Shantel Hebert-Magee, MD¹, Faisal Mukhtar, MD¹, Mohamad Eloubedi, MD, MPH², Isam Eltoum, MD, MBA¹

¹Departments of Pathology, ²Gastroenterology, University of Alabama at Birmingham, Birmingham, Alabama

Introduction: Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is an established diagnostic modality that reports high success in accurately diagnosing pancreatic adenocarcinoma. However, most EUS-FNA studies are traditionally restricted to single institutional data regarding diagnostic accuracy, frequently biased by combining clinical follow-up and surgical pathology as the ‘collective’ gold standards. To date, a meta-analysis has not been performed to critically evaluate the diagnostic accuracy of the EUS-FNA of pancreatic ductal carcinoma and determine the sources of heterogeneity among different studies. A meta-analysis was performed to determine if the variability of the gold standard and other sources of heterogeneity significantly impact diagnostic accuracy, following the cochrane collaboration guidelines.

Materials and Methods: We systematically searched MEDLINE and SCORPUS to identify the potential studies published between 1994 and 2011. The selection criteria were limited to studies consisting of consecutive patients who underwent EUS-FNA for suspicious solid pancreatic lesions and supported the construction of a 2 μ 2 contingency table for analysis. Two authors independently extracted data and entered it into a Microsoft Excel spread sheet, with the following parameters: Year, number of cases, TP, TN, FP, FN, presence of cytopathologist, sample adequacy, and the reference standard used. Computational analysis of the data was performed with the Cochrane Diagnostic Accuracy protocol using the Review Manager (RevMan5) and Statistical Analysis Software (SAS / STAT). RevMan uses a modified quality assessment tool for diagnostic accuracy studies, named QUADAS, to assess the quality of each study. Analysis and data presentation were performed as suggested by the Cochrane Diagnostic Accuracy. Covariates (assessing the presence / absence of a cytopathologist and the reference standard used) were evaluated for variations, utilizing the comparative ROC model and Forest plot summaries.

Results: We identified 783 (465 MEDLINE, 218 SCORPUS) published articles. Only 41 studies met the inclusion / exclusion criteria and included 3492 patients with 2119 pancreatic adenocarcinomas. The sensitivity ranged from 0.33 to 1.00, the specificity from 0.88 to 1.00. The lower sensitivity was consistent with the lack of onsite cytopathologist evaluation or sampling error. The overall diagnostic accuracy was found to be as follows: 0.91(CI : 0.87 – 0.94) for sensitivity and 1.00 (CI:0.99 – 1.00) for specificity. The gold standard was histology (n = 4), clinical course (n = 3), and histology and / or clinical (n = 34). Onsite evaluation by the cytopathologist was present (n = 18), sometimes (n = 4), not indicated (n = 9), and not present (n = 10). By receiver operating characteristic (ROC) curve, EUS-FNA performed with the assistance of a cytopathologist was greater than that of the assessment performed without onsite assistance. Results of the summary receiver operating characteristic (SROC) curve analysis, evaluating the reference standard, suggested lower diagnostic accuracy when the reference standard was the histological ‘gold standard’.

Conclusions: Our study shows that evaluating diagnostic accuracy in EUS-FNA of the pancreas using clinical follow-up or combined clinical and histological follow-up introduced a bias that falsely elevated the diagnostic accuracy. Our met-analysis also supports the previously published literature, that EUS-FNA evaluation for pancreatic adenocarcinoma is highly accurate and better when rapid onsite evaluation (ROSE) is performed.

Brittany Holmes, MD¹, Ralph Hruban, MD¹, Christopher Wolfgang, MD, PhD², Syed Ali, MD¹

¹Department of Pathology, The Johns Hopkins Hospital, Baltimore, Maryland; ²Department of Surgery, The Johns Hopkins Hospital, Baltimore, Maryland

Introduction: Formerly known as lymphoplasmacytic sclerosing pancreatitis (LSP), autoimmune pancreatitis (AIP) is an inflammatory, IgG4-associated condition that is primarily treated by non-surgical methods. However, the clinical and radiological features of AIP overlap with the more prevalent and often fatal pancreatic ductal adenocarcinoma, thus creating diagnostic confusion between the two entities on fine needle aspiration (FNA).

Materials and Methods: The pathology archives from a large tertiary care center were searched for a period of 26 years (1985 – 2011) for all surgical pancreatic specimens diagnosed as AIP or LSP, in patients with a prior FNA of the pancreas. FNA material was obtained by endoscopic ultrasound-guided FNA, and direct smears were prepared and stained with Diff-Quick as well as wet fixed for Papanicolaou staining. The surgical and cytopathological data were correlated.

Results: A total of 20 FNAs from 17 patients were identified. There were 12 males and five females (M : F, 2.4 : 1) with an age range of 29 to 76 years (mean 60 years). Imaging studies demonstrated a pancreatic mass or ill-defined density suspicious for malignancy in 65% (n = 11) of the patients with a size range of 2.1 to 5.2 cm (mean 3.5 cm, n = 4) and peripancreatic lymphadenopathy in 12% (n = 2). Of the 20 aspirates, 5% (n = 1) were diagnosed as adenocarcinoma, 5% (n = 1) were suggestive of a mucinous cystic neoplasm, 50% (n = 10) were atypical, 25% (n = 5) were benign, and 15% (n = 3) were scant or non-diagnostic. Of the 10 aspirates diagnosed as atypical, 70% (n = 7) were described as having rare or scattered atypical ductal cells, while 10% (n = 1) were markedly atypical, 10% (n = 1) were suspicious for malignancy, and 10% (n = 1) could not exclude a neuroendocrine neoplasm. Pancreatic intraepithelial neoplasia (PanIN) was present in the surgical specimens corresponding to 58% (n = 7) of the 12 aspirates diagnosed as atypical or neoplastic. Beginning in 2008, immunohistochemical labeling for IgG4 was performed, with nine out of ten demonstrating increased numbers of positive plasma cells.

Conclusions: On FNA, AIP more often leads to an ‘atypical’ cytopathological interpretation due to the presence of significant epithelial atypia. Thus, caution is warranted to avoid overdiagnosis on FNA, even when a mass lesion is reported by the imaging studies. Performing IgG4 immunohistochemistry on the cytologic cell block material may increase the diagnostic accuracy of AIP on fine needle aspiration.

Gillian Levy, MD¹, Alexander Finkelstein, DO, MD¹, Guoping Cai, MD¹, David Chhieng, MD¹, Uzma Siddiqui, MD², Harry Aslanian, MD², Malini Harigopal, MD¹

¹Pathology, Yale University School of Medicine, New Haven, Connecticut; ²Medicine and Digestive Diseases, Yale University School of Medicine, New Haven, Connecticut

Introduction: Endocrine neoplasms of the pancreas (PENs) are relatively uncommon, accounting for 1 – 2% of all pancreatic neoplasms. A diagnosis can be rendered cytologically for PENs with characteristic cytomorphological features. However, PENs may infrequently display unusual features, such as, oncocytic changes, mucin accumulation, intracytoplasmic inclusions, and clear cell or vacuolated cytoplasm. These morphological variations may pose a diagnostic challenge. Here we report three cases of PENs that had extensive cytoplasmic vacuolization, in which the differential diagnosis may include pancreatic adenocarcinoma, metastatic clear cell renal cell carcinoma, and adrenocortical carcinoma.

Materials and Methods: Three cases of PENs were diagnosed by endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) at our institution, between 2007 and 2010. The biopsy was performed in the endoscopy suite and a cytopathologist was available for a rapid onsite evaluation. The smears prepared from the aspirates were stained with Diff-Quik^® or Papanicolaou techniques for cytological evaluation. The additional aspirate material was fixed in CytoRich Red and processed for cell-block. Immunohistochemical studies were performed on the cell-block sections using a panel of antibodies including chromogranin, synaptophysin, CD56, AE1 / AE3, vimentin, CD10, inhibin, and Melan A. The final diagnosis was rendered based on cytomorphological and immunophenotypic features.

Results: The patients included one male and two females, aged 41, 46, and 61, respectively. The tumors were all located in the body / tail of the pancreas with sizes of 1.9 cm, 3.5 cm, and 14 cm each. One patient received a distal pancreatectomy. while the other two were inoperable due to documented liver metastasis or large tumor size with local invasion. Overall, the aspirates were moderately cellular and consisted of single and dyscohesive clusters of relatively uniform neoplastic cells. Extensive cytoplasmic vacuolization was seen in the tumor cells in all cases. The endocrine nature of neoplastic cells was confirmed by the positivity for chromogranin, synaptophysin, or CD56.

Conclusions: Extensive cytoplasmic vacuolization can rarely be seen as a cytomorphological feature in PENs, which can become a diagnostic pitfall for pancreatic adenocarcinoma and other metastatic tumors. Recognition of this rare occurrence with adjunct immunohistochemical studies will help render a correct diagnosis.

Faisal Mukhtar, MD, Shantel Hebert-Magee, MD, Isam Eltoum, MD, MBA

Department of Pathology, University of Alabama at Birmingham, Birmingham, Alabama

Introduction: Primary small cell carcinoma (SCC) of the pancreas is extremely rare, with approximately 35 cases reported in the literature. The aim of this study is to review a large center experience with SCC and if possible determine whether the origin of the tumor is primary or metastatic.

Materials and Methods: This a retrospective correlation study in which we reviewed electronic medical records, cytology, and histology reports of all endoscopic ultrasonography-guided fine needle aspiration (EUS-FNA) of the pancreas between 2000 and 2010, trying to find cases with a diagnosis of small cell carcinoma of the pancreas. We reviewed the history, imaging, and pathologica; findings to determine whether the lesion was primary or metastatic. A case was considered metastatic when the patient was known to have had a mass elsewhere, with immunostains supporting the origin of the tumor.

Results: Based on the search through medical records, we identified 2445 pancreatic FNAs during the study period from 2000 to 2010. One hundred and thirty-four cases were endocrine neoplasms with eight (6%) patients identified with small cell carcinoma (poorly differentiated endocrine carcinoma). Four of the eight patients were male, and four of the patients were known smokers, with one patient being a passive smoker. Median age at diagnosis was 62 years (range 39 – 81 years). Two of the patients were African American and six were Caucasians. The patients presented with vague abdominal pain (n = 3), jaundice (n = 2), and were asymptomatic (n = 3). The tumor size ranged from 3.0 cm to 8.0 cm. Four patients had a tumor in the head of the pancreas, one in the body of the pancreas, and the rest had tumors that were peripancreatic. Five patients had primary lung carcinoma, one had a primary cancer in the cervix, and the other two had primary lesions that could not be identified, making the prevalence of SCC 1.5% for all endocrine neoplasms. In all patients, the cytological features and endocrine markers were diagnostic of small cell carcinoma. With the exception of one patient, all patients died of the disease with survival ranging from 5 to 32 months (median 11 months) after the diagnosis.

Conclusions: SCC whether primary or metastatic to pancreas carries a grave prognosis. Primary pancreatic SCC is extremely rare. The majority of the SCC of pancreas is metastatic, therefore, a search for the origin is required, particularly given the recent reports, which show a comparatively good response of primary SCC to surgical resection.

Armanda Tatsas, MD¹, Joseph Herman, MD, MSc², Ralph Hruban, MD¹, Daniel Laheru, MD³, Christopher Wolfgang, MD, PhD⁴, Elliot Fishman, MD⁵, Syed Ali, MD¹

¹Departments of Pathology, ²Radiation Oncology and Molecular Radiation Sciences, ³Oncology, ⁴Surgery, ⁵Radiology, The Sol Goldman Pancreatic Cancer Research Center, The Johns Hopkins Hospital, Baltimore, Maryland

Introduction: Fine needle aspiration (FNA) of the pancreas and biliary brushing (BB) are commonly used techniques to preoperatively evaluate pancreaticobiliary disease. Patients with pancreatic and biliary tract neoplasms are often treated with radical surgery, or referred directly for chemo- and / or radiation therapy. Given the importance of accurate interpretation of the cytological specimens in these patients, we sought to examine the rate of diagnostic disagreement in cases of pancreatic FNA and BB referred to our institution for confirmation of diagnosis.

Materials and Methods: A retrospective search of the cytopathology archives at a large tertiary care center revealed 405 pancreatic FNA (88.7%) and BB (11.3%) cases submitted to our institution for confirmation of diagnosis from January 1, 2009 to December 31, 2010. A scanned copy of the diagnostic report from the originating institution was examined and that diagnosis was compared to the second opinion diagnosis rendered at our institution. Cases with incomplete or missing scanned diagnostic reports were excluded from the study. Each case was categorized as ‘no diagnostic disagreement,’ ‘minor diagnostic disagreement,’ or ‘major diagnostic disagreement.’ Minor or major disagreements were established based on a 1 or 2+ step discrepancy, respectively, on a scale that classified diagnoses as ‘benign, atypical, suspicious for malignancy / neoplasm, and malignant / neoplasm.’ Cases that did not fit readily into this classification scheme were assessed on an individual basis.

Results: In 318 (78.5%) cases, the second opinion diagnosis correlated closely with the original diagnosis, indicating no diagnostic disagreement. Minor diagnostic disagreements were found in 76 cases (18.8%). Major diagnostic disagreements were found in 11 cases (2.7%), nine of which had follow-up surgical pathology or cytopathology specimens. The second opinion diagnosis was supported in six of the nine cases with follow-up specimens. The most serious error detected was one case diagnosed as ‘benign, non-specific changes,’ at the originating institution, with a second opinion diagnosis of ‘adenocarcinoma.’ The lesion was resected and found to be a poorly differentiated infiltrating adenocarcinoma, thus the patient was spared a repeat procedure at our institution prior to definitive treatment.

Conclusions: The rate of major diagnostic disagreements in pancreatic FNA and BB samples in our study is 2.7%. Although this rate is slightly lower than the reported rate of major discrepancies in all cytopathologies, this still represents a significant discrepancy rate, given the serious clinical implications of these diagnoses. Our study supports the importance of mandatory second opinion review in pancreaticobiliary cytopathology as part of the ongoing efforts to improve patient safety and quality of care.