Surgical Management of Neuroendocrine Tumor Liver Metastases

INTRODUCTION

Neuroendocrine tumors (NETs) are a group of heterogeneous neoplasms arising from cells of the neuroendocrine system. The term neuroendocrine neoplasia encompasses both well-differentiated NETs and poorly differentiated neuroendocrine carcinomas. NETs produce neurosecretory granules, express characteristic neuroendocrine differentiation markers, and secrete vasoactive substances responsible for carcinoid syndrome. More than one-third of patients with NETs present with distant disease, and the liver is the most common site of distant metastasis.^1,2 Patients with NET liver metastases (NETLMs) can have significant hormonal symptoms and worse outcomes compared with patients with isolated locoregional disease.

Treatment of NETs therefore should include effective management of NETLMs, and multiple treatment modalities are available depending on primary site, disease extent, and tumor characteristics. There are 3 categories of treatment options: hepatic resection and cytoreduction, nonsurgical liver-directed therapies, and systemic therapies. This article reviews the epidemiology, presentation, diagnosis, and management of NETLMs.

EPIDEMIOLOGY

The incidence of NETs has increased from 1.09 per 100,000 persons in 1973 to 6.98 per 100,000 persons in 2012,² in part because of increased imaging, as well as increased screening and surveillance endoscopy. NET prevalence increased from a 20-year limited-duration prevalence of 0.006% in 1993 to 0.048% in 2012 as a result of increased incidence, the relatively indolent course of these tumors, and improvements in treatment.² Approximately 12% to 27% of all patients with NETs present with distant metastasis,^2–4 and the liver is the most frequent site for NET metastasis regardless of primary site.¹ The proportions of patients with pancreatic, cecal, colonic, and small bowel NETs presenting with distant disease are 64%, 44%, 32%, and 30%, respectively.³ Metastatic disease negatively affects survival, and patients presenting with distant metastases have a 4-fold increased risk of death compared with those with localized disease.²

CLINICAL PRESENTATION

Clinical presentation of NETs depends on their primary site, tumor functionality, and patterns of metastasis. Functional small bowel NETs (SBNETs) can secrete bioactive amines such as serotonin, histamine, kallikrein, and tachykinins, causing carcinoid syndrome.⁵ Carcinoid syndrome involves a classic triad of diarrhea, flushing, and wheezing. Less common symptoms include valvular heart disease and telangiectasias.⁶ Serotonin is thought to cause fibrotic endocardial thickening that distorts the tricuspid and pulmonic valves, leading to valvular regurgitation or stenosis. When severe, carcinoid heart disease causes right-sided heart failure.⁵ The liver often inactivates bioactive products released by functional NETs, so patients with carcinoid symptoms have a large tumor burden or liver metastases whose secretory products bypass hepatic inactivation.⁵ Pancreatic NETs (PNETs) are nonfunctional in more than 70% of cases, but may also make several hormones, including gastrin, which results in Zollinger-Ellison syndrome and severe ulcer diathesis; insulin, which causes recurrent hypoglycemia; vasoactive intestinal peptide (VIP), which causes hypokalemia, achlorhydria, and watery diarrhea (Verner-Morrison syndrome); glucagon, which leads to hyperglycemia, stomatitis, weight loss, and a migratory necrolytic rash; somatostatin, associated with diabetes, steatorrhea, diarrhea, and cholelithiasis; adrenocorticotrophic hormone (ACTH), producing Cushing syndrome; and parathyroid hormone-related peptide (PTHrP), leading to severe hypercalcemia.⁷

DIAGNOSIS

Diagnosis of NETLMs involves multiple modalities, including biochemical testing, imaging, and pathologic examination. In patients with classic carcinoid symptoms, biochemical testing can confirm a diagnosis of carcinoid syndrome. A 24-hour urinary collection for 5-hydroxyindoleacetic acid (5-HIAA), a product of serotonin metabolism, is the preferred method for carcinoid syndrome diagnosis,⁸ although the test is cumbersome and difficult to collect. Blood tumor markers are preferred for monitoring disease progression, recurrence, and response to treatment. Tumor biomarkers used for monitoring NETs include chromogranin A, pancreastatin, neurokinin A, pancreatic polypeptide, substance P, and neuron-specific enolase.⁹ Chromogranin A is the most commonly monitored biomarker, although recent data suggest pancreastatin is more sensitive, specific, and accurate for detecting disease progression.^10,11 Hormones can also serve as biomarkers in functional tumors, such as gastrin, insulin, and C peptide, VIP, glucagon, somatostatin, ACTH, or PTHrP.⁷

In patients without classic hormonal symptoms, imaging for abdominal pain or unrelated indications may reveal hepatic lesions that, on biopsy, reveal the diagnosis of NETLM. Work-up and staging of NETs may use anatomic and/or functional imaging. Anatomic imaging includes computed tomography (CT), MRI, and ultrasonography. Functional imaging uses radiolabeled somatostatin analogues, such as ¹¹¹indium pentetreotide scintigraphy (Octreoscan) and ⁶⁸gallium PET-CT (DOTATATE, DOTATOC, or DOTANOC). NETs with somatostatin receptors take up these radiolabeled somatostatin analogues, facilitating tumor localization and staging (Fig. 1A).

Fig. 1.

(A) Coronal ⁶⁸Ga-DOTATOC PET/CT image of numerous neuroendocrine liver metastases (white arrow) from a primary small bowel NET (red arrow). (B) Axial CT image of a hypodense neuroendocrine liver metastasis (white arrow). (C) Axial T1-weighted MRI of multifocal neuroendocrine liver metastases (white arrows). Neuroendocrine liver metastases appear hypointense on T1-weighted MRI. (D) Axial T2-weighted MRI of multifocal neuroendocrine liver metastases (white arrows). Neuroendocrine liver metastases appear hyperintense on T2-weighted MRI.

CT is often the initial imaging tool used and helps identify the primary tumor and characterize nodal metastases. However, MRI is more sensitive for detecting hepatic metastasis because of its high contrast resolution.^12,13 Sensitivities for MRI, CT, somatostatin scintigraphy using indium octreotide, and ⁶⁸Ga-DOTATATE PET-CT for detecting well-differentiated gastroenteropancreatic NETLMs are 95%, 79%, 49%, and 81%, respectively.^12,14 NETs tend to be hypervascular and are best seen on arterial-phase imaging with intravenous contrast. These metastases typically show bright enhancement on arterial-phase CT (Fig. 1B), hypointensity on unenhanced T1-weighted MRI (Fig. 1C), and hyperintensity on T2-weighted MRI (Fig. 1D).¹⁵

In addition to imaging, pathologic examination of liver biopsy specimens confirms an NET diagnosis. Immunohistochemistry (IHC) staining for neuroendocrine markers, such as synaptophysin and chromogranin A, as well as staining for keratin and somatostatin receptors distinguishes NETLMs from other neoplasms.¹⁶ Pathologic examination should include quantification of Ki-67 proliferative index, which determines the tumor’s grade. When the primary site remains unknown despite imaging and endoscopy, extended IHC panels or gene expression classifiers can distinguish small bowel from pancreatic NETs.¹⁷ Positive CDX2 staining suggests a midgut primary, whereas positive PAX6 or ISL1 staining suggests a pancreatic primary.¹⁸

HEPATIC RESECTION

Resection

Resection or cytoreduction is the preferred first-line treatment of NETLMs, when feasible. Resection or cytoreduction is associated with improved survival, offers relief from hormonal symptoms, and prevents sequelae of carcinoid syndrome, such as carcinoid heart disease. Randomized prospective data are lacking, but retrospective studies suggest that patients who undergo NETLM resection have better symptom relief and prolonged survival compared with patients not undergoing resection. In patients who can be optimally cytoreduced (>90% tumor burden), resection improves symptoms in up to 95% of patients,^19,20 whereas medical treatment alone improves symptoms in 25% to 80% of patients.^20,21 Although comparisons between nonrandomized surgical and medical studies are highly susceptible to the effects of selection bias and carry many caveats, resection of NETLMs is associated with reported 5-year overall survival (OS) rates of 61% to 74%,^19,20,22,23 whereas series of medical NETLM treatment report 5-year OS rates of 25% to 67%.^20,24,25

Complete resection is the only treatment that offers a potential for cure, although cure is rare. In general, complete resection is only achievable in about 44% to 53% of patients and almost all patients recur after resection.^19,23 Mayo and colleagues²² found a 5-year recurrence rate of 94% in their series of 339 patients that had resection of NETLMs. Sarmiento and colleagues¹⁹ reported a similar 5-year recurrence rate of 84%, and, even after apparent R0 resections (n = 75), the 5-year recurrence rate was 76%. This finding is caused in part by underestimation of tumor burden by preoperative imaging. Elias and colleagues²⁶ compared the number of tumors present in 5-mm slices of hepatectomy specimens with those detected by preoperative imaging in patients undergoing hepatectomy for NETLMs (n = 37). The investigators found that 50% of NETLMs discovered on pathologic examination were not detected by preoperative imaging or intraoperative exploration, which used manual palpation and ultrasonography. Half of patients had hepatic metastases smaller than 2 mm, and the accuracy of somatostatin receptor scintigraphy, CT, and MRI in detecting NETLMs was only 24%, 38%, and 29%, respectively.²⁶ These data show that unrecognized small metastases are present in most patients, and a realistic goal for resection should be to control, rather than to eliminate, disease.

Resection Thresholds

The optimal thresholds for cytoreduction of NETLMs have not been established. With regard to resection margin, more complete resection is associated with improved progression-free survival (PFS) in some series.^19,23 Sarmiento and colleagues¹⁹ found that complete (R0) resection was associated with longer PFS compared with incomplete resection (5-year recurrence rate 76% vs 91%, median PFS 30 vs 16 months, P<.001). In the series by Elias and colleagues,²³ the percentages of patients without progression at 5 years for R0, R1, and R2 resections were 66%, 46%, and 30%, respectively.

In other studies, margin status was not clearly associated with risk of death or progression.^22,27,28 For example, Mayo and colleagues²² found that margin status was not associated with a statistically significant difference in recurrence, although R0/R1 resection was associated with improved OS compared with R2 resection. Glazer and colleagues²⁷ did not find a significant correlation with margin status and risk of death, and Graff-Baker and colleagues²⁸ found that R0 resection was not associated with disease-specific survival compared with R2 resection. The existence of unrecognized low-burden disease even in patients with apparently R0 resections likely explains such findings. Given these data, the North American Neuroendocrine Tumor Society (NANETS) recommends pursuing surgical cytoreduction even when complete resection cannot be obtained, because removal of most of the gross disease can improve symptoms and extend survival.²⁹

The optimal threshold of expected cytoreduction to justify an attempt at R2 resection is controversial and continues to evolve. Historically, liver-directed surgery was pursued when at least 90% of liver metastases could be resected. This threshold is attributed to the study by McEntee and colleagues³⁰ in 1990, but as early as 1977 Foster and Berman³¹ and Foster and Lundy³² suggested that 95% cytoreduction of NETLMs was necessary to palliate symptoms in patients with NETLMs. The notion of a threshold for cytoreduction does not appear in the body of the McEntee and colleagues³⁰ article, but the discussion introduces it, where the investigators note that patients only experienced symptom relief when 90% of visible tumors were removed.²⁹ Que and colleagues³³ (n = 74) and Sarmiento and colleagues¹⁹ (n = 170) used this 90% debulking threshold in their series, finding that resection of NETLMs improved symptoms and had favorable outcomes compared with previous studies of patients with unresected NETLMs.

However, recent data suggest that lower debulking thresholds still offer improvement in symptoms and are associated with prolonged survival. In a series by Chambers and colleagues,³⁴ a 70% or greater reduction of tumor burden effectively palliated carcinoid syndrome symptoms. Graff-Baker and colleagues²⁸ also used this 70% threshold and compared patients (n = 52) based on the percentage of gross NETLMs resected: 70% to 89%, 90% to 99%, and 100%. Importantly, percentage of disease resected did not correlate with rate of progression or Kaplan-Meier estimates of PFS or disease-free survival. Median liver PFS was 71.6 months and 5-year disease-specific survival was 90%. These outcomes were not significantly improved with increased resection percentage.

Maxwell and colleagues³⁵ also found associated oncologic value with this 70% threshold in a study of 108 patients with small bowel and pancreatic NETLMs. This series included patients who had a wide range of percentage of NETLMs debulked, from less than 50% to greater than 90%. Patients who had greater than or equal to 70% cytoreduction of NETLMs had improved PFS (median PFS 3.2 vs 1.3 years, P<.001) and OS (median OS not reached vs 6.5 years, P = .009) compared with patients who had less than 70% cytoreduction. Patients who had greater than or equal to 90% cytoreduction did have prolonged PFS compared with those who had less than 90% cytoreduction (median PFS 3.8 vs 1.5 years, P<.001), but no difference in OS (median OS not reached vs 6.5 years, P = .12). This series also showed that parenchymal- sparing procedures, such as ablation and enucleation, could achieve the 70% cytoreduction threshold, resulting in survival outcomes comparable with those from resection.

In an update and expansion of this series, Scott and colleagues³⁶ found that this 70% threshold could be reliably obtained in patients with 1 to 5, 6 to 10, or greater than 10 hepatic lesions treated (n = 188; 128 SBNETs, 41 PNETs, and 19 other primary sites). Furthermore, there was no significant difference in PFS in the 3 groups (median PFS 23.3 vs 21.6 vs 21.8 months, P = .55) or OS (median OS 80 vs 131.5 vs 89 months, P = .55). There were no differences in biochemical response in tumor markers (defined by reduction in hormone levels by >50%, which occurred in 69% of patients), and complication rates between these groups. Grade 3 tumors were associated with worse OS compared with grade 1 (hazard ratio [HR], 11.69; P<.01). Overall, greater than or equal to 70% cytoreduction was achieved in 79% of patients, which was determined through careful comparison of preoperative and postoperative anatomic imaging studies. This outcome could be achieved more often in patients with less than 45% liver replacement and when there were fewer than 30 lesions. The median OS was 37.6 months for the less than 70% cytoreduction group, 134.3 months for the 70% to 90% group, and was not reached in those with greater than 90% cytoreduction (P<.01 for <70% vs 70%−90% cytoreduction; P = .6 for 70%−90% vs >90% cytoreduction). Corresponding PFS was 10.8 months, 20.6 months, and months, respectively, for these 3 categories of cytoreduction (P<.01 for each comparison). This study showed that cytoreduction was possible in most patients, even when many lesions were present, and that, although PFS improved with increasing levels of cytoreduction, OS was similar whether 70% to 90% or greater than 90% was achieved.

Morgan and colleagues³⁷ studied the potential value of the 70% threshold in a study of 44 patients with duodenal and pancreatic NETs, which found no significant differences in progression rates between patients who underwent greater than or equal to 70% debulking versus greater than or equal to 90% debulking. Median liver PFS was much lower in these tumors from this primary site (11 months) than their previous reports on SBNETs (71.6 months),²⁸ but 5-year OS was still high at 81%. Only size greater than 5 cm of the largest resected liver metastasis and formal hepatic resection correlated with rates of liver progression.

Guidelines now recommend considering resection when a 70% debulking threshold is possible.^29,38 Patients with numerous NETLMs, Ki-67 less than 20%, and extrahepatic disease may still be candidates for resection if at least 70% of their NETLMs can be cytoreduced based on careful preoperative imaging evaluation.³⁶ Guidelines also recommend parenchymal-sparing procedures when possible. Superficial metastases can be enucleated or resected with wedge resections, whereas radiofrequency or microwave ablation can reach deeper lesions, without sacrifice of large volumes of normal liver that anatomic resections might require.³⁵ Patients with poor performance status, significant comorbidities, severe hepatic dysfunction, or untreated carcinoid heart disease may not benefit from surgery. Patients with high-grade NETLMs or extensive liver replacement with tumor (tumor burden >50%−70%) are also unlikely to benefit from resection.^29,38

Resection of Primary Tumors with Unresectable Metastases

Whether to resect the primary tumor in patients with unresectable metastases depends on the site, functional status, and symptoms from the primary. In patients with SBNETs, the primary tumor can cause obstruction, mesenteric ischemia, mesenteric fibrosis, and gastrointestinal bleeding. Resection of the small bowel primary can relieve or prevent these symptoms³⁴ and may improve survival even with unresectable NETLMs. A systematic review of 6 retrospective studies found that resection of the primary in midgut NETs was associated with a trend toward improved survival compared with nonsurgical management of the primary tumor (median OS, 75–139 vs 50–88 months).³⁹ In comparison, Daskalakis and colleagues⁴⁰ performed propensity matching in their retrospective series and concluded that prophylactic resection of asymptomatic primary tumors in patients with stage IV SBNETs was not associated with improved survival compared with no surgery or surgery delayed for greater than 6 months. However, the fact that 58% of patients in the delayed surgery group eventually underwent surgery makes it difficult to conclude that there was no benefit to surgery. Other retrospective series showing benefit are affected by selection bias, so this remains an unanswered question.

Resection of PNETs can improve hormonal symptoms and may extend survival. Franko and colleagues⁴¹ found that patients with nonfunctional PNETs and distant metastases (n = 614) had an associated prolonged survival with resection of the primary tumor compared with no surgical treatment (median survival 8.4 vs 1.1 years, P<.001). A recent study evaluating the benefit of peptide receptor radionuclide therapy (PRRT) alone versus resection of the PNET followed by PRRT found longer survival of the latter group (median of 140 vs 58 months in 335 patients; HR, 2.91).⁴² Because there are no randomized data regarding the removal of primary PNETs in the setting of unresectable NETLMs, and selection bias is present in retrospective series, these results are difficult to interpret. Guidelines recommend evaluating patients’ symptoms, overall health, and tumor characteristics, such as grade and the location of the pancreatic tumor (distal pancreatectomy is less morbid than a Whipple procedure), and potential to improve response to PRRT.³⁸

Other Surgical Options

Cytoreduction with heated intraperitoneal chemotherapy (HIPEC) is used in the treatment of appendiceal, colorectal, and mesothelial neoplasms with peritoneal metastases.^43–46 A retrospective French study compared patients with NET peritoneal metastases (n = 41) treated with cytoreduction and HIPEC versus those treated with cytoreduction alone; the 28 HIPEC plus cytoreduction patients were treated from 1994 to 2007, and the 13 cytoreduction-only patients from 2008 to 2012, after the authors had stopped doing HIPEC in these patients.⁴⁷ The 2 groups had similar 2-year OS of 81% and 73% (P = .73), and liver recurrence rates (2-year liver metastasis-free survival rates of 61% vs 38%, respectively; P = .12). The study had several significant limitations, including short follow-up time, heterogeneous treatment groups, and treatment of cohorts at different time periods. Current guidelines from neuroendocrine and peritoneal surface malignancy societies do not support HIPEC for treatment of NET peritoneal metastases.^29,48

The impact of liver transplantation is unknown and its role in the management of NETLM remains controversial. In 1 retrospective European study of 213 patients undergoing liver transplant for NETLMs, median OS was 67 months and PFS was 24 months.⁴⁹ Postoperative mortality was 10%, and 11% of patients had to be retransplanted during the first 3 months.⁴⁹ A systematic review found that recurrence after liver transplant for NETLMs ranged from 31% to 57%.⁵⁰ Given the high rate of recurrence, patient selection is paramount with the understanding that disease-free survival is likely to be limited. Patients considered for transplant should have low-grade tumors, stable disease, their primary tumor removed before transplant, and absence of extrahepatic disease.⁵¹

Perioperative Considerations

Patients with carcinoid heart disease have increased operative complications, and significant valvular disease should be treated before surgery.^29,52 Even in the absence of hormonal symptoms of carcinoid syndrome, physicians should prepare for possible carcinoid crisis, which is hemodynamic instability thought to be caused by release of vasoactive substances resulting from anesthesia or tumor manipulation.⁵³ Features of carcinoid crisis include hypotension, flushing, chest pain, pruritus, and paresthesias. The incidence of carcinoid crisis is estimated between 3% and 35% in operations for SBNETs, and octreotide has been used to prevent carcinoid crisis.^54,55 The hormonal basis of carcinoid crisis has not been proved, and some investigators question the role of perioperative octreotide.^54,56 Nevertheless, octreotide may decrease the incidence of carcinoid crisis, the drug is inexpensive, and it does not seem to increase operative complications.^52,55 It is our practice to infuse octreotide at 100 μg/h throughout operations for NETLMs and additional fluids and vasopressors are provided if necessary. We wean the octreotide infusion postoperatively by 25 μg/h every 6 to 8 hours.⁵⁷

LIVER-DIRECTED THERAPY

Ablation

Ablation can be performed during laparotomy for cytoreduction, laparoscopically, or percutaneously by interventional radiology. Ablative techniques include radiofrequency ablation, microwave ablation, cryotherapy, and irreversible electroporation. As an adjunct to resection, ablation can minimize the loss of normal liver tissue and allows improved cytoreduction by destroying tumors not amenable to resection. Intraoperative ultrasonography guides identification of lesions and ablation probe placement.^35,58 Smaller tumors (<3 cm in diameter) are more amenable to ablation and have decreased likelihood of local recurrence.⁵¹ Ablation is not recommended for tumors near vital structures or tumors greater than 5 cm in diameter because of increased risk of complication and recurrence, based on studies of hepatocellular carcinoma and colorectal liver metastases.^59,60

Prospective data are lacking, but ablation alone seems to be effective for palliation of symptoms and may provide some survival benefit. Akyildiz and colleagues⁶¹ reported in a retrospective study of 89 patients that ablation of NETLM improved hormonal symptoms in 97% of patients for a median of 14 months. Disease-free survival in that study was 15 months, with 60% of patients developing new liver or extrahepatic metastases, and OS was 6 years.⁶¹ The 5-year survival rate was 57%, which is lower than the 5-year survival rate of 74% reported in the study by Mayo and colleagues²² of patients undergoing surgical resection and ablation.⁶¹ Ablation seems relatively safe, with a morbidity rate around 6%.⁶¹ Most of the patients reported by Scott and colleagues³⁶ had ablations performed with or without resection, with median OS of 134 months when greater than 70% cytoreduction could be achieved, with no mortality and a complication rate of 52% (15% Clavien-Dindo grade III-IV). Complications include hemorrhage, hematoma, pneumothorax, and neuritis where the probe enters the skin.^61,62

Intra-arterial Therapies

Hepatic arterial embolization is another useful option for improving hormonal symptoms and slowing tumor growth in patients with NETLMs. Hypervascular NETLMs derive most of their blood supply from the hepatic artery, whereas the hepatic parenchyma has a dual blood supply but receives most from the portal vein.⁶² Occlusion of arterial branches to NETLMs therefore induces tumor ischemia and necrosis with relative sparing of hepatic parenchyma. Types of embolization include bland transarterial embolization (TAE); transarterial chemoembolization (TACE), which administers doxorubicin or other chemotherapeutic agents to the tumor; and radioembolization with radionuclide ⁹⁰yttrium, which uses smaller bead sizes to enhance distribution of particles in more distal tumor vessels.

Intra-arterial therapies are generally used for palliation in patients whose hepatic tumor burden precludes surgical cytoreduction, and for less invasive treatment of symptoms. Bland embolization and chemoembolization can induce symptomatic improvements in 53% to 100% of patients for a duration of 10 to 55 months.⁶² Five-year OS rates range from 40% to 83%.⁶² Mayo and colleagues⁶³ compared 753 patients with NETLMs who underwent resection with those who underwent intra-arterial therapy (TAE, TACE, placement of drug-eluting beads, or selective internal radiation therapy with ⁹⁰yttrium). Propensity score matching was used to determine a subset of patients who underwent intra-arterial therapy with similar baseline characteristics to the surgical cohort. The study found that surgery and intra-arterial therapy had no difference in survival for patients with greater than 25% liver tumor burden and no symptoms, suggesting intra-arterial therapy may be an appropriate strategy in this cohort of patients. In comparison, surgery was associated with better survival in patients with less than 25% liver tumor burden (survival 84 vs 39 months for embolization, P = .045).

Morbidity with intra-arterial therapies occurs in approximately 25% of patients. A common complication is postembolization syndrome, which manifests as fever, leukocytosis, and abdominal pain and can be particularly sever in patients with high tumor burdens.^62,64 Absolute and relative contraindications to TAE and TACE include portal vein thrombosis, hepatic insufficiency, and history of a Whipple procedure (because of increased risk of hepatic abscess formation).⁵¹

SYSTEMIC THERAPIES

Somatostatin Analogues

In addition to treating hormonal symptoms, somatostatin analogues (SSAs) can stabilize tumor size and delay progression. The antitumor effect of SSAs was established by the phase III Placebo-Controlled, Double-Blind, Prospective, Randomized Study on the Effect of Octreotide LAR in the Control of Tumor Growth in Patients with Metastatic Neuroendocrine Midgut Tumors (PROMID) trial.⁶⁵ This study found that long-acting octreotide (30 mg intramuscularly every 4 weeks) stabilized tumor growth and decreased the risk of progression (HR, 0.33; P<.001) in patients with advanced (unresectable or metastatic) well-differentiated midgut NETs (n = 85).⁶⁵ Longer-term follow-up did not reveal a statistically significant difference in OS between placebo-treated and octreotide-treated groups, but the high rate of crossover to the treatment group on progression limited the ability to interpret the OS outcomes.²⁴ The Controlled Study of Lanreotide Antiproliferative Response in Neuroendocrine Tumors (CLARINET) trial confirmed the antitumor effects of SSAs in metastatic gastroenteropancreatic NETs (n = 204), finding that lanreotide (120 mg subcutaneously every 4 weeks) improved PFS compared with placebo (median PFS not reached vs 18 months, P<.002).⁶⁶ As in the PROMID study, there were no significant differences in quality of life or OS, and the CLARINET study was also complicated by crossover from the placebo to the treatment group after progression. Taken together, these studies show that SSAs are an effective first-line treatment of hormonal symptoms and stabilization of disease in metastatic gastroenteropancreatic NETs.

Molecularly Targeted Therapies

Everolimus, a mammalian target of rapamycin (mTOR) inhibitor, is a treatment option for patients who progress on SSAs. The RAD001 in Advanced Neuroendocrine Tumors (RADIANT)-3 trial studied the efficacy of everolimus (10 mg/d orally) in patients with advanced low-grade or intermediate-grade PNETs (n = 410).⁶⁷ This study found that PFS was longer in patients treated with everolimus compared with placebo (median PFS, 11 vs 4.6 months; P<.001). Treated patients were also less likely to progress or die (HR, 0.35; P<.001). In the RADIANT-4 trial, the efficacy of everolimus was confirmed for advanced, well-differentiated, grade 1 or 2 NETs of the lung and gastrointestinal tract. In this phase III trial (n = 302), patients treated with everolimus had longer median PFS compared with those treated with placebo (median PFS, 11.0 vs 3.9 months; P<.001) and decreased risk of progression or death (HR, 0.48; P<.001).⁶⁸

Another molecularly targeted therapy for advanced NETs is sunitinib, a vascular endothelial growth factor inhibitor. Sunitinib was studied in a randomized controlled study of 171 patients with progressive, advanced, well-differentiated pancreatic NETs. Patients treated with sunitinib (37.5 mg/d orally) had improved PFS compared with those in the placebo group (median PFS, 11.4 vs 5.5 months; HR for progression or death, 0.42; P<.001).⁶⁹ Sunitinib also stabilized disease and improved OS (HR of death, 0.41; P = .02). The study was terminated early because of the risk of adverse events, progression, and death in the placebo group. Based on this trial, sunitinib was US Food and Drug Administration approved for use in progressive, unresectable, and advanced pancreatic NETs.

Peptide Receptor Radionuclide Therapy

PRRT is a treatment option for patients who progress on SSAs, and involves the administration of a radionuclide, primarily ⁹⁰yttrium or ¹⁷⁷lutetium, conjugated to an SSA to selectively deliver radiation to NET cells.⁷⁰ The phase III Neuroendocrine Tumors Therapy (NETTER)-1 trial randomized patients to ¹⁷⁷Lu-DOTATATE and low-dose octreotide or high-dose long-acting octreotide (60 mg every 4 weeks) in 299 patients. The ¹⁷⁷Lu-DOTATATE treatment group had improved 20-month PFS rate compared with the control group (65% vs 11%), with HR for progression or death of 0.21 (P<.001). Median PFS was improved in the ¹⁷⁷Lu-DOTATATE group (median PFS not reached vs 8.4 months), and the HR for death in the ¹⁷⁷Lu-DOTATATE group compared with control was 0.4 (P = .004). There was also improved tumor response with PRRT compared with octreotide (objective response rate 18% vs 3%, P<.001). Therefore, PRRT is an effective treatment of metastatic well-differentiated NETs that express somatostatin receptors and have progressed despite SSA therapy.

Chemotherapy

Chemotherapy is generally indicated for higher-grade NETs such as neuroendocrine carcinomas, and advanced PNETs.^51,71 Neuroendocrine carcinomas (poorly differentiated, Ki-67>20%) are generally treated with platinum-based regimens, such as cisplatin or carboplatin in combination with etoposide.^72,73 Response rates to this regimen are only about 30% and prognosis remains poor, with median PFS of 4 months and median OS of 11 months.⁷⁴ In PNETs, streptozocin-based regimens with 5-fluorouracil or doxorubicin have been used with response rates of around 40% in older series, but are not commonly used currently.⁷⁵ The regimen of capecitabine and temozolamide (CAPTEM) has promising results in preliminary studies of advanced PNETs, with response rates greater than 50%.^76,77 The Eastern Cooperative Oncology Group (ECOG) 2211 trial resulted in preliminary data that CAPTEM (capecitabine 750 mg/m² by mouth twice day on days 1–14, temozolomide 200 mg/m² by mouth days 10–14) improved PFS compared with temozolamide alone (median PFS, 22.7 vs 14.4 months; P = .023).⁷⁸ CAPTEM shows lower efficacy in gastrointestinal NETs, with overall response rates of around 15%.⁷⁹ It remains an option for patients with progressive intestinal NETs who have failed other treatments.⁵¹

SUMMARY

A significant number of patients with NETs develop hepatic metastases, which can cause hormonal symptoms and negatively affect survival. Although prospective data comparing resection with other treatment modalities are lacking, resection and cytoreduction seem to improve symptoms and are associated with better survival outcomes. Cytoreduction should be considered when at least 70% of a patient’s NETLMs can be resected, with treatment decisions depending on patient factors, tumor characteristics, and tumor burden. Instead of major formal hepatic resections, parenchymal-sparing procedures, such as wedge resections, enucleations, and ablations, should be used when possible. Liver-directed and systemic therapies can palliate symptoms and improve survival in patients who are not surgical candidates. Treatment of NETLMs requires a multidisciplinary approach, and patients with NETLMs should be referred to centers with experience caring for such patients to optimize patient quality of life and outcomes.

The medical oncologist’s perspective.

Surgical management of liver metastases is considered in select gastrointestinal cancers, including colon cancer, gastrointestinal stromal tumors, and NETs. In addition to resectability, many other factors, such as the primary site of origin, grade and differentiation of the tumor, hormone hypersecretion, and patient symptoms, have to be taken into account before considering surgical resection of NETLMs. Although a randomized controlled clinical trial will best answer the question of whether surgery for NETLMs improves OS in advanced gastroenteropancreatic (GEP) NETs, conducting such a trial is impractical because of the indolent nature of the tumor and inherent selection bias, among other factors. Thus, the authors rely largely on the retrospective single-institutional and multi-institutional series discussed in detail in this article. While reviewing these data, remember that some series predate widespread availability of newer therapeutic options such as everolimus, sunitinib, or PRRT that have shown improvement in PFS and a trend toward improved OS in rigorous randomized clinical trials. Furthermore, with the increasing use of the sensitive ⁶⁸gallium-DOTATATE PET/CT imaging for initial staging, surgical plans may be altered to include more invasive or less invasive approaches based on detection of additional sites of disease.^80,81

In a recent meta-analysis of 11 cohort studies that included 1108 patients, 5 studies (n = 506) reported outcomes on liver metastases resection versus no liver metastases resection.⁸² Although the meta-analysis showed that 5-year OS was in favor of liver metastases resection (OR, 0.15; 95% confidence interval, 0.05–0.42; P = .0003), the authors noted a high risk of selection bias and significant heterogeneity. The European working group on neuroendocrine liver metastases recommended liver resection as the first choice for patients with completely resectable grade 1 and grade 2 liver metastases and no resectable extrahepatic disease. However, the group also noted that the role of cytoreductive surgery when alternative nonsurgical options are available is unknown.⁸³ The recently published NANETS guidelines on surgical management of pancreatic NETs addressed the question of liver cytoreduction in detail.³⁸ Although more than half the panel thought that symptom control and survival could be improved with more than 70% cytoreduction, this was noted to be a controversial area with level III evidence. There is perhaps even more controversy about whether resection of the primary tumor is warranted in patients with unresectable liver metastases because of the lack of prospective data. Although removal of the primary tumor may have a role in preventing symptoms caused by the location of the tumor, especially in SBNETs, the premise that this may prevent further liver metastases, or seeding, has very weak evidence in the literature.

Because most patients undergoing surgery for NETLM have recurrent disease and surgery is not curative, the morbidity of the surgery has to be weighed against other available therapeutic options in individual patients. When surgery is offered, it is also important to discuss with the patient the goal of this intervention: whether the primary objective of surgical cytoreduction is symptom control or for a potential survival advantage. A one-size-fits-all approach based on an algorithm is clearly not suitable for this disease. All patients with NET deserve a multidisciplinary approach to help them make an informed choice regarding their care. Despite the lack of prospective trial data, surgery at high-volume centers with expertise remains an important part of the treatment paradigm in advanced gastroenteropancreatic NETs.

CLINICS CARE POINTS.

• Cytoreduction of neuroendocrine tumor liver metastases can improve symptoms and survival.

• Systemic therapies are associated with improved survival outcomes in patients with metastatic neuroendocrine tumors.

• Patients with neuroendocrine tumors often benefit from multidisciplinary care at specialized centers.

KEY POINTS.

• Cytoreduction of neuroendocrine tumor liver metastases can relieve symptoms and is associated with improved survival.

• Data suggest the target for cytoreduction threshold should be lowered from 90% to 70% because cytoreduction of at least 70% of neuroendocrine liver tumor burden is associated with improved outcomes.

• Parenchymal-sparing procedures (eg, enucleation, ablation, and wedge resections) should be used when possible for cytoreduction.

• For patients who are not surgical candidates, liver-directed and systemic therapies can palliate symptoms and can prolong survival.