Association between lower plasma adiponectin levels and higher liver stiffness in type 2 diabetic individuals with nonalcoholic fatty liver disease: an observational cross-sectional study

In the last few years, there has been steadily increasing scientific interest in nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH) because of their growing burden on public health worldwide [1, 2]. NAFLD has reached epidemic proportions in most high-income countries, affecting up to a third of the general adult population [3, 4] and up to ~ 70% of patients with type 2 diabetes mellitus (T2DM) [5], as well as almost all patients with severe obesity [6].

Although the precise mechanisms involved in the development and progression of NAFLD remain to be fully elucidated, it has become increasingly clear that insulin resistance and low-grade inflammation are two key factors involved in the pathophysiology of NAFLD [7]. Crosstalk between adipokines and proinflammatory cytokines may play a role in the development of this liver disease [8, 9]. Adiponectin is a hormone mainly produced by adipocytes, which circulates in three different isoforms (i.e., low-molecular-weight (LMW), middle-molecular-weight (MMW), and high-molecular-weight (HMW) adiponectin) in the bloodstream [8‐10]. Specifically, circulating levels of adiponectin are closely and inversely associated with insulin resistance, excess body weight, and ectopic fat deposition [8, 9].

In recent years, many observational studies [11‐16] and some meta-analyses [17] have reported that lower plasma adiponectin levels are significantly associated with the presence and severity of NAFLD, especially in patients without T2DM, thereby suggesting that hypoadiponectinemia might represent a risk factor for NAFLD. However, little information is available to date on the association between lower plasma adiponectin levels and the presence of NAFLD or NASH in patients with T2DM [18]. Nor, to our knowledge, are any data available assessing whether this association is at least partially mediated by coexistence of the rs738409 genetic variant in the patatin-like phospholipase domain-containing 3 (PNPLA3) gene or other genetic polymorphisms that confer greater susceptibility to NAFLD and NASH [19, 20].

Of the 79 men with non-insulin-treated T2DM included in the study (mean age 67 ± 10 years, BMI 27.7 ± 4 kg/m², HbA1c 52 ± 6 mmol/mol), 28 (35.5%) did not have NAFLD on ultrasonography, 32 (40.5%) had hepatic steatosis alone, and 19 (24%) had NAFLD with coexisting significant fibrosis (i.e., LSM ≥ 7.0 kPa on Fibroscan®), respectively.

Table 1 shows the main clinical, biochemical, and genetic characteristics of participants who were simultaneously stratified by the presence and severity of NAFLD. Compared with those without NAFLD, patients with NAFLD, regardless of the presence or absence of significant fibrosis, were more likely to be younger, centrally obese, and more insulin resistant (as reflected by a higher HOMA-IR score) and also had greater diastolic blood pressure and higher serum ALT and triglyceride concentrations. Notably, compared to those without NAFLD, patients with NAFLD had markedly lower high-molecular-weight adiponectin levels (5.5 [IQR 2.3–7.6] vs. 2.4 [1.8–3.7] vs. 1.6 [1.0–2.9] μg/mL; p < 0.001 for trend). No significant intergroup differences were observed in terms of diabetes duration, smoking history, systolic blood pressure, glycemic control, total and HDL cholesterol, platelet count, AST, GGT, albumin, kidney function parameters, and comorbidities (such as prior ischemic heart disease and diabetic retinopathy), as well as the current use of lipid-lowering, anti-hypertensive, antiplatelet, or glucose-lowering agents (except for metformin use). Moreover, there were no significant intergroup differences in the distribution of the PNPLA3, TM6SF2, and MBOAT7 genetic variants among the patient subgroups. No significant intergroup differences were also found in the other six genotyped NAFLD-related polymorphisms (data not shown).

Figure 1 shows that there was a significant inverse association between plasma adiponectin levels and Fibroscan-assessed LSM in the whole group of patients.

The independent association between plasma adiponectin levels and the presence and severity of NAFLD is illustrated in Table 2. In the unadjusted regression model, lower plasma adiponectin levels were significantly associated with nearly threefold and sixfold odds of either hepatic steatosis alone or NAFLD with coexisting significant fibrosis. These results remained statistically significant after adjustment for age, BMI, and HOMA-IR score (model 1). The additional adjustment for either the PNPLA3 (model 2), TM6SF2 (model 3), or MBOAT7 (model 4) genetic variants did not substantially change the results. Almost identical results were also observed even when we included waist circumference as a covariate (instead of BMI), when we additionally adjusted for duration of diabetes, or when we excluded those with Fibroscan-assessed LSM ≥ 20 kPa (n = 2) (data not shown). Similar results were also observed when we excluded patients treated with pioglitazone (n = 13) (Supplementary Table 1), those treated with SGLT-2 inhibitors (n = 12) (Supplementary Table 2), or those treated with GLP-1 receptor agonists (n = 21) (Supplementary Table 3). However, in this latter case, the significance of the results was weaker in fully adjusted regression models; we believe that this is likely due to the smaller sample size available for this latter analysis (58 patients included). Finally, as reported in Supplementary Table 4, the further adjustment for the other six genotyped NAFLD-related polymorphisms did not change the results of the study.

In a meta-analysis of 27 observational studies that reported data on 2243 individuals (1545 patients with NAFLD and 698 controls), Polyzos et al. reported that control subjects had significantly higher plasma adiponectin levels compared to patients with simple steatosis alone (random-effects weighted mean difference [WMD] 3.0, 95% CI 1.57–4.43, I² = 80.4%) or those with NASH (random-effects WMD 4.70, 95% CI 3.71–5.78, I² = 84.1%) [17]. By performing a meta-regression analysis, age, sex, BMI, and pre-existing diabetes failed to account for the observed heterogeneity [17]. In another observational study, plasma adiponectin levels were significantly lower in patients with NAFLD than in those with viral hepatitis or other chronic liver diseases [29]. In a cross-sectional study including 84 Brazilian obese patients with T2DM and biopsy-proven NAFLD, Leite et al. reported that plasma adiponectin levels were lower in patients with NASH than in their counterparts without NASH [18]. Finally, in a recent systematic review and meta-analysis of 12 case–control studies, Zheng et al. reported that rs266729 and rs3774261 polymorphisms in the adiponectin gene were associated with a higher risk of having NAFLD among Asian, Chinese, and Caucasian populations [30].

Collectively, our findings confirm and extend the aforementioned observations, showing a strong association between lower plasma adiponectin levels and the presence and severity of NAFLD (as assessed with liver ultrasonography and Fibroscan®, which are the two most widely used diagnostic tools to non-invasively diagnose and stage NAFLD in clinical practice [31]) in Caucasian outpatients with non-insulin-treated T2DM. Some studies showed that NAFLD patients carrying the PNPLA3 G/G genotype had lower plasma adiponectin levels compared to those with the PNPLA3 C/C genotype [32], whereas others did not [14]. Notably, in our study, we showed for the first time that the inverse association between plasma adiponectin levels and the presence/severity of NAFLD remained statistically significant even after adjusting for the PNPLA3 rs738409 variant or other less common NAFLD-related genetic polymorphisms. Although the results of our study do not have clear implications for clinical practice, they may be important for guiding future mechanistic and intervention studies. In fact, the evidence from this and other studies [11‐16, 18] suggests a possible role of decreased adiponectin levels in the pathophysiology of NAFLD, thus underlining the need for future large-scale studies assessing the predictive as well as the therapeutic role of this adipokine in the spectrum of NAFLD. Additionally, these results may encourage the designing of diagnostic accuracy studies to examine whether plasma adiponectin levels may contribute (alone or in combination with other biomarkers) to the achievement of a non-invasive diagnosis of liver fibrosis.

To date, the role of adiponectin in the pathogenesis of NAFLD has not been fully elucidated. It is well known that adiponectin is one of the most important and abundant adipose-tissue-secreted adipokines [33]. Adipocytes secrete adiponectin into the bloodstream as three oligomeric complexes, including a low-molecular-weight (LMW) trimer, a medium-molecular-weight (MMW) hexamer, and a high-molecular-weight (HMW) multimer [10]. HMW adiponectin is the predominant isoform in circulation and has been identified as the most active biological isoform. HMW adiponectin exerts beneficial systemic metabolic and anti-inflammatory effects, as it promotes improvements in hepatic and systemic insulin sensitivity, glucose uptake, and lipid metabolism [10]. Adiponectin acts by binding and activating two different receptor isoforms, namely AdipoR1 and AdipoR2, expressed in skeletal muscle and the liver, among other tissues [34, 35]. Decreased plasma adiponectin levels are strongly associated with abdominal obesity, ectopic fat deposition, and greater insulin resistance [8, 9]. However, adiponectin may also signal in the liver, exerting a beneficial insulin-sensitizing effect [8, 9, 15]. Indeed, adiponectin regulates glucose and lipid metabolism by stimulating hepatic fatty acid oxidation and inhibiting hepatic fatty acid synthesis, mostly via the activation of the AMP-activated protein kinase [15]. Adiponectin may also attenuate liver fibrosis by inducing nitric oxide production of hepatic stellate cells that constitutively express both AdipoR1 and AdipoR2 [36]. Adiponectin also stimulates TIMP metallopeptidase inhibitor-1 (TIMP-1) secretion by hepatic stellate cells to retard their migration, thereby further contributing to the anti-fibrotic effect of adiponectin [37]. Overall, therefore, hypoadiponectinemia might promote the development of NAFLD and liver fibrosis [35, 38]. It is well known that glitazones (especially pioglitazone) have demonstrated promising results in randomized controlled trials for treatment of NASH [39]. Interestingly, parallel increases in plasma adiponectin levels and histological improvement of NASH were also observed in a systematic review of four randomized clinical trials, providing data on 187 patients with NASH treated up to 12 months [40]. Consequently, some authors have proposed that administration of adiponectin or an adiponectin analog (e.g., osmontin) might be an attractive pharmacological strategy for management of conditions characterized by adiponectin deficiency, such as NAFLD or NASH [8, 41]. Another potential therapeutic option would be the upregulation of endogenous adiponectin. Although the use of glitazones, such as rosiglitazone or pioglitazone, has been restricted due to moderate weight gain, peripheral fluid retention, and an increase in myocardial infarction risk (rosiglitazone), the non-thiazolidinedione, selective peroxisome proliferator-activated receptor-γ modulators, like INT131 besylate, could be promising candidates for randomized controlled trials with the potential to improve both glucose metabolism and NAFLD/NASH (while minimizing the side effects of full PPAR-γ agonists) [8, 42]. However, further research is needed to examine the therapeutic role of administration of adiponectin or an adiponectin analog in patients with NAFLD.

Our study has some limitations that should be mentioned. Firstly, the cross-sectional design of this single-center study limits our ability to establish temporal or causal associations between lower adiponectin levels and presence/severity of NAFLD. Secondly, the sample size of our study was small and was comprised of Caucasian men with metabolically well-controlled T2DM. Thus, these results cannot necessarily be extrapolated to other ethnic groups of patients, to women with T2DM (notably, the exploration of sex differences is today a priority of NAFLD research [43]), or to patients with uncontrolled glycemia. On the other hand, this latter limitation can represent a specific strength of our study in that it shows that lower adiponectin levels are associated with NAFLD and significant fibrosis in the absence of major changes in glycemia. Thirdly, we did not perform a liver biopsy or magnetic resonance elastography for staging liver fibrosis. Consequently, we were not able to compare the results of liver stiffness obtained by VCTE with histology data. However, both liver ultrasonography and VCTE are widely used for the diagnosis and staging of NAFLD in routine clinical practice [31]. In addition, a meta-analysis of 12 observational studies (published from January 2011 to February 2021) showed that conventional liver ultrasonography allows for reliable and accurate detection of ≥ 5% histologically defined hepatic steatosis (82% sensitivity and 80% specificity), as well as moderate-severe hepatic steatosis (85% sensitivity and 85% specificity), compared to liver histology [44].

Despite these limitations, our study has some important strengths, including the consecutive enrolment of the study population, its data completeness, and the adjustment for diabetes-related variables and other potential confounders, including a large panel of specific NAFLD-related genetic polymorphisms. In addition, the liver ultrasound and VCTE examinations were performed by a single trained physician, who was blinded to participants’ clinical and biochemical details, thereby eliminating both assessment bias and interobserver variability. Finally, we excluded patients with important comorbidities (for example, cirrhosis, advanced kidney disease, or cancer), deeming that including patients with such comorbidities might have confounded the interpretation of data.