A remarkably increasing incidence of gastroentero-pancreatic neuroendocrine neoplasms has been reported over the past several decades throughout the world [1-10]. In our previous nationwide study, we showed that the incidence has also been increasing in Korea over the last decade [3]. The most significant increase was found in the rectum, particularly for well-differentiated neuroendocrine tumors (NETs). In fact, the most common neuroendocrine neoplasm (NEN) site in Korea is the rectum, similar to that reported in Japan [7]. Recently the Surveillance Epidemiology and End Result (SEER) data of United States [5] and data of European Neuroendocrine Tumor Society (ENETS) [11] also described remarkable increases in the incidence of rectal NETs, although the most common site in western countries remains the small intestine [6]. The increasing incidence of rectal NETs in Korea may relate to ascertainment bias from an increasing awareness of this tumor and improved availability of endoscopy, although a true increase in incidence cannot be excluded. Rectal NETs are usually small and often incidentally discovered on endoscopy. Simple endoscopic excision is the typical mode of treatment. The prognosis for this type of local disease is known to be excellent [5,11]; however, confusion arises in cancer registries and over medical insurance compensation, particularly in relation to the 2010 World Health Organization (WHO) classification recommendations.

The WHO updated the classification of these tumors in 2010 and categorized them all as malignant, except for L-cell type and tubular NETs [12], which were designated as having uncertain malignant behavior. L-Cell NENs that are glucagon like peptide (GLP) or PP/PYY producing tumors have been approved for classification as uncertain malignancy by the International Agency for Research on Cancer (IARC)/WHO Committee for International Classification of Disease for Oncology, third edition (ICD-O3) in 2010 [12]. Coding of these tumors according this classification greatly affects the results of cancer registration and implicates L-cell phenotyping of NENs for pathology reports. Potential issues are raised because the diagnostic criteria for L-cell phenotyping are not clear-cut and the relation to other well-known prognostic factors, such as tumor grade, size, and depth remain unclear. As a practical point, this difference in classification has created confusion in the diagnosis, management, and coding of small rectal NETs that are < 10 mm, have low mitoses (WHO G1, < 2 mitosis/10 high-power field [HPF]), confined to the mucosa-submucosa and lacking additional risk factors. These lesions were formerly categorized as benign by the 2000 WHO system; however they should be classified as malignancy if it is not confirmed as L-cell type by the 2010 WHO classification. However, the National Comprehensive Cancer Network (NCCN) clinical practice guidelines recommend simple excision of small NETs [13], irrespective of lymphatic invasion or cell type of origin. Lee et al. [14] recently reported that 79% of rectal NETs were immunohistochemically defined as L-cell type but they showed no clinical significance for L-cell type in their reported rectal NETs.

In our previously published large nationwide multicenter study, rectal NENs showed the best prognosis among all gastroenteropancreatic (GEP)-NENs. However, it was unclear whether or not cell of origin affected the prognosis because information on cell type was unavailable in that study.

Here questions arise about the necessity of L-cell type immunophenotyping in daily practice for better prediction of patient’s prognosis. In addition, it is unclear whether small rectal (WHO G1) NETs by the 2010 WHO classification should be categorized as malignant, if not of L-cell type. To investigate these uncertainties, we analyzed the prognostic significance of L-cell phenotype in rectal NETs, along with previously known factors affecting biologic behavior, to assess the validity of malignancy criteria of rectal NENs.

Among all NENs in the body, rectal NETs are known to have the best prognosis, with a 5-year survival rate of 88.3% for all stages [2] and 90% for localized tumors with increased survival rates over time [1]. WHO classification based mitotic rate (grade) and stage (size and extent) are well known to predict patient’s survival but none are independent factors [16]. The 2010 WHO categorized L-cell type tumors, as uncertain malignancies, while all other NENs were classified as malignant. However, no concrete criteria were provided for defining L-cell type NET by histology or immunophenotyping and none clearly exist in the literature either. The defining L-cell type of rectal NENs by IHC stain also needs clarification as do other prognostic factors that may guide diagnosis and management. In relation to cancer registry data and medical insurance compensation, ICD-O behavior code “/1” means low malignant potential or uncertain malignant potential, and behavior code “/3” means frank malignancy. Therefore, dividing malignant tumors with low malignant potential (/1), very good prognosis and long survival from frankly malignant tumors (/3) is important.

In this study, we analyzed the 5- and 10-year OS of rectal NETs in relation to the pathological parameters, including L-cell typing. A full of 80.61% (n=1,151) of rectal NENs were less than 10 mm and these patients had an excellent prognosis with 10-year OS of more than 99%, if the tumor was NET G1 and confined to the mucosa-submucosa (small NET). Although univariate analysis showed significant prognostic significance of tumor grade, size, extent, lymph node metastasis and L-cell type, no independent prognostic factors were identified among them in multivariate analysis. The Colonoscopy Study Group of the Korean Society of Coloproctology reported lymph node metastasis in 5.1% (21 out of 414 rectal NETs) and was related to tumor size [17]. Lee et al. [14] reported that 6/151 patients with rectal NETs (4%) from 4 to 9 mm in diameter had lymph node metastasis; however patient outcome was not provided. Yamagishi et al. [18] also reported that lymph node metastasis was found in 8/20 patients with surgically resected rectal NETs, smaller than 10 mm, but no disease-related deaths were reported with a mean follow-up period of 68 months. In contrast, Jernman et al. [16] found that none of 61 patients with NET G1 had metastatic disease during follow-up (mean, 124 months). In a Japanese report, metastasis was significantly correlated with the tumor size, importantly, tumors less than 10 mm and confined to the mucosa-submucosa without vessel invasion did not show metastasis [19]. Although cases of small rectal carcinoid (< 10 mm) accompanied by multiple liver metastases have rarely been reported in the literature [20], but the pathological characteristics of these tumor, such as grade and extent are not clearly described. In the survival analysis of our data, 5- and 10-year OS were more than 99% if the tumor size was less than 10 mm, irrespective of lymph node metastasis. As mentioned, lymph node metastasis in patients with NET G1 and G2 did not show any prognostic significance in the survival analysis, however it is very difficult to determine the true incidence of lymph node metastasis in small NET because most of these NET were treated by endoscopic mucosal resection without lymph node dissection. We disagree with the 2000 WHO classification considering well-differentiated NETs as benign, because all NETs have malignant potential of varying degree as seen in our data. On the other hand, we reconfirm the excellent prognosis for patients with small rectal NETs, showing that this series is representative of prior literature and not biased by the Korean nationality of this study population. Therefore we believe that NETs have distinguished biological characteristics from NEC, frank malignancy (ICD-O code /3), although their pathogenetic mechanism is not yet clear.

In this study, to define the L-cell phenotype by immunohistochemistry, we compared three primary antibodies (2 glucagons and 1 PYY). The glucagon I primary antibody can detect the GLP-1 but the glucagon II can detect both GLP- 1 and 2. Gastric and duodenal NENs were almost all negative for all L-cell markers (except for 3 duodenal NENs), while 33.33% of non-rectal, colonic NENs and 83.7% of rectal NENs were defined as L-cell type. Even though the immunoscores of the three markers were significantly correlated with each other and all were prognostically significant by univariate analysis (p=0.0018, p < 0.0001, and p=0.0209, respectively), the patterns of expression were heterogeneous across cases (Fig. 3). In multivariate analysis of three markers, only glucagon II was prognostically significant (p=0.0232; HR, 3.669; 95% CI, 0.094 to 11.269). Immunoscoring was divided into four groups of increasing positivity from 0-3 and the 5-year OS were 48.33%, 91.47%, 94.44%, and 95.29%, respectively. These survival data strongly support the reliability of using a cutoff score group of more than 0 (immunoscore < 10) for these three markers for defining the L-cell type. Lee et al. [14] reported that the L-cell type was 79% in rectal NETs with varying immunoscores of 1-300. They divided the intensities into three groups while we divided them into two. We evaluated the intensity of immunoexpression by comparing tumor cells to normal L-cells in the non-neoplastic mucosa, classifying expression as weaker than or equal to that of L-cells in normal mucosa, because L-cells in normal mucosa showed uniform strong staining. Differences in L-cell tumor incidence may be related to use of different L-cell markers and absence of a standardized definition. In this study, we described the differences in immunoexpression patterns across L-cell markers. We found that glucagon II, which detects GLP-1 and GLP-2, was better than glucagon I and PYY because of the highest hazard ratio in multivariate analysis. In NET G1 and G2 tumors, L-cell type made up 96.56% and 95.24%, respectively, and 96.7% of mucosa/submucosa tumors and 92.1% of tumors < 1 cm. These findings may support that 2010 WHO classification describes L-cell type as uncertain malignancy/low malignant potential (ICD-O code/1).

Although the univariate analysis suggested size, grade, extent, lymph node metastasis, and L-cell phenotype as significant prognostic factors, multivariate analysis failed to show that only a single factor may not adequately predict prognosis. This also validates the conclusion that the suggested combined criteria are suitable for assessment of the biological behavior of rectal NENs. Therefore, in practice, we recommend perform IHC staining for L-cell markers for phenotyping of rectal NET and L-cell type best define classification as uncertain malignancy or low malignant potential (ICD-O code /1) [21,22] as indicated by the 2010 WHO classification if it is small (< 10 mm), G1 (< 2 mitoses per 10 HPF's), and with mucosa/submucosa-limited invasion.

We reconfirm the excellent prognosis for patients with rectal NET G1 if the tumor size was less than 10 mm and confined to the mucosa/submucosa, irrespective of lymph node metastasis. L-Cell type made up 96.56% of NET G1, 96.7% of mucosa/submucosa tumors, and 92.1% of tumors < 1 cm. These findings may support that 2010 WHO classification describes L-cell type as uncertain malignancy/low malignant potential (ICD-O code /1). Because multivariate analysis failed to show that only a single factor can adequately predict patient’s prognosis, we suggest combined criteria are suitable for assessment of the biological behavior of rectal NENs.