Efficacy and safety of neuromodulators combined with proton pump inhibitors for non-erosive reflux disease: A randomized clinical trial

Abstract

BACKGROUND

Treatment with proton pump inhibitors in non-erosive reflux disease (NERD) remains unsatisfactory, imposing a considerable burden due to persistent symptoms, impaired quality of life, and recurrent healthcare use. Because the pathogenesis of NERD often involves a complex interplay among acid reflux, esophageal hypersensitivity, and altered gut-brain interactions, the potential value of early adjunctive neuromodulation in treatment-naive patients remains uncertain.

AIM

To investigate the efficacy, safety, and cost-effectiveness of low-dose flupentixol-melitracen (FM) combined with lansoprazole (LPZ) as an initial regimen for patients with NERD.

METHODS

Patients with NERD were randomly assigned to 2 groups for 2 weeks of initial treatment with 30 mg LPZ plus 10.5 mg FM once daily (FM + LPZ) or 30 mg LPZ plus placebo. NERD was diagnosed based on typical reflux symptoms, gastroesophageal reflux disease questionnaire score, and the absence of mucosal erosions on endoscopy. Twenty-four-hour pH-impedance monitoring was not performed. During the subsequent 10-week on-demand treatment period, patients were advised to take the assigned initial treatment dose for 3 consecutive days if symptoms recurred. The primary endpoint was the percentage of patients achieving adequate relief on day 14. Secondary endpoints included health-related quality of life, psychological condition, sleep quality, and cost-effectiveness, which were evaluated at weeks 2 and 12. Treatment satisfaction and adherence were also assessed.

RESULTS

From April 25 to September 5, 2025, 164 patients were included in the intent-to-treat (ITT) population, and 154 were included in the per-protocol population. Combination therapy demonstrated superior efficacy in the ITT population at week 2 (86.6% vs 62.2%; P < 0.001). Significant differences were observed in most secondary endpoints. The cost-effectiveness ratio was lower in the FM + LPZ group than in the LPZ plus placebo group (47.5 vs 95.2). The incidence of adverse events did not differ between the groups.

CONCLUSION

Compared with placebo plus LPZ, combination therapy with FM and LPZ as an initial treatment regimen showed better therapeutic efficacy and a lower economic burden for patients with NERD during the 12-week study period.

Key Words: Non-erosive reflux disease; Neuromodulator; Lansoprazole; Flupentixol-melitracen; Symptom relief; Proton pump inhibitors

Core Tip: This randomized, double-blind, placebo-controlled trial evaluated the efficacy, safety and cost-effectiveness of low-dose flupentixol-melitracen combined with lansoprazole as an initial regimen for non-erosive reflux disease. The findings demonstrate that combining short-term, low-dose neuromodulation with a proton pump inhibitor as initial therapy in treatment-naive patients provides rapid, cost-effective symptom relief during the 12-week study period. This early-intervention strategy may help reduce symptom burden and offer a potential alternative to standard proton pump inhibitor monotherapy for patients with non-erosive reflux disease.

INTRODUCTION

Non-erosive reflux disease (NERD) refers to the presence of typical symptoms, such as acid reflux and heartburn, without obvious esophageal mucosal erosions or ruptures on endoscopy[1]. As one of the major subtypes of gastroesophageal reflux disease (GERD), NERD accounts for approximately 70% of patients with GERD[2,3]. In recent decades, the prevalence of GERD has increased markedly. The overall global burden of GERD has continued to worsen, with prevalence increasing by 77.53%, from 441.57 million cases in 1990 to 783.95 million cases in 2019[4]. Today, drug evaluation is not limited to efficacy and safety. Comprehensive pharmacoeconomic analyses are also essential to guide clinical decision-making among multiple therapeutic options[5].

Currently, proton pump inhibitor (PPI)-based acid suppression remains the cornerstone of NERD management recommended by clinical guidelines[6,7]. However, relevant studies have demonstrated that patients with NERD have a 20%-30% lower response rate to standard acid suppression therapy than patients with reflux esophagitis[8,9]. This disparity may stem from the distinct pathogenesis of NERD, which is characterized by substantial heterogeneity of visceral hypersensitivity[10,11]. Indeed, esophageal hypervigilance persists in patients with reflux, irrespective of acid burden and symptom-reflux correlation[12]. Importantly, amplification of this abnormal perception is closely linked to neuropsychiatric factors through the bidirectional brain-gut axis[13,14]. For instance, anxiety and depression may lower esophageal sensory thresholds via hypothalamic-pituitary-adrenal axis dysregulation[15], while persistent symptoms conversely exacerbate psychological distress[16,17]. These bidirectional interactions increase the risk of psychiatric comorbidity and healthcare utilization, providing a rationale for symptom-directed neuromodulator therapy in patients with NERD. However, whether earlier use of a low-dose neuromodulator in combination with PPI therapy can improve symptom control in treatment-naive patients remains unclear.

Multiple long-term studies[18,19] have shown that a few patients with NERD (less than 20%) are likely to progress to reflux esophagitis. Therefore, therapy should prioritize symptom control and quality-of-life improvement over mucosal healing. Central neuromodulators, including tricyclic antidepressants, selective serotonin reuptake inhibitors, and serotonin-norepinephrine reuptake inhibitors, are increasingly used as pain modulators in the treatment of esophageal diseases beyond functional esophageal disorders, including NERD[20]. However, these conventional neuromodulators typically require 4-6 weeks for therapeutic onset, which creates challenges for patient adherence and increases the risk of antidepressant discontinuation syndrome (ADS)[21]. This limitation highlights the need for low-dose, fast-acting and short-course neuromodulator therapies. In recent years, flupentixol-melitracen (FM), a compound preparation combining a tricyclic antidepressant with a dopamine D1 and D2 receptor antagonist, has been widely used in disorders of gut-brain interaction and refractory GERD[22,23]. Melitracen, as the primary component of FM, inhibits serotonin and norepinephrine reuptake, augments descending pain-inhibitory pathways, and attenuates ascending esophageal afferent signals, whereas flupentixol modulates dopaminergic dysregulation, a key contributor to dysfunction of the brain-gut axis[21,23]. Previous studies have shown that short-course FM therapy elicited rapid-onset symptom responses in disorders of gut-brain interaction at low doses, with a favorable safety profile[24]. Therefore, we hypothesize that co-administration of FM with a PPI as initial therapy for NERD could provide rapid symptom relief and mitigate visceral hypersensitivity by modulating the brain-gut axis. This approach has the potential to improve treatment response rates while possibly reducing overall healthcare costs.

This trial aimed to evaluate the efficacy and safety of combined short-term, low-dose FM and lansoprazole (LPZ) as an initial therapeutic regimen for patients with NERD. The findings of this study are expected to provide preliminary clinical evidence for a more individualized and cost-effective first-line management strategy for NERD.

MATERIALS AND METHODS

Participants

This study was approved by the Clinical Research Ethics Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University (approval No. LY2025-040-A) and was prospectively registered before patient enrollment. Written informed consent was obtained from each patient before enrollment. Patients were diagnosed with NERD if they met the following criteria: (1) Heartburn or regurgitation, either as the main symptom, occurring at least 3 days per week and present for ≥ 3 months before enrollment[25,26]; and (2) No esophageal mucosal lesions on endoscopy performed within 6 months before enrollment or at the time of enrollment[1]. Routine 24-hour pH-impedance monitoring was not performed due to limited clinical feasibility and patient acceptance[27].

Subjects who met all of the following criteria were eligible to enter this study: (1) Age 18-65 years; (2) Fulfillment of the above diagnostic criteria; and (3) Gastroesophageal reflux disease questionnaire (GerdQ) score ≥ 8, a well-validated tool for supporting the diagnosis and assessment of NERD[28,29]. The exclusion criteria were (1) Previous NERD-related treatment, including acid-suppressive agents, antacids, prokinetics, neuromodulators, probiotics, or other medications used for symptom relief; (2) Endoscopic evidence of erosive esophagitis or Barrett’s esophagus; (3) Other organic gastrointestinal diseases, such as peptic ulcer, inflammatory bowel disease, or malignancy; severe dysfunction or congenital disease of major organs; (4) Pregnancy or lactation; current or recent use of monoamine oxidase inhibitors within 5 weeks; (5) Severe anxiety or depressive symptoms, defined as a Generalized Anxiety Disorder 7-item scale (GAD-7) score ≥ 15 or Patient Health Questionnaire-9 (PHQ-9) score ≥ 15; (6) Participation in another clinical trial within the preceding 3 months; and (7) Known allergy to the study medications.

During the study, patients were not permitted to use other medications that could affect efficacy evaluation, such as histamine-2 receptor antagonists, prokinetic agents, or other neuromodulators. Any use of these prohibited medications was recorded as a protocol deviation.

Study design and treatments

Participants were randomly assigned in a 1:1 ratio to the FM + LPZ group or the LPZ plus placebo (P + LPZ) group. The randomization sequence was generated by designated staff independent of the research team using a computer-based permuted block randomization scheme with a fixed block size of 4. Allocation concealment was ensured through sequentially numbered, opaque, sealed envelopes prepared by a research assistant who was not involved in patient recruitment or clinical assessment. Investigators, participants, and outcome assessors were blinded to group assignment. Data collection and analysis were also performed in a blinded manner.

The FM + LPZ group received FM (10.5 mg once daily; Haisco Pharmaceutical Group Co., Ltd) and LPZ enteric-coated capsules (30 mg once daily; Takeda Pharmaceutical Company Limited). The P + LPZ group received LPZ enteric-coated capsules (30 mg once daily) and placebo (10.5 mg once daily). The initial treatment duration was 2 weeks in both groups. After the initial treatment period, all patients were switched to on-demand therapy for 10 weeks[27]. Symptom relapse was defined as bothersome reflux or heartburn symptoms occurring more than twice per week, indicated by a GerdQ heartburn or regurgitation item score ≥ 2, or the use of medication for reflux or heartburn symptoms for more than 7 days within 4 weeks[30,31]. If symptoms recurred, patients were instructed to take the initial treatment dose for 3 consecutive days[31]. During the trial, participants were required to maintain medication and symptom records. These records included reflux-related symptoms, sleep status, emotional changes, adverse events (AEs), and the date of each use of on-demand medication. The records were collected and verified by the investigators through scheduled follow-up interviews. To maintain blinding, FM and placebo tablets were assessed by two physicians not involved in the trial and were considered indistinguishable in appearance, color, odor, and packaging. During the trial, all participants were instructed to maintain their usual dietary habits and lifestyles. Participants in both groups received general lifestyle advice, such as avoiding excessive alcohol consumption, smoking, caffeine intake, and extremely spicy or greasy foods.

Data collection

All patients were required to complete a standardized questionnaire at enrollment to obtain complete medical history and demographic data. Baseline symptoms (GerdQ), psychological status (PHQ-9 and GAD-7), sleep status [Pittsburgh Sleep Quality Index (PSQI)], and quality of life [Gastroesophageal Reflux Disease Health-Related Quality of Life (GERD-HRQL)] were assessed at the initial visit[32,33]. GerdQ was assessed every 2 weeks during the study. At the end of initial treatment and on-demand therapy, PHQ-9, GAD-7, PSQI, GERD-HRQL, and cost data were collected at the end of the initial treatment period and on-demand therapy (Figure 1). Treatment satisfaction was evaluated using a 5-point Likert scale: (1) Very dissatisfied (1 point); (2) Dissatisfied (2 points); (3) Somewhat satisfied (3 points); (4) Satisfied (4 points); and (5) Very satisfied (5 points). All AEs were recorded, including severity, relationship to the study drug, management and outcomes.

Figure 1 The protocol of the study. Gastroesophageal reflux disease questionnaire scores were collected every 2 weeks during the study. Patient Health Questionnaire-9, Generalized Anxiety Disorder 7-item scale, Pittsburgh Sleep Quality Index, Gastroesophageal Reflux Disease Health-Related Quality of Life were collected at baseline, week 2 and week 12. Medication possession ratio was calculated at week 2. Cost-effectiveness analysis was calculated at week 12. CEA: Cost-effectiveness analysis; GAD-7: Generalized Anxiety Disorder 7-item scale; GerdQ: Gastroesophageal Reflux Disease Questionnaire; GERD-HRQL: Gastroesophageal Reflux Disease Health-Related Quality of Life; PHQ-9: Patient Health Questionnaire-9; PSQI: Pittsburgh Sleep Quality Index.

Outcome measures

The primary endpoint was the treatment response rate, defined as the percentage of patients achieving adequate symptom relief at the end of the initial 2-week treatment period. Adequate symptom relief was defined as the absence of bothersome heartburn or regurgitation, indicated by individual GerdQ item scores ≤ 1 during the last 7 days of the initial 2-week therapy[1,34,35].

Secondary endpoints included the treatment response rate at week 12, changes in GerdQ scores, number of days requiring on-demand therapy, quality of life as assessed by changes in GERD-HRQL scores, psychological status as assessed by changes in PHQ-9 and GAD-7 scores, sleep status as assessed by changes in PSQI scores, cost-effectiveness analysis (CEA), and treatment satisfaction.

Additionally, to investigate the impact of comorbid psychological symptoms on drug efficacy, participants with a PHQ-9 score ≥ 5 or GAD-7 score ≥ 5 were classified into the subpopulation with psychological symptoms.

CEA

Costs: Direct medical costs during the study period were included in the analysis. Direct medical costs refer to expenses directly related to the diagnosis and treatment of the disease, including consultation, examination, medication, hospitalization, and adverse event management. No NERD-related hospitalizations were observed during the study, and the AEs reported by patients were mild and self-limited. Therefore, these cost components were excluded. Although indirect costs and productivity losses are important economic dimensions, they were excluded to enhance comparability because of substantial interindividual variability in employment status and regional disparities in healthcare access. All costs were standardized to 2024 United States dollars (USD) using the annual average exchange rate (1 CNY = 0.140 USD). Discounting was not applied because the study period was shorter than 1 year.

Sensitivity analysis: To assess the robustness of the results, a nonparametric bootstrap probabilistic sensitivity analysis with 1000 stratified replications was performed in this study[36]. The results were shown in the scatter plots of incremental cost-effectiveness ratio (ICER)[37].

Statistical analysis

The sample size was calculated assuming a 20% difference between the two groups, with a significance level of 0.05 (α) and 80% statistical power in a two-sided test[35,38]. Accounting for an anticipated dropout rate of approximately 10%, the total sample size was 164 subjects, with 82 allocated to each arm. The primary efficacy analysis was performed in the intent-to-treat (ITT) population, which included all randomized patients analyzed according to their originally assigned groups. Missing data were imputed using the last-observation-carried-forward method. Medication adherence was assessed using the medication possession ratio (MPR). During the 2-week phase, adherence was monitored using patient medication diaries and confirmed during follow-up interviews at week 2. MPR was defined as the number of pills taken relative to the prescribed dose during the initial treatment period[39]. Participants who received less than 80% of the total prescribed medication during the initial treatment phase, were lost to follow-up, or had major protocol deviations, including the use of prohibited medications were excluded from the per-protocol (PP) analysis.

All statistical analyses were performed using SAS 9.4 software. Continuous variables are presented as the mean ± SD if normally distributed, whereas non-normally distributed variables are presented as median [interquartile range (IQR)]. For the primary efficacy endpoint, relative risks (RRs) and 95%CIs were calculated. The Wilcoxon signed-rank test or paired t-test was used to analyze changes in post-treatment scores from baseline within each group. Comparisons between the two groups were performed using the Mann-Whitney U test or unpaired t-test, depending on data distribution. Longitudinal changes in clinical outcomes across multiple time points were analyzed using linear mixed-effects models. Multiple comparisons across time points were controlled using the Bonferroni correction. Categorical variables were compared using the χ² test.

RESULTS

Demographics and baseline clinical characteristics

This study included 164 patients with NERD, who were randomly assigned in a 1:1 ratio to either the FM + LPZ group or the P + LPZ group. The screening and recruitment process for study participants is shown in Figure 2. In total, 164 patients were included in the ITT analysis. Ten participants were excluded from the PP analysis: (1) Six in the FM + LPZ group; and (2) Four in the P + LPZ group. Reasons for exclusion included loss to follow-up due to inability to contact the participants in seven cases (four in the FM + LPZ group and three in the P + LPZ group) and poor adherence to study medication in three cases (two in the FM + LPZ group and one in the P + LPZ group). Finally, 154 patients (76 in the FM + LPZ group and 78 in the P + LPZ group) completed the treatment and follow-up procedures and were included in the PP analysis. Groups were well-balanced in terms of demographic characteristics and baseline clinical features (Table 1).

Figure 2 Workflow of randomization and treatment procedures of enrolled patients. ITT: Intent-to-treat; PP: Per-protocol; FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

Table 1 Demographic and baseline clinical characteristics, median (interquartile range)/n (%).

Characteristics	FM + LPZ group (n = 82)	P + LPZ group (n = 82)	P value
Age (years), mean (SD)	47.0 (10.8)	46.9 (12.2)	0.941
Female	37 (45.1)	35 (42.7)	0.753
Weight (kg), mean (SD)	66.1 (11.9)	63.5 (10.9)	0.141
Height (cm), mean (SD)	167.8 (8.5)	166.9 (7.0)	0.436
BMI	23.2 (21.1-24.8)	22.8 (20.7-24.6)	0.329
Education level			0.439
Junior high school or below	21 (25.6)	23 (28.0)
Senior high school	16 (19.5)	10 (12.2)
Junior college or higher	45 (54.9)	49 (59.8)
Job category			0.655
Mainly manual labor	9 (11.0)	9 (11.0)
Mainly intellectual labor	33 (40.2)	30 (36.6)
Retired	25 (30.5)	32 (39.0)
Uncertain	15 (18.3)	11 (13.4)
Smoker	14 (17.1)	11 (13.4)	0.515
Alcohol drinker	12 (14.6)	9 (10.9)	0.483
Coffee	6 (7.3)	8 (9.8)	0.576
Spicy food	22 (26.8)	21 (25.6)	0.859
Sweet food	18 (21.9)	14 (17.1)	0.431
Greasy food	16 (19.5)	10 (12.2)	0.199
GerdQ scores	12.0 (10.0-14.0)	12.0 (10.0-15.0)	0.255
GERD-HRQL scores, mean (SD)	13.2 (5.0)	13.6 (4.7)	0.596
GAD-7 scores,	3.0 (1.0-7.0)	3.0 (2.0-5.0)	0.885
PHQ-9 scores	4.0 (1.0-8.0)	4.5 (2.0-8.0)	0.788
GAD-7 ≥ 5 or PHQ-9 ≥ 5	44 (53.7)	45 (54.9)	0.875
PSQI scores, mean (SD)	7.4 (4.4)	7.8 (4.1)	0.596

BMI: Body mass index; GAD-7: Generalized Anxiety Disorder 7-item scale; GerdQ: Gastroesophageal reflux disease questionnaire; GERD-HRQL: Gastroesophageal Reflux Disease Health-Related Quality of Life; PHQ-9: Patient Health Questionnaire-9; PSQI: Pittsburgh Sleep Quality Index; FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

Primary efficacy analysis

After the initial treatment, the proportion of patients who achieved the primary endpoint was significantly higher in the FM + LPZ group than in the P + LPZ group in the ITT population [86.6% (71/82; 95%CI: 77.6%-92.3%) vs 62.2% (51/82; 95%CI: 51.4%-71.9%); P < 0.001; Figure 3]. In the PP population, the difference in treatment response rate between the two groups remained consistent (85.5% vs 62.8%; P = 0.001). Compared with baseline, both groups showed significant reductions in GerdQ scores. Notably, the median decrease was greater in the FM + LPZ group than in the P + LPZ group [4.0 (IQR: 3.0-8.0) vs 3.0 (IQR: 2.0-6.0); P < 0.001]. In the exploratory subgroup analysis, FM combined with LPZ showed greater efficacy than placebo combined with LPZ in patients with psychological symptoms (RR = 1.44; 95%CI: 1.10-1.88) and in those without such symptoms (RR = 1.34; 95%CI: 1.03-1.75). No significant inconsistency in treatment effects was observed across psychological-symptom subgroups (interaction P = 0.706; Table 2).

Figure 3 Primary efficacy outcomes. Comparison of treatment response rate in intent-to-treat populations at week 2. Lansoprazole plus placebo: 51/82 (62.2%, 95%CI: 51.4%-71.9%); flupentixol-melitracen + lansoprazole: 71/82 (86.6%, 95%CI: 77.6%-92.3%). FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

Table 2 Subgroup analysis of the primary outcome.

Subgroup	Number of responsive/total patients		RR (95%CI)	P value for interaction
	FM + LPZ group	P + LPZ group
All patients	71/82	51/82	1.39 (1.15-1.68)
Mood disorders
With anxiety/depression	38/44	27/45	1.44 (1.10-1.88)	0.706
No anxiety/depression	33/38	24/37	1.34 (1.03-1.75)

RR: Relative risk; FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

During the on-demand maintenance phase, significant differences existed between the two groups in the changes of GerdQ scores over time (P < 0.001). Specifically, the P + LPZ group experienced significant symptom relapse, as indicated by an increase in scores from week 2 to week 10 (P = 0.011), whereas the FM + LPZ group maintained a stable response. At week 12, 84.1% of patients in the FM + LPZ group achieved adequate symptom relief in the ITT population (69/82; 95%CI: 76.4%-92.0%), compared with 57.3% in the P + LPZ group (47/82; 95%CI: 46.5%-68.1%; P < 0.001). Similar results were observed in the PP analysis, with a higher rate of adequate symptom relief in the FM + LPZ group than in the P + LPZ group [82.9% (63/76) vs 57.7% (45/78); P < 0.001]. The average number of days requiring on-demand medication was lower in the FM + LPZ group than in the P + LPZ group (8.2 ± 6.8 vs 21.3 ± 16.6; P < 0.001). The cumulative incidence of symptom relapse within 10 weeks after initial therapy was 38.2% (29/76) among patients who received FM with LPZ therapy and 70.5% (55/78) among those who received placebo with LPZ therapy. The FM + LPZ group had a lower symptom relapse rate than the P + LPZ group (P < 0.001).

Treatment outcomes of psychological conditions and sleep quality

The FM + LPZ combination demonstrated significantly greater benefits in alleviating psychological distress and improving sleep quality than P + LPZ from baseline to week 2 (P < 0.001; Table 3).

Table 3 Symptom scores at baseline, week 2 and week 12 in the per-protocol population, mean (SD)/median (interquartile range).

Variable	Time	FM + LPZ	P + LPZ	P value
GerdQ	Week 0	12.0 (10.0-14.0)	12.0 (10.0-15.0)
	Week 2	6.0 (6.0-8.0)	8.0 (7.0-10.0)	< 0.001
	Week 12	6.0 (6.0-8.0)	8.5 (8.0-10.0)	< 0.001
GAD-7	Week 0	3.0 (1.0-7.0)	3.0 (2.0-5.0)
	Week 2	2.0 (0.0-3.0)	3.0 (2.0-4.0)	< 0.001
	Week 12	2.0 (0.5-3.0)	3.0 (2.0-3.0)	< 0.001
PHQ-9	Week 0	4.0 (1.0-8.0)	4.5 (2.0-8.0)
	Week 2	2.0 (0.0-3.0)	3.5 (2.0-6.0)	< 0.001
	Week 12	2.0 (0.0-4.0)	3.5 (2.0-6.0)	< 0.001
PSQI	Week 0	7.4 (4.4)	7.8 (4.1)
	Week 2	5.4 (3.0)	7.6 (4.0)	< 0.001
	Week 12	4.7 (2.7)	7.7 (4.3)	< 0.001
GERD-HRQL	Week 0	13.2 (5.0)	13.6 (4.7)
	Week 2	3.8 (3.2)	7.7 (4.7)	< 0.001
	Week 12	2.7 (2.5)	7.3 (3.8)	< 0.001

GAD-7: Generalized Anxiety Disorder 7-item scale; GerdQ: Gastroesophageal reflux disease questionnaire; GERD-HRQL: Gastroesophageal Reflux Disease Health-Related Quality of Life; PHQ-9: Patient Health Questionnaire-9; PSQI: Pittsburgh Sleep Quality Index; FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

Specifically, the FM + LPZ group showed greater reductions in anxiety and depressive symptoms, as assessed by GAD-7 and PHQ-9 scores at week 2 (P < 0.001). Regarding sleep quality, PSQI scores decreased by 1.9 ± 3.1 in the FM + LPZ group and by 0.2 ± 0.7 in the P + LPZ group at week 2 (P < 0.001). From week 2 to week 12, the PSQI scores showed significant improvement in the FM + LPZ group (P = 0.010), whereas no obvious change was observed in the P + LPZ group (P = 0.581), indicating a sustained benefit of FM combined with PPI therapy on sleep quality.

Quality of life

The GERD-HRQL scores decreased significantly in both groups by the end of the initial 2-week treatment period. Notably, the magnitude of reduction was significantly greater in the FM + LPZ group than in the control arm (P < 0.001; Table 3). During the on-demand therapy, quality of life continued to improve in the FM + LPZ group (P = 0.006), whereas no significant improvement was observed in the P + LPZ group (P = 0.355; Table 3).

Treatment satisfaction and medication adherence

At the end of the initial (2 weeks) and the subsequent on-demand (10 weeks) treatment, participants were asked to rate their overall satisfaction. At week 2, the proportion of participants reporting positive satisfaction (“satisfied” or “very satisfied”) was significantly higher in the FM + LPZ group than in the P + LPZ group [81.7% (67/82) vs 64.6% (53/82); P = 0.014; Figure 4]. From week 2 to week 12, positive satisfaction rates remained stable in the FM + LPZ group (χ² = 0.400; P = 0.527), whereas the P + LPZ group showed a significant decline in satisfaction over the same period (χ² = 4.546; P = 0.033).

Figure 4 Treatment satisfaction of patients at week 2. FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo.

Medication compliance was also evaluated in the ITT population. The mean MPR was 0.86 ± 0.23 in the FM + LPZ group and 0.89 ± 0.21 in the P + LPZ group (P = 0.388). Both groups demonstrated excellent adherence, which may be attributable to the short treatment duration.

Result of CEA

The mean medication cost per patient was lower in the FM + LPZ group than in the P + LPZ group (USD 40.6 vs USD 59.8). The cost-effectiveness ratio was also lower in the FM + LPZ group than in the P + LPZ group (47.5 vs 95.2, Table 4).

Table 4 Cost-effectiveness between the two groups over 12 weeks.

Parameters	FM + LPZ group	P + LPZ group
Unit costs (USD)
LPZ (per tablet)	1.1	1.1
FM/placebo (per tablet)	0.1	0.0
Outpatient clinic visit (per visit)	7.0	7.0
Resource utilization (mean per patient)
Initial 2-week therapy (days)	14	14
On-demand therapy (days)	8.2	21.3
Outpatient clinic visits (times)	2	3
Cost-effectiveness
Total cost (USD)	40.6	59.8
CER (USD/treatment response rate)	47.5	95.2
ICER (USD/treatment response rate)	-84.6

Placebo tablets were produced at negligible cost and were therefore not included in the cost calculation. FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo; USD: United States dollars; CER: Cost-effectiveness ratio; ICER: Incremental cost-effectiveness ratio.

Furthermore, the ICER was calculated by comparing the additional cost per additional unit of effectiveness gained with one treatment over the other. The scatter plot in Figure 5 shows that all 1000 simulated points are distributed in the fourth quadrant, indicating that FM combined with LPZ was more effective at a lower cost than placebo combined with LPZ. Sensitivity analysis confirmed that this conclusion was robust to parameter fluctuations.

Figure 5 Scatter plot of incremental cost-effectiveness probability sensitivity analysis (n = 1000). The fourth quadrant indicated that the flupentixol-melitracen + lansoprazole regimen is an advantageous solution (with good performance and low cost).

Safety

Four patients [4.9% (4/82)] in the FM + LPZ group and two patients [2.4% (2/82)] in the P + LPZ group reported AEs during treatment (P = 0.682; Table 5). These events occurred primarily during the first 2 weeks.

Table 5 Summary of adverse events during treatment, n (%).

AEs	FM + LPZ (n = 82)	P + LPZ (n = 82)
Patients with AEs	4 (4.9)	2 (2.4)
Constipation	2 (2.4)	0 (0)
Nausea	0 (0)	1 (1.2)
Sleep disturbance	1 (1.2)	1 (1.2)
Increased heart rate	1 (1.2)	0 (0)

FM: Flupentixol-melitracen; LPZ: Lansoprazole; P + LPZ: Lansoprazole plus placebo; AEs: Adverse events.

Specifically, the reported events in the FM + LPZ group included constipation [2.4% (2/82)], sleep disturbance [1.2% (1/82)], and increased heart rate [1.2% (1/82)]. In the P + LPZ group, nausea [1.2% (1/82)] and sleep disturbance [1.2% (1/82)] were reported. Overall, all reported reactions were mild and resolved spontaneously by the end of the study. No extrapyramidal effects, withdrawal reactions, or life-threatening AEs were observed in either group. No patient discontinued treatment because of AEs.

DISCUSSION

Patients with NERD often experience chronic symptoms that are difficult to manage with standard therapeutic approaches[40]. Moreover, excessive dependence on PPIs may disrupt intestinal microbiota homeostasis and impair calcium absorption[41,42]. In this study, short-term, low-dose FM combined with LPZ was associated with greater symptom improvement and lower direct medical costs in treatment-naive patients during the 12-week study period. Our findings suggest the potential therapeutic value of early intervention with a short-term, low-dose neuromodulator in patients with NERD to achieve rapid symptom relief.

The major novelty of the present study lies in the use of a neuromodulator as an initial therapeutic option for treatment-naive patients with NERD. We found that patients in the FM + LPZ group showed a significantly higher response rate and a greater reduction in GerdQ scores after 2 weeks of treatment than patients in the P + LPZ group. Notably, this early benefit was sustained throughout the subsequent on-demand treatment period with a lower relapse rate. Previous studies have primarily focused on the application of neuromodulators in patients with refractory GERD[22], in whom the therapeutic response is often suboptimal. Whether earlier use of neuromodulators offers advantages over rescue use after PPI failure remains uncertain.

In the present study, a relatively short 2-week induction phase was followed by a 10-week on-demand regimen. This design was based on the clinical characteristics of NERD, the unique pharmacological profile of FM, and safety considerations. First, prior clinical studies have established that an early therapeutic response to acid suppression at week 2 is a crucial predictor of long-term outcomes in reflux disease[43]. Second, unlike conventional antidepressants, FM has a rapid onset of action and typically alleviates symptoms within 1–2 weeks[24,44]. Furthermore, recent evidence suggests that 2 weeks of FM therapy achieves efficacy comparable to that of a 4-week regimen in refractory functional gastrointestinal disorders while significantly reducing the risk of ADS[24,35]. Therefore, the 2-week time point may represent a pivotal clinical decision point for assessing the success of the induction phase before switching to on-demand therapy. Although this initial evaluation period was shorter than the standard 4-8-week continuous treatment regimens, our subsequent 10-week follow-up substantiated the stability and clinical relevance of this early rapid response, offering a more patient-centered and cost-effective therapeutic strategy without compromising clinical comparability.

Additionally, the benefits of FM extended beyond reflux symptoms. At weeks 2 and 12, patients receiving FM plus LPZ showed significantly greater improvements in GAD-7 and PHQ-9 scores compared with those receiving placebo plus LPZ, indicating a positive effect on psychological well-being. Improvements in sleep quality and overall quality of life were also observed in the FM + LPZ group, consistent with evidence that sleep deprivation exacerbates symptom perception in GERD[11].

Interestingly, in the exploratory subgroup analysis, the beneficial effects of FM were observed in NERD patients without overt psychological symptoms. This finding suggests that the therapeutic value of neuromodulators may extend beyond modulation of emotional or psychiatric factors and may also involve alterations in visceral perception, central pain processing, and gut-brain axis regulation[10,20,45]. This interpretation aligns with the Rome Foundation Working Team report, indicating that low-dose central neuromodulators may exert visceral analgesic effects independent of their mood-stabilizing properties[21]. Indeed, the clinically observed limited response to standard PPI monotherapy in NERD highlights the critical role of peripheral and central sensitization mechanisms in its pathogenesis[11]. As a combination of a tricyclic antidepressant (melitracen) and a typical antipsychotic agent (flupentixol), FM augments descending modulatory pain pathways by inhibiting serotonin and noradrenaline reuptake, thereby attenuating afferent nociceptive signals from the esophagus. Furthermore, these neuromodulators may help normalize altered central pain processing and reduce central hyperalgesia commonly observed in NERD. In combination with the dopamine receptor antagonism of flupentixol, these direct analgesic and desensitizing properties may account for the rapid relief of esophageal symptoms in patients with NERD.

Patient satisfaction serves as an alternative endpoint for assessing treatment response in the NERD population. Patients receiving the combination therapy with FM and LPZ displayed a significantly higher positive satisfaction rate than those receiving placebo plus LPZ. Moreover, the satisfaction rate in the FM + LPZ group remained high from week 2 to week 12, indicating sustained superiority over time.

Regarding the economic burden during the 12-week study period, the ICER analysis indicated that FM + PPI therapy was more cost-effective than P + PPI therapy. This advantage may result from a reduced risk of recurrence, decreased use of on-demand medication, and fewer clinic visits. Our results indicate that neuromodulators combined with PPIs as an initial treatment regimen may benefit patients with NERD, with respect to both clinical outcomes and short-term health care burden.

This study further validated the safety of FM therapy. Concerns regarding potential extrapyramidal side effects and withdrawal reactions associated with long-term use of FM have limited patient adherence in clinical practice[20]. To address these concerns, a low-dose, short-course regimen was adopted in the present study based on our previous findings[24]. This approach aimed to maximize therapeutic efficacy while minimizing AEs, thereby improving both safety and tolerability. All patients were closely monitored at 2-week intervals during follow-up, and no significant withdrawal symptoms or rebound psychological distress were observed. These findings indicate that 2-week low-dose exposure carries a minimal risk of ADS, allowing a smooth transition to on-demand maintenance therapy.

This research has several limitations. First, diagnosing NERD without 24-hour pH-impedance monitoring likely resulted in a heterogeneous cohort that included “true” acid reflux, reflux hypersensitivity, and functional heartburn[46]. In the absence of objective reflux phenotyping, we could not determine which specific subgroup benefited most from FM + LPZ therapy. This limitation may also have amplified the apparent benefit of neuromodulator therapy in symptom-driven phenotypes, thereby limiting the pathophysiological generalizability of our findings. Second, we did not formally assess the success of blinding. Although the risk of unblinding was low due to the short treatment period (2 weeks), low dose, and low incidence of AEs, the possibility of inadvertent unblinding of participants or investigators cannot be completely excluded. This may have introduced expectation bias into subjective endpoints such as GerdQ and psychological scores. Third, the exclusion of patients with severe anxiety or depression (GAD-7/PHQ-9 ≥ 15) limits the external validity of our findings. Therefore, the present results should be interpreted primarily in treatment-naive patients with NERD without severe psychological symptoms. Fourth, the single-center design and relatively short follow-up period limited the assessment of long-term safety, therapeutic efficacy, and economic outcomes. In addition, the pharmacoeconomic analysis included only short-term direct medical costs, and long-term overall health care utilization was not assessed.

CONCLUSION

This study provides novel evidence that early application of a neuromodulator–PPI combination regimen may offer superior clinical outcomes compared with conventional acid-suppressive therapy alone for the management of NERD. By employing a low-dose, short-course treatment strategy followed by 10 weeks of on-demand use, our approach achieved effective symptom relief with a favorable safety profile, high patient acceptability, and a lower economic burden. Future long-term, multicenter studies incorporating objective reflux phenotyping are imperative to confirm the durability and generalizability of these findings, thereby promoting broader adoption of this early combination therapy for treatment-naive patients with NERD.

ACKNOWLEDGEMENTS

We wish to express our sincere gratitude to Dr. Hong-Yi Qiu and Ping Xu for their valuable contributions to patient recruitment. We would also like to thank the dedication of the clinical staff involved in patient care and follow-up during the trial.