Primary therapeutic decision-making in inoperable non-small cell lung cancer

Purpose: To test the accuracy of our treatment decisions for patients with inoperable non-small cell lung cancer (NSCLC) using a prototype decision-su pport system (DSS) and a prognostic index (PI). Methods and Materials: To predict patient outcome and select optimal treatm ent, the systems protocol was tested retrospectively in 242 patients with S tage I-IV disease. The PI was determined in 184 patients with Stage I-IIIa, b disease. Survival was the final determinant of the accuracy of our treatm ent decisions. Results: Until 1996 it was our treatment policy to treat all Stage III pati ents with radical intent unless they had symptoms requiring palliation, In 1997, after the palliation concept of the DSS and the PI were changed to in clude all Stage In disease, there was considerable discordance between the rates of palliative treatment indicated by the DSS and the PI (69% and 99%, respectively) as well as that observed in our practice (30% in the DSS gro up and 20% in the PI group, respectively). There was also a significant dif ference in survival between the patients in the low- and high-risk categori es defined by the PI (median survival of 12 versus 6 months, respectively; p = 0.0001). In the group that received radical radiotherapy, there was als o a significant difference in the duration of survival between the low- and high-risk groups (median survival of 12 versus 8 months, respectively; p = 0.01). In addition, the risk categories proved to be the most important pr edictor of survival in the patients receiving radiotherapy longer than 2 we eks (median survival of 12 versus 7 months, respectively; p = 0.0001), In h igh-risk patients, however, the duration of radiotherapy did not have a sig nificant impact on survival (p = 0.25). Conclusion: Our data indicate that the PI is a useful method for selecting radical or palliative treatment modalities as well as for determining treat ment duration. (C) 2000 Elsevier Science Inc.