3-factor PPS based prognostic model (Model B) may serve as a useful alternative to the NRF for guiding radiation oncologists in prognostication and deciding the appropriate length of palliative radiotherapy. However, further validation study of the model is required. 

  1. Chow E, Abdolell M, Panzarella T, et al. Predictive Model for Survival in Patients with Advanced Cancer. J Clin Oncol 26: 5863-5869, 2008.