Predicting metastatic risk of gastrointestinal stromal tumors:: Role of cell proliferation and cell cycle regulatory proteins

Gastrointestinal stromal tumors (GIST) are a heterogeneous group of neoplasms whose biologic behavior is difficult to predict. The aim of this study is to evaluate the prognostic value in GIST of some oncoproteins involved in regulation of cell proliferation. Tumor size, mitosis, necrosis, and p53, c-myc, and bcl-2 protein expression of 32 GIST were studied. Proliferative index was assessed with Ki67. The 32 cases were grouped into the following clinical categories: (1) clinically benign (BN) were defined as disease-free survival greater than 3 years (n=10); (2) clinically malignant (MN) in which local recurrence or metastasis occurred regardless of the follow-up time (n=15); and (3) clinically indeterminate (LD) owing to follow-up <3 years without metastasis or local recurrence (n=seven). Discriminant analysis was used to allocate any tumor to one of the two prognostic groups (BN or MN). In univariate analysis all six factors studied above proved to be of significant prognostic value. Using a multivariate stepwise discriminant analysis to take into account the interrelationship between factors, we found that c-myc expression was the most important prognostic factor, followed, in order of statistical weight, by size and Ki67. These were combined to define a discriminant score ([10.75 x c-myc] + [0.39 x size] + [0.078 x Ki67]-15.54=score), which was capable of correctly identifying tumors in our series whose known clinical behavior was BN or MN in 92% of the cases. The classification score was applied subsequently to the seven clinically ID cases: Three (42.9%) were predicted as BN, and four (57.1%) were predicted as MN. Both expression of oncoprotein c-myc and the proliferative index provide prognostic information in GIST, in addition to morphologically established prognostic factors such as size. These factors in a discriminant analysis proved to be useful for the clinical classification of GIST into BN or MN and to predict the clinical outcome of clinically ID tumors.