Treatment sequence in patients with neuroendocrine tumours: a nationwide multicentre, observational analysis of the Swiss neuroendocrine tumour registry

BACKGROUND: In recent years, several treatment modalities have proved to be effective in the treatment of neuroendocrine tumours (NETs). However, there is currently no consensus on the sequence in which these options are best used. METHODS: In this observational study, we analysed the treatment modalities and sequences of all patients included in the Swiss NeuroEndocrine Tumour registry (Swiss-NET). SwissNET is a national registry, which has prospectively included patients with a NET from all regions of Switzerland since 2008. RESULTS: The registry includes 1366 patients; 1063 had documented therapies after the main diagnosis and were included in the analysis. The median follow-up time was 1.86 years. The most common primary site was the small intestine (291 patients, 27%) followed by pancreas (254 patients, 24%), lung (172 patients, 16%) and appendix (163 patients, 15%). A total of 167 different therapy sequences were observed. In 708 (67%) patients, surgery was the only treatment. The sequence of surgery followed by chemotherapy was most frequently documented in poorly (G3) differentiated (24 patients, 60%) and pancreatic (15 patients, 34%) NETs. Tumours treated with surgery followed by biotherapy or followed by peptide receptor radionuclide therapy (PRRT) were predominantly well-differentiated G1 NETs of the small intestine. In patients who were treated with either PRRT or systemic therapy (chemotherapy or molecular therapy) or both, PRRT was used more frequently than systemic therapy in patients with a small intestinal NET (35 patients, 62% vs 30, 54%), whereas the opposite held true in pancreatic (44 patients, 59% vs 56, 70%) and lung NETs (6 patients, 14% vs 40, 97%). If both chemotherapy and molecular therapy were used, chemotherapy was applied prior to molecular therapy in 13 of 19 (68%) patients with a pancreatic NET. CONCLUSION: Surgery represents the treatment of choice in most patients with a NET irrespective of tumour stage. In patients receiving additional treatment, an impressive variety of treatment sequences were documented. In small intestinal NETs, patients received PRRT more often than chemotherapy, whereas the opposite holds true for patients with pancreatic and lung NETs.


Introduction
Neuroendocrine tumours (NETs) represent a rare and highly heterogeneous tumour entity with increasing incidence [1].Based on population-based registry data, approximately 50% of NETs arise from the gastrointestinal tract, one quarter from the lungs and, in third place, 6% from the pancreas [2].The incidence rate of pancreatic NETs has significantly increased during the last decade, whereas the incidence rate of lung NETs is reported to be rather stable [3].
Goals of antiproliferative treatment options for patients with locally advanced and metastatic NETs include the reduction of tumour burden, delay of tumour progression, prolongation of life and improvement in quality of life.Current treatment modalities regularly used to achieve these aims consist mainly of surgery, antiproliferative drugs and peptide receptor radionuclide therapy (PRRT).Systemic drug therapy includes somatostatin analogues, cytotoxic drugs (chemotherapy) and targeted treatment options (molecular therapy).Because of the rarity of NET, large randomised, placebo-controlled trials assessing the efficacy of these treatments in the locally advanced, metastatic setting are limited.In the PROMID trial octreotide LAR significantly increased progression-free survival compared with placebo in patients with a metastatic midgut NET, irrespective of functionality [4].Correspondingly, lanreotide significantly prolonged progression-free survival among patients with a well-differentiated (Ki-67 <10%) metastatic enteropancreatic NET in the ran-domised, double-blind, placebo-controlled, multinational Clarinet trial [5].Chemotherapy regimens (streptozocin, 5-fluorouracil, doxorubicin, capecitabine, temozolomide, oxaliplatin) were investigated in retrospective and small prospective phase II trials [6][7][8].Sunitinib, an oral, smallmolecule, multi-targeted receptor tyrosine kinase inhibitor and everolimus, an inhibitor of mammalian target of rapamycin (mTOR) significantly improve progression-free survival in well-differentiated NETs [9,10].In the NET-TER-1 trial treatment with 177 Lu-Dotatate resulted in markedly longer progression-free survival and a significantly higher response rate than high-dose octreotide LAR among patients with an advanced midgut NET [11].
Currently, clinical trials addressing the issue of optimal treatment sequencing are limited.For example, the efficacy and safety of PRRT with 177 Lu-Edotreotide compared with targeted molecular therapy with everolimus is being studied in a prospective, randomised phase III trial in patients with inoperable NETs of gastroenteric or pancreatic origin (NCT03049189).Outside clinical trials, patient and tumour characteristics as well as treatment goals (disease stabilisation, tumour shrinkage) are taken into consideration to decide on the best treatment strategy.Additionally, physician expertise and regulatory issues with regard to access to novel treatments affect the choice of treatment.Based on those parameters, the ENETS (European Neuroendocrine Tumor Society) published guidelines to facilitate treatment decision [12].In view of the lack of evidence regarding the optimal treatment sequence to improve patient outcome, assessing the presence of a potential treatment consent in daily clinics is valuable.
Primarily, this research project aims to illuminate the NET treatment sequence used in the clinical setting outside of clinical trials based on a large, national, prospectively conducted registry (SwissNET).Secondly, we assessed the sequence of systemic drug therapy (chemo-and molecular therapy) and PRRT on the one hand and the sequence of chemo-and molecular therapy in respect to the different primary NET sites.

Study design and population
The SwissNET registry is a nationwide prospective database documenting data on patients with NET in Switzerland since 2008.Ethical approval to run the registry was obtained from the lead ethics committee in Bern (Kantonale Ethikkomission Bern; No: 395/2014).Currently, 56 participating hospitals and private practices are providing SwissNET with their patient information.Accrual sites in the SwissNET registry are presented in supplementary figure S1 (appendix 1).All documented patients have signed a written informed consent form, agreed to their medical records being collected within SwissNET and agreed that the pseudonymised data can be used for research purposes.Patients with a NET of the aerodigestive tract with the exclusion of small/large cell neuroendocrine carcinoma of the lung are registered in SwissNET.The current classification is based on the revised World Health Organization (WHO) criteria 2010 for NETs of gastroenteropancreatic origin and the 2004 WHO classification for lung NETs [13].Medical records are regularly screened and clinical information, including NET treatment modalities are documented by trained study nurses.
The main outcome was defined by the description of the patients' demographics (age), tumour characteristics (primary site, tumour grade, functionality), all documented treatment modalities and sequence in the main analysis set representing the entire SwissNET cohort (cut-off date 25 July 2017).Treatment modalities were divided into surgery, biotherapy (somatostatin analogues), chemotherapy, molecular therapy (sunitinib, everolimus), PRRT and local ablative therapy (percutaneous radiotherapy, radiofrequency ablation, etc.).Treatment sequencing was based on the date of therapy start.Repeated therapies of the same type were pooled if the time between the consecutive therapy starts was less than 6 months.Biotherapy, in particular the short-acting formulation, is often used concomitantly to other systemic therapies in functional NETs precluding sequence analysis with this regards [14].The limited documentation in the medical history of exact treatment details regarding biotherapy (mainly treatment dose, interval and end) hampered proper analysis of biotherapy in this context.Therefore, the use of biotherapy was analysed separately.
Based on a secondary analysis set, we assessed sequences used for small intestinal, pancreatic and lung NETs and studied the order of use of systemic therapy (chemotherapy and/or molecular therapy) and PRRT and the sequence of molecular therapy (sunitinib, everolimus) and chemotherapy.Local ablative and biotherapy as such were excluded in the secondary analysis set.Both treatment modalities are often used concomitantly with chemo-, molecular therapy and PRRT hampering sequence analysis.

Statistical analysis
Patient and disease characteristics were analysed using descriptive statistics.We reported the number of non-missing observations for patient and tumour characteristics.For the treatment sequences, we were limited to documented treatments.However, we expect that the documentation of the treatments is complete as trained study nurses regularly update all entries.

Treatment sequences in the SwissNET cohort
The different treatment modalities given to the patients in the entire SwissNET cohorts are documented in table 2. Most of the patients underwent surgery (935 patients; 88%) in the course of their disease.Only a minority of patients were received non-surgical treatment modalities.Overall, 158 (15%), 156 (15%) and 113 (11%) patients received biotherapy, chemotherapy or PRRT, respectively.
A total of 167 different therapy sequences were observed (table 3).Out of these, 40 (24%) were used in more than one patient.In 708 (67%) patients, surgery was the only treatment appendical NETs, representing the most common primary tumour site treated with surgery only.Twenty patients (2%) and 12 patients (1%) were treated only with chemotherapy and biotherapy, respectively.
Overall, 311 patients (29%) received more than one treatment modality.For the first two treatment modalities the most common sequences were surgery followed by biotherapy in 45 (14%), surgery followed by chemotherapy in 44 (14%) and surgery followed by PRRT in 36 (12%) patients (table 4).

Small intestinal neuroendocrine tumours
The treatment sequences of 288 small intestinal NET patients were analysed (figure 1).Surgery was the only treatment in 214 patients (74%).Systemic drug therapy only and PRRT only were given to 5 (2%) and 1 (0%) patients, respectively.After surgery, PRRT was applied most frequently (29 patients; 10%).Surgery followed by another tumour resection was noted in 21 (7%) and surgery followed by systemic drug therapy in 15 patients (5%).
The use of systemic drug therapy and PRRT and the use of chemo-or molecular therapy and the corresponding sequence of these two treatment modalities are shown in table 5. Fifty-six patients (19%) with small intestinal NETs

Pancreatic neuroendocrine tumours
In total, 251 pancreatic NET patients were analysed for their treatment sequence.Surgery was the only treatment for 169 patients (67%).Systemic drug therapy only and PRRT only were given to 4 (2%) and 3 (1%) patients, respectively.After surgery, systemic therapy was applied most frequently, in 17 patients (7%).The detailed treatment sequences are depicted in figure 2.
The use of systemic drug therapy and PRRT, and the use of chemo-or molecular therapy and the corresponding sequence of these two treatment modalities are shown in

Lung neuroendocrine tumours
In total, 170 lung NET patients were analysed for their therapy sequence.Surgery was the only treatment for 127 patients (75%).Systemic drug therapy only and PRRT only were given to 17 (10%) and 1 (1%) patients, respectively.Further treatment sequences are illustrated in figure 3.
The sequence whether systemic drug therapy or PRRT and whether chemo-or molecular therapy was given first are demonstrated in table 5. 41 patients (24%) with lung NET were treated with either PRRT or systemic therapy or both.

Original article
In this cohort systemic therapy was used more frequently than PRRT, in 40 patients (97%) vs 6 patients (14%), respectively.In the cohort treated with either molecular therapy or chemotherapy or both chemotherapy was used more frequently than molecular therapy (39 patients [98%] vs 5 patients [13%]).

Biotherapy (somatostatin analogues)
Treatment with somatostatin analogues was documented in 158 patients (15%).In 53 patients of these patients (34%) no surgical therapy was performed in the disease course.Surgery followed by biotherapy and biotherapy followed by surgery could be documented in 70 (44%) and 17 patients (11%), respectively.For functional NETs, biotherapy was given more often before surgery.There was no other obvious difference in patient and tumour characteristics ac-cording to the biotherapy sequence groups (supplementary table S1, appendix 1).

Discussion
The main findings of this study can be summarised as follows: 1. Surgery is the treatment of choice in most NETs irrespective of tumour stage.
2. Only a small proportion of NET patients receive other treatment modalities in their disease course.
3. If patients treated with surgery only are excluded, there is a seemingly unlimited variety of treatment sequences used in NET patients.Original article 4. PRRT seems to be preferred over therapy in patients with small intestinal NETs, whereas the opposite is true in pancreatic and lung NETs.
5. Chemotherapy seems to be preferred over molecular therapy regardless of tumour site.
Our analysis confirms that tumour resection plays a major role for the treatment of NETs.The broad indication for surgical treatment explains this observation.Surgery is considered to be the treatment of choice for patients who have a localised well-differentiated NET, for predominantly hepatic disease in the metastatic setting with potentially curative intent and, importantly, to reduce tumour burden and thus symptoms of carcinoid syndrome [15].
In the case of multifocal tumour progression, evidence is limited with regard to which treatment option should be preferred in which situation.Therefore, studying the sequence of treatment modalities (e.g., PRRT, chemotherapy) and to a lesser extent local therapies is highly relevant to preventing disease progression and improving outcome [14,16,17].Recently, treatment patterns in advanced NETs of the pancreas and potential differences between the treatment in an academic hospital and community oncology practices were reported.Patients treated within the academic tertiary cancer centre received more lines of therapy, were more likely to undergo surgery (47.8 vs 6.5%) including liver-directed therapy and were less often treated with somatostatin analogues (23.8 vs 45.8%).However, the number of patients assessed (n = 44) was small [18].
The diversity of nonsurgical treatment sequences in our study is impressive and might reflect firstly the complexity of the disease and secondly the lack of guidance from clinical studies.Nevertheless, we were able to find some commonalities in the use of different treatment approaches.Note, sunitinib and everolimus gained approval by the European Medical Agency for the treatment of pancreatic NETs in 2010 and 2011, respectively.In 2016, everolimus received approval for treatment of non-pancreatic types of NET, too.PRRT was available in Switzerland during the whole registry period.

Small intestinal neuroendocrine tumours
Tumours treated with surgery followed by biotherapy or followed by PRRT were mostly well-differentiated (G1) NETs of the small intestine.These findings are completely in line with the recently published tumour site-specific guidelines of the European Neuroendocrine Tumor Society (ENETS) [12,19,20].Chemotherapy is not recommended in non-pancreatic NETs unless for tumours with a high proliferation index indicating aggressive biological behaviour or those with somatostatin receptor negativity.Somatostatin analogues may be used in stable or progressive disease for antiproliferative purposes and are mainly recommended as a first-line therapy in midgut NETs.
Assessment of the treatment sequence for PRRT and systemic therapy showed that most patients received PRRT first.There are mainly two reasons for observation.Firstly, most studies in midgut NETs report on an only modest response rate of streptozocin-and temozolomidecontaining chemotherapy in this specific NET subtype [21,22].Additionally, the evidence for the use of molecular therapies such as sunitinib or everolimus in small intestinal NETs, was limited until recently when the results of the RADIANT-4 trial were published [10].Further evidence for the use of PRRT was provided by the phase III randomised controlled NETTER-1 trial, which assessed the efficacy and safety of 177 Lu-DOTATATE in patients with advanced, somatostatin receptor-positive, G1/G2 midgut NETs that were progressing on long-acting octreotide.Objective tumour responses were reported in 18% of patients who received PRRT compared with 3% in those who did not.At the time of primary endpoint analysis, the median progression-free survival had not been reached for the patients who received PRRT and was 8.4 months in the control group [11].Owing to the limited patient numbers no conclusion can be drawn regarding the sequence of chemoand molecular therapy.

Pancreatic neuroendocrine tumours
Based on the current evidence there are rather more accepted treatment options for patients with pancreatic NETs compared with NETs of other primary sites [12].The sequence of surgery followed by chemotherapy was most frequently documented in poorly differentiated (G3) and pancreatic NETs.These findings correlate with the current literature.Streptocotozin-and temozolomide-containing chemotherapy combinations are associated with a high tumour response rate (up to 70%) in NETs of pancreatic origin [6,7].In contrast, the chemo-responsiveness of gas-trointestinal NETs seems to be very limited [23].Additionally, Sorbye et al. reported a significant benefit of platinum-based chemotherapy in poorly differentiated NETs with a proliferation fraction of >55% [24].
Nevertheless, in patients receiving both PRRT and systemic therapies, no clear pattern in treatment sequences could be observed in our analysis.Given the different established and approved therapeutic options in pancreatic NETs and the lack of a prospective trial with PRRT in pancreatic NETs, PRRT is generally recommended in G1/ G2 NETs after failure of medical therapy including a somatostatin analogue, chemotherapy or novel targeted drugs [12].There are several prospective and retrospective studies looking at tumour response to PRRT and survival outcomes for patients with a pancreatic NET, [25,26].Ramage et al. reported a median disease control rate and objective response rate of 83% (range 50-94%) and 58% (range 13-73%), respectively [27].Notably, chemotherapy in pancreatic NETs was more commonly used first followed by molecular therapy than vice versa.This might be explained by the higher response rate with chemotherapy in pancreatic NETs when compared with molecular therapies, sunitinib and everolimus rather leading to disease stabilisation [9,10].

Lung neuroendocrine tumours
When compared with PRRT, systemic drug treatment modalities, and in particular chemotherapy, were more commonly used in lung NET patients.This is surprising as the evidence for the use of chemotherapy for pulmonary NETs is very low.Platinum-based chemotherapy and temozolomide were studied in rather small retrospective cohorts with a reported response rate in NETs of 0-67% and progression-free survival in the range of 10 months [28,29].Only a few phase II trials assessed the potential efficacy of cytotoxic drugs.Oxaliplatin-based chemotherapy was associated with a response rate of 14% [30].Notably, Imhof et al. investigated in a phase II trial the efficacy of PRRT in NETs and reported a response rate of 29% in the lung NET subgroup [31].The rarity of lung NETs has precluded further prospective trials in the past.Again, the small cohorts in our study hampered a meaningful sequence analysis of nonsurgical therapies and therefore further interpretation.

Strengths and limitations
The strength of our study is based on the large cohort, the high quality and completeness of data assessment within the SwissNET database.The prospective nature of the registry and data collection by highly dedicated, specialised study nurses ensures high data quality.Despite the rarity and heterogeneity of this disease, data on the treatment of NETs of the main primary sites are provided in this analysis.There are several limitations of our analysis.The nature of the data is observational and the follow-up time relatively short.Although the total number of intestinal, pancreatic and lung NET patients analysed is high, the small sample size in several treatment cohorts hampers drawing consequences.

Future prospects
Ongoing and future research should improve knowledge around the molecular biology of NET and efficacy of different sequencing or combination strategies.An ongoing Phase III clinical trial, the SEQTOR trial, studies the right treatment sequence for patients with NET of pancreatic origin assessing the best treatment strategy comparing the efficacy of everolimus followed by streptozotocin/fluorouracil or vice versa (NCT02246127).Additionally, the results of the COMPETE trial assessing prospectively the optimal first-line therapy (PRRT with 177Lu-Edotreotide compared to everolimus) in patients with an inoperable NET of gastroenteric or pancreatic origin are eagerly awaited (NCT03049189).A randomised phase III trial of lanreotide autogel versus placebo in advanced, unresectable lung NETS is ongoing (NCT02683941).

Conclusion
We present the largest cohort reporting on the treatment sequence in patients suffering from NETs of any primary site.Our report illustrates the omnium gatherum of used treatment modalities and sequences.Surgery is clearly the treatment option of choice when feasible.If additional therapies are required, PRRT seems to be preferred to systemic therapy in patients with small intestinal NETs, whereas systemic therapy is favourably used in pancreatic and lung NETs.

Figure 1 :
Figure 1: Tree plot for small intestinal neuroendocrine tumours based on the secondary analysis set.Each row represents a therapy line.Stop indicates that no more treatments were documented.(n = 288 patients).PPRT = peptide receptor radionuclide therapy

Figure 2 :
Figure 2: Tree plot for pancreatic neuroendocrine tumours based on the secondary analysis set.Each row represents a therapy line.Stop indicates that no more treatments were documented.(n = 251 patients).PPRT = peptide receptor radionuclide therapy

Figure 3 :
Figure 3: Tree plot for lung neuroendocrine tumours based on the secondary analysis set.Each row represents a therapy line.Stop indicates that no more treatments were documented.(n = 170 patients).PPRT = peptide receptor radionuclide therapy

Continuous and categorical variables are presented as me- dian and interquartile range (IQR) or number and percent
age of patients.All analyses were done in Stata Release 14 (Ref: StataCorp.2015.Stata Statistical Software: Release 14. College Station, TX: StataCorp LP.).

Table 2 :
Therapies in the main analysis set (n = 1063).

Table 3 :
All therapy sequences in the main analysis set (n = 1063).

Table 4 :
Patient characteristics for groups defined by the first two therapies in the main analysis set.

Table 5 :
Therapy sequence groups: peptide receptor radionuclide therapy (PRRT) versus systemic therapy (molecular or chemotherapy) and molecular versus chemotherapy.