The study was designed as a cross-sectional, single-center, matched case–control study. Acromegaly was diagnosed by failure of suppression of serum GH concentrations below 1 ng/mL after a 75-g oral glucose tolerance test (OGTT) together with fasting serum IGF-1 levels above the normal ranges for age and gender, and the presence of clinical features. The IGF-1 upper limit of normal (IGF-1 ULN) was calculated by dividing IGF-1 by the upper limit of IGF-1 based on age and gender [22]. Consecutive patients with newly diagnosed acromegaly admitted to our clinic were assessed for inclusion in the study. Two patients were excluded due to pre-existing conditions: one had chronic liver disease, and the other was taking insulin for poorly controlled diabetes. The control group consisted of individuals who attended our outpatient clinic for routine check-up. For each case, two age, sex, body mass index (BMI) and waist circumference (WC)-matched controls were selected.

Exclusion criteria for the selection of the patient and control groups were to be older than 65 years or younger than 18 years old and to have excess alcohol consumption, a history of toxic, autoimmune, viral or metabolic liver disease, secondary causes of fatty liver (e.g., use of systemic steroids, tamoxifen and methotrexate), chronic liver disease of any cause, cirrhosis, any malignancy, chronic renal failure, chronic respiratory failure and cardiac disease. Control subjects known to have NAFLD and diabetes were not included. The study only included patients with newly diagnosed acromegaly to eliminate any potential treatment effects. Since acromegaly often leads to diabetes due to increased insulin resistance, we only included patients with controlled diabetes (glycosylated haemoglobin (HbA1c) levels < 7%) who were not using insulin.

Blood pressures and mean arterial pressures were measured. Bioelectrical impedance analysis (BIA) was performed using the Tanita BC-418 MA Body Composition Analyzer (TANITA Corp., Tokyo, Japan). Height, weight, body mass index (BMI), WC, total body fat mass and percentage were all recorded.

Serum samples for analysis were obtained early in the morning after an overnight fast. Serum fasting glucose, insulin, total cholesterol (Tchol), high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and triglyceride (TG) levels, HbA1c, Alanine aminotransferase (ALT), Aspartate aminotransferase (AST), Gamma glutamyltransferase (GGT), Alkaline phosphatase (ALP) of the patient group and the control group, in addition to the GH and IGF-1 values of the patient group, were recorded. The Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) was calculated using the (glucose (mg/dL) x insulin (µIU/mL) /405) formula.

For CK18 and IGFBP7 measurements, patient and control samples were centrifuged and stored at −80 °C until the day of analysis. Before analysis, serum samples were first transferred to −20 °C and then to room temperature. Elabscience Human CK18 ELISA kit (detection range 6.25–400 mIU/mL, sensitivity < 3.75 mIU/ml, coefficient of variation (CV) < 10%) and Elabscience Human IGFBP7 ELISA (detection range 0.94–60 ng/ml, sensitivity < 0.56 ng/ml, CV < 10%) were used. The kits were used following the manufacturer's instructions, applying a 1:3 dilution to the serum samples.

The presence of any three of the five criteria was defined as metabolic syndrome (MS); elevated WC (≥ 102 cm in men or ≥ 88 cm in women), elevated TG (≥ 150 mg/dL), reduced HDL (< 40 mg/dL in men, < 50 mg/dL in women) elevated blood pressure (BP) (≥ 130 mm Hg systolic BP or ≥ 85 mm Hg diastolic BP), and elevated fasting glucose (≥ 100 mg/dL) [23].

Ultrasound examinations were performed by two radiologists, each with over 10 years of experience in abdominal radiology, who were blinded to the patients’ clinical status. All patients had fasted for at least 6 h before undergoing ultrasound examinations using a single ultrasound machine (RS85 Prestige, Samsung Medison Co. Ltd) equipped with a convex transducer (1–7 MHz). The imaging of patients was performed in the supine position with the right intercostal approach and the patient’s right hand positioned over the head to increase the width of the intercostal spaces. The radiologist, at first, examined the liver along its course and then measured the liver length in the midclavicular line in centimetres (cm), while the patient was in the supine position. Secondly, stiffness measurements of the liver using SWE imaging were obtained in accordance with the Society of Radiologists in Ultrasound (SRU) guideline [24]. Five consecutive stiffness measurements with breath-hold situation during neutral breathing. The radiologists paid particular attention to avoid applying any pressure with the transducer during the SWE imaging. One cm diameter circular regions of interest are used to measure the liver stiffness values. According to the manufacturer’s recommendations, the appropriate Reliability Measurement Index (RMI) value (> 0.4) is used as a quality indicator for stiffness measurements. Patients with increased liver stiffness (≥ 5kPa) and increased fibrosis (8 > kPa) were noted [13, 24]. The median value of five measurements in kilopascals (kPa) was noted for analyses. The stiffness measurements were considered as reliable if the interquartile range to median value was lower than 30%.

Lastly, the hepatic fat content was quantified by AC measurements using the TAI technique. The AC values were reported in units of dB/cm/MHz and only the values with R2 ≥ 0.6 were considered reliable. Five measurements at mid-breath hold were obtained for AC and the median of the measurements was noted. The cut-off value of 0.75 dB/cm/MHz, found in comparison with Magnetic Resonance Imaging Proton Density Fat Fraction (MRI-PDFF) for predicting increased hepatic steatosis (> 5%) with TAI technique, was accepted as the cut-off value for hepatic steatosis [25]. Cases with increased hepatic steatosis detected by TAI technique (AC > 0.75 dB/cm/MHz) were classified as NAFLD.

The study protocol was approved by Local Ethics Committee (Date: 27.06.22, No: 520). Written informed consent was obtained from all participants.

When determining the number of individuals, it was planned to include each case and two controls per case due to the high prevalence of NAFLD in the community. Based on similar studies, when the effect power was determined to be 1.63 using Cohen’s method, 95% power, and a Type I error (alpha) of 0.05 in the analysis performed with the G-Power 3.1.9.4 program, it was determined that at least 8 and 16 individuals were required for the acromegaly and control groups respectively, much lower than the actual study.

Commercial statistical software, Statistical Package for the Social Sciences (SPSS) version 22.0 (IBM Corp., Armonk, NY, USA) was used for statistical analyses. The Shapiro–Wilk test was used to assess the conformity of continuous variables to normal distribution, while the homogeneity of variance was evaluated using Levene’s test. Continuous variables with a normal distribution were presented as the mean ± standard deviation (SD). Continuous variables that were not normally distributed were presented as the median and interquartile range (25-75th percentile). Student’s t-test was used for normally distributed continuous data and Mann—Whitney U test for not-normally distributed data.

The relationships between variables that did not provide the assumption of normality were evaluated using Spearman’s Rho correlation coefficient while the relations between variables that provided the normality assumption were evaluated with the Pearson correlation coefficient.

The effects of BMI, HOMA-IR, HDL cholesterol, TG, CK18, and IGFBP7 variables, which are thought to be associated with liver stiffness and steatosis, were planned to be evaluated using a stepwise multiple linear regression analysis. To ensure the assumption of normality, liver stiffness measurement was included in the model with an inverse transformation. Linear regression analysis could not be performed for the TAI variable due to its lack of normality. Due to the high correlation coefficient between glucose, HOMA-IR and HbA1c, HOMA-IR was included in the model.

The error rate (α = 0.05) was determined in all tests, and the difference between the groups was considered statistically significant when p < 0.05.