This is the first report to include large amounts of data from a multicenter study in Japan and examine the clinicopathological features and prognosis of PSBA in detail. Previously, it was difficult to detect lesions on the anal side of the ligament of Treitz. Moreover, endoscopic biopsy was difficult; therefore, surgery resection was often required to make a diagnosis [16, 17, 20]. Owing to recent advances in the field of small-bowel endoscopy, especially in diagnostic abilities and therapeutic techniques, the incidence of small bowel tumors initially diagnosed using biopsy and treated with endoscopic resection has increased [18, 19]. However, PSBA is still often detected at an advanced stage, with metastasis to other organs or peritoneal dissemination, because of its rarity [2,3,4]. Therefore, the risk factors for PSBA should be examined, and high-risk cases should be treated at an early stage. Crohn’s and celiac diseases have been reported as risk factors for PSBA [7, 8, 12, 13]. These diseases result in PSBA against a background of chronic inflammation; however, appropriate small intestinal surveillance methods and the duration of these diseases have not been established [21, 24]. Similarly, reportedly, hereditary diseases, such as familial adenomatous polyposis [9], Peutz-Jeghers syndrome [10], and Lynch syndrome [11, 14, 24], are also risk factors for PSBA. Lynch syndrome is a disease in which germline mutations in mismatch-repair genes (MLH1, MSH2, MSH6, EPCAM, and PMS2) predispose patients to the development of various tumors [24]. Because of the extremely low incidence of PSBA in the general population, the proportion of tumors associated with Lynch syndrome is relatively high, at approximately 4–8% among small intestinal cancers [11, 14]. In the present study, the proportion of patients diagnosed with Lynch syndrome was 5%, which was similar to that reported previously. However, this was only a report of diagnosed cases, and the actual incidence of Lynch syndrome may have been higher. Lynch syndrome is predisposed to colorectal cancer; in fact, 15 of the 50 patients in this study with a history of colorectal cancer had Lynch syndrome. Therefore, patients with a history of colorectal cancer may include undiagnosed cases of Lynch syndrome. Although the usefulness of small-bowel capsules for surveillance of Lynch syndrome has been reported [25], the surveillance intervals or which cases of Lynch syndrome are at a high risk for PSBA remain unknown.

Approximately 75% of all patients had symptoms, with the most common being abdominal pain, followed by vomiting. These symptoms are nonspecific and have been reported in the previous studies [7, 16, 17, 20, 26, 27]. Talamonti et al. [26] reported that patients usually had symptoms for a long time before diagnosis, with a mean duration of 10 months. Thus, if these symptoms persist for a long time, an examination of the small intestine should be performed.

Small-bowel endoscopy (single-balloon endoscopy [28], double-balloon endoscopy [18, 19], and capsule endoscopy [29]) is useful not only for the diagnosis but also for the treatment of small intestinal lesions. No study has reported the actual rate of small-bowel endoscopy as a diagnostic device for PSBA. In this study, small-bowel endoscopies were used in approximately 60% of the cases (with some cases overlapping), indicating their widespread use. Moreover, the PSBA site was measured by using small-bowel endoscopy, intraoperative findings, and surgical specimens in this study; the frequent PSBA sites were the proximal jejunum within 40 cm of the ligament of Treitz in the jejunum and the distal ileum within 40 cm of the ileocecal valve in the ileum.

Most previous reports on the prognosis of PSBA have included the duodenum, and some of these studies have revealed the duodenum itself as a poor prognostic factor [2, 30]. However, other studies reported that duodenum is a favorable prognostic factor [4, 31]. This may reflect differences in the number of cases per stage in each report and the inclusion of ampullary carcinoma [32]. Therefore, whether the duodenum is a prognostic factor for PSBA remains controversial. The prognosis of PSBA should be analyzed, excluding the duodenum; however, there are few reports on the prognosis. Amin et al. [23] reported that the 5-year OS rates in patients with PSBA from graph data were approximately 80% for stage I, 60% for stage II, 35% for stage III, and 10% for stage IV. Limited to stages I-III, the 5-year OS rate was 43–59% [2, 4, 20]. In contrast, the 5-year OS rate was < 10% in patients with stage IV disease [4, 20]. Based on the findings of the previous studies and those of our study, the tumor stage is the most important prognostic factor in PSBA [16, 20, 32]. In our study, the 5-year OS and DSS rates for stage I were lower than those for stage II because of the small number of patients in stage I and the fact that stage I patients included many older patients with comorbidities. The reported rates of PSBA stage were 3–11% for stage 0–I, 23–38% for stage II, 22–30% for stage III, and 32–41% for stage IV [2,3,4, 7, 8, 17, 20, 31]. The rate of each stage in our study was similar to that reported in the previous studies; however, the survival rate was higher. The first reason for this was the high rate of surgical resection (approximately 80%) in our study. Curative resection is the main treatment strategy for PSBA, and many surgical resections are performed at the localized stage [21, 32]. In actual practice, in patients with PSBA, surgical resection is often performed even at stage IV if there are obstructive symptoms or perforation findings [17, 27]. In such cases, resection of the primary tumor (radical or non-radical) or bypass surgery is performed, followed by chemotherapy. Furthermore, surgical resection may be performed even for resectable distant metastases. These resections may have increased the efficacy of systemic chemotherapy, because they reduced the tumor volume or improved patients’ activities of daily life. Second, the extent of lymph-node dissection was determined according to the JCCAC, which may have contributed to the prognosis. In fact, lymph-node metastasis was an unfavorable prognostic factor for localized PSBA [16, 26], and several reports indicated that the number of lymph-node dissections was a prognostic factor [3, 33]. Third, various new chemotherapeutic regimens have been developed in recent years. In addition to 5-fluorouracil, capecitabine, oxaliplatin, cisplatin, and irinotecan, which were commonly used for PSBA treatment [21, 32], bevacizumab, regorafenib, or anti-EGFR monoclonal antibodies can be used. Moreover, immune checkpoint inhibitors have been available since 2014 in Japan and are expected to be effective for PSBA in the future. In particular, Lynch syndrome is correlated with mismatch-repair deficiency, which is a good indication for the use of immune checkpoint inhibitors [34]. The peritoneum was the most common site of PSBA metastasis, followed by the liver and lungs, and the same outcome was observed in this study [7, 27]. The rate of peritoneal metastasis was approximately 30–50% in PSBA [7, 27], which could cause obstructive symptoms and was a major factor that made curative surgical resection impossible. The small intestine has a thinner wall than the other gastrointestinal tracts and is presumably more prone to peritoneal dissemination.

The Kaplan–Meier method revealed that the presence of symptoms at the initial diagnosis and the tumor location in jejunum were associated with significantly worse prognosis. However, according to the multivariate analyses with Cox-hazard model, clinical stage was only significant predictor of DSS for patients with PSBC. Similar to the present study, several studies have reported that the presence of symptoms at the initial diagnosis is a poor prognostic factor [7, 27]. Since there were no specific symptoms of PSBA, the disease may have already progressed when symptoms, such as vomiting, appeared. However, Sakae et al. [7] reported that the presence of symptoms at diagnosis was an independent prognostic factor for the tumor stage. Tian et al. [27] also reported that the multivariate predictors of poor prognosis were intestinal obstruction or perforation at first diagnosis. Based on these reports, symptomatic PSBA itself may exhibit poor oncological behavior. Further research, including genetic analysis, is needed to confirm this hypothesis. Moreover, the DSS rate was significantly lower for the jejunum than for the ileum in our study. This result differed from that of previous reports in that the ileum was a poor prognostic factor [4]. This may be because the rate of stage IV tumors was higher in the jejunum than in the ileum in our study.

This study had some limitations. First, this study had an inevitable selection bias, because the data were collected retrospectively from relatively high-volume centers; a prospective multicenter study should be conducted to optimize the treatment for PSBA, a rare disease. Second, not all cases were genetically screened and may potentially include a greater number of patients with Lynch syndrome. Third, only the surgical technique was examined, and it is unknown whether curative surgery was performed. Finally, because there were no PSBA guidelines, treatment decisions were made at the discretion of each institution (most cases were treated in accordance with the JCCAC and JSCCR guidelines). Therefore, there is an urgent need to establish guidelines for PSBAs.

In conclusion, we identified the characteristics and prognoses of patients with PSBA in a large number of cases. To improve the PSBA prognosis, high-risk patients, such as those with Lynch syndrome, should be identified and screened, and PSBA should be detected and treated in the early stages before symptoms appear.