A risk prediction model for cisplatin-induced acute kidney injury in patients with head and neck cancer who receive chemoradiotherapy: A supplement analysis of a phase II/III trial of JCOG1008

12104 Background: Cisplatin-induced acute kidney injury (C-AKI) is a common reason for treatment interruption in head and neck cancer patients who receive chemoradiotherapy (CRT). We developed a risk prediction model for C-AKI in patients from a randomized phase II/III trial of JCOG1008, which demonstrated that CRT with weekly cisplatin at 40 mg/m 2 (weekly arm) was noninferior to 3-weekly cisplatin at 100 mg/m 2 (3-weekly arm) in terms of overall survival for postoperative high-risk head and neck cancer ( J Clin Oncol 2022; 40: 1980-90). Methods: Two hundred fifty-one patients received RT with a total of 66 Gy/33 fr concurrently with 3-weekly or weekly cisplatin. AKI was defined using the AKI Network classification/staging system as an increase in serum creatinine of ≥0.3 mg/dL or a 1.5-fold or greater increase from baseline (≥ stage I) by 30 days after the completion of CRT. AKI risk prediction models were developed by combining explanatory variables which were expected to contribute, including age, sex, ECOG performance status, primary site, pathological stage, diabetes mellites, hypertension, main nutritional method, treatment arm, serum albumin, serum magnesium, and creatinine clearance (CCr) at registration. A logistic regression model with the Akaike information criterion was used to explore the optimal model. Results: Ninety-four of 251 (37.5%) patients developed C-AKI in this study. The optimal C-AKI risk prediction model was composed of four factors, including a primary site (hypopharynx/larynx or oral cavity/oropharynx, odds ratio [OR] 1.92 [95% CI, 1.10 to 3.30]), treatment arm (3-weekly arm or weekly arm, OR 1.75 [95% CI, 1.02 to 3.00]), serum albumin (≤ 3.5 g/dL or > 3.5 g/dL, OR 2.06 [95% CI, 1.11 to 3.87]) and CCr ( < 90 mL/min or ≥ 90 mL/min, OR 1.85 [95% CI, 1.05 to 3.26]). The incidence of C-AKI increased in accordance with a total count of the four factors (Table): with a total count of 0-1, 2 and 3-4, the incidence of clinically relevant stage II/III C-AKI was 2.9%, 11.5%, and 22.6%, respectively. When total count was ≤1, the positive and negative predictive values for stage II/III C-AKI development were 16.1% and 97.0%, respectively. Conclusions: We developed a risk prediction model for C-AKI in head and neck cancer patients who receive CRT using primary site, cisplatin administration method, serum albumin, and CCr. The high negative predictive value of this prediction model may help rule out the development of moderate to severe C-AKI. Clinical trial information: jRCTs031180135 . [Table: see text]