Adenocarcinoma of the esophagus, or GEJ, has a poor prognosis. Early lesions [i.e. high grade dysplasia (HGD) or T1-carcinoma] are potentially curable. Local endoscopic therapies are promising treatment options for superficial lesions; however, for deeper lesions, surgical resection is considered to be the treatment of choice. To contribute to therapeutic decision-making, we retrospectively analysed the outcome of transhiatal esophagectomy in 120 patients with pathologically proven HGD (n=13) or T1-adenocarcinoma (n=107) of the distal esophagus or gastro-esophageal junction (GEJ). Tumors were subdivided into six different depths of invasion ('T1-mucosal' m1-m3, 'T1-submucosal' sm1-sm3), and the frequency of lymphatic dissemination and time to locoregional and/or distant recurrence were analysed. Only one of the 79 T1m1-3/sm1 tumors (1%) showed lymph node metastases as compared with 18 out of 41 T1sm2-3 tumors (44%). There was a significant difference in recurrence-free period between T1m1-m3/sm1 versus T1sm2-sm3 tumor patients (P log rank <0.0001), with 5-year recurrence-free percentages of 97% and 57%, respectively. In multivariate analysis including age, gender, tumor differentiation grade, N-stage and depth of invasion, only N-stage was an independent prognostic factor for recurrence-free period (hazard rate=5.9, 95% CI 1.7-20.7). However, if N-stage was excluded from analysis, only depth of invasion (T1sm2-3 versus T1m1-m3/sm1) was an independent prognostic factor for recurrence-free period (hazard rate=7.5, 95% CI 2.0-27.7). These data indicate that T1m1-m3/sm1 adenocarcinomas of esophagus or GEJ show a very low risk of lymphatic dissemination and are therefore eligible for local endoscopic therapy. After transhiatal surgical resection, almost half of the patients with T1sm2-sm3 lesions develop recurrent disease within 5 years, and therefore need additional therapy to improve survival.

SUV decrease 14 days after the start of chemoradiotherapy was significantly associated with histopathologic tumor response, but its accuracy in detecting nonresponders was too low to justify the clinical use of FDG-PET for early discontinuation of neoadjuvant chemoradiotherapy in patients with potentially curable esophageal cancer.

Survival after recurrent EC in patients who undergo FU without routine imaging after esophagectomy is approximately 3 months and has not improved over the past 18 years.

Barrett's esophagus (BE) is associated with increased risk of esophageal adenocarcinoma (EAC) and characterized by replacement of normal esophageal squamous epithelium by columnar epithelium. These alterations are also reflected in changes in the protein-expression profiles of the cell types involved. To separately investigate the proteomes of selected cell-types we combined laser-capture microdissection (LCM) and liquid chromatography-mass spectrometry (LC-MS). Aims were to determine the sensitivity, specificity, and technical reproducibility of the sampling method, and the biological variability within and between biopsies and patients. Frozen biopsies were cryo-sectioned, samples of around 2000 epithelial or stroma cells microdissected, digested and measured by Orbitrap LC-MS. Proteins were then identified by MS/MS database search and quantified by label-free analysis. An average of 366 protein-groups were identified per sample, and more protein-groups were found in epithelial samples than in stromal samples (442 vs 301, p < 0.0001). Altogether, 1254 distinct protein-groups were found, 289 and 88 of them significantly more often in epithelial and stroma samples, respectively. We assessed five different types of reproducibilities (run-to-run, intrabiopsy, biopsy-to-biopsy, experiment-to-experiment, and patient-to-patient) for protein identification and protein quantification. Reproducibility of protein identification ranged from 78 to 57%, and standard deviation of protein quantification was on patient-to-patient level four times higher than for run-to-run. We conclude that sampling around 2000 cells requires groups of 32 samples to detect significant, over 10-fold differences in protein abundances and thus creates a successful compromise between throughput and quality of results. We therefore believe that this method is suitable for investigating protein-expression profiles during carcinogenesis.