Development of prognostic index for primary supratentorial intracerebral tumours

The clinical course of intrinsic supratentorial tumours is variable. Prediction of outcome would be useful in defining patients for specific treatment policies. A retrospective analysis of 560 patients with intrinsic supratentorial tumours was performed. Proportional hazards models for survival were derived by using a stepwise selection procedure with only clinical and CT features as possible explanatory variables. The variables of prognostic importance were age, a first symptom of epilepsy, focal signs at presentation, a cystic lesion on CT scan, and duration of symptoms before presentation. The model defined a group with a good prognosis (score less than or equal to 9, n = 211) and a group with a poor prognosis (score greater than 9, n = 344). The median survival was 27 months for those with a score less than or equal to 9 or less and three months for those with score greater than 9. An alternative model, not including duration of symptoms, is also capable of defining groups with long (score less than or equal to 16, n = 234) and short (score greater than 16, n = 325) survival. The model may provide a means of classifying patients for inclusion in prospective randomised studies.