Aurantii Fructus Immaturus Flavonoid tablets for functional dyspepsia: A multicenter, double-blind, dose-exploring IIb randomized controlled trial

Abstract

BACKGROUND

Functional dyspepsia (FD) is a prevalent chronic upper gastrointestinal disorder with limited safe long-term treatment options, highlighting the need to explore novel therapies such as Aurantii Fructus Immaturus Flavonoid (AFIF).

AIM

To evaluate the efficacy and safety of AFIF tablets as a potential long-term treatment option for patients with FD in this multicenter, randomized, double-blind, placebo-controlled phase IIb trial.

METHODS

This double-blind randomized controlled trial used central stratified block randomization and was designed to enroll 400 eligible participants aged 18-65 years who met the Rome III criteria. Participants were allocated in a 1:1:1:1 ratio to receive high-dose AFIF (4 tablets/tid), medium-dose AFIF (3 tablets/tid), low-dose AFIF (2 tablets/tid), or placebo for 4 weeks. The primary efficacy endpoints were the resolution rates for all 4 FD symptoms (postprandial fullness, early satiety, epigastric burning sensation, and abdominal pain) and the individual symptom resolution rates after 4 weeks of medication. Secondary endpoints included the resolution rates of all four FD symptoms and individual symptoms at other visit time points, the postprandial 2-hour gastric emptying rate, and gastric half-emptying time. Safety was assessed through adverse events (AEs) and laboratory tests.

RESULTS

Of the 399 enrolled participants, 267 were included in the Full Analysis Set. In the Full Analysis Set, the 4-week resolution rates of all four symptoms differed significantly across groups [22.39% (high), 27.27% (medium), 2.99% (low), and 5.97% (placebo); P < 0.0001]. The high- and medium-dose AFIF groups showed higher rates than the low-dose and placebo groups, with no significant differences between the high- and medium-dose groups or between the low-dose and placebo groups. Individual symptom resolution rates showed similar patterns. AFIF improved gastric emptying, although the improvement was not significant compared with placebo. AE incidence was comparable across groups [5.97% (high), 6.06% (medium), 0.00% (low), and 7.46% (placebo); P = 0.112], with no serious AEs reported in the AFIF groups.

CONCLUSION

Medium- and high-dose AFIF significantly improved FD symptoms with good safety. The medium dose (3 tablets/tid) is recommended for phase III trials and may be a safer alternative to proton pump inhibitors or prokinetics. Limitations include small follow-up samples and restricted gastric emptying testing; therefore, larger long-term studies are needed.

Key Words: Functional dyspepsia; Traditional Chinese medicine; Aurantii Fructus Immaturus Flavonoid; Optimal dosage; Zhi Shi

Core Tip: This placebo-controlled, randomized phase IIb trial identified, for the first time, the optimal clinical dose of Zhi Shi Flavonoids for functional dyspepsia. Both the medium- and high-dose Aurantii Fructus Immaturus Flavonoid groups showed significant improvement in functional dyspepsia symptoms after 4 weeks compared with the placebo group, with good tolerability and no serious adverse events. These findings inform optimal dosing for phase III trials and support the integration of personalized traditional Chinese medicine into the modern management of gastrointestinal disorders.

INTRODUCTION

Functional dyspepsia (FD) is a disorder of gut-brain interaction characterized by chronic upper gastrointestinal symptoms, primarily postprandial fullness, early satiety, epigastric burning sensation, or pain, that cannot be explained by organic lesions, such as gastric ulcers or tumors, or by other systemic diseases[1]. The diagnosis of FD requires that symptoms persist for at least 6 months and that the above criteria be met during the preceding 3 months[2,3]. Its pathogenesis is complex and may involve delayed gastric emptying, visceral hypersensitivity, abnormal gastric acid secretion, and psychosocial factors[4]. Abnormal gut-brain interaction is considered the main pathogenic mechanism[5].

According to the World Health Organization, the global prevalence of FD exceeds 10%, and approximately 135 million FD patients are in China[6]. Studies have shown that among patients undergoing gastroscopy for dyspeptic symptoms, approximately 60%-70% are ultimately diagnosed with FD[7]. Although the disease has no impact on mortality[8,9], it seriously affects patients’ quality of life[10] and consumes substantial medical resources.

The 2022 Chinese Expert Consensus on the Diagnosis and Treatment of FD and the 2022 British Society of Gastroenterology Guidelines on the Management of FD both recommend that FD patients engage in regular aerobic exercise, whereas dietary therapy is not recommended[5,11]. Proton pump inhibitors (PPIs) are the first-line medications. For patients with Helicobacter pylori-positive coinfection, eradication therapy should be administered. Prokinetic agents may be effective for FD; however, because of their potential cardiotoxicity, these drugs are unavailable in regions outside Asia and the United States. In China, they are commonly used for FD treatment, but their use is strictly regulated.

Although PPIs are effective for short-term relief of FD symptoms, long-term use may be associated with multiple adverse reactions (ARs), including fractures, vitamin B12 and folate deficiencies, gastrointestinal infections, delayed gastric emptying, which may exacerbate dyspeptic symptoms, potential gastric tumor formation, and an increased risk of cardiovascular events[12,13]. Tricyclic antidepressants, as gut-brain neuroregulators, are effective second-line treatments for FD. However, they have significant anticholinergic side effects, may cause sleep disorders, and carry a risk of serious withdrawal reactions. Therefore, there remains a need to explore medications with good efficacy and fewer side effects for FD treatment.

Aurantii Fructus Immaturus Flavonoid (AFIF) tablets are flavonoid compounds extracted from the traditional Chinese medicine (TCM) Citrus aurantium, the dried fruit of Citrus aurantium L., and Citrus sinensis Osbeck, which is one of the main Chinese herbal medicines used to treat gastrointestinal disorders. Laboratory studies have shown that total flavonoids from Citrus aurantium tablets can treat FD by promoting gastrointestinal motility, reducing visceral hypersensitivity, regulating gastric emptying, and improving gastrointestinal adaptability[14,15].

AFIF was approved for clinical trials by China’s National Medical Products Administration (NMPA) in 2008 (approval No. 2008 L04094). Between 2009 and 2014, five clinical studies were completed: One phase I human tolerance trial, two phase II trials (phase IIa for syndrome exploration and phase IIb for dose finding), and two phase III trials (placebo-controlled and domperidone-controlled), which confirmed AFIF’s efficacy/effectiveness and safety.

In December 2015, due to concerns about data quality and integrity, the applicant voluntarily applied to the NMPA to withdraw the marketing application and conduct rigorous internal data inspections. In September 2022, a marketing application for AFIF was resubmitted to the NMPA. AFIF was successfully launched in China in October 2023 and was included in the National Medical Insurance Catalog in 2024, providing a more economical and effective treatment option for patients with FD.

The results of AFIF’s phase III domperidone-controlled clinical trial have been published[16], and the phase III placebo-controlled results are forthcoming. To ensure transparency of the complete trial data, the results of the phase IIb dose-finding study are now being disclosed, and the phase IIa TCM syndrome exploration results will be published simultaneously.

MATERIALS AND METHODS

Study design

A multicenter, randomized, double-blind, placebo-controlled, parallel-group trial was designed to evaluate the safety and efficacy of AFIF at doses of 2, 3, and 4 tablets per administration, administered three times daily, for the treatment of FD. The trial was registered at https://www.isrctn.com/, No. ISRCTN10266781. Ethical approval was obtained from the Clinical Trial Ethics Committee of West China Hospital of Sichuan University on July 28, 2011, No. 4/2011. Central stratified block randomization was used, with a 1:1:1:1 ratio among the three dose groups and the placebo group. Each group was planned to enroll 100 participants, for a total planned enrollment of 400 participants.

Participant criteria

Inclusion criteria: (1) Aged ≥ 18 years and ≤ 65 years, regardless of gender; (2) Willing to participate and sign a written informed consent form; (3) Met the Rome III diagnostic criteria[1,2] for FD (Supplementary Table 1); and (4) Had discontinued other prokinetic agents, gastric mucosal protectants, or acid-suppressing drugs for 4 weeks before enrollment.

Exclusion criteria: Subjects were excluded if they met any of the following criteria: (1) Organic lesions, including esophagitis, atrophic gastritis, gastric or duodenal ulcers, erosions, or tumors; (2) Organic diseases of the liver, gallbladder, or pancreas; (3) Diabetes, kidney disease, connective tissue disease, or psychiatric disorders; (4) Severe primary diseases involving the heart, brain, lung, kidney, hematopoietic system, or endocrine system; (5) A history of abdominal surgery; (6) Renal insufficiency, defined as serum creatinine above the upper limit of normal; (7) Hepatic insufficiency, defined as aspartate aminotransferase and/or alanine aminotransferase > 1.5 × the upper limit of normal; (8) Mental or legal disabilities; (9) Pregnancy, intention to become pregnant, or lactation; and (10) Participation in other drug clinical trials within the preceding 3 months.

During case inclusion, patients were assessed for “alarm symptoms and signs” suggestive of organic disease based on a comprehensive medical history and physical examination. Individuals were excluded from screening if they: (1) Were over 45 years old with recent-onset dyspepsia symptoms; (2) Had weight loss, anemia, hematemesis, melena, dysphagia, abdominal mass, jaundice, or other relevant findings; or (3) Experienced progressive exacerbation of dyspepsia symptoms. Before enrollment, participants were also required to undergo gastroscopy, occult blood testing of stool, and abdominal ultrasound to exclude possible organic lesions.

Withdrawal criteria

Investigator-initiated withdrawal occurred if the participant: (1) Was found not to meet the inclusion criteria or to meet the exclusion criteria during treatment; (2) Developed other diseases during the trial that affected the evaluation of efficacy or safety; (3) Could not tolerate the treatment, although the participant was included in the safety analysis; (4) Had a serious protocol violation, such as self-discontinuation of medication or misuse of other drugs affecting efficacy evaluation; or (5) Was deemed by the investigator to be inappropriate to continue the trial. Voluntary withdrawal occurred if the participant: (1) Withdrew consent to continue; or (2) Was lost to follow-up.

Intervention

Study drugs: Treatment: AFIF tablets, 0.29 g/tablet, manufactured and supplied by Jiangxi Qingfeng Pharmaceutical Co., Ltd., production batch number: 20110605. Control: AFIF placebo tablets; the specification and production batch number were the same as those of the treatment tablets.

Dosage and administration: High-dose group: AFIF 4 tablets; medium-dose group: AFIF 3 tablets + AFIF placebo 1 tablet; low-dose group: AFIF 2 tablets + AFIF placebo 2 tablets; placebo group: AFIF placebo 4 tablets. Patients were instructed to take the tablets with warm water 30 minutes before meals, three times daily. The treatment course was 4 weeks.

Outcome measures

Primary outcome measures: The primary efficacy endpoints were the complete resolution rate of all four FD symptoms and the individual symptom resolution rate after 4 weeks of treatment. Symptoms included postprandial fullness, early satiety, epigastric burning sensation, and abdominal pain. Patients were not required to have all four symptoms at baseline. For patients presenting with fewer than 4 symptoms at enrollment, “complete resolution of all four symptoms” was defined as the resolution of all baseline symptoms. The first three symptoms were scored on a 0-3 severity scale: 0 for asymptomatic; 1 for mild symptoms without impact on daily life; 2 for noticeable symptoms with no obvious impact on daily life; and 3 for persistent symptoms significantly affecting daily life (Supplementary Table 2). For epigastric pain, a visual analog scale was used, with 0 representing “no pain” and 10 cm representing “the most severe pain”. Patients recorded their pain intensity by marking the corresponding position on the line based on their subjective experience.

Secondary outcome measures: The resolution rates of all four FD symptoms and each symptom were assessed at other visit time points. Gastric emptying function: Changes in the 2-hour gastric emptying rate and gastric half-emptying time were compared with baseline values (Supplementary material). Gastric emptying tests were performed using radionuclide imaging. Because of limitations in hospital facilities during the trial, gastric emptying function assessments were conducted only in participants from two centers: West China Hospital of Sichuan University and Xiangya Hospital of Central South University.

Safety outcome measures

Safety outcome measures included: (1) Vital signs: Body temperature, respiration, blood pressure after 10 minutes of rest, including systolic and diastolic blood pressure, and resting heart rate; (2) Routine blood, urine, and stool tests; (3) Liver and kidney function, including alanine aminotransferase, aspartate aminotransferase, total bilirubin, direct bilirubin, blood urea nitrogen, and creatinine; (4) Fasting blood glucose; (5) Resting 12-lead electrocardiogram, including measurement of PR, QRS, and QTc intervals; and (6) Adverse events (AEs), serious AEs, ARs, and serious ARs. For participants with abnormal laboratory findings after treatment, follow-up testing was performed until the values normalized or stabilized.

Visiting time-point

FD clinical symptoms were assessed before treatment, and participants were visited once weekly after treatment initiation to collect unused drugs from the previous distribution and dispense the next required supply. Gastric emptying tests and safety index examinations were performed only before and after treatment. AEs were recorded daily during the study period. For participants with complete resolution of all FD symptoms, an additional 4-week observation period was conducted, with another evaluation performed on day 56.

Sample size calculation

The primary efficacy endpoint was the resolution rate of all FD symptoms: Postprandial fullness, early satiety, epigastric pain, and epigastric burning sensation. A two-sided significance level of α = 0.05 and a power of 1-β = 0.8 were set. Based on the results of the completed phase IIa clinical trial, on day 28 of treatment, the resolution rates in the AFIF 12-tablet/day group were 26.14% for postprandial fullness, 58.52% for early satiety, 74.43% for epigastric pain, and 81.25% for epigastric burning sensation, whereas those in the placebo group were 8.99%, 23.60%, 52.81%, and 61.80%, respectively. The sample size required for the 12-tablet/day group to demonstrate superiority over the placebo group with a superiority margin of 0 was calculated for each symptom, and the maximum value across the four estimates was selected, yielding a minimum of 82 cases per group. Considering dropout and exclusion factors, 100 cases were planned per group, for a total of 400 enrolled.

Randomization and blinding

Eligible patients were randomly assigned to the treatment groups, comprising three AFIF dose groups or the placebo group. Randomization was performed by a statistician using SAS software version 9.1, stratified by center, with a block size of 4. The placebo tablets had the same appearance, including size, shape, and color, as the AFIF tablets. Allocation concealment was maintained by blinding patients, clinicians, investigators, and statisticians throughout the study. Each patient had an unblinded emergency envelope held by the principal investigator at each center.

Data management

Laboratory tests: Uniform standards and requirements for laboratory testing indicators were established across all participating hospitals. All serum and urine specimens were transported to the respective central laboratories within 24 hours of collection. For gastric emptying tests, a standardized operating procedure was developed and implemented.

Clinical trial monitoring: Regular monitoring visits were conducted by monitors appointed by Jiangxi Qingfeng Pharmaceutical Co., Ltd. to assess trial progress and completion. The monitors reviewed the completeness of case records and the accuracy of case report forms (CRFs), verified trial data, ensured compliance with the protocol and Good Clinical Practice, tracked the status of enrolled patients, and confirmed the proper storage, distribution, and accountability of investigational drugs.

Data recording and management: During the trial, patient data were collected anonymously using CRFs. After review by the monitors, the CRFs were submitted to the designated data management center. The center performed double data entry and consistency checks, after which the dataset was locked for final statistical analysis (Supplementary material).

Statistical analysis

Continuous variables were expressed as means, SDs, medians, and P₂₅ and P₇₅ values. Categorical variables were expressed as n (%). For the primary outcomes, analyses based on the Full Analysis Set (FAS), which followed the Intention-to-Treat principle, and Per Protocol analyses were performed, whereas only Intention-to-Treat analyses were performed for secondary outcomes. One-way analysis of variance, the Kruskal-Wallis H test, the χ² test, Fisher's exact test, and analysis of covariance were used to compare differences between groups when baseline variables were imbalanced. Missing data were imputed using the last-observation-carried-forward method. Statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, United States) and R 4.1.1 (https://cran.r-project.org/bin/windows/base/) statistical software. A two-sided P value < 0.05 was considered statistically significant.

Changes in study procedures or analysis plan

No protocol amendments were made during the study. During the self-inspection process, data from two centers were excluded because of clinical trial quality control issues: Xiangya Hospital of Central South University (60 cases) and the Affiliated Hospital of Jiangxi University of Chinese Medicine (72 cases). Sensitivity analyses were performed on the FAS with and without data from these two centers. For participants identified during self-inspection as concurrently enrolled in other trials but without other protocol violations, sensitivity analyses were conducted on the Per-Protocol Set (PPS) with and without their inclusion.

RESULTS

The recruitment timeline

A total of 399 participants were enrolled across seven national subcenters: West China Hospital of Sichuan University, Affiliated Hospital of Chengdu University of Traditional Chinese Medicine, The First Affiliated Hospital of Hunan University of Chinese Medicine, Xiangya Hospital of Central South University, Affiliated Hospital of Jiangxi University of Chinese Medicine, The First Affiliated Hospital of Tianjin University of Traditional Chinese Medicine, and The Second Affiliated Hospital of Tianjin University of Traditional Chinese Medicine. Of these, 100 participants were assigned to the high-dose group, 99 to the medium-dose group, 100 to the low-dose group, and 100 to the placebo group.

Efficacy and safety data from two centers, comprising 132 cases in total, were excluded, resulting in 267 cases in both the Safety Analysis Set and the FAS: 67 in the high-dose group, 66 in the medium-dose group, 67 in the low-dose group, and 67 in the placebo group (Supplementary Table 3). Among these 267 participants, five withdrew early and had missing primary efficacy endpoints, 83 had major protocol violations (59 did not comply with the protocol, seven received concomitant medications affecting efficacy evaluation, and 17 participated in other trials), two had out-of-window time points, and one completed the trial but had missing primary efficacy endpoints. Consequently, 176 participants were included in the PPS: 46 in the high-dose group, 42 in the medium-dose group, 40 in the low-dose group, and 48 in the placebo group. The participant distribution flow diagram is shown in Figure 1.

Figure 1 Flow chart of patient selection. AFIF: Aurantii Fructus Immaturus Flavonoid; FAS: Full Analysis Set; SS: Safety Analysis Set; PPS: Per-Protocol Set.

Participant demographics and baseline characteristics

Participants’ age, gender, height, weight, case source, disease duration, history of Western medicine treatment, history of TCM treatment, smoking history, physical examination findings, TCM syndrome patterns, FD clinical symptoms, gastric emptying rate, and half-emptying time were comparable among groups (P > 0.05; Table 1).

Table 1 Demographics and baseline characteristics, mean ± SD/n (%).

Variables		High-dose group, n = 67	Median-dose group, n = 66	Low-dose group, n = 67	Placebo group, n = 67	Statistics	P value
Age (years)		40.51 ± 12.98	40.86 ± 13.95	41.56 ± 13.80	39.24 ± 12.87	F = 0.35	0.788
Gender	Male	21 (31.34)	22 (33.33)	29 (43.28)	30 (44.78)	χ2 = 3.962	0.266
	Female	46 (68.66)	44 (66.67)	38 (56.72)	37 (55.22)
Height (cm)		162.84 ± 7.68	163.03 ± 6.75	164.16 ± 8.54	163.76 ± 8.59	F = 0.414	0.745
Weight (kg)		59.25 ± 8.05	61.31 ± 8.07	59.43 ± 8.33	60.63 ± 9.25	F = 0.903	0.441
Source	Outpatient	67 (100.00)	66 (100.00)	67 (100.00)	67 (100.00)
Duration (months)	N (N miss)	45 (22)	51 (15)	51 (16)	49 (18)	H = 3.000	0.392
	Median (P25, P75)	28.10 (14.90, 48.10)	24.60 (12.90, 42.60)	25.40 (11.80, 55.30)	22.70 (10.90, 32.10)
Western medicine treatment history	No	48 (71.64)	51 (77.27)	53 (79.10)	53 (79.10)	Fisher’s exact test	0.727
	Yes	19 (28.36)	15 (22.73)	14 (20.90)	14 (20.90)
TCM treatment history	No	48 (71.64)	48 (72.73)	52 (77.61)	52 (77.61)	Fisher’s exact test	0.801
	Yes	19 (28.36)	18 (27.27)	15 (22.39)	15 (22.39)
Comorbidities	No	56 (83.58)	51 (77.27)	64 (95.52)	60 (89.55)	Fisher’s exact test	0.012
	Yes	11 (16.42)	15 (22.73)	3 (4.48)	7 (10.45)
Smoking history	No	60 (89.55)	59 (89.39)	64 (95.52)	58 (86.57)	Fisher’s exact test	0.321
	Yes	7 (10.45)	7 (10.61)	3 (4.48)	9 (13.43)
Physical examination	Normal	67 (100.00)	66 (100.00)	67 (100.00)	66 (98.51)	Fisher’s exact test	> 0.999
	Abnormal	0 (0.00)	0 (0.00)	0 (0.00)	1 (1.49)
TCM syndrome patterns	Liver-stomach disharmony	28 (41.79)	21 (31.82)	17 (25.37)	20 (29.85)	H = 11.431	0.247
	Spleen-stomach damp-heat	16 (23.88)	20 (30.30)	28 (41.79)	24 (35.82)
	Food stagnation	13 (19.40)	9 (13.64)	9 (13.43)	6 (8.96)
	Spleen-stomach weakness	10 (14.93)	16 (24.24)	13 (19.40)	17 (25.37)
Postprandial fullness	0 points	0 (0.00)	0 (0.00)	1 (1.49)	0 (0.00)	H = 2.198	0.531
	1 point	11 (16.42)	10 (15.15)	13 (19.40)	12 (17.91)
	2 points	38 (56.72)	38 (57.58)	39 (58.21)	43 (64.18)
	3 points	18 (26.87)	18 (27.27)	14 (20.90)	12 (17.91)
Early satiety	0 points	1 (1.49)	0 (0.00)	2 (2.99)	0 (0.00)	H = 2.672	0.446
	1 point	20 (29.85)	24 (36.36)	26 (38.81)	22 (32.84)
	2 points	35 (52.24)	35 (53.03)	33 (49.25)	39 (58.21)
	3 points	11 (16.42)	7 (10.61)	6 (8.96)	6 (8.96)
Upper abdominal burning sensation	0 points	7 (10.45)	6 (9.09)	8 (11.94)	4 (5.97)	H = 0.780	0.854
	1 point	28 (41.79)	28 (42.42)	25 (37.31)	30 (44.78)
	2 points	27 (40.30)	28 (42.42)	30 (44.78)	23 (34.33)
	3 points	5 (7.46)	4 (6.06)	4 (5.97)	10 (14.93)
Upper abdominal pain visual analog scale	Median (P25, P75)	5.00 (4.00, 6.00)	5.00 (4.00, 5.00)	4.00 (3.00, 5.20)	4.00 (3.00, 6.00)	H = 0.947	0.423
Gastric emptying rate (%)	N (N miss)	17 (50)	17 (49)	17 (50)	17 (50)	F = 0.962	0.419
		31.46 ± 9.14	29.43 ± 10.72	31.94 ± 7.30	34.92 ± 10.69
Gastric half-emptying time (hours)	N (N miss)	17 (50)	17 (49)	17 (50)	17 (50)	F = 1.707	0.175
		3.64 ± 1.07	4.07 ± 2.04	3.38 ± 0.95	3.09 ± 0.85

TCM: Traditional Chinese medicine.

Assessment of treatment compliance

Among the 267 subjects in the FAS, compliance was below 80% in 6 cases: 1 in the high-dose group, 1 in the medium-dose group, none in the low-dose group, and 4 in the placebo group. No cases had compliance above 120%. The mean compliance values (SD) in the high-dose, medium-dose, low-dose, and placebo groups were 99.16% (9.76%), 99.37% (7.18%), 100.54% (4.25%), and 98.31% (9.18%), respectively. There was no statistically significant difference in mean compliance among the four groups (F = 1.11, P = 0.347).

Effects of AFIF on the resolution rate of the four symptoms of FD after treatment

In the FAS, the between-group difference in the resolution rate of all four FD symptoms after 4 weeks of medication was statistically significant (χ² = 19.79, P < 0.0001). Pairwise comparisons showed that both the high-dose and medium-dose groups differed significantly from the low-dose and placebo groups. No significant differences were observed between the low-dose and placebo groups or between the high-dose and medium-dose groups. In the PPS, the between-group difference in the resolution rate of all symptoms was also statistically significant (χ² = 15.92, P = 0.001). The conclusions from the four-group comparisons and pairwise comparisons in the PPS were consistent with those in the FAS, indicating that the resolution rates of all four symptoms in the medium-dose and high-dose groups on day 28 were higher than those in the low-dose and placebo groups (Figure 2). At 4 weeks after drug withdrawal, the resolution rates of all symptoms in each group remained above 75%; however, no statistically significant between-group differences were observed because of the small follow-up sample size.

Figure 2 Resolution rate of all four symptoms of functional dyspepsia after 4 weeks. A: Full Analysis Set; B: Per-Protocol Set. FAS: Full Analysis Set; PPS: Per-Protocol Set.

After 4 weeks of medication, the between-group differences in the resolution rates of postprandial fullness, early satiety, epigastric burning sensation, and epigastric pain were all statistically significant (χ² = 14.061, P = 0.003; χ² = 18.703, P < 0.001; χ² = 17.205, P < 0.001; χ² = 17.552, P < 0.001, respectively). For individual symptom resolution rates, both the high-dose and medium-dose groups showed superior effects compared with the placebo and low-dose groups. No significant differences were observed between the low-dose and placebo groups or between the high-dose and medium-dose groups (Figure 3). At 4 weeks after drug withdrawal, the resolution rates of the four individual symptoms in each group remained above 75%; however, no statistically significant differences were observed because of the small follow-up sample size.

Figure 3 Resolution rate of a single symptom of functional dyspepsia after 4 weeks. A: Postprandial fullness; B: Early satiety; C: Upper abdominal burning sensation; D: Upper abdominal pain.

Effects of AFIF on the resolution rate of the symptoms of FD during the treatment

In the FAS, the resolution rates of all four symptoms in the four groups gradually increased from weeks 1 to 4 of medication. There were no statistically significant differences among the four groups at weeks 1 and 2, whereas significant differences were observed at weeks 3 and 4. Consistent conclusions were drawn in the PPS (Figure 4). The resolution rates of postprandial fullness, early satiety, epigastric burning sensation, and epigastric pain in the four groups also increased gradually. Statistically significant between-group differences were observed at weeks 3 and 4 of medication (Figure 5).

Figure 4 Resolution rate of all four symptoms of functional dyspepsia during treatment. A: Full Analysis Set; B: Per-Protocol Set. FAS: Full Analysis Set; PPS: Per-Protocol Set.

Figure 5 Resolution rate of a single symptom of functional dyspepsia during treatment. A: Postprandial fullness; B: Early satiety; C: Upper abdominal burning sensation; D: Upper abdominal pain.

Improvement of the postprandial 2-hour gastric emptying rate and the half-emptying time

The 2-hour postprandial gastric emptying rate after 4 weeks of medication increased in the AFIF groups and decreased in the placebo group; however, there was no significant between-group difference in the change in gastric emptying rate from before to after treatment. Gastric half-emptying time was shortened in the AFIF groups and increased in the placebo group; however, there was no significant between-group difference in the change in gastric half-emptying time from before to after treatment. Details are shown in Table 2.

Table 2 Changes of postprandial 2-hour gastric emptying rate (%) and gastric half-emptying time (hours) before and after treatment, mean ± SD.

Outcomes		High-dose group	Median-dose group	Low-dose group	Placebo group	F value	P value
Gastric emptying rate	N (N miss)	17 (50)	15 (51)	17 (50)	15 (52)	1.10	0.355
		5.78 ± 9.83	5.01 ± 10.07	4.24 ± 10.99	-0.35 ± 10.13
	MD (95%CI) vs placebo	6.13 (-3.49, 15.75)	5.35 (-4.56, 15.27)	4.58 (-5.04, 14.20)
Gastric half-emptying time	N (N miss)	17 (50)	15 (51)	17 (50)	15 (52)	1.02	0.389
		-0.75 ± 1.05	-0.75 ± 2.40	-0.23 ± 1.21	0.01 ± 0.96
	MD (95%CI) vs placebo	-0.77 (-2.17, 0.64)	-0.76 (-2.21, -0.69)	-0.24 (-1.65, 1.16)

CI: Confidence interval; MD: Mean difference.

Subgroup analysis of four TCM syndrome types

After 4 weeks of medication, there was no significant difference in the resolution rates of all four FD symptoms between the Ganwei Buhe (liver-stomach disharmony) and Piwei Xuruo (spleen-stomach weakness) subgroups. However, significant between-subgroup differences were observed in the Piwei Shire (spleen-stomach damp-heat) and Yinshi Tingzhi (food stagnation) subgroups (Figure 6).

Figure 6 Subgroup analysis of four traditional Chinese medicine syndrome types. CI: Confidence interval.

Sensitivity analysis

Analysis without excluding center 04 and center 05 from the FAS: The resolution rates of all four symptoms in the high-dose, medium-dose, low-dose, and placebo groups were 27.00%, 29.29%, 2.00%, and 5.00%, respectively, after 4 weeks of medication. The between-group comparison showed statistically significant differences (P < 0.0001), consistent with the conclusions obtained when these centers were excluded. For individual symptoms, all four symptoms showed statistically significant between-group differences at week 4 (all P < 0.01), with conclusions consistent with those from the exclusion analysis.

Analysis without excluding participants enrolled in other trials within 3 months prior to enrollment or during the trial from the PPS: The resolution rates of all four symptoms in the high-dose, medium-dose, low-dose, and placebo groups were 22.45%, 28.57%, 0.00%, and 8.00%, respectively, after 4 weeks of medication. The between-group comparison showed statistically significant differences (P < 0.0001), consistent with the conclusions obtained when these participants were excluded. For individual symptoms, all four symptoms showed statistically significant between-group differences at week 4 (all P < 0.001), with conclusions consistent with those from the exclusion analysis.

Safety analysis

In the Safety Analysis Set, 267 participants experienced 13 AEs across 21 event instances. Specifically, the high-dose group had 4 participants with 5 AEs (5.97%), the medium-dose group had 4 participants with 10 AEs (6.06%), the low-dose group had no AEs (0.00%), and the placebo group had 5 participants with 6 AEs (7.46%). There was no significant difference in AE incidence among the four groups (P = 0.1122). One severe AE led to trial discontinuation. A participant in the placebo group experienced reflux, with specific symptoms of stomach pain accompanied by acid regurgitation and belching, with an incidence of 1.49%. No serious AEs or adverse reactions were reported in the AFIF groups (Supplementary Table 4).

In the high-dose group, AEs with an incidence ≥ 1% included urinary red blood cell positivity, decreased blood glucose, chronic bronchitis, acne, and fatigue, each occurring in 1.49% of participants (1/67). In the medium-dose group, AEs with an incidence ≥ 1% included cough (4.55%, 3/66), oropharyngeal pain (3.03%, 2/66), fever, pain, upper respiratory tract infection, goiter, and lymphadenitis, each occurring in 1.52% of participants (1/66). In the placebo group, AEs with an incidence ≥ 1% included decreased platelet count, cough, rhinitis, circumcision, reflux, and upper abdominal pain, each occurring in 1.49% of participants (1/67).

DISCUSSION

Main findings

This study demonstrated that both the medium-dose (3 tablets/time, tid) and high-dose (4 tablets/time, tid) groups of AFIF significantly improved clinical symptoms of FD after 4 weeks of treatment. No significant difference was observed between the low-dose group (2 tablets/time, tid) and the placebo group. In terms of safety, no adverse reactions were observed in any AFIF dose group (4, 3, or 2 tablets/time, tid). Neither the incidence of AEs/adverse reactions nor laboratory parameters showed clinically or statistically significant differences compared with the placebo group. Therefore, the recommended dose for phase III clinical trials is 3 tablets/time, tid.

Possible mechanisms of action of AFIF

Citrus aurantium (Zhishi) is an authentic medicinal herb from Jiangxi Province, characterized by a bitter and pungent taste and a slightly cold nature. The latest edition of the Chinese Pharmacopoeia identifies it as the dried young fruit of the Rutaceae plant Citrus aurantium L.[14]. First documented in Shennong Ben Cao Jing, Zhishi was noted in Ming Yi Bie Lu for relieving flatulence and stabilizing gastric qi. It acts on the spleen, stomach, and large intestine meridians and is traditionally used to break qi, eliminate stagnation, resolve phlegm, and relieve abdominal fullness. With nearly 2000 years of clinical use in TCM formulas, Zhishi is a key ingredient in classic prescriptions, and the modern Chinese Pharmacopoeia lists 44 patent formulations containing Zhishi, such as Zhizhu pill[15].

Animal studies have demonstrated the efficacy of Zhishi in ameliorating gastrointestinal dyskinesia[17-19]. Its therapeutic mechanism involves inhibition of the PINK1/Parkin signaling pathway, thereby modulating mitochondrial autophagy, enhancing gastrointestinal peristalsis, and increasing gastrointestinal hormone levels in FD rats[19]. Zhishi also regulates gastric Cajal mesenchymal stromal cells by activating the SCF/c-Kit and TLR4/MyD88/NF-κB signaling pathways, thereby improving gastrointestinal dysfunction, duodenal inflammation, and barrier damage[20].

Zhishi contains high levels of flavonoids, which are among its active ingredients[21]. AFIF, developed based on TCM theory and clinical experience, is formulated from total flavonoid glycosides extracted from Zhishi. Its therapeutic effects on FD may involve the following multitarget mechanisms: (1) Improving gastric emptying and compliance. FD patients exhibit gastric dysfunction, including slowed peristalsis and decreased compliance[22-24]. AFIF enhances gastric antral contractility and intestinal peristalsis by activating the cholinergic pathway or inhibiting dopamine D2 receptors, thereby improving delayed gastric emptying[25]. Animal studies have shown that AFIF significantly increases duodenal smooth muscle tension and contraction frequency in rats, with effects superior to those of domperidone[26]. Experimental models indicate that AFIF markedly increases duodenal smooth muscle tension and contraction frequency while enhancing gastric accommodation in FD rats[27]. Additionally, AFIF promotes motilin secretion while inhibiting vasoactive intestinal peptide secretion, a peptide neurotransmitter that inhibits gastrointestinal motility, thereby enhancing gastric emptying and promoting small intestinal peristalsis in FD rats[28]; (2) Reducing visceral hypersensitivity. AFIF may modulate 5-hydroxytryptamine 3/4 receptors or inhibit transient receptor potential vanilloid 1 channels to reduce visceral pain signal transmission, thereby alleviating epigastric pain and discomfort in FD patients[29]; (3) Regulating the brain-gut axis. Gastric dysfunction in FD may reflect neuromuscular dysfunction[24]. Flavonoid components, such as hesperidin and naringin, exhibit anxiolytic and mild antidepressant effects, potentially improving FD-associated psychological symptoms via the hypothalamic-pituitary-adrenal axis[27]. AFIF also upregulates miR-5100 expression and inhibits Fzd2, attenuating calcium overload and autophagic death in intraclass correlation coefficient, modulating neurotransmitter levels, and enhancing enteric nervous system integrity[30]; and (4) Anti-inflammatory and mucosal-protective effects. Gastric mucosal nerve density is reduced in FD[28], possibly because of mucosal inflammation. AFIF reduces low-grade gastric mucosal inflammation and repairs the gastrointestinal barrier through its antioxidant properties[25,27]. Alterations in duodenal microbiota are associated with gastric emptying and FD symptoms[31]. AFIF also decreases interleukin-6, interleukin-1β, and tumor necrosis factor-α levels, thereby ameliorating intestinal inflammation and modulating microbial diversity and abundance, ultimately improving FD symptoms[32].

Possible explanations of the non-monotonic dose-response relationship

Because the present study observed a non-monotonic dose-response pattern, possible explanations are discussed as follows. Three aspects may explain this pattern, in which the medium-dose group showed a numerically higher resolution rate than the high-dose group. (1) Biological considerations. Non-monotonic curves, including plateau and U-shaped patterns, have been documented for various therapeutic agents, often because of receptor saturation or compensatory feedback mechanisms. As reviewed by Calabrese[33], the biphasic dose-response model, or hormesis, in which low to moderate doses may stimulate or optimize biological responses while higher doses become inhibitory, has been observed across a wide range of biological models and agents. From a TCM perspective, a higher dose of AFIF might exert excessive “purging” or “drying” effects, which could be less suitable for patients with spleen-stomach disharmony and may contribute to the plateaued efficacy at higher doses; (2) Methodological considerations. The numerically higher response rate in the medium-dose group than in the high-dose group may be partially attributable to random variation, as small sample sizes increase the vulnerability of effect estimates to chance[34]. Unmeasured differences in baseline patient characteristics, such as symptom severity or TCM pattern differentiation, across dose groups may also have influenced the observed pattern. Additionally, the selected dose range may not have fully captured the linear portion of the dose-response curve; the high dose may have already reached the plateau phase, where further dose escalation yields no additional benefit; and (3) Post hoc nature and cautious interpretation. Importantly, this observation was post hoc, as it was not a prespecified comparison. As emphasized by Burke et al[35], post hoc subgroup observations should be explicitly labeled as hypothesis-generating and interpreted with appropriate caution. Therefore, we explicitly acknowledge that the true optimal biological dose remains somewhat uncertain. Our recommendation of the medium dose for future phase III trials is based on an integrated assessment of efficacy trends, safety, and feasibility, and warrants prospective validation.

Study limitations

Sample size constraints: Although the study enrolled participants from multiple centers, the final sample sizes of the FAS and PPS were reduced after data from two centers, Xiangya Hospital and the Affiliated Hospital of Jiangxi University of Chinese Medicine, were excluded. Data collection had been completed at these two centers; however, during the subsequent rigorous data cleaning and quality control process, we identified numerous protocol deviations at these sites. Collectively, these deviations indicated systemic problems in trial conduct, including frequent violations of the inclusion/exclusion criteria, protocol noncompliance, and breaches of visit windows. Given the extent and nature of these deviations, which compromised the integrity and reliability of the data from these sites, we excluded all data from these two centers from the final analysis to protect the overall validity of the study. Consequently, the reduced sample size may have affected the statistical power of the results. The small sample size in the 4-week post-withdrawal follow-up also limited the generalizability of conclusions regarding long-term efficacy. The actual sample sizes for the high-dose, medium-dose, low-dose, and placebo groups were 67, 66, 67, and 67, respectively, with response rates of 22.39%, 27.27%, 2.99%, and 5.97%. With a two-sided α of 0.05, the power of this study was calculated to be 96.2% based on the above results. This indicates that the study had sufficient power to detect the overall difference in response rates among the four groups; therefore, the reduction in sample size had little impact on the analysis results.

Limitations of the diagnostic criteria: This study used the Rome III criteria for diagnosing FD, which were the standard at study initiation in 2011. The Rome IV criteria, published in 2016, were not yet available. Unlike Rome III, Rome IV requires more frequent symptoms (m³ days/week rather than ≥ 1 day/week) and emphasizes gut-brain interactions. Consequently, applying Rome IV might have excluded patients with milder symptoms, potentially resulting in a study population with higher baseline symptom severity and affecting the observed effect sizes. Our findings should therefore be interpreted in the context of Rome III-defined FD.

Limitations in gastric emptying function testing: Due to equipment constraints, gastric emptying function tests were conducted at only two centers, West China Hospital of Sichuan University and Xiangya Hospital of Central South University. The insufficient sample size may have limited the ability to detect statistically significant between-group differences, potentially masking the drug’s effects on gastric motility.

Limitations in subgroup analysis by TCM syndrome types: Although subgroup analyses by TCM syndrome type were performed, some subgroups had small sample sizes, limiting cautious interpretation of the dose-response efficacy results across syndrome types. The efficacy of AFIF across common TCM syndrome types can be inferred from the IIa clinical study, which demonstrated favorable efficacy in three common syndrome types.

Short duration: The study observed the treatment effect over only 4 weeks and the short-term maintenance of effects for 4 weeks after drug withdrawal. It lacked safety and efficacy data for long-term use, such as 6 months or longer, which prevented evaluation of the drug’s long-term effects.

Limitation in AE assessment: Notably, the low-dose AFIF group had an AE rate of 0.00%, compared with 5.97%-7.46% in the other groups. Although this difference did not reach statistical significance, likely due to random variation associated with the modest sample size, this observation should be interpreted with caution. Future large-scale phase III trials with adequate power are needed to provide a more precise estimate of the AE profile of low-dose AFIF.

Limitation regarding protocol violations: Many participants were excluded from the PPS because they were concurrently enrolled in other trials. It should be clarified that these concurrent trials were unrelated to the treatment of FD and did not involve medications that could have affected the efficacy evaluation of our study intervention. Therefore, we believe these protocol violations had minimal impact on data integrity. Nevertheless, the exclusion of these participants constitutes a methodological limitation, and the generalizability of our findings to patients with similar protocol nonadherence remains uncertain. Future studies should implement stricter control over enrollment criteria to minimize such violations.

Implications for clinical practice

This study demonstrated that both the medium-dose regimen of AFIF (3 tablets per dose three times daily) and the high-dose regimen (4 tablets per dose three times daily) significantly improved FD symptoms and had good safety profiles. Considering both efficacy and safety trends, the medium-dose regimen is recommended as the preferred clinical option. Compared with PPIs and prokinetic agents, such as domperidone, AFIF was not associated with significant adverse reactions, such as cardiotoxicity or risks related to long-term PPI use, thereby providing a safer treatment option for FD patients, especially those requiring long-term management.

Suggestions for future research

Larger-scale phase IV clinical trials or registration studies should be conducted, with the follow-up period extended to 6 months to 1 year, to clarify the long-term efficacy and safety of AFIF. Further in-depth mechanistic research is also required to explore the specific molecular mechanisms by which AFIF regulates gastrointestinal motility, visceral sensitivity, and the brain-gut axis. Future studies should investigate how the gut microbiota interacts with these bioactive compounds to further elucidate the mechanism of action of AFIF in treating FD. For example, animal models or cell experiments could be used to verify its effects on 5-hydroxytryptamine receptors, dopamine pathways, and other relevant targets. By combining modern technologies, such as electrogastrography and high-resolution manometry, a multimodal approach should be adopted to comprehensively evaluate the effects of AFIF on gastrointestinal function.

CONCLUSION

Medium- and high-dose AFIF significantly improved FD symptoms with good safety. The medium dose (3 tablets/tid) is recommended for phase III trials and may be a safer alternative to PPIs or prokinetics.