Prognostic predictors of gastrointestinal stromal tumors: a multi-institutional analysis of 102 patients with definition of a prognostic index

Abstract

Aims: The aim of this study is to identify prognostic factors influencing survival in patients with gastrointestinal stromal tumors (GISTs) and to identify a mathematical model that can predict lifetime expectation.

Methods: One hundred and two patients with GISTs, were followed retrospectively for a median period of 32 months (from 1 to 82 months). Complete follow-up data were available in 72 cases. All tumors were surgically resected and examined by conventional light microscopy, immunohistochemistry and image analysis. The tumors' location, size, histologic characteristics, immunophenotype, proliferative activity index (assessed by proliferating cell nuclear antigen (PCNA) and Ki-67 immunoreactivity) and the apoptotic markers bcl-2 and bax, were considered as potential prognostic factors and were correlated with patient survival.

Results: Tumor size >8 cm (p<0.03), presence of necrosis (p<0.02), number of mitoses >5/10 HPF (p<0.01), metastasis (p<0.001), and PCNA index >10% (p<0.004) were significant predictors of poor survival. Bcl-2 protein (p<0.0007) was a favorable prognostic indicator. If all tumors were treated as of uncertain malignant potential, the following mathematic model named GISTs Prognostic Index (GPI), could be formed by the linear regression technique: GPI exp=(49.6 months−Status of metastasis×22.9185−Size in cm×0.6801+bcl-2 expression%×0.2569) (r²=0.67) (Prob>F=0.0001).

Conclusions: Tumors' size, necrosis, mitoses, metastasis and PCNA index are independent poor prognosticators, while bcl-2 protein is associated with favorable prognosis. An interesting equation for survival in patient with GISTs has been reported.

Introduction

Stromal tumors of the gastrointestinal tract (GISTs) are uncommon visceral neoplasms, derived from connective tissue elements and located along the entire length of the gut (60% in the stomach, 30% in the small intestine, 10% elsewhere).¹ During the past three decades there has been remarkable controversy regarding their nomenclature, exact cellular origin, and criteria used for diagnosis of malignancy and prognosis.² The majority correspond to a heterogeneous group of lesions that have as a common denominator an immature proliferation of epithelioid or spindle cells arising from the gastrointestinal muscle layer and showing partial or incomplete myoid, neural, ganglionic or mixed features of differentiation.¹

The cell origin for GISTs is not fully understood, but resemblance of the neoplastic cells to interstitial cells of Cajal, expression of some smooth muscle markers and occurrence outside of the GI-tract, suggest origin from multipotential cells that can differentiate into Cajal and smooth muscle cells.3., 4. The current World Health Organization classification refers to these tumors collectively as smooth muscle/stromal tumors.⁵ The term ‘gastrointestinal stromal tumors’ is being applied to cover all such mesenchymal tumors of the gastrointestinal tract, which could be neither neurogenic nor of smooth cell origin.³ The designation of GIST was used in the present study, as it has been accepted as an umbrella term to cover all mesenchymal tumors, although recent evidence indicates that GISTs may alternatively be defined as c-kit (CD117, stem cell factor receptor)-positive mesenchymal neoplasms.⁶

Criteria claimed to predict the biological behavior of GISTs remain vague and do not even permit accurate discrimination between benign and malignant lesions. Several gross and histological parameters such as tumor's size, location, necrosis, histological grade and mitotic count have been reported to influence the course of the disease.2., 4. In addition, parameters pertaining to the tumor biology such as immunophenotype and immunohistochemical assessment of proliferation index has been applied with encouraging results.7., 8. Also mutations on c-kit gene were reported recently as a potential marker of poor prognosis but the results are still controversial.9., 10., 11.

It has been argued that tumor growth influences survival in various human malignancies. Tumor growth depends on two distinct pathways: cell proliferation and apoptosis.¹² The potential of apoptosis can be at least in part evaluated by the comparative immunohistochemical assessment of apoptosis-related proteins with either an enhancing or an inhibiting role (i.e. bax and bcl-2 proteins, respectively).¹³

The bcl-2 expression has been studied in GISTs,¹⁴ but the significance of its expression in relation to prognosis remains uncertain. The demonstration of high expression of bcl-2 in these lesions suggest that reduced rate of cell death may be the mechanism responsible for the accumulation of cells and tumor formation but the value of expression of bcl-2 protein as prognostic marker in GISTs remains unknown.

In the current investigation, we intend to detect the prevalence and confirm the prognostic significance various clinicopathological predictors of survival in patients with such neoplasms, as well as to investigate the prognostic significance of tumors' proliferative fraction and apoptosis-related proteins' immunoreactivity in a well documented series of GISTs.