Nonalcoholic Fatty Liver Disease Impairs the Liver–Alpha Cell Axis Independent of Hepatic Inflammation and Fibrosis

Subjects and Study Design

We recruited 33 adult individuals with obesity who were scheduled for bariatric surgery at Copenhagen University Hospital Hvidovre (Hvidovre, Denmark) and 9 healthy controls.

Study participants were required to fulfill the general criteria for bariatric surgery (laparoscopic gastric bypass [Roux-en-Y gastric bypass; RYGB] or laparoscopic sleeve gastrectomy [SG]) according to the Danish national bariatric guidelines, namely a body mass index (BMI) greater than 35 kg/m² and the presence of at least one obesity-related comorbidity (T2D, uncontrolled hypertension, polycystic ovarian syndrome, documented sleep apnea, or arthrosis in the lower extremities) or a BMI greater than 40 kg/m² in the absence of the aforementioned comorbidities. In addition, eligible bariatric candidates must complete an 8% mandatory weight loss before referral for the RYGB or SG procedure in Denmark. Choice of surgical procedure (RYGB or SG) was based on the patients’ own preference, the endocrinologist’s recommendation, and any relevant comorbidities. A further study requirement was that participants be of North European ethnicity. Exclusion criteria were former or ongoing regular alcohol intake (2.5 units/day for men and 1.5 units/day for women), former or ongoing long-term use of medication known to induce secondary NAFLD/NASH, known preexisting liver disease (except preoperative NAFLD), disease in the lipid metabolism, and/or acute or chronic inflammatory disease.

Participants in CON consisted of 9 healthy adults scheduled for planned (not acute) laparoscopic cholecystectomy. CON participants were required to be between 18 and 65 years old, have a BMI between 18.0 and 27.5 km/m², and have a waist circumference less than 88 cm for women and less than 102 cm for men. Additionally, this group had to be healthy (apart from gall bladder stones) and with zero intake of prescribed medication other than allergy medicine, mild pain killers, and/or oral contraceptives.

Ethics

The study was conducted in accordance with the Helsinki Declaration and approved by the Ethics Committee for the Capital Region of Denmark (H-16030784 and H-16030782). All participants received oral and written information regarding the experimental procedures and their potential risks. Oral and written consent were obtained from all participants.

Study Design and Anthropometrics

Study participants with obesity were assessed twice: at baseline (on the morning of the day of bariatric surgery) and 12 months after surgery (follow-up). At both visits, we registered anthropometrics and phenotypic data and collected fasting blood samples.

At baseline, a wedged liver biopsy from the edge of the right liver lobe was collected intraoperatively immediately after trocar placement but before the actual bariatric procedure. The same surgical team, consisting of three senior bariatric surgeons, performed all biopsies.

At the 12-month follow-up, an ultrasound-guided TruCut percutaneous liver biopsy was sampled from the right liver lobe in all 33 study participants with obesity.

Participants in CON were studied once, on the day of their planned cholecystectomy. Anthropometrics, phenotypic data, blood samples, and an intraoperatively wedged liver biopsy were collected as described above.

Three study subjects were enrolled before their required 8% weight loss. Fasting blood samples were collected and a liver biopsy was obtained by TruCut, before the weight loss, at the time of surgery, and 12 months after surgery.

Biochemical Analyses

Plasma alanine aminotransferase (ALT), gamma-glutamyltransferase (GGT), aspartate aminotransferase (AST), plasma-glucose, low-density lipoproteins (LDLs), very low-density lipoproteins (VLDLs), high-density lipoproteins (HDLs), and triglycerides were measured using the Roche/Hitachi Cobas c 8000 system (Roche Diagnostics GmbH, Mannheim, Germany) and Cobas calibrators and reagents according to the manufacturer’s instructions. Serum insulin and C-peptide concentrations were measured by immunoassay with the Cobas e 602.

Hemoglobin A1c (HbA1c) was measured in plasma with the Tosoh TSKgel G8 Variant His on the Tosoh Automated Glycohemoglobin Analyzer HLC-723G8 (Tosoh Corporation, Tokyo, Japan)

Plasma Glucagon and Targeted Metabolomics

Plasma concentrations of glucagon were quantified using a validated enzyme-linked immunosorbent assay (catalog no. 10-1271-01; Mercodia, Uppsala, Sweden).⁽²¹⁾ Plasma concentrations of l-amino acids (termed total AAs) were quantified using the enzymatic l-Amino Acid Assay Kit (catalog no. ab65347; Abcam, Cambridge, United Kingdom). The assay was evaluated by recovery experiments using pooled human plasma (n = 4) with known amounts of AAs (catalog no. A6282; Sigma Aldrich, Copenhagen, Denmark). Recovery of AAs was on average (± SD) 79 ± 9% in human plasma. The individual AAs were measured in a targeted metabolomic approach (using retention time and mass to charge) by mass spectrometry, as described elsewhere.⁽²⁰⁾

Liver Biopsies and Histologic Examination

The liver biopsies were fixed in neutrally buffered formalin at room temperature for 24-48 hours, processed, and embedded in paraffin. Slides were cut at 3 µm and evaluated according to standard pathology guidelines. The biopsies were examined independently by three experienced liver pathologists who were blinded to clinical details. Consensus was sought in the event of disagreement among scores. Liver fibrosis (Fib) was staged from F0 to F4 based on sirius red staining, and the NAFLD activity score (NAS) was calculated using standard guidelines for histologic scoring of NAFLD.⁽²²⁾

To identify how NAFLD severity affects metabolism, we subdivided the group of study subjects with (bariatric) obesity (OBE; n = 33) into three groups (termed NAFLD groups) based on liver histology: (1) OBE-no steatosis comprised 17 patients without liver steatosis but with a median NAS of 2; (2) OBE+steatosis comprised 8 patients with liver steatosis but an overall NAS of less than 5; (3) OBE-NASH-Fib comprised 8 patients with a NAS of 5 or above and/or fibrosis grade 2.

CON consisted of 9 healthy, normal weight/near normal weight, control subjects without hepatic steatosis and with NAS between 0.5 and 1.5

Calculations

To generate a surrogate marker for the hepatic actions of glucagon on ureagenesis, we calculated a previously evaluated glucagon-alanine index using the following formula: glucagon-alanine index = (fasting plasma glucagon [pmol/L] × fasting plasma alanine [µmol)/L]).⁽⁸⁾

Because C-peptide and insulin are secreted from pancreatic beta cells in equimolar amounts while only insulin binds to hepatic insulin receptors and hence is subjected to first-pass metabolism in the liver, calculating the hepatic insulin extraction (insulin clearance) can be used as a proxy for insulin sensitivity of the liver.⁽²³⁾ Hepatic insulin clearance is calculated as the ratio between C-peptide and insulin. A higher ratio indicates a higher clearance and a higher degree of hepatic insulin sensitivity.

The Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) was calculated according to the formula (fasting glucose [mg/dL] × insulin [mU/L]/405), which primarily is an index of hepatic insulin resistance.⁽²⁴⁾

At the 12-month follow-up visit, we calculated percentage excess body weight loss (%EBWL) as ([{baseline BMI – follow-up BMI}/{baseline BMI-25}] × 100%).

Statistics

The criteria for normal distribution were not met, therefore we carried out nonparametric testing. Comparisons between more than two groups were made using the Kruskal-Wallis one-way analysis of variance (ANOVA) test, and comparisons within groups (baseline versus 12-month follow-up) were made using the Wilcoxon signed-rank test for paired data. The results of the Kruskal-Wallis tests were adjusted with the Bonferroni correction for multiple comparisons in cases of significant overall outcome. Correlations were assessed using Spearman’s rank correlation.

Study participant characteristics are presented as medians with interquartile range (IQR) or as frequencies. P < 0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics 25 64-bit. Graphs and illustrations were made using GraphPad Prism version 8.1.1 for Windows 10 (GraphPad Software, Inc., www.graphpad.com).

Participants in OBE are Characterized by Hyperglucagonemia and Hyperaminoacidemia

OBE and CON were matched for age and sex. Not surprisingly, participants in OBE were metabolically deranged, reflected by significantly increased fasting glucose, insulin, C-peptide, and HBA1c levels compared with CON (Table 2; Fig. 1E). In total, 10 subjects in OBE had T2D; 6 were treated only with metformin, 2 were treated only with liraglutide, and 2 were treated with a combination of metformin, liraglutide, and long-lasting insulin. Liver histology in OBE was characterized by a significantly higher NAS than in CON, with median NAS scores of 3 (IQR, 2-4) versus 1 (IQR, 0.5-1.5; P < 0.001) and driven by the presence of steatosis and ballooning. Correspondingly, ALT levels were higher in OBE (median, 32 U/L; IQR, 24-39) than in CON (median, 19 U/L; IQR, 18-27; P < 0.05).

OBE was characterized by increased plasma concentrations of glucagon (60%), i.e., hyperglucagonemia, compared to CON (Fig. 2A; P < 0.05). Similarly, plasma concentrations of the AA pool (termed total AAs) were 56% higher in OBE (Fig. 2C; P < 0.001).

Severity of NAFLD Does Not Worsen Hyperglucagonemia in OBE

We found no significant difference in glucagon concentrations between OBE-no steatosis (median, 8.0 pmol/L; IQR, 6.0-11.5), OBE+steatosis (median, 7.5 pmol/L; IQR, 6.0-9.8), and OBE-NASH-Fib (median, 7.5 pmol/L ; IQR, 6.0-9.8; P = 0.195).

Insulin resistance was numerically higher in OBE-NASH-Fib (median HOMA-IR, 6.9; IQR, 6.0-10.9) versus OBE-no steatosis (median HOMA-IR, 4.0; IQR, 3.0-6.6; P = 0.118). Fasting insulin secretion, evaluated from plasma C-peptide levels, was not found to be different across NAFLD groups (P = 0.486).

The OBE-NASH-Fib group was, as expected, the NAFLD group with the most pronounced metabolic derangement, displaying simultaneous and significantly elevated waist:hip ratio, HbA1c, fasting plasma glucose, insulin, C-peptide, and HOMA-IR levels (Table 1).

Hyperaminoacidemia Is Increased Comparably Among All NAFLD Groups

All NAFLD groups had marked hyperaminoacidemia, with 54% (P < 0.05), 59% (P = 0.01), and 60% (P < 0.01) higher plasma concentrations of AAs in OBE-no steatosis, OBE+steatosis, and OBE-NASH-Fib, respectively. The presence of hepatic steatosis and/or fibroinflammation did not appear to aggravate the hyperaminoacidemia (Fig. 2C; P = 1.0 ).

Liver–Alpha Cell Axis Is Disrupted in Patients With Obesity and NAFLD

The glucagon-alanine index was significantly increased in all NAFLD groups (Fig. 2E) compared with CON, but the severity of NAFLD did not significantly impact the glucagon-alanine indexes (P = 0.109). Hepatic insulin resistance, reflected by fasting HOMA-IR, was increased and the hepatic insulin extraction ratio was markedly lower in OBE (Fig. 1C; P < 0.05), worsening in step with increasing NAFLD severity.

To investigate whether glucagon resistance toward AA metabolism is associated with hepatic insulin resistance, we correlated the glucagon-alanine index to fasting HOMA-IR and found a positive correlation (Spearman’s rank coefficient R = 0.165; P < 0.05) (Fig. 3).

Hyperaminoacidemia Persists in All NAFLD Groups Despite Improvement in NAFLD and Metabolic Derangement 12 Months After Bariatric Surgery

Twelve months after bariatric surgery, all 33 OBE subjects had experienced pronounced weight loss, with a median reduction in %EBWL of 50.3 (IQR, 38.7-90.2) in OBE-NASH-Fib to 57.7 (IQR, 42.8-86.2) in OBE+steatosis (P = 0.850 among NAFLD groups) (Table 1) and an overall decrease in median BMI from 42 to 33 kg/m² (Table 2).

NAS declined significantly in all three NAFLD groups (delta NAS baseline to 12 months: OBE-no steatosis, 1, IQR, 1-2, P < 0.001; OBE+steatosis, 2.5, IQR, 1.3-3, P < 0.05; OBE-NASH-Fib, 3, IQR, 3-4.8, P < 0.05). OBE-NASH-Fib had a larger reduction in NAS compared to the OBE-no steatosis group (P < 0.001). Of the 17 study participants with steatosis at baseline, 13 had reached grade zero steatosis at 12 months and all three NAFLD groups saw a significant reduction in ballooning grade. Furthermore, the OBE-NASH-Fib group achieved reduced inflammation (P < 0.05), whereas fibrosis did not decrease significantly (P = 0.096). Metabolically, all groups displayed significant and comparable improvement in HOMA-IR, C-peptide, and insulin (Table 1; Fig. 1B). Three participants still had manifest T2D and needed treatment with metformin (2 participants) and a combination of liraglutide, metformin, and long-lasting insulin (1 participant).

Interestingly, bariatric surgery and weight loss had no impact on total AA concentrations, which were elevated and unchanging in all NAFLD groups (Table 1; Fig. 2D).

Hyperglucagonemia Resolves After Bariatric Surgery in Patients With NAFL But Not in Those With NASH

Plasma concentrations of glucagon were significantly reduced from a median of 8.0 pmol/L (IQR, 6.0-10.0) to 5.0 pmol/L (IQR, 4.0-8.5; P < 0.01) 12 months after bariatric surgery. The OBE-no steatosis and OBE+steatosis group normalized their plasma concentrations of glucagon relative to CON. In contrast, glucagon concentrations did not change in OBE-NASH-Fib after 12 months compared with their baseline (OBE-NASH-Fib 12-month median, 7.0 pmol/L, IQR, 5.0-9.8 versus baseline median, 7.5 pmol/L, IQR, 6.0-9.8; P = 0.799) (Table 1; Fig. 2B).

Improvements in Glucagon and Insulin Resistance After Bariatric Surgery

Twelve months after bariatric surgery, the glucagon-alanine index was significantly reduced in OBE-no steatosis and OBE+steatosis but not in OBE-NASH-Fib (Fig. 2F). The decreased glucagon-alanine index was driven by concomitant decreases in both glucagon and alanine plasma concentrations.

Pronounced reductions in HOMA-IR were seen in OBE-no steatosis (P < 0.01) and OBE-NASH-Fib (P < 0.05), but only negligible reductions were seen in OBE+steatosis. Both C-peptide and insulin were significantly lowered from baseline to 12 months; insulin decreased by 52% and C-peptide by 35%, indicating a significant increase in the hepatic insulin extraction in OBE from median 9.51 (IQR, 7.86-11.42) at baseline to 12.78 (IQR, 10.24-15.71) 12 months after surgery (P < 0.01) (Fig. 1C,D).

No Differences Were Observed in the Effect on Biochemical, Anthropometric, or Clinical Parameters Between RYGB and SG Individuals

When comparing the effect of RYGB with SG, we found no difference between the delta changes among any of the parameters assessed (Table 2).