Patient-reported KPS and other measures as predictors of survival in patients with advanced cancer.

Abstract

e19606 Background: Understanding determinants of survival remains a challenge in patients(pts) with advanced cancer. Prognostic factors of interest include pts rating of their own performance status, and patient responses from the EQ-5D. Methods: This prospective study was approved by the VA New Jersey HCS IRB. Pts with metastatic cancer whose cancer had already been treated with standard or experimental chemotherapy with KPS < 80%, or who did not wish to receive systemic chemotherapy, were recruited in a specified manner. At entry, Karnofsky performance status (KPS) was estimated, pts rated their own KPS, and answered a modified version of the EQ-5D. Cox regression survival analyses were performed. Results: Of 242 pts enrolled, 237 pts were analyzable. Median (M) age was 67 years (range 44-88), with 56% white, 41% black and 3% other; lung (26%) and prostate (18%) were the 2 most common primary sites and M KPS was 60% (range 30-100%). The majority (97%) of pts have died, with M survival of 95 days, range 4-2032 days. Higher KPS is associated with decreased risk of death (p<.0001). Both patient KPS (p<0.0319) and physician rated KPS were predictive of survival (p<0.004). Discrepancy between physician and pt KPS was noted, with upstaging by pts 48%, same for 27%, and downstaging in 25%, with no effect on survival. Physician KPS was a better predictor than pt KPS at each level. The EQ-5D pain item showed an increased risk of death with increasing pain (p<0.0001). The pain item was associated with KPS: pts with no pain had average KPS 75; moderate pain average KPS 63; extreme pain average KPS 48. The patient’s EQ-5D health rating was positively correlated with survival (p=0.0054), and with KPS (r=0.36). The other items in the EQ-5D did not predict survival. When all the factors (physician KPS, pt KPS, pain, health) were incorporated into a Cox model, only physician KPS was statistically significant (p=0.0033). Conclusions: Pts ratings of health and pain are significantly associated with KPS. Pts have a more positive outlook on their performance status. Physician KPS may be a better predictor because physicians have a wider frame of reference. Physician KPS can contribute to determination of hospice eligibility. Supported by VA HSRD IIR 2-103