Boolean searches can be executed based on the mention of any word or combination of words in admission notes, discharge summaries, radiology reports, and other documentation.

To meet HIPAA guidelines and insure patient confidentiality, all data was de-identified using an honest broker system. This study met the criteria for exemption of informed consent by the University of Pittsburgh Institutional Review Board. We identified patients who underwent esophagectomy for high grade dysplasia in the setting of Barrett’s esophagus between January 1993 and June 2007. The search terms used variations of Barrett’s, high grade dysplasia, adenocarcinoma of esophagus, and esophagectomy. For inclusion, subjects had to have a preoperative diagnosis of high grade dysplasia confirmed by the pathology department at the University of Pittsburgh Medical Center. Patients with a preoperative diagnosis of low grade dysplasia or invasive adenocarcinoma or who underwent esophagectomy for other indications were excluded. Cases were identified by retrospective review of preoperative pathology reports of biopsy specimens obtained at endoscopy. After identifying the cohort of patients undergoing resection, all available preoperative endoscopy, surgical, and radiology reports for each of the patients was reviewed.

Postoperative pathology reports were reviewed to determine whether the final pathologic diagnosis remained high grade dysplasia, was upgraded to adenocarcinoma, or was downgraded to low grade or no dysplasia.

In the 8 patients with a postoperative diagnosis of invasive cancer, the size of the tumor ranged from 0.3 cm to 5 cm, with the average 1.86 cm. The TNM staging of the tumors revealed 5 patients with T1bN0Mx, 1 with T1bN1M1, 1 with T3N1M1, and 1 with T3N1M0. The 4 patients with intramucosal cancer had tumor sizes ranging from 0.1 to 1.2 cm, with an average of 0.61 cm.

The 2 tumors with T3 staging postoperatively had tumor sizes of 4 cm and 5 cm. The patient with the 4 cm tumor had evidence of malignancy on a preoperative positron emission tomography – computerized tomography (CT) scan. On endoscopic ultrasound, this patient had multiple enlarged thoracic lymph nodes. The patient with the 5 cm tumor had a preoperative CT scan revealing a 3.1 cm mass with multiple mediastinal lymph nodes. This same patient had a preoperative barium esophogram suggestive of an esophageal stricture. Two other patients had preoperative findings suggestive of invasive disease. The patient with a 1.6 cm tumor staged as T1b had esophageal thickening up to 7 mm noted on a preoperative computed tomography scan and on preoperative endoscopy multiple esophageal nodules were noted. The patient with T1b staging and a 1.5 cm tumor had nodular, ulcerated lesions on endoscopy and a preoperative endoscopic ultrasound was suggestive of submucosal involvement (Table 2). None of the other patients with invasive cancer or intramucosal carcinoma had radiologic or endoscopic evidence suggestive of cancer on preoperative testing.

Despite a preoperative diagnosis of HGD 2 patients staged as T3 had radiologic and endoscopic evidence to suggest invasive cancer. Two patients with subsequent T1b staging postoperatively also had preoperative suspicion for malignancy. Thus, 4 patients with preoperative HGD had occult carcinoma detected postoperatively, for an occult incidence rate of 5.9% (4/68).

Several predictors of invasive carcinoma in the setting of HGD have been recognized. Nodular lesions in HGD have been shown to be at a higher risk for adenocarcinoma (15). A recent study analyzed pooled data from multiple studies, and showed that visible lesions at endoscopy are associated with a higher risk of submucosal invasion, although statistical significance was not reached (12).

Determining the true rate of occult adenocarcinoma in Barrett’s with HGD is important because it impacts on the recommendations for management. If the rate of malignancy were 40% or even higher, than the associated risk of mortality to adenocarcinoma would be substantial enough that surgery would be the optimal choice. However, if the rate is more on the order of 8% -12%, than the risk of surgery must be weighed against the risk of the operation, and the potential response to less invasive treatments such as endoscopic therapy, including mucosal resection or photoablative or radioablative treatment. Esophagectomy is a procedure with a mortality risk of 3% to 8%, and with risk for significant morbidity, even at the most experienced centers. In a lower volume center, these risks are higher (10,16). A recent study from the University of Pittsburgh reported a 30 day mortality of 0% for T1 cancer patients undergoing esophagectomy, so local expertise may affect the clinical approach (17). Multiple patient factors including patient age and health status must be considered when deciding on the management of patients with HGD.

A recent review of 1074 patients from 16 studies, concluded that endoscopic therapy including photodynamic therapy, argon plasma coagulation, or radiofrequency ablation, can eradicate Barrett’s disease and dysplasia, and were generally well tolerated (18). It is possible that endoscopic therapy might have been successful in the 4 patients in our cohort with T1a stage intramucosal disease.

One limitation of our study is the lack of standardized preoperative testing for the patients in our cohort. A lack of comprehensive preoperative testing may have contributed to a higher rate of occult cancer, by increasing patients in the group with no suspicion for invasive cancer. Three of the four subjects with occult invasive adenocarcinoma did not undergo radiologic assessment at our center. Because of our using deidentified data, we could not rereview the outside studies. However, these subjects had very small tumors without lymph node involvement, and the likelihood that they were truly occult is high. As with all retrospective studies, selection bias remains a concern, although we attempted to minimize bias by searching our electronic medical records using comprehensive inclusion and exclusion criteria.

The rate of invasive adenocarcinoma in association with HGD and Barrett’s in this series was 11.7% with 5.9% having occult adenocarcinoma. When analyzed based on the date of surgery, we did not find any significant difference in the rate of detection of postoperative adenocarcinoma in patients with HGD over time, indicating that rate of cancer detection did not change in more recent years with the advent of more modern endoscopic techniques and imaging.