Medical Management of Gastroenteropancreatic Neuroendocrine Tumors: Current Strategies and Future Advances

SSA THERAPY

Somatostatin is a peptide hormone that binds to SSTR subtypes 1–5 and inhibits the secretion of other hormones such as serotonin, gastrin, vasoactive intestinal peptide, and glucagon (9–12). Because of its short half-life of several minutes, administration of somatostatin on an outpatient basis is not practical. Long-acting SSAs include octreotide long-acting release (LAR) and lanreotide. Both analogs bind primarily to SSTR subtypes 2 and 5 (13,14). SSAs were initially shown to alleviate hormonal symptoms such as flushing and diarrhea associated with carcinoid syndrome; necrolytic migratory erythema, cachexia, and hyperglycemia in glucagonoma syndrome; and severe watery diarrhea associated with vasoactive-intestinal-peptide–oma syndrome. SSAs were subsequently found to inhibit tumor growth despite very rare objective radiographic responses.

The phase III PROMID study was the first randomized trial to evaluate the antiproliferative effects of an SSA (15). The trial randomized 85 patients with midgut NETs to receive octreotide LAR 30 mg every 4 wk versus placebo, with time to progression as the primary endpoint. The study met its primary endpoint with significant improvement in time to progression (14.3 vs. 6 mo; hazard ratio [HR], 0.34; P = 0.000072). More recently, the phase III CLARINET study evaluated lanreotide depot 120 mg every 4 wk against placebo in hormonally nonfunctioning, SSTR-positive GEP-NETs with a Ki-67 index of less than 10% (16). In this trial, lanreotide was associated with a significantly prolonged progression-free survival (PFS) compared with placebo (median not reached vs. 18 mo; HR, 0.47; P = 0.001). Of note, the HR for PFS in the midgut population of the CLARINET trial was nearly identical to the HR for time to progression in the PROMID trial.

SSAs are generally well tolerated and associated with fewer side effects or risks than other antineoplastic therapies. Steatorrhea is one of the chief toxicities and can usually be managed effectively with pancreatic digestive enzymes. Cholelithiasis is a long-term complication caused by inhibition of physiologic gallbladder contraction. Mild hyperglycemia and bradycardia are other side effects but are rarely of clinical significance.

At this time, there are no convincing data to favor use of one SSA versus the other, and guidelines endorse use of either octreotide or lanreotide, both for syndrome and for tumor control. There is some evidence to suggest that escalation of the dose or frequency of SSA beyond label doses can be used to control refractory carcinoid syndrome: this strategy seems to be particularly useful when symptoms worsen toward the end of each 4-wk injection cycle (17,18). Patients with suboptimal syndrome control can also use rescue injections of short-acting octreotide. There is no evidence to support switching from one SSA to another at a time of radiographic progression. Because of their benign side effect profile and evidence of efficacy in multiple clinical trials, SSAs are often the first-line treatment of choice for well-differentiated NETs. Whether evidence of SSTR expression is required as a criterion for use is a question of some controversy.

TELOTRISTAT ETHYL

Telotristat ethyl is an oral inhibitor of tryptophan hydroxylase, the rate-limiting enzyme in serotonin biosynthesis. It was developed to treat refractory diarrhea related to carcinoid syndrome. The phase 3 TELESTAR study assessed the safety and efficacy of 2 doses of telotristat (250 and 500 mg, each taken 3 times daily) combined with SSA versus placebo plus SSA in patients with well-differentiated metastatic NETs who experienced uncontrolled diarrhea (≥4 bowel movements daily) on SSA (19). There was a statistically significant reduction in bowel movement frequency using telotristat compared with placebo averaged over a 12-wk double-blind period (−0.81 bowel movement per day for 250 mg [P < 0.001 vs. placebo], and –0.69 for 500 mg [P < 0.001]). A significant reduction in urine 5-hydroxyindoleacetic acid level was observed with either dose of drug (decreased by a mean of 40 and 57.7 mg per 24 h with 250 and 500 mg, respectively). The drug was well tolerated and safe overall. There was a slightly higher incidence of nausea and depression with the 500-mg dose of telotristat than with either the lower 250-mg dose or placebo. These results supported the efficacy and safety of telotristat, and a dosage of 250 mg 3 times per day was granted approval by the Food and Drug Administration for refractory carcinoid-syndrome–related diarrhea. There is no clear evidence that telotristat affects flushing, which is not thought to a serotonergic symptom, nor is there evidence that it inhibits tumor growth. It is possible that telotristat can inhibit progression of carcinoid heart disease, but clinical evidence for this is limited.

EVEROLIMUS

The mammalian target of rapamycin (mTOR) enzyme is a threonine kinase that is crucial in regulation of cell proliferation and metabolism. It also mediates signaling from growth factor receptors, including insulinlike growth factor receptor (20,21). Previous preclinical studies on pancreatic neuroendocrine cell lines (BON-1) have shown that mTOR inhibitors (rapamycin and everolimus) can decrease NET cell growth (22,23). Everolimus is an oral mTOR inhibitor that has been studied extensively in GEP‐NETs. On the basis of favorable single-arm data, several phase III trials were launched to evaluate everolimus in various NET populations. The RADIANT 3 study evaluated everolimus 10 mg versus placebo in patients with low‐ and intermediate‐grade pancreatic NETs (24). The primary endpoint was PFS, and results demonstrated a statistically significant PFS improvement from a median of 4.6 mo on the placebo arm to 11 mo on the everolimus arm (HR, 0.35; P < 0.001). The objective response rate (ORR) was only 5% on the everolimus arm (vs. 2% in the placebo arm). Although the study was not powered to evaluate OS, there was a mild and statistically insignificant trend toward OS benefit with everolimus in this population.

The RADIANT 2 study evaluated everolimus plus octreotide versus placebo plus octreotide in patients with advanced gastrointestinal and lung NETs (nonpancreatic) and with a history of carcinoid syndrome (25). Because of the latter eligibility criterion, the study predominantly enrolled a population of patients with relatively slow-growing midgut NETs. This study fell just short of meeting its endpoint of statistically significant improvement in PFS (HR, 0.77; P = 0.026, with prespecified statistical significance threshold of P < 0.024) and thus did not lead to Food and Drug Administration approval of everolimus in this population of patients. Moreover, subsequent survival analysis showed a nonsignificant, mild trend toward decreased OS with everolimus versus placebo (26).

The final RADIANT study, RADIANT 4, evaluated everolimus versus placebo in hormonally nonfunctional low- and intermediate-grade gastrointestinal and lung NETs with radiographic progression over 6 mo (27). Concurrent SSAs were prohibited. This trial enrolled a population of patients with relatively aggressive disease and met its primary endpoint of improvement in PFS (11 vs. 3.9 mo; HR, 0.48; P < 0.00001) with promising interim OS (HR, 0.64; P = 0.037). Everolimus received Food and Drug Administration approval for gastrointestinal and lung NETs based on results of the RADIANT 4 study.

The RADIANT 3 and RADIANT 4 trials showed that everolimus has efficacy across a wide spectrum of GEP-NETs and should be especially considered in patients with clinically significant disease progression. However, the results of the RADIANT 2 study suggest caution when prescribing the drug to patients with slow-growing midgut NETs and carcinoid syndrome.

The most common side effects of everolimus include oral aphthous ulcers, rash, diarrhea, hyperglycemia, hyperlipidemia, pneumonitis, and immunosuppression resulting in atypical infections. However, with dose reductions and appropriate supportive measures, such as a dexamethasone mouth rinse for oral ulcer prevention, most patients can tolerate everolimus for a long treatment period.

ANGIOGENESIS INHIBITORS

GEP-NETs are highly vascular cancers. They frequently express the vascular endothelial growth factor (VEGF) ligand and its receptors, and high levels of circulating VEGF are associated with tumor progression (28). The tyrosine kinase inhibitor sunitinib, which targets VEGF receptors 1, 2, and 3, as well as platelet‐derived growth factor receptor, was evaluated in a phase II trial of pancreatic and nonpancreatic (carcinoid) NETs. ORR in the pancreatic NET cohort was 16.7% (vs. 2% in carcinoid tumors) (29). These results led to a randomized phase 3 trial of sunitinib in patients with advanced, well-differentiated pancreatic NETs (30). In total, 171 patients were randomized to receive sunitinib 37.5 mg daily versus placebo. There was a significant PFS benefit with sunitinib, and there was an improvement in median PFS from 5.5 to 11.4 mo (HR, 0.42; P < 0.001), with an ORR of 9.3%. The study was not powered to show OS improvement; however, there was a nonsignificant trend toward improved OS with sunitinib (median OS, 38.6 mo for sunitinib vs. 29.1 for placebo; P = 0.094). The most documented side effects of sunitinib were diarrhea, nausea, vomiting, palmar–plantar erythrodysesthesia, fatigue, and hypertension.

Other antiangiogenic agents for management of GEP-NETs have been investigated in phase II and III trials. The phase III SWOG S0518 trial evaluated the efficacy of bevacizumab (anti-VEGF‐A antibody) and octreotide versus interferon and octreotide in GEP-NET patients (31). Despite promising phase II data, there was no significant improvement in PFS with bevacizumab (16.6 mo in bevacizumab arm vs. 15.4 mo with interferon; HR, 0.93; P = 0.55). Other antiangiogenic tyrosine kinase inhibitors, including pazopanib, axitinib, and sorafenib, have been investigated in several phase II trials (32,33). An ongoing prospective randomized phase II trial of pazopanib versus placebo has enrolled patients with progressive nonpancreatic NETs (NCT01280201). In addition to monotherapy, the combination of VEGF and mTOR inhibition has shown promising results. For example, a randomized study in 150 patients with progressive pancreatic NETs compared everolimus monotherapy with everolimus plus bevacizumab (34). Although there was remarkable improvement in response rate in the combination arm, with 31% ORR, there was only a modest PFS benefit (16.7 vs. 14 mo; P = 0.12).

At this time, with are no prospective studies comparing everolimus with sunitinib in pancreatic NET patients, a population for whom both drugs are approved for progressive disease. The results of the RADIANT 3 study and the phase III sunitinib study in pancreatic NETs suggest similar benefit and tolerability. Biomarkers predicting response to either everolimus or antiangiogenic drugs have not been identified. At this time, patient comorbidities may dictate the choice of treatment. For example, sunitinib is relatively contraindicated in patients with uncontrolled hypertension or cardiovascular disease, and use of everolimus is discouraged in patients with advanced diabetes.

CYTOTOXIC CHEMOTHERAPY IN WELL-DIFFERENTIATED NETS

Streptozocin-based chemotherapy regimens have been a standard treatment of pancreatic NETs for several decades (35). In recent years, the oral alkylating drug temozolomide has been found to be active in a variety of NETs, particularly pancreatic NETs. One retrospective analysis of temozolomide in a heterogenous population of NETs demonstrated an ORR of 14% (36). A prospective phase II study of temozolomide plus thalidomide demonstrated a response rate of 45% in pancreatic NETs versus 7% in carcinoid tumors (37). Another phase II study of temozolomide combined with bevacizumab yielded similar results: an ORR of 33% in pancreatic NETs versus 0% in carcinoid tumors (38).

The combination of temozolomide and capecitabine was shown to be synergistic in cell line studies (39). On the basis of promising phase I data, an analysis of temozolomide–capecitabine in 30 consecutively treated pancreatic NET patients reported an ORR of 70% and median PFS of 18 mo (40). On the basis of these results, an Eastern Cooperative Oncology Group trial (ECOG 2211) randomized 144 patients with progressive pancreatic NETs to received temozolomide monotherapy versus temozolomide–capecitabine, with PFS as the primary endpoint (41). The results showed that the temozolomide–capecitabine regimen was associated with significant improvement in PFS (median PFS was 22.7 mo for temozolomide–capecitabine vs. 14.4 mo for temozolomide; HR, 0.58; P = 0.023) and OS (median OS not reached for temozolomide–capecitabine vs. 38.0 mo for temozolomide; HR, 0.41; P = 0.012). Overall, the temozolomide–capecitabine regimen is relatively tolerable, with cytopenias, particularly thrombocytopenia, representing the most clinically significant grade 3 or 4 toxicities. There is significant controversy as to whether expression of O⁶-methyl-guanine-methyl-transferase predicts response to temozolomide-based chemotherapy (42).

The ECOG 2211 trial and single-arm data indicate that temozolomide–capecitabine can be considered a standard-of-care regimen in pancreatic NETs. The activity of this regimen in nonpancreatic NETs is almost certainly inferior but has yet to be clearly established. Intravenous streptozocin–based regimens, such as streptozocin plus 5-fluorouracil, streptozocin plus doxorubicin, or combinations of all 3 drugs, remain a valid option for pancreatic NETs (43).

CYTOTOXIC CHEMOTHERAPY IN POORLY DIFFERENTIATED NECA

Poorly differentiated NECAs are treated similarly to small cell lung cancer, with etoposide plus platinum (either cisplatin or carboplatin) as the standard first-line regimen. This combination has been used to induce response rates of about 45%–60%, but with a short duration of response (8–9 mo), median survival rates of less than 2 y, and significant toxicity (44,45). With the lack of supportive literature for the standard of care, most patients who progress after a platinum/etoposide regimen will have limited therapeutic options. There are small trials and case series suggesting that regimens used in gastrointestinal malignancies (i.e., 5-fluorouracil plus oxaliplatin or 5-fluorouracil combined with oxaliplatin and irinotecan) are active as well (46,47). Unfortunately, preliminary data from immunotherapy trials suggest that poorly differentiated NECAs are less immunosensitive than small cell lung cancer (48,49).

RADIOLABELED SSA THERAPY

Radiolabeled SSA therapy (also known as peptide receptor radionuclide therapy [PRRT]) is an emerging form of targeted radiotherapy. Radiolabeled SSAs consist of a radionuclide isotope, an SSA (peptide), and a chelator that binds them, enabling the delivery of radioactive isotopes to SSTR-expressing tumors. Octreotide or the modified octapeptide octreotate are the most widely used SSAs, and DOTA is most commonly used as the chelating molecule. PRRT activity correlates with levels of SSTR uptake on SSTR imaging (50).

Data from early studies of PRRT using octreotide radiolabeled with high doses of Auger-electron–emitting ¹¹¹In showed evidence of symptom palliation but low rates of radiographic response (51,52). Subsequent single-arm studies showed much higher radiographic response rates using SSAs radiolabeled with the β-emitters ⁹⁰Y or ¹⁷⁷Lu and encouraging durations of median PFS (53). Long-term toxicity analyses have demonstrated a roughly 2.5% risk of myelodysplastic syndrome or acute leukemia (54,55). However, these single-arm studies were mostly large institutional series rather than prospective phase II studies and often lacked strict eligibility criteria, intention-to-treat data analysis, and independent radiographic review.

The phase III NETTER-1 was the first randomized, prospective trial of a radiolabeled SSA (56). In total, 231 patients with SSTR-expressing midgut NETs progressing on standard-dose octreotide were randomized to receive ¹⁷⁷Lu‐DOTATATE (4 doses of 7.4 GBq every 8 wk) combined with standard-dose octreotide LAR (30 mg) or high‐dose octreotide LAR (60 mg). The primary endpoint was PFS. The study met its primary endpoint with a 79% reduction in risk of progression or death (median PFS not reached vs. 8.4 mo; HR, 0.21; P < 0.00001). ORR was significantly higher in the ¹⁷⁷Lu-dotatate group than in the control group (18% vs. 3%; P < 0.001). Interim analysis of OS, at the time of PFS analysis, demonstrated preliminary improvement of OS in the PRRT arm (HR, 0.4; P = 0.004).

Overall, treatment was well tolerated, with the most documented side effects consisting of grade 1–2 nausea and vomiting, attributable to commercial amino acids used for renal prophylaxis. There was no excess nephrotoxicity observed in the ¹⁷⁷Lu-DOTATATE arm. Treatment with ¹⁷⁷Lu-DOTATATE was also associated with a significant delay in time to deterioration in key quality-of-life domains, including global health, physical functioning, role functioning, diarrhea, pain, and fatigue (57). The NETTER-1 results and single-arm registry data led to the approval of ¹⁷⁷Lu-DOTATATE for treatment of advanced GEP-NETs.

LIVER-DIRECTED THERAPY

The liver is the dominant site of metastases for GEP-NETs, and liver-directed treatments, including cytoreductive surgery and hepatic arterial embolization, have been used for decades despite limited prospective trial data.

Cytoreductive surgery can be palliative for patients with symptomatic metastases and is associated with favorable long-term survival outcomes (58,59). Traditionally, the ability to resect 90% of tumors has been considered a critical eligibility criterion. Decisions regarding the surgical resectability of disease are often made early in the course of treatment. Data supporting surgical cytoreduction are exclusively retrospective, and consequently the level of evidence supporting this approach is inherently limited.

Liver embolization therapy relies on the fact that metastatic tumors in the liver derive their blood supply primarily from the hepatic arterial circulation, whereas normal liver parenchyma is supplied primarily from the portal venous circulation. Embolization of microparticles without chemotherapy (bland embolization), or admixed with chemotherapy (chemoembolization), have both been described extensively in the NET literature. Evidence supporting liver embolization originates primarily from retrospective institutional series, which demonstrate ORRs of roughly 50% and an even higher rate of symptomatic improvement. Median durations of PFS are generally in the 12- to 24-mo range (60).

Liver embolization is primarily recommended for patients with progressive, unresectable, bilobar liver metastases. There are no completed randomized prospective trials comparing various embolization modalities; the currently open randomized RETNET study is comparing bland embolization versus chemoembolization and drug-eluting bead embolization for treatment of NETs metastatic to the liver (NCT02724540).

A more recent liver-directed approach consists of embolization of ⁹⁰Y glass or resin microspheres to liver metastases. This approach is also known as selective internal radiotherapy (61). Although initial data showed promising response rates and favorable short-term toxicities, long-term follow-up has demonstrated relatively high rates of radioembolization-induced liver disease, which can lead to jaundice, ascites, and progressive hepatic dysfunction (62).

SELECTION AND SEQUENCING OF THERAPIES

The development of randomized prospective studies has led to approval of multiple new drugs for various indications within the GEP-NET field (Table 1). In nearly all randomized studies, the control arm has consisted of a placebo or a nonstandard treatment. As a result, there have been few studies comparing 2 active drugs, and there has been an absence of high-level evidence regarding selection and sequencing of treatments. At this time, some treatment recommendations can be based only on cross-trial comparisons.

For nearly all well-differentiated SSTR-expressing GEP-NETs, first-line systemic treatment should consist of an SSA. This recommendation is endorsed by guidelines (63) and is based on the exceptional tolerability and safety of octreotide and lanreotide, as well as a high level of evidence for their antiproliferative effect as demonstrated in the PROMID and CLARINET trials.

Treatment selection is more complicated beyond the first-line setting. For midgut NETs, high-level evidence exists for the use of ¹⁷⁷Lu-DOTATATE based on the NETTER-1 study. There is weaker evidence to support use of everolimus based on negative results from the RADIANT 2 study, which enrolled primarily patients with hormonally functioning midgut NETs. Liver embolization remains an important treatment modality for patients with liver-dominant midgut NETs, despite lack of high-level trial evidence. Use of radioembolization should probably be limited in this population, particularly in light of likely overlap in toxicity with ¹⁷⁷Lu-DOTATATE and the potential for long-term radiation-induced liver disease.

For gastrointestinal NETs not in the midgut, evidence for use of everolimus is high, based primarily on the results of the phase III RADIANT 4 study. However, outcomes with ¹⁷⁷Lu-DOTATATE based on single-arm data are also encouraging, and ultimately prospective clinical trials should compare radiolabeled SSAs with everolimus in this population. One such trial, the COMPETE study, is investigating everolimus versus ¹⁷⁷Lu-DOTATOC in nonfunctioning SSTR-positive GEP-NETs and should provide much-needed data comparing efficacy and toxicity of 2 active treatment regimens (NCT03049189).

The systemic treatment options for progressive pancreatic NETs are even more diverse. High-level evidence now supports the use of everolimus, sunitinib, and capecitabine plus temozolomide, whereas single-arm studies support the use of ¹⁷⁷Lu-DOTATATE in this population (64). The results of the ECOG 2211 study demonstrate a median PFS of nearly 2 y in patients with progressive pancreatic NETs treated with capecitabine and temozolomide, an outcome unmatched in randomized clinical trials evaluating similar populations. Nonetheless, randomized clinical trials are needed to compare capecitabine/temozolomide with other standard regimens such as everolimus or sunitinib in order to begin to establish evidence-based treatment-sequencing guidelines.

More data are also needed to establish predictive markers in order to appropriately match patients with treatments. At this time, PRRT is the only treatment with a predictive clinical biomarker: SSTR expression on somatostatin-receptor imaging. A predictive blood marker for ¹⁷⁷Lu-DOTATATE response has also been developed and should be validated in additional studies (65). The role of O⁶-methyl-guanine-methyl-transferase expression for prediction of response to capecitabine/temozolomide will be evaluated in the ECOG 2211 study. Predictive markers for everolimus therapy, including mutations of mTOR pathway enzymes, have not been validated.

FUTURE DIRECTIONS

PRRT remains an important area of investigation for patients with GEP-NETs. Retrospective studies evaluating PRRT in select high-grade tumors have shown encouraging results, but prospective studies are needed. SSTR antagonist (as opposed to agonist)–based PRRT has been evaluated in several small, early-phase studies with mixed results (66,67). New classes of isotopes, such as α-emitters, have been evaluated in only small institutional series, and formal prospective trials are needed to assess the efficacy of α-emitting radiolabeled SSAs (68). Finally, the role of fixed-dose versus dosimetrically calculated administration of PRRT has yet to be evaluated in a randomized trial.

Other novel therapies also exploit high levels of SSTR expression in GEP-NETs. A miniaturized antibody–drug conjugate, PEN-221, has been developed for treatment of NETs, and phase II studies are ongoing (NCT02936323). Bispecific antibodies targeting SSTR2 and CD3 represent another novel form of therapy (69). Finally, chimeric antigen receptor T-cell technology targeting SSTR is in early phases of preclinical development.

Immunotherapy has radically altered the landscape of cancer treatment, but GEP-NETs appear to be relatively resistant to immunotherapy with PD-1 inhibitors, possibly because of their low mutational burden (48,49). Further studies evaluating inhibition of additional checkpoints, as well as combination studies evaluating immune sensitizers, are warranted. Studies evaluating the combination of PRRT with immunotherapy are currently open, but interpretation of results of single-arm studies may be difficult.

DISCLOSURE

No potential conflict of interest relevant to this article was reported.