Prognostic factors in 677 patients in Singapore with nondisseminated nasopharyngeal carcinoma

BACKGROUND. The objective of the current study was to describe the survival of nasopharyngeal carcinoma (NPC) patients in Singapore, verify the progno stic value of the revised 1997 TNM staging system, and develop a multivaria te prognostic model for NPC. In addition, the authors also examined the pro gnostic value of characteristics of lymph node spread and parapharyngeal in volvement. METHODS, A prospectively maintained database containing clinical and comput ed tomography scan data was used to reclassify 677 NPC patients treated bet ween 1992 and 1994 according to the new staging system. Records were linked with the death registry to ascertain the patient's vital status and date o f death. Overall and stage specific survival were analyzed using the Kaplan -Meier method and the log rank test. Univariate and multivariate Cox propor tional hazards regression analysis were used to obtain prognostic models. RESULTS. Two hundred seventy-four deaths (40.5%) occurred. The 5-year survi val rate was 56.6% (95% confidence interval [95% CI], 52.3%, 60.7%). The st age specific 5-year survival rates were: Stage I, 88%; Stage IIA, 75%; Stag e IIB, 74%; Stage III, 60%; Stage IVA, 35%; and Stage IVB, 28%. TNM stage w as found to be a statistically significant prognostic factor (P < 0.0001). Cranial nerve (hazard ratio [HR]: 2.77), orbit (HR: 5.71), and intracranial involvement (HR: 2.46) conferred a particularly bad prognosis in univariat e analysis. Independently significant prognostic factors were age; lymph no de status; and paraoropharyngeal, cranial nerve, orbit, and nasal involveme nt. Among lymph node positive patients, independently significant prognosti c lymph node characteristics were Ho level and laterality. Although parapha ryngeal involvement appeared to be prognostically unimportant, paraorophary ngeal involvement distinguished a subgroup with a poorer prognosis (HR: 1.8 4; 95% CI, 1.45, 2.34; P < 00001). Lateral spread to the medial infratempor al fossa and beyond also was found to confer a poorer prognosis. CONCLUSIONS. The results of the current study show that the revised 1997 TN M staging system is prognostically useful. Subdivision into paraoropharynge al involvement and using the medial infratemporal fossa to delineate progno stically significant lateral spread should be considered in future revision s. Cancer 1999;86:1912-20. (C) 1999 American Cancer Society.