Prognostic value of clinical variables in invasive cervical cancer

To test the hypothesis that clinical variables, such as patients' symptoms, symptom severity, and co-morbidity, affect the survival rate of patients with invasive cervical cancer.

From the medical records of 251 cases of invasive cervical cancer treated at Yale-New Haven Hospital between 1984 and 1988, information was extracted for patients' demographic characteristics, symptoms, symptom severity, comorbidity, physical findings, laboratory data, treatment, and subsequent course.

Three-year survival data were available for 250 (99%) of the 251 cases. For a composite clinical predictive system based on symptom status and co-morbidity, the 3-year survival rates were as follows: 85% (64 of 75) for the patients who were asymptomatic without co-morbidity; 63% (58 of 92) for the group that was either symptomatic or co-morbid, but not both; and 40% (33 of 83) for symptomatic patients with co-morbidity or patients with systemic, metastatic, or severe symptoms (P < .0001, chi 2 for linear trend). When entered into a Cox proportional hazards model along with other variables that might affect prognosis, including International Federation of Gynecology and Obstetrics stage, the composite symptom-co-morbidity stage remained statistically significant.

Our findings demonstrate the importance of clinical variables, such as symptoms and co-morbidity, in estimating prognosis in cervical cancer, even after stage and other factors are controlled. Unless the clinical variables are suitably analyzed, prognostic estimates based on morphology alone will be imprecise and therapeutic evaluations may be misleading.