Does Immunohistochemistry Provide Additional Prognostic Data in Gastrointestinal Stromal Tumors?

Abstract

<b>Background:</b> To investigate the predictive and prognostic effects of clinicopathologic and immunohistochemical(IHC) features in patients with gastrointestinal stromal tumours (GISTs). Materials and <br/><b>Methods</b>: Fifty-sixpatients who were diagnosed with GIST between 2002 and 2012 were retrospectively evaluated. Relationshipsbetween clinicopathologic/immunohistochemical factors and prognosis were investigated. <br/><b>Results</b>: Medianoverall survival (OS) of the whole study group was 74.9 months (42.8-107.1 months), while it was 95.2 monthsin resectable and 44.7 months in metastatic patients respectively (p=0.007). Epitheliolid tumor morphology wassignificantly associated with shortened OS as compared to other histologies (p=0.001). SMA(+) tumours weresignificantly correlated with low (<10/50HPF) mitotic activity (p=0.034). Moreover, SMA(+) patients tendedto survive longer and had significantly longer disease-free survival (DFS) times than SMA (-) patients (37.7months vs 15.9 months; p=0.002). High Ki-67 level (≥30%) was significantly associated with shorter OS (34 vs95.2 months; 95%CI; p=0.001). CD34 (-) tumours were significantly associated with low proliferative tumours(Ki-67<%10) (p=0.026). Median PFS (progression-free survival) of the patients who received imatinib was 36months (27.7-44.2 months). CD34 (-) patients had significantly longer PFS times than that of negative tumours;(50.8 vs 29.8 months; p=0.045). S100 and desmin expression did not play any role in predicting the prognosisof GISTs. Multivariate analysis demonstrated that ≥10/50HPF mitotic activity/HPF was the only independentfactor for risk of death in GIST patients. <br/><b>Conclusions</b>: Despite the negative prognostic and predictive effect ofhigh Ki-67 and CD34 expression, mitotic activity remains the strongest prognostic factor in GIST patients. SMApositivity seems to affect GIST prognosis positively. However, large-scale, multicenter studies are required toprovide supportive data for these findings.