We appreciate Dr. Markman's thoughtful comments regarding our article, “Lack of reliability of CA125 response criteria with anti-VEGF molecularly targeted therapy.”1 We agree that monitoring disease response in ovarian cancer is problematic and controversial; various studies have produced differing results regarding whether changes in CA-125 concentrations or radiologic measurements are better predictors of disease behavior on a given therapy. However, multiple studies have demonstrated that changes in CA-125 concentrations do correlate well with changes in radiologic measurements when cytotoxic chemotherapeutics are employed. Taking these data into consideration, the Gynecologic Cancer Intergroup has endorsed using the Rustin CA-125 criteria as secondary endpoints in first-line therapy randomized trials.2
The thrust of our article is that these CA-125 criteria have not been validated in treatment trials of targeted agents. On the basis of the results from our small sample, the same correlation between CA-125 and radiologic response may not be present in targeted therapy trials—a phenomenon that also has been described in the prostate (prostate-specific antigen) literature.3 Accordingly, we believe that clinical trialists need to be cautious in designing trials for patients with ovarian cancer so that patients are not removed prematurely from study or, alternatively, kept on study well after clinical benefit has ceased. Dr. Markman's points regarding the common sense approach of not removing patients from a treatment because of modest increases in either tumor size on imaging or CA-125 concentrations are well taken; however, clinical trials may not allow such leeway if strict Rustin CA-125 criteria are written into the study initially. Had we written the Rustin criteria into our trial as response criteria, as stated in our article, then 3 of 7 partial responders would have been taken off study prematurely: One of those patients currently remains on study 2 years after she would have had to cease therapy. The objective of our article was not to prove that changes in imaging parameters are better assessors of disease behavior than CA-125 concentrations. Rather, we emphasized that the 2 parameters may not correlate with one another in molecularly targeted therapies the way they do in cytotoxic chemotherapy. Instead of arbitrarily choosing 1 method over the other, a better alternative would be to test both CA-125 criteria and imaging criteria in a prospective, properly powered study of targeted therapy correlated with patient outcome.