Exploring the current status of neuroendocrine tumours: a population-based analysis of epidemiology, management and use of resources

Patient records were extracted from the database under the supervision of the Spanish Ministry of Health, which compiles information from 192 private and 313 public hospitals, covering all Spanish regions. Patients admitted with a neuroendocrine neoplasm in secondary care (inpatient and outpatient) between 2010 and 2015 were identified with the code 209 from the 9th revision of the International Statistical Classification of Diseases and Related Health Problems, Clinical Modification (ICD9-CM). Prior to the extraction, parameters such as health centres and medical history identifiers were re-coded to maintain records anonymised in accordance with the principles of Good Clinical Practice and the Declaration of Helsinki, and thus, ethics committee approval was not required (Ley 14/2007, de 3 de julio, de Investigación biomédica).

All admission data was used to evaluate information on hospital admission and discharge, days of stay, services that attended the patients, and medical procedures utilised. The extraction of single-patient information was carried out to analyse patient characteristics by eliminating repeated records corresponding to separated admissions, relying on the first admission as the index event.

Within the database, tumours were codified and classified following the ICD9 criteria, firstly according to their behaviour and in malignant well-differentiated NETs, malignant poorly differentiated neuroendocrine carcinomas (NECs), benign well-differentiated NETs and metastatic neuroendocrine tumours (mNETs). Then, tumours were classified according to their origin site. ICD9-CM codes identified patient diagnoses, and medical procedures data was claimed by means of the 9th revision of the International Statistical Classification of Diseases and Related Health Problems, Procedure Classification System (ICD9-PCS). The expenses associated with secondary healthcare were calculated based on the mean costs of medical procedures determined by the Spanish ministry of health, which include the costs related to medical examination, all in-hospital medication, surgery, nutrition, personnel, medical equipment and resources in both inpatient and outpatient admissions. The individual cost of somatostatin analogues and targeted therapy could not be calculated due to the lack of a specific ICD9-PCS code and the costs of prescription drugs were not available in this database.

Data presentation is mainly descriptive. Statistical analyses were performed using Microsoft Excel© Professional Plus 2010 (Microsoft Corporation, Redmond, WA, USA).

Between 2010 and 2015, 9120 patients were diagnosed with a neuroendocrine neoplasm, with an increasing tendency over the years (Fig. 1a). A 2 fold increase was observed in the number of diagnosed patients over a 6 year period. Overall, females represented the 42.25% of these patients, while males were the 57.75%. The different types of neoplasms were diagnosed at a mean age of 62.58 years (SD = 14.65), with virtually no differences between males and females. Tumours were diagnosed in patients ranging from 1 to 104 years, and the analysis per age groups revealed a peak in the group of patients between 66 and 70 years old (Fig. 1b).

The database permitted the analysis of tumour typology, considering that one patient could be diagnosed with distinct tumour types. In total, 12,804 neoplasms were registered; 46.86% of the patients had a malignant NET, 32.02% had a poorly differentiated NEC and 42.93% of patients developed mNETs (Table 1). In addition, 18.59% of patients were diagnosed with benign neuroendocrine tumours.

The most common NETs were those of the bronchus and lung and of other sites, which includes pancreatic tumours, mNETs were principally tumours of the liver and distant lymph nodes and for malignant poorly differentiated neuroendocrine tumours site was generally not specified.

In all cases, neuroendocrine neoplasms and metastatic tumours were the hospitalisation motive; yet, secondary diagnoses registered during the admission were also available. Disease comorbidities affecting more than 5% of patients were analysed to further characterise the population (Table 2). In total, 41.62% of the patients had essential hypertension, 18.38% were diagnosed with diabetes mellitus (type I or II), and other disorders that may influence patients treatment were observed, as anaemia, hypercholesterolemia, chronic airway obstruction or renal failure. Additionally, tumour metastases were registered in a large number of patients; presumably attributable to metastatic neuroendocrine tumours affecting principally the liver, lymph nodes and bone.

Between 2010 and 2015, 11,637 admissions were registered in the database corresponding to patients affected by neuroendocrine neoplasms. Less than one admission was registered per patient per year and mean hospitalisation time was of 11 days. The average readmission rate, understood as a readmission attributable to the same motive within 30 days after discharge, was 16.10%.

To assess disease management several parameters were evaluated, including admission origin, the service to attend the patients, medical procedures, and patients’ condition at discharge. According to the registry, 55.99% of hospital admissions corresponded to scheduled visits, while 43.83% were non-scheduled or urgent. After discharge, the vast majority of patients were discharged into their residence, 83.94%, less than 2% of patients were transferred to other hospitals or extended care facilities, while 12.66% of patients died during the hospitalisation event.

Oncology services registered 19.56% of the admissions, followed by admissions into digestive surgery services, 18.86% (Fig. 2).

The analysis of the common medical procedures showed a persistent presence of diagnostic procedures, especially focused on the abdomen and thorax, and a propensity to surgically remove tumours and affected regions (Table 3). Interestingly, the injection of chemotherapy was registered in only 6.85% of total admissions. The injection of other therapeutic substances was registered in 11.40% of admissions, which is likely to include somatostatin analogues and targeted therapies. On the other hand, pulmonary lobectomies and operations were a minority, which suggests that other treatment options were considered for bronchus and lung NETs. A further analysis showed that only 7% of patients with malignant lung NETs were subjected to lobectomies and the same percentage received chemotherapy, while about 10% of patients received another therapeutic substance.

The direct cost of NETs associated to the use of healthcare resources in secondary care (inpatient and outpatient) was €15,373,961 per year, corresponding to €9093 per patient per year between 2010 and 2015. Within this cost, surgical procedures, and chemotherapy and radiotherapy administration accounted for most of the expenses (Table 4). The individual cost of somatostatine analogues and targeted therapy could not be assessed. Patients with malignant NETs of the ileum, cecum, thymus, benign NETs of the transverse colon and other sites (including pancreas) displayed the highest costs per patient.

In addition, 98.39% of the patients were attended under social security financing, and the number of patients under private or mutual care was less than 1%.

The prevalence of NETs in the general population has just recently started being evaluated. These malignancies are overall rare, although research suggest an increase in their diagnosis. A previous retrospective study in Canada displayed a marked increase, of 2.5 fold, of NETs incidence over the course of 15 years [5], comparable to that found herein, probably a consequence of the improvement in their detection. Disease predominance in males has been as well previously described, similarly to the peaking incidence in patients over 65 years [3]. A large number of patients, 78.88%, had a malignant neoplasm, considering NETs and NECs, and the importance of tumours in the bronchus and lung and the gastrointestinal tract support previous evidence that should direct research and diagnostic advances [1, 6].

The presence of comorbidities as essential hypertension and diabetes possibly reflects the ageing nature of the population of study, as for hyperlipidaemia. Alternatively, metabolic disorders have been described in patients treated with certain inhibitors of serine/threonine-specific protein kinases [15]. The present database does not include information regarding patients’ state prior to diagnosis; nonetheless, a previous presence of obesity and metabolic syndrome has been linked with NETs incidence which would require further research [16]. Altogether, comorbidities have a direct influence in the therapeutic choice [2] and should be considered in the principal medical protocols.

The use of molecular tumour markers was not evaluable via this database, although, as expected, diagnostic imaging procedures were preferred in patients with suspected NETs, favouring tomography or MRI techniques [2]. Secondarily, biopsies were employed. In terms of treatment, surgery appeared as the primary therapeutic option, and pancreatectomies were predominant, which indicates a prevalence of pancreatic tumours in these patients although the exact number could not be determined due to the lack of a specific code for this site.

Less than one new admission per year was registered after the diagnosis of a NET, yet the readmission rate was 16.10%. Previous follow-up studies suggest that readmissions are likely related to postoperative complications after pancreatectomy [17], and patients’ admissions into oncology or surgery services are in line with cancer management protocols. The percentage of registered chemotherapy sessions was surprisingly low; it is possible that its importance was diluted in total admission data, nonetheless limitations due to the database characteristics cannot be ruled out. Moreover, the database characteristics impeded a direct analysis of the use of targeted therapies and somatostatin analogues which should be addressed in further studies; currently, the use of these therapies is codified as the injection of another therapeutic substance not specified in another code.

In terms of prognosis and survival, mortality rates vary enormously with tumour characteristics. In patients with pancreatic NETs 5-year survival rates of 53.9% have been registered [18], although previous follow-up studies of patients after aggressive resection suggest an increase in the 5-year survival up to 80% [19]. To infer survival in the population of study, discharge data was evaluated. Herein patients’ mortality could only be measured when it was registered during hospitalisation; yet, in-hospital mortality reached the 12.66%.

The healthcare usage cost associated with NETs is largely unknown, and it originates rising interest given the increased number of diagnosis in the past years. Herein, the €9093 per patient per year are the sum of the direct costs of hospital admissions (inpatient and outpatient care) including all medical procedures, with almost 3 million euros destined to pancreatic surgery only, which represents 19% of total annual secondary care expenses for the disease. A preceding study in the United States suggested that those patients with pancreatic tumours account for the highest costs [20]. Similarly, data obtained from a cancer registry in Canada supports this statement [21]. Our data suggests that this cost could be linked to the resection of benign tumours. Primary care costs and pharmaceutical expenses should be considered in further analysis to calculate the total medical costs of NETs. In addition, Hallet and colleagues predicted an increase of treatment costs over time due to drug maintenance therapy in patients with grastrointestinal NETs, obtaining an inpatient care cost that reached the €9469 per patient at the time of diagnosis [22].

Presumably, patients’ old age increases overall costs as it raises the number of chronic diseases that are diagnosed in these patients, given the augmented needs and high costs that have been associated with multimorbidities including diabetes, anaemia and hypertension [22, 23].

A series of factors limit the conclusions were subjected to a series of limitations that derive from the database characteristics. Data is codified via ICD9 codes that not correlate with the most recent NET classification criteria as expressed by the World Health Organisation. The lack of specific ICD9 codes to identify somatostatin analogues and targeted therapies impedes the quantification of these drug-related costs individually. Equally, primary care and pharmaceutical data was not available. This study provides novel data on the minimal existing literature regarding the management of NETs and their costs; nevertheless, further research will be required to determine the total burden that NETs represent for the Spanish National Healthcare System.