Role of lymphovascular invasion and immunohistochemical expression of IMP3 in the risk stratification of superficially invasive pT1 esophageal adenocarcinoma

Abstract

BackgroundA problem in the management of patients with Barrett's esophagus-related pT1 esophageal adenocarcinoma is to distinguish those who should be treated conservatively (endoscopic mucosal resection and/or radiofrequency ablation) from those who require esophago-gastrectomy. Recently, lymphovascular invasion (LVI) has emerged as one of the best predictors of regional lymph node metastasis (LNM) and recurrence-free survival (RFS) in pT1 EAC. However, LVI may be underestimated, both because of interobserver variability and incomplete sampling. The aim of our study was to correlate the presence of LVI, with the immunohistochemical expression of IMP3 in pT1 EAC and assess their role in further stratifying these lesions into high and low risk groups based on the potential for lymph node metastasis and poor outcome. DesignDepth of invasion, assessed in five sublevels (m2, m3, sm1, sm2, and sm3), LVI, and expression of IMP3 were studied in 30 patients who underwent esophagogastrectomy for pT1 EAC (2001–2010) at Hartford Hospital, and correlated with LNM and RFS. IMP3 was considered positive when expressed in >50% of the malignant cells with an intensity of stain of 2–3+. ResultsTen of 18 (55.5%) cases with IMP3 expression demonstrated LVI and 2/10 (20%) showed LNM and died of disease. In contrast, none of the 12 IMP3 negative cases showed LVI (p<0.004; 2-tailed Fisher exact test) or had LNM/DOD. ConclusionsIn pT1 EAC, (1) based on IMP3 expression, pT1 EAC may be divided into high risk (LVI+/IMP3+) and low risk (LVI−/IMP3−) categories. (2) Absence of IMP3 expression is associated with a significantly reduced risk of LVI (Negative Predictive Value: 100%). (3) Since identifying lymphovascular invasion and other morphological parameters is prone to significant inter-observer variation, IMP3 may be useful as an ancillary marker especially in these pT1 lesions in predicting their clinical behavior, the risk stratification and potentially on the type of treatment.