Ki-67 staining is an independent correlate of biochemical failure in prostate cancer treated with radiotherapy.

PURPOSE To determine the significance of Ki-67/MIB1 staining as a marker of patient outcome for prostate cancer patients treated with radiotherapy. EXPERIMENTAL DESIGN Pretreatment archival prostate biopsy tumor tissue was available from 106 stage T1-T4 prostate cancer patients treated with external beam radiotherapy between 1987 and 1993 at M. D. Anderson Cancer Center. Diagnosis was made from prostate needle biopsy in 64 cases and from transurethral resection of the prostate (TURP) in 42 cases. All patients had a pretreatment prostate-specific antigen (PSA), and no patient had evidence of metastasis. Immunohistochemical staining for MIB1 was used to determine the percentage of Ki-67-positive tumor cells, the Ki-67 labeling index (Ki67-LI). Biochemical failure after radiotherapy was defined as three rises in PSA on follow-up. Median follow-up was 62 months. RESULTS The mean and median Ki67-LI for the entire cohort was 3.2 and 2.3. The mean and median Ki67-LIs for those diagnosed by needle biopsy were 3.2 and 2.3, and by TURP were 3.1 and 2.4. For all patients, mean Ki67-LI levels were significantly higher with stage T3/T4 disease, Gleason 7-10 disease, and in those that developed treatment failure. Similar relationships were observed when the Ki67-LI was dichotomized into low (< or =3.5%) and high (>3.5%) groups. Actuarial freedom from biochemical failure (bNED) when Ki67-LI was low and high was 76 and 33% at 5 years (P < 0.0001, log rank). Similar statistically significant differences were observed when the TURP and needle biopsy groups were analyzed separately. Cox proportional hazards regression showed that dichotomized Ki67-LI was an independent correlate of bNED, along with pretreatment PSA, Gleason score, and clinical stage. CONCLUSIONS The Ki67-LI obtained from pretreatment prostate cancer tissue is a strong independent predictor of failure after radiotherapy using biochemical criteria. This prognostic factor was equally valuable for patients diagnosed by TURP or needle biopsy.