TRANSFORMATION IN RECURRENT OVARIAN GRANULOSA-CELL TUMORS - KI67 (MIB-1) AND P53 IMMUNOHISTOCHEMISTRY DEMONSTRATES A POSSIBLE MOLECULAR-BASIS FOR THE POOR HISTOPATHOLOGIC PREDICTION OF CLINICAL BEHAVIOR

Ovarian granulosa cell tumors (GCTs) behave unpredictably. Stage I pat ients suffer recurrences many years after treatment, and histopatholog ic evaluation of the primary GCT offers only a few clues. Grading, in particular, is largely ineffective. Ri67 (MIB-1) and p53 monoclonal an tibodies (active on paraffin embedded tissues) provide insight into nu clear proliferation and control, respectively. In this study, the auth ors hypothesized that these molecular markers will help predict the cl inical behavior of GCTs. Paraffin sections from 68 GCTs (arising in 56 patients: 53 primary and 15 recurrent) including 34 typical and 27 di ffuse adult, and seven juvenile types were immunostained for Ki67 (MIB -1 clone; Immunotech, Westbrook, ME) and p53 (DO7 clone; Novacastra La boratories, UK). The Ki67 proliferation index (Ri67PI = percentage imm unoreactive on a count of at least 400 nuclei) ranges from 1 to 50% (m ean, = 12.2%; median, 9.3%). Nineteen percent of GCTs exhibited focal p53 immunoreactivity; the number of GCTs and proportion of nuclei deco rated were as follows: four, <1%; seven, 1% to 10%; and one, 20%. Ki67 PI was higher in recurrent tumors (P < .001) and correlated with mitot ic rate (r = .75; P < .0001). p53 staining was associated with juvenil e type GCTs (P < .001) and higher Ki67PI (P < .005). Other histopathol ogic features exhibited no association with p53 staining or Ki67PI. Fo llow-up was available for 54 of 56 patients: 18 suffered recurrences a fter 16 to 229 months (mean, 72.1 months; median, 59 months), and 36 w ere disease free 16 to 369 months (mean, 78.2 months; median, 70 month s) after diagnosis. Curiously, high Ki67PI and mitotic rates of the pr imary GCT correlated weakly with a disease-free course (P = .03 and .0 7, respectively). Disease recurrence was associated with stage > I (P < .0005), vessel invasion in the capsule (P < .001), ruptured tumors ( P < .005), and older patients (P < .02). p53 staining and size or subt ype of GCT exhibited no prognostic value. For 12 patients, paired prim ary and first recurrence of GCT showed a striking increase in Ki67PI ( P < .00005) and mitotic activity (P < .02) in the recurrence. p53 expr ession also appeared (de novo) in two recurrent GCTs. The interpretati on of focal p53 staining (>10% nuclei decorated in only one GCT) is co ntroversial. Some investigators suggest that this represents overexpre ssion of wild type p53 rather than p53 gene mutation. Primary GCTs exh ibit a wide spectrum of proliferative activity, and the seven juvenile GCTs (the most proliferative type) demonstrated no recurrences in thi s study. Recurrent GCTs displayed a transformation of molecular marker s to increased proliferative activity and overexpression of p53, funda mentally, by these markers, a different GCT than the primary one. Thes e findings suggest a molecular basis for the lack of histopathologic p redictors for recurrence. Factors other than proliferation of the prim ary GCT (which relates most closely to grading) either extrinsic to th e neoplasm (host dependent) or as yet undetectable must determine mali gnant behavior. (C) 1996 by W.B. Saunders Company