In patients with endoscopically resected T1 disease (40 patients in 2 years) if available, we will stain the initial endoscopic tumor specimen, as well as any subsequent specimen obtained at each routine 3(+/- 2) month interval endoscopy.

In patients who undergo surgery as their primary therapy, serum will be obtained at the time of surgical resection, and at each subsequent long-term disease status follow-up visit every 4 (+/- 2) months.

a serum sample will be obtained : 1) prior to initiation of therapy, 2) following the completion of induction chemotherapy, 3) at the time of surgical resection, and 4) at each subsequent long-term disease status follow-up visit every 4 (+/- 2) months. The availability of tissue for staining will determine whether or not patients are evaluable for Group 3.

Tissue mesothelin staining at the time of the initial endoscopy, and of any subsequent biopsy specimen during the endoscopic screening period.

Serum mesothelin level at the time of initial endoscopy, and at each subsequent endoscopy for two years or until disease recurs.

The investigators will examine the range and variability in the percentage on cells stained (for TM) and in the absolute value (for SM). TM expression is commonly analyzed in a binary fashion, with 25% of cells stained indicating positive expression (per standard pathological guidelines for tissue staining)

The investigators will explore the optimal cut point that defines positive expression. Serum mesothelin (SM) will be measured and analyzed whenever possible on a continuous scale.

First, the investigators will use two-sample t-tests to determine whether the initial responders to induction chemotherapy (defined as >30% decrease in SUV at the repeat PET) differ from non-responders in their 1) baseline (pre-induction) SM value, and 2) percent change in SM value between pre-induction and mid-induction (after 2 cycles) evaluations.

The investigators will take the following steps in order to assess the ability of SM at the time of resection with curative intent to predict disease recurrence: 1) we will examine the association between SM and the risk of recurrence by fitting a Cox proportional hazards model (after appropriate transformation of the SM value and checking of the PH assumption

The investigators will Wilcoxon test to investigate the association between serum mesothelin and tissue mesothelin at each time point where both markers are evaluated, and will further attempt to obtain an aggregate measure of this correlation using clustered Wilcoxon test, which accounts for multiple measurements per patient (14).