Chemotherapy Response Score: Development and Validation of a System to Quantify Histopathologic Response to Neoadjuvant Chemotherapy in Tubo-Ovarian High-Grade Serous Carcinoma.

PURPOSE: To develop and validate a histopathologic scoring system for measuring response to neoadjuvant chemotherapy in interval debulking surgery specimens of stage IIIC to IV tubo-ovarian high-grade serous carcinoma.

PATIENTS AND

METHODS

A six-tier histopathologic scoring system was proposed and applied to a test cohort (TC) of 62 patients treated with neoadjuvant chemotherapy and interval debulking surgery. Adnexal and omental sections were independently scored by three pathologists. On the basis of TC results, a three-tier chemotherapy response score (CRS) system was developed and applied to an independent validation cohort of 71 patients.

RESULTS

The initial system showed moderate interobserver reproducibility and prognostic stratification of TC patients when applied to the omentum but not to the adnexa. Condensed to a three-tier score, the system was highly reproducible (kappa, 0.75). When adjusted for age, stage, and debulking status, the score predicted progression-free survival (PFS; score 2 v 3; median PFS, 11.3 v 32.1 months; adjusted hazard ratio, 6.13; 95% CI, 2.13 to 17.68; P < .001). The three-tier CRS system applied to omental samples from the validation cohort showed high reproducibility (kappa, 0.67) and predicted PFS (CRS 1 and 2 v 3: median, 12 v 18 months; adjusted hazard ratio, 3.60; 95% CI, 1.69 to 7.66; P < .001). CRS 3 also predicted sensitivity to first-line platinum therapy (94.3% negative predictive value for progression < 6 months). A Web site was established to train pathologists to use the CRS system.

CONCLUSION

The CRS system is reproducible and shows prognostic significance for high-grade serous carcinoma. Implementation in international pathology reporting has been proposed by the International Collaboration on Cancer Reporting, and the system could potentially have an impact on patient care and research.