Scoring System May Predict Chemo Response in Ovarian Cancer Patients

April 14, 2012

A study published today details a scoring system that may predict which ovarian cancer patients responded to first-line platinum chemotherapy based on a DNA-repair pathway-focused score. The score is based on a gene expression profile of 23 DNA-repair genes that normally function to repair platinum-induced DNA damage.

Platinum-based chemotherapy regimens are typically the first-line of treatment for women diagnosed with late-stage ovarian cancer postsurgery. While some women have a robust response (at least 12 months) to platinum based-chemotherapy such as cisplatin, carboplatin, or oxaliplatin, about 30% are chemoresistant patients who will relapse prior to 12 months.

A scoring system based on a DNA-repair pathway-focused score may predict which ovarian cancer patients will respond to first-line platinum chemotherapy

Currently, there is no way to predict which patients will respond. The high toxicity of the chemotherapy regimen allows less than half of patients to complete the 6- to 8-course treatment, further underscoring the need to single-out the patients most likely to respond.

A study published today in the Journal of the National Cancer Institute details a scoring system that may predict which patients responded to first-line platinum chemotherapy based on a DNA-repair pathway-focused score. The score is based on a gene expression profile of 23 DNA-repair genes that normally function to repair platinum-induced DNA damage.

Josephine Kang, MD, PhD, department of radiation oncology at Dana Farber Cancer Institute, and colleagues tested the score on a set of 304 patients with advanced epithelial ovarian cancer who were treated with a platinum-based regimen and whose tumor expression profiles were available through the Cancer Genome Atlas (TCGA) database. A 161-patient group who did not receive platinum chemotherapy served as the control group. The prognostic value of the patient score-either high or low-was evaluated based on overall survival and progression-free survival.

The Score Results

Patients with a high score had improved overall survival (40% had a 5-year overall survival) which was statistically significant compared to those patients that had a low score (17% had a 5-year overall survival, P < .05). Of the patients treated with platinum chemotherapy, 71% in the data set achieved a complete response. The researchers looked specifically at the association of BRCA somatic and germline mutations and outcomes but did not find a statistically significant correlation.

Still, the ovarian cancer response score, after further validation and development, would need to be validated in a prospective trial. Dr. Kang and colleagues believe that the lowest score patients, who have a median overall survival of only 2.1 years have the most to gain from establishing a predictive approach to treatment. “With additional prospective validation in clinical trials, we hope that the score can become a powerful tool that is useful in stratifying advanced-stage ovarian cancer patients toward optimal treatments incorporating new treatment regimens vs current standard of care,” the authors conclude.

In an accompanying editorial, Elizabeth M. Swisher, Toshiyasu Taniguchi, and Beth Y. Karlan note that because almost 80% of primary ovarian cancers initially respond to chemotherapy, a prospective biomarker would need to be very highly predictive. The alternatives to platinum-based chemotherapy are experimental clinical trials. The editorial strongly states that this test is nowhere near “ready for prime time.”

Swisher, of the department of obstetrics and gynecology at the University of Washington and her coauthors believe that the study’s hypothesis-that malignant cells deficient in DNA-repair are more likely to demonstrate a more durable response to platinum-based chemotherapy-while reasonable, still needs fine-tuning. Specifically, how to define DNA-repair deficient, and which gene sequences to assess is not yet clear cut.

The outcome of this study was surprising according to the editorial: those patients whose ovarian tumors had high expression of DNA-repair genes had worse outcomes-the opposite of the study hypothesis. Swisher and colleagues conclude that the TCGA data is complex and may not be specifically useful for extracting predictive biomarkers. It is likely necessary to better understand molecular phenotypes before the creation of a prediction score. Swisher et al suggest that a functional DNA-repair test may have more utility that a gene expression signature.