Type 2 Diabetes in Neuroendocrine Tumors: Are Biguanides and Statins Part of the Solution?

Abstract

Context

Biguanides and statins exert beneficial effects on various cancer types. Their precise effects and underlying molecular mechanisms are poorly understood.

Materials and Methods

We analyzed the relationship between metabolic syndrome and histological, epidemiological, and prognosis variables in two cohorts of patients with neuroendocrine tumors (NETs): those with lung carcinoids (LCs; n = 81) and those with gastroenteropancreatic NET (GEP-NET; n = 100). Biguanide and statin antitumor effects were investigated by evaluating proliferation, migration, secretion, gene expression, and involved molecular pathways in BON1/QGP1 cell cultures.

Results

Pleura invasion was higher (LCs group; P < 0.05) and tumor diameter tended to be increased (GEP-NET group) in patients with type 2 diabetes (T2DM) than in those without. Somatostatin and ghrelin systems mRNA levels differed in tumor tissue of patients with T2DM taking metformin or not. Biguanides decreased proliferation rate in BON1/QGP1 cells; the effects of statins on proliferation rate depended on the statin and cell types, and time. Specifically, only simvastatin and atorvastatin decreased proliferation in BON1 cells, whereas all statins decreased proliferation rate in QGP1 cells. Metformin and simvastatin decreased migration capacity in BON1 cells; biguanides decreased serotonin secretion in BON1 cells. Phenformin increased apoptosis in BON1/QGP1 cells; simvastatin increased apoptosis in QGP1 cells. These antitumor effects likely involved altered expression of key genes related to cancer aggressiveness.

Conclusion

A clear inhibitory effect of biguanides and statins was seen on NET-cell aggressiveness. Our results invite additional exploration of the potential therapeutic role of these drugs in treatment of patients with NETs.

Biguanides comprise a class of drugs with relevant effects as insulin-sensitizing agents; consequently, they are used to treat type 2 diabetes (T2DM), a severe disease with distinct comorbidities and whose incidence, along with that of the associated metabolic syndrome and other concomitant diseases, is growing worldwide (1). The inflammation and insulin resistance present in patients with T2DM or metabolic syndrome have been associated with increased incidence of neoplasms (2). Thus, some treatment options targeting related pathways, as may be the case of biguanides, could be beneficial in some types of cancer. In this context, a putative specific relationship between T2DM–metabolic syndrome and neuroendocrine tumors (NETs) has not been established yet.

Among biguanides, only metformin is commercially available for medical use, because it has a safe profile and is well tolerated. Phenformin and buformin were withdrawn in the early 1970s because of an association with lactic acidosis and increased risk for cardiac death (3, 4). Interestingly, a putative association between metformin treatment and cancer prevention and treatment was suggested in 2005 (5), and multiple investigations have been published subsequently on this topic. Specifically, results of some epidemiological studies have suggested a decreased risk for pancreatic, liver, colon, lung, and breast cancers in patients with diabetes treated with metformin (6–9). This protective effect of metformin for cancer also has been found in patients with diabetes, according to several meta-analyses (9–11). Moreover, biguanides can inhibit cell proliferation in vitro in several cancer cell lines, including pancreatic and NET cells (12, 13). In terms of signaling, biguanides stimulate AMP-activated protein kinase (AMPK), reduce hepatic gluconeogenesis and glycogenolysis, and increase glucose uptake in the muscle (14, 15). AMPK activation also suppresses the mammalian target of rapamycin (mTOR) 1, which is a key regulator of proliferation in cancer cells. AMPK induces cell cycle arrest and reduces insulin and insulin like growth factor 1 signaling (16, 17). Metformin-mediated AMPK activation may also result in p53-mediated cell cycle arrest or apoptosis (18, 19). It has been also suggested that metformin could inhibit cell proliferation by G0/G1, G2/M, or S phase arrest (20). However, metformin may also exert antineoplastic properties in an AMPK-independent manner (21).

Statins are also commonly used drugs in the therapeutic arsenal for patients with metabolic syndrome or T2DM. Statins inhibit the enzyme 3-hydroxy-3-methylglutaryl-coenzyme A reductase, affecting the rate-limiting step in cholesterol synthesis, but they also exert other clinical effects related to immunomodulatory mechanisms in vascular diseases, autoimmune diseases, and organ transplantation (22). In addition, statins also reduce bone marrow stimulation and exert antiproliferative effects on smooth muscle cells (23–26). The antitumor mechanisms of statins may include induced cell-cycle arrest, apoptosis induction and activation of the signaling of c-Jun N-terminal kinases (JNKs), decreased invasion or metastasis capacity, and decreased MKI67 expression (27–31). These antitumor effects have been described in several tumor types, including melanoma, colon cancer, and breast cancer (30–33). Moreover, statins have been proposed as an useful treatment option to induce apoptosis and decrease proliferation in pheochromocytomas and paragangliomas (34, 35); to the best of our knowledge, however, studies with statins have not yet been reported in NETs.

Because antineoplastic therapy in advanced NETs is still unsatisfactory, novel drugs for tumor growth control are required, especially in progressive and hereditary NETs, which are characterized by early onset and multiple lesions (12). Therefore, based on the potential association among T2DM, metabolic syndrome, and cancer, we explored this association in a well-characterized cohort of lung carcinoids (LCs) and gastroenteropancreatic NETs (GEP-NETs). In addition, we analyzed the use of antidiabetic drugs and statins in these cohorts and explored their putative relationship with clinical and histological characteristics. Finally, we also investigated the potential in vitro antitumoral effects of different biguanides (namely, metformin, buformin, and phenformin) and statins (namely, atorvastatin, lovastatin, rosuvastatin, and simvastatin) in two different NET-cell models: BON1 and QGP1 cell lines.

Materials and Methods

Patients and tissue samples

This study was approved by the ethics committee of the Reina Sofia University Hospital (Córdoba, Spain), and was conducted in accordance with the Declaration of Helsinki and with national and international guidelines. A written informed consent was signed by every individual before inclusion in the study. A total of 181 patients (n = 81 with LCs; n = 100 with GEP-NETs) who underwent surgery at the Reina Sofia University Hospital from 2005 to 2015 were included in the study. Clinical records were used to collect patients’ full medical history. Endocrine-associated syndromes were excluded. Patients with T2DM before the diagnosis of NET (n = 31 patients) were analyzed separately; these included 14 patients in the LCs group, 6 of whom were treated with metformin, and 17 patients in the GEP-NETs group, 9 of whom were treated with metformin. A similar analysis was performed in those patients treated with statins: four in the LCs group and six in the GEP-NETs group. Demographic and clinical characteristics of both cohorts are summarized in Tables 1–4.

Table 1.

General Characteristics of the Population of Patients With LCs

General Characteristic	Total (n = 81 )	Patients Without Diabetes (n = 61)	Patients With T2DM (n = 14)	Pa
Sex
Male	51.8 (42/81)	52.5 (32/61)	50.0 (7/14)	0.55
Female	48.1 (39/81)	47.5 (29/61)	50.0 (7/14)
Age, y	56.4 ± 15.6	56.1 ± 2.7	58.3 ± 3.7	0.75
Personal history of other tumors	18.7 (14/75)	18.3 (11/60)	21.4 (3/14)	0.52
Smoking habit (active/exsmoker)	65.5 (38/58)	68.1 (32/47)	60 (6/10)	0.44
Family history of neoplasms	55.6 (5/9)	80.0 (4/6)	20.0 (1/3)	0.40
Weight loss	14.6 (7/48)	8.6 (3/35)	36.4 (4/11)	0.046
Functionality	55.6 (5/9)	80.0 (4/6)	20.0 (1/3)	0.40
Incidental	14.6 (7/48)	8.6 (3/35)	36.4 (4/11)	0.046
Maximal tumor diameter, cm	7.5 (4/53)	7.7 (3/39)	8.3 (1/12)	0.67
Multiple tumors	19.2 (10/52)	21.1 (8/38)	16.7 (2/12)	0.55
Vascular invasion	2.9 ± 2.4	2.8 ± 0.3	3.5 ± 0.8	0.36
Neural invasion	6.9 (5/72)	5.5 (3/55)	8.3 (1/12)	0.55
Metastasis	16.1 (5/50)	17.4 (4/23)	0 (0/6)	0.37
Bronchial infiltration	11.8 (2/17)	8.3 (1/12)	25.0 (1/4)	0.45
Parenchyma infiltration	25.0 (17/68)	25.0 (13/52)	25.0 (3/12)	0.65
Pleura infiltration	6.8 (4/59)	2.2 (1/46)	37.5 (3/8)	0.008
Classification				0.17
Typical	69.4 (34/49)	65.8 (25/38)	77.8 (7/9)	0.69
Atypical	30.6 (15/49)	34.2 (13/38)	22.2 (2/9)	0.69
Relapsed disease	11.8 (6/51)	12.5 (5/40)	0 (0/9)	0.34
Disease free during follow-up	77.6 (45/58)	77.8 (35/45)	81.8 (9/11)	0.56
Mortality rate	19.4 (13/67)	15.4 (8/52)	35.7 (5/14)	0.09

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

^a

For the comparison between patients who are not diabetic and those with T2DM.

Table 2.

Metformin Treatment in Patients With T2DM and LCs

General Characteristic	Metformin (n = 6)	Other Antidiabetic Treatment (n = 5)	P
Weight loss	0 (0/4)	100 (4/4)	0.04
Maximal tumor diameter, cm	2.6 ± 0.3	6.7 ± 2.7	0.38
Metastasis	0 (0/6)	60.0 (3/5)	0.12
Bronchial infiltration	50.0 (2/4)	50.0 (1/2)	0.8
Parenchyma infiltration	7.05 (3/4)	25.0 (1/4)	0.6
Pleura infiltration	50.0 (2/4)	50.0 (1/2)	0.8
Disease free during follow-up	100 (4/4)	50.0 (2/4)	0.21
Mortality rate	33.3 (2/6)	60.0 (3/5)	0.39

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

Table 3.

General Characteristics of the Population of Patients With GEP-NETs

General Characteristic	Total (n = 100)	Patients Without Diabetes (n = 70)	Patients With T2DM (n = 17)	Pa
Sex				0.28
Male	57.0 (57/100)	58.6 (41/70)	47.1 (8/17)
Female	43.0 (43/100)	41.4 (29/70)	52.9 (9/17)
Age, y	55.7 ± 17.5	57.8 ± 3.1	55.7 ± 2.2	0.80
Personal history of other tumors	20.7 (18/87)	25.7 (18/70)	0 (0/17)	0.012
Smoking habit	67.4 (29/43)	65.7 (23/35)	71.4 (5/7)	0.57
Family history of neoplasms	46.4 (13/28)	45.0 (9/20)	50.0 (4/8)	0.56
Weight loss	38.5 (20/52)	41.5 (17/41)	25.0 (2/8)	0.32
Incidental tumor	40.3 (29/72)	42.3 (22/52)	33.3 (4/12)	0.41
Functionality	31.5 (23/73)	32.1 (17/53)	33.3 (4/12)	0.59
Primary tumor localization
Pancreas	36.0 (36/99)	34.3 (24/70)	50.0 (8/16)
Stomach	6.0 (6/99)	4.3 (3/70)	6.3 (1/16)
Small bowel	19.0 (22/99)	16.3 (14/70)	25.1 (4/16)
Colon and rectum	36.0 (36/99)	33.4 (29/70)	0 (0/16)
Maximal tumor diameter, cm	2.6 ± 2.2	2.5 ± 0.2	3.4 ± 0.5	0.06
Free surgical border	87.3 (62/71)	93.8 (45/48)	69.2 (9/13)	0.032
Multiple tumors	7.5 (4/53)	7.5 (3/40)	0 (0/6)	0.65
Local infiltration	53.1 (43/81)	57.1 (32/56)	42.9 (6/14)	0.25
Vascular invasion	28.4 (21/74)	30.8 (16/52)	25.0 (3/12)	0.50
Neural invasion	29.6 (21/71)	32.7 (16/49)	16.7 (2/12)	0.24
Metastasis	47.4 (45/95)	45.5 (30/66)	47.1 (8/17)	0.56
Grading (WHO 2010 criteria)
Low	46.4 (32/69)	52.2 (24/46)	23.1 (3/13)	0.08
Intermediate	39.1 (27/69)	34.8 (16/46)	69.2 (9/13)	0.08
High	14.5 (10/69)	13.0 (6/46)	7.7 (1/13)	0.27
Relapsed disease	34.2 (27/79)	36.8 (21/57)	27.3 (3/11)	0.41
Disease free during follow-up	61.4 (43/70)	60.8 (31/51)	70.0 (7/10)	0.43

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

Abbreviation: WHO, World Health Organization.

^a

For comparison between patients who are not diabetic and those with T2DM.

Table 4.

Metformin Treatment Patients With T2DM and GEP-NETs

General Characteristic	Metformin (n = 9)	Other Antidiabetic Treatment (n = 7)	P
Weight loss	25.0 (1/4)	33.3 (1/3)	0.71
Maximal tumor diameter, cm	3.5 ± 0.8	3.5 ± 0.9	0.95
Necrosis	25.0 (1/4)	0 (0/4)	0.50
Vascular invasion	40.0 (2/5)	16.7 (1/6)	0.42
Neural invasion	0 (0/5)	33.3 (2/6)	0. 27
Metastasis	44.4 (4/9)	57.1 (4/7)	0.50
Mortality rate	55.6 (5/9)	28.6 (2/7)	0.29

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

Patients were treated following the available guidelines and recommendations. After surgery, if residual or relapsed disease was observed, adjuvant treatment with or without surgery was prescribed. To confirm the neuroendocrine nature of all tumors, chromogranin A, synaptophysin, cytokeratin 7, cytokeratin 20, CD56, and neuronal-specific enolase levels were determined by immunohistochemistry, which was performed following the standardized diagnosis protocol of our hospital and evaluated by two experienced pathologists. Formalin-fixed paraffin-embedded (FFPE) samples were available for 46 LCs and 55 GEP-NETs cases; total RNA was isolated from these. Tumor samples were reevaluated by two experienced pathologists before RNA isolation. Only primary tumor samples were included. Samples were analyzed individually and mRNA expression levels were correlated with the clinical and histological characteristics of the corresponding patient.

Culture of cell lines

We used two human pancreatic NET (PNET) cell lines: BON-1 and QGP-1 (36–39). BON-1 cells were cultured in DMEM-F12 (Life Technologies, Barcelona, Spain) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, Madrid, Spain), and 0.2% antibiotic (gentamicin/amphotericin B; Life Technologies). QGP-1cells were cultured in RPMI 1640 (Lonza, Basel, Switzerland), supplemented with 10% FBS, 1% glutamine, and 0.2% antibiotic. Cells were harvested with trypsin (0.05%) and EDTA (0.53 mM) (Sigma-Aldrich) and resuspended in culture medium. Cell viability always exceeded 85%. Both cell lines were cultured in 75 cm² flasks at 37 °C in a 5% CO₂ incubator. All experimental procedures in both cell lines were performed at least three times.

Drugs and reagents

Metformin, phenformin, simvastatin, atorvastatin, lovastatin, and rosuvastatin were purchased from Sigma-Aldrich. Buformin was purchased from Santa Cruz Biotechnology (Dallas, TX). All treatments were dissolved in the respective FBS-free medium and diluted until final concentrations were obtained before use (metformin, 10⁻² M; phenformin and buformin, 5 × 10⁻³ M; statins, 10⁻⁵ M). IGF1 and paclitaxel were purchased from Sigma-Aldrich. Drug doses were selected on the basis of in vitro dose-response curves (unpublished results) or in previous studies (40, 41).

Cell viability assay

Cells were plated in 100 µL of medium in 96-well plates at the density necessary to obtain a 65% to 70% cell confluence in the control groups at the end of the experiment. Twenty-four hours later, serum-free medium was added over 24 hours. After this, biguanides and statins were added into wells in medium with 5% serum. Cell viability was measured using the alamarBlue assay (Thermo Fisher Scientific, Waltham, MA) at basal, and 24, 48, and 72 hours of incubation by measuring the fluorescent signal exciting at 560 nm and reading at 590 nm (Flex Station 3; Molecular Devices, San Jose, CA). On the day of each measurement, cells were incubated for 3 hours in 10% alamarBlue/serum-free medium and then alamarBlue reduction was measured. After each measurement, medium was replaced immediately by fresh medium. In all cases, cells were seeded per quadruplicate and all assays were repeated a minimum of four times. IGF1 and paclitaxel treatment were used as positive and negative controls, respectively.

Migration capacity assay

The ability of BON-1 cells to migrate after 24 hours of treatment with biguanides and statins was evaluated by a wound-healing technique. Briefly, cells were plated at subconfluence in 12-well plates. Confluent cells were serum starved for 24 hours; after synchronization, the wound was made using a 100-μL sterile pipette tip. Cells were incubated for 24 hours in FBS-free medium. Wound healing was calculated as the area of a rectangle centered in the picture 24 hours after the wound vs the area of the rectangle just after wounding, as previously reported (42). At least three experiments per cell line were performed on independent days, in which three random images along the wound were acquired per well.

RNA isolation and reverse transcription

Total RNA from FFPE samples was isolated using the RNAeasy FFPE Kit (Qiagen, Limburg, Netherlands) according to the manufacturer’s instructions. Quantification of the recovered RNA was assessed using a NanoDrop2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, NC). One microgram of total RNA was retrotranscribed to cDNA with the First Strand Synthesis kit using random hexamer primers (Thermo Fisher Scientific), as previously reported (43, 44).

Quantitative real-time PCR

cDNA was amplified with the Brilliant III SYBR Green Master Mix (Stratagene, La Jolla, CA) using the Stratagene Mx3000p system and specific primers for each transcript of interest. Specifically, expression levels (absolute mRNA copy number per 50 ng of sample) of insulin receptor (INSR) and GLUT4 genes were measured in the cells using previously validated primers (41), and the expression level of each transcript was adjusted by the expression of β actin (BACT; used as housekeeping gene). Experiments were performed at least three times.

In human tumor samples, somatostatin system [i.e., somatostatin (SST), cortistatin (CORT), and their receptors SSTR1, SSTR2, SSTR3, SSTR4, SSTR5, and sst5TMD4], and ghrelin system [i.e., ghrelin (GHRL), In1-ghrelin variant, ghrelin-o-acyltransferase enzyme (GOAT), and the receptors GHSR1a/GHSR1b] were evaluated using previously validated primers (41, 45–48). mRNA levels were normalized by 18S in GEP-NETs and by BACT in LCs, as previously described (45, 46). In NET cell lines, the expression of INSR and glucose transporter GLUT4 was analyzed in response to biguanides and statins treatment using previously validated primers (49, 50).

All samples were run, in the same plate, against a standard curve to estimate mRNA copy number and a no reverse transcriptase sample as a negative control. Thermal profile consisted of an initial step at 95°C for 30 seconds, followed by 50 cycles of denaturation (95°C for 20 seconds) and annealing and elongation (60°C for 20 seconds), and finally, a dissociation cycle (melting curve; 55°C to 95°C, increasing 0.5°C/30 seconds) to verify that only one product was amplified.

Serotonin assay

BON-1 and QGP-1 cells were cultured in 12-well plates. At 70% confluence, cells were serum starved and after a 24-hour incubation period with specific treatments or with vehicle, media were collected and stored at −20°C until measurements were made. Secretion of serotonin was detected using a serotonin ELISA kit (ALPCO, Salem, NH) following the instructions of the manufacturer.

Apoptosis assay

BON-1 and QGP-1 cells were cultured in 24-well plates. At 70% confluence, cells were serum starved and after a 48-hour incubation with specific treatments or with vehicle-treated controls in 5% FBS medium, apoptosis levels were measured using a cell-death detection ELISA kit (Sigma-Aldrich) following the manufacturer’s instructions.

Measurement of ERK1/2 and AKT signaling pathways by western blotting

A total of 500,000 of each cell type (BON-1 and QGP-1) were cultured in six-well plates and incubated for 8 minutes with specific treatments and vehicle-treated controls. Briefly, after the corresponding treatment, medium was removed and cells were washed twice using PBS, detached using a scraper, and immediately lysed in prewarmed SDS-dithiotreitol sample buffer at 65°C (62.5 mM Tris-HCl, 2% SDS, 20% glycerol, 100 mM dithiotreitol, and 0.005% bromophenol blue) followed by sonication for 10 seconds and boiling for 5 minutes at 95°C, as previously described (51, 52). Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes (Millipore, Darmstadt, Germany). Then membranes were blocked with 5% nonfat dry milk in Tris-buffered saline/0.05% Tween 20 and incubated with the primary antibodies for total-ERK1/2, p-ERK1/2, total-AKT, p-AKT (Santa Cruz, CA) and then with the appropriate secondary antibodies (anti-rabbit antibody; Cell Signaling, Danvers, MA). Protein analyses were developed using an enhanced chemiluminescence detection system (GE Healthcare, Chicago, IL) with dyed molecular-weight-markers. A densitometric analysis of the bands was carried out with ImageJ software (53). Relative phosphorylation of ERK and AKT was obtained from normalization of p-ERK1/2 or p-AKT against the total ERK1/2 or AKT, respectively.

Statistical analysis

In functional experiments, results are expressed as percentage vs vehicle control (nontreated cells). mRNA levels are expressed as mean ± SEM. Cell survival rate compared with control was assessed by multiple comparison tests. Mann-Whitney U tests were used to evaluate clinical relations within LC and GEP-NET samples. χ² test was used to compare categorical data. All statistical analyses were performed using SPSS statistical software, version 20, and GraphPad Prism, version 6. P values < 0.05 were considered statistically significant.

Results

Clinical evolution in patients with LCs and GEP-NETs and correlations with T2DM

In the LC group, weight loss was more frequently observed in patients with T2DM than in patients without diabetes (36.4% vs 8.6%; P < 0.05; Table 1). Likewise, pleura invasion was also higher in patients with T2DM (37.5% vs 2.2%; P < 0.05; Table 1). Despite the increased incidence of weight loss in patients with diabetes with LC, none of the metformin-treated patients with T2DM exhibited this symptom (P < 0.05; Table 2). In this cohort, the clinical outcome did not differ in those patients receiving metformin or other antidiabetic treatment (Table 2). The death rate tended to be increased in patients with T2DM (P = 0.09; Table 1).

In the GEP-NETs group, an increased incidence of a second neoplasm was observed in the nondiabetic group (25.7% vs 0%; P < 0.05; Table 3). Tumor diameter tended to be greater in patients with T2DM compared with patients without diabetes (3.4 ± 0.5 vs 2.5 ± 0.2 cm; P = 0.06). In addition, the proportion of patients with complete surgical resection was lower in the T2DM group compared with the nondiabetic group (69.2% vs 93.8%; P < 0.05; Table 3). In this cohort, the clinical outcome of patients treated with metformin was also similar to those treated with other antidiabetic drugs or insulin (Table 4).

None of the other clinical parameters evaluated (including functionality or incidental findings), histopathological variables (including necrosis, local invasion, presence of metastasis, and vascular or nerve invasion), tumor grading, or evolution parameters (including relapsed disease, disease-free survival, and mortality) were associated with T2DM or the use of metformin in the cohorts of patients with LCs or GEP-NETs.

No clinical, histological, or molecular variable was associated with the presence of hyperlipidemia in the cohort of patients with LCs (Table 5) or GEP-NETs (Table 6). A higher proportion of patients treated with statins were free of disease during the follow-up (χ², 7.07; P < 0.05). None of the other clinical, histological, or evolution parameters were associated with the use of statins in the cohorts of patients with LCs or GEP-NETs.

Table 5.

Statin Treatment in Patients With LCs

General Characteristic	No Statins (n = 71)	Statins (n = 4)	P
Previous other tumor	16.9 (12/71)	50.0 (2/4)	0.16
Weight loss	13.0 (6/46)	50.0 (1/2)	0.27
Maximal tumor diameter, cm	2.8 ± 0.3	4.6 ± 2.3	0.42
Necrosis	33.3 (8/24)	0 (0/1)	0.68
Metastasis	25.0 (16/64)	25.0 (1/4)	0.74
Bronchial infiltration	74.1 (43/58)	100 (2/2)	0.55
Parenchyma infiltration	40.4 (23/57)	0 (0/2)	0.37
Pleura infiltration	7.0 (4/57)	0 (0/2)	0.87
Disease free during follow-up	78.6 (44/56)	50 (1/2)	0.40
Mortality rate	17.5 (11/63)	50 (2/4)	0.167

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

Table 6.

Statin Treatment in Patients With GEP-NETs

General Characteristic	No Statins (n = 80)	Statins (n = 6)	P
Weight loss	36.4 (16/44)	60.0 (3/5)	0.29
Maximal tumor diameter, cm	2.6 ± 0.2	4.3 ± 1.2	0.13
Necrosis	32.0 (8/25)	0 (0/1)	0.69
Peritumoral invasion	52.3 (34/65)	75.0 (3/4)	0.36
Vascular invasion	28.3 (17/60)	33.3 (1/3)	0.64
Neural invasion	28.1 (16/57)	33.3 (1/3)	0.64
Metastasis	44.7 (34/76)	66.7 (4/6)	0.27
Mortality rate	28 (21/75)	66.7 (4/6)	0.07

Data given as % (no./total) or mean ± SD, unless otherwise indicated.

mRNA expression of SST and ghrelin system components in LCs and GEP-NETs, and their correlations with T2DM

The mRNA levels of several genes of interest were measured in the tumor tissue obtained from patients with LCs (Fig. 1A) and GEP-NET (Fig. 1B). In the LCs group, mRNA levels of SST and several receptor subtypes (i.e., SSTR1, SSTR2, SSTR4, SSTR5, and sst5TMD4), but not of CORT or SSTR3, were numerically, albeit nonsignificantly, decreased in patients with T2DM compared with the patients without diabetes (Fig. 1A). A similar pattern of expression was observed in the GEP-NETs group, except for the mRNA levels of SSTR5 (Fig. 1B). Interestingly, an overall decrease in the mRNA levels of all SSTRs was found in the LC and GEP-NET groups, but this difference only reached statistical significance in the GEP-NET group (Fig. 1C).

Figure 1.

mRNA expression of SST system components in patients with and without T2DM with (A) LCs and (B) GEP-NETs. (C) Total mRNA expression of somatostatin receptors in patients with LCs or GEP-NETs. The absolute mRNA expression of the different components of the SST system was determined by quantitative real-time PCR in tumor samples (values were normalized to BACT in LCs and 18S in GEP-NETs). Data represent the mean ± SEM. **P < 0.01.

Similarly, mRNA levels of all the components of the ghrelin system (i.e., GHRL, In1-ghrelin, GOAT, and the receptors GHSR1a and GHSR1b) displayed nonsignificant lower levels in patients with diabetes in the LC group compared with patients with LCs but not diabetes (Fig. 2A). In GEP-NETs, the mRNA levels of In1-ghrelin, GOAT, and GHSR1b, but not GHRL or GHSR1a, also tended to be lower in patients with T2DM (Fig. 2B).

Figure 2.

mRNA expression of ghrelin system components in patients with and without T2DM with (A) LCs and (B) GEP-NETs. The absolute mRNA expression of the different components of the ghrelin system was determined by quantitative real-time PCR in tumor samples. Data represent the mean ± SEM.

Interestingly, a subanalysis showed that although the overall expression of SSTRs was significantly lower in the GEP-NETs group with T2DM compared with patients without diabetes, these levels were not decreased in patients with T2DM treated with metformin compared with patients who were not diabetic (Fig. 3A). Specifically, a nonsignificant increase in the mRNA levels of SST, CORT, SSTR1, SSTR2, and SSTR3 was observed in patients with T2DM treated with metformin compared with patients with T2DM not treated with metformin (Fig. 3B), as well as in the mRNA levels of GHRL, In1-ghrelin, and GHSR1a (Fig. 3C).

Figure 3.

Effects of metformin in patients with T2DM and GEP-NETs. (A) Total mRNA expression of somatostatin receptors in GEP-NETs. Specific (B) SST and (C) ghrelin system components in GEP-NETs. The absolute mRNA expression of the different components of the SST system was determined by quantitative real-time PCR and normalized to 18S. mRNA expression was assessed in patients with GEP-NET with and without T2DM. Among patients with GEP-NET with and without T2DM, two subgroups were analyzed: those treated with metformin vs those treated with other antidiabetic drugs or insulin. mRNA expression was compared with that in controls. Data represent the mean ± SEM. *P < 0.05.

Cell survival of PNET cells after treatment with biguanides or statins

All biguanides tested clearly decreased the survival rate in both BON-1 (Fig. 4A) and QGP-1 (Fig. 4B) cell lines in a time-dependent manner. The most remarkable effect was observed in BON-1 cells with phenformin (5 × 10⁻³ M), which decreased the survival rate by 76.6%, 93.1%, and 97.13% after 24, 48, and 72 hours of incubation, respectively. Metformin (10⁻² M) decreased the survival rate in these cells by 25.1%, 38.1%, and 49.4%, whereas buformin (5 × 10⁻³ M) reduced it by 36.9%, 37.1%, and 56.3% after 24, 48, and 72 hours of incubation, respectively (Fig. 4A).

Figure 4.

Time-dependent effect on cell viability of biguanides in (A) BON-1 and (B) QGP-1 cell lines; and statins in (C) BON-1 and (D) QGP-1. Cell viability is expressed as cell survival percentage after 24, 48, and 72 hours. Cell proliferation rate compared with that of controls was assessed by multiple comparison tests. *P < 0.05; **P < 0.01.

A similar effect was observed in the QGP-1 cell line. Specifically, phenformin was also the most effective biguanide; survival rates with phenformin decreased by 68.2%, 87.4%, and 96.9% after 24, 48, and 72 hours of incubation, respectively, whereas, metformin decreased the survival rate by 24.9%, 45%, and 60%, respectively, and buformin by 30.7%, 53.0%, and 69.7%, respectively (Fig. 4B).

We also analyzed the effect of different statins on cell survival in BON-1 (Fig. 4C) and QGP-1 (Fig. 4D) cell lines. Specifically, a decreased survival rate was observed after 48 and 72 hours of treatment with simvastatin (10⁻⁵ M; 21.4% and 34.5%, respectively), and after 72 hours of treatment with atorvastatin (10⁻⁵ M; 15.2%) in BON-1 cells (Fig. 4C). The effect of simvastatin was more pronounced than that of atorvastatin at 72 hours (P < 0.05; Fig. 4C). In QGP-1 cells, a reduction in the proliferation rate was observed after 48 and 72 hours of incubation with all the statins tested (Fig. 4D). Thus, simvastatin, atorvastatin, lovastatin, and rosuvastatin (10⁻⁵ M) decreased survival rates by a range of 14.7% to 17.2% after 72 hours.

Based on their antiproliferative effects, phenformin and simvastatin were chosen as representative compounds of these two classes of drugs to perform further functional experiments (i.e., migration, apoptosis, and serotonin secretion). Moreover, metformin was also included in these analyses because of its relevance in clinical practice.

Migration capacity in PNET cells in response to metformin, phenformin, and simvastatin treatment

Metformin and simvastatin (after 24 hours of incubation) significantly decreased the migration capacity of BON-1 cells (100% and 38.6%, respectively; representative images are presented in Fig. 5A). In contrast, it was not possible to measure the migration capacity in response to phenformin in BON-1 cells, perhaps due to a treatment-related toxicity of this compound (discussed later in Results). As previously reported (54), it was not feasible to measure QGP-1 cells with this functional assay, because these cells form aggregates or clusters in culture, which do not allow correct measurement of the migration capacity under basal conditions or in response to any given treatment.

Figure 5.

Effect of biguanides and statins on (A) cell migration in BON-1 cells; (B) apoptosis rate in BON-1 and (C) QGP-1 cells; and (D) serotonin secretion in (D) BON-1 and (E) QGP-1 cell lines. Migration and serotonin secretion were assessed after 24 hours of incubation; apoptosis rate was evaluated after 48 hours. Representative images of wound healing after 24 hours of treatment are presented in (A), lower panels. Treatment rates were compared with those in controls by multiple comparison tests. *P < 0.05; **P < 0.01; ***P < 0.001.

Effect of metformin, phenformin, and simvastatin treatment on apoptosis

In BON-1 cells, phenformin caused a threefold increase in apoptosis (Fig. 5B). However, metformin or simvastatin treatment did not alter apoptosis in BON-1 cells. In QGP-1 cells, a twofold increase in apoptosis was also observed in response to phenformin (Fig. 5C). In addition, simvastatin increased the apoptotic rate in QGP-1 cells by 58.1% (Fig. 5C). Conversely, metformin treatment did not alter apoptosis in QGP-1 cells.

Effect of biguanides and statins on serotonin secretion in PNET cell lines

In BON-1 cells, phenformin, but not simvastatin, decreased serotonin secretion after 24 hours of incubation (P < 0.05; Fig. 5D). Metformin treatment also tended to decrease serotonin release (P = 0.06; Fig. 5D). In contrast, none of these treatments altered serotonin secretion from QGP-1 cells (Fig. 5E).

Effects of metformin, phenformin, and simvastatin on ERK1/2 and AKT signaling pathways

To start exploring the signaling pathways affected by biguanides (i.e., metformin and phenformin) and simvastatin to induce their functional actions in NET cells, the levels of phosphorylation of AKT and ERK were evaluated. In BON-1 cells, both biguanides and simvastatin similarly decreased phosphorylation levels of AKT and ERK compared with controls (Fig. 6A). In marked contrast, in QGP-1 cells, only phenformin and simvastatin decreased phosphorylation levels of ERK without altering those of AKT (Fig. 6B).

Figure 6.

Effects of biguanides and statins on AKT and ERK phosphorylation in (A) BON-1 and (B) QGP-1 cells. Phosphorylation levels compared with those in controls was assessed by multiple comparison tests. *P < 0.05; **P < 0.01; ***P < 0.001.

Effect of metformin, phenformin, and simvastatin in the expression of key genes in PNET pathophysiology

In BON-1 cells, metformin and phenformin severely decreased the mRNA levels of INSR (P < 0.001; Fig. 7A). Also, a trend toward an increase in the expression GLUT4 was observed in response to phenformin (Fig. 6A). In QGP-1 cells, GLUT4 expression was increased in response to both biguanides and simvastatin, but this difference only reached statistical significance in the case of phenformin (Fig. 7B). No significant changes were observed in the expression of INSR in QGP-1 cells in response to these compounds (Fig. 7B). Finally, metformin treatment did not significantly alter the expression of SSTRs in BON-1 and QGP-1 cells (data not shown).

Figure 7.

Effects of biguanides and statins on mRNA expression in (A) BON-1 and (B) QGP-1 cells. mRNA expression compared with that in controls was assessed by multiple comparison tests. *P < 0.05; ***P < 0.001.

Discussion

NETs are a widely heterogeneous group of neoplasms that are frequently diagnosed at an advanced stage of disease (55). Therapeutic options for advanced, metastasized NETs include somatostatin analogs, interferon-α, chemotherapy, and peptide receptor therapy (55, 56). Target-directed therapies have increased the therapeutic spectrum for progressive NETs (e.g., sunitinib as tyrosine kinase inhibitor and everolimus for mTOR pathway inhibition) (57–59). However, despite the improvements in progression-free survival with these therapeutic options, their effect on overall survival is still controversial (60). Therefore, novel treatments are still required, especially for patients with advanced disease.

The incidence of metabolic syndrome is continuously increasing, reaching almost 35% in some countries, with a consequent increase in the prevalence of some types of cancer (61). T2DM has been also related to an increased risk of malignancies (62, 63) and is frequently developed in patients using everolimus and some somatostatin analogs (64, 65). In this context, metformin is one of the most widely prescribed oral hypoglycemic agents and it has received increased attention because of its potential antitumorigenic effects (5, 66). Likewise, some publications have described an inhibitory effect exerted by statins on tumor-induced angiogenesis and an antitumor effect in cellular and animal models of human cancer (22, 67). However, other studies have also suggested a potential risk of cancer when statins are used (68). In this sense, the current study, although using a limited number of samples, is, to our knowledge, the first in which (i) the association of T2DM with clinical evolution parameters in patients with different NET types (LCs and GEP-NETs) was assessed; (ii) the expression levels of all the components of two key regulatory systems (SST and ghrelin) in the tumor tissue of patients with these two NET types was evaluated in relation to T2DM and metformin treatment; and (iii) the effects of different biguanides and statins in key functional parameters were analyzed and compared in two representative models of NET cell lines: BON-1 and QGP-1 cells.

T2DM is linked to relevant defects in the INSR signaling pathway, which regulates growth and metabolic responses in insulin-targeted cells and tissues (69). Some epidemiological studies have described an increased risk for several types of cancer (i.e., breast, colon, rectum, liver, and pancreas) in patients with insulin resistance (70). Remarkably, in our cohort of patients, a more aggressive pattern in LCs (increased incidence of pleura invasion) and increased tumor size in GEP-NETs were observed in patients with T2DM when compared with patients without diabetes, suggesting a possible association between T2DM and NET pathophysiology.

Because insulin is related to increased risk of cancer, treatment options targeting this pathway could be effective in cancer prevention (71). Indeed, a meta-analysis showed a 31% reduction in overall cancer incidence and a 34% decline in cancer mortality in patients with diabetes treated with metformin (72). A retrospective study in patients with T2DM and NETs showed a lower recurrence rate in those treated with metformin compared with patients not treated with metformin or those without diabetes (72). Moreover, in a cohort of patients with PNETs who were receiving everolimus and octreotide LAR, progression-free survival was longer in patients treated with metformin compared with other drugs (73). To the best of our knowledge, no other specific reports in NETs have been published yet. In our cohort, metformin appeared to avoid weight loss in patients with LCs and T2DM. Interestingly, the numerical records assessed in patients with LCs receiving metformin suggested that tumors were smaller, incidence of metastasis was lower, and disease-free follow-up was longer than in patients not treated with metformin; however, these results did not reach statistical significance, likely due to the limited size of the groups. In contrast, no association was observed between clinical or histological variables and the use of metformin in the GEP-NETs group. We should underline that the main limitation of this work might be the limited number of patients with T2DM and those treated with metformin included in the analysis, although the size of the total cohort evaluated was large enough for making general comparisons. Therefore, like other studies reporting a limited cohort of samples (73), the results of this study should be interpreted with caution.

Novel mechanisms of action have been proposed for metformin in recent years. Among them, the induction of the expression of the glucagon-like peptide 1 receptor on pancreas β cells was described (74). Our report shows that the expression of several SSTR subtypes is reduced in NETs from patients with diabetes (in the LC and GEP-NET groups) compared with those from patients without diabetes, and, most importantly, that the overall expression of SSTR is significantly increased in LCs from patients with diabetes treated with metformin compared with LCs from patients with diabetes without metformin treatment. In fact, these expression levels of SSTR in LCs from patients with diabetes treated with metformin achieved the levels observed in LCs from patients without T2DM. These results provide suggestive evidence that metformin treatment could increase SSTR expression in NETs in patients with diabetes, which might be important from a clinical point of view, in that a previous study has suggested that metformin could have a potential synergistic effect when combined with somatostatin analogs via the inhibition of PI3K/AKT/mTOR axis (73). Thus, it will be worth elucidating the mechanisms involved in the capacity of metformin to regulate the expression of SSTRs as well as a putative synergistic effect between somatostatin analogs and metformin. In this sense, we analyzed the SSTR expression in BON1 and QGP1 cells after metformin treatment, but we did not observe changes in SSTRs mRNA expression levels, which could be in line with the idea that metformin could reverse the changes previously altered under diabetic conditions and maybe only have reduced potential to modulate basal expression of SSTRs. In addition, it has to be noted that pre- or cotreatment with biguanides and statins was not evaluated in this study, because the antiproliferative in vitro response to somatostatin analogs is limited in these NET cell lines (45, 75, 76).

Biguanides increase insulin sensitivity as well as glucose use by peripheral tissues (3). Antitumoral effects of metformin and phenformin have been evaluated in in vitro and in vivo studies, and metformin is also being tested as an adjuvant therapy to classic chemotherapeutic regimens (66, 77). Specifically, an earlier study showed that metformin inhibited cell proliferation in pancreatic, bronchopulmonary, and midgut NET cell lines in a dose-dependent manner, wherein these antitumoral effects appeared to be mediated via inhibition of mTORC1 signaling (12). Metformin also inhibits breast cancer cell growth in vitro in an AMPK-dependent manner in association with a decreased mTOR activation (78). In our study, we also observed a time-dependent antiproliferative effect of different biguanides in PNET cell lines. Similarly, by measuring other relevant functional end points (i.e., migration capacity and apoptotic rate), our study revealed that biguanides could exert additional, beneficial effects on NET cell function. These results support and extend previous data showing that metformin exerted antitumoral actions in vitro by modulating cell proliferation and apoptosis in breast cancer cells (79). However, we found that phenformin, but not metformin, increased apoptosis in both NET cell lines, which is partially in agreement with previous data indicating that apoptosis induced by metformin would differ depending on the NET cell type (12). We also found that metformin and phenformin decreased serotonin secretion in BON-1, but not in QGP-1, cells. Although the exact mechanisms are still to be elucidated, these results could be clinically relevant for patients with carcinoid syndrome, because elevated serotonin levels are directly associated with symptoms in this pathology (78). In this sense, we should remark that this is not the first time that different results have been observed in the functional response of BON-1 and QGP-1 cells (45, 48, 81, 82), which further emphasizes their potential distinct value to study the intrinsic heterogeneity of NETs. Indeed, the reason for these differences is still unknown but could be related to the distinct expression pattern of key regulatory systems (e.g., SST, ghrelin, IGF-I) (48, 81–83) and/or to the different activation or signaling of these NET cells in response to the same treatment as it has been previously observed, for instance, for SST analogs (i.e., octreotide and pasireotide) (48, 81).

Statins can also exert antitumoral actions. Thus, a phase II trial has reported a statin-induced antiproliferative effect in breast cancer (84). As well, the antiproliferative effect of statins has also been reported in several cancer cell lines, including cervical (85), leukemic natural killer (86), cholangiocarcinoma (87), and prostate (88). In line with these studies, we observed here that different statins exerted a clear antiproliferative effect in NET cells. In addition, we found that simvastatin significantly increased apoptosis levels in QGP-1 cells, an effect that has been described in cervical cancer, leukemia, natural killer, and cholangiocarcinoma cell lines (84–86).

It is well known that the PI3K/AKT/mTORC1 pathway exerts important roles in NETs pathogenesis (89). In LCs, metformin inhibited AKT, ERK, and mTOR pathways, suggesting that its antiproliferative effects can be both AMPK dependent and independent (90). In fact, Vlotides et al. (12) suggested that the functional effect of metformin is cell-type dependent; they reported that AMPK and AKT phosphorylation was elevated in pancreatic and midgut NET cell lines in response to metformin (after 48 hours of incubation), but this effect was not observed in bronchopulmonary neuroendocrine cells (12). Interestingly, it was also suggested that the inhibition of the mTOR pathway was associated to the induction of GSK3 phosphorylation following the ERK or AKT pathway. In our study, we observed an inhibition of phosphorylated AKT and ERK pathways after treating cells with biguanides (and also with simvastatin), which also reveals the AMPK-dependent and -independent effects of these drugs in NET cells. It should be mentioned that the differences between our results and those reported by Vlotides et al. (12) may be related to the drug-incubation period (8 minutes vs 48 hours). However, cell inhibition of the ERK pathway has been also reported in non–small lung cancer and cholangiocarcinoma cell lines with concomitant induction of apoptosis (87, 91).

The mechanisms linking T2DM and cancer are a most exciting and interesting research topic. It has been proposed that chronic hyperinsulinemia may promote the development of neoplasms via abnormal stimulation of multiple cellular signaling cascades by insulin, enhancing growth factor–dependent cell proliferation and/or modifying cell metabolism (66). In our study, we observed changes in the molecular expression of key genes involved in tumor aggressiveness (e.g., INSR, GLUT4) in response to metformin or phenformin, but not simvastatin, suggesting a putative modulatory effect of biguanides in these signaling pathways. In line with this, some studies have suggested that the antiproliferative effect of statins in cancer cell lines might be associated with cell cycle regulatory effects (84), epigenetic alterations (92), or with gene expression modifications of cancer signaling (93). However, the effects of simvastatin treatment on these regulatory end points, as well as whether metformin and simvastatin could have synergistic effects in NETs [which has been demonstrated in different tumor pathologies (94–96)], could not be evaluated in our study, but deserve further attention.

In sum, our study, using a limited cohort of patients, revealed a potential association between key clinical parameters of NET aggressiveness (e.g., incidence of pleura invasion or metastasis, tumor size) and the presence of diabetes and/or treatment with antidiabetic drugs in patients with different NET types (i.e., LCs and GEP-NETs). Moreover, this study provides evidence that the expression of multiple components of two key regulatory systems for the pathophysiology of NETs, the SST and ghrelin systems, are modulated in patients with diabetes with LCs and GEP-NETs compared with patients without diabetes. Finally, our results also showed that different biguanides and statins can directly exert clear antitumoral actions in NET cells, probably due to their effect on cell survival, cell migration, apoptosis, gene expression, and metabolic pathway modifications. Therefore, because metformin and statins are low-cost, commercially available drugs with a safe profile and large experience in their clinical use, our present results invite further exploration of their potential value as adjuvant therapy for the treatment of patients with NETs.

Abbreviations:

Abbreviations:
 
• AMPK

  AMP-activated protein kinase

 
• CORT

  cortistatin

 
• FBS

  fetal bovine serum

 
• FFPE

  formalin-fixed paraffin-embedded

 
• GEP

  gastroenteropancreatic

 
• GHRL

  ghrelin

 
• GLUT4

  glucose transporter

 
• GOAT

  ghrelin-o-acyltransferase enzyme

 
• INSR

  insulin receptor

 
• LC

  lung carcinoid

 
• mTOR

  mammalian target of rapamycin

 
• NET

  neuroendocrine tumor

 
• PNET

  pancreatic neuroendocrine tumor

 
• SST

  somatostatin

 
• T2DM

  type 2 diabetes

Acknowledgments

Financial Support: This work was funded by Instituto de Salud Carlos III [cofunded by European Union (ERDF/ESF, “Investing in your future”): PI16/00264, PI17/02287, CM17/00104, CP15/00156], MINECO (BFU2016-80360-R), Junta de Andalucía (BIO-0139, CTS-1406), GETNE Grant 2014 and CIBERobn. CIBER is an initiative of Instituto de Salud Carlos III, Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain.

Disclosure Summary: The authors have nothing to disclose.

Author notes

M.A.G.-M., J.P.C., and R.M.L. contributed equally to this study.