Objectives: In the search for an early measure of response in hormone refractory prostate cancer (HRPC), most have targeted changes in serum prostate-specific antigen (PSA). Up to this point, no one has targeted changes in serum hemoglobin during treatment. If dynamic changes in hemoglobin after treatment provide additive prognostic information to dynamic changes in PSA, then we should consider and test ways to incorporate serum hemoglobin into measures of response in HRPC.
Methods: Our patients consisted of 321 men who were studied on Cancer and Leukemia Group B protocols 9181 and 9182. We fit serial values of PSA and hemoglobin with an exponential model: y = exp(alpha + beta*t + gamma*t(2)) with y symbolizing either PSA or hemoglobin and t denoting time. We then used the Cox proportional hazard model to relate the parameters of the model (alpha, beta, and gamma) to subsequent survival.
Results: We found that the exponential model fit serial measurements of both PSA and serum hemoglobin well, and all three of the parameters for both markers related closely to subsequent survival (P < or = 0.003). The Cox model suggested a composite hazard score (HS) as a way to consolidate the information from serial measurements of both serum markers, and we observed that those with HS < 0 enjoyed a longer survival.
Conclusion: Because serial measurements of serum hemoglobin during treatment of HRPC add prognostic information to serial measurements of PSA, we hypothesize that combining the dynamic changes in serum hemoglobin with those of PSA could lead to an improved measure of response in HRPC.