Abstract
Objective
Obesity and associated comorbidities, such as NAFLD impose a major healthcare burden worldwide. Bariatric surgery remains the most successful approach for sustained weight loss and the resolution of obesity-related complications. However, the impact of preexisting NAFLD on weight loss after bariatric surgery has not been previously studied. The goal of this study is to assess the impact of preexisting NAFLD on weight loss outcomes up to 5 years after weight loss surgery.
Research Design and Methods
Data from the Michigan Bariatric Surgery Cohort (MI-BASiC) was extracted to examine the effect of baseline NAFLD on weight loss outcomes. The cohort included a total of 714 patients older than 18 years of age undergoing gastric bypass (GB; 380 patients) or sleeve gastrectomy (SG; 334 patients) at University of Michigan between January 2008 and November 2013. Repeated measures analysis was used to determine if preexisting NAFLD was a predictor of weight loss outcomes up to 5 years post-surgery.
Results
We identified 221 patients with an established clinical diagnosis of NAFLD at baseline. Multivariable repeated measures analysis with adjustment for covariates shows that patients with preexisting NAFLD had a significantly lower percentage of total and excess weight loss compared to patients without preexisting NAFLD. Further, our data show that baseline dyslipidemia is an indicator of the persistence of NAFLD after bariatric surgery.
Conclusions
Our data show that patients’ body weight loss in response to bariatric surgery is impacted by factors such as preexisting NAFLD. Additionally, we show that NAFLD may persist or recur in a subset of patients after surgery, and thus careful continued follow-up is recommended.
Keywords: Bariatric Surgery, Dyslipidemia, Nonalcoholic Fatty Liver Disease, Persistence, Weight Loss
Introduction
Obesity is a chronic complex multifactorial disease that affects multiple systems in the body [1, 2]. It is defined as a body mass index (BMI) of 30 kg/m2 or higher [3]. Obesity is highly prevalent, with nearly over a third of adults having obesity in 2008, increasing to around 40% in 2016, and current projections are that nearly one in every 2 adults in the United States will have obesity by 2030 [4, 5]. Nonalcoholic Fatty Liver Disease (NAFLD) is a spectrum of liver diseases characterized by hepatic fat accumulation without secondary causes such as significant alcohol consumption, drug-induced steatosis, or inherited disorders [6]. It can progress histologically from simple steatosis to steatohepatitis (nonalcoholic steatohepatitis (NASH)), advanced fibrosis, cirrhosis, and hepatocellular carcinoma [7]. NAFLD is a highly prevalent disease that is estimated to affect over 30% of the worldwide population [8].
NAFLD and obesity are strongly associated with each other, with around 70% of patients with NAFLD displaying obesity, and 65% of patients with obesity having NAFLD [9]. Multiple studies have shown that 3-5% weight reduction demonstrates improvement in liver fat measurements, and 5% reduction in BMI leads to approximately 25% reduction in liver fat [10, 11]. Bariatric surgery is the most effective treatment to achieve sustained weight loss, and also results in better glycemic control compared to alternative weight-loss therapies [12-15]. Bariatric surgery also measurably improves NAFLD-related morbidity and mortality, but the mechanisms behind the improved lipid deposition after surgery are not well understood [16].
Our previous study showed a significant effect of preexisting diabetes on weight loss over 5 years following bariatric surgery [17]. Since NAFLD can moderate the development of metabolic disease including clinical diabetes [18], the aim of this study is to assess the impact of preexisting NAFLD on weight loss outcomes up to 5 years post-surgery. In addition, we examined the improvement in NAFLD using clinical surrogates and looked at the persistence of NAFLD in a real-world setting, evaluating the factors that may be driving the persistence of NAFLD. Our study is the first study that places NAFLD as an important metabolic indicator that may impact weight loss outcomes after bariatric surgery. In addition, it is one of the first studies to assess the persistence of NAFLD after bariatric surgery.
Research Design and Methods
Patients
The Michigan Bariatric Surgery Cohort has been previously described (MI-BASiC) [19, 20]. Our retrospective study includes patients older than 18 years who underwent GB or SG surgery at the University of Michigan between January 2008 and November 2013. Patients had either body mass index (BMI) > 40 kg/m2 or BMI > 35 kg/m2 with an obesity-related comorbidity. We excluded patients with previous bariatric surgery, and patients undergoing bariatric surgery for reasons other than obesity and its related comorbidities, patients with incomplete surgeries due to unexpected operative findings. 714 individuals were included based on these criteria, with 564 (79%) being females. 380 of these individuals had GB, whereas 334 had SG. Prior to data collection, an institutional review board (IRB) permission was acquired. Informed consent was not necessary because of the nature of the inquiry.
Data Collection
Patients in the multidisciplinary Adult Bariatric Surgery Program at the University of Michigan receive a preoperative evaluation and are followed up at two weeks and two months post-surgery by the surgeon and a dietitian. Patients are seen in the Post-Bariatric Endocrine Clinic at six and twelve months post-surgery and annually in the following years to address long-term care problems with the endocrinologist and nutritionist.
A retrospective electronic medical review was performed, and data were gathered from preoperative and yearly postoperative visits at years one to five. Our previous publication has a detailed overview of the data collection [19].
Definitions and Time of Follow up
Retrospective electronic medical review was performed, and data was abstracted from visits that occurred preoperatively (within 60 days prior to surgery) and postoperatively at one to five years (± 180 days). Data up to five years are included to ensure equal follow-up time, even though follow-up is typically longer than five years. Baseline demographic characteristics collected included sex and age. Data collected at baseline and follow-up, when available, included weight, height, BMI, blood pressure, and HbA1c. We also collected total cholesterol, triglyceride, HDL, LDL, ALT, and AST levels. Total weight loss (TWL) is calculated as the difference between preoperative weight and postoperative weight at any time. Total weight loss percent (TWL%) is calculated as TWL divided by preoperative weight. Excess weight (EW) is the weight above the ideal weight (BMI at 25kg/m2). Excess weight loss percent (EWL%) is calculated as TWL divided by EW.
NAFLD at baseline was defined based on the presence of ICD-9 or ICD-10 diagnosis, evidence of hepatic steatosis on imaging studies (US, CT, or MRI), or liver biopsy consistent with NAFLD or NASH. To screen for the presence of NAFLD, a natural language-processing tool EMERSE was utilized, and results were verified manually by a board-certified hepatologist with expertise in NAFLD. Patients with a history of significant alcohol use or misuse defined by ICD-9 or ICD-10 diagnosis or other mention of this in their clinical chart were excluded. Patients with additional causes of chronic liver disease such as viral or autoimmune hepatitis or hereditary forms of liver disease were also excluded. Persistent NAFLD was assessed based on the presence of hepatic steatosis on abdominal imaging (ultrasound, CT, or MRI) over follow-up. The emergence of new alcohol use disorder post-surgery was also assessed to ensure there was not a secondary cause of hepatic steatosis post-surgery. NAFLD fibrosis score (NFS) was calculated using age, hyperglycemia, body mass index, platelet count, albumin, and AST/ALT ratio [21].
Baseline dyslipidemia was defined based on one or more of the following conditions: presence of the corresponding ICD-9 and ICD-10 codes, presence of lipid lowering medication prescriptions, triglyceride levels >155 mg/dL or cholesterol level>200 mg/dL. The presence of type 2 diabetes was defined based on the presence of the diagnosis in the chart or an HbA1c of at least 6.5% (48 mmol/mol) prior to surgery. The diagnosis of hypertension was defined by one of the following conditions: having corresponding ICD-9 or ICD-10 codes in the EMR, having a prescription for anti-hypertensive medication, having a systolic blood pressure >135 mm, Hg or diastolic blood pressure >85 mm, Hg at baseline visit. Comorbidities were determined by conducting a manual review of the EMR with at least two team members because some patients had been given medication for preventative purposes. To code the patient with type 2 diabetes, both members of the team were required to approve the condition.
Statistical Analysis
Continuous data were expressed as mean ± standard error of the mean. Categorical data were summarized by counts and percentages. Based on the distribution of the data for continuous variables, a two-sample Student t-test or Wilcoxon rank-sum test was used to compare baseline characteristics between the GB and SG surgery groups. Chi-square or Fisher's exact tests, as appropriate, were used to compare the categorical baseline characteristics. When comparing patients with and without NAFLD, the repeated measure analyses of the continuous weight-related outcomes (TWL, BMI (kg/m2), TWL%, and EWL%) were performed using a linear mixed model. Conditional Generalized Linear Mixed Models (GLMM) were used to calculate the odds of achieving excess weight loss of more than 50%, achieving BMI less than 30 kg/m2, or either of these milestone outcomes. Odds ratios (OR) and 95% confidence interval (CI) were calculated. All weight-related models were adjusted for baseline weight. Models for metabolic parameters were also adjusted for their baseline values.
All the above hypothesis tests were considered significant at a p-value of 0.05 or lower. Statistical analyses were carried out utilizing SAS (version 9.4; SAS Institute, Inc., Cary, NC).
Results
Baseline patient characteristics and description of two subgroups (NAFLD and non-NAFLD)
As previously published, the entire study cohort included 714 patients, 79% of which were females (n=564), with a median age of 45 (37-53) years. Median BMI was 47.5 (42.4-53.4) kg/m2. 380 patients underwent gastric bypass (GB) surgery, while 334 underwent sleeve gastrectomy (SG) surgery. Baseline characteristics between both groups were similar except for body mass index (BMI) and weight which were both significantly higher in the SG group (p-values <0.001), while baseline hemoglobin A1C, total cholesterol, and triglycerides were significantly higher in the GB group (p-values 0.026, 0.042, and <0.001 respectively) (Table 1). The baseline differences between the two surgery types are due to the evolution of surgery criteria exclusions as only patients with BMI greater than 50 kg/m2 were eligible for SG historically until the end of 2011 after which all individuals meeting surgical criteria could have SG.
Table 1.
Demographic characteristics for patients with and without NAFLD undergoing gastric bypass (GB) and sleeve gastrectomy (SG)
| Entire | NAFLD | non-NAFLD | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Characteristic | Total | GB | SG | P-value | NAFLD | Non-NAFLD | P-value | GB | SG | P-value | GB | SG | P-value |
| (n=714) | (n=380) | (n=334) | (n=221) | (n=493) | (n=128) | (n=93) | (n=252) | (n=241) | |||||
| Age, median | 45 [37 - 53] | 45 [35 - 53] | 46 [39 - 54] | 0.056 | 47 [40 - 55] | 44 [35 - 53] | <.001 | 46 [39 - 54] | 48 [41 - 55] | 0.204 | 43 [33 - 52] | 45 [37 - 53] | 0.086 |
| Gender (F) | 564 (79.0%) | 305 (80.3%) | 259 (77.5%) | 0.374 | 161 (72.9%) | 403 (81.7%) | 0.007 | 90 (70.3%) | 71 (76.3%) | 0.32 | 215 (85.3%) | 188 (78.0%) | 0.036 |
| Diabetes Mellitus | 257 (36.0%) | 149 (39.2%) | 108 (32.3%) | 0.056 | 124 (56.1%) | 133 (27.0%) | <.001 | 74 (57.8%) | 50 (53.8%) | 0.549 | 75 (29.8%) | 58 (24.1%) | 0.154 |
| Hypertension | 392 (54.9%) | 209 (55.0%) | 183 (54.8%) | 0.955 | 146 (66.1%) | 246 (49.9%) | <.001 | 84 (65.6%) | 62 (66.7%) | 0.872 | 125 (49.6%) | 121 (50.2%) | 0.893 |
| Baseline Dyslipidemia | 210 (29.4%) | 110 (28.9%) | 100 (29.9%) | 0.771 | 89 (40.3%) | 121 (24.5%) | <.001 | 50 (39.1%) | 39 (41.9%) | 0.667 | 60 (23.8%) | 61 (25.3%) | 0.699 |
| Systolic Blood Pressure | 137 [123 - 148] | 136 [123 - 149] | 138 [125 - 148] | 0.833 | 138 [123 - 148] | 136 [124 - 149] | 0.972 | 136 [123 - 148] | 138 [123 - 147] | 0.635 | 136 [122 - 149] | 138 [125 - 149] | 0.576 |
| Body Mass Index (BMI) | 47.45 [42.40 - 53.40] | 46.4 [41.6 - 51.9] | 49.7 [43.7 - 54.1] | <.001 | 46.40 [41.80 - 52.40] | 47.90 [42.60 - 53.70] | 0.07 | 45.45 [41.05 - 50.65] | 48.70 [42.90 - 53.20] | 0.012 | 46.65 [41.85 - 52.10] | 50 [43.90 - 54.30] | <.001 |
| Weight (KG) | 134 [116 - 155] | 129 [114 - 151] | 140 [120 - 159] | <.001 | 132 [115 - 153] | 135 [117 - 155] | 0.193 | 129 [114 - 149] | 135 [117 - 154] | 0.189 | 129 [114 - 151] | 142 [122 - 161] | <.001 |
| Hemoglobin A1C | 6 [5.60 - 6.80] | 6.10 [5.70 - 6.90] | 5.90 [5.60 - 6.70] | 0.026 | 6.20 [5.70 - 7.50] | 5.90 [5.60 - 6.40] | <.001 | 6.20 [5.70 - 7.30] | 6.30 [5.70 - 7.60] | 0.869 | 6 [5.60 - 6.60] | 5.80 [5.60 - 6.30] | 0.014 |
| Total Cholesterol | 175 [154 - 203] | 176 [159 - 204] | 173 [147 - 200] | 0.042 | 170 [149 - 193] | 178 [156 - 208] | 0.013 | 173 [156 - 200] | 168 [144 - 188] | 0.206 | 180 [161 - 209] | 176 [148 - 205] | 0.086 |
| Triglycerides | 134 [99 - 202] | 144 [109 - 212] | 119 [94 - 175] | <.001 | 156 [106 - 218] | 128 [96 - 172] | 0.002 | 170 [122 - 241] | 128 [94 - 195] | 0.01 | 138 [102 - 188] | 116 [92 - 152] | 0.01 |
| LDL | 99 [81 - 124] | 100 [81 - 123] | 98.50 [78 - 125] | 0.561 | 95.50 [76 - 117] | 102 [83 - 128] | 0.021 | 96 [77.50 - 117] | 94.50 [73 - 118] | 0.775 | 102 [84 - 129] | 103 [82 - 127] | 0.527 |
| HDL | 42.50 [36 - 50] | 42 [36 - 49] | 43 [37 - 50.50] | 0.272 | 40 [35 - 46] | 44 [37 - 52] | <.001 | 40 [34 - 46] | 41 [35 - 49] | 0.118 | 44 [37 - 52.50] | 44 [38 - 52] | 0.877 |
| AST | 25 [20 - 32] | 24 [20 - 32] | 25 [20 - 33] | 0.202 | 29 [23 - 42] | 23 [19 - 29] | <.001 | 29.50 [23 - 43] | 29 [24 - 42] | 0.997 | 23 [19 - 28] | 24 [20 - 30.50] | 0.036 |
| ALT | 28 [21 - 41] | 28 [21 - 42] | 29 [21 - 40] | 0.895 | 39 [26 - 54] | 26 [20 - 33] | <.001 | 41 [26.50 - 59] | 37 [25 - 49] | 0.116 | 25 [19 - 32] | 26 [20 - 35] | 0.161 |
We identified 221 patients with an established clinical diagnosis of NAFLD at baseline (GB =128, SG =93) using the methods described above. Among patients with NAFLD, the median age was 47 (40-55) years with 72.9% females (n=161) and a median BMI of 46.4 (41.8-52.4) kg/m2. 38 patients (17.2%) had biopsy-proven diagnosis, among which 25 (65.8%) showed non-alcoholic steatohepatitis (NASH) and 3 (7.9%) demonstrated liver cirrhosis. The remainder of the patients (n=493, GB=252, and SG= 241) constituted a non-NAFLD cohort; these patients were significantly younger (median age 44 (35-53) versus 47 (40-55) years) and had a higher female representation (81.7 percent versus 72.9 percent females) compared to those with clinically significant NAFLD. Similarly, diabetes, hypertension, and baseline dyslipidemia were more prevalent at baseline in the group with NAFLD (p-values of <0.001 for the three factors) (Table 1).
We also looked at the demographic characteristics and the presence of comorbidities in the two groups with respect to surgery type. Within the cohort with NAFLD, baseline characteristics between both groups were similar except for BMI and triglycerides which were significantly elevated in SG and GB groups respectively (p-values of 0.012 and 0.01 respectively). Among the group without NAFLD, there were differences in gender, BMI, and baseline HbA1c, triglycerides, and AST between the two surgery types (p-values 0.036, <0.001, 0.014, 0.01, and 0.036 respectively) (Table 1).
Weight and dyslipidemia improved postoperatively in all patients
We first looked at changes in parameters of interest across time with respect to surgery type. Improvement in percent total weight loss, hepatic, and lipid biomarkers such as liver enzymes and LDL, respectively, was observed in all patients (with and without NAFLD) (Figures 1 and 2). These improvements were maximal at the end of the first year after surgery, with a gradual increase in the mentioned measured parameters thereafter. However, there was a persistent net improvement in most of the studied parameters in both groups compared to the baseline persisting at five years.
Figure 1. Improvements in weight (%TWL), hepatic, and lipid biomarkers were observed maximally at the end of the first year after surgery.
A. Percentage of total weight loss (unadjusted) in patients undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) in total population as well as by presence of NAFLD at each year. B. Alanine Aminotransferase (ALT) and C. Aspartate Aminotransferase (AST) outcomes in patients undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) in total population as well as by presence of NAFLD at each year. All graphs show absolute value over time. Error bars represent IQR. Red four-pointed star indicated significant p-value (<0.05).
Figure 2. Improvements in LDL and were observed maximally at the end of the first year after surgery.
A. Low-Density Lipoprotein (LDL) and B. Triglycerides outcomes in patients undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) in total population as well as by presence of NAFLD at each year. All graphs show absolute value over time. Error bars represent IQR. Red four-pointed star indicated significant p-value (<0.05).
When comparing the weight between each surgery cohort, our results show the superiority of gastric bypass over sleeve gastrectomy throughout the follow-up period. Looking at the cohorts of patients with and without NAFLD, all patients had a maximal reduction in the mean of their weight-related measures after one-year post-surgery with both cohorts having clear weight regain/obesity recidivism over time toward baseline. Figure 1a shows how patients undergoing GB in both cohorts lost more weight compared to the group undergoing SG during the entire follow-up period (p-values of <0.001 for all years for both cohorts) (Figure 1a).
Within the cohort with NAFLD, the difference in the change from the baseline alanine transaminase (ALT) and aspartate aminotransferase (AST) levels between GB and SG was significant only during years one and three post-surgery (p-value <0.001 and 0.034 for ALT and 0.023 and 0.037 respectively) (Figure 1b, 1c). Among the cohort without NAFLD, the difference between GB and SG groups was significant for ALT during years one and two (p-values 0.007 and 0.003 respectively) and significant for AST levels during years one, four, and five (p-values 0.036, 0.013, and 0.001) (Figures 1b, 1c).
Figure 2a shows low-density lipoprotein (LDL) measurements comparison between both cohorts post-surgery. Within the cohort with NAFLD, the difference between GB and SG groups was significant between years one, two, and five (p-values of 0.007, 0.015, 0.035). In the cohort without NAFLD, the difference was significant during the entire follow-up period (p-values of <0.001, 0.002, 0.016, <0.001, <0.001) (Figure 2a). In the NAFLD cohort, there was no significant difference in triglyceride levels post-surgery between GB and SG groups (Figure 2a). Among the cohort without NAFLD, the difference between triglyceride measurements was significant during years one and three (p-values of 0.012 and 0.022) (Figure 2b).
Our results show the superiority of GB over SG with respect to the improvement in NAFLD fibrosis score over the entire follow-up period in the entire cohort, the difference between GB and SG groups was significant during the five years of follow-up (p-values of <0.001, 0.0113, 0.002, 0.013, 0.013). The median NFS effect does decay toward baseline at year five (Figure 5).
Figure 5. The comparison of NAFLD Fibrosis Score between GB vs. SG groups over five years follow-up.
Comparison of NAFLD Fibrosis Score (NFS) outcomes in patients with NAFLD undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) at each year. All graphs show absolute value over time. Error bars represent IQR. Red four-pointed star indicated significant p-value (<0.05).
Changes in clinical outcomes across time depending on the surgery type and the presence of NAFLD
We then looked at changes in parameters of interest over time by surgery type with the respect to presence of NAFLD. When comparing the impact of NAFLD on weight loss across surgery types, we found that the impact was more pronounced in patients undergoing GB than SG (Figure 3). Looking at patients undergoing GB, we can see that patients without NAFLD lost significantly higher percent total weight compared to patients with NAFLD across the five years follow-up period (p-values of 0.014, 0.002, 0.001, 0.002, and 0.008 respectively). The effect of preexisting NAFLD on patients undergoing SG was less evident, as patients without NAFLD undergoing SG lost a higher percent total weight compared to patients with NAFLD only during the first two years of follow-up (p-values of <0.001 and 0.02 respectively) (Figure 3a).
Figure 3. The comparison of %TWL, ALT, and AST between NAFLD vs. non-NAFLD groups by surgery type over five years follow-up.
A. Percentage of total weight loss (unadjusted) in patients with NAFLD vs. without NAFLD undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) at each year. B. Alanine Aminotransferase (ALT) and C. Aspartate Aminotransferase (AST) outcomes in patients with NAFLD vs. without NAFLD undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) at each year. All graphs show absolute value over time. Error bars represent IQR. Red four-pointed star indicated significant p-value (<0.05).
We also looked at the impact of NAFLD on liver enzymes by surgery type. This analysis revealed that the decrease in ALT over time is more pronounced in patients undergoing GB and patients without NAFLD improved their liver enzymes more significantly than patients with NAFLD during years one, two, three, and five of follow-up (p-values of <0.001, 0.002, 0.007, and 0.019 respectively). The effect of baseline NAFLD on patients undergoing SG was also less evident, as it was significant only during the second year of follow-up with a p-value of <0.001. The impact of baseline NAFLD on AST showed a similar pattern between patients undergoing GB vs SG, as it was only evident during the first two years of follow-up (p-values of 0.004 and 0.018 versus 0.045 and 0.025 respectively) (Figures 3b, 3c).
When looking at the effect of baseline NAFLD on reduction in LDL between patients undergoing GB or SG, we found no significant differences, suggesting that the impact of surgery type may be more significant than the presence of NAFLD diagnosis across patients undergoing the same surgery type (Figure 4a). The effect of NAFLD on reduction in triglycerides between the same groups was only significant between patients with and without NAFLD undergoing GB surgery during year two post-surgery (p-value 0.007) (Figure 4b).
Figure 4. The comparison of LDL and Triglycerides between NAFLD vs. non-NAFLD groups by surgery type over five years follow-up.
A. Alanine Aminotransferase (ALT) and B. Aspartate Aminotransferase (AST) outcomes in patients with NAFLD vs. without NAFLD undergoing Sleeve gastrectomy (SG, orange line) vs. Roux-en-Y gastric bypass (GB, blue line) at each year. All graphs show absolute value over time. Error bars represent IQR. Red four-pointed star indicated significant p-value (<0.05).
Baseline dyslipidemia impacted adequate versus suboptimal weight loss after surgery
Third, we looked at the baseline characteristics affecting the percentage of excess weight loss post-surgery. We identified 29 patients as a suboptimal loss group as they lost ≤30% of their excess weight at some point post-surgery, and 301 patients as an adequate loss group as they have lost ≥70% of their excess weight at some point post-surgery.
Assessing the entire cohort, we found older age was significantly higher in the suboptimal loss group versus the adequate loss group (52 (36-57) years versus 43 (35-51) years (p =0.010). Baseline weight and BMI were also significantly higher in the suboptimal loss group, with baseline weight and BMI medians of 149 (127-167) kg and 53.5 (47.9-59.6) kg/m2 respectively versus 124 (112-141) kg and 44.7 (40.5-48.9) kg/m2 respectively in the adequate loss group (p <0.001 for both measures). The Baseline AST level was significantly higher in the suboptimal loss group with a median of 31 (23-38) IU/L versus 24 (19-32) IU/L in the adequate loss group (p-value of 0.012). The presence of NAFLD was significantly higher in the suboptimal loss group as 48.3% (n=14) of them had NAFLD versus only 28.6% (n=86) of the adequate loss group (p-value of 0.027). Furthermore, baseline total cholesterol was significantly higher in the adequate loss group with a median of 178 (159-204) mg/dL versus 152 (139-163) mg/dL in the suboptimal loss group (p-value of 0.012) (Table 2a).
Table 2a.
Comparison of baseline characteristics b/n %EWL<30 and %EWL>=70 for the entire cohort
| Characteristic | %EWL<30 | %EWL>=70 | Total | P-value |
|---|---|---|---|---|
| (n=29) | (n=301) | (n=330) | ||
| Age, median | 52 [36 - 57] | 43 [35 - 51] | 43 [35 - 52] | 0.01 |
| Gender (F) | 21 (72.4%) | 248 (82.4%) | 269 (81.5%) | 0.186 |
| Diabetes Mellitus | 12 (41.4%) | 98 (32.6%) | 110 (33.3%) | 0.336 |
| Hypertension | 18 (62.1%) | 142 (47.2%) | 160 (48.5%) | 0.125 |
| Baseline Dyslipidemia | 12 (41.4%) | 75 (24.9%) | 87 (26.4%) | 0.055 |
| Systolic Blood Pressure | 141 [124 - 145] | 135 [122 - 147] | 136 [122 - 147] | 0.483 |
| Body Mass Index (BMI) | 53.50 [47.90 - 59.60] | 44.70 [40.50 - 48.90] | 45 [40.70 - 50.30] | <.001 |
| Weight (KG) | 149 [127 - 167] | 124 [112 - 141] | 127 [112 - 144] | <.001 |
| Hemoglobin A1C | 5.90 [5.70 - 7.50] | 5.90 [5.50 - 6.60] | 5.90 [5.60 - 6.60] | 0.136 |
| Total Cholesterol | 152 [139 - 163] | 178 [159 - 204] | 175 [156 - 203] | 0.012 |
| Triglycerides | 99 [90 - 126] | 134 [104 - 191] | 131 [98 - 188] | 0.13 |
| LDL | 81 [76 - 97] | 103 [85 - 125] | 99.50 [83 - 124] | 0.058 |
| HDL | 40 [34 - 53] | 44 [37 - 51] | 43 [37 - 52] | 0.426 |
| AST | 31 [23 - 38] | 24 [19 - 32] | 25 [20 - 33] | 0.012 |
| ALT | 32 [25 - 42] | 27 [20 - 42] | 28 [20 - 42] | 0.091 |
| Presence of NAFLD | 14 (48.3%) | 86 (28.6%) | 100 (30.3%) | 0.027 |
We next studied the baseline characteristics affecting the percentage of excess weight lost post-surgery for the NAFLD cohort. We identified 14 patients as a suboptimal loss group (≤30% EWL) and 86 patients as an adequate loss group (≥70% EWL). Age was significantly higher in the suboptimal loss group versus the adequate loss group (55 (50-58) years versus 45 (38-52) years) (p-value of 0.007). Baseline weight and BMI were both significantly higher in the suboptimal loss group with baseline medians of 149 (124-167) kg and 53.95 (48.3-59.6) kg/m2 respectively versus 121 (109-138) kg and 42.9 (39.8-46.9) kg/m2 respectively in adequate loss group (p-values of 0.006 and <0.001 respectively). Baseline dyslipidemia was observed in the suboptimal loss group, as 71.4% (n=10) of them had baseline dyslipidemia versus 36% (n=31) in the adequate loss group (p-value of 0.013) (Table 2b).
Table 2b.
Comparison of baseline characteristics b/n %EWL<30 and %EWL>=70 for the NAFLD cohort
| Characteristic | %EWL<30 | %EWL>=70 | Total | P-value |
|---|---|---|---|---|
| (n=14) | (n=86) | (n=100) | ||
| Age, median | 55 [50 - 58] | 45 [38 - 52] | 46 [38.50 - 54.50] | 0.007 |
| Gender (F) | 8 (57.1%) | 60 (69.8%) | 68 (68.0%) | 0.367 |
| Diabetes Mellitus | 10 (71.4%) | 42 (48.8%) | 52 (52.0%) | 0.117 |
| Hypertension | 10 (71.4%) | 50 (58.1%) | 60 (60.0%) | 0.347 |
| Baseline Dyslipidemia | 10 (71.4%) | 31 (36.0%) | 41 (41.0%) | 0.013 |
| Systolic Blood Pressure | 142 [132 - 145] | 135 [123 - 145] | 136 [123 - 145] | 0.27 |
| Body Mass Index (BMI) | 53.95 [48.30 - 59.60] | 42.90 [39.80 - 46.90] | 43.95 [40.05 - 48.50] | <.001 |
| Weight (KG) | 149 [124 - 167] | 121 [109 - 138] | 127 [110 - 142] | 0.006 |
| Hemoglobin A1C | 7.20 [6.40 - 7.90] | 5.90 [5.50 - 7.20] | 5.95 [5.60 - 7.30] | 0.059 |
| Total Cholesterol | 157 [138 - 171] | 178 [156 - 202] | 175 [150 - 195] | 0.058 |
| Triglycerides | 109 [92 - 174] | 174 [116 - 234] | 160 [115 - 226] | 0.084 |
| LDL | 81 [74 - 97] | 103 [80.50 - 119] | 97 [79 - 118] | 0.149 |
| HDL | 40 [35 - 47] | 39 [34 - 46] | 40 [34 - 46] | 0.734 |
| AST | 37 [30 - 52] | 28 [23 - 43] | 30 [23 - 44] | 0.091 |
| ALT | 41 [32 - 56] | 41.50 [25 - 62] | 41 [27 - 61] | 0.929 |
We also explored baseline characteristics affecting the percentage of excess weight lost post-surgery for the non-NAFLD cohort. We identified 15 patients as the suboptimal loss group (≤30% EWL) and 215 patients as the adequate loss group (≥70% EWL). Baseline weight and BMI were the only significantly different covariates, as they were higher in the suboptimal loss group with baseline medians of 149 (129-179) kg and 50.50 (45.90-62.20) kg/m2 respectively versus 125 (113–143) kg and 45.30 (40.80-50.40) kg/m2 respectively in the adequate loss group (p-values of 0.014 and 0.014 respectively) (Table 2c).
Table 2c.
Comparison of baseline characteristics b/n %EWL<30 and %EWL>=70 for the non-NAFLD cohort
| Characteristic | %EWL<30 | %EWL>=70 | Total | P-value |
|---|---|---|---|---|
| (n=15) | (n=215) | (n=230) | ||
| Age, median | 48 [35 - 53] | 42 [33 - 50] | 42 [33 - 50] | 0.381 |
| Gender (F) | 13 (86.7%) | 188 (87.4%) | 201 (87.4%) | 0.93 |
| Diabetes Mellitus | 2 (13.3%) | 56 (26.0%) | 58 (25.2%) | 0.367 |
| Hypertension | 8 (53.3%) | 92 (42.8%) | 100 (43.5%) | 0.426 |
| Baseline Dyslipidemia | 2 (13.3%) | 44 (20.5%) | 46 (20.0%) | 0.741 |
| Systolic Blood Pressure | 131 [122 - 145] | 135 [121 - 148] | 135 [122 - 147] | 0.992 |
| Body Mass Index (BMI) | 50.50 [45.90 - 62.20] | 45.30 [40.80 - 50.40] | 45.65 [41.10 - 50.80] | 0.014 |
| Weight (KG) | 149 [129 - 179] | 125 [113 - 143] | 126 [113 - 145] | 0.014 |
| Hemoglobin A1C | 5.85 [5.60 - 5.90] | 5.80 [5.50 - 6.40] | 5.80 [5.60 - 6.40] | 0.825 |
| Total Cholesterol | 148 [147 - 154] | 178 [159 - 207] | 176 [159 - 206] | 0.119 |
| Triglycerides | 86 [74 - 123] | 126 [95 - 151] | 124 [90.50 - 151] | 0.284 |
| LDL | 91 [76 - 96] | 104 [87 - 129] | 101 [86.50 - 129] | 0.362 |
| HDL | 37 [34 - 53] | 45 [39 - 53] | 45 [38 - 53] | 0.483 |
| AST | 24 [20 - 31] | 23 [18 - 28.50] | 23 [19 - 29] | 0.328 |
| ALT | 27 [24 - 35] | 25 [19 - 34] | 25 [20 - 34] | 0.317 |
Multivariable Analysis Models identified baseline type 2 diabetes and dyslipidemia as significant covariates contributing to percent total weight loss after surgery
In order to investigate the impact of baseline factors contributing to weight loss, percentage over time, we constructed a multivariable model for percent total weight loss. Preexisting NAFLD remained a highly significant covariate when adjusted for baseline BMI, age, gender, surgery type, time (year of follow-up), time by surgery type, preexisting diabetes, baseline cholesterol and triglyceride levels (Beta −2.288, 95% CI (−3.938, −0.639), p-value = 0.0067). Of note, the only other significant covariate with regard to comorbidities was the presence of baseline diabetes as we have previously reported REF 14 (Table 3).
Table 3.
Multivariable logistic regressions of total percent weight loss vs covariates adjusted for baseline BMI, age, gender, surgery type, time (year of follow up), time by surgery type, preexisting diabetes, baseline cholesterol and triglyceride levels
| Effect | Type of Surgery | Time (Years) | Beta [95%CI] | pvalue |
|---|---|---|---|---|
| Year | <.0001 | |||
| 2 vs 1 | −1.660 [−2.642, −0.679] | 0.001 | ||
| 3 vs 1 | −4.526 [−5.702, −3.349] | <.0001 | ||
| 4 vs 1 | −6.034 [−7.391, −4.677] | <.0001 | ||
| 5 vs 1 | −6.353 [−7.755, −4.951] | <.0001 | ||
| Surgery type | GB vs SG | 5.947 [4.300, 7.594] | <.0001 | |
| Surgery_type*Year interaction | 0.0047 | |||
| GB vs SG | 2 vs 1 | 1.416 [0.139, 2.694] | 0.0299 | |
| GB vs SG | 3 vs 1 | 1.991 [0.455, 3.527] | 0.0112 | |
| GB vs SG | 4 vs 1 | 0.814 [−0.959, 2.587] | 0.3674 | |
| GB vs SG | 5 vs 1 | −0.267 [−2.101, 1.566] | 0.7745 | |
| Baseline BMI | 0.073 [−0.028, 0.174] | 0.1574 | ||
| Baseline Age (years) | −0.185 [−0.264, −0.106] | <.0001 | ||
| Sex (Male vs. Female) | −0.648 [−2.600, 1.303] | 0.514 | ||
| Baseline DM (yes vs. no) | −2.811 [−4.602, −1.021] | 0.0022 | ||
| NAFLD cohort | −2.288 [−3.938, −0.639] | 0.0067 | ||
| Baseline Triglycerides | 0.007 [−0.002, 0.017] | 0.131 | ||
| Baseline Total Cholesterol (mg/dL) | −0.016 [−0.039, 0.007] | 0.1655 |
Baseline dyslipidemia is an indicator of the persistence of NAFLD after surgery
Figure 6 demonstrates the persistence of NAFLD over time. 37.5% (n=83) of patients who had NAFLD at baseline had follow-up abdominal imaging with 33.7% of those having imaging evidence of persistent NAFLD (Figure 6a). Among those with persistent NAFLD, imaging post-surgery was done at 1-2 years in 28.7%, 2-4 years in 16.3%, and >4 years in 55.0%. (Figure 6b).
Figure 6. Follow up imaging among NAFLD cohort.
A. Data shows percentage of patients with NAFLD persistence vs. NAFLD resolution among patients with NAFLD who had follow-up imaging. B. Data shows the timing at which NAFLD persistence was documents for patients with NAFLD who had follow-up imagining after surgery.
Examination of baseline predictors of persistent NAFLD five years post-surgery using multivariable analysis demonstrated that baseline dyslipidemia was the only predictive factor for the persistence of NAFLD (p-value of 0.047) (Table 4). Important to note that dyslipidemia was also the only significant univariate determinant. Thus, for patients with NAFLD, baseline dyslipidemia is a very important factor for both, the persistence of NAFLD and predicting suboptimal weight loss compared to other baseline factors.
Table 4.
Univariable and multivariable logistic regressions of NAFLD follow-up imaging vs change in weight at year 5 and covariates
| Effect | Univariable | Multivariable | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Global model | LASSO-selected | ||||||||||||||
| OR | LCL | UCL | p-value | N Used |
OR | LCL | UCL | p-value | N Used |
OR | LCL | UCL | p-value | N Used |
|
| Surgery Type (SG vs GB) | 0.62 | 0.23 | 1.64 | 0.332 | 83 | 0.41 | 0.05 | 3.162 | 0.39 | 43 | . | . | . | . | . |
| Baseline Diabetes Mellitus | 1.2 | 0.48 | 2.98 | 0.7 | 83 | 0.38 | 0.03 | 5.091 | 0.463 | 43 | . | . | . | . | . |
| Baseline Dyslipidemia | 2.81 | 1.1 | 7.23 | 0.032 | 83 | 8.51 | 1.02 | 70.66 | 0.047 | 43 | 3.48 | 0.62 | 19.45 | 0.155 | 50 |
| Baseline Systolic Blood Press | 0.99 | 0.96 | 1.01 | 0.32 | 83 | 0.98 | 0.93 | 1.027 | 0.395 | 43 | 1 | 0.97 | 1.04 | 0.941 | 50 |
| Baseline Body Mass Index | 1.02 | 0.96 | 1.08 | 0.626 | 83 | 1.16 | 0.98 | 1.382 | 0.094 | 43 | 1.08 | 0.93 | 1.259 | 0.298 | 50 |
| Baseline Weight | 1 | 0.98 | 1.02 | 0.954 | 83 | 0.98 | 0.92 | 1.04 | 0.485 | 43 | 1 | 0.96 | 1.044 | 0.988 | 50 |
| Baseline HBA1C | 1.18 | 0.79 | 1.75 | 0.425 | 60 | 0.88 | 0.42 | 1.813 | 0.724 | 43 | 1.02 | 0.54 | 1.915 | 0.959 | 50 |
| Baseline Total Cholesterol | 1 | 0.98 | 1.02 | 0.855 | 59 | . | . | . | . | . | . | . | . | . | . |
| Baseline Triglycerides | 1 | 1 | 1.01 | 0.778 | 59 | 1 | 1 | 1.011 | 0.469 | 43 | 1 | 0.99 | 1.007 | 0.834 | 50 |
| Baseline LDL | 1 | 0.98 | 1.01 | 0.604 | 58 | 0.98 | 0.95 | 1.016 | 0.306 | 43 | 1 | 0.98 | 1.024 | 0.964 | 50 |
| Baseline HDL | 1.03 | 0.98 | 1.08 | 0.273 | 59 | 1.06 | 0.99 | 1.138 | 0.108 | 43 | 1.06 | 0.99 | 1.135 | 0.077 | 50 |
| Baseline AST | 1.01 | 0.99 | 1.03 | 0.424 | 83 | 0.92 | 0.8 | 1.049 | 0.204 | 43 | 0.9 | 0.8 | 1.013 | 0.081 | 50 |
| Baseline ALT | 1 | 0.99 | 1.02 | 0.823 | 83 | 1.06 | 0.97 | 1.151 | 0.203 | 43 | 1.06 | 0.99 | 1.132 | 0.096 | 50 |
| Weight Change at Year 5 | 0.99 | 0.97 | 1.02 | 0.608 | 68 | 1 | 0.93 | 1.074 | 0.99 | 43 | . | . | . | . | . |
| Max Weight Change | 1 | 0.97 | 1.02 | 0.722 | 83 | . | . | . | . | . | . | . | . | . | . |
Discussion
Obesity and its related comorbidities have become a major healthcare burden worldwide. Non-alcoholic fatty liver disease (NAFLD), and nonalcoholic steatohepatitis (NASH) are strongly linked to obesity and type 2 diabetes. While these conditions are now also being referred to as metabolic dysfunction associated steatotic liver disease (MASLD), this study uses the old terminology since this work started earlier than this consensus [22]. Bariatric surgery has shown great promise in the treatment of NAFLD [16]. This study aimed to assess the impact of preexisting NAFLD on weight loss outcomes up to 5 years post GB and SG surgeries in a cohort of 714 patients that underwent bariatric surgery at University of Michigan (MI-BASiC cohort). Our analysis showed that patients with preexisting NAFLD achieved less weight loss compared to patients without NAFLD when undergoing bariatric surgery. These differences were apparent within one year after surgery and persisted over five years. Notably, we observed the persistence of NAFLD in a subset of patients, and they appear to be different in their metabolic abnormalities. Our analyses demonstrated that baseline dyslipidemia was independently associated with both lower percentage weight loss and NAFLD persistence.
In our dataset, patients with NAFLD had a greater reduction in weight after GB compared to SG. Our observations also show improvement in clinical NAFLD after bariatric surgery, just as in other previously published studies [23-25]. Importantly, however, we also observed the persistence of NAFLD in a subset of patients. Our further analyses showed that baseline dyslipidemia predicts the persistence of NAFLD. This outcome was in addition to patients with NAFLD achieving lower weight loss compared to other baseline factors. These findings suggest that the baseline state of the patient has important implications for post-bariatric surgery outcomes. Therefore, the baseline workup of the patient should alert the clinician in setting realistic expectations for weight loss and improvement of hepatic lipid parameters after bariatric surgery. These new data identify the necessity of an individualized and patient-focused approach and develop individualized plans for patients with a specific set of metabolic complications as they contemplate bariatric surgery. Appropriate selection of pharmacological and lifestyle interventions both pre- and post-operatively is often required based on the patient history and comorbidities. Therefore, enhancing outcomes through combined therapies (bariatric surgery and pharmacological interventions) should be recommended for patients with baseline dyslipidemia before and after undergoing bariatric surgery.
The MI-BASiC cohort, consisting of 714 patients, is one of the largest single-institution cohorts of the most commonly performed bariatric procedures: sleeve gastrectomy and gastric bypass. This cohort includes a balanced distribution of patients across both types of surgeries and evaluates various aspects such as weight loss parameters, metabolic parameters, and the impact on obesity-related comorbidities. Here, we examined how preexisting clinical NAFLD impacted weight loss outcomes using percent weight loss as a continuous measure. We then evaluated the weight loss outcomes of the two surgeries using preexisting NAFLD as a confounder. Both sets of analyses demonstrated a significant impact of NAFLD on the ultimate weight loss achieved. A recently published study by Rheinwalt and colleagues evaluated the impact of baseline NAFLD on weight loss after RYGB surgery, and the study concluded that baseline histopathological presence of NAFLD might lead to inferior postoperative weight reduction after gastric bypass surgery [26]. However, the study included 143 patients, with one year of follow-up, and one type of surgery which limits the understanding of the sustained effects of weight loss surgery [26]. The majority of weight loss operations in the United States today are sleeve gastrectomy (SG) and thus our analysis usefully compares the effects of GB and SG on multiple metabolic parameters [27]. Most studies evaluate the resolution of NAFLD with bariatric surgery, but to our knowledge, this is the first study to evaluate if baseline dyslipidemia impacts weight loss outcomes in a larger cohort with two surgery types and the persistence or reemergence of NAFLD upon longer-term follow-up.
These data also expand our knowledge of persistent or recurrent NAFLD among individuals who undergo bariatric surgery. In our study, 33.7% of individuals with baseline NAFLD and follow-up imaging had evidence of persistent NAFLD with a higher percentage of steatosis noted at longer follow-up intervals (4 years post-surgery). A study done by Jimenez and colleagues on 90 patients undergoing RYGB showed that patients with obesity recidivism or regain above 20% of the maximal weight loss have significantly increased their fibrosis score compared to other groups at year three post-surgery [28]. A meta-analysis of 2649 paired liver biopsies in individuals with NAFLD undergoing bariatric surgery with a median follow-up of 15 months noted new or worsening features of NAFLD in 12% of patients (95% CI 5-20%) [24]. Mathurin and colleagues followed 211 patients with severe obesity (mean BMI 50) for 5 years post bariatric surgery and noted 37.7% of patients had persistent steatosis at year 5 with refractory insulin resistance as an independent predictor of persistent steatosis and ballooning at year 5 [29]. Of note, in that cohort 56.2% underwent gastric band, 22.8% biliointestinal bypass, and 21% underwent gastric bypass. The prevalence of persistent steatosis in this paired biopsy cohort is consistent with that noted in our study using non-invasive assessments, though independent predictors of persistent steatosis varied, perhaps as a reflection of differences in surgical types. Our finding that baseline dyslipidemia was independently associated with persistent NAFLD post-bariatric surgery may reflect the role of excess hepatic free fatty acids in hepatic cell injury and apoptosis [30], with hepatic diacylglycerol content being the best predictor of insulin resistance in non-diabetic patients [31], and the correlation between insulin resistance, increased fatty acid beta-oxidation, and hepatic oxidative stress and NAFLD disease spectrum [32].
There are several limitations to note in our study. This study is an observational retrospective cohort study, though the data capture and follow-up in our bariatric surgery program is quite rigorous as part of routine clinical care and thus these data were available for analysis retrospectively. The diagnosis of NAFLD was predominantly made non-invasively, though that is consistent with clinical practice guidelines and routine clinical care. Prior studies have analyzed the ability of ICD-9 and ICD-10 coding as a mechanism to capture patients with NAFLD and have assessed this method to have sufficient accuracy for EHR-based analysis when accounting for other potential causes of steatosis and chronic liver disease as was done in this study [33]. Patients were not randomized to surgery type and there may be selection bias in the groups. Patients undergoing SG had more weight at baseline owing in part to the insurance approval criteria from 2008 to 2012 for SG in Michigan and nationwide. Lastly, our evaluation of persistent NAFLD was limited by only a small proportion of our cohort undergoing imaging or liver biopsy post-surgery and thus there is potential to miss individuals who may have had persistent or recurrent disease.
Conclusion
Our data show that preexisting NAFLD decreases the weight loss effect of bariatric surgery. Although the mechanistic implications behind these results are not well understood, these data can guide healthcare providers who may need to intensify pharmacological and lifestyle therapies for patients with diagnosed or suspected NAFLD prior to surgery. Further, individuals with preexisting dyslipidemia are prone to persistent or re-emergent NAFLD and their follow-up program should focus on careful evaluation of liver health.
Key Points:
Since preexisting NAFLD may impact the outcomes of bariatric surgery, it is important to screen for this comorbidity prior to weight loss surgery to set realistic expectations.
While NAFLD may improve after weight loss surgery, there is a subset of individuals where persistence may be observed.
Concurrent dyslipidemia is an important predictor of persistent NAFLD post bariatric surgery.
Acknowledgments
We are grateful for our patients and families for seeking care at University of Michigan.
Funding
This study has no external sponsor that could have influenced the design and conduct of the study or the writing of the manuscript. Work was enabled through a generous gift funding to Dr. Oral from grateful patients (Baker family and Parker Family).
Footnotes
Data sharing and guarantor ship
EAO has full access to raw data and assumes the final responsibility to submit for publication.
Conflict of Interest
EAO has received funding from Novo Nordisk, Rhythm Pharmaceuticals, GID and Fractyl Health for trials on the obesity space but these have no relationship to the work.
Ethics statement
This study was approved by the Institutional Review Board of Michigan Medicine Medical School.
Informed Consent
Informed consent was not necessary because of the nature of the inquiry.
References
- 1.Bray GA, Kim KK, Wilding JPH, and World Obesity F, Obesity: a chronic relapsing progressive disease process. A position statement of the World Obesity Federation. Obes Rev, 2017. 18(7): p. 715–723. [DOI] [PubMed] [Google Scholar]
- 2.Bluher M., Obesity: global epidemiology and pathogenesis. Nat Rev Endocrinol, 2019. 15(5): p. 288–298. [DOI] [PubMed] [Google Scholar]
- 3.Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. World Health Organ Tech Rep Ser, 1995. 854: p. 1–452. [PubMed] [Google Scholar]
- 4.Hales CM, et al. , Trends in Obesity and Severe Obesity Prevalence in US Youth and Adults by Sex and Age, 2007-2008 to 2015-2016. JAMA, 2018. 319(16): p. 1723–1725. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ward ZJ, et al. , Projected U.S. State-Level Prevalence of Adult Obesity and Severe Obesity. N Engl J Med, 2019. 381(25): p. 2440–2450. [DOI] [PubMed] [Google Scholar]
- 6.Chalasani N., et al. , The diagnosis and management of non-alcoholic fatty liver disease: practice Guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association. Hepatology, 2012. 55(6): p. 2005–23. [DOI] [PubMed] [Google Scholar]
- 7.Powell EE, et al. , The natural history of nonalcoholic steatohepatitis: a follow-up study of forty-two patients for up to 21 years. Hepatology, 1990. 11(1): p. 74–80. [DOI] [PubMed] [Google Scholar]
- 8.Riazi K., et al. , The prevalence and incidence of NAFLD worldwide: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol, 2022. 7(9): p. 851–861. [DOI] [PubMed] [Google Scholar]
- 9.Zou B., et al. , Prevalence, characteristics and mortality outcomes of obese, nonobese and lean NAFLD in the United States, 1999-2016. J Intern Med, 2020. 288(1): p. 139–151. [DOI] [PubMed] [Google Scholar]
- 10.Thoma C, Day CP, and Trenell MI, Lifestyle interventions for the treatment of non-alcoholic fatty liver disease in adults: a systematic review. J Hepatol, 2012. 56(1): p. 255–66. [DOI] [PubMed] [Google Scholar]
- 11.Patel NS, et al. , Effect of weight loss on magnetic resonance imaging estimation of liver fat and volume in patients with nonalcoholic steatohepatitis. Clin Gastroenterol Hepatol, 2015. 13(3): p. 561–568 e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Adams TD, et al. , Weight and Metabolic Outcomes 12 Years after Gastric Bypass. N Engl J Med, 2017. 377(12): p. 1143–1155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Schauer PR, et al. , Bariatric Surgery versus Intensive Medical Therapy for Diabetes - 5-Year Outcomes. N Engl J Med, 2017. 376(7): p. 641–651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Mingrone G., et al. , Metabolic surgery versus conventional medical therapy in patients with type 2 diabetes: 10-year follow-up of an open-label, single-centre, randomised controlled trial. Lancet, 2021. 397(10271): p. 293–304. [DOI] [PubMed] [Google Scholar]
- 15.Colquitt JL, Pickett K, Loveman E, and Frampton GK, Surgery for weight loss in adults. Cochrane Database Syst Rev, 2014. 2014(8): p. CD003641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Aminian A., et al. , Association of Bariatric Surgery With Major Adverse Liver and Cardiovascular Outcomes in Patients With Biopsy-Proven Nonalcoholic Steatohepatitis. JAMA, 2021. 326(20): p. 2031–2042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Luo Y., et al. , Impact of diabetes on weight loss outcomes after bariatric surgery: Experience from 5-year follow-up of Michigan Bariatric Surgery Cohort. Clin Endocrinol (Oxf), 2023. 99(3): p. 285–295. [DOI] [PubMed] [Google Scholar]
- 18.Anstee QM, Targher G, and Day CP, Progression of NAFLD to diabetes mellitus, cardiovascular disease or cirrhosis. Nat Rev Gastroenterol Hepatol, 2013. 10(6): p. 330–44. [DOI] [PubMed] [Google Scholar]
- 19.Lager CJ, et al. , Metabolic Parameters, Weight Loss, and Comorbidities 4 Years After Roux-en-Y Gastric Bypass and Sleeve Gastrectomy. Obes Surg, 2018. 28(11): p. 3415–3423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lager CJ, et al. , Roux-En-Y Gastric Bypass Vs. Sleeve Gastrectomy: Balancing the Risks of Surgery with the Benefits of Weight Loss. Obes Surg, 2017. 27(1): p. 154–161. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Angulo P., et al. , The NAFLD fibrosis score: a noninvasive system that identifies liver fibrosis in patients with NAFLD. Hepatology, 2007. 45(4): p. 846–54. [DOI] [PubMed] [Google Scholar]
- 22.Rinella ME, et al. , A multi-society Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol, 2023. [DOI] [PubMed] [Google Scholar]
- 23.Fakhry TK, et al. , Bariatric surgery improves nonalcoholic fatty liver disease: a contemporary systematic review and meta-analysis. Surg Obes Relat Dis, 2019. 15(3): p. 502–511. [DOI] [PubMed] [Google Scholar]
- 24.Lee Y., et al. , Complete Resolution of Nonalcoholic Fatty Liver Disease After Bariatric Surgery: A Systematic Review and Meta-analysis. Clin Gastroenterol Hepatol, 2019. 17(6): p. 1040–1060 e11. [DOI] [PubMed] [Google Scholar]
- 25.Mummadi RR, Kasturi KS, Chennareddygari S, and Sood GK, Effect of bariatric surgery on nonalcoholic fatty liver disease: systematic review and meta-analysis. Clin Gastroenterol Hepatol, 2008. 6(12): p. 1396–402. [DOI] [PubMed] [Google Scholar]
- 26.Rheinwalt KP, et al. , Baseline Presence of NAFLD Predicts Weight Loss after Gastric Bypass Surgery for Morbid Obesity. J Clin Med, 2020. 9(11). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.English WJ, et al. , American Society for Metabolic and Bariatric Surgery 2018 estimate of metabolic and bariatric procedures performed in the United States. Surg Obes Relat Dis, 2020. 16(4): p. 457–463. [DOI] [PubMed] [Google Scholar]
- 28.Jimenez LS, et al. , Impact of Weight Regain on the Evolution of Non-alcoholic Fatty Liver Disease After Roux-en-Y Gastric Bypass: a 3-Year Follow-up. Obes Surg, 2018. 28(10): p. 3131–3135. [DOI] [PubMed] [Google Scholar]
- 29.Mathurin P., et al. , Prospective study of the long-term effects of bariatric surgery on liver injury in patients without advanced disease. Gastroenterology, 2009. 137(2): p. 532–40. [DOI] [PubMed] [Google Scholar]
- 30.Li ZZ, Berk M, McIntyre TM, and Feldstein AE, Hepatic lipid partitioning and liver damage in nonalcoholic fatty liver disease: role of stearoyl-CoA desaturase. J Biol Chem, 2009. 284(9): p. 5637–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kumashiro N., et al. , Cellular mechanism of insulin resistance in nonalcoholic fatty liver disease. Proc Natl Acad Sci U S A, 2011. 108(39): p. 16381–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Sanyal AJ, et al. , Nonalcoholic steatohepatitis: association of insulin resistance and mitochondrial abnormalities. Gastroenterology, 2001. 120(5): p. 1183–92. [DOI] [PubMed] [Google Scholar]
- 33.Boursier J., et al. , Non-invasive tests accurately stratify patients with NAFLD based on their risk of liver-related events. J Hepatol, 2022. 76(5): p. 1013–1020. [DOI] [PubMed] [Google Scholar]