Human papillomavirus type 18: association with poor prognosis in early stage cervical cancer

Cervical carcinoma is a leading cause of mortality from cancer among women worldwide, accounting for approximately 160,000 deaths annually. Prognosis in patients with this disease is dependent on several well-established clinical features (stage of disease and age of patient) and pathologic features (lymph node status, grade of tumor, and depth of invasion). Although the features associated with poor clinical outcome have been well studied, molecular markers such as human papillomavirus (HPV) type that may reflect the underlying biologic basis for clinical behavior are poorly understood.

To test the hypothesis that differences in survival among patients with cervical carcinoma are associated with HPV DNA type, we conducted a historical cohort study of patients treated at our institutions over a 10-year period.

Fresh primary tumor tissue samples from 291 women with all stages of cervical carcinoma diagnosed from April 1983 through August 1993 were rapidly frozen and stored at -70 degrees C until analysis. High-molecular-weight DNA was extracted and purified by homogenization, proteinase K digestion, phenol extraction, ammonium acetate salt displacement, ethanol precipitation, and ribonuclease treatment. HPV nucleotide sequences were amplified from tumor DNA samples by polymerase chain reaction with the use of both consensus L1 (MY09/MY11) primers that recognize more than 25 HPV types and modifications of type-specific primers developed for HPV types 16, 18, and 6. Clinical data were abstracted from hospital, office, and tumor registry records. Univariate analysis was conducted using Student's t test and chi-squared tests. Survival curves were estimated by use of the Kaplan-Meier method; differences between groups were examined by the logrank test. Multivariate survival analysis was performed according to the Cox proportional hazards model.

HPV DNA was detected in 247 (85%) of 291 tumors: HPV16 in 52%, HPV18 in 20%, other HPV types in 13%, and no HPV DNA in 15%. Eighty-eight percent of squamous tumors contained HPV DNA compared with 79% of adenocarcinomas, the latter harboring predominantly HPV18. Women 45 years old or younger with a history of cigarette smoking tended to have HPV DNA in their tumors, but the HPV type was not associated with established prognostic factors such as stage, grade, lymph node metastasis, or depth of stromal invasion. After a median follow-up of 38.9 months, among potential prognostic factors of patient age, histologic cell type, grade, and HPV DNA status, only stage was predictive of survival in the entire study population. However, among the 171 patients treated with type III radical hysterectomy (removal of uterus and upper vagina along with other tissues extending to the pelvic wall) and pelvic lymphadenectomy (removal of all lymphatic tissue in the pelvis), multivariate analysis determined that lymph node status (adjusted risk ratio [RR] = 3.12; 95% confidence interval [CI] = 1.35-7.21), depth of stromal invasion (adjusted RR = 3.14; 95% Cl = 1.05-9.34), and the presence of HPV18 DNA (adjusted RR = 2.59; 95% CI = 1.08-6.22) were statistically significant predictors of survival.

HPV18 DNA type is an independent prognostic factor in patients with cervical carcinomas treated with radical hysterectomy and pelvic lymphadenectomy.

The use of molecular markers such as HPV DNA type may allow the identification of patients with early stage cervical cancer at high risk for disease recurrence.