The role of tumor cell subtype in classification, assessing the risk stratification, and determining the management of appendiceal NETs has not been well defined. In this study, we show that immunohistochemical staining can be used to subclassify appendiceal NETs into at least three subtypes that have distinct clinical and prognostic features. Just over half of these tumors were EC-cell tumors that express serotonin; they resemble EC-cell tumors of the small bowel. Around 30% of appendiceal NETs were L-cell tumors that have a tubular architecture and correspond to the tumor previously known as “tubular carcinoid” [15]. These tumors may be negative for chromogranin [16], and therefore, they may be misdiagnosed in the absence of other stains. Like L-cell tumors elsewhere in the gastrointestinal tract, they can express PP, PYY, and/or glucagon-like peptides that are detected by antibodies to glucagon. The analysis showed that PYY served as the best biomarker of these tumors, being expressed in 91% of L-cell appendiceal NETs. Interestingly, we also detected a significant proportion of tumors (17%) that show evidence of mixed L- and EC-cell differentiation based on overlapping immunoprofiles with expression of serotonin as well as one or more of PP, PYY, and glucagon. All of these subtypes of appendiceal NETs express CDX2 and SATB2, consistent with origin in the distal gastrointestinal tract [17], and PSAP expression was also frequent.

The analysis of clinicopathological features showed interesting correlations between cell type and behavior. EC-cell tumors were significantly larger with more extensive invasion and more often involved the muscularis propria, subserosa, and mesoappendix compared with L-cell tumors. The biologically indolent nature of appendiceal NETs that are less than 5 mm [18] may also be explained by their increased proportion of L-cell tumors as identified in our series. Mixed tumors were intermediate in all of these parameters. Both EC-cell and mixed tumors had more frequent lymphatic and/or vascular invasion than L-cell tumors. The tumor type correlated with pT stage and the only patient with metastatic disease in this series had an EC-cell tumor. Unlike EC-cell tumors, L-cell tumors did not show metastatic nodal involvement. These data further support the findings of the only previous series that performed hormone stains [10]; that study also reported lack of nodal metastasis in patients with glucagon-positive appendiceal NETs (presumed to be L-cell tumors) compared to serotonin-positive tumors. Our findings indicate that tumor subtyping may be an important pathological variable for risk stratification and prognosis in patients with appendiceal NETs. Current guidelines do not include any reference to tumor cell type or hormone product in patient management [6, 19] but this should be reconsidered.

The importance of cell type in tumor prognosis is also associated with an important message for clinical surveillance. It is generally thought that patients with gastrointestinal NETs can undergo surveillance by measurement of either circulating serotonin or, more commonly, its metabolite urinary 5HIAA (5-hydroxyindoleacetic acid). While this may be true for patients with ileal NETs that are almost exclusively EC-cell tumors, our data show that this test would be misleading in almost half of patients with appendiceal NETs since their tumor did not produce abundant serotonin. Instead, they may have other circulating biomarkers including PYY, PP, and glucagon, which would be reliable only if proven to be diffusely and strongly expressed by the tumor.

Our data showing a less aggressive behavior of L-cell appendiceal NETs are similar to those obtained for rectal NETs where L-cell tumors have been shown to be less aggressive. Thus, a simple appendectomy would be sufficient for the treatment of virtually all appendiceal L-cell NETs. Moreover, biochemical surveillance using circulating L-cell biomarkers may be considered in patients with other questionable risk factors. Since L-cell NETs may rarely show high-risk features such as mesoappendix involvement and intermediate Ki67 proliferation index, it would be prudent to offer surveillance in such cases until more long-term follow-up data are available. Recent data have also identified a subset of rectal NETs that express somatostatin [20]; in this study, we did not examine other hormones but future studies should focus on this and other potential hormone products to further refine the prognosis and management of patients with appendiceal NETs.

In conclusion, our study confirms that appendiceal NETs are not a homogeneous tumor population and are composed of at least three types including EC-cell, L-cell, and mixed tumors, each with unique clinicopathologic characteristics. Cell subtype should be taken into consideration in risk stratification and patient management.