Value of cytology in detecting intestinal metaplasia and associated dysplasia at the gastroesophageal junction.

Tissue sampling is essential for detecting intestinal metaplasia in the distal esophagus (Barrett's esophagus), because symptoms and endoscopy are not reliable in making this diagnosis. The utility of cytology in this process is unknown. All adult patients having elective upper gastrointestinal endoscopy over a 6-month period were invited to participate in a prospective study whose aim was to determine the prevalence of intestinal metaplasia in the distal esophagus in an adult population with diverse upper gastrointestinal symptoms. Clinical data and endoscopic findings were recorded. Brush cytology and biopsy specimens were obtained from both sides of the apparent squamocolumnar junction. The cytology specimens were processed routinely, stained with the Papanicolaou technique, and reviewed blinded to the clinical information and the histological findings in the corresponding biopsy specimens. One hundred fifty-five patients (81 women, 74 men; 137 whites, 11 blacks, 7 others; mean age, 52 years) were included. Glandular epithelium/cells were present on both histology and cytology in 147 specimens. Thirty-two patients (22%) showed intestinal metaplasia on histology. Of the cytology specimens from these 32 patients, 6 contained definite goblet cells (19%), 7 probable goblet cells, and 19 no goblet cells. Goblet cells and probable goblet cells were observed on cytology in 7 and 11 additional specimens, respectively. One was from a patient known to have intestinal metaplasia in the esophagus. Follow-up endoscopy with biopsy was performed in two of these latter 18 patients and did not show intestinal metaplasia. One case of high-grade dysplasia, two of low-grade dysplasia, and three indefinite for dysplasia were diagnosed on histology. All three cases of dysplasia were also identified on cytology. The three indefinite cases on histology were considered reactive in two and unremarkable in one on cytology. Low-grade dysplasia was diagnosed on cytology alone on two cases. Follow-up endoscopy with biopsy was performed in one patient, and low-grade dysplasia was found. Cytology using the Papanicolaou stain is not as sensitive and specific as histology for detecting intestinal metaplasia in the distal esophagus. However, it may be at least as useful as tissue sampling in detecting dysplasia.

ThinPrep processing of endoscopic brushing specimens.

To evaluate ThinPrep (Cytyc, Marlborough, MA) processing of endoscopic brushing specimens for cytologic examination, ThinPrep slides and direct smears of 29 gastrointestinal (GI) and 22 bronchial brushings were compared. Clinicians prepared the direct smears. The brush was then immersed in CytoLyt (Cytyc) and one ThinPrep slide made. All cases had corresponding biopsies. Smears and ThinPrep slides were screened and reviewed independently. Screening time per case was recorded. All slides were evaluated for cellularity, quality, cellular preservation, and quantity of diagnostic cells. A diagnosis was rendered for each case. Cytologic and histologic diagnoses were correlated. Follow up was obtained for cases with discrepant histologic and cytologic diagnosis. Twenty-three brushings were from esophagus, 5 stomach, 1 duodenum, and 22 lung. An average of 3.6 direct smears (range 2-6) was made for each case. Average screening time per case was 12 minutes for GI direct smear, 15 minutes bronchial direct smear, 4 minutes GI ThinPrep slide, and 9 minutes bronchial ThinPrep slide. ThinPrep slides were superior to direct smears in cellularity, quantity of diagnostic cells, and quality of slides. ThinPrep slides and direct smears showed comparable cellular preservation. The sensitivity of detecting malignancy by biopsy, direct smears, and ThinPrep slides was 81%, 75%, and 75%, respectively. One false-positive diagnosis was made on cytology with both direct smear and ThinPrep slide, a case with radiation atypia. In conclusion, ThinPrep slides are at least comparable to direct smears in cytologic examination of brushings. However, false-positive diagnosis is a possible potential pitfall.