Ovarian Cancer, CA-125 Addiction, and Informed Confusion: Much Ado About Less

Abstract

Biomarkers can be useful in risk stratification of a disease or as a surrogate end point in determining the effectiveness of treatment. CA-125 has been proposed as a surrogate for response in ovarian cancer, and in most clinical trials, clinical complete response is defined as radiographic disappearance of all disease, resolution of all disease-related symptoms, and normalization of the CA-125 if it was elevated at therapy initiation. Less well accepted is assignment of complete response in a patient whose only evidence of disease is an elevated tumor marker that subsequently normalizes on therapy. It is possible that this type of serologic definition of complete response is as good as or better than conventional radiographic imaging in patients with advanced ovarian cancer. CA-125 has also been shown to be an accurate marker to define progression of ovarian cancer, and becomes elevated on an average of 3 to 4 months (range, 1 to 15 months) before clinically assessable (symptomatic, palpable, or visible by imaging) disease in 70% of patients. Furthermore, an increasing CA-125 serum level alone is sometimes being used to initiate cytotoxic systemic therapy for recurrence in the absence of clinically demonstrable disease, despite the lack of evidence demonstrating a survival advantage for this approach. Many investigators have attempted to codify what degree of CA-125 elevation can predict progression of ovarian cancer with acceptable specificity and sensitivity. Although this serologic approach to assess disease progression may eventually prove useful in the conduct of clinical trials, it may not demand routine treatment initiation, given that there is wide variability in lead time between biochemical and clinical progression. Furthermore, there is significant variation among investigators in terms of when to initiate treatment for recurrent disease based on personal bias and patient preference. The Gynecologic Cancer Intergroup (GCIG) defines progressive disease after a complete response to primary therapy based on CA-125 criteria: the date of first elevation of CA-125 to two-fold the upper limit of normal (documented on two occasions at least a week apart). For those with persistently elevated CA-125 levels, progression of disease is defined as the first date of CA-125 2 the nadir value documented on two occasions no less than a week apart. A number of ongoing and planned clinical trials incorporated these existing GCIG criteria in conjunction with the standard Response Evaluation Criteria in Solid Tumors Group (RECIST) classifications in an attempt to validate the use of serum CA-125 levels as surrogate markers for defining disease progression in ovarian cancer. Recently, the GCIG criteria of progression were validated using data from an Intergroup trial that randomly assigned patients to frontline combination chemotherapy with cisplatin and paclitaxel compared with cisplatin and cyclophosphamide. The analysis using GCIC criteria showed that the magnitude of therapeutic benefit was similar using either CA-125 or RECIST criteria. From an economic point of view, use of a biomarker seems preferable to more expensive radiographic testing that is often used to determine response and progression in the context of clinical trials. However, outside of a clinical trial, what should one do when the marker signals progression but the patient’s review of systems is negative and the imaging study of choice is within normal limits? In such a situation the false-positive rate of the marker signal becomes paramount because this type of information could lead to inappropriate treatment. In addition, even if clinical recurrence is predicted by the marker signal, there are no data demonstrating a survival or quality-of-life benefit for the patient treated before clinical progression. In the validation trial by Rustin et al, the false-positive rate was 2%. The only two clinical trials addressing immediate versus delayed therapy at the time of biochemical progression are the Medical Research Council (MRC) 05 study combined with the European Organization for the Research and Treatment of Cancer (EORTC) study 55955, and Gynecologic Oncology Group (GOG) trial 198. The first trial simply randomly assigns notification or lack thereof of the patient when there is biochemical evidence of recurrence. Therapy is not controlled. In GOG 198, patients with biochemical recurrence are randomly assigned to receive either tamoxifen or thalidomide until clinical progression, with progression-free survival (PFS) as the end point. What might this new and potentially more sensitive indicator of recurrence do? First, it will shorten the PFS of the prior therapy. Second, it may alter the definition of platinum resistance, given that currently, platinum-free intervals are based largely on intervals between discontinuation of primary therapy and presentation with clinical recurrence. Third, it will lead to inappropriate therapy in those patients with false-positive results and possibly also in those patients where there is a long lead time between biochemical and clinical progression. These changes in definitions are largely surmountable because clinical trials typically are randomized JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 25 NUMBER 24 AUGUST 2