5109 Background: Numerous well-designed retrospective studies of prognostic factors (pf) for survival (S) in metastatic renal cell carcinoma (mRCC) patients (pts) have been conducted since 1986. However, no single model for describing S in this population has been universally accepted. Methods: Authors of several existing prognostic indices, and others, formed the IKCWG to develop a single validated S model. The IKCWG has established a comprehensive database of previously reported clinical pf from 3748 previously untreated mRCC pts entered on institution review board approved clinical trials conducted by 11 centers in Europe and the United States from 1975–2002. Results: Median age at study entry was 58, 70% of pts were male, 89% had ECOG performance status (PS) 0 or 1; 75% had prior nephrectomy. 72%, 30%, and 19% of pts had lung, bone, and liver metastases (mets), respectively. 72% received interferon-a and/or interleukin-2 based treatments (tx); 25% were txd with chemotherapy/hormones only; 3% received other tx. 88% of pts have died; median S was 11.1 months (m). All examined factors except sex, age, and histology impacted S at p<.001 in univariable analysis. Multivariable analysis using a log-logistic model stratified by center and multivariable fractional polynomials was performed to identify independent predictors of S. Missing data were handled using multiple imputation methods. Using p=.0044 as the criterion for variable selection to avoid overly complex models, a model comprising tx, PS, number of met sites, interval from diagnosis to tx, and pre-tx hemoglobin, WBC, LDH, alkaline phosphatase and calcium was identified. The 25th and 75th percentiles of the prognostic index formed by multiplying each factor by its regression coefficient were used as cutpoints to form three risk (r) groups with median S times (SE) of: favorable r (n=937; 27.8 (0.4) m), intermediate r (n=1874; 11.4 (0.2) m), and poor r (n=937; 4.1 (0.1) m). Conclusions: 9 clinical factors can be used to model S in mRCC and form 3 distinct prognostic groups. Additional model building to determine if model complexity can be reduced further, validation in independent data and comparison to existing prognostic models are underway. No significant financial relationships to disclose.
Precision Surgery and Kidney Cancer: Knowledge of Genetic Alterations Influences Surgical Management
