Management of Small Bowel Neuroendocrine Tumours: 10 Years’ Experience at a Tertiary Referral Centre

Simple Summary Neuroendocrine tumours in the small bowel are often diagnosed when they are at an advanced stage. Treatment for these tumours can be a challenge, and there are multiple types of treatment available, such as surgery, medical options, and targeted radiotherapy. This study sought to report the outcomes of patients with small bowel neuroendocrine tumours treated at a specialist centre, where combined treatment strategies have been increasingly used. Abstract Background: Neuroendocrine tumours (NET) arising from the small bowel are clinically challenging and are often diagnosed at advanced stages. Disease control with surgery alone can be demanding. Multimodal treatment concepts integrating surgery and non-surgical modalities could be of benefit. Method: Retrospective review of consecutive adult patients with SB NET treated at Imperial College Healthcare NHS Trust between 1 January 2010 and 31 December 2019. Data regarding clinicopathological characteristics, treatments, and disease trajectory were extracted and summarised. Overall and progression/recurrence-free survival were estimated at 5 and 10 years. Results: 154 patients were identified, with a median age of 64 years (range 33–87); 135/154 (87.7%) had stage III/IV disease at diagnosis. Surgery was used in 125 individuals (81.2%), typically with either segmental small bowel resection (60.8%) or right hemicolectomy (33.6%) and mesenteric lymphadenectomy for the primary tumour. Systemic and/or liver-directed therapies were used in 126 (81.8%); 60 (47.6%) had more than one line of non-surgical treatment. Median follow-up was 67.2 months (range 3.1–310.4); overall survival at 5 and 10 years was 91.0% (95% CI: 84.9–94.7%) and 82.5% (95% CI: 72.9–88.9%), respectively. Imaging-based median progression-free survival was 42.7 months (95% CI: 24.7 to 72.4); 5-year progression-free survival was 63.4% (95% CI: 55.0–70.6%); 10-year progression-free survival was 18.7% (95% CI: 12.4–26.1). Nineteen patients (12.3%) reached 10 years follow-up without disease recurrence and therefore were considered cured. Conclusions: Most patients with SB NET present in a metastasised stage. Multimodal treatment concepts may be associated with excellent clinical outcomes. Future work should explore optimal approaches to treatment sequencing and patient selection.


Introduction
Neuroendocrine tumours (NET) of the small bowel (SB), recently renamed under the umbrella term neuroendocrine neoplasms (NEN), present manifold clinical challenges: their incidence is steadily rising (0.2 per 100,000 in 1973 compared to 1.05 per 100,000 in 2012) [1,2], they are multifocal in 30-56% of cases [3][4][5], they frequently metastasise, and up to 70% of patients with SB NET are under-staged on morphologic imaging [6].Somatostatin receptor (SSTR) based functional imaging, preferably in positron emission tomography (PET)/computed tomography (CT) techniques have evolved to become gold standard imaging tools since they yield significantly higher diagnostic sensitivity in the detection of both locoregional disease and distant metastases [7,8].Whilst the vast majority of SB NET are well-differentiated grade (G) 1 or G2 lesions, approximately 90% still have mesenteric lymph node metastases and 45-70% have liver metastases at diagnosis [9][10][11].
Radical resection with an aim to eliminate the primary tumour/s including locoregional mesenteric metastases (and, if applicable, complete resection of liver metastases) is the cornerstone of efficient treatment, yielding long-term survival of patients with SB NET [11].An analysis of the Surveillance, Epidemiology, and End Results (SEER) database v18 spanning the period 2000-2014 revealed a 5-year overall survival of 76% for SB NET patients who had any tumour-directed surgery, compared to 45.30% for those who were never treated surgically [12].In the largest single-centre series to date on 603 SB NET patients who were diagnosed between 1985 and 2010 at Uppsala University, the 5-year overall survival was 67% [11].Therein, locoregional resection was reported as a significant positive prognostic factor associated with increased survival compared to no surgery or explorative laparotomy only.
Complete resection of the primary tumour and attendant locoregional disease may be technically demanding [13].Not only are the primary lesions frequently very small and difficult to identify even on gold standard imaging, but lymphadenectomy also presents challenges to surgery aiming to follow small intestine-sparing principles due to frequent encasement of vital mesenteric vascularity accompanied by desmoplastic reaction or mesenteric fibrosis [9,14].In SB NET patients with liver metastases, gross bilobar multifocal spread accompanied by microscopic deposits is a common finding, reducing the chance of complete elimination of hepatic disease burden to less than 30% [11,[15][16][17].As treatment of metastatic disease occupies a pivotal role in the management of patients with SB NET, outcomes of recent landmark randomised controlled trials on somatostatin analogues (PRO-MID [18], CLARINET [19]), mTOR inhibitors (RADIANT IV [20]), and peptide receptor radionuclide therapy (NETTER-1 [21]) have contributed to the incremental implementation of multimodal treatment concepts combining surgery with non-surgical therapeutic strategies.
In this retrospective study, we report the outcomes of patients with SB NET in the era of multimodal treatment concepts and precision medicine in a tertiary referral centre and highlight the benefits and pitfalls of surgical approaches.

Study Population
A prospectively maintained database of consecutive patients with SB NET referred to the Imperial College Health Care NHS Trust (an ENETS Center of Excellence) between 1 January 2010 and 31 December 2019 was used for this retrospective case series study.Small bowel NET were identified as those with a primary tumour site from the duodeno-jejunal junction to the ileo-caecal junction.The diagnosis of a NEN was established histologically on surgical specimens or on biopsy; in most instances, this was on material from liver metastases.All but one patient had a follow-up of at least six months.All patients were discussed in local multidisciplinary team meetings.Patients consented to participation in clinical research.Regarding ethical approval, the institutional approval for the prospectively maintained patient database is under GEN_16; the collection and research use of integrated imaging, genomic, and other clinico-biological data is under 07/MRE09/54 (REC 22/WA/2836), and this specific study was approved by the HRA (IRAS ref: 35117, REC: 23/EM/0102).

Staging and Grading
Tumour staging and grading were according to the European Neuroendocrine Tumour Society (ENETS) tumour-node-metastasis (TNM) staging and grading system [23].The World Health Organisation (WHO) classifications 2017 and 2019, respectively, were considered [24,25].Liver metastases were considered synchronous if present at the time of primary tumour diagnosis and metachronous if diagnosed at least 6 months after primary tumour diagnosis.

Surgical Procedures
Emergency and elective cases were included, as were patients who were referred to us after incomplete/futile surgeries performed elsewhere.Our standard surgical approach included open laparotomy followed by systematic palpation of the entire small bowel starting at the Treitz ligament and ending at the caecal valve, assessment of mesenteric lymph node metastases involvement (stage I-IV) [26], and segmental small bowel resection/s including mesenteric loco-regional disease.A lymph node-first, intestinal-sparing principle was followed.For tumours localised in the terminal ileum, concomitant oncologic right hemicolectomy was performed.Some patients who had their initial surgery at our referring institutions had ileocecal resections.In recent years we have applied a modified approach in selected patients by starting the procedure laparoscopically, mobilising the bowel, and manually exploring the intestine after longitudinal enlargement of the port site incision for the camera and evisceration of all small bowel loops and mesentery.Cytoreduction of peritoneal carcinomatosis was applied as appropriate (peritoneal stripping and/or local electrocautery).In patients considered for liver surgery, the panel of procedures included resections according to Brisbane nomenclature [27], ≥ 70% resections (debulking), and segmental resections combined with intraoperative radiofrequency ablation (RFA).Liver resection was performed as an isolated measure or in combination with primary tumour resection.Selected patients with non-resectable liver metastases and no extrahepatic disease manifestation were evaluated for liver transplantation.Selected patients with advanced loco-regional disease were considered for multivisceral resections.In some cases, the tumour was discovered incidentally at the time of laparotomy for non-NET-related disorders.A patient with stage IV non-resectable mesenteric lymph node metastases and a multifocal primary tumour was investigated for multivisceral intestinal transplantation.
All patients considered for elective surgery were treated with octreotide (50 microgram/hour intravenously) for 12 h prior to surgery and 24 h thereafter to minimise the risk of carcinoid crisis.The definition of carcinoid crisis was according to criteria set by Fouche et al. [28].Resected specimens were subjected to immuno-histochemical examination including assessment of Ki67% for tumour grading.Surgical morbidity (90 days) was assessed according to the Clavien-Dindo classification [29] and 90-day mortality was recorded.

Monitoring of Response to Treatment and Follow-Up
Regular blood tests for assessment of adverse events and imaging were carried out.Final restaging and assessment of response to treatment were performed with SSTR-PET/CT at 6 months after the last PRRT cycle.Responses were evaluated with the European Organisation for Research and Therapy of Cancer (EORTC) criteria [32] (PET component of PET/CT) as well as by the Response Evaluation Criteria in Solid Tumours (RECIST) [33] (CT component of PET/CT or MRI).Adverse events were assessed from laboratory data at the time of occurrence and graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 (NCI, Bethesda, Rockville, MD, USA).Patients were followed up in our clinic until death or study end date (31 December 2020).Follow-up encompassed standard biochemistry, assessment of blood and urinary standard tumour markers for NEN, morphologic imaging every 3-6 months, and SSTR-PET/CT-based imaging every 6-12 months or anytime earlier if suspicious findings on morphologic imaging were evident.

Statistical Analysis
Patient demographics and tumour characteristics were reported with descriptive statistics.Recurrence-free survival (RFS), progression-free survival (PFS), and overall survival (OS) were assessed by using the Kaplan-Meier methodology.Recurrence-free survival was calculated from the date of surgery to the date of disease recurrence evident on imaging.Progression-free survival was calculated from the date of initial diagnosis to the date of progression evident on imaging on first-line systemic treatment or liver-directed treatment.Overall survival was calculated from initial diagnosis to the date of death or the last follow-up visit, respectively.According to a statistical definition of cure previously described [34], patients alive and recurrence-free after 10 years were deemed cured.Logrank tests were used to compare Kaplan-Meier curves, with p < 0.05 set as the threshold for significance.
We performed univariable and multivariable analyses of factors potentially associated with OS and PFS/RFS using Cox proportional hazards regression; however, these were purely exploratory due to the low number of events for most analyses.Stata V17 was used for all analyses.

Results
In total, 154 patients, of them 81 females, were included in the study (Table 1).The median age of the study cohort was 64 years (range: 33-87 years).One patient had hereditary multifocal small bowel NET (with multiple siblings affected).Two patients had more than one primary NET: one a metachronous pancreatic NET and the other a synchronous appendix NET.Both were negative on genetic screening for multiple endocrine neoplasia type 1 and 4, respectively.Fifteen patients (9.7%) had the primary tumour resected in their local hospital prior to referral to our centre for further management.Tissue blocks of all of them were reexamined by a dedicated NET pathologist.Symptoms of carcinoid syndrome were present in 46 patients (29.9%).Various symptoms caused by local effect of the primary tumour such as pain, bleeding, and/or occlusion were recorded in 137/154 (89.0%).In 10/154 cases (6.5%), carcinoid heart disease was evident.In patients with the primary tumour still in place at the time of referral, the site of the primary tumour (small intestine) was identified on imaging and/or endoscopy in 87.1% (121/139).In total, 135 cases (87.5%) were in metastatic stage.Of them, 44 (28.6%) had nodal disease only, 2 (1.3%) had distant metastases only, and in 89 (57.8%) nodal disease and distant metastases were present.Liver metastases were evident in 88 patients (57.1%).The 'classical' tumour markers chromogranin A and 5-HIAA proved non-informative for confirmation of the diagnosis, disease stage, and/or follow-up.Eight patients (5.2%) had synchronous double malignancies, a NET, and an adenocarcinoma.The initial diagnosis of a NET was made in 13 patients (10.4%) at emergency laparotomy for intestinal obstruction or bleeding.

Surgical Treatment
Surgery for the primary tumour including loco-regional lymphadenectomy (and, in some cases, liver resection) took place in 125 patients (81.2%) (Table 2).In all patients with a unifocal tumour, the primary was in the ileum.Multifocal primary tumours were found in 46 cases (36.5%).The median number of tumours in multifocal cases was eight (range, .In all but two cases, multifocal primaries were missed on preoperative imaging.The number of primary tumours detected intraoperatively by surgeons corresponded with the numbers from the histologic analysis.In the vast majority of cases, either segmental small bowel resection (60.8%) or right hemicolectomy (33.6%) was carried out.The grade of the primary tumour was G1 in 102 cases (81.6%),G2 in 22 cases (17.6%), and G3 in 1 case (0.8%).In cases of multifocal tumours with inter-tumoural grade variation, the highest grade was considered.Multivisceral resections were required in three cases to achieve an R0 situation.Thirteen patients (10.4%) underwent emergency laparotomy for intestinal obstruction caused by widespread metastases and/or peritoneal carcinomatosis.In four patients, a NET was discovered incidentally at the time of abdominal surgery for non-NET conditions.One patient had explorative laparotomy and biopsy only.In one patient with a multifocal ileal primary and extensive conventionally unresectable metastatic bulk in the mesentery, a radical resection followed by multivisceral liver-free intestinal transplantation took place after neoadjuvant peptide receptor radionuclide therapy (for more details see previous reports [35]).
In 90 patients (72.0%) locoregional lymph node metastases were present.Eighty-five patients (55.2%) and sixty patients (39.0%) had lymphovascular and perineural invasion, respectively.Complete resection of lymph node metastases was carried out in 84/90 (93.3%).In six patients with extensive mesenteric lymph node metastases (all also had liver metastases), only an incomplete resection of lymph nodes circumferentially encasing the mesenteric root (stage IV) was technically feasible without taking a high risk of postoperative short bowel syndrome.Two patients with an initial R0 resection of locoregionally limited disease had a reoperation for recurrence of mesenteric lymph node metastases after 12 months and 23 months, respectively.
Of patients with liver metastases, 18 (20.5%)(18/88 with LM) were considered as suitable for liver resection, either concomitant with primary tumour resection or as a separate procedure.All had type II liver metastases.Four patients with type II and type III liver metastases, respectively, were evaluated for liver transplantation.None passed the evaluation process as they did not meet the Milan criteria for liver transplantation of neuroendocrine liver metastases.
Five patients were referred to us after laparoscopic procedures with an intent to resect NET at hospitals elsewhere.In two, no tumour was found at the index procedure and in three only incomplete tumour resections were performed.The residual disease was evident on 68 Ga-DOTATATE PET/CT performed for staging purposes after referral.All five patients underwent laparotomy and radical resection of primary tumours and locoregional lymph node metastases.Both patients in whom initially no tumour was identified had multifocal NET with 9 and 12 primary ileal tumours, respectively.
Two patients had cardiac surgery for advanced carcinoid heart disease.In both, multiple valve replacement was required.In one of them, right hemicolectomy, mesenteric lympadenectomy, and liver debulking followed cardiac surgery, and the other remained on systemic treatment with somatostatin analogues.

Non-Surgical Treatment
Systemic and/or liver-directed therapies were administered to 125 patients (81.2%) during the study period, either as an adjunct to surgical treatment for control of residual non-resectable disease and/or for disease progression (the results are not presented in detail since they were published previously either as single centre series or within multicentre trials) [20,21,31].In one of the patients, PRRT was used in the neoadjuvant setting prior to multivisceral transplantation.In 60/125 surgically treated patients (48.0%) more than one line of non-surgical treatment was required.

Follow-Up and Survival Outcomes
The median follow-up was 67.2 months (range, 3.1-310.4months).No patient was lost to follow-up.At the last follow-up, 29 patients (18.8%) were alive with no evidence of NET, 102 (66.2%) were alive with NET, 18 (11.7%)died of NET, and 5 (all of them with stable disease, 3.3%) died of non-NET related causes.
The univariate and multivariable analyses of factors associated with OS and PFS are reported in Tables 4 and 5, respectively.The only significant associations (i.e., p-value < 0.05) identified were those for grade 2 tumours having reduced OS and PFS, but this must be interpreted in the context of the low numbers of events affecting precision and the extensive use of multiple lines of therapy.

Discussion
In this study comprising 154 consecutive patients with SB NET, we have demonstrated the following: (a) the vast majority of SB NET is present in stage III or IV although classified as low-grade neoplasia in 70% of cases; (b) despite high tumour stages and high hepatic tumour burden, radical surgery embedded within multimodal treatment concepts achieves favourable long-term outcomes with acceptable treatment-associated morbidity and nil mortality.The 5-year and 10-year OS of 91.0% and 82.5%, respectively, achieved in our cohort support the notion that cytoreductive surgery should be considered in all SB NET patients fit for surgical intervention [36][37][38].Of note, a cure was likely achieved in 15%.Overall survival did not differ between patients with liver metastases and those with no hepatic involvement.Overlapping confidence intervals of the two groups likely reflect heavy use of successful, continued lines of systemic and/or liver-directed treatment in combination with surgery.No 'predictors' of survival were identified-we hypothesise that multimodal treatment approaches contributed to the favourable long-term survival in this series compared with the 5-year survival (all stages) in the range between 53% and 63% noted in earlier reports [39,40] and the 5-year and 10-year OS of 67% and 37%, respectively, reported in a systematic review in patients in metastasised stages [41].
The outcome of randomised controlled trials for systemic treatment of SBNET, PRO-MID [18], CLARINET [19], RADIANT IV [20], and NETTER-1 [42], which all failed to show impact on OS, further underpins the pivotal role of surgery.
When planning surgery for SB NET, several issues should be considered.Preoperative imaging frequently understages the disease, both in terms of the primary tumour and the metastases.In our series, 36.5% had multifocal primaries that were only detectable in their full extent by meticulous intra-operative palpation.In the series reported by Pasquer et al. [10], multicentric primaries were diagnosed in 33% and missed in 61% on preoperative imaging.Keck et al. found multifocal primaries in 56% of them, with 72% located within 100 cm of the ileocolic valve [3].Clinical behaviour and survival are comparable between multifocal tumours and unifocal counterparts [4,5].Based on the results of complete genome sequencing of tumours from patients with multicentric NET, Elias et al. suggested an independent clonal origin of such lesions [43].
Other underestimated findings on preoperative imaging are mesenteric tumour deposits and peritoneal carcinomatosis.The latter was observed at laparotomy in 12.8% of our patients and in 20-30% in other series [11,44].Fata et al. [45] found mesenteric tumour deposits described as discrete but irregular mesenteric tumour nodules frequently located adjacent to neurovascular bundles and discontinuous from the primary neoplasm in 68% of 132 resected small bowel NET.Since both peritoneal carcinomatosis and mesenteric tumour deposits have been reported as independent factors of prognostic relevance, they should be considered when planning cytoreduction [10,11,44,45].
In contrast to NET originating from the pancreas or lung, high-grade NET or neuroendocrine carcinoma (NEC) are rarely seen in the small intestine.In line with observations made by others, the majority of our patients had G1 or G2 with low Ki67 tumours [11,36,45].When assessing the Ki67 index, intra-tumoural and inter-tumoural grade heterogeneity should be considered.In our previous study, we have shown discrepancies in grading between primary gastroenteropancreatic NET and their metastatic sites in 35.3% of cases [46].In 77.8% of patients with liver metastases, the Ki67 % was higher in liver lesions than in the primary tumour.
An open approach as recommended in current guidelines [13,47] is still considered to be a gold standard since it facilitates extensive manual palpation of the entire small intestine, secure access to lymph node metastases at the root of the mesentery, and synchronous liver debulking in selected cases.Laparoscopic resection is feasible [48,49]; however, solid evidence for the true benefit for the patient is still lacking.Conversion rates of 30% [50] and as high as 80% in patients with stage III or stage IV mesenteric lymph node metastases [51] have been reported.
Abdominal lymph node metastases are present in 75-90% of SB NET at initial diagnosis [9,10,52].Their complete resection is pivotal for favourable long-term survival; however, this is challenging when there is an encasement of the mesenteric vessel origin and/or retropancreatic space, particularly in the presence of retractile fibrotic mesenteritis and/or desmoplastic reaction [11,53].An optimal lympadenectomy should comprise at least eight lymph nodes under consideration of limited resection of the small intestine and utilisation of intestinal sparing techniques [54][55][56].In patients with inadequate primary resection, a reintervention irrespective of negative imaging results should be considered [57].Although 72% of patients in our series had lymph node metastases, 93.3% of these had them completely resected, and only two showed evidence of recurrence in the lymph nodes thereafter.This is in contrast to the published experience in patients with neuroendocrine liver metastases, which recur in virtually all cases [58].
The present study has limitations.The study sample comprises those treated at a tertiary referral centre; this may affect the transportability of findings to other settings, such as the high incidence of distantly metastatic disease in this group, and that all patients underwent elective surgery.Further, due to being managed in a tertiary centre, all patients underwent 68GA DOTATATE PET/CT prior to treatment decisions, which could have increased disease detection (e.g., identifying deposits that could have been 'missed' by other imaging modalities) and therefore exerted influence on treatment selection.The retrospective nature of this case series is also unable to ascertain the effects of specific treatment decisions on outcomes.
A substantial number of reports has shown that hepatic resection for neuroendocrine liver metastases provides favourable outcome with 5-year overall survival rates of 60-100% [15,16,38,59].As shown in this series and in the experience of others, only about 30% are good candidates for hepatic resection (< 45% liver involvement, ≥ 70% cytoreduction, resectable/limited extrahepatic spread).In a systematic review and meta-analysis of outcomes of liver resection in patients with stage IV pancreatic NET, the median 1-year, 3-year, and 5-year overall survival in the resected group was 92.7%, 76.9%, and 67.5%, respectively, compared with 77.3%, 40.9%, and 26.6%, respectively, in the non-resected group [60].As the authors rightly highlight, these data must be taken with caution since there were no randomised controlled trials, the selection for patients was biased by their suitability for surgery, and the treatment in the no liver-resection group was heterogeneous, ranging from primary tumour resection as the only measure to multimodal treatment concepts.Not achieving hepatic resection was the only independent factor of survival in the retrospective series in patients with SB NET and synchronous liver metastases reported by Addeo et al. [61].
In the light of high recurrence rates [62][63][64] and apparently no difference in the longterm survival outcome between patients who had R0 liver resection and those with R1 or R2 status [65,66], a shift from classical major resections [67] to more parenchyma preserving procedures [66,68] and acceptance of a debulking threshold of 70% in contrast to previous 90% has evolved over the last decade.In the series of Scott et al. encompassing 184 patients who had 188 cytoreductive surgeries, the median overall survival and progression freesurvival was 89.4 months and 22.5 months, respectively [69].In addition, 70% debulking was associated with better OS compared with < 70%, and 90% cytoreductive surgery was not associated with improved OS when compared with 70-90%.
The appropriateness of primary tumour resection in asymptomatic SB NET patients with non-resectable liver metastases is debated controversially.While proponents argue that primary tumour resection avoids local complications and prolongs survival, opponents suggest resection in this setting only if symptoms occur.The results of a systematic review were in favour of a proactive approach; median overall survival was 112 months in the primary tumour resection group compared with 60 months in the conservatively managed group.Five-year overall survival rates were 74% and 44%, respectively [70].A study performed by Bennet et al. demonstrated that upfront small bowel resection was associated with lower rates of readmission and intervention compared to non-operative management [71].In a single-centre series reported by Daskalakis et al., prophylactic upfront locoregional surgery had no survival advantage [72].Moreover, delayed surgery was associated with fewer laparotomies for intestinal obstruction.Of note, emergency procedures were included in the latter group.In the experience of Ahmed et al. [15], in contrast, primary tumour resection and the Ki67 index were the only independent predictors of survival in the group of 360 patients with SBNET metastasised to the liver.Results reported by Kaemmerer et al. in patients with stage IV SBNET considered for peptide receptor radionuclide therapy supported upfront primary tumour resection leading to a median overall survival and progression-free survival of 134 months and 18 months, respectively, compared to 67 months and 18 months, respectively [73].
Frequently metastatic disease, high recurrence rates after hepatic resections, and promising results of systemic medical therapies and interventional liver directed procedures call for the consideration of multimodal concepts (combined surgical and non-surgical modalities) in the management of SB NET.In this series, multimodal treatment was applied in more than two third of patients.Of them, 87% had somatostatin receptor targeted therapy.Although in NEN, a neoadjuvant approach utilising peptide receptor radionuclide therapy has been mainly used in those originating from the pancreas [74], results in small bowel NET have been shown in few reports [30,75,76].The recent case series of Fisher and colleagues comprised 17 patients with metastatic ileal NET and found a commendable 93% 5-year overall survival, and a 70% progression-free survival with the use of SSAs after surgery [76].
Genomic, transcriptomic, metabonomic, and epigenetic studies of SB NET have contributed to the delineation of novel molecular subgroups with differing risks of metastases and aggressive behaviour [77][78][79].Future avenues of research should focus on the selection of treatment based on individual patient and tumour characteristics, optimal sequence of treatment modalities, and effective monitoring of treatment response utilising molecularbased markers [80,81].

Figure 1 .
Figure 1.Overall survival for (a) (top) the study sample; (b) (middle) stratified by liver metastases status; (c) (bottom) whether or not patients with liver metastases underwent hepatic surgery.The pvalues for parts (b,c) were obtained from a log-rank test.

Figure 1 .
Figure 1.Overall survival for (A) (top) the study sample; (B) (middle) stratified by liver metastases status; (C) (bottom) whether or not patients with liver metastases underwent hepatic surgery.The p-values for parts (B,C) were obtained from a log-rank test.

Table 1 .
Demographic and clinicopathological characteristics of the study cohort.The stage is as per the AJCC staging system; the grade is as per the ENETS/WHO grading system.

Table 2 .
Surgical and histological characteristics of patients with small bowel neuroendocrine tumours.All individuals undergoing emergency surgery did so prior to referral to the tertiary centre.'Not available' for pT staging refers to the unavailability of the full pathological report of surgical specimens at data extraction.* Surgery was performed at referring hospitals (while the pathologist at our centre re-reviewed the histopathology slides to confirm the diagnosis of well-differentiated NET, there was insufficient tissue remaining for Ki67 staining).

Table 3 .
Kaplan-Meier estimates of survival functions at 1 year and 3, 5, and 10 years after initial diagnosis.

Table 4 .
Associations between selected clinicopathological characteristics and 10-year overall survival were assessed using hazard ratios estimated using Cox proportional hazards modelling.Not estimable: non-convergence of the Cox proportional hazards model (small cell counts).N/A: not applicable.

Table 5 .
Associations between selected clinicopathological characteristics and 10-year progressionfree survival were assessed using hazard ratios estimated using Cox proportional hazards modelling.Not estimable: non-convergence of the Cox proportional hazards model (small cell counts).N/A: not applicable.