Effect of Butyrate‐Producing Enterobacteria and Proton Pump Inhibitors on Advanced Hepatocellular Carcinoma Treatment With Durvalumab and Tremelimumab

ABSTRACT

Aim

The gut microbiome modulates immune responses, and butyrate‐producing bacteria have been linked to improved immune checkpoint inhibitor (ICI) efficacy. Conversely, proton pump inhibitors (PPIs) may negatively impact ICI outcomes by altering gut microbiota. This study aims to elucidate their effects in hepatocellular carcinoma (HCC).

Methods

This retrospective multicenter cohort study included 208 HCC patients treated with durvalumab plus tremelimumab at 25 hospitals in Japan. Patients were classified into a butyric acid group (n = 27), who ingested drugs containing butyrate‐producing enterobacteria, and a non‐butyric acid group (n = 181), as well as a PPI group (n = 107) and a non‐PPI group (n = 101). Overall survival (OS) was analyzed using inverse probability of treatment weighting, and risk factors were assessed with Cox proportional hazards modeling. Tumor response was evaluated by RECIST v1.1.

Results

No significant OS differences were observed between the butyric acid and non‐butyric acid groups (p = 0.921), or between PPI and non‐PPI groups (p = 0.917). The objective response rate was 3.7% in the butyric acid group versus 15.5% in the non‐butyric acid group (p = 0.543) and 15.8% in the PPI group versus 12.1% in the non‐PPI group (p = 0.222). Disease control rates were comparable. Multivariate analysis identified ECOG performance status (p = 0.019) and ALBI score (p < 0.001) as independent prognostic factors, while butyrate‐producing bacteria and PPI use were not associated with survival outcomes.

Conclusion

Neither butyrate‐producing bacteria nor PPI use significantly influenced the efficacy of durvalumab plus tremelimumab in HCC. The liver's immunotolerant microenvironment may limit the impact of microbiome modulation on ICI efficacy.

Abbreviations

AE

adverse event

ALBI

albumin‐bilirubin

BCLC

barcelona clinic liver cancer

CI

confidence interval

DCR

disease control rate

ECOG

eastern cooperative oncology group

HCC

hepatocellular carcinoma

HR

hazard ratio

ICI

immune checkpoint inhibitor

IPTW

inverse probability of treatment weighting

ORR

objective response rate

OS

overall survival

PPI

proton pump inhibitor

PS

performance status

RCC

renal cell carcinoma

RECIST

response evaluation criteria in solid tumors

UDCA

ursodeoxycholic acid

VEGF

vascular endothelial growth factor

1. Introduction

The gut microbiome plays a crucial role in modulating the host immune response, particularly in the context of cancer immunotherapy [1]. Butyrate‐producing enterobacteria are bifidogenic and promote dendritic cell function and T cell recruitment, potentially enhancing anti‐tumor immunity [2]. Recent studies have highlighted the potential of butyrate‐producing bacteria in enhancing the efficacy of immune checkpoint inhibitors (ICIs) [3, 4, 5]. Specifically, in renal cell carcinoma (RCC), the administration of butyrate‐producing bacteria has been shown to improve treatment outcomes by modulating the tumor immune microenvironment [4].

However, our previous research demonstrated that the addition of butyrate‐producing bacteria did not enhance the efficacy of atezolizumab plus bevacizumab therapy in patients with advanced hepatocellular carcinoma (HCC) [6]. A potential explanation for this observation is that bevacizumab, a vascular endothelial growth factor (VEGF) inhibitor, may counteract the beneficial effects of butyrate‐producing bacteria on the immune system. VEGF signaling is known to influence the tumor microenvironment by suppressing antigen presentation and promoting an immunosuppressive milieu, which could interfere with microbiome‐mediated immunomodulation [7, 8].

Given this, we hypothesized that butyrate‐producing bacteria might exert a positive effect in an ICI regimen that does not include a VEGF inhibitor. Durvalumab plus tremelimumab, a combination of PD‐L1 and CTLA‐4 blockade, or durvalumab monotherapy has been approved for the treatment of advanced HCC in Japan and offers a distinct immunological mechanism compared to atezolizumab plus bevacizumab [9]. Unlike VEGF‐targeted therapies, this regimen relies solely on immune checkpoint blockade to enhance anti‐tumor immunity.

In addition to the gut microbiome, recent evidence suggests that proton pump inhibitors (PPIs) may negatively impact the efficacy of ICIs. PPIs alter the gut microbiome composition by reducing microbial diversity and increasing the prevalence of potentially immunosuppressive bacteria [10]. Several studies have reported that PPI use is associated with poorer outcomes in cancer patients receiving ICIs, potentially due to microbiome dysbiosis and impaired immune activation [11]. However, in our previous study, we found that PPI use did not influence the efficacy of atezolizumab plus bevacizumab therapy in patients with advanced HCC [12]. This suggests that, similar to butyrate‐producing bacteria, the effect of PPIs on immunotherapy may be influenced by the presence of a VEGF inhibitor.

In this study, we aim to investigate two key microbiome‐related factors influencing ICI efficacy: the administration of butyrate‐producing bacteria and the use of PPIs. By evaluating clinical and immunological outcomes, we seek to clarify the role of microbiome‐based interventions in immune checkpoint blockade and determine whether the absence of VEGF inhibition allows for a microbiome‐driven enhancement of anti‐tumor responses. Furthermore, we aim to assess whether PPI use negatively impacts treatment efficacy in this setting, given the conflicting evidence in different therapeutic contexts.

2. Materials and Methods

2.1. Study Design and Patient Selection

This study was a retrospective cohort study conducted at 25 different hospitals in Japan between January 2023 and July 2024 (Okayama City Hospital, Ehime Prefectural Central Hospital, Nippon Medical School, Kagawa University, Japanese Red Cross Society Himeji Hospital, Toyama University Hospital, Kansai Medical University, Ehime University Graduate School of Medicine, Asahi General Hospital, National Hospital Organization Takasaki General Medical Center, Saiseikai Niigata Hospital, Niigata Cancer Center Hospital, Kindai University, Takarazuka City Hospital, Gunma Saiseikai Maebashi Hospital, Teine Keijinkai Hospital, Hamamatsu University School of Medicine, Kagawa Prefectural Central Hospital, Japanese Red Cross Takamatsu Hospital, Matsuyama Red Cross Hospital, Hanwa Memorial Hospital, Iwate Medical University, Hyogo Prefectural Harima‐Himeji General Medical Center, Ogaki Municipal Hospital, Hyogo Medical University). Patients with hepatocellular carcinoma (HCC) who received durvalumab plus tremelimumab or durvalumab monotherapy were enrolled (n = 208). The eligibility criteria required histologically or radiologically confirmed HCC, tumors unsuitable to curative or locoregional treatment or that had progressed thereafter, an ECOG performance status of 0 or 1, and Child‐Pugh class A or B liver function. Patients with other malignancies were excluded from the analysis.

2.2. Treatment Protocol

Of the 208 enrolled patients with HCC, 189 received durvalumab (1500 mg) plus tremelimumab (300 mg) intravenously every 4 weeks in accordance with standard clinical practice, whereas the remaining 19 patients received durvalumab monotherapy. Treatment was discontinued if the tumor progressed or if unacceptable adverse effects (AEs) were determined by each physician based on the patient's age and status. Moreover, AEs were managed by interrupting treatment or reducing the dose according to the instructions provided by the supplier.

2.3. Butyrate‐Producing Bacteria and PPI Use

To evaluate the impact of butyrate‐producing bacteria on treatment efficacy, patients who ingested drugs containing butyric acid‐producing enterobacteria (MIYA‐BM, MIYARISAN Pharmaceutical CO. LTD. or BIO‐THREE, TOA Pharmaceuticals CO. LTD.) for at least 1 month before and during immunotherapy were classified as the butyric acid group. Similarly, the influence of proton pump inhibitors (PPIs) was assessed by categorizing patients based on PPI usage. The PPI group was defined as patients who had been receiving PPIs for at least 1 month before and during immunotherapy.

In most cases, MIYA‐BM was initiated for the treatment of irritable bowel syndrome or for gastrointestinal adverse events related to previous therapies such as lenvatinib, whereas PPIs were prescribed for common gastrointestinal disorders such as peptic ulcer or gastroesophageal reflux disease, at the discretion of the treating physician based on the patient's clinical condition.

2.4. Efficacy Assessments

Treatment efficacy was evaluated based on response rates and survival outcomes. Tumor response was assessed via computed tomography or magnetic resonance imaging every 2–3 months, according to the Response Evaluation Criteria in Solid Tumors version 1.1 [13]. Overall survival (OS) was calculated from the initiation of durvalumab treatment to the date of the last follow‐up or death from any cause.

2.5. Statistical Analysis

Baseline characteristics between groups were compared using the chi‐square test or Fisher's exact test for categorical variables and the Student's t‐test or Mann–Whitney U test for continuous variables. Kaplan–Meier survival curves were generated and compared using the log‐rank test. To adjust for potential confounding factors, inverse probability of treatment weighting (IPTW) was applied, considering Barcelona Clinic Liver Cancer (BCLC) stage, Albumin‐Bilirubin (ALBI) score, performance status (PS), alpha‐fetoprotein (AFP), and whether durvalumab plus tremelimumab was used as first‐line therapy. Covariate balance was assessed using standardized mean differences (Table S1). Hazard ratios (HRs) were estimated using a Cox proportional hazards model. Statistical analyses were performed using JMP software (version 17.0; JMP Statistical Discovery LLC, Tokyo, Japan), except for IPTW, which was conducted using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria).

2.6. Ethical Considerations

The protocol of this multicenter study complied with the Helsinki Declaration and was approved by each participating institution (Central institutional review board: NHO Takasaki General Medical Center Clinical Research Ethics Committee approved No. TGMC2024‐003). Written informed consent was waived due to the retrospective nature of the study.

3. Results

3.1. Patient Characteristics

Of the 208 patients with HCC enrolled in this study, 27 patients (13.0%) were classified as the butyric acid group, while 181 patients (87.0%) were in the non‐butyric acid group. Similarly, 107 patients (51.4%) were classified as the PPI group, and 101 patients (48.6%) were in the non‐PPI group.

The baseline characteristics of patients in each group are summarized in Table 1. The median age was comparable between the butyric acid group and the non‐butyric acid group (73 vs. 73 years, p = 0.349) and between the PPI group and the non‐PPI group (73 vs. 74 years, p = 0.435). The proportion of male patients was similar in both comparisons. There were no significant differences in ECOG performance status, Child‐Pugh classification, ALBI score, BCLC stage, AFP, C‐reactive protein, neutrophil lymphocyte ratio, or first‐line systemic therapy use across groups.

TABLE 1.

Characteristics of the patients.

Variable	Butyric acid group (n = 27)	Non‐butyric acid group (n = 181)	p	PPI group (n = 107)	Non‐PPI group (n = 101)	p
Age, years	74 (43–80)	74 (39–93)	0.127	74 (39–90)	75 (43–93)	0.497
Male sex	24 (88.9%)	151 (83.4%)	0.468	86 (85.1%)	89 (83.2%)	0.697
Cause of hepatocellular carcinoma			0.322			0.639
Hepatitis B	6 (22.2%)	29 (16.0%)		15 (14.9%)	20 (18.7%)
Hepatitis C	11 (40.7%)	55 (30.4%)		32 (31.7%)	34 (31.8%)
Hepatitis B + C	1 (3.7%)	3 (1.7%)		1 (1.0%)	3 (2.8%)
Nonviral	9 (33.3%)	94 (51.9%)		53 (52.5%)	50 (46.7%)
ECOG performance status, 0/1	22/5	147/34	0.973	81/20	88/19	0.705
Use in first‐line systemic therapy	5 (18.5%)	67 (37.0%)	0.059	30 (29.7%)	42 (39.3%)	0.147
BCLC stage, A/B/C	1/11/15	13/57/111	0.558	6/33/62	8/35/64	0.936
Intrahepatic tumor size, mm	37 (0–167)	36 (0–190)	0.757	36 (0–190)	38 (0–168)	0.959
Intrahepatic tumor number (> 5nodules)	13 (52%)	51 (34.0%)	0.083	36 (39.6%)	28 (0%–33.3%)	0.392
Portal vein invasion (0/1/2/3/4)	21/1/2/3/0	148/1/9/10/12	0.224	84/0/5/6/6	85/2/6/7/6	0.725
Presence of extrahepatic spread	10 (37.0%)	74 (40.9%)	0.703	44 (43.6%)	40 (37.4%)	0.363
ALBI score	−2.14 (−3.16–1.00)	−2.22 (−3.36–0.79)	0.749	−2.17 (−3.18 to −1.00)	−2.31 (−3.36 to −0.79)	0.185
Child–Pugh stage A	23 (85.2%)	147 (81.2%)	0.618	79 (78.2%)	91 (85.1%)	0.202
Alpha‐fetoprotein (ng/mL)	22 (0.6–47 731)	36 (0.6–348 209)	0.848	30.5 (0.9–348 209)	41.5 (0.6–97 315)	0.638
C‐reactive protein (mg/dL)	1.49 (0.06–4.92)	0.40 (0.02–9.53)	0.073	0.45 (0.02–8.05)	0.4 (0.02–9.53)	0.292
Neutrophil lymphocyte ratio	3.27 (1.15–9.94)	2.85 (0.74–19.6)	0.208	3.17 (1.15–17.1)	2.83 (0.75–19.6)	0.459

Note: Values are indicated as median (range) unless otherwise noted.

Abbreviations: ALBI, albumin‐bilirubin; BCLC, barcelona clinic liver cancer; ECOG, eastern cooperative oncology group; PPI, proton pump inhibitor.

Regarding tumor burden, the proportion of patients with more than five intrahepatic tumors did not differ significantly between the butyric acid and non‐butyric acid groups (52% vs. 34%, p = 0.083). Similarly, intrahepatic tumor size, presence of portal vein invasion, and extrahepatic spread were comparable between groups. No significant differences in tumor burden parameters were observed between the PPI group and non‐PPI group.

Overall, patient characteristics were well‐balanced between groups, with no statistically significant differences in clinical and tumor burden parameters.

3.2. Tumor Response

Best tumor response, as evaluated by RECIST v1.1, is shown in Table 2. The objective response rate (ORR) was 3.7% in the butyric acid group and 15.5% in the non‐butyric acid group (p = 0.543). The disease control rate (DCR) was 29.6% and 46.4%, respectively. Among PPI users, the ORR was 15.8% compared to 12.1% in non‐PPI users (p = 0.222), whereas the DCR was 42.5% and 45.7%, respectively. There were no statistically significant differences in treatment response between groups.

TABLE 2.

Best response evaluated by RECIST.

	CR	PR	SD	PD	NE	ORR	DCR
Butyric acid group	0 (0%)	1 (3.7%)	7 (25.9%)	17 (63.0%)	2 (7.4%)	1 (3.7%)*	9 (29.6%)
Non‐butyric acid group	1 (0.6%)	27 (14.9%)	56 (30.9%)	78 (43.1%)	19 (10.5%)	28 (15.5%)*	84 (46.4%)
PPI group	0 (0%)	16 (15.8%)	27 (26.7%)	46 (45.5%)	12 (11.9%)	16 (15.8%)**	43 (42.5%)
Non‐PPI group	1 (0.9%)	12 (11.2%)	36 (33.6%)	49 (45.8%)	9 (8.4%)	13 (12.1%)**	49 (45.7%)
Total	1 (0.5%)	28 (13.5%)	63 (30.3%)	95 (45.7%)	21 (10.1%)	29 (14.0%)	92 (44.3%)

Abbreviations: CR, complete response; DCR, disease control rate; NE, not evaluable; ORR, objective response rate; PD, progressive disease; PPI, proton pump inhibitor; PR, partial response; SD, stable disease.

^* p = 0.543.

^** p = 0.222.

3.3. OS Analysis

Kaplan–Meier survival curves comparing overall survival (OS) between groups after IPTW adjustment are presented in Figure 1. The OS did not significantly differ between the butyric acid and non‐butyric acid groups (p = 0.921). Similarly, no significant difference was observed between the PPI and non‐PPI groups (p = 0.917). No significant OS differences were observed between the butyric acid and non‐butyric acid groups, either within the PPI group or the non‐PPI group.

FIGURE 1.

Overall survival of patients with hepatocellular carcinoma treated with durvalumab and tremelimumab combination therapy. Inverse probability of treatment weighting was applied, accounting for Barcelona Clinic Liver Cancer stage, Albumin‐Bilirubin score, alpha‐fetoprotein, performance status, and whether durvalumab plus tremelimumab was used as first‐line therapy. (a) Overall survival did not significantly differ between the butyric acid and non‐butyric acid groups (p = 0.921). (b) Similarly, no significant difference was observed between the proton pump inhibitor (PPI) and non‐PPI groups (p = 0.917).

In the multivariate analysis (Table 3), ECOG performance status of 1 (HR = 2.02, 95% CI: 1.11–3.69, p = 0.021) and ALBI score > −2.22 (HR = 2.39, 95% CI: 1.40–4.09, p = 0.001) were identified as independent prognostic factors for OS. However, the presence of butyrate‐producing bacteria (HR = 0.94, 95% CI: 0.42–2.13, p = 0.886) or PPI use (HR = 1.11, 95% CI: 0.66–1.88, p = 0.685) was not associated with survival outcomes.

TABLE 3.

Factors correlated with patients' survival (multivariate analysis).

Variables	Univariate analysis			Multivariate analysis
	Hazard ratio	95% CI	p	Hazard ratio	95% CI	p
Butyric acid group	1.09	0.49–2.41	0.828	0.94	0.42–2.13	0.886
PPI group	1.11	0.67–1.85	0.668	1.11	0.66–1.88	0.685
ECOG performance status 1 (vs. 0)	2.39	1.35–4.22	0.002	2.02	1.11–3.69	0.021
BCLC stage C (vs. A/B)	1.47	0.85–2.54	0.165	1.17	0.65–2.08	0.598
AFP > 40 ng/mL (vs. ≦ 40 ng/mL)	1.62	0.97–2.73	0.062	1.48	0.88–2.50	0.143
ALBI score > −2.22 (vs. ≦ −2.22)	2.48	1.46–4.20	< 0.001	2.39	1.40–4.09	0.001
Use in first‐line systemic therapy	0.57	0.31–1.07	0.079	0.61	0.33–1.15	0.126

Abbreviations: AFP, alpha‐fetoprotein; ALBI, albumin‐bilirubin; BCLC, barcelona clinic liver cancer; ECOG, eastern cooperative oncology group; PPI, proton pump inhibitor.

4. Discussion

In this retrospective cohort study, we investigated the impact of butyrate‐producing bacteria and PPI use on the efficacy of durvalumab plus tremelimumab therapy in patients with HCC. Contrary to our initial hypothesis, neither butyrate‐producing bacteria nor PPI use significantly influenced OS or tumor response rates. These findings contrast with prior studies demonstrating a beneficial role of gut microbiota modulation in immune checkpoint inhibitor (ICI) therapy for other malignancies, such as renal cell carcinoma and melanoma [1, 4, 14].

One possible explanation for the lack of effect observed in our study is the unique immunological environment of the liver. The liver is typically considered an immunotolerant organ, which may contribute to the reduced efficacy of ICI therapy in HCC compared to other cancers [15]. This phenomenon is thought to be driven by the hepatic microenvironment, which is rich in regulatory immune cells, immunosuppressive cytokines, and metabolic factors that dampen anti‐tumor immune responses [16]. Recent research has highlighted that bile acids, particularly primary and conjugated bile acids, accumulate in the HCC tumor microenvironment and impair CD8+ T cell function [17]. These findings suggest that the liver's intrinsic immunosuppressive mechanisms are different from that of renal cell carcinoma or melanoma, and may override potential microbiome‐mediated immunomodulation, thereby diminishing the impact of butyrate‐producing bacteria on ICI efficacy.

Furthermore, while previous studies have suggested that PPI use may negatively affect ICI outcomes by altering gut microbiota composition and reducing microbial diversity [11], our study did not observe a significant impact of PPIs on treatment efficacy. This may be due to the robust nature of the liver's immunosuppressive microenvironment, which might mitigate the microbiota‐dependent effects observed in other cancers.

A potential avenue for improving ICI efficacy in HCC may involve targeting bile acid metabolism. Recent findings indicate that ursodeoxycholic acid (UDCA), a secondary bile acid, may enhance anti‐tumor immunity by counteracting the immunosuppressive effects of conjugated bile acids [17]. In preclinical models, UDCA has been shown to alleviate endoplasmic reticulum stress in CD8+ T cells, thereby enhancing their anti‐tumor function. Given these promising findings, future studies should explore whether UDCA supplementation could serve as an adjunctive strategy to improve responses to immune checkpoint blockade in HCC.

Our study has several limitations. First, the butyrate group was substantially smaller than the control group (27 vs. 181), which may have reduced the statistical power of our analyses. This imbalance could limit the ability to detect clinically meaningful differences, and a type II error remains possible. Second, while we adjusted for potential confounders using IPTW, unmeasured variables such as baseline microbiota composition, dietary habits, or concomitant medications may still have influenced our findings. In addition, the indications for prescribing butyrate‐producing bacteria or PPIs were heterogeneous and incompletely separable, and the small numbers in each subgroup precluded sensitivity analyses. Thus, residual confounding by indication cannot be excluded. Third, we did not assess specific alterations in the gut microbiome composition, which may have provided mechanistic insights into the lack of observed effects. Future prospective studies incorporating microbiome sequencing and bile acid profiling may help clarify the complex interactions between the gut‐liver axis and immunotherapy efficacy in HCC.

In conclusion, our study indicates that butyrate‐producing bacteria and PPI use do not significantly influence the efficacy of durvalumab plus tremelimumab therapy in HCC. The intrinsic immunosuppressive properties of the liver may limit the impact of microbiome modulation in this context. Future research should focus on strategies to overcome hepatic immune resistance and enhance immune responses in liver cancer.

Funding

The authors have nothing to report.

Ethics Statement

Approval of the Research Protocol: The study was approved by the NHO Takasaki General Medical Center Clinical Research Ethics Committee (approved No. TGMC2024‐003) and by each participating institution.

Consent

All patients provided consent to review their clinical records for this study.

Conflicts of Interest

Kazuhiro Nouso has received lecture fees from Eisai. Atsushi Hiraoka has received lecture fees from Chugai, AstraZeneca, and Eli Lilly. Toshifumi Tada has received lecture fees from AbbVie, Eisai, AstraZeneca, and Chugai. Satoru Kakizaki has received lecture fees and research funding from AbbVie. Kazuhito Kawata has received lecture fees from AbbVie. Hidekatsu Kuroda has received lecture fees from Chugai, Eisai, and AstraZeneca. Masatoshi Kudo has received fees for advisory role from Chugai, Roche, Eisai, and AstraZeneca, MSD, lecture fees from Chugai, Eisai, and AstraZeneca, research funding from Otsuka, Taiho, Chugai, GE Healthcare, and Eisai.

Supporting information

Table S1: Standardized mean differences before and after IPDW matching.

Nouso K., Wakuta A., Shiota S., et al., “Effect of Butyrate‐Producing Enterobacteria and Proton Pump Inhibitors on Advanced Hepatocellular Carcinoma Treatment With Durvalumab and Tremelimumab,” JGH Open 10, no. 1 (2026): e70346, 10.1002/jgh3.70346.

Data Availability Statement

The data is not open to the public.