Survival outcome and prognostic factors of patients with nasopharyngeal cancer in Yogyakarta, Indonesia: A hospital-based retrospective study

Abstract

Purpose This study aimed to determine the survival outcome and prognostic factors of patients with nasopharyngeal cancer accessing treatment in Yogyakarta, Indonesia. Methods Data on 759 patients with NPC diagnosed from 2007 to 2016 at Dr Sardjito General Hospital were included. Potential prognostic variables included sociodemographic, clinicopathology and treatment parameters. Multivariable analyses were implemented using semi-parametric Cox proportional hazards modelling and fully parametric survival analysis. Results The median time of observation was 14.39 months. In the whole cohort the median observed survival was 31.08 months. In the univariable analysis, age, education status, insurance type, BMI, ECOG index, stage and treatment strategy had an impact on overall survival (OS) (p values <0.01). Semi-parametric multivariable analyses with stage stratification showed that education status, ECOG index, and treatment modality were independent prognostic factors for OS (p values <0.05). In the fully parametric models age, education status, ECOG index, stage, and treatment modality were independent prognostic factors for OS (p values <0.05). For both multivariable analyses, all treatment strategies were associated with a reduced hazard (semi-parametric models, p values <0.05) and a better OS (parametric models, p values <0.05) compared with no treatment. Furthermore, compared with radiation alone or chemotherapy alone, a combination of chemotherapy and radiation either in a form of concurrent chemoradiotherapy (CCRT), sequential chemotherapy and radiation, or induction chemotherapy followed by CCRT demonstrated a reduced hazard (hazard ratio/HR 0.226, 95% confidence interval/CI 0.089–0.363, and HR 0.390, 95%CI 0.260–0.519) and a better OS (time ratio/TR 3.108, 95%CI 1.274–4.942 and TR 2.531, 95%CI 1.829–3.233) (p values < 0.01). Conclusions Median OS for the cohort was low compared to those reported in both endemic and non-endemic regions. By combining the findings of multivariable analyses, we showed that age, education status, ECOG index, stage and first treatment modality were independent predictors for the OS.