Nomogram Prediction of Response to Neoadjuvant Chemotherapy Plus Pembrolizumab in Locally Advanced Hypopharyngeal Squamous Cell Carcinoma.

Abstract

Importance: The objective response (ORR) rate in patients with locally advanced hypopharyngeal squamous cell-carcinoma (LA-HPSCC) following neoadjuvant chemotherapy (NACT) of albumin-bound paclitaxel plus carboplatin is low. At present, it is unclear whether the addition of pembrolizumab could increase the ORR or not.

Objective: To investigate whether the addition of pembrolizumab could increase the ORR, and to develop a nomogram to predict the response of pembrolizumab addition.

Design: Retrospective cohort study.

Setting: This study was conducted at a single institution.

Participants: This study included 129 patients who conformed to the inclusion criteria.

Intervention or exposures: NACT with or without pembrolizumab for patients with LA-HPSCC.

Main outcome measures: The ORR was analyzed according to the RECIST 1.1 criteria and a nomogram was developed based on least absolute shrinkage and selection operator and multivariable Cox regression analysis. Predictive accuracy and discriminative ability of the nomogram were evaluated by receiver operating characteristics, precision recall, calibration curves, and decision curve analysis.

Results: Eighty-two patients received NACT and 47 also received pembrolizumab. ORR was higher in patients receiving additional pembrolizumab (66.0% vs 47.6%, χ² = 4.074, P = .044). The nomogram identified pretreatment levels of lymphocytes and red blood cells as independent predictors of a high ORR, while basophil levels were an independent predictor of a low ORR. Calibration curve showed that the nomogram-based predictions corresponded well with actual observations. C-index of the nomogram was 0.925 (0.848-1.002) and the area under curve was 0.925. Decision curve analysis affirmed that the nomogram had important clinical value.

Conclusions and relevance: Pembrolizumab could improve the ORR in LA-HPSCC patients treated with NACT. Furthermore, a risk-prediction nomogram incorporating readily assessable routine pretreatment blood parameters can accurately estimate the response to NACT with pembrolizumab, leading to precise treatment and minimizing the waste of medical resources.