Introduction
As the prevalence of obesity and insulin resistance continues to rise, nonalcoholic fatty liver disease (NAFLD), now rebranded as metabolic dysfunction-associated fatty liver disease (MAFLD), has emerged as the most prevalent parenchymal liver disease worldwide and explains 9% of deaths from liver cirrhosis.1–3 Currently, there are no approved pharmacotherapies for fatty liver disease.4 Bariatric surgery for fatty liver disease has enjoyed a high profile due to its remarkable capacity for improving liver enzyme, NAFLD activity score, and fibrosis.5,6 However, unexpected rates of liver fibrosis progression in patients who undergo bariatric surgery and excessive risks of postoperative complications limit the acceptance of bariatric surgery.7,8 Additionally, lifestyle modification strategies are difficult to address the disadvantage regarding treatment compliance.9,10 As a result, novel therapeutic applications, which take all efficacy, safety, and treatment compliance into account, are urgently needed for all MAFLD patients.
Recently, the potential role of endoscopic bariatric and metabolic therapies (EBMT) in the management of fatty liver disease has been highlighted.11,12 EBMT are developed to avoid the invasive nature of laparoscopic or open bariatric surgery, in contrast, reproducing similar gastrointestinal physiological alterations and therapeutic effects.13 Among these interventions, intragastric balloon (IGB), as a space-occupying EBMT device with proven efficacy in inducing weight loss, has been used in diminishing liver volume to reduce the risks of subsequent bariatric surgery and has met with success.14,15 Prior study has demonstrated that the change in liver volume was positively correlated with the change in intrahepatic fat,16 which suggested the potential therapeutic effect of using IGB in fatty liver disease. In terms of current evidence, a randomized controlled trial (RCT) evaluated changes in histological scores after 6-month IGB therapy and showed a beneficial effect on the severity of fatty liver disease.17 However, due to the limited sample size of this trial, we still need to combine the existing RCT findings with observational longitudinal studies to present the effectiveness of IGB in larger sample size, before it is widely recommended for the treatment of MAFLD. Therefore, we performed a systematic review with meta-analyses to evaluate the therapeutic effect of IGB on the markers of MAFLD, such as homeostasis model assessment of insulin resistance (HOMA-IR) index, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transpeptidase (GGT). Furthermore, to identify patients most appropriate for IGB therapy, stratified analyses and meta-regression were both implemented.
Methods
Data sources and search strategy
This systematic review was performed according to the preferred reporting items for systematic reviews and meta-analysis statement (see Table S1).18 The protocol for this review is registered in PROSPERO (no. CRD42020214315).
To collect all full-text articles describing the effect of IGB on the markers of MAFLD, we performed a search of the Medline, Cochrane Library, and Web Of Science with English-language restriction and up to September 2020 using the following strategy: (“Intragastric balloon” OR “Gastric balloon”) AND (“Alanine aminotransferase” OR “Alanine transaminase” OR “ALT” OR “Liver” OR “Nonalcoholic fatty liver disease” OR “Non-alcoholic fatty liver disease” OR “NASH” OR “NAFLD” OR “HOMA-IR” OR “Homeostasis model assessment” OR “Insulin resistance”). The detailed search strategy is summarized in Table S2. Furthermore, the reference lists of each article were manually searched to prevent the omission of any pertinent study.
Study eligibility and selection criteria
Only observational longitudinal studies and RCTs were included. Inclusion criteria of the articles were as follows: (a) population: all patients who are obese or in need of obesity treatment; (b) intervention: liquid-filled IGB procedure; (c) comparator: the participants at baseline before IGB placement; and (d) outcome: the decrease of ALT, AST, GGT, or HOMA-IR index in all the participants treated with IGB. Moreover, the studies which recruited only pediatric patients or utilized the gas-filled IGB as an intervention were excluded to prevent bias.
Data extraction and quality assessment
Data extraction was performed independently by two investigators (ZYZ, JZ). The information and characteristics extracted from the included study were first author, year of publication, study design, country, study size of participants with IGB therapy, IGB type, dwelling time of IGB, filling of IGB, method of IGB implantation, additional nutrition and exercise prescription, description of liver disease in exclusion criteria, percentage of male individuals, prevalence of diabetes, participants’ age and body mass index (BMI) at baseline, and participants’ ALT, AST, GGT and HOMA-IR before and after IGB therapy. When standard deviation was unavailable, it was replaced with a quarter of the range.19 The risk of bias of the selected studies was evaluated using the modified Newcastle-Ottawa scale (NOS) for observational longitudinal studies20 and Cochrane Collaboration’s tool for RCT.21
Data analysis
Using R software version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) and Review Manager version 5.3 (Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark), meta-analyses (quantitative synthesis) were performed to evaluate the pooled mean difference (MD) in HOMA-IR, ALT, AST and GGT from baseline to end of IGB therapy using the inverse variance method and random-effect model, with 95% confidence interval (CI) and p-value. A p-value <0.05 was considered statistically significant. Publication bias was evaluated by Egger’s test and funnel plot.22,23 Heterogeneity was evaluated with inconsistency index (I2), classified as a low (I2≥25%), substantial (I2≥50%), or considerable (I2≥75%).24 Stratified analyses were conducted to investigate sources of heterogeneity based on the following characteristics: method of IGB implantation; mean basal level of serum markers (HOMA-IR, ALT, AST, or GGT); age and BMI of the participants; study region; and NOS score. When meta-regression analysis was performed, univariate and multivariate linear regression models were utilized to evaluate the slope coefficient between the reduced value of serum marker (HOMA-IR, ALT, AST, or GGT) after IGB therapy and the following covariates: mean basal level of serum marker; percentage of male individuals; and age and BMI of the participants. To summarize the results, the scatter plots were mapped to materialize the linear relationship between the changed value after IGB therapy and covariates which had statistical significance with both univariate and multivariate meta-regression analysis (p<0.05). Each study was represented by a circle of size proportional to the inverse of the variance of MD.
Results
Literature search results
Figure 1 summarizes the flow diagram of the selection process performed to identify eligible studies in this systematic review. Out of 152 references, a total of 14 studies25–38 comprising 624 participants met the predefined inclusion criteria. All studies were published prior to September 13, 2020.
Improvement of insulin resistance after IGB on therapy
Summary of study characteristics
Eight studies25–27,29,33,35,36,38 with a total of 352 individuals were included in this meta-analysis of HOMA-IR level, and their characteristics are summarized in Table 1. All included studies were published after 2007. Of these, one38 was a two-arm RCT, and the rest25–27,29,33,35,36 were observational longitudinal studies, meaning that a total of nine intervention arms were included in this analysis. The participants came from three countries (Brazil, Italy, Japan). Seven intervention arms25–27,29,33,38 applied the Orbera IGB system, one arm36 used the Orbera/Spatz IGB system, and the single remaining arm35 reported results with the Elipse IGB system. Furthermore, the range of average baseline HOMA-IR was from 2.36 to 12.30. The results of the quality assessment using the modified NOS and Cochrane Collaboration’s tool can be found in Table S3 and Figure S1.
Table 1Characteristics of the included studies
| Author | IGB group, n | Type of study; Country; Prevalence of diabetes | Nutrition and exercise prescription | IGB type; IGB duration; Filling; Implantation of IGB | Liver disease excluded |
|---|
| Bazerbachi et al.37 | 21 | Observational study; USA; 52% | Low-calorie diet + lifestyle therapy | Orbera; 6 months; Liquid-filled; Endoscopically | Other primary causes of liver disease |
| Maekawa et al.38 | 18 | RCT; Japan; Not described | Low-carbohydrate diet | Orbera; 6 months; Liquid-filled; Endoscopically | Not described |
| Maekawa et al.38 | 13 | RCT; Japan; Not described | Low-calorie diet | Orbera; 6 months; Liquid-filled; Endoscopically | Not described |
| Guedes et al.36 | 42 | Observational study; Brazil; Not described | Low-calorie diet | Orbera/Spatz; 6 months; Liquid-filled; Endoscopically | Not described |
| Genco et al.35 | 38 | Observational study; Italy; Not described | Low-calorie diet + exercise counseling | Elipse; 4 months; Liquid-filled; Swallowable | Not described |
| Raftopoulos et al.34 | 12 | Observational study; Greece; Not described | Diet and exercise counseling | Elipse; 4 months; Liquid-filled; Swallowable | No history of alcohol |
| Folini et al.32 | 13 | Observational study; Italy; Not described | Low-calorie diet + exercise counseling | Orbera; 6 months; Liquid-filled; Endoscopically | Alcohol consumption, presence of any predisposing disorders for liver diseases, pregnancy, and lactation |
| Takihata et al.33 | 8 | Observational study; Japan; Not described | Low-calorie diet | Orbera; 6 months; Liquid-filled; Endoscopically | Not described |
| Tai et al.31 | 28 | Observational study; China (Taiwan); Not described | Low-calorie diet | Orbera; Median 200 days; Liquid-filled; Endoscopically | Alcoholism or drug addiction |
| Nikolic et al.28 | 33 | Observational study; Croatia; Not described | Low-calorie diet | Orbera; 6 months; Liquid-filled; Endoscopically | Not described |
| Sekino et al.29 | 8 | Observational study; Japan; Not described | Not described | Orbera; 6 months; Liquid-filled; Endoscopically | Not described |
| Stimac et al.30 | 165 | Observational study; Croatia; Not described | Not described | Orbera; 6 months; Liquid-filled; Endoscopically | Present alcohol or drug abuse |
| Forlano et al.27 | 120 | Observational study; Italy; 13.3% | Low-calorie diet | Orbera; 6 months; Liquid-filled; Endoscopically | Use of drugs reported to cause liver damage, alcohol intake of 30 g/day or more, and viral hepatitis |
| Donadio et al.26 | 40 | Observational study; Italy; Not described | Not described | Orbera; 6 months; Liquid-filled; Endoscopically | Alcoholism |
| Ricci et al.25 | 65 | Observational study; Italy; Not described | Low-calorie diet | Orbera; 6 months; Liquid-filled; Endoscopically | Positivity for hepatitis B virus or hepatitis C virus, previous or current alcohol consumption >30 g/day, use of medications with reported hepatosteatogenic effect (amiodarone, tamoxifen, estrogens), and type 1 diabetes |
Quantitative synthesis and stratified analyses
Nine intervention arms25–27,29,33,35,36,38 of 352 participants evaluated the effect of IGB on HOMA-IR. The pooled mean decrease in HOMA-IR levels with IGB therapy was 1.56 (95% CI=1.16–1.95, I2=61.1; Fig. 2A). According to the Egger’s test and funnel plot, no significant publication bias was present (p=0.2665; Fig. S2A). Table 2 presents the results of the stratified analyses. Both endoscopic IGB (MD=1.68, 95% CI=1.24–2.11) and swallowable IGB (MD=0.90, 95% CI=0.26–1.54) were effective in inducing HOMA-IR loss. There were trends showing the advanced age group had less change in HOMA-IR (MD=1.07, 95% CI=0.57–1.56) compared to those ≤40 years (MD=1.82, 95% CI=1.25–2.40), but the findings were not statistically significant (p=0.0502). Higher baseline HOMA-IR (>5) was associated with more significant reductions in HOMA-IR [MD=3.48 (95% CI=2.46–4.50) vs. MD=1.40 (95% CI=1.25–1.54), p<0.0001)]. Consequently, intra-subgroup heterogeneity was significantly diminished and almost absent with different basal HOMA-IR (basal HOMA-IR ≤5: I2=0.0; basal HOMA-IR >5: I2=0.0).
Table 2Pooled change in HOMA-IR, ALT, AST, and GGT after IGB treatment and removal: Stratified analyses
| Intervention arm, n | MD (95% CI) | I2 |
|---|
| Pooled change in HOMA-IR after IGB treatment and removal |
| Insertion of IGB (IGB type) | | | |
| Endoscopic (Orbera/Spatz) | 8 | 1.68 (1.24–2.11) | 60.3 |
| Swallowable (Elipse) | 1 | 0.90 (0.26–1.54) | – |
| Basal HOMA-IR | | | |
| ≤5 | 7 | 1.40 (1.25–1.54) | 0.0 |
| >5 | 2 | 3.48 (2.46–4.50) | 0.0 |
| Mean age, years | | | |
| ≤40 | 4 | 1.82 (1.25–2.40) | 82.0 |
| >40 | 5 | 1.07 (0.57–1.56) | 0.0 |
| Mean BMI, kg/m2 | | | |
| ≤40 | 4 | 1.35 (1.19–1.51) | 0.0 |
| >40 | 5 | 2.01 (1.25–2.77) | 66.4 |
| Region | | | |
| Asia | 4 | 1.35 (1.19–1.51) | 72.4 |
| Europe | 4 | 1.41 (1.01–1.81) | 29.1 |
| South America | 1 | 1.39 (1.22–1.56) | – |
| NOS scale | | | |
| High | 3 | 1.37 (0.88–1.87) | 52.0 |
| Fair | 4 | 2.16 (0.87–3.44) | 81.2 |
| Pooled change in ALT after IGB treatment and removal |
| Insertion of IGB (IGB type) | | | |
| Endoscopic (Orbera) | 10 | 10.85 (6.31–15.39) | 55.9 |
| Swallowable (Elipse) | 1 | 20.27 (6.49–34.05) | – |
| Basal ALT, U/L | | | |
| ≤40 | 7 | 9.58 (6.18–12.98) | 38.7 |
| >40 | 4 | 32.43 (18.49–46.37) | 0.0 |
| Mean age, years | | | |
| ≤40 | 6 | 10.40 (5.38–15.41) | 54.6 |
| >40 | 5 | 15.57 (5.20–25.93) | 64.6 |
| Mean BMI, kg/m2 | | | |
| ≤40 | 2 | 22.61 (11.49–33.74) | 0.0 |
| >40 | 9 | 9.98 (5.59–14.38) | 53.7 |
| Region | | | |
| Asia | 3 | 25.80 (9.69–41.91) | 0.0 |
| Europe | 7 | 9.58 (6.18–12.98) | 38.7 |
| North America | 1 | 9.88 (7.33–12.44) | – |
| NOS scale | | | |
| High | 4 | 12.71 (5.27–20.16) | 78.0 |
| Fair | 7 | 10.59 (4.84–16.35) | 29.4 |
| Pooled change in AST after IGB treatment and removal |
| Insertion of IGB (IGB type) | | | |
| Endoscopic (Orbera) | 6 | 6.74 (0.53–12.96) | 60.4 |
| Swallowable (Elipse) | 1 | 8.60 (2.41–14.79) | – |
| Basal AST, U/L | | | |
| ≤40 | 6 | 4.52 (1.05–7.99) | 29.8 |
| >40 | 1 | 36.18 (13.62–58.74) | 0 |
| Mean age, years | | | |
| ≤40 | 4 | 3.30 (−0.66 to 7.26) | 29.9 |
| >40 | 3 | 14.54 (−0.04 to 29.12) | 63.5 |
| Mean BMI, kg/m2 | | | |
| ≤40 | 2 | 8.77 (2.95–14.58) | 0.0 |
| >40 | 5 | 6.64 (−0.20 to 13.49) | 66.8 |
| Region | | | |
| Asia | 3 | 11.15 (1.77–20.53) | 0 |
| Europe | 3 | 3.59 (−0.34 to 7.52) | 52.1 |
| North America | 1 | 36.18 (13.62–58.74) | – |
| NOS scale | | | |
| High | 2 | 17.67 (−14.52 to 49.86) | 87.7 |
| Fair | 5 | 6.17 (0.53–11.81) | 38.2 |
| Pooled change in GGT after IGB treatment and removal |
| Insertion of IGB (IGB type) | | | |
| Endoscopic (Orbera) | 8 | 9.45 (4.46–14.45) | 53.0 |
| Swallowable (Elipse) | 0 | – | – |
| Basal GGT, U/L | | | |
| ≤40 | 6 | 8.74 (2.89–14.59) | 66.2 |
| >40 | 2 | 12.96 (−0.23 to 26.15) | 0.0 |
| Mean age, years | | | |
| ≤40 | 5 | 8.75 (1.71–15.79) | 71.3 |
| >40 | 3 | 8.80 (2.02–15.58) | 0.0 |
| Mean BMI, kg/m2 | | | |
| ≤40 | 8 | 9.45 (4.46–14.45) | 53.0 |
| >40 | 0 | – | – |
| Region | | | |
| Asia | 2 | 12.96 (−0.23 to 26.15) | 0.0 |
| Europe | 6 | 8.74 (2.89–14.59) | 66.2 |
| NOS scale | | | |
| High | 3 | 10.10 (2.49–17.72) | 80.1 |
| Fair | 5 | 7.88 (1.86–13.89) | 0.0 |
Meta-regression
Table 3 presents the meta-regression findings of HOMA-IR. In univariate meta-regression, basal HOMA-IR of the participants (slope coefficient=0.3966, 95% CI=0.1119–0.6814, p=0.0063) and percentage of male individuals (slope coefficient=0.0433, 95% CI=0.0183 to 0.0684, p=0.0007) seemed to be factors significantly associated with reductions in HOMA-IR. Subsequently, using a multivariate meta-regression approach, our final model consisted of four covariates: basal HOMA-IR, percentage of male individuals, age and BMI of the participants. Greater HOMA-IR loss was predicted by younger age (slope coefficient=−0.0932, 95% CI=−0.1647 to −0.0216, p=0.0107).
Table 3Univariate and multivariate meta-regression analyses on the mean deference of HOMA-IR, ALT, AST, and GGT after IGB treatment and removal
| Moderators | Intervention arm, n | Univariable analysis
| Multivariable analysis
|
|---|
| Slope coefficient (95% CI) | p | Slope coefficient (95% CI) | p |
|---|
| Mean deference of HOMA-IR after IGB treatment and removal |
| Basal HOMA-IR | 9 | 0.3966 (0.1119–0.6814) | 0.0063 | 0.2361 (−0.2764 to 0.7487) | 0.3665 |
| Mean age | 9 | −0.0753 (−0.1891 to 0.0386) | 0.1950 | −0.0932 (−0.1647 to −0.0216) | 0.0107 |
| Mean BMI | 9 | 0.0975 (−0.0288 to 0.2238) | 0.1302 | −0.0473 (−0.1298 to 0.0351) | 0.2607 |
| Male | 9 | 0.0433 (0.0183–0.0684) | 0.0007 | 0.0381 (−0.0094 to 0.0856) | 0.1159 |
| Mean deference of ALT after IGB treatment and removal |
| Basal ALT | 11 | 0.7314 (0.3862–1.0767) | <0.0001 | 0.7135 (0.3213–1.1057) | 0.0004 |
| Mean age | 11 | 0.8603 (−0.3376 to 2.0582) | 0.1593 | 0.3408 (−0.5602 to 1.2417) | 0.4585 |
| Mean BMI | 11 | −1.1489 (−2.6746 to 0.3767) | 0.1399 | −0.5035 (−1.6750 to 0.6681) | 0.3996 |
| Male | 11 | 0.0851 (−0.3480 to 0.5181) | 0.7002 | −0.0652 (−0.3742 to 0.2437) | 0.6791 |
| Mean deference of AST after IGB treatment and removal |
| Basal AST | 7 | 0.7650 (0.3319–1.1982) | 0.0005 | 0.5438 (−0.0501 to 1.1378) | 0.0727 |
| Mean age | 7 | 1.4430 (0.5644–2.3216) | 0.0013 | 0.5348 (−0.8576 to 1.9272) | 0.4516 |
| Mean BMI | 7 | −0.1374 (−1.5495 to 1.2747) | 0.8488 | −0.1086 (−0.8199 to 0.6028) | 0.7648 |
| Male | 7 | 0.1362 (−0.2020 to 0.4744) | 0.4299 | 0.0699 (−0.2378 to 0.3777) | 0.6562 |
| Mean deference of GGT after IGB treatment and removal |
| Basal GGT | 8 | 0.7968 (0.2032–1.3904) | 0.0085 | 1.3773 (0.4793–2.2754) | 0.0026 |
| Mean age | 8 | 0.5022 (−2.0156 to 3.0201) | 0.6958 | 0.3219 (−1.9139 to 2.5577) | 0.7778 |
| Mean BMI | 8 | −0.2371 (−4.2184 to 3.7441) | 0.9071 | −1.3277 (−4.6165 to 1.9612) | 0.4288 |
| Male | 8 | 0.0996 (−0.3201 to 0.5193) | 0.6418 | −0.4310 (−0.9670 to 0.1051) | 0.1151 |
Decrease in ALT after IGB therapy
Summary of study characteristics
Eleven observational longitudinal studies25–34,37 with a total of 513 individuals were included in this meta-analysis of ALT level, and their characteristics are summarized in Table 1. All included studies were published after 2007. The participants included in the meta-analysis of ALT level came from six countries (China, Croatia, Greece, Italy, Japan, USA). Ten studies25–33,37 applied the Orbera IGB system, and one study34 reported results with the Elipse IGB system. Furthermore, the range of average baseline ALT was from 26.0 to 91.6 U/L. The results of the quality assessment using the modified NOS can be found in Table S3.
Quantitative synthesis and stratified analyses
Eleven studies25–34,37 of 513 participants evaluated the effect of IGB on ALT. The pooled mean decrease of ALT with IGB therapy was 11.53 U/L (95% CI=7.10–15.96, I2=55.4; Fig. 2B). According to the Egger’s test and funnel plot, no significant publication bias was present (p=0.2422; Fig. S2B). Table 2 presents the results of the stratified analyses. Both endoscopic IGB (MD=10.85 U/L, 95% CI=6.31–15.39) and swallowable IGB (MD=20.27 U/L, 95% CI=6.49–34.05) were effective in inducing ALT loss. The advanced age group had similar change in ALT (MD =15.57 U/L, 95% CI=5.20–25.93) compared to those ≤40 years (MD =10.40 U/L, 95% CI=5.38–15.41). Higher baseline ALT (>40 U/L) was associated with more significant reductions in ALT [MD=32.43 U/L (95% CI=18.49–46.37) vs. MD=9.58 U/L (95% CI=6.18–12.98), p=0.0018]. Overall, intra-subgroup heterogeneity in different basal ALT diminished significantly and was classified as a low (basal ALT ≤40 U/L: I2=38.7; basal ALT >40 U/L: I2=0.0).
Meta-regression
Table 3 presented the meta-regression findings of ALT. In univariate meta-regression, basal ALT of the participants (slope coefficient=0.7314, 95% CI=0.3862–1.0767, p<0.0001) seemed to be a factor significantly associated with reductions in ALT. Subsequently, using a multivariate meta-regression approach, our final model consisted of four covariates: basal ALT; percentage of male individuals; age; and BMI. Effectiveness on ALT was predicted by basal ALT (slope coefficient=0.7135, 95% CI=0.3213–1.1057, p=0.0004). The scatter plot showed a linear trend towards increasing effectiveness of IGB therapy with increasing basal ALT of the participants (Fig. 3A).
Decrease in AST after IGB therapy
Summary of study characteristics
Seven observational longitudinal studies26,28,29,31,33,34,37 with a total of 150 individuals were included in this meta-analysis of AST level, and their characteristics are summarized in Table 1. The participants included in the meta-analysis of AST level came from six countries (China, Croatia, Greece, Italy, Japan, USA). Six studies26,28,29,31,33,37 applied the Orbera IGB system, and one study34 reported results with the Elipse IGB system. Furthermore, the range of average baseline AST was from 21.7 to 67.5 U/L. The results of the quality assessment using the modified NOS can be found in Table S3.
Quantitative synthesis and stratified analyses
Seven studies of 150 participants evaluated the effect of IGB on AST. The pooled mean decrease of AST with IGB therapy was 6.79 U/L (95% CI=1.69–11.90, I2=59.9; Fig. 2C). According to the Egger’s test and funnel plot, no significant publication bias was present (p=0.3768; Fig. S2C). Table 2 presents the results of the stratified analyses. Both endoscopic IGB (MD=6.74 U/L, 95% CI=0.53–12.96) and swallowable IGB (MD=8.60 U/L, 95% CI=2.41–14.79) were effective in inducing AST loss. The advanced age group had a similar change in AST (MD =14.54 U/L, 95% CI=−0.04 to 29.12) compared to those ≤40 years (MD=3.30 U/L, 95% CI=−0.66 to 7.26). Higher baseline AST (>40 U/L) was associated with more significant reductions in AST [MD=36.18 U/L (95% CI=13.62–58.74) vs. MD=4.52 U/L (95% CI=1.05-7.99, p=0.0065)]. Overall, intra-subgroup heterogeneity in different basal AST diminished significantly and was classified as a low (basal AST ≤40 U/L: I2=29.8; basal AST >40 U/L: I2=0.0).
Meta-regression
Table 3 presents the meta-regression findings of AST. In univariate meta-regression, basal AST of the participants (slope coefficient=0.7650, 95% CI=0.3319–1.1982, p=0.0005) and age of the participants (slope coefficient=1.4430, 95% CI=0.5644–2.3216, p=0.0013) seemed to be factors significantly associated with reductions in AST. Subsequently, using a multivariate meta-regression approach, our final model consisted of four covariates: basal AST; percentage of male individuals; age; and BMI. Effectiveness on AST could not be predicted by all of the above covariates.
Decrease in GGT after IGB therapy
Summary of study characteristics
Eight observational longitudinal studies25–30,32,33 with a total of 452 individuals were included in this meta-analysis of GGT level, and their characteristics are summarized in Table 1. The participants included in the meta-analysis of GGT level came from three countries (Croatia, Italy, Japan). All eight studies25–30,32,33 applied the Orbera IGB system. Furthermore, the range of average baseline GGT was from 29.8 to 53.0 U/L. The results of the quality assessment using the modified NOS can be found in Table S3.
Quantitative synthesis and stratified analyses
Eight studies25–30,32,33 of 452 participants evaluated the effect of IGB on GGT. The pooled mean decrease of GGT with IGB therapy was 10.54 U/L (95% CI=6.32–14.75, I2=37.6; Fig. 2D). According to the Egger’s test and funnel plot, no significant publication bias was present (p=0.8620; Fig. S2D). Table 2 presented the results of the stratified analyses. The advanced age group had a similar change in GGT (MD =8.80 U/L, 95% CI=2.02–15.58) compared to those ≤40 years (MD=8.75 U/L, 95% CI=1.71–15.79). There were trends showing that the higher basal GGT group had more change in GGT (MD=12.96, 95% CI=−0.23 to 26.15) compared to those ≤40 U/L (MD=8.74, 95% CI=2.89–14.59) but the findings were not statistically significant (p=0.6919). Overall, intra-subgroup heterogeneity diminished significantly in the higher basal GGT group (I2=0.0).
Meta-regression
Table 3 presents the meta-regression findings of GGT. In univariate meta-regression, basal GGT of the participants (slope coefficient=0.7968, 95% CI=0.2032–1.3904, p=0.0085) seemed to be a factor significantly associated with reductions in GGT. Subsequently, using a multivariate meta-regression approach, our final model consisted of four covariates: basal GGT; percentage of male individuals; age; and BMI. Effectiveness on GGT was predicted by basal GGT (slope coefficient=1.3773, 95% CI=0.4793–2.2754, p=0.0026). The scatter plot showed a linear trend towards increasing effectiveness of IGB therapy with increasing basal GGT of the participants (Fig. 3B).
Discussion
Principal findings and relevant mechanisms
IGB is the most widely available EBMT with proven efficacy in inducing weight loss. According to the IGB type, an empty balloon is introduced into the stomach by an upper gastrointestinal endoscopy or by swallowing the balloon capsule directly. The liquid-filled IGB is inflated with saline and methylene blue to occupy the space in the stomach. After that, the IGB dwells in the stomach for 4 to 6 months until it ruptures or is removed.14,39 Due to its moderate efficacy of weight loss and excellent safety profiles, the potential utility of IGB was mentioned by the Asian-Pacific clinical practice guideline on MAFLD.40 IGB has also been employed for clinical research of fatty liver disease. However, there is still substantial heterogeneity in results across studies. One explanation is that patients with fatty liver disease can be subdivided into IGB responder and non-responder groups. In this systematic review with meta-analysis, we demonstrated that IGB could reverse the serum markers of MAFLD, including HOMA-IR, ALT, AST, and GGT levels. Furthermore, the change of ALT and GGT with IGB therapy had a positive linear relationship with the basal value. This means that even at higher levels of disease severity, abnormal liver enzymes can be controlled within the reported range of included studies (ALT: 26.0–91.6 U/L; GGT: 29.8–53.0 U/L).
Due to the dearth of eligible studies, the histological and radiological findings cannot be quantitatively pooled through meta-analyses and can only be described in the discussion. In terms of histological variables, a small RCT,17 with 18 patients who completed the study, reported that NAFLD activity score at post-therapy was significantly lower among the IGB-treated compared with the sham-treated arm. On the other hand, there seemed to be no difference between the IGB-treated arm and the sham-treated arm in improving fibrosis. Consistent with this finding, according to another observational study,37 significant improvement of NAFLD activity score was reached in most NAFLD patients treated with IGB (p<0.001). Apart from these, some of the studies assessed non-invasive radiological parameters of NAFLD. A prospective single-arm study27 showed that after 6 months of IGB therapy, the number of patients with severe hepatic steatosis confirmed by abdominal ultrasound decreased from 52% to 4%. Two other clinical studies,32,37 respectively, demonstrated that hepatic fat fraction and fibrosis by magnetic resonance imaging could be significantly alleviated by IGB therapy. Taken together, these histological and radiological findings were consistent with the results of serum markers (HOMA-IR, ALT, AST, and GGT) in our meta-analyses.
To date, no study has looked at the impact of age on insulin resistance amelioration in patients receiving IGB therapy. In our meta-analysis, multivariate linear meta-regression and stratified analyses indicated that participants with advanced age had less change in HOMA-IR after IGB therapy. Several weight-dependent and non-weight-dependent hypotheses may explain this phenomenon. A previously published study reported that advanced age was significantly correlated with less excess weight loss in females after IGB intervention.41 Given that clinically significant weight loss can alleviate insulin resistance,42 age-related differences in insulin resistance outcomes might be partly attributed to the different weight loss during treatment. Additionally, both obesity and aging are linked to and engender insulin resistance.43 Among elderly patients, the effect of aging is strongly amplified and cannot be eliminated by the obesity management tools. Taken together, age might be considered as a predictor of insulin resistance amelioration in patients undergoing IGB therapy.
Comparison with other studies or reviews
In terms of the impact of IGB on liver enzymes, a commendable meta-analysis published in 2016 showed that the use of IGB could decrease ALT (MD=10.02, 95% CI=6.8–13.2),19 which was in line with our findings. When their meta-analysis was published, swallowable IGB had not been widely used and investigated.14 To help clinicians and researchers keep up to date with current evidence, we performed this systematic review including more updated studies. Our stratified analysis revealed that the improvement of ALT, AST, and HOMA-IR with swallowable IGB therapy was no worse than that with endoscopic IGB. Future RCTs are needed to comprehensively compare the efficacy and safety between these two IGBs.
Limitations and strengths
Our systematic review does have some shortcomings. First, although our review included studies of both endoscopic and swallowable IGB, there were still a number of IGB types (such as ReShape Duo Balloon and Obalon Gastric Balloon) not mentioned in the current review due to the lack of relevant clinical research.14 Second, at the time of the preliminary search, we found that most of the clinical studies in this field were of longitudinal observational design. Thus, when formal screening of the search was performed, we defined the patient at baseline, but not the sham-treated group, as comparators. However, this approach ignored the potential for spontaneous remission of the disease.44 Despite these limitations, our systematic review provides the most comprehensive evaluation of the effect of IGB on the serum markers of MAFLD, with low intra-subgroup heterogeneity in stratified analysis, suggesting that the evidence is highly credible. More impressively, our observations demonstrate for the first time that age has an adverse effect on IGB treatment of insulin resistance.
Conclusions and perspectives
IGB therapy has led to improvements in the serum markers of MAFLD, including HOMA-IR, ALT, AST, and GGT. Significant reductions in HOMA-IR and liver biochemical parameters were seen across different methods of balloon implantation and different age/BMI classes. The improvement of insulin resistance and liver biochemistries with swallowable IGB therapy was no worse than that with endoscopic IGB. Furthermore, greater insulin resistance amelioration with IGB therapy was predicted by younger age and the relevant mechanism needs further investigation. Although IGB has the potential to become a multidisciplinary management tool of MAFLD, it cannot be ignored that IGB is a temporary measure. If the patient cannot maintain an active lifestyle after the first balloon is removed, relapse of MAFLD is an expected result. In this regard, IGB combined with other pharmacotherapy or sequential IGB therapy could be a potential solution, and further RCT is warranted.
Supporting information
Supporting Fig. 1
Quality assessment of included RCTs.
(TIF)
Supporting Fig. 2
Funnel plots.
HOMA-IR (A), ALT (C), AST (C), and GGT (D) decreased after IGB treatment and removal.
(TIF)
Supporting Table 1
Reporting checklist for a systematic review with meta-analysis.
(DOCX)
Supporting Table 2
Search strategy.
(DOCX)
Supporting Table 3
NOS quality assessment scale.
(DOCX)
Abbreviations
- ALT:
alanine aminotransferase
- AST:
aspartate aminotransferase
- BMI:
body mass index
- CI:
confidence interval
- EBMT:
endoscopic bariatric and metabolic therapies
- GGT:
gamma-glutamyl transpeptidase
- HOMA-IR:
homeostasis model assessment of insulin resistance
- IGB:
intragastric balloon
- I2:
inconsistency index
- MD:
mean difference
- NAFLD:
nonalcoholic fatty liver disease
- NOS:
Newcastle-Ottawa scale
- MAFLD:
metabolic dysfunction-associated fatty liver disease
- RCT:
randomized controlled trial
Declarations
Data sharing statement
No additional data are available.
Funding
This study was supported by the National Key R&D Program of China (2017YFC0908903), National Natural Science Foundation of China (81873565, 81900507), Shanghai Leading Talent Plan 2017, Innovative Research Team of High-Level Local Universities in Shanghai, and Hospital Funded Clinical Research, Clinical Research Unit, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine (17CSK04).
Conflict of interest
The authors have no conflict of interests related to this publication.
Authors’ contributions
Conception and design (JGF), funding acquisition and supervision (JGF, TYR), collection and assembly of data (ZYZ, JZ), data analysis and interpretation (ZYZ), manuscript writing (ZYZ, JZ, TYR, YWS, RXY, JGF), final approval of manuscript (ZYZ, JZ, TYR, YWS, RXY, JGF).