This study found that the overall incidence of GEP-NENs in Japan was 3.53 per 100,000 population in 2016. Median age at onset was around the 60s, but patients with rectal NEN were younger than NEN patients with other primary sites. In terms of malignant potential, NENs in the esophagus, stomach and colon were likely to show a more aggressive phenotype, while most NENs in the duodenum, ileum, appendix and rectum were of low grade. Accordingly, distant metastasis was more common with the former NENs, while the latter NENs were likely to remain at a local site.
The number of NEN patients appears to be increasing recently worldwide6,7. In Japan, two reports by Ito et al. in 2005 and 2010 used questionnaire surveys to estimate the incidence of newly diagnosed GEP-NENs as 1.27 per 100,000 population in PanNENs and 3.51 per 100,000 population in GI-NENs in 20104,5. Our current estimation was 0.70 per 100,000 population in PanNENs and 2.83 per 100,000 population in GI-NENs, slightly lower than the estimations of Ito et al.
These discrepancies might be attributable to the different methods of estimating incidence, resulting in difficulty in estimating trends for NEN patients in Japan. However, when referring to the age distribution and primary site distribution within NEN patients, median age at onset for individuals with PanNEN and GI-NEN in 2005 was 57.6 years and 59.8 years5, compared to 65 years for GEP-NEN in the current study. This difference of approximately 5 years could be attributable to the aging of the population in Japan. Indeed, median age in Japan was 42.9 years in 2005, compared to 46.5 years in 2016 according to a population survey by the Statistics Bureau of Japan, representing an increase similar to that seen in NEN patients.
The incidence of GEP-NENs in each age provided three categories according to primary site: 1) decrease group: appendix and rectum; 2) increase group: esophagus, stomach and colon; and 3) steady group: pancreas, ileum and duodenum. Because the increasing group showed a high proportion of NEC (Figure 2), these different tendencies might imply different mechanisms of development for NENs.
In the 2010 survey, Pancreatic NEC (WHO2010) accounted for 7.5% and GI-NEC for 6.2%5. The current analysis revealed that 20.4% of PanNENs and 26.6% of GI-NENs were NEC, including MANEC, suggesting that aggressive phenotypes are more prevalent than expected. Compared to the SEER database as reported by Dasari6, the overall incidence of GEP-NEC is around 15–20%, comparable to that in our study.
The distribution of primary tumor sites was similar to that in a previous survey from Japan. Half of GEP-NENs were rectal NENs, followed by PanNEN. Comparing to analyses from Europe8-11, the ileum was much less frequent as a primary site, at 1%. This is consistent with a report from Taiwan12, which suggested interactions with ethnic and/or regional factors in the development of NENs. Yao et al.3 presented the incidence of ileal/jejunal NENs among Asian/Pacific Islander and Indian/Alaskan natives as 0.09 per 100,000 population, compared to that in White and African-American individuals as 0.71 and 0.88 per 100,000 population, respectively, suggesting ethnic impacts on the development of NENs.
A tendency toward a male predominance was identified among newly developed GEP-NENs in Japan. The overall incidence among males was 1.73 per 100,000 population, as compared to 1.05 per 100,000 population among females. This is similar to two results using the SEER database in the United States, with 3.65 per 100,000 population among males vs. 3.08 per 100,000 among females in 2000-20043 and males representing 57.75% compared to 42.25% females in 2010–201513.
The current NCR includes data for initial treatment for every patient. As in the figure3, most of the initial treatments for GEP-NEN were surgical or endoscopic resection in every NEN grade even in NEC. It is reasonable that the tumor was resected palliatively or curatively in GI-NEN, because the tumor could cause bowel obstruction. However, even in pancreatic NEC, 31.9% of the tumor were resected. Because 38.0% of the pancreatic NEC was within regionally invasive (figure2), more than 80% of the pancreatic NEC without metastasis were initially resected in Japan. In contrast, treatment of 11.5% of the pancreatic NETG1 was not registered as resection, radiation, chemotherapy, other treatments or unknown. Because NCR registered every element with yes, no or unknown, no treatments above suggest under surveillance or best supportive care. Therefore, for pancreatic NETG1, it is likely that around 10% of the tumor were carefully under surveillance for some reasons in Japan.
This study had several limitations that require consideration when interpreting the results. First, given that NENs may not have been reported to cancer registries unless considered malignant, we have likely underestimated the true incidences and prevalences. Second, the NCR was started in 2016 and we currently lack a system for checking and omitting cases mistakenly registered as newly identified NEN before 2015, implying a potential for underestimation of incidence. This will be corrected as the time progresses. Third, because the NCR was started in 2016, we do not yet have sufficient data on survival. We plan to clarify these issues after we accumulate sufficiently mature data. Finally, several important epidemiological factors are not captured by the NCR. For example, the database does not provide information regarding the hormonal functional status of the NEN, as well as the genetic background. These may affect treatment decisions and survival outcomes that will be analyzed later. Such drawbacks are inherent to any retrospective, population-based study and raise obvious concerns about the generalizability of the findings. However, the size of the present study compensates for this weakness to a great extent and provides a comprehensive epidemiological understanding of NENs in Japan.
In conclusion, we have provided information on the epidemiological status quo among GEP-NEN patients in Japan by utilizing population-based NCR data for the first time. We plan to expand this study to serial years, and to analyze survivals and trends in mortality afterwards. The current study serves as an important first step to determine the exact etiology and trends in Japan.