Prediction of tumour regrowth of pontine glioma using a two-term model.

Background: Mode lling tumour regrowth kinetics and determining tumour burden is critically important in making accurate patient prognoses. Patients and Methods: Our two-term model, which describes tumour regrowth after chemotherapy was tested by analyzing tumour sizes in patients with pontine gliomas treated at a major cancer centre during the past 40 years. Tumour measurements at four early time points were used to fit the model, which w as then used to predict later tumour sizes. The predictions were compared with observed measurements from clinical charts and this model 's performance was compared to that of a quadratic model. Results: The mean of relative errors using the two- term model was 0.186; standard deviation, 0.176. For the quadratic model, the mean of relative errors was 0.412; standard deviation, 0.521. Conclusion: The two-term model fits the tumour regrowth data reasonably well and is more accurate than a competing quadratic model used in the same way (p<0.05). Accurately determining tumour burden and evaluating