Several molecular-genetic alterations in epithelial ovarian cancer, including expression of glutathione S-transferase-pi(GST-pi), p-glycoprotein(p-gly), and metallothionein(MT), have been associated with drug resistance and poor prognosis. To deterrmine the importance of molecular-genetic factors relative to more traditional surgical-pathologic prognostic factors, a multivariable analysis was performed.

Immunohistochemical staining for GST-pi, p-gly, and MT was performed in relation to the response to chemotherapy with cisplatin in 87 patients with primary epithelial ovarian cancer who had not received any chemotherapy before surgery. In addition, information regarding surgical-pathologic features of the cancers was obtained. Univariabe analysis was performed followed by multivariable analysis using Cox's proportional hazards model to identify variables predictive of poor prognosis.

With univariable analysis, age, FIGO stage, grade, MT, ascites, residual tumor, and initial value of CA 125 were predictive of patient's survival. In the multivariable analysis, only age, grade, stage, histologic type nad residual tumor were significant. Cox's regression model fitting stratified by stage, MT(p=0.0362) was significant in early stage of cancer, however, histologic type(p=0.0001)was significant in advanced ovarian cancer. Conclusions. Among factors for drug resistance to chemotherapy, MT was the most strongly predictive of survival in early ovarian cancer.