Randomized Controlled Trial of the Gastrin/CCK₂ Receptor Antagonist Netazepide in Patients with Barrett’s Esophagus

Abstract

Hypergastrinemia has been associated with high grade dysplasia and adenocarcinoma in patients with Barrett’s esophagus (BE), and experimental studies suggest pro-inflammatory and pro-neoplastic effects of gastrin on BE. This is of potential concern, as BE patients are treated with medications that suppress gastric acid production, resulting in increased physiologic levels of gastrin. We aimed to determine whether treatment with the novel gastrin/CCK₂ receptor antagonist netazepide reduces expression of markers associated with inflammation and neoplasia in BE. This was a randomized, double-blind, placebo-controlled trial of netazepide in patients with BE without dysplasia. Subjects were treated for 12 weeks, with endoscopic assessment at baseline and at end of treatment. The primary outcome was within-individual change in cellular proliferation as assessed by Ki67. Secondary analyses included changes in gene expression, assessed by RNA-sequencing, and safety and tolerability. A total of 20 subjects completed the study and were included in the analyses. There was no difference between arms in mean change in cellular proliferation (netazepide: +35.6 Ki67+ cells/ mm², SD 620.7; placebo: +307.8 Ki67+ cells/ mm², SD 640.3; p=0.35). Netazepide treatment resulted in increased expression of genes related to gastric phenotype (TFF2, MUC5B) and certain cancer-associated markers (REG3A, PAX9, MUC1), and decreased expression of intestinal markers MUC2, FABP1, FABP2, and CDX1. No serious adverse events related to study drug occurred. The gastrin/CCK₂ receptor antagonist netazepide did not reduce cellular proliferation in patients with non-dysplastic BE. Further research should focus on the biological effects of gastrin in Barrett’s esophagus.

INTRODUCTION

Esophageal adenocarcinoma (EAC) has risen dramatically in incidence over the past half century.[1, 2] Barrett’s esophagus (BE) is the precursor lesion to EAC, and patients diagnosed with BE undergo endoscopic surveillance with the goal of early detection of dysplasia and cancer. Endoscopic surveillance is associated with detection of earlier stage EAC in BE patients[3], but EAC continues to be associated with a dismal prognosis.[4] Thus, agents that could lower the risk of EAC would be extremely valuable. A recent large trial demonstrated that high-dose esomeprazole was more effective than low-dose for increasing time to all-cause mortality or progression to high grade dysplasia or EAC in BE patients.[5] However, there are no agents with proven ability to decrease the risk of EAC, and to date trials of potential chemopreventive agents have proved disappointing.[6–8]

There is a body of observational and experimental data to suggest a potential pro-neoplastic effect of gastrin on BE. This is of potential concern, as BE patients are typically treated with proton pump inhibitors (PPIs) for the long-term, which results in moderate hypergastrinemia.[9] In a study of 95 patients with BE, those with a history of high-grade dysplasia or EAC had a 5-fold increased odds of having high serum gastrin.[10] In a follow-up study in a subset of those patients, serum gastrin correlated significantly with cellular proliferation as assessed by Ki67 in BE biopsies.[11] The gastrin receptor, CCK2R, is highly expressed in BE tissue, and exposure of BE to gastrin in vitro induces proliferation as well as PGE-2 and COX-2.[12, 13] In a series of experiments using a mouse model of BE/EAC (termed L2-IL-1B), both omeprazole-treated mice as well as L2-IL-1B mice crossed with hypergastrinemic INS-GAS mice had significantly accelerated neoplasia.[14] Further, treatment with a gastrin/CCK₂ receptor antagonist resulted in decreased metaplasia and dysplasia as well as reduced proliferation. More recently, a CCK2R-associated single nucleotide polymorphism was identified as a risk locus for gastro-esophageal reflux disease (GERD) and for BE and EAC.[15]

Thus, targeting the effects of gastrin in BE could in theory represent an effective means of chemoprevention. We conducted a randomized clinical trial to assess the effects of gastrin/CCK₂ receptor antagonist in patients with BE. Netazepide (also known as YF476) is a benzodiazepine derivative and is a potent and selective CCK2R antagonist; in clinical trials in healthy subjects and in patients with gastric neuroendocrine tumors, netazepide has been well-tolerated and biologically active at relatively low doses.[16–23] For the current trial, the effects of netazepide were assessed in BE patients; the primary outcome was cellular proliferation, and the secondary outcomes were effects on other biomarkers associated with esophageal neoplasia and the safety and tolerability of netazepide.

MATERIALS AND METHODS

Study Description

This was a randomized, double-blind, placebo-controlled trial conducted in patients with Barrett’s esophagus without dysplasia. Subjects were treated with netazepide 25 mg daily or placebo for 12 weeks, with upper endoscopy performed at baseline and at the end of the treatment period. The study was conducted at Columbia University Irving Medical Center (New York, USA) and Addenbrooke’s Hospital (Cambridge, UK). The study was conducted from January 2013 through February 2019 (Addenbrooke’s Hospital joined in November 2014). The study was approved by the Institutional Review Board of Columbia University and the Ethical Board of NRES Committee South Central - Oxford A (UK) and was registered with clinicaltrials.gov (NCT01298999). The study was conducted in accordance with the U.S. Common Rule. Informed written consent was obtained from all study subjects.

Eligibility Criteria

Subjects were eligible for the study if they had a documented history of Barrett’s esophagus of at least 1 cm circumferential or at least 2 cm maximal length (C ≥ 1 or M ≥ 2) and with intestinal metaplasia detected on esophageal biopsies. A prior history of indefinite for dysplasia was permitted if the subject’s most recent endoscopy prior to study enrollment had shown no evidence of dysplasia. Subjects had to be on continuous PPI therapy at least once daily for at least 12 months prior to study entry, and on a stable PPI dose for 3 months leading up to enrollment. Subjects were excluded if they had: a history of BE with histologically confirmed low grade dysplasia, high grade dysplasia, or adenocarcinoma; prior endoscopic therapy for BE; history of gastric or esophageal surgery; history of atrophic gastritis, pernicious anemia, or Zollinger-Ellison syndrome. One subject was diagnosed with recurrent, metastatic prostate cancer shortly after enrollment. Subsequently, the following additional exclusion criteria was added: a history of cancer >3 years from the time of enrollment, and the patient is not up to date with surveillance for that cancer (based on national guidelines), or has evidence of cancer at the time of enrollment.

Study Drug

Netazepide (also known as YF476; Trio Medicines, London, UK) is a potent, orally active, highly selective, competitive antagonist of CCK2 receptors. In the current trial, subjects were treated with spray-dried netazepide 25 mg or matching placebo (supplied by Trio Medicines, London, UK) once daily by mouth for 12 weeks. At the time, long-term toxicology studies had not been completed, and 12 weeks was the longest period of netazepide treatment allowed. Treatment with netazepide 25 mg has a biological effect in humans through gastrin receptor inhibition, and has been shown to reduce gastric neuroendocrine tumor size with durations as short as 12 weeks.[18, 23]

Procedures and Visits

After enrollment, subjects were randomized 1:1 to either netazepide 25 mg or placebo. Blocked randomization was performed, stratified by PPI frequency (once or twice daily) and by study site. The study investigators and staff and the study subjects were blinded to treatment assignment. Upper endoscopy was performed at baseline, with biopsies taken in a four-quadrant fashion every 2 cm along the length of Barrett’s esophagus. Subjects then took study drug for 12 weeks, with a repeat endoscopy with biopsies using the same protocol performed at the end of that period. There were study visits during the 12-week treatment period and one month after treatment finished to assess any adverse effects. Blood samples were collected at each study visit for serum gastrin and plasma chromogranin A (CgA) measurement, as well as for safety analyses. Treatment compliance was assessed by patient diary cards and pill counts.

Ki67 Assessment

In order to assess Ki67 density, the initial intent was to calculate the total number of BE epithelial Ki67+ cells divided by the total number of BE epithelial cells. Initial artificial intelligence (AI) algorithms resulted in undercounting of Ki67-negative cells based on visual inspection. Thus, the decision was made to estimate density as Ki67+ cells/mm² BE epithelium. To do this, the immunostain for pan-cytokeratin was added to aid in detection of epithelial cells. Sections were stained with hematoxylin, and sequentially immunostained with Ki67 in brown (Ultraview DAB system/CC1 pretreatment solution; clone 30–9; Ventana Medical) and pan-cytokeratin in red (CK, Refine Red AP system/ER2 pretreatment solution; clone AE-1/3; Leica Biosystems). AI algorithms were developed using HALO-AI (v2.3.2089.30, Indica Labs) to classify tissues into CK-strong epithelium, CK-weak epithelium, stroma, and glass on the annotated images. Nuclear segmentation was performed and thresholds were determined for nuclear Ki67 positivity (nuclear counting algorithm). For each image, representative glands were manually counted and compared to the AI nuclear counting algorithms in order to optimize the algorithms within 5% error of manual count per image.

Gastrin and CgA Levels

Serum gastrin levels was measured by chemiluminescent assay on the Immulite® 2000 (Siemens Healthcare Diagnostics, Gwynedd, United Kingdom). Plasma CgA levels was measured by ELISA (NEOLISA^™ Chromogranin A; SVAR Life Science AB, Malmö, Sweden) on the Triturus ELISA analyser. The manufacturers’ normal ranges for gastrin and CgA are 6.2–54.8 pmol/L and ≤ 3 nmol/L, respectively.

Gene Expression Analyses

Changes in gene expression were assessed by RNA-Seq. One sample had insufficient RNA for analysis, and thus gene expression changes from 10 subjects in the netazepide arm and 9 in the placebo were analyzed. Extraction was performed with Qiagen’s miRNeasy micro kit. The quality of purified RNA samples was determined using a Bioanalyzer 2100 (Agilent). The mean RNA Integrity Number (RIN) was 8.9, and the minimum was 7.5. PolyAAA pulldown, amplification, and labeling were performed with the TruSeq Stranded mRNA kit. RNA expression was measured on an Illumina NovaSeq 6000 at the Columbia Genome Center. 20M PE 100 BP reads per sample were taken. RTA (Illumina) was used for base calling and bcl2fastq2, version 2.20 (Illumna) for converting BCL to fastq format. Adaptors were trimmed with Trimmomatic[24], fastq files were aligned to the GRCh38 assembly of the human genome with STAR[25], and genes were quantified with featureCounts.[26] Data was deposited in the Gene Expression Omnibus[27], with accession number GSE155665.

Differential expression was analyzed using weighted Limma-Voom.[28–30] Unwanted variation was taken into account with RUVSeq.[31] Estimation of the surrogate variables was obtained by using the transcription of all but the top 5000 differentially expressed genes as negative controls.[31] 15 surrogate variables were found to be optimal based upon visual inspection of p-value histograms.[32] Reactome Pathway analysis was performed with CAMERA.[33]

Study Outcomes

The primary outcome for the study was change in cellular proliferation, as assessed by Ki67 density on immunohistochemistry. Additional secondary outcomes included change in serum gastrin and plasma CgA, as well as change in expression of genes that, at the time the study protocol was originally written, were felt to be potentially relevant to gastrin/CCK₂ receptor inhibition in BE: the gastrin receptor CCK2R, cyclooxygenase-2 (PTGS2), and the tuft cell marker DCLK1. Exploratory analyses were performed to assess for additional changes in gene expression. Adverse events for netazepide and placebo were compared.

Power and Statistical Analyses

Sample size estimates were based the primary outcome, within-individual change in Ki67 density (ΔKi67). Based on prior work[11], it was estimated that baseline Ki67 density would be 45% with a common standard deviation (SD) of 14–18%. Assuming a type I error rate of 0.05, a sample size of 14 (7 per arm) would have 90% power to detect a 30% difference in ΔKi67 between the two arms if the common SD was 16%, and 87% power for a 25% difference with a 14% SD. Additional funding was obtained after the study was underway. At this point, a second study site (Cambridge, UK) was added, and the sample size was increased to 20 subjects. This sample size provided 84% power to detect a 25% difference in ΔKi67 between the two groups if the common SD was 18%, equivalent to an effect size of 1.39 SDs. A 10% drop-out rate was anticipated, and it was expected that 22 subjects would need to be enrolled.

Modified intent-to-treat analyses were performed for the primary outcome and for gene expression outcomes, with inclusion of those subjects who completed both endoscopies. All other analyses were performed on an intent-to-treat basis. Baseline characteristics such as age, gender, race, body mass index (BMI), waist circumference, smoking history, proton pump inhibitor (PPI) use (once or twice daily), and aspirin use, were compared between netazepide and placebo arms using t-tests for normally distributed continuous variables, Wilcoxon rank sum tests for non-normally distributed continuous variables, or Fisher’s exact tests for categorical variables. Linear regression models were used to compare the primary endpoint for treatment effect (Ki67 density). Factors associated with baseline serum gastrin and plasma CgA levels in all subjects and within-individual ΔKi67 in the netazepide arm were also investigated by linear regression models. A p-value <0.05 was considered statistically significant. All analyses were performed using SAS version 9.4.

RESULTS

A total of 27 subjects were enrolled in the study, and 20 completed the study and were included in the primary analyses. (Figure 1) Three subjects were screen failures, and 24 subjects were randomized. Three subjects were subsequently withdrawn when pathology from the baseline endoscopy was read as indefinite for dysplasia (2) or low grade dysplasia (1). An additional patient was withdrawn shortly after the baseline endoscopy and starting study drug when he was diagnosed with recurrent, metastatic prostate cancer. The patient characteristics are shown in Table 1. There was a significant difference between the two groups in the length of BE (p=0.008) and hiatal hernia size (p=0.003).

Figure 1.

CONSORT diagram for randomized, double-blind, placebo-controlled trial of netazepide in patients with Barrett’s esophagus without dysplasia.

Table 1.

Characteristics of all subjects randomized (n=24).

	netazepide (n=13)	placebo (n=11)	p-value
Age, years	64.4 (±7.2)	68.6 (±6.6)	0.15
Male sex	11 (84.6%)	10 (90.9%)	0.64
Race, white	13 (100%)	11 (100%)	1.00
BMI	28.9 (±4.7)	26.7 (±3.1)	0.20
Waist circumference, cm	95.7 (±17.8)	98.1 (±10.9)	0.71
Current smoker	0 (0%)	0 (0%)	1.00
PPI frequency			0.73
Once daily	8 (61.5%)	6 (54.6%)
Twice daily	5 (38.5%)	5 (45.5%)
Aspirin use	3 (23.1%)	2 (18.2%)	0.77
BE length†, cm	3 (IQR 4–2=2)	6 (IQR 9–4=5)	0.008
Hiatal hernia size, cm	2 (IQR 4–2=2)	4 (IQR 5–3=2)	0.003
Serum gastrin‡, pmol/L	60.3 (±41.7)	51.8 (±29.1)	0.58
Plasma CgA‡, nmol/L	12.9 (±17.2)	9.1 (±5.0)	0.46

^†Maximal BE length (Prague M)

^‡At baseline

There were no differences between the two arms in baseline gastrin (p=0.58) or CgA (p=0.46). Baseline serum gastrin was significantly higher in patients on twice daily PPIs (twice daily: mean 75.4 pmol/L, SD 45.2; once daily: mean 42.9 pmol/L, SD 20.1; p=0.03) but was not associated with other patient characteristics. Baseline CgA was also significantly higher in twice daily PPI users (twice daily: mean 17.3 nmol/L, SD 18.5; once daily: mean 6.8 nmol/L, SD 3.4; p=0.049). In multivariable analyses, increased BMI (p=0.003) and decreased waist circumference (p=0.0005) were significantly associated with baseline CgA. After 12 weeks of treatment, subjects on netazepide had significantly increased serum gastrin (p=0.015) and decreased plasma CgA (p=0.001) compared to those on placebo (Supplementary Figure 1), suggesting that netazepide effectively blocked gastrin receptor activity in the stomach. All of the subjects had >97% compliance taking study drug based on pill count data.

Change in Cellular Proliferation

Ki67 density was assessed as the number of Ki67+ cells per mm² of Barrett’s epithelium. (Figure 2) There was no difference in baseline Ki67 density between the netazepide and placebo arms (1539 cells/ mm², SD 514 vs. 1556 cells/ mm², SD 622, respectively; p=0.95). None of the patient characteristics, medications, or endoscopic features was associated with baseline Ki67 density. There was no significant correlation between baseline Ki67 and baseline gastrin (p=0.16) or baseline CgA (p=0.53).

Figure 2.

Examples of immunostaining and automated analysis of Barrett’s esophagus biopsies immunostained for Ki67 (brown) and pan-cytokeratin (red). (A) Annotation of biopsy to include Barrett’s epithelium (40x magnification); (B) BE immunostained for Ki67 (brown) and pan-cytokeratin (red) (400x magnification); (C) AI algorithms used to identify Ki67-positive nuclei (red) (400x magnification).

There was no difference in the study primary outcome, change in cellular proliferation (ΔKi67), comparing the two arms (netazepide: +35.6 cells/mm², SD 620.7; placebo: +307.8 cells/mm², SD 640.3; p=0.35). (Figure 3) There was a non-significant, positive correlation between baseline serum gastrin and ΔKi67 (r= 0.55, p=0.099). No other factors were associated with ΔKi67.

Figure 3.

There was no significant difference in mean within-individual change in cellular proliferation in Barrett’s esophagus epithelium after 12 weeks’ treatment with netazepide compared to placebo.

Changes in Gene Expression

To explore additional effects of gastrin/CCK₂ receptor antagonism in Barrett’s esophagus tissue, changes in gene expression were assessed using RNA-Seq. There was no evidence of clustering by treatment arm. Treatment with netazepide did not result in significant changes to any of the genes chosen for a priori comparisons: CCK2R, log-fold change −1.19, p=0.16; PTGS2, log-fold change 0.31, p=0.54; DCLK1, log-fold change −0.34, p=0.33. In global gene expression analyses, 62 genes were differentially expressed in the netazepide group as compared with placebo. (Supplementary Table 1) Interestingly, netazepide treatment resulted in increased expression of genes related to gastric phenotype (TFF2, MUC5B) and certain cancer-associated markers (REG3A, PAX9, MUC1), and decreased expression of intestinal markers such as MUC2, FABP1, FABP2, and CDX1. In pathway analyses, netazepide treatment was notably associated with upregulation of IL-1, NOD 1/2, interferon alpha/beta, and interferon gamma signaling pathways, and with downregulation of various pathways associated with DNA replication and repair. (Supplementary Table 2)

Safety

No subjects developed dysplasia during the study period. One serious adverse event occurred in the netazepide arm; a subject required hospitalization for treatment of a scrotal abscess, which was deemed unrelated to treatment. The most common adverse events were headache, diarrhea, and abdominal pain; there were no differences in the frequency of adverse events between the netazepide and placebo arms. (Supplementary Table 3) All adverse events were mild-to-moderate and self-limited.

DISCUSSION

In this randomized, double-blind, placebo-controlled trial of a gastrin receptor antagonist in patients with non-dysplastic Barrett’s esophagus, there was no effect of netazepide as compared to placebo on cellular proliferation. In exploratory gene expression analyses, treatment with netazepide resulted in increased expression of certain genes associated with gastric differentiation and cancer-associated markers, while there were decreases in markers of intestinal differentiation. Netazepide induced a rise in serum gastrin and decrease in plasma CgA, indicating that the drug was biologically active. Overall, the drug was well tolerated, and there were no serious adverse events related to study drug.

There are several possible explanations for the apparent lack of effect of netazepide on proliferation in BE tissue. One possible explanation is that patients who might benefit from the drug would be those with a much higher baseline circulating gastrin. In the current study, the mean baseline serum gastrin was only slightly above the normal range, and substantially lower than levels in gastric neuroendocrine tumor (gNET) patients.[18] In our previous retrospective study of BE patients on PPIs, only patients in the highest quartile of gastrin had a significantly increased odds of advanced neoplasia (HGD or EAC).[10] Similarly, in animal experiments with the BE/EAC mouse model, gastrin/CCK₂ receptor antagonism had anti-neoplastic effects only in the L2-IL-1B mice treated with PPIs or crossed with hypergastrinemic INS-GAS mice, but not in normogastrinemic L2-IL-1B mice.[14] Second, the choice of netazepide dose and duration of therapy may not have been appropriate. Netazepide 25 mg was top of the dose-response curve for inhibition of gastric acid production in healthy subjects[16], and similar dose and duration had marked biologic effects in the treatment of gNETs[18, 23]; however, perhaps higher doses are required to observe biologic effects in BE. The current trial differs from previous preclinical BE/EAC studies and the gNET trials in that these studies examined the effects of gastrin antagonism on late time points (e.g. HGD or EAC, or gNETs), and it is conceivable that gastrin may promote neoplasia only at later stages. Thus, some beneficial effects of netazepide may have been more apparent if tested in patients with low- or high-grade dysplasia. Finally, cellular proliferation was the primary outcome measure, and this marker may have been too insensitive to detect important biological effects of gastrin/CCK₂ receptor antagonism.

However, another possibility is that gastrin/CCK₂ receptor antagonism may simply have no benefit in human BE. The drug was highly effective in hypergastrinemic mice, in which CCK2R marks gastric cardia progenitor cells that show robust and long-term lineage tracing at the GE junction.[14] But while the gastrin/CCK₂ receptor marker is highly expressed and upregulated in human Barrett’s esophagus[12, 14], it remains uncertain whether CCK2R marks a gastric cardia progenitor in humans. Furthermore, even though BE cells in humans express CCK2R, the origin of these cells is currently unknown and the nature of the in vivo gastrin response uncertain. Indeed, Lavery et al. have shown that Barrett’s esophageal glands show the proliferative and stem cell architecture of pyloric glands[34], and our group recently showed in murine models that proliferation and symmetric cell division of CCK2R-expressing antro-pyloric stem cells is inhibited, rather than stimulated, by gastrin.[35]

The results of the gene expression analyses did not identify significant changes in other potential gene markers that might suggest a chemopreventive effect in BE, such as PTGS2, CCND1, or NFKB1. The sample size was relatively small, and the study may have been underpowered to detect smaller yet still biologically meaningful effects. Nevertheless, while gene expression analyses did not demonstrate large-scale effects of netazepide on BE, netazepide did result in increased expression of markers of gastric differentiation and certain potential cancer-associated markers, along with reductions in markers of intestinal differentiation. This pattern contrasts with what might be expected for a chemopreventive agent in BE. Several studies from our group[36] as well as others[37] have shown that intestinal differentiation is a negative prognostic marker for progression to EAC, and that progression to EAC is associated with reduced goblet cell density and increased levels of Notch expression.[38, 39] Thus, the decreased expression of intestinal type genes such as MUC2, FABP1, FABP2, and CDX1 together with increased expression of gastric differentiation genes would if anything suggest the opposite of chemoprevention.

The current study has several strengths. The study design permitted the assessment of gastrin/CCK₂ receptor antagonism as compared to placebo, with tissue analysis before and after treatment. Measurement of cellular proliferation was highly rigorous, with automated analyses calibrated by sample to minimize the error rate. Both compliance assessment and gastrin and CgA measurements served to confirm that netazepide was being taken and had biological activity. The RNA sequencing analyses provided detailed insight into the effects of gastrin/CCK₂ receptor antagonism in BE tissue, which had not previously been elucidated.

There were also certain limitations. The sample size was relatively small, and it is possible that the study was underpowered to detect smaller yet still biologically meaningful changes. Despite randomization, patients treated with placebo had longer BE segments and larger hiatal hernias. However, it is unlikely that these differences impacted the results; there were no baseline differences in Ki67 between the groups, and BE length and hiatal hernia size were not associated with change in cellular proliferation. Only one dose was tested, and higher doses of netazepide may have been more bioactive in BE tissue. However, the dose and duration used in the current study were the same as used in prior trials of gastric neuroendocrine tumors, where biological effects were clearly observed.[18, 23] The study was restricted to BE patients without dysplasia, and thus potential chemopreventive effects of gastrin receptor antagonism at later stages along the pathway to EAC could not be assessed. Other medications such as aspirin could interact with the effects of netazepide, although this could not be assessed in the current study. Finally, drug compliance may have been poor, although this was not supported by pill count data.

In conclusion, in this randomized, placebo-controlled, double-blind trial of a gastrin/CCK₂ receptor antagonist in patients with non-dysplastic BE, netazepide had no effect on cellular proliferation, the primary endpoint. Overall, netazepide was safe and well-tolerated. Exploratory gene expression analyses suggested that netazepide may possibly promote gastric and not intestinal differentiation. Gastrin may not have pro-neoplastic effects in Barrett’s esophagus, although future studies are warranted to explore this further.