Immunohistochemically detected p53 mutations in epithelial tumors and results of treatment with chemotherapy and radiotherapy. A treatment-specific overview of the clinical data.
BACKGROUND: The aim was to ascertain whether many hundreds of clinical reports over the last decade are consistent with the prediction of a poorer outcome in cancer patients with p53 abnormalities treated with cytotoxic drugs and radiation. MATERIAL AND METHOD: There are 301 studies on the influence of p53 overexpression published through summer 2000, in which chemotherapy or radiotherapy was used alone or in combination with surgery. From 45 reports meeting stringent selection rules, comparison groups are identified in whom the same measure of outcome was reported for the same treatment applied to the same tumor, with results corrected for important prognostic factors. Metaanalysis techniques are then applied to the comparison groups. Attention was limited to reports using immunohistochemical techniques, to form comparison groups of sufficient size. RESULTS: Four comparison groups were identified by treatment and endpoint: 1) Stage I-III breast cancer (surgery and chemotherapy, disease-free survival, seven studies); 2) stage I-III breast cancer (surgery and chemotherapy, overall survival, six studies); 3) stage II-IV head and neck cancer (radiotherapy and chemotherapy, overall survival, five studies); 4) FIGO I-IV ovarian cancer (surgery and chemotherapy, overall survival, six studies). In the breast (disease-free survival) and ovarian (overall survival) comparison groups, the hazard ratio for a deleterious effect of p53 overexpression was significant or marginally significant, depending on assumed ranges for unreported hazard ratios in non-significant studies. CONCLUSIONS: Despite the many caveats related to metaanalysis applied to retrospective data, high variability of immunohistochemical technique, etc., a nearly significant negative effect of p53 overexpression on outcome of treatment with cytotoxic drugs and radiation emerges in the few studies where heterogeneity can be sufficiently reduced or accounted for.