Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is the most sensitive and specific test for establishing a tissue diagnosis for many gastrointestinal malignancies; however, cytologic morphology alone may not be definitive for subsets of tumors. Our aim was to quantify the impact of the broad application of flow cytometry (FC) and immunohistochemistry (IHC) on EUS-FNA diagnostic yield. A retrospective chart review was performed on EUS procedures at a tertiary referral, academic medical center. All EUS-FNA cases performed over a 21-month period were examined. Of 606 EUS procedures reviewed during the period of study, 264 utilized FNA. After pancreatic cyst cases were excluded, 235 EUS-FNA cases for 221 patients were selected for analysis. For cases with subsequent histological evaluation, including the subset utilizing FC/IHC, the sensitivity of EUS-FNA was 89%, specificity was 100%, and accuracy was 91%. One quarter (58/235, 25%) of the tissue specimens underwent further testing by FC/IHC. There were 48 definitive diagnoses made in the subset utilizing FC/IHC. In 20 of the 48 diagnoses (42%), FC/IHC was deemed critical to the diagnosis, and without FC/IHC testing in those cases, the overall sensitivity and accuracy of EUS-FNA would be reduced to 74 and 77%, respectively. FC/IHC allowed for six diagnoses rarely or not previously described by EUS-FNA. Application of FC/IHC improves characterization predominantly for nonadenocarcinoma tumor subtypes and may lead to a diagnostic result for tumors not previously characterized by EUS-FNA. With an adequate tissue sample, broad application of FC/IHC increases the diagnostic yield of EUS-FNA.

A 53-year-old woman with a history of acute lymphocytic leukemia (ALL) was admitted to an outside hospital in Florida. She had received a matched unrelated donor allogeneic peripheral blood stem cell transplant 13 months prior and was on steroid treatment for graft-versus-host disease involving skin, eyes, and liver. She presented with symptoms of bacteremia, subacute left cerebellar infarct with neurological symptoms, and depression secondary to her medical condition. Three sets of blood cultures (Bactec; Becton, Dickinson and Company, Sparks, MD), collected at an outside hospital, were positive (aerobic bottles) for a Gram-negative rod that tested susceptible (MIC Ͻ 3 g/ml) to piperacillin-tazobactam but intermediate or resistant to all other antibiotics as determined by a MicroScan Gram-negative panel (Siemens Healthcare Diagnostics, Inc., West Sacramento, CA) and Clinical and Laboratory Standards Institute (CLSI) interpretations for nonEnterobacteriaceae. The patient was treated with piperacillintazobactam for 9 days without improvement. Susceptibility testing was repeated at the outside hospital on a follow-up culture and showed an MIC of 64 g/ml for piperacillin-tazobactam but no change in the results for the other antibiotics. The patient was subsequently transferred to our tertiary care medical center for further management. Blood cultures (BacT/ Alert; bioMérieux, Durham, NC) were redrawn on admission and again grew a Gram-negative rod in the aerobic bottle that appeared thin and filamentous on initial Gram staining of the broth (Fig. 1). The organism produced pale yellow colonies on sheep blood agar and chocolate agar after overnight incubation at 35°C in 5% CO 2 . Growth on MacConkey agar was poor. The organism was nonmotile, catalase positive, and oxidase positive. A cerebrospinal fluid specimen collected 4 days after admission also showed Gram-negative rods, some of which were dumbbell shaped and grew the same organism as the blood cultures.