Clinical Gastroenterology and Hepatology 2019;17:1040–1060

SYSTEMATIC REVIEWS AND META-ANALYSES

Siddharth Singh, Section Editor
Complete Resolution of Nonalcoholic Fatty Liver
Disease After Bariatric Surgery: A Systematic Review
and Meta-analysis
Yung Lee,*,‡ Aristithes G. Doumouras,‡ James Yu,* Karanbir Brar,§ Laura Banfield,k
Scott Gmora,‡ Mehran Anvari,‡ and Dennis Hong‡
*Michael G. DeGroote School of Medicine, ‡Division of General Surgery, Department of Surgery, kHealth Sciences Library,
McMaster University, Hamilton, Ontario, Canada; §Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada

This article has an accompanying continuing medical education activity, also eligible for MOC credit, on page e61. Learning Objective–Upon
completion of this activity, successful learners will be able to demonstrate an increase in, or affirmation of, their clinical knowledge of nonalcoholic
fatty liver disease and identify the role of bariatric surgery in nonalcoholic fatty liver disease treatment.

BACKGROUND & AIMS: Bariatric surgery has been reported to lead to complete resolution of nonalcoholic fatty liver
disease (NAFLD) following the sustained weight loss induced in obese patients. We performed a
systematic review and meta-analysis to evaluate the effects of bariatric surgery on NAFLD in
obese patients.

METHODS: We searched MEDLINE, EMBASE, CENTRAL, and Web of Science databases through May 2018 for
studies that compared liver biopsy results before and after bariatric surgery in obese patients.
Primary outcomes were biopsy-confirmed resolution of NAFLD and NAFLD activity score. Sec-
ondary outcomes were worsening of NAFLD after surgery and liver volume. The Grading of
Recommendations, Assessment, Development, and Evidence approach was conducted to assess
overall quality of evidence.

RESULTS: We analyzed data from 32 cohort studies comprising 3093 biopsy specimens. Bariatric surgery
resulted in a biopsy-confirmed resolution of steatosis in 66% of patients (95% CI, 56%–75%),
inflammation in 50% (95% CI, 35%–64%), ballooning degeneration in 76% (95% CI, 64%–
86%), and fibrosis in 40% (95% CI, 29%–51%). Patients’ mean NAFLD activity score was
reduced significantly after bariatric surgery (mean difference, 2.39; 95% CI, 1.58–3.20; P <
.001). However, bariatric surgery resulted in new or worsening features of NAFLD, such as
fibrosis, in 12% of patients (95% CI, 5%–20%). The overall Grading of Recommendations,
Assessment, Development, and Evidence quality of evidence was very low.

CONCLUSIONS: Through this systematic review and meta-analysis, we found that bariatric surgery leads to
complete resolution of NAFLD in obese patients. However, some patients develop new or
worsened features of NAFLD. Randomized controlled trials are needed to further examine the
therapeutic benefits of bariatric surgery for patients with NAFLD.

Keywords: Metabolic Surgery; Morbid Obesity; Hepatic Histology; NASH.

onalcoholic fatty liver disease (NAFLD) has

N become one of the most common chronic liver
diseases in the world, affecting 20% to 30% of the gen-
eral population in Western countries and costing the US
health care system $32 billion annually.1,2 NAFLD is the
hepatic manifestation of metabolic syndrome and its
prevalence has increased with the obesity epidemic,
affecting up to 90% of morbidly obese patients.3,4 NAFLD
is characterized by hepatic steatosis that can progress to
nonalcoholic steatohepatitis (NASH), characterized by
inflammation and hepatocellular injury. NASH is the
second most common indication for liver transplantation
Abbreviations used in this paper: BMI, body mass index; BPD, bil-
iopancreatic diversion; GRADE, Grading of Recommendations, Assess-
ment, Development, and Evidence; I2, inconsistency; JIB, jejunoileal
bypass; MINORS, Methodological Index for Non-Randomized Studies;
NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohe-
patitis; NAS, nonalcoholic fatty liver disease activity score; RCT, ran-
domized controlled trial; RYGB, Roux-en-Y gastric bypass.
Most current article
© 2019 by the AGA Institute
1542-3565/$36.00
https://doi.org/10.1016/j.cgh.2018.10.017
May 2019
in the United States,5 and affected patients are also at a
higher risk for liver-related death owing to progression
to fibrosis, cirrhosis, and hepatocellular carcinoma.6
Currently, there are no approved therapies to treat
NAFLD, and most treatments involve controlling under-
lying factors of the metabolic syndrome.7 The first-line
treatment is weight loss; however, the 10% weight loss
required to reduce inflammation and fibrosis in patients
with NASH is difficult to maintain.8,9 Pharmacotherapy
for NAFLD includes insulin sensitizers such as pioglita-
zone, antioxidants including vitamin E, and glucagon-like
peptide-1 analogues, which have limited evidence but are
the subject of ongoing trials.7,10–12 Bariatric surgery has
proven to be effective for achieving sustained weight loss
in patients and can reverse risk factors that contribute to
the pathogenesis of NAFLD, including dyslipidemia, in-
sulin resistance, and inflammation, making it a promising
treatment option for NAFLD.13,14
Initial studies raised concerns that rapid weight loss
from bariatric surgery could exacerbate steatohepatitis
or acute liver failure in morbidly obese patients,15,16 but
more recent surgical techniques such as Roux-en-Y
gastric bypass have shown improvements in liver histo-
pathologic scoring after 5 years of follow-up evalua-
tion.17,18 Although the majority of studies have reported
that fibrosis scores improve after bariatric surgery, a
number of studies also have noted development de novo
and worsening of fibrosis in a small subset of their
patients.18–21
Despite the lack of randomized trials, guidelines have
indicated that bariatric surgery reduces liver fat and is
likely to reduce the progression of NAFLD.22,23 However,
a nationwide analysis found that despite an association
between prior bariatric surgery with decreased mortality
in 45,462 morbidly obese patients with NAFLD, the
proportion of NAFLD patients receiving bariatric surgery
decreased from 2004 to 2012.24 Given the considerable
number of studies with larger sample sizes that have
accumulated in recent years, this systematic review and
meta-analysis aims to establish the harms and benefits of
bariatric surgery on histologically confirmed resolution
of NAFLD (steatosis, inflammation, ballooning degener-
ation, and fibrosis), NAFLD activity score (NAS), and
histologic worsening of NAFLD.
Methods
Search Strategy
We searched the following databases covering the
period from database inception through May 2018:
MEDLINE, EMBASE, Web of Science, and Cochrane Cen-
tral Register of Controlled Trials. The search was
designed and conducted by a medical librarian with
input from study investigators (the complete search
strategy is available in Supplementary Table 1). We also
searched the references of published studies and
Resolution of NAFLD After Bariatric Surgery 1041
What You Need to Know
Background
Nonalcoholic fatty liver disease (NAFLD) is a com-
mon chronic liver disease with substantial burden
and no approved treatments. The prevalence of
NAFLD has increased substantially with the obesity
epidemic.
Findings
Our systematic review and meta-analysis found that
weight loss induced by bariatric surgery leads to
biopsy-confirmed resolution of NAFLD and
improvement of the NAFLD Activity Score in a sub-
stantial proportion of patients.
Implications for patient care
If bariatric surgery should be found to be safe and
effective for NAFLD through larger randomized trials,
consideration should be given to treating obese pa-
tients with NAFLD with bariatric surgery.
searched grey literature (eg, conference abstracts, pre-
sentations, proceedings, unpublished trial data) manu-
ally to ensure that relevant articles were not missed. We
did not discriminate full texts by language. This sys-
tematic review and meta-analysis is reported in accor-
dance with the Preferred Reporting items for Systematic
Reviews and Meta-Analyses.25 The protocol of this study
was registered before commencement in the Prospective
Register of Systematic Reviews (CRD42018093012).
Eligibility Criteria and Data Abstraction
Articles were eligible for inclusion if the studies
examined the effect of bariatric surgery on NAFLD. We
included both single-arm studies (effect of bariatric
surgery on NAFLD status before and after surgery) or
double-arm studies (bariatric surgery vs placebo or
medical therapy). However, there were no double-arm
studies identified in the current literature. Exclusion
criteria were as follows: (1) case-series/reports, expert
opinions, basic science, and review articles; (2)
nonhuman studies; (3) studies with fewer than 10
eligible patients; and (4) patients with cirrhosis or a
history of liver transplants.
At least 2 reviewers independently screened the
searched titles, abstracts, and full texts following the
inclusion and exclusion criteria. Reviewers were not
blinded to authors, institution, or the journal in which
the article was published. Discrepancies that occurred at
the title and abstract screening stages were resolved by
automatic inclusion to ensure that all relevant articles
were not missed. Discrepancies at the full-text or data
abstraction stage were resolved by consensus between 2
reviewers, and if disagreement persisted, a third
reviewer was consulted. Two reviewers independently
1042 Lee et al
conducted data abstraction onto a standardized spread-
sheet designed a priori. The following data were
abstracted from included studies: study characteristics
(author, country, year of publication, study design, and
funding source), patient demographics (mean age at time
of surgery, percentage of female patients, number of
patients included, comorbidities, and mean body mass
index [BMI] before and after surgery), follow-up time
points, type of bariatric surgery, biopsy description (type
of histopathologic grading system, biopsy method,
reason for follow-up biopsy, number of paired liver bi-
opsy specimens for histologic analysis before and after
surgery), and outcomes.
Outcomes Assessed
Articles included in the systematic review and meta-
analysis reported at least 1 of our 4 outcomes of inter-
est before and after surgery, which included the
following: (1) histologic (biopsy) features of NAFLD such
as steatosis, inflammation, ballooning degeneration, and
fibrosis; (2) NAS, which is a sum of individual NAFLD
biopsy scores; (3) worsening of NAFLD after bariatric
surgery; and (4) change in liver volume outcomes after
surgery. The primary outcome of our review was biopsy-
proven complete resolution of NAFLD. Complete resolu-
tion of NAFLD was defined as the absence of pathologic
biopsy results after bariatric surgery. If a histopathologic
grading system had a scale of 0 to 4, then 0 was
considered to be complete resolution and 1 to 4 were
categorized as disease. Dichotomization of the grading
system allowed us to pool and analyze how many pa-
tients with NAFLD had complete resolution of NAFLD
after bariatric surgery across studies that used different
histopathologic grading systems. In several studies, the
investigators only reported improvement of NAFLD or
did not report complete resolution and improvement of
NAFLD separately. We did not include these outcomes
because our review strictly focused on the effect of
bariatric surgery on complete resolution of NAFLD,
without improvement data inflating the results of the
overall effect estimate. We also assessed the NAS score,
which combines steatosis, lobular inflammation, and
hepatocyte ballooning scores to provide a more accurate
and complete picture of liver disease than individual
histopathologic results.26,27
Statistical Analysis
The pooled proportion of patients with complete
resolution of steatosis, inflammation, ballooning degen-
eration, and fibrosis after bariatric surgery was calcu-
lated using the Freeman–Tukey double arcsine
transformation of proportions. The DerSimonian and
Laird random-effects meta-analysis of proportions was
used to generate the overall effect size of each outcome.
Clinical Gastroenterology and Hepatology Vol. 17, No. 6
The same method was applied for the histologic wors-
ening of NAFLD-related outcomes (eg, worsening or new
development of fibrosis) after bariatric surgery. We also
performed pairwise meta-analyses using a DerSimonian
and Laird random-effects model for continuous variables
such as NAS before and after surgery. Pooled effect es-
timates were obtained by calculating the mean difference
in outcomes along with their respective 95% CIs to
confirm the effect size estimation. In addition, for studies
that did not report change in SD, a mean between the
baseline SD and the end point SD was calculated.
Assessment of heterogeneity was completed using the
inconsistency (I2) statistic. We considered I2 higher than
50% to represent considerable heterogeneity.28 Publi-
cation bias was assessed using a funnel plot. All statis-
tical analyses and meta-analyses were performed on
STATA, version 14 (StataCorp, College, TX) and Cochrane
Review Manager 5.3 (London, UK), with a level of sig-
nificance set at a P values less than .05. In addition, we
performed subgroup analyses based on different types of
bariatric surgery, method of biopsy, and type of histo-
pathologic grading systems, study designs including
prospective vs retrospective studies, short term (<1 y)
vs long-term follow-up evaluation (>1 y), and baseline
NAS. A sensitivity analysis was conducted by identifying
potential outliers that could contribute to heterogeneity.
Risk of Bias Assessment and Certainty
of Evidence
Risk of bias for individual studies was assessed using
the Methodological Index for Non-Randomized Studies
(MINORS) tool.29 Quality of evidence for estimates
derived from meta-analyses were assessed by Grading of
Recommendations, Assessment, Development and Eval-
uation (GRADE).30,31
Results
Study Characteristics
From 1695 potentially relevant citations received
from electronic databases and searches from reference
lists, 32 studies met the inclusion criteria (15 retro-
spective and 17 prospective cohort studies).13,15–21,32–52
There were no randomized controlled trials identified.
All studies were single-arm studies examining the effect
of bariatric surgery on NAFLD before and after surgery
with no comparators. Figure 1 shows a Preferred
Reporting Items for Systematic Reviews and Meta-
Analyses flow diagram of the study selection process,
and study characteristics are reported in detail in
Table 1. Of the 32 studies included, there was a total of
3093 liver biopsies at baseline and 2649 biopsies at
follow-up evaluation (85.65% follow-up rate). Included
studies were conducted between 1995 and 2018, with a
May 2019
Figure 1. Preferred
Reporting Items for Sys-
tematic Reviews and
Meta-Analyses diagram.
Transparent reporting of
systematic reviews and
meta-analysis flow dia-
gram outlining the search
strategy results from initial
search to included studies.
median follow-up period of 15 months (range, 3–55 mo)
across all outcome measurements. The weighted mean
age of the patients at the time of surgery was 41.38 
4.18 years. The weighted mean BMI at baseline was
48.68  2.92 kg/m2 and 34.2  3.53 kg/m2 at follow-up
evaluation, with an absolute percentage reduction of
24.98% after surgery. For studies with multiple time
points, we analyzed the time point closest to 1 year. Past
studies have shown that in long-term prospective eval-
uations, histologic characteristics of NAFLD and NAS
improved primarily within the first year after bariatric
surgery and persisted for up to 5 years.17,18 Therefore, to
allow sufficient time for liver function to improve after
surgery, we chose 1 year as the follow-up time for our
analyses.
Biopsy method and its subsequent histopathologic
grading system varied between studies. Histopathologic
grading systems used in the studies were Brunt, Dixon,
Lieber, Brazilian Pathology Society, and the NASH Clinical
Research Network grading system (Kleiner) (Table 2).
However, there were many studies that did not clarify
Resolution of NAFLD After Bariatric Surgery 1043
the type of grading system or had customized their own
method of categorizing the liver biopsy results. In addi-
tion, studies used a mix of needle and wedge methods for
collecting liver biopsy specimens. The initial number of
biopsy specimens for each study and relevant histologic
features are reported in Supplementary Table 2. Most
studies did not report individual patient data before and
after bariatric surgery and instead chose to provide the
number of patients with a specific numeric score for each
histologic feature before and after surgery. Moreover, all
included studies did not explicitly report the number of
patients with NAFLD or NASH separately before and af-
ter surgery. Bariatric surgeries conducted in the included
studies were laparoscopic Roux-en-Y gastric bypass (10
studies), open Roux-en-Y gastric bypass (open RYGB; 14
studies), laparoscopic adjustable gastric banding (9
studies), laparoscopic sleeve gastrectomy (4 studies),
sleeve gastrectomy (4 studies), gastroplasty (3 studies),
bilio-intestinal bypass (3 studies), biliopancreatic diver-
sion (BPD; 1 study), jejunoileal bypass (JIB; 1 study), and
gastric balloon (1 study).
Table 1. Study Characteristics

1044 Lee et al
Center type, Surgery Study Follow-up Pre-BMI, Post-BMI, Reduction in
Study country type design Study population details duration means  SD means  SD BMI, % Funding

Silverman, Multi, United RYGB R Inclusion criteria: weigh 200% or 45 kg more 18.4 mo 47 33.5 28.72 Public
199519 States than ideal body weight, no success with
conservative weight loss therapies
Exclusion criteria: N/A
Mean age at surgery, 39 y (17–56 y)a
Female: 95.6%
Luyckx, Single, Belgium GP R Inclusion criteria: severe obesity, resistance to 27 mo 43.9  8.3 31.7  4.1 27.79 Public
199815 dietary and pharmacologic treatments, or
associated comorbid conditions
Exclusion criteria: alcohol abusers (>20
glasses/wk), drug addicts
Mean age at surgery: 36 y (11 y)
Female: 85.5%
Dixon, Single, Australia LAGB R Inclusion criteria: BMI >35 kg/m2, significant 25.6 mo 47  10.6 34  5.5 27.66 Private
200465 medical, physical, or psychosocial
disabilities
Exclusion criteria: alcoholism with >200 g of
alcohol/wk, evidence of hepatitis B or C,
taking known hepatotoxic medications,
other specific liver disease
Mean age at surgery: 43 y (10.3 y)
Female: 69.4%
Kral, 200450 Single, Canada BPD R Inclusion criteria: patients undergoing 41 mo 47  8.4 31  7.9 34.04 N/A
biliopancreatic diversion for severe obesity
from 1984 to 1994 at Laval Hospital,

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Quebec, Canada
Exclusion criteria: use of hepatotoxic
medications, hepatitis, prior weight-loss
surgery, consumption of >100 g/wk of
alcohol
Mean age at surgery: 36.9 y (9 y)
Female: 80.8%
Clark, 200551 Single, United RYGB R Inclusion criteria: BMI >40 kg/m2 or >35 kg/ 305 d 51.1  6.1 32.9  5.1 35.62 Public
States m2 with comorbidities, prior unsuccessful
attempts of weight loss
Exclusion criteria: N/A
Mean age at surgery: 43.9 y (8.1 y)
Female: 50%
 May 2019
Mattar, Single, United LRYGB, LAGB, R Inclusion criteria: patients with increased liver 15 mo 56  11 39  10 30.36 Public
200513 States LSG function tests or gross features of fatty
liver
Exclusion criteria: history of alcoholism, >20 g
alcohol/d, evidence of autoimmune
hepatitis, chronic hepatitis B or C, HIV,
genetic hemochromatosis, a1 antitrypsin
deficiency, Wilson disease, or hepatotoxic
drugs; patients with time between initial
and repeat liver biopsy <3 mo
Mean age at surgery: 49 y (9 y)
Female: 68.6%
Mottin, Single, Brazil RYGB R Inclusion criteria: N/A 1y 46.7  0.88 N/A N/A Public
200552 Exclusion criteria: patients with no histologic
alterations on intraoperative biopsy;
presence of histologic alteration other than
steatosis, alcohol abuse, other hepatic
damage, or insufficient material at biopsy
Mean age at surgery: 35.6 y (1.1 y)
Female: 71.1%
Stratopoulos, Single, Greece GP P Inclusion criteria: morbid obesity, BMI >40 kg/ 18 mo, 35 mo 52.8  1 N/A N/A N/A
200532 m2, at least 2 dietary trials of weight
reduction, or associated comorbid
conditions
Exclusion criteria: type 2 diabetes, average
daily consumption of alcohol >20 g for
women and >30 g for men, evidence of
drug-induced liver disease, hepatitis B and
C, another specific liver disease, heart
failure, organic renal disease, cancer, or
other major disease
Mean age at surgery: N/A

Resolution of NAFLD After Bariatric Surgery 1045

Female: 64.7%
Barker, Single, United RYGB R Inclusion criteria: obese patients who 21.4 mo 47  4.4 29  5.2 38.30 N/A
200633 States underwent RYGB from October 2001 to
September 2003 and had concomitant
intraoperative needle liver biopsies
performed
Exclusion criteria: N/A
Mean age at surgery: 48.6 y (35–58 y)a
Female: 89.45%
Csendes, Single, Chile RYGB P Inclusion criteria: N/A 22 mo 44.3 (37–60)a 28.6 (22.5–37)a 35.44 Public
200634 Exclusion criteria: N/A
Mean age at surgery: 46.2 y (21–65 y)a
Female: 93.75%
 1046 Lee et al
Table 1. Continued

Center type, Surgery Study Follow-up Pre-BMI, Post-BMI, Reduction in

Study country type design Study population details duration means  SD means  SD BMI, % Funding

De Almeida, Single, Brazil RYGB P Inclusion criteria: patients between January 23.5 mo 53.4  8.8 31.1  4.7 41.76 N/A
200635 2001 and December 2002 with a diagnosis
of NASH who had undergone RYGB >1 y
before and had a second liver biopsy; an
international normalized ratio 1.4,
platelet count 80,000/mm3, partial
thromboplastin time 10 s
Exclusion criteria: alcoholic patients (ingestion
of >40 g alcohol/d for men and >20 g/
d for women), drug use; use of
anticoagulant or antiplatelet-aggregation
drugs
Mean age at surgery: 41.5 y (9.1 y)
Female: 87.5%
Mathurin, Single, France BLB, LAGB P Inclusion criteria: BMI >35 kg/m2 with 12 mo 47.1 (46–48.4)a 38.1 (36.6–39.9)a 19.11 Public
200618b comorbidity/comorbidities or BMI >40 kg/
m2 for at least 5 years and resistance to
medical treatment
Exclusion criteria: average daily consumption
of alcohol of 20 g/d for women and 30 g/
d for men, past excessive drinking for >2
years within past 20 years, use of
hepatotoxic drugs, chronic liver diseases,
including hepatitis B/C
Mean age at surgery: 40.6 y (39–42.8 y)a

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Female: 80%
Meinhardt, Single, JIB R Inclusion criteria: morbid obesity, undergone 57.6 mo 52.8  7.5 35.7  7.5 32.4 N/A
200667 Brazil JIB at Hospital Conceição from 1987 to
2002
Exclusion criteria: N/A
Mean age at surgery: 37.9 y (7.6 y)
Female: 84%
Furuya, Single, Brazil RYGB, GP P Inclusion criteria: NAFLD patients BMI >40 2y 51.7  7 32.43  6 37.27 N/A
200737 kg/m2; resistance to classic dietary and
pharmacologic treatments
Exclusion criteria: liver abnormalities such as
alcoholic liver disease (consumption of
>100 g/wk of alcohol), and drug-induced
liver disease
Mean age at surgery: 46.6 y (7.3 y)
Female: 94.4%
 May 2019
Liu, 200738 Single, United LRYGB R Inclusion criteria: morbid obesity, undergone 18 mo 47.7  6.2 29.5  5.6 38.16 Public
States LRYGB at University of Alabama-
Birmingham from 2001 to 2006
Exclusion criteria: previously existing liver
disease
Mean age at surgery: 41.4 y (9 y)a
Female: 84.6%
Mathurin, Single, France BLB, LAGB, P Inclusion criteria: BMI >35 kg/m2 with 1 y, 5 y 50  7.6 39  8.2 22 Public
200936 RYGB comorbidities or BMI >40 kg/m2 for at
least 5 years and resistance to medical
treatment
Exclusion criteria: average daily consumption
of alcohol of 20 g/d for women and 30 g/
d for men, excessive drinking for >2 years
within past 20 years, use of hepatotoxic
drugs, chronic liver diseases, including
hepatitis B/C
Mean age at surgery: 41.5 y (9.6 y)
Female: 77.2%
Moschen, Single, LAGB P Inclusion criteria: BMI >35 kg/m2 with no 6 mo 42.6  0.7 33.8  0.9 20.7 Public
200948 Austria significant medical, physical, or
psychosocial disabilities
Exclusion criteria: average daily consumption
of alcohol 20 g/wk, positive hepatitis B/C,
autoimmune hepatitis, primary biliary
cirrhosis, and other chronic liver diseases
Mean age at surgery: 37.5 y (10.7 y)
Female: 80%
Moretto, Single, Brazil RYGB R Inclusion criteria: N/A N/A 45.4  8.1 29.3  5.8 35.46 Public
201220 Exclusion criteria: N/A
Mean age at surgery: 39.5 y (11.4 y)
Female: 75.6%

Resolution of NAFLD After Bariatric Surgery 1047

Tai, 201239 Single, Taiwan LRYGB R Inclusion criteria: Chinese patients with BMI 1y 43.8  7.5 28.3  4.6 35.39 Public
40 kg/m2 or BMI 35 kg/m2 with
comorbidities
Exclusion criteria: presence of daily alcohol
intake of >20 g/d, chronic hepatitis B/C, use
of hepatotoxic drugs, other known liver
diseases, autoimmune hepatitis, or
malignant diseases
Mean age at surgery: 29.9 y (8.1 y)
Female: 61.9%
 1048 Lee et al
Table 1. Continued

Center type, Surgery Study Follow-up Pre-BMI, Post-BMI, Reduction in

Study country type design Study population details duration means  SD means  SD BMI, % Funding

Vargas, N/A, Spain RYGB P Inclusion criteria: BMI >40 kg/m2 16.3 mo 49.3  4.8 30.9  4.3 37.32 Public
201216 who had significant medical,
physical or psychosocial disabilities were
considered for entry into
the study
Exclusion criteria: history of alcoholism,
consuming more than 200 g/wk of alcohol,
evidence of hepatitis B/C, history of
another specific liver
disease
Mean age at surgery: 45 y (2 y)
Female: 73.1%
Caiazzo, Single, France LAGB, RYGB P Inclusion criteria: obese patients who had the 1 y, 5 y 46.8  6.5, 39.9  6.7 19.88 Public
201417 RYGB or adjustable gastric banding 49.8  8.2 (1 y, AGB), (1 y, AGB),
surgery between December 1996 and 36  6.9 24.70
June 2012 at Lille University Hospital and (5 y, AGB), (5 y, AGB),
had liver biopsies performed 37.5  7.7 27.7
Exclusion criteria: removal of AGB before 5 (1 y, RYGB), (1 y, RYGB),
years in patients undergoing surgery 38.5  8.6 22.69
before 2008 because of complications or (5 y, RYGB) (5 y, RYGB)
insufficient weight loss
Mean age at surgery: 40.3 y (11.4 y), 41.1 y
(11.1 y)
Female: 80% (AGB), 72.7% (RYGB)
Lassailly, Single, France BLB, RYGB, P Inclusion criteria: BMI >40 kg/m2 or >35 kg/m2 12 mo 49.3  8.2 37.4  6.9 24.14 Public

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

201540 LAGB, SG with at least 1 comorbid factor for at least 5
y, age >18 y, Social Security insurance
coverage
Exclusion criteria: excessive drinking >2 y in
past 20 years, long-term use of hepatotoxic
drugs, chronic liver
disease
Mean age at surgery: 46.2 y (10.5 y)
Female: 63.3%
Raj, 201541 Single, India RYGB, SG P Inclusion criteria: initial biopsy specimen with 15 d, 1 mo, N/A N/A N/A Public
any degree of NAFLD and were re- 1.5 mo,
biopsied at 6 mo 3 mo, 6 mo
Exclusion criteria: excessive alcohol use >20
g/d if male or 10 g/d if female, hepatotoxic
medications, chronic liver disease, prior
weight loss surgery, normal liver histologic
findings or findings suggestive of liver
disease other than NAFLD
Mean age at surgery: 44 y (21–70 y)a
Female: 50%
 May 2019
Taitano, Single, United LRYGB, RYGB, R Inclusion criteria: patients who underwent 1 mo 52  10 33  8 36.54 Public
201542 States LAGB bariatric surgery and a subsequent
abdominal surgery from 1999 to 2013 and
had liver biopsies during the 2 procedures
Exclusion criteria: history of alcoholism >60 g/
d of alcohol, evidence of autoimmune
hepatitis, chronic hepatitis B/C, HIV,
hemochromatosis, a1 antitrypsin
deficiency, or Wilson disease
Mean age at surgery: 47 y (12 y)
Female: 83%
Froylich, Single, United LRYGB, SG R Inclusion criteria: patients with NAFLD 1.7 y (RYGB), 51  13.5 34.5  11.3 32.35 Public
201643 States who had both intraoperative and 1.2 y (SG) (RYGB), 72.7 (RYGB), 53.9 (RYGB),
postoperative liver biopsies and  21.1 (SG)  12.6 (SG) 25.86 (SG)
had bariatric surgery between
2005 and 2012
Exclusion criteria: 2 patients were excluded
because of active
hepatitis C and 4 were excluded because
of autoimmune
hepatitis, insufficient tissue
biopsy, biliary cirrhosis, and
biliary cholangitis
Mean age at surgery: 56.2 y (8.6 y), 46.3 y
(11.7 y)
Female: 57% (RYGB), 73% (SG)
Schneck, Single, France LRYGB P Inclusion criteria: met the 1992 NIH >40 mo 42 27.1 35.48 Public
201644 consensus guidelines for bariatric surgery (median,
Exclusion criteria: alcohol abuse, history of 55 mo)
inflammatory disease, cancer with 5 years,
severe pulmonary or cardiac disease,
hepatitis B or C, autoimmune hepatitis

Resolution of NAFLD After Bariatric Surgery 1049

Mean age at surgery: 51 y (35–59 y)a
Female: 100%
Aldoheyan, Single, Saudi SG P Inclusion criteria: age between 18 and 60 y, 3 mo 44.6  7.8 34.2  6.3 23.32 Public
201745 Arabia BMI >30 kg/m2, weight reduction of 10%
before surgery, ultrasound diagnosis of
NAFLD
Exclusion criteria: alcohol intake
>20 g/d for >5, autoimmune hepatitis,
chronic hepatitis B/C, HIV, other specific
liver diseases, cirrhosis, pregnancy, taking
hepatotoxic medications, and failure to
attend follow-up evaluation >1 y
Mean age at surgery: 35 y (8 y)
Female: 66.66%
 1050 Lee et al
Table 1. Continued

Center type, Surgery Study Follow-up Pre-BMI, Post-BMI, Reduction in

Study country type design Study population details duration means  SD means  SD BMI, % Funding

Hedderich, Single, Germany LRYGB, LSG R Inclusion criteria: BMI between 35 and 60 kg/ 6 wk, 12 wk, 24 44.1  5.2 39.8  4.8 (6 9.75 (6 wk), Public
201746 m2 (body weight <200 kg), waist wk wk), 37.1  15.87 (12
circumference <136 cm, age >18 y 4.9 (12 wk), wk), 23.36
Exclusion criteria: contraindications for MRI 33.8  5.6 (24 wk)
such as claustrophobia or implanted (24 wk)
electrical devices
Mean age at surgery: 41.42 y (12.54 y)
Female: 78.9%
Manco, Multi, Italy/ LSG, gastric P Inclusion criteria: age 13–17 y, BMI 35 kg/ 1 mo, 3 mo, 6 48.56  4.15 38.53  3.51 20.65 Public
201747 Germany balloon m2, biopsy-proven NAFLD, failure to mo, 12 mo
achieve 10% weight loss using lifestyle
intervention alone
Exclusion criteria: any endocrine or systemic
disease uN/Aelated to obesity, severe
gastroesophageal reflux disease or
esophagitis, large sliding hiatal hernia, or
paraesophageal hernia type III, psychiatric
disorder, previous gastrointestinal surgery,
use of recreational drugs or alcohol abuse
>140 g/wk
Mean age at surgery: 16.71 y (1.44 y)
Female: 65%
Parker, Multi, United LRYGB P Inclusion criteria: BMI 40 kg/m2 or >35 kg/ 487 d 48  8 32.4  4.2 32.5 Public
201749 States m2 with comorbidities, failure of
nonsurgical treatment for morbid obesity

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Exclusion criteria: other specific liver disease
(chronic viral hepatitis, autoimmune
hepatitis) uN/Aelated to obesity, end-stage
liver disease, known iodine sensitivity or
allergy
Mean age at surgery: 46 y (11 y)
Female: 69%
Luo, 201853 Single, United LRYGB, LSG, P Inclusion criteria: baseline BMI >35 kg/m2, 1 mo, 3 mo, 6 45.3  5.9 34.4  5.1 24.06 Public
States LAGB willingness to participate in follow-up mo
evaluation
Exclusion criteria: contraindications to MRI,
evidence of liver diseases other than
NAFLD
Mean age at surgery: 50.9 y (10.8 y)
Female: 85.7%
May 2019 Resolution of NAFLD After Bariatric Surgery 1051

Effects on Complete Resolution of

laparoscopic Roux-en-Y gastric bypass; LSG, laparoscopic sleeve gastrectomy; MRI, magnetic resonance imaging; NAFLD, nonalcoholic fatty liver disease; N/A, data not available; NIH, National Institutes of Health; P,
BLB, bilio-intestinal bypass; BMI, body mass index; BPD, biliopancreatic diversion; GP, gastroplasty; HIV, human immunodeficiency virus; JIB, jejunoileal bypass; LAGB, laparoscopic adjustable gastric banding; LRYGB,
Public

Nonalcoholic Fatty Liver Disease,

Nonalcoholic Fatty Liver Disease Activity
Score, and Liver Volume
28.29

From the 32 studies included, 25 studies reported

steatosis (n ¼ 1329), 21 reported inflammation (n ¼
657), 15 reported ballooning degeneration (n ¼ 320),
and 22 reported fibrosis (n ¼ 619) at baseline. A meta-
analysis of proportions showed a complete resolution
34.53

of steatosis in 66% of patients (95% CI, 56%–75%),

inflammation in 50% of patients (95% CI, 35%–64%),
ballooning degeneration in 76% of patients (95% CI,
64%–86%), and fibrosis in 40% of patients (95% CI,
29%–51%) (Figure 2). Similar to individual biopsy re-
sults, bariatric surgery resulted in a significant decrease
48.15

in NAS compared with baseline (mean difference, 2.39;

95% CI, 1.58–3.20; P < .001; 11 studies) (Figure 3).
Heterogeneity was high across all outcomes, ranging
from I2 of 77.15% to 99%. Changes in liver volume were
reported in only 2 studies. Both studies measured liver
12 mo

volume by magnetic resonance imaging and showed

significant reductions in liver volume 6 months after
bariatric surgery.46,53
mellitus, patients needed to be on a stable
alcohol consumption <20 g/d; if known to
Inclusion criteria: BMI >40 kg/m2 or >35 kg/

etiology, medication known to precipitate

Histologic Worsening of Nonalcoholic Fatty

have hyperlipidemia or type 2 diabetes
m2 with comorbidities, age 18 y and

Liver Disease
inflammatory drugs, type 1 diabetes,
Exclusion criteria: liver disease of other

smoking, pregnancy, breastfeeding

steatohepatitis, nonsteroidal anti-

Histologic worsening in regard to NAFLD-relevant

outcomes was reported in a limited number of studies.
Mean age at surgery: 47.2 y

Nineteen studies reported histologic worsening after the

surgery and reported worsening or new development of
NAFLD characteristics such as fibrosis (12 studies),
prospective; R, retrospective; RYGB, open Roux-en-Y gastric bypass; SG, sleeve gastrectomy.

steatosis (2 studies), and inflammation (4 studies) after

drug regimen

Female: 76.2%

surgery. The development or worsening of NAFLD

occurred in 12% of patients (95% CI, 5%–20%)
(Supplementary Figure 1). Other adverse effects,
including postoperative complications, are reported in
Supplementary Table 3.
P

Subgroup Analyses

The majority of bariatric procedures conducted

across all meta-analyzed outcomes were RYGB or lapa-
roscopic Roux-en-Y gastric bypass. Other procedures
LRYGB

Overlapping population with Mathurin 2009.18

such as laparoscopic adjustable gastric banding, laparo-

scopic sleeve gastrectomy, sleeve gastrectomy, gastro-
plasty, bilio-intestinal bypass, BPD, and JIB each were
Single, Canada

reported in fewer than 5 studies across all outcomes.

When more than 1 bariatric procedure was conducted,
studies often did not report separate outcomes. There-
fore, we only conducted the subgroup analysis for
studies that solely conducted RYGB. Complete resolution
Range shown.

was higher in proportion for RYGB across all histologic

Schwenger,
201821

features compared with combined analyses (steatosis,

80%; range, 66%–91%; inflammation, 57%; range, 29%–
83%; ballooning, 80%; range, 65%–91%; and fibrosis,
b
a
Table 2. Biopsy Characteristics

1052 Lee et al
Type of biopsy Type of Reasons for Follow-up
Histopathologic at time of biopsy at Reasons for Baseline biopsy Follow-up loss to follow-up time
Study grading system surgery follow-up follow-up biopsies specimens, n biopsy, n biopsy point, mo

Silverman, 1995 N/A N N During surgical procedure for 91 91 - 18.4 (2–61)

various conditions
Luyckx, 1998 N/A W W Second surgical procedure for 69 69 - 27  15
staple line disruption,
stenosis, pouch
enlargement, slipping
stomach, cholecystectomy
Dixon, 2004 Dixon N N Revisional surgery for 36 36 - 25.6 (9–51)
prolapse, band slippage,
band explanation,
cholecystectomy, or
prospective (study
protocol)
Kral, 2004 Lieber W W Reoperation for dissatisfaction 104 104 - 41 (6–111)
with weight loss, diarrhea,
malabsorption, other
complications unrelated to
liver
Clark, 2005 Brunt W W Repair of incisional hernia 16 16 - 10 (5.8–14.5)
Mattar, 2005 Brunt N N During follow-up surgery 70 70 - 15  9
Mottin, 2005 N/A W and N N Prospective (study protocol) 90 90 - 12
Stratopoulos, 2005 Brunt W N Prospective (study protocol) 51 51 - 18  9.6
Barker, 2006 Dixon N N Prospective (study protocol) 19 19 - 21.4 (13.3–31.7)
Csendes, 2006 Brunt W W Incisional hernia surgery 16 16 - 22 (9–33)

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

De Almeida, 2006 Modified Brunt W and N N Incisional hernioplasty, 16 16 - 23.5 (12.8–36.6)
cholecystectomy, and
prospective (study
protocol)
Mathurin, 2006b Brunt N/A N/A Prospective (study protocol) 185 121 14 did not have 12
bariatric surgery,
11 clinicians
forgot to
conduct biopsy,
3 patients
refused, 3 LTFU,
2
contraindicated,
9 sample too
small, 22 lost to
a different study
Meinhardt, 2006 Brunt W and N W and N During surgery for various 30 30 - 57.6  48
conditions
 May 2019
Furuya, 2007 NASH CRN W N Prospective (study protocol) 18 18 - 24
Liu, 2007 Dixon N N Surgery for internal hernias, 39 39 - 18 (6–41)
cholecystectomy,
abdominal pain, and
miscellaneous reasons
Mathurin, 2009 Brunt N/A N/A Prospective (study protocol) 362 267 51 included in a 12
substudy
evaluating QOL,
2 had cirrhosis,
18 refused
biopsy, 5 LTFU,
12 insufficient
size, 11
forgotten by
clinicians, 3
contraindicated,
7 early surgical
complication, 5
others
Moschen, 2009 Brunt N N Prospective (study protocol) 30 18 None given 6
Moretto, 2012 Brazilian Pathology N N Prospective (study protocol) 78 78 - N/A
Society
Tai, 2012 NASH CRN W N N/A 21 21 - 12
Vargas, 2012 NASH CRN N N Prospective (study protocol) 26 26 - 16.3 (12–22)
Caiazzo, 2014 NASH CRN N/A N/A Prospective (study protocol) 1201 1164 9 AGB removal, 4 12
deaths, 24
unclear
Lassailly, 2015 Brunt, NASH CRN, N N Prospective (study protocol) 109 82 5 did not reach 1- 12
Metavir year follow-up at
time of analysis,
2 deaths, 11
refused, 3 LTFU,

Resolution of NAFLD After Bariatric Surgery 1053

3
contraindicated,
2 early surgical
complication, 1
forgotten by
clinicians
Raj, 2015 NASH CRN N N Prospective (study protocol) 88 30 35 declined, 23 6 (6–8)
LTFU
Taitano, 2015 Brunt N N Subsequent abdominal 160 160 - 1 (1–111)
surgeries
Froylich, 2016 Brunt N/A N/A N/A 25 25 - 20.4  8.4
(RYGB), 14.4
 7.2 (SG)
Schneck, 2016 NASH CRN W N Prospective (study protocol) 9 9 - >40 (median, 55)a
1054 Lee et al Clinical Gastroenterology and Hepatology Vol. 17, No. 6

51%; range, 39%–63%) (Supplementary Figure 2).

CRN, Clinical Research Network; LTFU, loss to follow-up evaluation; MRI, magnetic resonance imaging; N, needle; N/A, not available; NASH, nonalcoholic steatohepatitis; O, other; QOL, quality of life; RYGB, Roux-en-Y gastric
Follow-up Moreover, the range of heterogeneity decreased to I2 of
point, mo

16  2.9
time 58.80% to 94.85%. In addition, liver side effects

5.5
12

12
3

6
decreased to 8% (range, 2%–15%) for RYGB
(Supplementary Figure 3). Subgroup analyses were
conducted based on the histopathologic grading systems
(Brunt, Dixon, and NASH Clinical Research Network),
loss to follow-up

LTFU, 4 health
procedure, 10
methods of biopsy (needle vs wedge), durations of

pregnant, 58
refused, 26
25 had sleeve
Reasons for

issues, 1
None given

unclear
follow-up evaluation (short vs long term), study designs
biopsy

(prospective studies that performed biopsies per proto-

-

-
-

col vs retrospective studies that used opportunity bi-

opsies), and NAS score before surgery (low NAS, score of
0–4; high NAS, score >4). According to the subgroup
analyses, steatosis resolved in as high as 89% and in as
Follow-up
biopsy, n

low as 51% of patients, lobular inflammation resolved in

27

20
15

42
19

49

as high as 70% and in as low as 40% of patients,

ballooning degeneration resolved in as high as 93% and
in as low as 66% of patients, and fibrosis resolved in as
high as 51% and in as low as 27% of patients. Our an-
Baseline biopsy

alyses did not show any substantial difference compared

specimens, n

with the original pooled effect estimates. A comprehen-

106

166
27

20
19

87

sive table of subgroup analyses and results are available

in Table 3.
Upon visual examination of the forest plots, the sub-
stantial heterogeneity present across all outcomes was
owing to more than 1 study. Even after removing outlier
Prospective (study protocol)

Prospective (study protocol)

Prospective (study protocol)

Prospective (study protocol)

studies that had comparably lower or higher values than

N/A (MRI for liver volume)

N/A (MRI for liver volume)

follow-up biopsies

other studies, heterogeneity did not decrease to less than

Reasons for

50% in any of the meta-analyzed outcomes. Therefore,

we did not conduct a sensitivity analysis by removing
specific data points because it likely would have lead to
biased results.54,55

Quality of Evidence
The mean MINORS score of included studies was
follow-up
biopsy at
Type of

12.40  1.08, which indicates a fair quality of evidence

N/A

N/A
N
N

N
N

for nonrandomized studies.29 There was a substantial

level of agreement among quality assessment scores
using the MINORS criteria (intraclass correlation coeffi-
Type of biopsy

cient, 0.98; 95% CI, 0.98–0.99). A comprehensive list of

at time of
surgery

MINORS for included studies are available in

N/A

N/A
W
N

N
N

Supplementary Table 4. In brief, all 32 studies were

observational and had a clearly stated objective. Most of
the studies included consecutive patients (29 of 32
studies), had an established protocol before the study
Histopathologic
grading system

Overlapping population with Mathurin 2009.

(20 of 32 studies), used unbiased assessment of the

study end point such as blind evaluation of outcomes (29
NASH CRN
NASH CRN
NASH CRN

of 32 studies), and had less than 5% to 10% of loss to

follow-up evaluation (29 of 32 studies). The mean
Brunt
Brunt
N/A

follow-up period was longer than 12 months in 29 of 32

of studies. However, studies lacked a prospective calcu-
Table 2. Continued

lation of study size (2 of 32 studies). The GRADE quality-

Schewenger, 2018

bypass; W, wedge.
Aldoheyan, 2017
Hedderich, 2017

Median is shown.

of-evidence profile is summarized in Supplementary

Manco, 2017
Study

Parker, 2017

Tables 5 and 6 in detail. Because of high heterogeneity

Luo, 2018

present in all outcomes, evidence was rated down for

inconsistency. Despite the large magnitude of effect, low
risk of bias, and adequate event rate and sample size in
b
a
May 2019 Resolution of NAFLD After Bariatric Surgery 1055

Figure 2. Proportion meta-analysis forest plot of biopsy-proven complete resolution of NAFLD features. (A) Steatosis, (B)
inflammation/steatohepatitis, (C) ballooning degeneration, and (D) fibrosis. ES, effect size.

most outcomes, the certainty of evidence was not Thus, further high-quality research is very likely to have
upgraded because of the major concerns with inconsis- an important impact on our confidence in the estimate of
tency.56 Overall, there was a very low certainty of evi- effect and is likely to change the estimate. Symmetry
dence, suggesting the effect of estimate is uncertain. shown in our funnel plot suggests that there is a low

Figure 3. Random effects

meta-analysis forest plot
of NAS before and after
surgery. IV, inverse vari-
ance; Sx, surgery.
1056 Lee et al Clinical Gastroenterology and Hepatology Vol. 17, No. 6

possibility of publication bias, which might mean that

<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
there is a low number of unpublished negative studies57

P
(Supplementary Figure 4).
mean difference

(1.58–3.20)
(1.43–3.07)
(0.94–4.13)
(1.76–2.25)
(1.49–3.46)
(1.43–3.41)
(1.92–3.43)

(0.81–3.61)

(1.04–3.86)
(1.25–2.21)
(2.13–4.21)
NAS score,

Discussion
(95% CI)

N/A

N/A
This comprehensive systematic review investigated
2.39
2.25
2.54
2.01
2.48
2.42
2.67

2.21

2.45
1.73
3.17
the effects of bariatric surgery on biopsy-confirmed
NAFLD. Bariatric surgery leads to a complete resolution
of steatosis, inflammation, ballooning, and fibrosis in

CRN, Clinical Research Network; N/A, <2 studies according to this criterion performed a meta-analysis; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis.
(5%–20%)
(2%–15%)
(5%–28%)
(0%–34%)
(7%–22%)
(2%–15%)
(4%–22%)
(2%–20%)
(2%–28%)
(7%–27%)
(0%–21%)
(4%–21%)
(4%–22%)
66% (95% CI, 56%–75%), 50% (95% CI, 35%–64%),
% (95% CI)
Worsening
of NAFLD,

76% (95% CI, 64%–86%), and 40% (95% CI, 29%–51%)

of patients, respectively. Hepatic fibrosis has long been
known to play an important role in the progression
12%

15%

14%

12%

12%
16%

11%
11%
7%

9%

8%

9%

5%

to advanced liver disease, and the effects of bariatric

surgery on fibrosis has been unclear.20 Importantly,
this systematic review found that fibrosis completely
(29%–51%)
(27%–58%)
(22%–54%)
(18%–46%)
(30%–63%)
(39%–63%)
(30%–52%)
(17%–59%)
(18%–36%)
(27%–53%)
(21%–79%)
(23%–67%)
(0%–94%)
% (95% CI)
Resolution
of fibrosis,

resolves in 40% of patients. Complementing the indi-

vidual histologic improvements of NAFLD status, this
review also shows a reduction in NAS in patients after
40%
42%
37%
31%
46%
51%
41%
37%
27%
40%
50%
44%
38%

bariatric surgery (mean difference, 2.39; 95% CI,

1.58–3.20; P < .001).
Previous studies have variably explored the effect of
(72%–100%)

bariatric surgery on NAFLD. A 2008 meta-analysis of 15

(64%–86%)
(61%–88%)
(51%–93%)
(41%–87%)
(69%–91%)
(65%–91%)
(63%–87%)
(53%–81%)
(40%–88%)
(76%–96%)
(62%–91%)
(60%–91%)
degeneration,
Resolution of

% (95% CI)
ballooning

cohort studies by Mummadi et al58 found an improve-

ment or resolution in steatosis in 91.6% of patients,
steatohepatitis in 81.3%, and fibrosis in 65.5% after
76%
76%
75%
66%
81%
80%
76%
68%
66%
87%
78%
77%
93%

bariatric surgery. However, resolution and improvement

of characteristics were reported together, inflating the
true benefit experienced by patients. Although there are
Resolution of lobular

still no randomized controlled trials (RCTs) 10 years

(35%–64%)
(39%–73%)
(17%–64%)
(30%–70%)
(32%–75%)
(29%–83%)
(36%–66%)
(30%–86%)
(28%–61%)

(39%–83%)
(57%–83%)
(0%–100%)
inflammation,

(6%–94%)

later, our review of 32 studies provides further evidence

% (95% CI)
Table 3. Results of Subgroup Analyses for Resolution of NAFLD and NAS

that bariatric surgery is effective by focusing on the

complete resolution of NAFLD. Furthermore, although
Mummadi et al58 claimed that “the risk of progression of
50%
57%
40%
50%
54%
57%
51%
60%
44%
41%
62%
70%
41%

inflammatory changes and fibrosis seems to be minimal,”

our review found a lower resolution of fibrosis and a
12% chance of worsening NAFLD after bariatric surgery.
A prospective cohort study of 381 patients by Mathurin
(53%–100%)

(45%–100%)
(56%–75%)
(53%–80%)
(48%–79%)
(54%–78%)
(50%–79%)
(66%–91%)
(56%–78%)
(46%–77%)
(41%–62%)

(74%–89%)
(69%–88%)

et al18 found an improvement in steatosis, ballooning,

of steatosis,
% (95% CI)
Resolution

and overall NAS, with a significant reduction in the

percentage of NASH patients at 5 years compared with
before surgery. However, 5 years after surgery, there
66%
67%
64%
67%
65%
80%
67%
62%
51%
89%
82%
79%
84%

was significant worsening of fibrosis in 19.8% of patients

for unknown reasons.18 Our finding that NAFLD wors-
ened or developed after bariatric surgery in 12% of pa-
Short-term follow-up evaluation (<1 y)
Long term follow-up evaluation (>1 y)

tients pooled from 19 studies confirms the existence of

liver side effects such as fibrosis. A Cochrane systematic
NASH CRN grading system only
Low baseline NAS (score, 0–4)
High baseline NAS (score, >4)

review by Chavez-Tapia et al59 found 21 cohort studies

Subgroup Name

Dixon grading system only

Retrospective studies only

Brunt grading system only

in which improvement of steatosis or inflammation were

Prospective studies only

reported after bariatric surgery. Although the review did

Wedge biopsy only
Needle biopsy only

not include a meta-analysis or examine complete reso-

Original analysis

lution of histologic characteristics, the investigators did

identify 4 studies describing some deterioration in the
RYGB only

degree of fibrosis.18,32,50,59 A 2015 meta-analysis by

Bower et al60 associated bariatric surgery with a reduc-
tion in the incidence of steatosis, hepatocyte ballooning,
May 2019
lobular inflammation, fibrosis, and reductions in liver
enzyme levels. However, the use of liver enzyme levels
correlates poorly to histologic findings, and are not
reliable in identifying NAFLD and NASH.61 Nearly 80% of
NAFLD patients have normal-range alanine aminotrans-
ferase levels, and alanine aminotransferase level typi-
cally decreases as fibrosis progresses to cirrhosis.62,63
Liver biopsies are the gold standard for diagnosing
NASH and assessing NAFLD, and it is for this reason that
the present review primarily examines histologic char-
acteristics after bariatric surgery.61
This study showed a substantial effect of bariatric
surgery on the resolution of histopathologic features of
NAFLD. Importantly, this study better contextualizes the
benefit of bariatric surgery compared with previous
studies because it reports complete resolution rather
than including improvements, which may or may not be
clinically relevant. Accordingly, this study provides a
conservative estimate of the proportion of patients with
NAFLD who would benefit from bariatric surgery. This
study supports the current guidelines,22,64 which state
that NAFLD as a comorbidity should prompt bariatric
surgery in patients with a BMI of 35 to 40 kg/m2. Despite
the overall favorable histopathologic outcomes, among
19 studies that reported histologic worsening of NAFLD,
12% of patients developed de novo or worsened NAFLD
after bariatric surgery. This figure is reported with less
precision than the benefits because many studies did not
report liver side effects and follow-up biopsies were
conducted in a proportion of studies for revision sur-
geries or unspecified conditions related to the
liver.14,50,65 Occasional worsening of NAFLD also may be
attributed to the type of bariatric procedures undergone
by patients and the extent of malnutrition and malab-
sorption.66 JIB and BPD both have been associated with
higher liver function morbidity, although the 1 study
containing JIB in our review did not report any wors-
ening or adverse events.66,67 Despite this, there appears
to be a clear net benefit to bariatric surgery for patients
with NAFLD. Furthermore, subgroup analyses for RYGB
showed a greater reduction of liver side effects and also a
higher proportion of complete resolution of NAFLD fea-
tures. Thus, our analyses continue to support RYGB as
the gold standard of bariatric procedures, with the most
data to support its safety for the liver.
Future clinical studies should focus on comparative
randomized trials that study the effect of bariatric sur-
gery compared with current medical therapy. This is
important because although there is a small proportion
of patients with progression of disease after bariatric
surgery, this proportion likely is smaller than the num-
ber of patients who progress on current medical therapy.
Therefore, future trials would better elucidate the risks
and benefits of surgery compared with standard medical
therapy. Longer-term studies also are needed to better
understand the benefit of bariatric surgery on liver dis-
ease in the context of longer-term outcomes such as liver
transplantation, cirrhosis, and liver failure. Finally, if
Resolution of NAFLD After Bariatric Surgery 1057
bariatric surgery should be found to be safer and more
effective than current medical therapy, consideration
may be given to treating patients with aggressive NAFLD
and BMIs lower than 35 kg/m2 with bariatric surgery,
which mirrors current recommendations for type II
diabetes.
The key strengths of our review included the evalu-
ation of NAS, the most widely used histologic grading and
staging system for NAFLD,61 to offer a broad view of
NASH and NAFLD improvement that complements data
on individual histologic features. The reporting of com-
plete biopsy-confirmed resolution rather than a com-
bined outcome of improvement and resolution provides
a more conservative estimate of patients who absolutely
benefit from the procedure. This review also meta-
analyzed histologic worsening of NAFLD, focusing on
events relevant to NAFLD outcomes that have been
described frequently in individual studies. Our review
also differs from previous reviews in its rigorous
assessment of included studies, both on the individual
study level for risk of bias using MINORS, and on the
body of evidence level using GRADE. The current
consensus for ideal follow-up evaluation rate is 80% or
greater of any original cohort in both randomized and
prospective studies.68,69 The overall follow-up biopsy
rate was 85.65% (81.6% for prospective studies only),
which indicates that loss to follow-up evaluation was not
substantial enough to cause significant attrition bias and
impact our results.68,69
Our study findings should be interpreted in the light
of the following limitations. First, heterogeneity between
included studies was high for all outcomes. We attemp-
ted to address the heterogeneity by conducting subgroup
and sensitivity analyses, but our results failed to explain
why heterogeneity is present across pooled effect esti-
mates. A potential cause could be the wide range of
follow-up time points across included studies or other
comorbidities at an individual patient level. Our study
also dichotomized the histopathologic grading system of
patients based on individual histologic components; as
such, we were unable to determine whether a specific
patient suffered from only 1 or multiple histologic fea-
tures owing to the lack of individual patient data re-
ported by included studies. In addition, the lack of
individual patient data precludes us from conducting an
individual patient meta-analysis to reliably assess all
outcomes and address confounders (eg, difference in
follow-up time, type of bariatric surgery, biopsy method,
degree of obesity) or conduct a subject-level meta-
regression to evaluate the percentage of BMI loss
correlating with histologic improvement of NAFLD.
Finally, all studies were observational with no compar-
ators, which ultimately led to a very low certainty of
evidence in all outcomes according to GRADE. There are
no RCTs available for this research question to date
because blinded RCTs face ethical issues related to
equipoise and sham surgery.70 Open RCTs still would
face challenges related to the loss to follow-up evaluation
1058 Lee et al
of patients from competing treatments related to car-
diovascular risks and liver comorbidities.59
Conclusions
The current body of evidence shows bariatric surgery
to be beneficial for NAFLD and NASH. Our review shows
that bariatric surgery leads to complete resolution in
histologic features of NAFLD as well as a significant
reduction of NAS in a substantial proportion of patients.
Furthermore, the role of RYGB was cemented further as
the gold standard procedure for the treatment of NAFLD.
However, with the discovery of potential histologic
worsening of NAFLD and adverse events as well as the
certainty of evidence being very low, further high-quality
studies, preferably RCTs, are warranted to recommend
bariatric surgery as a therapy for NAFLD remission.
Supplementary Material
Note: To access the supplementary material accom-
panying this article, visit the online version of Clinical
Gastroenterology and Hepatology at www.cghjournal.org,
and at https://doi.org/10.1016/j.cgh.2018.10.017.
References
1. Younossi ZM, Koenig AB, Abdelatif D, et al. Global epidemi-
ology of nonalcoholic fatty liver disease—meta-analytic
assessment of prevalence, incidence, and outcomes. Hepatol-
ogy 2016;64:73–84.
2. Haskins O. US economic burden of fatty liver disease is $32
billion annually. Bariatric News 2018. http://www.bariatricnews.
net/?q¼news/113158/us-economic-burden-fatty-liver-disease-
32-billion-annually.
3. Lazo M, Clark JM. The epidemiology of nonalcoholic fatty liver
disease: a global perspective. Semin Liver Dis 2008;28:339–350.
4. Machado M, Marques-Vidal P, Cortez-Pinto H. Hepatic histol-
ogy in obese patients undergoing bariatric surgery. J Hepatol
2006;45:600–606.
5. Wong RJ, Aguilar M, Cheung R, et al. Nonalcoholic steatohe-
patitis is the second leading etiology of liver disease among
adults awaiting liver transplantation in the United States.
Gastroenterology 2015;148:547–555.
6. Townsend SA, Newsome PN. Non-alcoholic fatty liver disease in
2016. Br Med Bull 2016;119:143–156.
7. Singh SL, Agarwal A, Mishra S. Treatment options for alcoholic
and non-alcoholic fatty liver. World J Clin Cases 2017;
16:258–306.
8. Promrat K, Kleiner DE, Niemeier HM, et al. Randomized
controlled trial testing the effects of weight loss on nonalcoholic
steatohepatitis. Hepatology 2010;51:121–129.
9. Harrison SA, Fecht W, Brunt EM, et al. Orlistat for overweight
subjects with nonalcoholic steatohepatitis: a randomized, pro-
spective trial. Hepatology 2009;49:80–86.
10. Aithal GP, Thomas JA, Kaye PV, et al. Randomized, placebo-
controlled trial of pioglitazone in nondiabetic subjects with
nonalcoholic steatohepatitis. Gastroenterology 2008;
135:1176–1184.
Clinical Gastroenterology and Hepatology Vol. 17, No. 6
11. Nakade Y, Murotani K, Inoue T, et al. Ezetimibe for the treatment
of non-alcoholic fatty liver disease: a meta-analysis. Hepatol
Res 2017;47:1417–1428.
12. Armstrong MJ, Gaunt P, Aithal GP, et al. Liraglutide safety and
efficacy in patients with non-alcoholic steatohepatitis (LEAN): a
multicentre, double-blind, randomised, placebo-controlled
phase 2 study. Lancet 2016;387:679–690.
13. Mattar SG, Velcu LM, Rabinovitz M, et al. Surgically-induced
weight loss significantly improves nonalcoholic fatty liver dis-
ease and the metabolic syndrome. Ann Surg 2005;242:610–620.
14. Weiner RA. Surgical treatment of non-alcoholic steatohepatitis
and non-alcoholic fatty liver disease. Dig Dis 2010;28:274–279.
15. Luyckx FH, Lefebvre PJ, Scheen AJ. Non-alcoholic steatohe-
patitis: association with obesity and insulin resistance, and in-
fluence of weight loss. Diabetes Metab 2000;26:98–106.
16. Vargas V, Allende H, Lecube A, et al. Surgically induced weight
loss by gastric bypass improves non alcoholic fatty liver disease
in morbid obese patients. World J Hepatol 2012;4:382.
17. Caiazzo R, Lassailly G, Leteurtre E, et al. Roux-en-Y gastric
bypass versus adjustable gastric banding to reduce nonalco-
holic fatty liver disease. Ann Surg 2014;260:893–899.
18. Mathurin P, Hollebecque A, Arnalsteen L, et al. Prospective
study of the long-term effects of bariatric surgery on liver injury
in patients without advanced disease. Gastroenterology 2009;
137:532–540.
19. Silverman EM, Sapala JA, Appelman HD. Regression of hepatic
steatosis in morbidly obese persons after gastric bypass. Am J
Clin Pathol 1995;104:23–31.
20. Moretto M, Kupski C, da Silva VD, et al. Effect of bariatric sur-
gery on liver fibrosis. Obes Surg 2012;22:1044–1049.
21. Schwenger KJP, Fischer SE, Jackson T, et al. In nonalcoholic
fatty liver disease, Roux-en-Y gastric bypass improves liver
histology while persistent disease is associated with lower im-
provements in waist circumference and glycemic control. Surg
Obes Relat Dis 2018;14:1233–1239.
22. Chalasani N, Younossi Z, Lavine JE, et al. The diagnosis and
management of nonalcoholic fatty liver disease: practice guid-
ance from the American Association for the Study of Liver Dis-
eases. Hepatology 2018;67:328–357.
23. De Luca M, Angrisani L, Himpens J, et al. Indications for
surgery for obesity and weight-related diseases: position
statements from the International Federation for the Surgery
of Obesity and Metabolic Disorders (IFSO). Obes Surg 2016;
26:1659–1696.
24. McCarty TR, Echouffo-Tcheugui JB, Lange A, et al. Impact of
bariatric surgery on outcomes of patients with nonalcoholic fatty
liver disease: a nationwide inpatient sample analysis,
2004–2012. Surg Obes Relat Dis 2018;14:74–80.
25. Moher D, Liberati A, Tetzlaff J, et al. The PRISMA Group.
Preferred Reporting Items for Systematic Reviews and Meta-
Analyses: the PRISMA Statement. Ann Intern Med 2009;
151:264–269.
26. Brunt EM, Kleiner DE, Wilson LA, et al. Nonalcoholic fatty liver
disease (NAFLD) activity score and the histopathologic diag-
nosis in NAFLD: distinct clinicopathologic meanings. Hepatol-
ogy 2011;53:810–820.
27. Kleiner D, Brunt E, Van Natta M, et al. Design and validation of a
histological scoring system for nonalcoholic fatty liver disease.
Hepatology 2005;41:1313–1321.
28. Higgins JPT, Thompson SG, Deeks JJ, et al. Measuring incon-
sistency in meta-analyses. BMJ 2003;327:557–560.
May 2019
29. Slim K, Nini E, Forestier D, et al. Methodological index for non-
randomized studies (minors): development and validation of a
new instrument. ANZ J Surg 2003;73:712–716.
30. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging
consensus on rating quality of evidence and strength of rec-
ommendations. Chin J Evidence Based Med 2009;9:8–11.
31. The Cochrane Collaboration. GRADEpro GDT. Cochrane Com-
munity 2017.
32. Stratopoulos C, Papakonstantinou A, Terzis I, et al. Changes
in liver histology accompanying massive weight loss after
gastroplasty for morbid obesity. Obes Surg 2005;15:
1154–1160.
33. Barker KB, Palekar NA, Bowers SP, et al. Non-alcoholic stea-
tohepatitis: effect of Roux-en-Y gastric bypass surgery. Am J
Gastroenterol 2006;101:368–373.
34. Csendes A, Smok G, Burgos AM. Histological findings in the
liver before and after gastric bypass. Obes Surg 2006;
16:607–611.
35. De Almeida SR, Savassi Rocha PR, Dias Sanches M, et al.
Roux-en-Y gastric bypass improves the nonalcoholic steato-
hepatitis (NASH) of morbid obesity. Obes Surg 2006;
16:270–278.
36. Mathurin P, Gonzalez F, Kerdraon O, et al. The evolution of
severe steatosis after bariatric surgery is related to insulin
resistance. Gastroenterology 2006;130:1617–1624.
37. Furuya CK, De Oliveira CPMS, De Mello ES, et al. Effects of
bariatric surgery on nonalcoholic fatty liver disease: preliminary
findings after 2 years. J Gastroenterol Hepatol 2007;22:510–514.
38. Liu X, Lazenby AJ, Clements RH, et al. Resolution of nonalco-
holic steatohepatitis after gastric bypass surgery. Obes Surg
2007;17:486–492.
39. Tai C-M, Huang C-K, Hwang J-C, et al. Improvement of
nonalcoholic fatty liver disease after bariatric surgery in
morbidly obese Chinese patients. Obes Surg 2012;22:
1016–1021.
40. Lassailly G, Caiazzo R, Buob D, et al. Bariatric surgery reduces
features of nonalcoholic steatohepatitis in morbidly obese pa-
tients. Gastroenterology 2015;149:379–388.
41. Praveen Raj P, Gomes RM, Kumar S, et al. The effect of surgi-
cally induced weight loss on nonalcoholic fatty liver disease in
morbidly obese Indians: “nASHOST” prospective observational
trial. Surg Obes Relat Dis 2015;11:1315–1322.
42. Taitano AA, Markow M, Finan JE, et al. Bariatric surgery im-
proves histological features of nonalcoholic fatty liver disease
and liver fibrosis. J Gastrointest Surg 2015;19:429–437.
43. Froylich D, Corcelles R, Daigle C, et al. Effect of Roux-en-Y
gastric bypass and sleeve gastrectomy on nonalcoholic fatty
liver disease: a comparative study. Surg Obes Relat Dis 2016;
12:127–131.
44. Schneck AS, Anty R, Patouraux S, et al. Roux-En Y gastric
bypass results in long-term remission of hepatocyte apoptosis
and hepatic histological features of non-alcoholic steatohepa-
titis. Front Physiol 2016;7:344.
45. Aldoheyan T, Hassanain M, Al-Mulhim A, et al. The effects of
bariatric surgeries on nonalcoholic fatty liver disease. Surg
Endosc 2017;31:1142–1147.
46. Hedderich DM, Hasenberg T, Haneder S, et al. Effects of bar-
iatric surgery on non-alcoholic fatty liver disease: magnetic
resonance imaging is an effective, non-invasive method to
evaluate changes in the liver fat fraction. Obes Surg 2017;
27:1755–1762.
Resolution of NAFLD After Bariatric Surgery 1059
47. Manco M, Mosca A, De Peppo F, et al. The benefit of sleeve
gastrectomy in obese adolescents on nonalcoholic steatohe-
patitis and hepatic fibrosis. J Pediatr 2017;180:31–37.e2.
48. Moschen AR, Molnar C, Wolf AM, et al. Effects of weight loss
induced by bariatric surgery on hepatic adipocytokine expres-
sion. J Hepatol 2009;51:765–777.
49. Parker BM, Wu J, You J, et al. Reversal of fibrosis in patients
with nonalcoholic steatohepatosis after gastric bypass surgery.
BMC Obes 2017;4:32.
50. Kral JG, Thung SN, Biron S, et al. Effects of surgical treatment of
the metabolic syndrome on liver fibrosis and cirrhosis. Surgery
2004;135:48–58.
51. Clark JM, Alkhuraishi ARA, Solga SF, et al. Roux-en-Y gastric
bypass improves liver histology in patients with non-alcoholic
fatty liver disease. Obes Res 2005;13:1180–1186.
52. Mottin CC, Moretto M, Padoin AV, et al. Histological behavior of
hepatic steatosis in morbidly obese patients after weight loss
induced by bariatric surgery. Obes Surg 2005;15:788–793.
53. Luo RB, Suzuki T, Hooker JC, et al. How bariatric surgery affects
liver volume and fat density in NAFLD patients. Surg Endosc
Other Interv Tech 2018;32:1675–1682.
54. John GH. Robust decision trees: removing outliers from data-
bases. First Int Conf Knowl Discov Data Min 1995;174-179.
55. Beniger JR, Barnett V, Lewis T. Outliers in statistical data.
Contemp Sociol 1980;9:560.
56. Guyatt GH, Oxman AD, Sultan S, et al. GRADE guidelines: 9.
Rating up the quality of evidence. J Clin Epidemiol 2011;
64:1311–1316.
57. Sedgwick P, Marston L. How to read a funnel plot in a meta-
analysis. BMJ 2015;351:h4718.
58. Mummadi RR, Kasturi KS, Chennareddygari S, et al. Effect of
bariatric surgery on nonalcoholic fatty liver disease: systematic
review and meta-analysis. Clin Gastroenterol Hepatol 2008;
6:1396–1402.
59. Chavez-Tapia NC, Tellez-Avila FI, Barrientos-Gutierrez T, et al.
Bariatric surgery for non-alcoholic steatohepatitis in obese pa-
tients. Cochrane Database Syst Rev 2010;1:CD007340.
60. Bower G, Toma T, Harling L, et al. Bariatric surgery and
non-alcoholic fatty liver disease: a systematic review of
liver biochemistry and histology. Obes Surg 2015;25:
2280–2289.
61. Dyson JK, Anstee QM, McPherson S. Non-alcoholic fatty liver
disease: a practical approach to diagnosis and staging. Front
Gastroenterol 2014;5:211–218.
62. Browning J, Szczepaniak L, Dobbins R, et al. Prevalence of
hepatic steatosis in an urban population in the United States:
impact of ethnicity. Hepatology 2004;40:1387–1395.
63. Mofrad P, Contos MJ, Haque M, et al. Clinical and histologic
spectrum of nonalcoholic fatty liver disease associated with
normal ALT values. Hepatology 2003;37:1286–1292.
64. European Association for the Study of the Liver, European As-
sociation for the Study of Diabetes, European Association for
the Study of Obesity. EASL–EASD–EASO Clinical Practice
Guidelines for the management of non-alcoholic fatty liver dis-
ease. J Hepatol 2016;64:1388–1402.
65. Dixon JB, Bhathal PS, Hughes NR, et al. Nonalcoholic fatty liver
disease: improvement in liver histological analysis with weight
loss. Hepatology 2004;39:1647–1654.
66. Verna EC, Berk PD. Role of fatty acids in the pathogenesis of
obesity and fatty liver: impact of bariatric surgery. Semin Liver
Dis 2008;28:407–426.
1060 Lee et al
67. Meinhardt NG, Souto KEP, Ulbrich-Kulczynski JM, et al. Hepatic
outcomes after jejunoileal bypass: is there a publication bias?
Obes Surg 2006;16:1171–1178.
68. Kristman V, Manno M, Cote P. Loss to follow-up in cohort studies:
how much is too much? Eur J Epidemiol 2004;19:751–760.
69. Fewtrell MS, Kennedy K, Singhal A, et al. How much loss to
follow-up is acceptable in long-term randomised trials and
prospective studies? Arch Dis Child 2008;93:458–461.
70. McCulloch P, Taylor I, Sasako M, et al. Randomised trials in
surgery: problems and possible solutions. BMJ 2002;
324:1448–1451.
Clinical Gastroenterology and Hepatology Vol. 17, No. 6
Reprint requests
Address requests for reprints to: Dennis Hong, MD, MSc, FRCSC, FACS,
Division of General Surgery, Department of Surgery, McMaster University, St.
Joseph’s Healthcare, 50 Charlton Avenue, East Hamilton, Ontario, L8N 4A6
Canada. e-mail: dennishong70@gmail.com; fax: (905) 521-6123.
Acknowledgments
The authors wish to thank Dr Gordon Guyatt, Dr Stefan Schandelmeier, and Dr
Vahid Ashoorion for their insights on research methods and using Grading of
Recommendations, Assessment, Development, and Evaluation for this review.
Conflicts of interest
The authors disclose no conflicts.
May 2019 Resolution of NAFLD After Bariatric Surgery 1060.e1

Supplementary
Figure 1. Proportion meta-
analysis forest plot of his-
tologic worsening of
NAFLD (liver side effects)
after surgery. ES, effect
size.
1060.e2 Lee et al Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Supplementary Figure 2. Subgroup analysis of biopsy-proven complete resolution of NAFLD features in patients who
received RYGB. (A) Steatosis, (B) inflammation/steatohepatitis, (C) ballooning degeneration, and (D) fibrosis. ES, effect size.
May 2019 Resolution of NAFLD After Bariatric Surgery 1060.e3

Supplementary
Figure 3. Subgroup anal-
ysis of histologic wors-
ening of NAFLD (liver side
effects) in patients who
received RYGB. ES, effect
size.

Supplementary Figure 4. Funnel plot assessing publication

bias of complete resolution of steatosis (meta-analysis
outcome with the most number of studies included).
Supplementary Table 1. Search Strategy up to May 2018

1060.e4 Lee et al
Embase MEDLINE Web of Science CENTRAL

1. Obesity/ 1. Obesity/ (obes* or over weight or overweight) AND (NAFL 1. Obesity/

2. Obes*.mp. 2. Obes*.mp. or non-alcoholic fatty liver disease* or 2. Obes*.mp.
3. OVERWEIGHT/ 3. OVERWEIGHT/ nonalcoholic fatty liver disease* or NASH or 3. OVERWEIGHT/
4. Over weight.mp. 4. Over weight.mp. non-alcoholic steato* or nonalcoholic 4. Over weight.mp.
5. Overweight.mp. 5. Overweight.mp. steato*) AND (bariatric* or gastroplast* or 5. Overweight.mp.
6. Or/1-5 6. Or/1-5 ((gastric or jejunoileal or jejuno-ileal or 6. Or/1-5
7. Non-alcoholic fatty liver disease/ 7. Non-alcoholic fatty liver disease/ ileojejunal or ileo jejunal or gastroileal or 7. Non-alcoholic fatty liver disease/
8. NAFLD.mp. 8. NAFLD.mp. roux-en-y or jejunal-ileal or biliopancreatic or 8. NAFLD.mp.
9. Non-alcoholic fatty liver disease*.mp. 9. Non-alcoholic fatty liver disease*.mp. bilio-pancreatic) NEAR2 bypass*) or 9. Non-alcoholic fatty liver disease*.mp.
10. Nonalcoholic fatty liver disease*.mp. 10. Nonalcoholic fatty liver disease*.mp. gastrojejunostom* or intestinal bypass* or 10. Nonalcoholic fatty liver disease*.mp.
11. NASH*.mp. 11. NASH*.mp. lipectom* or lipoplast* or lipolysis or 11. NASH*.mp.
12. Non-alcoholic steato*.mp. 12. Non-alcoholic steato*.mp. liposuction* or gastric band* or 12. Non-alcoholic steato*.mp.
13. Nonalcoholic steato*.mp. 13. Nonalcoholic steato*.mp. biliopancreatic diversion* or bilio-pancreatic 13. Nonalcoholic steato*.mp.
14. Or/7-13 14. Or/7-13 diversion or gastrectom* or duodenal switch 14. Or/7-13
15. Exp bariatric surgery/ 15. Exp bariatric surgery/ or gastric plication or gastric placation or 15. Exp bariatric surgery/
16. Bariatric*.mp. 16. Bariatric*.mp. gastric balloon* or gastric bubble* or 16. Bariatric*.mp.
17. Gastroplast*.mp 17. Gastroplast*.mp ballobes balloon* or duodeno-ileostomy) 17. Gastroplast*.mp
18. ((gastric or jejunoileal or jejuno-ileal or 18. ((gastric or jejunoileal or jejuno-ileal or 18. ((gastric or jejunoileal or jejuno-ileal or
ileojejunal or ileo ileojejunal or ileojejunal or ileo jejunal or gastroileal or
jejunal or gastroileal or roux-en-y) adj2 ileo jejunal or gastroileal or roux-en-y) roux-en-y) adj2 bypass*).mp.
bypass*).mp. adj2 bypass*).mp. 19. Gastrojejunostom*.mp.
19. Gastrojejunostom*.mp. 19. Gastrojejunostom*.mp. 20. Intestinal bypass*.mp.
20. Intestinal bypass*.mp. 20. Intestinal bypass*.mp. 21. Lipectom*.mp.
21. Lipectomy/ 21. Lipectomy/ 22. Lipoplast*.mp.
22. Lipectomy*.mp. 22. Lipectomy*.mp. 23. Lipolysis.mp.
23. Lipoplasty/ 23. Lipoplasty/ 24. Liposuction*.mp.

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

24. Lipoplast*.mp. 24. Lipoplast*.mp. 25. Gastric band*.mp.
25. Lipolysis/ 25. Lipolysis/ 26. Biliopancreatic bypass/
26. Lipolysis/.mp. 26. Lipolysis/.mp. 27. Biliopancreatic diversion*.mp.
27. Liposuction/ 27. Liposuction/ 28. Bilio-pancreatic Diversion.mp.
28. Liposuction*.mp. 28. Liposuction*.mp. 29. Gastrectomy/
29. Gastric band*.mp. 29. Gastric band*.mp. 30. Gastrectom*.mp.
30. Biliopancreatic bypass/ 30. Biliopancreatic bypass/ 31. Biliopancreatic diversion/
31. Biliopancreatic diversion*.mp. 31. Biliopancreatic diversion*.mp. 32. Duodenal switch.mp.
32. Bilio-pancreatic Diversion.mp. 32. Bilio-pancreatic Diversion.mp. 33. Gastric plication.mp.
33. Gastrectomy/ 33. Gastrectomy/ 34. Gastric placation.mp.
34. Gastrectom*.mp. 34. Gastrectom*.mp. 35. Gastric balloon/
35. Biliopancreatic diversion/ 35. Biliopancreatic diversion/ 36. Gastric bubble*
36. Duodenal switch.mp. 36. Duodenal switch.mp. 37. Ballobes balloon*.mp.
37. Gastric plication.mp. 37. Gastric plication.mp. 38. Duodeno-ileostomy.mp.
 May 2019
38. Gastric placation.mp. 38. Gastric placation.mp. 39. Or/15-38
39. Gastric balloon/ 39. Gastric balloon/ 40. 6 and 14 and 39
40. Gastric bubble* 40. Gastric bubble*
41. Ballobes balloon*.mp. 41. Ballobes balloon*.mp.
42. Duodeno-ileostomy.mp. 42. Duodeno-ileostomy.mp.
43. Or/15-42 43. Or/15-42
44. 6 and 14 and 43 44. 6 and 14 and 43
1762 668 841 39

CENTRAL, Cochrane Central Register of Controlled Trials.

Resolution of NAFLD After Bariatric Surgery 1060.e5

Supplementary Table 2. Initial Number of Patients With Steatosis, Steatohepatitis (Inflammation), Ballooning Degeneration, Fibrosis, and NAS

1060.e6 Lee et al
Lobular Baseline NAS score, Follow-up NAS score,
Study Steatosis, n inflammation, n Ballooning, n Fibrosis, n means  SD means  SD

Silverman, 1995 71 37 N/A 13 N/A N/A

Luyckx, 1998 57 10a 0a 1 N/A N/A
Dixon, 2004 35 24 26 23 N/A N/A
Kral, 2004 104a N/A N/A N/A N/A N/A
Clark, 2005 16 15 14 14 N/A N/A
Mattar, 2005 69 65 N/A 56 N/A N/A
Mottin, 2005 90 N/A N/A N/A N/A N/A
Stratopoulos, 2005 50 50 N/A 48 N/A N/A
Barker, 2006 19 19 N/A 14 N/A N/A
Csendes, 2006 15 N/A N/A N/A N/A N/A
De Almeida, 2006 16 15 16 4 N/A N/A
Mathurin, 2006 20 N/A N/A N/A N/A N/A
Meinhardt, 2006 36 13 N/A 16 N/A N/A
Furuya, 2007 18 18 17 9 N/A N/A
Liu, 2007 35 39 23 18 4.87  1.89 1.97  0.58
Mathurin, 2009 N/A N/A N/A 83a 1.97  1.33 1.07  1.26 (1 y), 1  1.33 (5 y)
Moschen, 2009 14 12 14 10 N/A N/A
Moretto, 2012 78 N/A 35 35 N/A N/A
Tai, 2012 19 15 18 18 3.33 (1–5)b 0.857 (0–2)b
Vargas, 2012 26 25 16 25 N/A N/A
Caiazzo, 2014 166 (AGB), 220 (RYGB) N/A N/A N/A 1.7  1.4 (AGB), 2  1.5 (RYGB) 1.1  1.2 (1 y, AGB), 0.7  1 (1 y,
RYGB), 1  1.3 (5 y, AGB), 0.7
 1.2 (5 y, RYGB)
Lassailly, 2015 N/A 81 81 73 5 1
Raj, 2015 30 14 10 14 2.6  1.3 0.57  0.97

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Taitano, 2015 160 156 N/A 158 N/A N/A
Froylich, 2016 N/A N/A N/A N/A 4.4  1.7 (RYGB), 2.6  1.6 (SG) 1.4  1.7 (RYGB), 0.9  1.2 (SG)
Schneck, 2016 9 9 9 8 5.11  0.33 0.67  1
Aldoheyan, 2017 N/A N/A N/A N/A 4 (3–5)b 2 (1–3)b
Hedderich, 2017 N/A N/A N/A N/A N/A N/A
Manco, 2017 20 17 20 20 4.15  0.67 1.6  1
Parker, 2017 13 12 12 15 N/A N/A
Luo, 2018 N/A N/A N/A N/A N/A N/A
Schewenger, 2018 27 21 9 27 2.07  1.53 0.33  0.78

a
Data could not be meta-analyzed because the study only reported baseline values.
b
Range shown.
Supplementary Table 3. Number of Histologic Worsening of NAFLD and Description of Postoperative Adverse Events After Bariatric Surgery

May 2019
NAFLD
Study Type of surgery Adverse event description complications, n Biopsies,

Silverman, 1995 RYGB 2 patients had symptoms of obstruction because of gallstones that developed after gastric bypass, 3 10 91
patients had increased steatosis, 1 patient developed slight perisinusoidal fibrosis, and 6 patients
developed slight portal fibrosis
Luyckx, 1998 GP A significant increase in hepatitis was observed in 18 patients (26%) compared with 10 patients (14%) 18 69
before gastroplasty (P < .05); 87.5% of the cases were graded as mild, 12.5% as moderate, and no
severe hepatitis was observed
Dixon, 2004 LAGB No adverse events were reported N/A N/A
Kral, 2004 BPD 1 patient with cirrhosis died of postoperative pulmonary embolism, 42 patients increased fibrosis 47 104
grade, 10 patients developed mild inflammation, 1 patient died of liver failure owing to
hemosiderosis after having her surgery reversed, 1 patient died of liver failure after dismantling
surgery, and 3 patients developed new cirrhosis
Clark, 2005 RYGB No patients had worsening of liver histologic results 0 16
Mattar, 2005 RYGB, LAGB, LSG Overall complication rate, 7%; 1 patient sustained a pulmonary embolus that required embolectomy, 1 N/A N/A
patient suffered acute renal failure that resolved with intravenous hydration, 1 patient developed a
splenic abscess that was treated with percutaneous drainage, 1 acalculous cholecystitis was
treated with intravenous antibiotics, and 1 patient with a urinary tract infection was treated with
antibiotics
Mottin, 2005 RYGB 1 patient died from complications during the immediate postoperative period, no patients had 0 90
worsening of liver histologic results
Stratopoulos, 2005 GP Liver fibrosis progressed in 6 patients, lipogranulomas worsened in 3 patients, regeneration worsened 6 51
in 6 patients
Barker, 2006 RYGB Only 1 patient had a worsening in lobular fibrosis from stage 0 to 1 1 19
Csendes, 2006 RYGB 1 patient showed histologic progression from mild steatosis to pericellular fibrosis 1 16
De Almeida, 2006 RYGB No adverse events were reported N/A N/A
Mathurin, 2006 BLB, LAGB The fibrosis score increased significantly 1 year after surgery from 0.14  0.39 to 0.38  0.64; N/A N/A
however, the mean magnitude of the increase (0.24) and the mean score of fibrosis on the second

Resolution of NAFLD After Bariatric Surgery 1060.e7

biopsy (0.38) were not clinically relevant
Meinhardt, 2006 JIB No patients had worsening of liver histologic results 0 41
Furuya, 2006 RYGB No adverse events were reported N/A N/A
Liu, 2007 RYGB 2 patients had an increase in lobular inflammation from grades 1 to 2, 6 patients developed portal 9 39
inflammation, 1 patient developed fibrosis
Mathurin, 2009 LAGB, GB, BLB 51 patients had worsening of fibrosis after 1 year; 47 patients had worsening of fibrosis after 5 years; an 51 267
increase in fibrosis was observed in the 5 years after surgery (from 0.27  0.55 to 0.36  0.59; P <
.001)
There was a significant increase in fibrosis 1 year after surgery, from 0.27  0.55 to 0.41  0.69 (P <
.002), whereas there was no significant difference between 1 and 5 years
Moschen, 2009 LAGB No adverse events were reported N/A N/A
Moretto, 2012 RYGB 5 patients developed hepatic fibrosis, portal fibrosis worsened or developed in 11 patients, lobular 22 126
fibrosis worsened or developed in 6 patients
Tai, 2012 RYGB 5 patients had worsened lobular inflammation after surgery, 2 patients had stenosis of 5 21
gastrojejunostomy, and 1 patient had bleeding from an ulcer at gastrojejonostomy
Vargas, 2012 RYGB No adverse events were reported N/A N/A
Caiazzo, 2014 LAGB 48 gastric bands were removed before 5 years owing to complications N/A N/A
 1060.e8 Lee et al
Supplementary Table 3. Continued

NAFLD
Study Type of surgery Adverse event description complications, n Biopsies,

Lassailly, 2015 RYGB, LAGB, SG No adverse events were reported N/A N/A
Raj, 2015 RYGB, LSG None had worsening of liver histologic results 0 30
Taitano, 2015 RYGB or LAGB Steatosis worsened in 5 patients, and developed in 3 patients, steatohepatitis developed in 3 patients, 78 160
lobular inflammation developed in 15 patients, chronic portal inflammation worsened in 9 patients,
and developed in 18 patients, fibrosis worsened in 12 patients, and developed in 12 patients,
cirrhosis developed in 1 patient who previously had grade 3 fibrosis
Froylich, 2016 RYGB 5 patients had minimal increase of ALT, 3 patients had worsening fibrosis postoperatively 3 25
Schneck, 2016 RYGB or SG 1 patient progressed from fibrosis F0 to F1A; all other histologic results improved or remained the same 1 9
Aldoheyan, 2017 SG No adverse events were reported N/A N/A
Hedderich, 2017 RYGB, SG No adverse events were reported N/A N/A
Manco, 2017 Gastric balloon, LSG No major perioperative complications were recorded N/A N/A
Of the 7 patients with gastric balloon, 3 patients (42.8%) required a second balloon, 1 patient presented
with asymptomatic deflation of a balloon that was evacuated
1 patient developed persistent vomiting that resolved after 5 days of treatment
1 patient developed dysphagia that resolved completely after 15 days
1 patient required hospital readmission for bronchopneumonia and pleural effusion 2 weeks after LSG
Parker, 2017 RYGB None had worsening of liver histologic results 0 15
Luo, 2018 RYGB or LSG 1 patient had significant hepatic bleeding from an unspecified surgery N/A N/A
Schewenger, 2018 RYGB Fibrosis worsened in 4 (9.5%) patients 4 42

ALT, alanine aminotransferase; BLB, bilio-intestinal bypass; GP, gastroplasty; LAGB, laparoscopic adjustable gastric banding; LSG, laparoscopic sleeve gastrectomy.

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Supplementary Table 4. MINORS Assessment of Included Studies

May 2019
MINORS criteria

Inclusion of Prospective End points Unbiased Follow-up period Loss to follow-up Prospective
A clearly consecutive collection of appropriate to the assessment of appropriate to the evaluation calculation of the
Study stated aim patients data aims of the study the study end point aim of the study less than 5% study size Total

Silverman, 1995 2 2 0 2 2 2 2 0 12
Luyckx, 1998 2 0 0 2 2 2 2 0 10
Dixon, 2004 2 2 1 2 2 2 1 0 12
Kral, 2004 2 2 2 2 2 2 2 0 14
Clark, 2005 2 2 0 2 2 2 2 0 12
Mattar, 2005 2 2 0 2 2 2 2 0 12
Mottin, 2005 2 2 0 2 2 2 2 0 12
Stratopoulos, 2005 2 1 2 2 2 2 2 0 13
Barker, 2006 2 2 0 2 2 2 2 0 12
Csendes, 2006 2 2 2 2 2 2 2 0 14
De Almeida, 2006 2 2 1 2 0 2 0 0 9
Furuya, 2006 2 2 2 2 2 2 2 0 14
Mathurin, 2006 2 2 2 2 2 2 1 0 13
Meinhardt, 2006 2 2 2 2 2 1 1 0 12
Liu, 2007 2 2 0 2 2 2 2 0 12
Mathurin, 2009 2 2 2 2 2 2 1 0 13
Moschen, 2009 2 1 2 2 2 1 1 2 13
Moretto, 2012 2 2 0 2 2 1 2 0 11
Tai, 2012 2 0 0 2 2 2 2 0 10
Vargas, 2012 2 0 2 2 2 2 2 0 12
Caiazzo, 2014 2 2 2 2 2 2 1 0 13
Lassailly, 2015 2 2 2 2 2 2 1 0 13
Raj, 2015 2 1 2 2 2 1 1 0 11

Resolution of NAFLD After Bariatric Surgery 1060.e9

Taitano, 2015 2 2 0 2 2 2 2 0 12
Froylich, 2016 2 2 0 2 0 2 2 0 10
Schneck, 2016 2 1 2 2 2 2 1 0 12
Aldoheyan, 2017 2 2 2 1 2 1 0 2 11
Hedderich, 2017 2 1 0 2 0 1 1 0 7
Manco, 2017 2 2 2 2 2 2 2 0 14
Parker, 2017 2 2 2 2 2 2 1 0 13
Luo, 2018 2 2 2 1 2 1 1 0 11
Schewenger, 2018 2 1 2 2 2 2 0 0 11
Supplementary Table 5. GRADE Certainty of Evidence Summary Table for Meta-Analysis of Proportions

1060.e10 Lee et al
GRADE certainty assessment Summary of findings

Anticipated
Patients, n effects

Large Dose- Plausible Overall Biopsies Pooled Resolution

Risk of Publication magnitude response confounders certainty of Complete with proportion with
Studies, n bias Inconsistency Indirectness Imprecision bias of effect gradient or biases evidence resolution, n disease, n (95% CI) surgery

Complete resolution of biopsy-proven steatosis

25 cohorts Not serious Serious Not serious Not serious None Yes N/A No 4 840 1318 66% (56–75) Resolution in
very low steatosis
was 66%
(660 per 1000)
Complete resolution of biopsy-proven inflammation
21 cohorts Not serious Serious Not serious Not serious None Yes N/A No 4 268 667 50% (35–64) Resolution in
very low inflammation
was 50%
(500 per 1000)
Complete resolution of biopsy-proven ballooning degeneration
15 cohorts Not serious Serious Not serious Not serious None Yes N/A no 4 237 320 76% (64–86) Resolution in
very low ballooning
degeneration
was 76%
(760 per 1000)
Complete resolution of biopsy-proven fibrosis
22 cohorts Not serious Serious Not serious Not serious None Yes N/A No 4 220 702 40% (29–51) Resolution in
very low fibrosis was 40%
(400 per 1000)

Clinical Gastroenterology and Hepatology Vol. 17, No. 6

Histologic worsening of NAFLD
19 studies Not serious Serious Not serious Not serious None None N/A No 4 247 1231 12% (5–20) Histological
very low worsening of
NAFLD was 12%
(120 per 1000)
Supplementary Table 6. GRADE Certainty of Evidence Summary Table Meta-Analysis of Continuous Variables

May 2019
GRADE certainty assessment Summary of findings

Anticipated effects

Large Plausible Overall Mean Reduction of

Publication magnitude Dose-response confounders certainty difference NAS with
Studies, n Risk of bias Inconsistency Indirectness Imprecision bias of effect gradient or biases of evidence (95% CI) surgery

Reduction of NAS score

11 studies Not serious Serious Not serious Not serious None Yes N/A No 4 2.39 (1.58–3.20) Reduction in NAS
very low score was 2.39
(95% CI, 1.58–3.20)

Resolution of NAFLD After Bariatric Surgery 1060.e11