Circulating Biomarkers of Response to Sunitinib in Gastroenteropancreatic Neuroendocrine Tumors

Abstract

After years of limited progress in the treatment of patients with advanced-stage gastroenteropancreatic neuroendocrine tumors (GEP-NETs), strategies using targeted agents have been developed on the basis of increased knowledge of the biology of these tumors. Some of these agents, targeting vascular endothelial growth factor (VEGF) and the mammalian target of rapamycin (mTOR) pathway, have shown efficacy in randomized clinical trials. The tyrosine kinase inhibitor sunitinib and the mTOR inhibitor everolimus have received international approval for the treatment of advanced well differentiated pancreatic NETs after showing survival benefit in randomized phase III trials. There is now an imperative need to identify biomarkers of the biologic activity of such targeted therapies in specific disease contexts, as well as new markers of response and prognosis. This approach may allow rational development of drugs and early identification of patients who may obtain benefit from treatments. In this article, we review recent developments in circulating biomarkers of the clinical benefit of targeted therapies for GEP-NET, including soluble proteins and circulating cells, with an emphasis on sunitinib. No validated molecular biomarkers are yet integrated into clinical practice for sunitinib in NET, although some markers have shown correlation with clinical outcomes and may be implicated in resistance. The VEGF-pathway proteins and interleukin-8 (IL-8) are possibly prognostic in GEP-NET; other possible soluble markers of the activity of sunitinib and everolimus include stromal cell-derived factor 1α, chromogranin A, and neuron-specific enolase. We additionally discuss treatment-induced modulation of circulating endothelial cells and progenitors and subpopulations of cells of the myeloid lineage. These candidate markers should be considered in the development of future combination or sequential therapies.

1. Introduction

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are mainly classified as well or poorly differentiated. Well differentiated GEP-NETs are rare neoplasms arising from neuroendocrine cells in the pancreas (pNET) or elsewhere in the intestine (where they are commonly known as carcinoid tumors). Some of these tumors are able to produce and secrete a number of amines and hormones and are therefore known as functioning GEP-NETs, whereas the rest are considered nonfunctioning. Although carcinoid syndrome is present only in a minority of patients with GEP-NETs, it is the most common clinical presentation of the functioning type, causing diarrhea and flushing, among other symptoms.

The terms used to characterize GEP-NET can be confusing because different classifications use a variety of terms. The last World Health Organization classification, published in 2010,^[1] recommends the terms ‘neuroendocrine neoplasm grade 1 and 2′ for low and intermediate grades of differentiated GEP-NETs, respectively, and the term ‘neuroendocrine carcinoma grade 3′ for poorly differentiated GEP-NETs, which are classified as small-cell and large-cell carcinomas.

Complete resection is the only curative treatment, but it is feasible only in cases of localized disease. Unfortunately, GEP-NETs are frequently metastatic by the time the patient first seeks attention, and although an indolent clinical course is not rare, a cure is no longer an option in most cases. Somatostatin analogs (e.g. lanreotide [Somatuline®; Ipsen Pharma Biotech, Boulogne, France] and octreotide acetate [Sandostatin®; Novartis Pharmaceuticals Corporation, Basel, Switzerland]) are the first line of treatment for functioning GEP-NETs because about 90% of them express somatostatin receptors.^[2] In addition to the proven ability of the somatostatin analogs to control symptoms in patients with functioning tumors, the results of several studies have suggested that they have additional direct antitumor activity.^[3,4] Although the exact mechanisms of that activity remain unclear, one possibility is inhibition of insulin-like growth factor 1, which in turn affects the cell cycle and tumor angiogenesis.^[5,6] A randomized placebo-controlled phase III trial of octreotide LAR [long-acting release] in patients with well differentiated metastatic midgut NETs demonstrated benefit in progression-free survival (PFS).^[7] Similar placebo-controlled studies for exclusively nonfunctioning pNETs are currently ongoing to analyze this potential antitumor effect.

Whereas chemotherapy is the standard of care for poorly differentiated NET, the rate of responses in well differentiated pNETs is low (although the efficacy of chemotherapy in carcinoid tumors is even lower). Regimens based on streptozocin combined with either 5-fluorouracil or doxorubicin have been widely used as first-line treatment or after treatment with somatostatin analogs since the early 1990s.^[8] A few phase II trials of combinations of capecitabine and oxaliplatin^[9,10] and capecitabine plus temozolamide^[11] yielded responses in patients with pNETs.

Local procedures such as radiofrequency ablation and chemoembolization are an option when the disease has metastasized to the liver. Targeted radionuclide therapy is another systemic palliative treatment option for symptomatic patients with progressing non-resectable or metastatic disease. Few therapeutic advances have been made in this field until very recently, and few randomized clinical trials have been conducted.

Increasing knowledge about the biology of NETs, however, has led to several studies exploring targeted therapies against vascular endothelial growth factor (VEGF) and the mammalian target of rapamycin (mTOR) pathway. Thus, after nearly two decades of limited progress, two drugs have received international approval for the treatment of metastatic or unresectable progressing pNET: sunitinib (sunitinib malate [Sutent®; Pfizer, Inc., New York, NY, USA]) and everolimus (RAD001 [Afinitor®; Novartis Pharmaceuticals Corporation]). The clinical trials that led to the approval of both drugs are summarized in section 3 (see also table I).

Table I.

Pivotal clinical trials of sunitinib and everolimus in advanced pancreatic neuroendocrine tumors

Treatment	Patients	n	Median PFS (months)	Median OS (months)	Comments	References
Sunitinib 37.5mg daily vs placebo	Well differentiated pNET	154 patients randomized (1 : 1)	11.4 vs 5.5; HR 0.42; p < 0.001	30.5 vs 24.4; HR 0.73; p = 0.19	Trial stopped at interim analyses. 69% of patients in the placebo group received sunitinib at progression or after the trial was stopped	12,13
Everolimus 10mg daily vs placebo	Low- or intermediate- grade pNET	410 patients randomized (1 : 1)	11.0 vs 4.6; HR 0.35; p < 0.001	Not reported yet		14

HR = hazard ratio; OS=overall survival; PFS=progression-free survival; pNET=pancreatic neuroendocrine tumor.

Identification of biomarkers predictive of clinical benefit is necessary for more effective application of targeted therapies to patients. Early indicators of clinical benefit are especially important in slowly progressing cancers like NET because traditional endpoints in clinical trials, such as overall survival (OS), may take a long time to be achieved. Moreover, benefit from some of these treatments might not correlate with objective radiologic responses.

The results of an analysis of circulating biomarkers in patients with NET who were treated with sunitinib in a phase II clinical trial were recently presented. A panel of soluble angiogenic proteins and circulating cells was evaluated, and new candidate markers of clinical benefit and biologic activity of sunitinib in NET were identified.^[15] In this article, we review recent developments in biomarkers of clinical benefit of targeted therapies for GEP-NET, with an emphasis on sunitinib.

2. Sunitinib Targets Multiple Receptor Tyrosine Kinases

Sunitinib is a small molecule that has antitumor and antiangiogenic effects. Its activity is mediated by the inhibition of a wide spectrum of receptor tyrosine kinases (RTKs), including VEGF receptors (VEGFR-1, VEGFR-2, and VEFGR-3), platelet-derived growth factor (PDGF) receptors (PDGFR-α and PDGFR-β), fetal liver tyrosine kinase receptor (FLT3), and c-Kit. Sunitinib’s efficacy has been demonstrated in several randomized phase III clinical trials in patients with different types of cancer, including metastatic renal cell carcinoma (mRCC), metastatic or unresectable gastrointestinal stromal tumors (GIST) after progression with imatinib treatment (imatinib mesilate [Gleevec® or Glivec®; Novartis Pharmaceuticals Corporation]), and unresectable or metastatic pNET.^[16,17]

Sunitinib (identified during its development as SU011248) is orally administered and metabolized to an active metabolite, SU012662. Data from preclinical studies suggested that the plasma concentration of sunitinib plus SU012662 necessary to achieve in vivo efficacy is in the range of 50–100 ng/mL.^[18] Results from the first-in-human study were initially reported in 2006.^[19] Twenty-eight patients were enrolled, and a phase II dosage of 50 mg/day in a 4-week-on and 2-week-off cycle was recommended. Six patients experienced radiologic responses (three with mRCC, one with NET, one with GIST, and one with an unknown primary adenocarcinoma). The pharmacokinetic profile was non-linear, with area under the concentration-time curve (AUC) values increasing less than proportionally with higher dosages. The recommended phase II dosage corresponded with plasma concentrations within the predicted range. Most of the patients who experienced dose-limiting toxicities had plasma concentrations of greater than 100 ng/mL.

3. Clinical Development of Sunitinib and Its Position in the Gastroenteropancreatic Neuroendocrine Tumor (GEP-NET) Therapeutic Landscape

The first specific phase II trial of sunitinib in the setting of GEP-NET, using the regimen discussed above, enrolled 107 patients (66 with pNET and 41 with carcinoid tumors). Eleven radiologic responses (16.7%) were observed in the pNET group, whereas only one objective response (2.5%) was detected in the non-pancreatic carcinoid NET cohort. The time to disease progression was not significantly different between the groups (7.7 months for pNET vs 10.2 months for carcinoid tumors).^[15]

In a randomized, double-blinded, placebo-controlled, international phase III trial of sunitinib, given as a continuous dose of 37.5 mg daily in patients with well differentiated metastatic or unresectable pNET,^[12] a significant benefit in PFS time was found in the sunitinib-treated patients (11.4 months vs 5.5 months; hazard ratio [HR] 0.41; p = 0.0001). The radiologic response rate in the sunitinib arm was 9.3%, compared with no responses in the placebo arm. That trial was stopped early because of evidence of PFS benefit in a planned intermediate safety analysis, and that led to the crossover to sunitinib of most of the patients in the placebo arm. In a recent update of that trial, the investigators reported a trend toward a longer OS time in the sunitinib-treated group than in the originally placebo-treated patients (30.5 months vs 24.4 months; HR 0.73; p = 0.19). Although this difference did not reach the level of statistical significance, it is notable that up to 69% of patients in the original placebo group received sunitinib at the progression of their disease or after the trial interruption.^[13]

On the basis of those phase III trial results, sunitinib has been approved internationally for the treatment of unresectable or metastatic progressive well differentiated pNET.

In 2008, the results of the first phase II trial of everolimus in patients with NET were published. That single-center trial evaluated 60 patients (30 with carcinoid tumors and 30 with pNET), including patients with well and intermediately differentiated tumors, who were treated with everolimus and octreotide. The objective response rate was 27% in patients with pNET (median PFS 63 weeks) and 17% in those with nonpancreatic NET (median PFS 50 weeks).^[20] Everolimus was approved for this indication after the publication of results from an international, randomized, double-blinded, placebocontrolled phase III trial at a dosage of 10 mg/day in patients with progressive well or intermediately differentiated pNET. The primary endpoint in that trial was PFS. The everolimustreated group experienced significant benefit relative to that in the placebo-treated group: PFS of 11 months versus 4.6 months (HR 0.35; p < 0.001).^[14]

The recent approval of sunitinib and everolimus in this setting has yielded new active treatment options for patients with GEP-NET. The best sequence of treatment for progressive pNET is still unclear, and a face-to-face comparison of these two drugs may be required. By now, the sequence of treatment should be decided case by case, considering comorbidities, the toxicity profile, and drug availability.

There are as yet no validated molecular or cellular biomarkers for sunitinib response in clinical practice. Alternative markers of clinical benefit such as the early development of adverse side effects (e.g. hypertension^[21]) and new imaging techniques are beyond the scope of this review.

More results of studies with sunitinib treatment for clinical indications such as mRCC and GIST have been reported than for GEP-NET because of the longer experience with the drug in those diseases.

4. Circulating Angiogenesis-Related Proteins as Biomarkers of Response to Sunitinib in Tumor Types Other than GEP-NET

VEGF plays an important role not only in physiologic and pathologic angiogenesis but also in processes leading to the survival, growth, and metastatic ability of tumors.^[22,23] The VEGF family consists of several similar proteins (VEGFs A through F, although the term VEGF is most commonly used to indicate VEGF-A) and placental growth factor (PGF).^[24] VEGF is considered the most important regulatory factor of tumor angiogenesis. Three different cell-membrane receptors have been described for the VEGF family: VEFGR-1, -2, and -3.^[25] Soluble forms of these receptors can be detected in peripheral blood.

The first-in-human trial of sunitinib^[19] included an analysis of plasma levels of VEGF and soluble VEGFR (sVEGFR)-2, in which an enzyme-linked immunosorbent assay (ELISA) was used to evaluate samples both pretreatment and after 28 days of treatment. A progressive increase in VEGF and a decrease in sVEGFR-2 concentrations were observed, demonstrating ontarget effects of the drug.

In the setting of mRCC, the results of several studies showed changes in the levels of VEGF, PGF, sVEGFR-2, and sVEFGR-3 during therapy and also that there was correlation between some of those changes and the response to treatment. For example, DePrimo et al.^[26] analyzed candidate circulating biomarkers of response to therapy in 63 patients with mRCC enrolled in a phase II study of sunitinib given at 50 mg/day as the starting dosage in a 4-week-on, 2-week-off regimen.^[27] Greater than 3-fold increases from baseline in VEGF and PGF concentrations were found at the end of the first cycle in 44% and 40% of the patients, respectively, as well as significant decreases of more than 30% from the baseline concentrations of sVEGFR-2 (in 91% of the patients) and sVEGFR-3 (in 87%) during the same interval. The concentrations of all factors tended to return to near baseline levels at the end of the first 2-week-off period, strongly suggesting a drug-related effect. Only 27% of baseline samples reached the threshold level for ELISA-based quantification of PGF, whereas up to 95% of the patients had concentrations of PGF that were within the detectable range on day 28, suggesting an overall increase with treatment. Similar results were observed with further cycles of treatment. Larger changes in VEGF, sVEGFR-2, and sVEGFR-3 on day 28 of treatment, relative to baseline concentrations, showed significant correlation with radiologic response to treatment; PGF was not included in this analysis.

In a subsequent study in patients with mRCC who had been pretreated with bevacizumab, similar changes from baseline in the concentrations of VEGF, PGF, and sVEGF-3 were detected in response to sunitinib treatment.^[28] Correlations between baseline concentrations of VEGF-C and sVEGFR-3 and radiologic responses were reported, but since this was a singlearm clinical trial, it is unclear whether pretreatment VEGF-C and sVEGFR-3 concentrations are associated with sunitinib treatment benefit or could be prognostic factors independent of treatment type in mRCC.

A phase II trial of sunitinib in patients with previously treated metastatic breast cancer revealed a correlation between OS and reductions of 20% below baseline in sVEGFR-3 concentrations at the end of the first cycle, as well as a correlation between time to disease progression and decreases of more than 50% below baseline in circulating Kit concentrations.^[29] Soluble Kit has also been shown to be associated with sunitinib benefit in patients with GISTs treated after failure of imatinib treatment.^[30]

Another phase II trial of sunitinib in patients with urothelial cancer revealed a correlation between relatively lower levels of interleukin-8 (IL-8) at baseline and increased time to disease progression.^[31] After 2 weeks of treatment, there was a significant decrease in the sVEGFR-2 concentration and a significant increase in that of stromal cell-derived factor (SDF)-1α. Of interest, the patients who experienced clinical benefit in this trial had higher median increases in SDF-1α and -β concentrations than did those who experienced no benefit from sunitinib treatment.

Analysis of patients with hepatocellular carcinoma treated with sunitinib in a different clinical trial showed that a relatively higher baseline SDF-1α concentration was a factor of a poor prognosis.^[32]

5. Circulating Biomarkers of Sunitinib Clinical Benefit and/or Biologic Activity in GEP-NET

Several circulating biomarkers have been identified as indicators of the pharmacodynamics of sunitinib (table II) and/or prognostic value in GEP-NET (table III).

Table II.

Circulating pharmacodynamic biomarkers of sunitinib in patients with gastroenteropancreatic neuroendocrine tumors

Biomarker	Data reported	Reference
VEGF	>3-fold increase in VEGF levels in 53% of patients after 4 weeks on treatment (p <0.00001), returning to near baseline after 2 weeks off treatment	33
	Higher changes observed when sunitinib levels in plasma were >50 ng/dL	33
	3.7-fold median increase after 4 weeks of treatment (p < 0.0001); no significant differences between pNET and carcinoids	34
sVEGFR	≥30% decrease in sVEGFR-2 levels (57.6% of patients) and sVEGFR-3 levels (69.5% of patients), returning to near baseline after 2 weeks off treatment	33
	Median 40% decrease in sVEGFR-3 levels after 4 weeks of treatment; no significant differences among tumor types. Correlation between decrease in sVEGFR-3 levels (ρ=−0.51; p < 0.0001) and plasma drug levels; not present for sVEGFR-2 levels	34
IL-8	>2-fold increase in 43% of patients and >3-fold in 23% of patients after 4 weeks of treatment, returning to near baseline after 2 weeks off treatment	33
	1.8-fold median increase after 4 weeks on treatment (p = 0.008)	34
SDF-1α	20% median increase after 4 weeks on treatment (p = 0.027)	34
Circulating endothelial cells	Significant decrease during the first cycle of treatment (p = 0.037 at day 14 and p = 0.045 at day 28)	34
Myeloid populations	7% decrease observed after 2 weeks (p = 0.04) and 4 weeks (p = 0.08) in the CD14+ monocyte count, returning to baseline after 2 weeks off treatment	34

ρ=Spearman’s rank correlation coefficient; IL-8 = interleukin-8; pNET=pancreatic neuroendocrine tumor; SDF-1α=stromal cell-derived factor-1α; sVEGFR = soluble VEGF receptor; VEGF=vascular endothelial growth factor.

Table III.

Soluble biomarkers and correlations with outcomes in gastroenteropancreatic neuroendocrine tumors

Biomarker	Patients	Data reported	Reference
VEGF	Low grade GEP-NET (resected)	VEGF levels inversely correlated with PFS (strong vs weak vs negative expression, 29 months vs 81 months vs not reached; p = 0.02; 50 months of follow-up)	35
	NET	Higher VEGF levels in patients with stable disease vs patients with progressive disease (p < 0.001)	36
sVEGFR	pNET treated with sunitinib	Higher baseline sVEGFR-3 levels correlated with PFS (p=0.04). Greater reductions in sVEGFR-3 levels were correlated with objective responses and improved PFS (p = 0.04) in pNET	33
	pNET treated with sunitinib	Correlation between high baseline sVEGFR-2 levels and OS (HR 0.22; p = 0.01) in pNET	34
IL-8	NET	Higher IL-8 levels in patients with progressive disease vs stable disease (p > 0.018). Similar IL-8 levels among patients with stable disease and controls (p=0.23). Patients with OS >2 years had lower IL-8 levels than those with OS <2 years (p=0.009)	36
	pNET treated with sunitinib	Lower baseline IL-8 levels among patients with radiological stability for >6 months compared with those with stable disease for <6 months (p=0.009)	33
	Carcinoid tumors treated with sunitinib	Correlation between lower baseline IL-8 levels and ‘clinical benefit response’ [radiological response or stable disease for <6 months] (p=0.005) in carcinoids	34
SDF-1αa	pNET treated with sunitinib	High baseline SDF-1α levels correlated with shorter PFS (HR 3.59; p = 0.005) and OS (HR 2.34; p = 0.02)	34
CgA	pNET treated with everolimus	Baseline elevated CgA levels (>2-fold upper normal limits) correlated with reduced PFS (HR0.55; p = 0.03) and OS (HR0.3; p=0.01). Early reductions in CgA levels (>30% reduction after 4 weeks compared with baseline) correlated with improved PFS (HR 0.25; p < 0.001) and OS (HR 0.4; p = 0.01)	37
	pNET treated with everolimus+octreotide vs placebo+octreotide	Baseline elevated CgA levels (>2-fold upper normal limits) correlated with reduced PFS (HR0.43; p = 0.001) in the overall population of the trial. Benefit in PFS for patients in the everolimus + octreotide arm vs the placebo+octreotide arm was significant in the subgroup with baseline elevated CgA levels (HR 0.66; p=0.003), but benefit in PFS was not significant in the subgroup with non-elevated baseline CgA levels	38
NSE	pNET treated with everolimus	Baseline elevated NSE levels (over upper normal limits) correlated with reduced PFS (HR 0.52; p = 0.01) and OS (HR 0.44; p=0.005). Early reductions in NSE levels (>30% reduction after 4 weeks compared with baseline) correlated with improved PFS (HR 0.25; p <0.001) but not with significant changes in OS	37
5-HIAA	pNET treated with everolimus +octreotide vs placebo+octreotide	Benefit in PFS for patients in the everolimus + octreotide arm vs the placebo+octreotide arm was significant in the subgroup with baseline elevated 5-HIAA levels [>2-fold upper normal limits] (HR 0.66; p = 0.007) but benefit in PFS was not significant in the subgroup with non-elevated baseline 5-HIAA levels. Correlation between elevated 5-HIAA levels and decreased PFS was not significant for the overall population of the trial	38

5-HIAA=5-hydroxyindoleacetic acid; CgA=chromogranin A; GEP-NET=gastroenteropancreatic neuroendocrine tumor; HR = hazard ratio; IL-8=interleukin-8; NET=neuroendocrine tumor; NSE = neuron-specific enolase; OS=overall survival; PFS=progression-free survival; pNET = pancreatic neuroendocrine tumor; SDF-1a=stromal cell-derived factor-1g; sVEGFR = soluble VEGF receptor; VEGF=vascular endothelial growth factor.

5.1 Vascular Endothelial Growth Factor and Soluble Receptors

Well differentiated GEP-NETs have been reported to be highly vascularized and to express high levels of VEGF and its receptors. Moreover, increased expression of VEGF in NET has been associated with poor outcomes.^[35,39]

Pavel et al.^[36] analyzed the levels of circulating VEGF in 38 patients with NET and compared them with those in 23 agematched healthy control subjects. Eleven patients had pNET, 23 had non-pancreatic GEP-NET, and four had bronchial carcinoid tumors. Concentrations of VEGF were significantly higher in the patients with GEP-NET than they were in the control group. In addition, the mean circulating VEGF concentration was significantly higher in the group of 17 patients with progressive disease at the time of the study than it was in patients with stable disease. The VEGF concentrations in control subjects and in patients with stable disease were similar. This study included patients with functioning and non-functioning tumors, as well as patients both on and off available therapies (somatostatin analogs, chemotherapy, and radioisotopes) at the time. Neither of these variables was associated with differences in circulating VEGF concentrations.

In the case of GEP-NET, a single analysis has been reported that describes modulation of angiogenesis-linked proteins before and during treatment with sunitinib in the group of patients enrolled in a phase II trial of sunitinib in NETs.^[33,34]

Plasma concentrations of VEGF, sVEGFR-2, sVEGFR-3, and IL-8 were measured by performing ELISA at baseline and after 28 days of continuous treatment in cycles 1-3. Patients with pNET had higher sVEGFR-2 and a trend toward higher baseline VEGF concentrations than did patients with carcinoid tumors. On analyses for correlations between baseline biomarker concentrations and clinical outcomes, higher baseline sVEGFR-2 concentrations correlated with longer OS in the pNET subgroup, whereas low sVEGFR-3 baseline concentrations (as well as IL-8, discussed below) were associated with longer PFS and OS in the carcinoid tumor group.^[34] These data suggested that VEGF-pathway proteins and IL-8 may be prognostic in GEP-NET.

Consistent with the results observed in mRCC,^[27] VEGF and sVGEFR-2 and -3 were significantly modulated during treatment with sunitinib. However, no differences were observed between patients with pNET and those with carcinoid tumor. Plasma VEGF concentrations increased more than 3-fold above baseline in about half of the samples after 28 days, whereas sVEGFR-2 and sVEGFR-3 decreased, with more than 30% reductions in 60% of samples for sVEGFR-2 and 70% of samples for sVEGFR-3. As in the previous studies in mRCC, the concentrations of VEGF, sVEGFR-2, and sVEGFR-3 tended to return to baseline values during the 2-week-off-treatment period, confirming that sunitinib’s biologic activity is mediated by the VEGF signaling pathway in NET patients.

5.2 Interleukin-8

IL-8 is a member of the CXC cytokine family that has proangiogenic as well as mitogenic and motogenic effects. These effects are mediated through its interaction with two different receptors, IL-8RA and IL-8RB (also known as CXCR1 and CXCR2). Its primary physiologic function is as a mediator in the inflammatory response through induction of neutrophil chemotaxis. IL-8 is up-regulated by a number of stress factors, such as hypoxia, acidosis, nitric oxide, and cell density. It has also been shown to be constitutively overexpressed in different cancer types.^[40-42]

Although normal pancreatic cells do not express IL-8, samples from both adenocarcinoma and pNETs have shown increased expression of IL-8 and its receptors, especially CXCR2.^[43,44] In a preclinical study, IL-8 was shown to inhibit apoptosis, to enhance proliferation and expression of several matrix metalloproteinases in CXCR1- and CXCR2-expressing endothelial cell lines, and to regulate angiogenesis-inducing capillary tube formation in a concentration-dependent manner.^[45] And in studies using murine models, antibodies against IL-8 inhibited tumor growth and angiogenesis in different tumor types, such as melanoma and RCC.^[46,47]

In another study, among 38 patients with NET, higher plasma concentrations of IL-8 were detected in patients whose disease was progressing than in healthy control subjects, whereas no significant differences were identified between the controls and the patients whose disease was stable.^[36] Plasma IL-8 concentrations were also significantly associated with the 2-year survival rate: all 24 patients considered ‘long survivors’ had IL-8 concentrations below the detectable range (i.e. <10 ng/L), whereas seven of the 14 of patients who died within 2 years had elevated concentrations (p = 0.009). This study included patients with pNETs and non-pancreatic NETs, but the data were not reported independently for each tumor type.

Modulation of IL-8 concentrations in response to sunitinib treatment of patients with NETs was recently reported.^[33,34] No significant differences in baseline IL-8 concentrations or in the pattern or degree of modulation during treatment were observed between patients with carcinoid tumors and those with pNET. IL-8 increased after the first 4 weeks of treatment by an average of 1.8-fold over baseline and tended to return to the baseline concentration during the following 2-week-off-treatment period. A modest correlation between plasma IL-8 and trough concentrations of sunitinib (including its active metabolite, SU012662) was observed.^[33] In the group with carcinoid tumors, baseline IL-8 concentrations were lower in patients who experienced clinical benefit (defined as an objective response or stable disease for more than 6 months) than in those who did not (11.3 vs 30.8 pg/mL; p = 0.005), and there was an inverse correlation between baseline IL-8 concentrations and both PFS and OS.

Together, these results suggest that IL-8 is prognostic in patients with carcinoid tumors and/or can determine those who will experience clinical benefit from sunitinib treatment. The increase in IL-8 concentrations during sunitinib treatment could represent a mechanism of drug resistance. For example, Huang et al.^[47] reported up-regulation of IL-8 on development of resistance to sunitinib in clear-cell RCC cell lines, as well as a correlation between high baseline IL-8 concentrations and primary resistance to sunitinib treatment in patients’ samples.

5.3 Stromal Cell-Derived Factor-1α

SDF-1α, also known as CXCL12, is a small cytokine and part of the chemokine CXC family. Like IL-8, SDF-1α plays an important role in promoting cell migration, proliferation, survival, and angiogenesis. Its role in proangiogenic processes seems to be related to the promotion of migration and particularly to tumor homing of endothelial progenitor cells.^[48,49] Changes in plasma SDF-1α have been shown to be related to the biologic activity of sunitinib. SDF-1α concentrations increased in tumor-bearing mice after their exposure to sunitinib. No changes were seen after incubation of whole blood with sunitinib, though, suggesting that the observed result was probably not related to an effect on peripheral blood cells.^[50]

In a subset of 28 patients from the phase II trial of sunitinib in GEP-NET,^[34] SDF-1α levels were significantly higher in the pNET group than they were in the carcinoid tumor group. A median 1.2-fold increase in SDF-1α levels was detected after 4 weeks of sunitinib treatment. An inverse correlation was observed between SDF-1α levels and clinical outcome, since patients with high SDF-1α levels had a significantly shorter time to disease progression. Moreover, the group of patients who experienced clinical benefit (defined as objective response or disease stabilization over 6 months) had lower SDF-1α levels.^[34]

Whether baseline SDF-1α levels or the increase in SDF-1α detected after exposure to sunitinib may have predictive value is currently unclear. SDF-1α may participate in mechanisms of resistance to antiangiogenic therapy and may be a potential therapeutic target.^[51,52] Arvidsson et al.^[53] studied the responses of carcinoid tumor cells to hypoxia. They observed significant down-regulation of SDF-1α and up-regulation of its receptor, CXCR4, among the genes affected in hypoxic carcinoid cells. Stimulation of hypoxic cells with SDF-1α led to activation of the mitogen-activated protein kinase (MAPK) pathway and increased cell migration, suggesting a possible role of the SDF-1α-CXCR4 axis in resistance to antiangiogenic therapy.

6. Modulation of Cell Populations in Peripheral Blood as Markers of Sunitinib’s Pharmacodynamics

6.1 Circulating Endothelial Cells and Circulating Endothelial Progenitors

The circulating endothelial cells (CECs) detected in peripheral blood are known to be relatively increased in cancer patients and, in some cases, are associated with a poor prognosis.^[54,55] There are at least two distinct subpopulations of CECs: endothelial precursors derived from the bone marrow (CEPs); and mature CECs, which are thought to be shed from existing vessels.^[56-58] It has been shown that VEGFR inhibitors can impact these two subpopulations differently,^[59] as discussed below. Multicolor flow cytometry is a sensitive method for detecting CECs and CEPs in the peripheral blood,^[60,61] but reliable quantification is often difficult because the antigens used for their detection are not entirely specific, and these cells typically exist in the circulation in very low numbers.

CECs express VEGFR and may therefore be targeted by sunitinib. The release of endothelial cells from blood vessels into the circulation may reflect damage to the tumor vasculature caused by antiangiogenic treatment.^[59] On the other hand, mobilization of bone marrow-derived CEPs may be induced by VEGF; in that case, CEPs would be expected to decrease with anti-VEGF therapies.^[62]

In patients treated with sunitinib for mRCC, the mean number of CECs detected increased by 2.2-fold after a cycle of treatment, and patients with relatively higher CEC counts after the first 2 weeks of treatment had longer PFS.^[63] Similar results were reported in patients with GIST who were undergoing treatment with sunitinib.^[64] In patients treated with sunitinib for GEP-NET, the number of CECs decreased significantly during the first 4 weeks of treatment, although no differences were detected after the 2-week-off-therapy period relative to baseline, suggesting a drug-related effect. No changes in CEPs were observed.^[34]

6.2 White Blood Cell Subtypes

Cells of the myeloid lineage have been identified as being important in facilitating angiogenic processes, tumor progression, and metastasis. Expression of sunitinibys target receptors, such as VEGFR-1, PDGFR-β, KIT, and CXCR4, has been described on the surface of these cells.^[65-67]

Modulation of subpopulations of the myeloid lineage has been described in patients with mRCC exposed to neoadjuvant sunitinib.^[68] In addition, Zurita et al.^[34] reported observing a distinct decrease in monocytes after exposure to sunitinib in patients with GEP-NETs, especially CD14+ monocytes expressing VEGFR-1 and CXCR4, suggesting that specific monocyte subpopulations are a marker of sunitinib’s pharmacodynamics.

7. Circulating Biomarkers of Clinical Benefit and Biologic Activity of Everolimus in GEP-NET

Baseline and post-treatment levels of VEGF, PGF, basic fibroblast growth factor, sVEGFR-1, and sVEGFR-2 were analyzed in patients treated with everolimus in the RADIANT-3 (RAD001 in Advanced Neuroendocrine Tumors) phase III randomized clinical trial.^[69] In relation to the placebo, everolimus treatment led to a significant and progressive reduction in sVEGFR-2 and to an early but unsustained decrease in PGF. No significant differences were observed in circulating concentrations of VEGF or sVEGFR-1. These data suggest a possible antiangiogenic effect of everolimus as a consequence of mTOR inhibition.

Chromogranin A (CgA), which can be detected in the secretory granules of neuroendocrine cells, and neuron-specific enolase (NSE), a protein present in their cytoplasm, have been retrospectively studied as biomarkers of prognosis in patients with GEP-NET.^[70-72] An analysis of CgA and NSE in 114 patients enrolled in a phase II study of everolimus in NET showed that relatively higher baseline concentrations of CgA were associated with shorter PFS. The patients with the shortest PFS had relatively elevated concentrations of both CgA and NSE at baseline.^[37] CgA and NSE responses were defined as a 50% or greater reduction from baseline or normalization, and early CgA and NSE responses were defined as a 30% or greater decrease from baseline or normalization after 4 weeks of treatment. Of note, among patients with elevated CgA and/or NSE, an early decrease predicted clinical benefit. For CgA, an early response was associated with both longer PFS (13.3 vs 7.5 months; HR 0.25; p < 0.001) and longer OS (24.9 vs 12.7 months; HR 0.4; p = 0.01). For NSE, an early response was associated with prolonged PFS (8.5 vs 2.8 months; HR 0.25; p < 0.001) but no difference in OS.

A recently reported analysis of the patients included in the phase III RADIANT-2 clinical trial confirmed that early modulation of CgA by everolimus can be used as a surrogate marker of PFS in this setting, and also strongly suggested that 5-hydroxyindoleacetic acid (5-HIAA) levels have prognostic value.^[38]

To our knowledge, no data have been reported for CgA or NSE in patients treated with sunitinib for GEP-NET.

8. Conclusions

GEP-NETs constitute a heterogeneous and relatively rare group of malignancies that have suffered from little therapeutic progress until recently. The international approval of sunitinib and everolimus for pNET and the promising results observed with these two drugs in carcinoid tumors mean new options for patients, even for those in whom conventional treatments have failed. The current challenge for investigators is to identify and validate markers to better assess prognosis, to decide who and when to treat, and to optimize the clinical application (selection, sequence, and dosing) of targeted drugs. Candidate markers of prognosis and sunitinib clinical benefit and pharmacodynamics in blood from pNET and carcinoid patients were recently identified. Similar efforts are being conducted for everolimus and other drugs under development. Since blood makes it possible to assess and monitor host and tumor-derived factors that may affect prognosis and response to targeted therapies, the evaluation of circulating biomarkers in randomized clinical trials in GEP-NET patients seems a mandatory step toward the incorporation of molecular markers into clinical practice.