The biomarker CA125 has been validated as an indicator of recurrence and progression of disease in patients with ovarian cancer being treated with cytotoxic chemotherapy. More recently, the Rustin criteria have been established to use CA125 concentrations to determine disease response and progression in clinical trials. However, it is uncertain how CA125 concentration is affected by molecularly targeted agents. Bevacizumab (Bev) is an antibody against vascular endothelial growth factor (VEGF) and sorafenib (Sor) is a small molecule that targets the VEGFR-2 and Raf kinase. We analyzed our preliminary results from a phase I trial combining these 2 agents to determine if there was a correlation between CA125 concentration and disease response by RECIST criteria.


Patients were first enrolled on Sor and Bev at half the single-agent FDA approved doses (dose level 1), 200 mg PO BID and 5 mg/kg IV q2 weeks, respectively. Dose limiting toxicity was reached at dose level 2, with Sor 200 mg PO BID and Bev 10 mg/kg IV q2 weeks. Fourteen of 38 patients (DL1: 13 patients, DL2: 1 patient) enrolled had a diagnosis of recurrent epithelial ovarian cancer. CT scans were performed prior to therapy and then q2 cycles. Baseline CA125 concentration was measured on study and then monthly. CA125 changes were not used as a determinant of disease response or progression. We retrospectively analyzed the CA125 concentrations and compared them to tumor measurements from physical examination and CT scan.


Thirteen of 14 patients had abnormal CA125 concentrations at study entry (median 1056 U/mL; range 67 - 9813). Six patients (42%) had confirmed partial response (PR) by RECIST criteria; these patients had a median CA125 decrease of 70%. Four of these 6 patients would be considered to have PR by Rustin criteria (66% sensitivity). Seven patients would have been partial responders by Rustin criteria (50%); only 4 were confirmed by CT (specificity of 63%). The positive and negative predictive value of CA125 concentrations as a surrogate for response in this small cohort are 57% and 71%, respectively. RECIST criteria combined with CA125 criteria yielded a response rate of 62%. Three patients with PR documented by CT scan lasting 12+, 16+, and 20+ months would have been removed from study erroneously 2-6 months earlier if we had employed Rustin criteria. The patient at 20+ months withdrew from treatment due to travel fatigue and continues on therapy at home, now 22+ months with continued PR.


CA125 concentration does not have adequate sensitivity or specificity to be used as response criterion in this trial of combination anti-VEGF therapy with sorafenib and bevacizumab. Researchers should be cautioned before using CA125 concentration as a response criterion in trials of molecularly targets agents until prospective studies validate whether changes in CA125 concentrations mirror clinical and imaging response criteria.

98th AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA