Prognosis Model Identifies Four Risk Factors Associated with Overall Survival

Chicago—Researchers have identified four readily available clinical and laboratory factors to generate a prognosis index model (PIM) for overall survival (OS) in patients with metastatic castration-resistant prostate cancer (mCRPC). That was one of the findings of modeling conducted by researchers utilizing the final analysis dataset from COU-AA-302.

Charles J. Ryan, MD, and colleagues reported their findings during a poster session at the ASCO 2015 Annual Meeting. The poster was titled Prognostic Index Model (PIM) for Overall Survival (OS) in Metastatic Castration-Resistant Prostate Cancer (mCRPC) Patients (pts) without Prior Chemotherapy Treated with Abiraterone Acetate (AA).

Treatment decisions for patients with mCRPC may be assisted with the use of models that predict prognosis. The researchers previously presented a PIM for radiographic progression-free survival from the phase 3 COU-AA-302 trial in chemotherapy-naïve mCRPC patients.

The final analysis of COU-AA-302 data demonstrated a significant prolonged median OS in patients treated with abiraterone acetate (AA) plus prednisone (P) compared with P alone (34.7 months vs 30.3 months; hazard ratio [HR], 0.81; 95% confidence interval, 0.7-0.93; P=.0033). Using the final analysis dataset, the researchers developed a PIM for OS in patients with mCRPC in the AA plus P arm.

Of 546 patients, the complete data set was available for 90% of patients (n=493); these data formed the basis for the modeling. Laboratory factors were dichotomized using accepted values for lower (LLN) and upper (ULN) limits of normal; other factors were dichotomized for ease of interpretation.

A univariate Cox model was used to assess factors for association with OS; the factors were used in a multivariate Cox model with a stepwise procedure. A bootstrap resampling procedure was used for internal validation of the predictive value of the final value. Model discriminatory power was estimated by the C-index.

The final model included four factors associated with poor prognosis: (1) Brief Pain Inventory score of 2 or 3 (HR, 1.71; P<.0001); (2) lactate dehydrogenase >ULN (234 IU/L) (HR, 2.03; P<.0001); (3) alkaline phosphatase >ULN (131 IU/L; HR, 1.6; P=.0004); and (4) ≥10 bone metastases (HR, 1.92; P<.0001).

Using number of risk factors, patients were categorized into three risk groups: good (n=296); intermediate (n=117); and poor (n=131). Median OS was calculated for each group. For patients in the good group (0 or 1 risk factor), median OS was 53.6 months. For those in the intermediate group (2 risk factors), median OS was 38.4 months. For patients in the poor group (3 to 5 risk factors), median OS was 26.6 months. The C-index was 0.67.

The researchers concluded, “We identified four readily available clinical and laboratory factors to generate a PIM for OS in mCRPC, and categorized patients into three distinct risk groups. If validated in an independent dataset, this PIM may be useful for estimating OS in mCRPC and designing risk-adapted treatment strategies.”

Source: Ryan CJ, Kheoh T, Molina A, et al. Prognostic index model (PIM) for overall survival (OS) in metastatic castration-resistant prostate cancer (mCRPC) in patients (pts) without prior chemotherapy treated with abiraterone acetate (AA). Abstract of poster presentation at the American Society of Clinical Oncology 2015 Annual Meeting, Chicago, Illinois, May 30, 2015.