Abstract
BACKGROUND
Low-grade-glioma (LGG) is often a chronic disease requiring multiple treatment-interventions. The multi-state-model (MSM) analyzes almost unlimited disease-states that patients may undergo.
PURPOSE
We analyzed disease-state probability of German LGG-patients within the SIOP-LGG 2004 study. Starting-state was defined as “diagnosis” and final-state as “death” with in-between “disease-states”, i.e. surgery, chemotherapy or radiation. We then evaluated clinical risk factors associated with multiple interventions.
RESULTS
We identified 1587 patients and a total of 23 different states (median follow-up 6.2 years). For robust statistical calculation, we reduced the model to 10 states. Five years after diagnosis state probabilities of patients were as follows: 0.11 no therapy at all, 0.49 one and 0.10 two surgeries-only, 0.19 one, 0.04 two and 0.02 even three non-surgical interventions with or without prior surgery, 0.005 had developed a malignant tumor and 0.02 had deceased. Risk-group analysis revealed higher probability for multiple interventions for infants (0.40), tumor location in supratentorial midline (0.21) or cerebellum (0.20), astrocytic tumors (0.22) and degree of resection, i.e. partial resection (0.31) and biopsy-only (0.35) five years after diagnosis. Probabilities for patients with neurofibromatosis type-1 were most likely not to be treated at all (0.36) or to receive non-surgical therapy, if needed (0.45). Sex had no influence.
CONCLUSION
This is the first MSM analysis of a LGG clinical trial cohort reflecting natural history and transitions to multiple interventions. A substantial number of patients is at risk for successive treatments with associated neurotoxicity. This underscores the necessity that LGG-management attempts to save function.