BACKGROUND. Prognostic variables for idiopathic (primary) osteomyelofi
brosis (IMF) are ill-defined because of the lack of large control stud
ies based on uniform diagnostic criteria. METHODS. A retrospective cli
nicopathologic study was performed on 250 consecutively recruited pati
ents (115 males and 135 females) with an established diagnosis of IMF.
In contrast to previous studies, the current study cohort encompassed
the full spectrum of initial to advanced stages of the disease proces
s according to laboratory data and particularly histology. Because of
the relatively high patient age on admission (median, 66.5 years), rel
ative survival rates with corresponding life expectancies and disease
specific life loss were calculated. Moreover, a classification and reg
ression tree (CART) analysis was performed to segregate the study pati
ents into subgroups with significantly different prognosis. RESULTS. A
nalysis of the life expectancy and the proportion of deaths attributab
le to IMF showed a global reduction in life expectancy of 31%. Further
calculation disclosed a consistently greater impact of disease in old
er patients. Age, hemoglobin level on admission, and leukocyte and thr
ombocyte counts remained as the most relevant parameters for prognosis
in multivariate consideration (CART analysis) and facilitated a clear
-cut separation into three risk groups. The life expectancy of low ris
k patients was approximately 10 times higher than that of high risk pa
tients (22.07 years vs. 2.25 years). CONCLUSIONS. These results are in
keeping with the assumption that features signaling bone marrow insuf
ficiency are associated with a worsening of survival. Generalization,
indicated by myeloid metaplasia, can occur at every stage, even in so-
called hypercellular phases of IMF. Conversely, myelofibrosis alone is
not necessarily predictive of poor survival. (C) 1997 American Cancer
Society.