Markov model-based estimation of individual survival probability for medullary thyroid cancer patients

The relatively benign, but occasionally rapidly fatal clinical course of medullary thyroid cancer (MTC) has raised the need for individual survival probability estimation. A retrospective study on 91 MTC clinical case histories with a mean follow-up of 6 years indicated prevalences of local, regional and distant residual tumor on primary care completion of 23%, 54% and 54%, respectively. Local, regional and distant relapses during follow-up occurred in 8%, 23% and 26% of the patients, with a cause-specific death in 26% of the cases. Prognostic factors statistically significantly influencing the cause-specific survival were selected by uni- and multivariate analysis. A Markov method-based model was developed for the estimation of individual time-dependent local, regional and distant relapse-free and cause-specific survival probability functions, with parameters numerically determined via a maximum likelihood procedure. These parameters include relative risk factors related to prognosticators, a residual or recurrent local/regional/distant tumor, and combinations of these entities. In multivariate studies, the patient s age and gender, the genetic basis of the dis-ease, lymph node involvement, the existence of a general symptom (diarrhoea) at presentation, and the dosage of external irradiation proved to be prognosticators. The cause-specific survival function of the study population indicated mean 5, 10 and 15-year survival probabilities of 69%, 62% and 58%.

Survival probabilities can be predicted for extrastudy cases provided that the same laws and principles govern the clinical course of these cases and those comprising the study. For individual survival probability estimation, a Pascal program (MEDUPRED) was written and is available on the home page of the National Institute of Oncology, Budapest (www.oncol.hu).