Hepatic Fibrosis: A Manifestation of the Evolution of Liver Disease in Patients with Ataxia-Telangiectasia

Talita Lemos Neves Barreto (  talita.lemos@unifesp.br ) UNIFESP: Universidade Federal de Sao Paulo https://orcid.org/0000-0001-5440-4615 Roberto José de Carvalho Filho UNIFESP: Universidade Federal de Sao Paulo David Carlos Shigueoka UNIFESP: Universidade Federal de Sao Paulo Fernando Luiz Affonso Fonseca Faculdade de Medicina do ABC Ariel Cordeiro Ferreira Faculdade de Medicina do ABC Cristiane Kochi Faculdade de Ciencias Medicas da Santa Casa de Sao Paulo Carolina Sanchez Aranda UNIFESP: Universidade Federal de Sao Paulo Roseli Oselka Saccardo Sarni Faculdade de Medicina do ABC

proposal especially for A-T patients [17,18]. These noninvasive methods have been recommended for the identi cation of hepatic brosis, with varied availability and accuracy and that allow longitudinal evaluation and monitoring the liver disease progression [19][20][21][22][23].
To our knowledge, there are no publications to date evaluating imaging and liver brosis biomarkers in A-T patients. Thus, the aim of this study was to identify the frequency of signi cant hepatic brosis and to verify an association with metabolic changes and the degree of ataxia in A-T patients.

Methods
This is a cross-sectional study that included 25 A-T patients from both genders, aged between 5 and 31 years, who met the diagnostic clinical criteria of the European Society for Immunode ciencies (ESID), under a multidisciplinary follow-up at the Division of Allergy, Clinical Immunology, Rheumatology of the Department of Pediatrics of Universidade Federal de São Paulo (UNIFESP/EPM) [24]. Individuals undergoing oncological treatment, using hepatotoxic drugs or those with hepatitis B or C were excluded.
The outpatient clinic currently follows 26 A-T patients and only one patient was excluded from the study for undergoing cancer treatment.
The study was approved by UNIFESP Research Ethics Committee (0081/2018). All the caregivers of patients signed an informed consent to be enrolled in this study.
Anthropometric and pubertal development assessment Anthropometric measurements included weight, height and skinfolds (tricipital, bicipital, subscapular and suprailiac) and the mid-upper arm circumference and waist circumference (WC) [25,26]. The patients who were unable to stand upright were weighed on a digital wheelchair scale (Micheletti® electronic weighing platform for up to 500 kg). Recumbent height measurements were taken in supine position on a at, rm surface by using a non-extensible tape (in millimeters).
For nutritional status classi cation, body mass index (BMI)-for-age and height-for-age z-scores for children and adolescents were calculated. Adults were classi ed according to BMI [25,27]. The sum of the skinfold thickness was used to estimate the body fat percentage [28][29][30][31].
The WC was classi ed as altered when the WC to height ratio (WHtR) was equal or higher than 0.5 [32,33]. Mid-upper arm circumference combined to the tricipital skinfold was used to estimate the mid-upper arm muscle circumference (MUAMC) [26,31].
The stage of pubertal development was self-assessed according to Marshall & Tanner [34].

Neurological assessment
The International Cooperative Ataxia Rating Scale (ICARS) was applied in the adapted and validated version for the Brazilian culture to evaluate ataxia severity in all patients by a skilled physical therapist. ICARS has 19 items subdivided into four subscales with a maximum score of 100: posture and gait disturbances; kinetic functions; speech disorders and oculomotor disorders. For the severity classi cation, the following cut-off points were adopted: mild ataxia (1 to 30 points), moderate ataxia (31 to 60 points) and severe ataxia (> 60 points) [35].

Laboratory tests
Blood was collected, after an 8-hour fasting, by peripheral venous puncture for biomarker analysis. All analyses were performed with standardized methods and according to good practices in clinical analyses.

Lipid metabolism biomarkers
Triglycerides (TG); total cholesterol (TC); low-density lipoprotein (LDL-c); high-density lipoprotein of cholesterol (HDL-c) and non-HDL-c cholesterol (NHDL-c) which was calculated by subtracting the TC values from the HDL-c values.
For classi cation, the cut-off points suggested by the American Academy of Pediatrics and the National Cholesterol Education Program (NCEP) were adopted [39,40]. The NHDL-c was classi ed according to Bogalusa and the NCEP [40,41].

Glucose metabolism biomarkers
The oral glucose tolerance test (OGTT) was performed with insulin dosage. After an 8-hour fast, 1.75 g/kg of dextrose (maximum dose of 75 g) was administered orally. Blood samples for glycemic and insulin curves were obtained immediately after dextrose intake, in fasting and at 30, 60, 90 and 120minutes intervals.
Glucose intolerance was considered when values between 140 mg/dL and 199mg/dL were found in 120 minutes. Diabetes was considered when fasting or at 120 minutes, values equal to or greater than 126mg/dL and equal to or greater than 200 mg/dL, respectively [42].
Insulin resistance (IR) was assessed using the following indices: Homeostasis Model Assessment -Insulin Resistance (HOMA-IR), Homeostasis Model Assessment -Adiponectin (HOMA-AD) and Matsuda that estimates whole-body insulin sensitivity [43][44][45]. For HOMA-AD classi cation, the cut-off points >8.6 for children and >14.3 for adolescents and adults were considered as suggestive of IR [46].

Liver imaging tests
Hepatic ultrasonography Ultrasonography was performed for grading liver steatosis. The examination was performed by a single examiner (radiologist) using GE healthcare equipment, Logiq P6 model, 5 MHz multifrequency convex transducer. The changes in the liver parenchyma were strati ed in the form of score (1-9): normal liver (0), slight diffuse steatosis (1-3), moderate (4-6) and accentuated (7)(8)(9). This classi cation uses as parameters the liver echogenicity in comparison with renal parenchyma and posterior sound attenuation in the liver parenchyma, which can be determined by the degree of visibility of the diaphragm and hepatic vessels [47].
After a 2-hours-minimum fast, the patients were put in dorsal decubitus, with the right arm raised to facilitate access to the right hepatic lobe. A M-size probe was positioned between the 9th and 11th intercostal spaces between the anterior and middle axillary lines. In each examination, to obtain at least 10 valid measurements, a maximum of 20 measurements were performed.
For reliability analysis of the TE, the following parameters were used: success rate (SR) which would be the equivalent of the quotient between the number of valid measurements and the total number of obtained measurements expressed in percentage and interquartile range to median of measurements ratio (IQR/Md).
The following reliability criteria were adopted: minimum of 10 valid measurements; SR ≥ 60%; IQR/Md < 30% for any liver stiffness value; and IQR/Md > 30% for liver stiffness values below 7.1 kPa [52]. TE results that did not meet these criteria were disregarded.
A-T patients who presented by TE ≥ 7 kPa (cutoff point used for NAFLD) or alteration of the APRI and one of the scores evaluated (NFS or BARD) simultaneously were considered with signi cant liver brosis [19,[36][37][38]. TE results were assessed and interpreted along with the clinical and laboratory aspects of each patient.

Statistical analysis
Data were entered and consolidated in an Excel spreadsheet (O ce Microsoft ®) and analyzed using the statistical package SPSS 19.0 (IBM®). Categorical variables were presented as absolute and percentage values. Continuous variables were analyzed for its normality using Shapiro-Wilk test. For comparisons between the two groups of patients, with or without signi cant liver brosis, the variables with parametric distribution were presented in the form of mean and standard deviation and compared by the independent t-Student test and the variables with non-parametric distribution were presented as median (minimum and maximum) and compared by the two-tailed Mann-Whitney U test. Spearman's correlation coe cient was used to assess the correlations. The statistical signi cance level of 5% (p<0.05) was adopted. (42%) had short stature for age. According to body composition, only 4/24 (16.7%) had a low percentage of body fat. However, lean mass impairment was observed in 13/24 (54%) of the patients. Regarding pubertal development 8/25 (32%) were pubescent.

Results
According to metabolic biomarkers, 16/25 (64%) had dyslipidemia, 4/22 (18%) were diabetic (three diagnosed by OGTT performed in this study), 5/17 (29%) had IR according to HOMA-AD and 5/25 (20%) had values equal to or greater than twice the upper limit of normal for AST and ALT enzymes. Hepatic steatosis by ultrasonography was found in 13/20 (65%) and suggestive diagnosis of signi cant liver brosis in 5/25 (20%) of A-T patients. TE result of only one patient was disregarded for not meeting the adopted reliability criteria. Table 1 shows the characteristics of A-T patients.  Table 2 shows the values of the parameters obtained by TE and the suggestive scores of liver brosis in patients considered to have suggestive diagnosis of signi cant liver brosis. p=0.005), HOMA-AD (rho=0.485; p=0.041) and indirect with adiponectin (rho= -0.613; p=0.003); a fact not observed for the APRI (rho=0.432; p=0.051) and ALT (rho=0.417; p=0.060). Figure 1 shows the correlation between the median LSM with the HOMA-AD, NFS and the ICARS score. A-T patients were divided into two groups according to the presence or absence of suggestive of signi cant liver brosis to compare the variables evaluated in this study. A signi cant difference was found between groups for age (p<0.001), MUAMC (p<0.001), WC (p=0.008), ICARS score (p=0.009), platelets (p=0.027), albumin (p= 0.019), HDL-c (p=0.013), LDL-c (p=0.049), AST (p=0.001), ALT (p=0.002), GGT (p=0.001), ferritin (p=0.001), 120-minutes glycemia (p=0.049), Matsuda index (p=0.044) and HOMA-AD (p=0.016) ( Table 3). Table 3. Comparison of the patients with Ataxia-telangiectasia according to the presence or absence of a suggestive of signi cant liver brosis Liver biomarkers, in ammation and HOMA-IR, HOMA-AD and Matsuda indices were compared between patients with and without liver steatosis assessed by ultrasound, with no statistically signi cant difference being observed between groups for any of the variables.
In calculating the accuracy of the ultrasound assessment of steatosis as a predictor of signi cant liver brosis, a sensitivity of 60% (CI95% 14.7-94.7 and a speci city of 33% (CI95% 11.8-61.6) was observed. This nding suggests that the presence of hepatic steatosis on ultrasound does not seem to contribute to the screening of hepatic brosis in these patients.  (Figure 2). Figure 2 Median values of the controlled attenuation parameter (CAP) between patients without (n=17) and with suggestive of signi cant brosis (n=4). *Signi cance level by Student's t-test (a). Median CAP values between patients without (n=7) and with ultrasound liver steatosis (n=11). *Signi cance level of the Mann-Whitney U test (b).
In calculating the accuracy of values equal to or greater than twice the upper limit of normal for AST and ALT enzymes as predictors of signi cant liver brosis, a sensitivity of 80% (CI95% 28.4-99.5) and speci city of 95% (CI95% 75.1-99.9), which suggests that the increased values of these enzymes may contribute to the screening of liver brosis in these patients.

Discussion
The present study showed that 20% of the A-T patients evaluated presented a suggestive diagnosis of signi cant liver brosis. It is important to emphasize that all the patients with suggestive of liver brosis had dyslipidemia and four had diabetes. The presence of suggestive diagnosis of brosis was associated with the severity of ataxia and with higher values of liver enzymes, ferritin, 120-minutes glycemia by OGTT and the HOMA-AD and lower values for the Matsuda index, compared to A-T patients without liver brosis.
Regarding the liver biomarkers evaluated, only liver enzymes (ALT, AST and GGT) were elevated in patients with a suggestive diagnosis of signi cant liver brosis. Values equal to or greater than twice the upper limit of normal for AST and ALT enzymes as predictors of signi cant liver brosis revealed a sensitivity of 80% and a speci city of 95%. In a previous study performed by our group, the values of the ALT and AST have shown alterations mainly during adolescence in A-T patients [5]. In a retrospective study with A-T patients conducted by Donath et al. a steady upward evolution of ALT and GGT was observed, especially from the age of 12 [9]. In another retrospective study, elevation of liver enzymes was observed in younger A-T patients with a mean age of 9.97±5.09 years and a signi cant association with the presence of dyslipidemia [10]. Therefore, it is recommended periodic monitoring of the AST and ALT liver enzymes from the age of 10, which may contribute to the screening of liver brosis in A-T patients In the present study, patients diagnosed with diabetes had fasting glucose within the normal range and were only identi ed by OGTT with 120-minutes glycemia above 200 mg/dL. The ndings suggest the importance of performing the OGTT for early identi cation and treatment of diabetes, especially from adolescence. A cohort of A-T patients showed a progressive increase in glycated hemoglobin (HbA1c) and fasting glucose with advancing age. OGTT was considered to have good sensitivity for IR screening and the HbA1c a marker to assess the therapy response [53].
Regarding the indices used to assess IR, the values of the HOMA-AD and Matsuda differed signi cantly between patients with and without a suggestive diagnosis of signi cant liver brosis; fact not observed for HOMA-IR. A controlled study has also veri ed lower values of Matsuda index in A-T patients compared to healthy individuals (5.96 6 ± 0.77 vs.11.03 ± 1.69; p=0.019) and similar values of HOMA-IR between the groups [54]. A recent study conducted with Brazilian children and adolescents for metabolic syndrome screening has found a better performance of HOMA-AD compared to HOMA-IR [46]. Furthermore, Hung et al. observed that HOMA-AD appears to be sensitive in detecting small changes in insulin sensitivity in patients with or without diabetes [55]. Therefore, to assess IR in A-T patients by HOMA-IR seems not to be the most appropriate method.
Regarding in ammatory biomarkers, only ferritin had higher values in patients with suggestive of liver brosis compared to those who did not. Experimental study with the aim of investigating the iron regulation, regulatory genes, and markers of oxidative stress in the liver tissue of ATM-de ciency mice, described higher values of serum and hepatic iron, ferritin, and hepcidin when compared to controls. This study suggested that the increase in tissue iron would be associated with hepatic oxidative stress resulting from iron-induced increase in hepcidin, which can suppress its export by ferroportin, which is considered a protective mechanism in response to oxidative stress [56]. Therefore, it is suggested that the increase in ferritin may contribute to the chronic oxidative stress presented by A-T patients and, consequently, to the development of liver disease.
Another noteworthy fact was the signi cant and direct correlation between the median LSM obtained by TE and the ICARS score. In a retrospective study aiming to determine the evolution of liver disease and its relationship with age and neurological impairment in A-T patients, a signi cant and direct correlation was found of the Klockgether Ataxia Scale (KAS) score with age, AFP, GGT, and ALT [9]. In a recent study conducted by our group, correlations were found between the severity of ataxia with age and metabolic changes including impairment of liver damage markers and IR in A-T patients [57].
Regarding the CAP values obtained by TE, no signi cant difference was observed between patients with and without suggestive diagnosis of signi cant brosis. By ultrasound, three of the ve patients with suggestive of liver brosis presented steatosis. Therefore, only the presence of hepatic steatosis by both ultrasound and CAP was not related to hepatic brosis in A-T patients, which suggests that the presence of steatosis is not a predictive factor for brosis in these patients.
One of the mechanisms of ATM protein activation is attributed to the action of reactive oxygen species (ROS), resulting in increased concentrations of antioxidants and the repair of oxidative DNA damage. In the absence/de ciency of ATM, A-T patients have low antioxidant capacity and as a result macromolecules, lipids and DNA are exposed to constant oxidative stress and its damages [58-60]. NAFLD and A-T share a similar pathogenic mechanism of ROS generation and mitochondrial dysfunction that contributes to the development of lesions [61]. Thus, it is postulated that oxidative stress has a relevant contribution in the genesis of liver disease associated with A-T.
The mechanisms involved in the evolution of NAFLD are not fully known. Changes in proteins such as ATM could play a role in these mechanisms since this protein is associated with DNA integrity and mitochondrial homeostasis. An experimental study showed that ATM-de cient mice presented a reduction and delay in DNA replication during liver regeneration. Furthermore, when partial hepatectomy was performed, an increase in apoptosis was observed, which indicates that the ATM protein is involved in the regeneration and survival of hepatocytes [62].
A recent study that analyzed the expression of messenger RNAs, total proteins, or phosphoproteins related to the ATM pathway of individuals with healthy liver, hepatic steatosis, and NASH, found a causal association between the ATM pathway and NAFLD. During the steatosis phase, there was low ATM activation, which caused mitochondrial dysregulation and greater DNA damage, in addition to reduced growth of the hepatocyte. In NASH, there was greater ATM depletion, with a greater degree of DNA damage and cell growth arrest due to the action of ATM in the cell cycle. In addition, to compensate for hepatocyte growth arrest, pre-oncogenic cells appeared, with a high rate of proliferation [63]. Therefore, the ATM protein appears to play an important role both in the beginning and progression of NAFLD, including its evolution to HCC.
This study has as strengths the fact that it is unprecedented in the sense of describing a suggestive diagnosis of liver brosis in A-T patients through non-invasive methods and prospective data collection. As limitations, we can consider the absence of genotyping of the ATM gene variants and the small number of patients with suggestive of liver brosis, which limited the analysis.

Conclusions
Suggestive ndings of signi cant liver brosis by non-invasive markers were observed in 20% of A-T patients. The presence of brosis was associated with alterations in liver enzymes, ferritin, increase in the HOMA-AD, and severity of ataxia. The ndings lead to the importance of monitoring liver and metabolic changes through non-invasive imaging and laboratory methods, especially from adolescence.

Consent for publication
Patients and parents gave consent to be included in the study through consent form.

Availability of data and material
All data generated or analysis during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests. All authors read and approved the nal manuscript.   Abbreviations: CAP (controlled attenuation parameter), IQR (interquartile range), LSM (liver stiffness measurement), IQR/Md (interquartile range to median measurements ratio), APRI (aspartate aminotransferase to platelet ratio index) and NFS (nonalcoholic fatty liver disease brosis score). *Transient elastography was not performed in patient 5.   Figure 1