Non-ampullary-duodenal carcinomas: clinicopathologic analysis of 47 cases and comparison with ampullary and pancreatic adenocarcinomas.

Literature on non-ampullary-duodenal carcinomas is limited. We analyzed 47 resected non-ampullary-duodenal carcinomas. Histologically, 78% were tubular-type adenocarcinomas mostly gastro-pancreatobiliary type and only 19% pure intestinal. Immunohistochemistry (n=38) revealed commonness of 'gastro-pancreatobiliary markers' (CK7 55, MUC1 50, MUC5AC 50, and MUC6 34%), whereas 'intestinal markers' were relatively less common (MUC2 36, CK20 42, and CDX2 44%). Squamous and mucinous differentiation were rare (in five each); previously, unrecognized adenocarcinoma patterns were noted (three microcystic/vacuolated, two cribriform, one of comedo-like, oncocytic papillary, and goblet-cell-carcinoid-like). An adenoma component common in ampullary-duodenal cancers was noted in only about a third. Most had plaque-like or ulcerating growth. Mismatch repair protein alterations were detected in 13% (all with plaque-like growth and pushing-border infiltration). When compared with ampullary (n=355) and pancreatic ductal (n=227) carcinomas, non-ampullary-duodenal carcinomas had intermediary pathologic features with mean invasive size of 2.9 cm (vs 1.9, and 3.3) and 59% nodal metastasis (vs 45, and 77%). Its survival (3-, 5-year rates of 57 and 57%) was similar to that of ampullary-duodenal carcinomas (59 and 52%; P=0.78), but was significantly better than the ampullary ductal (41 and 29%, P<0.001) and pancreatic (28 and 18%, P<0.001) carcinomas. In conclusion, non-ampullary-duodenal carcinomas are more histologically heterogeneous than previously appreciated. Their morphologic versatility (commonly showing gastro-pancreatobiliary lineage and hitherto unrecognized patterns), frequent plaque-like growth minus an adenoma component, and frequent expression of gastro-pancreatobiliary markers suggest that many non-ampullary-duodenal carcinomas may arise from Brunner glands or gastric metaplasia or heterotopic pancreatobiliary epithelium. The clinical behavior of non-ampullary-duodenal carcinoma is closer to that of ampullary-duodenal subset of ampullary carcinomas, but is significantly better than that of ampullary ductal and pancreatic cancers. The frequency of mismatch repair protein alterations suggest that routine testing should be considered, especially in the non-ampullary-duodenal carcinomas with plaque-like growth and pushing-border infiltration.

Immunohistochemical Classification of Ampullary Carcinomas: Critical Reappraisal Fails to Confirm Prognostic Relevance for Recently Proposed Panels, and Highlights MUC5AC as a Strong Prognosticator.

Recently, immunohistochemistry-based classifications of ampullary carcinomas have been proposed (Ang and colleagues [PMID: 24832159]; Chang and colleagues [PMID: 23439753]). In this study, the prognostic value of Ang/Chang panel markers (CK20, MUC1, MUC2, CDX2) as well as other markers (CK7, MUC5AC, and MUC6) were tested on full-faced sections of 136 ampullary carcinoma resections with substantial (>5 mm) invasion. Immunohistochemistry was correlated with both histologic classification (intestinal [INT], pancreatobiliary [PB], or nontubular based on ≥3/5 observer agreement) and clinical outcome. No prognostic correlation was found with MUC1, CDX2, MUC2 or CK20 despite testing with different quantitative cutoffs. CK7 and CK20 were nonspecific. Ang classification had reasonable correlation with histologic subclassification of tubular cases as INT versus PB with high specificity but low sensitivity and ambiguous category was large (29%) and included also some classical cases. Prognostically, Ang classification approached but did not reach statistical significance, even when their large "ambiguous" group was eliminated and only tubular cases were analyzed (Ang-INT vs. Ang-PB; P=0.08). The Chang panel, in which the definition of the INT subcategory is not clearly defined, only marginally reached prognostic significance when tested as MUC1+/CDX2- versus MUC1-/CDX2+ and only by Wilcoxon test (P=0.0485) but 31% of the cases were "unclassifiable." The only individual marker that was found to have direct and strong correlation with the clinical outcome was MUC5AC (not used in the Ang or Chang panels), with statistically significant survival differences found with various cutoffs tested (for 20% cutoff, 5-y survival, 68% vs. 31%; P=0.0002). In addition, MUC5AC significantly stratified the histologically PB and INT cases (P=0.01 and 0.03, respectively), as well as Ang's ambiguous and Chang's unclassified cases (P=0.006 and 0.007, respectively). In conclusion, the widely used putative lineage markers, MUC1/MUC2/CK7/CK20/CDX2, do not seem to have direct/significant prognostic correlation either individually or in combination of Ang and Chang panels. Ang panel is helpful as an adjunct in determining the cell lineage with a few caveats. MUC5AC proves to be a significant independent prognosticator and should be incorporated into evaluation of ampullary carcinomas.