Page 1 of 25

Accepted Preprint first posted on 10 January 2014 as Manuscript EJE-13-0609 1

1

VITAMIN D LEVELS AND LIVER HISTOLOGICAL ALTERATIONS IN CHILDREN

2

WITH NON-ALCOHOLIC FATTY LIVER DISEASE (NAFLD)

3

Short Title: Vitamin D and NAFLD

4

5

6

Valerio Nobili1, Valentina Giorgio1, Daniela Liccardo1, Giorgio Bedogni1, 2, Giuseppe Morino3,

7

Anna Alisi1, and Stefano Cianfarani4, 5

8

9

1

Hepato-Metabolic Disease Unit, Bambino Gesù Children’s Hospital – IRCCS, Rome, Italy.

10

2

Clinical Epidemiology Unit, Liver Research Center, Basovizza, Trieste, Italy.

11

3

Dietetics Unit , Bambino Gesù Children’s Hospital – IRCCS, Rome, Italy.

12

4

D.P.U.O. “Bambino Gesù” Children’s Hospital – “Tor Vergata” University, Rome, Italy.

13

5

Department of Women’s and Children’s Health, Karolinska Institutet, Stockholm, Sweden.

14

Send correspondence to:

15

Stefano Cianfarani, M.D.

16

“Bambino Gesù” Children’s Hospital,

17

Piazza S. Onofrio 4, 00165-Rome, Italy.

18

Tel. +39 06 6859 3074, Fax. +39 06 6859 2508

19

E-mail: [email protected]

20 21 22 23 24 25 26

Key words: obesity; non-alcoholic fatty liver disease; non-alcoholic steatohepatitis; liver fibrosis; vitamin D.

Word count: 2,501 words

Copyright © 2014 European Society of Endocrinology.

Page 2 of 25

2

1 2 3 4 5 6

List of abbreviations BMD, bone mineral density; BMI, body mass index; CRN, clinical research network; DXA, dual-

7

energy X-ray absorptiometry; GFR, glomerular filtration rate; MS, metabolic syndrome; NAFLD,

8

nonalcoholic fatty liver disease; NAS, non-alcoholic steatohepatitis score; NASH, non-alcoholic

9

steatohepatitis; PTH, parathyroid hormone; SDS: standard deviation score; VD, vitamin D; VDR,

10 11 12

vitamin D receptor;

Page 3 of 25

3

1 2

ABSTRACT

3

Objective: To investigate the association between plasma vitamin D (VD) levels and histological

4

liver damage in children with nonalcoholic fatty liver disease (NAFLD).

5

Patients and Methods: In this cross-sectional study, performed in a tertiary care center for obesity,

6

73 consecutive overweight and obese children with persistently elevated serum aminotransferase

7

levels and diffusely hyperechogenic liver on ultrasonography underwent liver biopsy. Non-

8

alcoholic steatohepatitis (NASH) and fibrosis were histologically diagnosed using NAFLD Clinical

9

Research Network (CRN) criteria. Plasma levels of 25-OH-VD were measured by HPLC. Bone

10

mineral density (BMD) of lumbar spine was evaluated by dual-energy-X-ray absorptiometry.

11

Multiple linear regression analysis was used to evaluate the association between 25-OH-VD and the

12

predictors of interest after correction for age, gender, waist circumference, body mass index and

13

other potential confounders.

14

Results: The children (64% males) were aged 8 to 18 years and their median BMI was 2.45 SDS.

15

Both PTH and BMD were within the normal range. All cases of fibrosis were detected in children

16

with NASH. At multivariable linear regression with correction for age, gender and BMI, 25-OH-

17

VD levels were 9 (95%CI 12 to 6) ng/ml lower in NASH+ than in NASH- children (p < 0.001) and

18

9 (12 to 6) ng/ml lower in those with stage 1 vs. stage 0, and 9 (13 to 6) ng/ml lower in those with

19

stage 2 vs. stage 0 fibrosis (p < 0.001 for both).

20

Conclusion: Vitamin D concentrations are inversely associated with NASH and fibrosis in children

21

with NAFLD.

22

Page 4 of 25

4

1 2

INTRODUCTION

3

Vitamin D is a key nutrient for both healthy and chronically ill children (1). Sources of vitamin D

4

are diet and dietary supplements as well as skin 7-dehydrocholesterol after exposure to UVB

5

radiation. Despite this large availability, it is estimated that one billion people worldwide are

6

vitamin D deficient (2).

7

Vitamin D2, ergocalciferol, derives from plant sources and dietary supplements; Vitamin D3,

8

cholecalciferol, is produced in human skin or derives from animal sources or dietary supplements.

9

Vitamin D3 undergoes a three-step activation before interacting with its specific receptor (VDR).

10

After the conversion of skin 7-dehydrocholesterol into pre-vitamin D, there is a 25-hydroxylation in

11

the liver and a further 1-hydroxylation in the kidney. Vitamin D then interacts with VDR and

12

regulates the expression of more than 200 genes, mostly involved in apoptosis, cell growth and cell

13

differentiation.

14

NAFLD is the leading chronic liver disease worldwide, with an increasing prevalence that mirrors

15

that of obesity (3). NAFLD prevalence in obese children is estimated to be between 36 and 44%,

16

regardless of the method used to diagnose the disease (4,5). It is noteworthy that vitamin D

17

deficiency is more common in obese than in normal weight subjects (6). And a recent meta-analysis

18

has shown that NAFLD patients have decreased serum vitamin D concentrations (7).

19

In adults affected by NAFLD, vitamin D levels have been reported to be inversely associated with

20

liver steatosis, necroinflammation and fibrosis (8). In rats exposed to Westernized diets, vitamin D

21

deficiency exacerbates NAFLD through the activation of the toll-like receptor, possibly by means of

22

endotoxin exposure, causing insulin resistance, over expression of hepatic resistin, and up-

23

regulation of hepatic inflammatory genes (9, 10).

Page 5 of 25

5

1

To date, no robust data on the association between vitamin D and NAFLD are available in

2

childhood because of the difficulty in obtaining liver tissue specimens to be related to vitamin D

3

circulating levels. We previously reported preliminary data suggesting an inverse association

4

between vitamin D concentrations and liver fibrosis and necroinflammation in obese children (11).

5

The aim of the present study was to confirm and extend that preliminary observation by performing

6

a large-scale study in children with biopsy proven NAFLD. We investigated the association of

7

vitamin D with the histopathological features of NAFLD, taking into account potential confounders

8

such as age, gender, waist circumference and body mass index (BMI).

9

Page 6 of 25

6

1 2

3

SUBJECTS AND METHODS

4

Subjects

5

Seventy-three overweight or obese Caucasian children with biopsy-proven NAFLD were

6

consecutively enrolled into this cross-sectional study between January and March 2012. The setting

7

was a tertiary referral center for the study and treatment of obesity and metabolic syndrome.

8

Children with persistently elevated serum aminotransferase levels and/or diffusely hyperechogenic

9

liver on ultrasonography, were selected for liver biopsy. Exclusion criteria were hypothyroidism,

10

Wilson disease, HBV or HCV infection, cystic fibrosis, celiac disease, alpha-1-antitrypsin

11

deficiency, autoimmune hepatitis, use of known steatogenic drugs, and parenteral nutrition. Patients

12

taking vitamin and /or mineral supplement and/or medications known to influence vitamin D status

13

were excluded from the study.

14

Vitamin D levels were also assessed in two control groups. Control group A, sixty-four (40 males

15

and 24 females) age matched (median age, 12.8 years, 10 to 15 years) normal weight children

16

(mean BMI, 20 ± 2.1), recruited from the Ear Nose and Throat Pediatric Outpatient Clinic. Control

17

group B, 21 (13 males and 8 females), age matched (median age, 11.5 years, 6 to 15 years) obese

18

children (mean BMI, 29.3 ± 2.9), without NAFLD..

19

The study was approved by the Ethics Committee of the Bambino Gesù Children’s Hospital and

20

informed consent was obtained from the parents or legal guardians of the patients, or the patients

21

themselves when aged 18 years.

22

23

Page 7 of 25

7

1

2

Anthropometric and laboratory assessment

3

Weight, height and waist circumferences were measured following standard guidelines (12). BMI

4

was calculated and converted to standard deviation scores (SDS) using Italian reference data (13).

5

Fasting blood samples collected at the time of liver biopsy were obtained to measure serum levels

6

of glucose, insulin, HbA1c, cholesterol, LDL, HDL, triglycerides, creatinine, calcium, phosphate,

7

VD, PTH, AST, ALT, GGT. Simultaneous spot urinary samples were collected to measure urinary

8

calcium. Insulin resistance (IR) was estimated from the homeostasis model assessment (HOMA)

9

(14). Glucose was measured by standard methods, and insulin by means of radioimmunoassay

10

(MYRIA Technogenetics, Milan, Italy). Glomerular filtration rate (GFR) was estimated from

11

Schwartz’s formula (15). Blood fatty acids were analyzed in a drop of whole blood absorbed on a

12

strip and transmethylated for gas-chromatography analysis. 25-OH-VD was measured by HPLC

13

(Bio-Rad, Italy). This method measures 25-OH Vitamin D3 and 25-OH Vitamin D2 levels. Vitamin

14

D deficiency was defined as a value of 25-OH-VD < 20 ng/ml (16). Intact PTH was measured by

15

chemiluminescence immunoassay (Siemens). Normal values of PTH in our laboratory are

16

comprised between 9 and 55 pg/ml. The metabolic syndrome was diagnosed according to the

17

International Diabetes Federation (17).

18

19

Dual-energy X-ray absorptiometry

20

DXA measurements of lumbar spine were performed using a QDR-4500 scanner (Hologic Inc.,

21

Waltham, MA) to obtain bone mineral density (BMD, g/cm2) at L1–L4 level. BMD was calculated

22

using the reference values provided by the scanner software (18).

23

Page 8 of 25

8

1

2

Liver ultrasonography

3

Liver ultrasonography was performed at most 1 month before liver biopsy using an Acuson S2000

4

system (Siemens, Munich, Germany) with linear and convex transducers (frequency bandwidth 4–

5

14 MHz). Absent steatosis (grade 0) was defined as normal liver echo-texture; mild steatosis (grade

6

1) as slight and diffuse increase in fine parenchymal echoes with normal visualization of diaphragm

7

and portal vein borders; moderate steatosis (grade 2) as moderate and diffuse increase in fine echoes

8

with slightly impaired visualization of diaphragm and portal vein borders; and severe steatosis

9

(grade 3) as fine echoes with poor or no visualization of diaphragm, portal vein borders and

10

posterior portion of the right lobe (19).

11

12

Liver histopathology

13

Liver tissue specimens were fixed in buffered formalin, embedded in paraffin, sliced into 3-mm

14

sections, and stained with hematoxylin and eosin, periodic Schiff acid with and without diastase,

15

and Van Gieson trichrome stains. The histological features of NAFLD were classified using the

16

NAFLD Clinical Research Network (CRN) system (20). Steatosis was graded as: 0 = involving

17

66% of hepatocytes.

18

Lobular inflammation was graded as: 0 = no foci; 1 = < 2 foci per 200X field; 2 = 2–4 foci per

19

200X field; 3 = > 4 foci per 200X field. Portal chronic inflammation was also evaluated as: 0 = no;

20

1 = yes. Hepatocyte ballooning was graded as: 0 = none, 1 = few balloon cells, 2 = many/prominent

21

balloon cells. Fibrosis was staged as: F0 = no fibrosis; F1 = perisinusoidal or periportal; F2 =

22

perisinusoidal and portal/periportal; F3 = bridging; and F4 = cirrhosis. The non-alcoholic

23

steatohepatitis score (NAS) was obtained by summing steatosis, lobular inflammation and

Page 9 of 25

9

1

ballooning scores. NASH was diagnosed by a liver pathologist following the NASH CRN

2

recommendations (20).

3

4

Statistical analysis

5

Most continuous variables were not normally distributed and all are reported as 50th, 25 th and 75 th

6

percentiles. Categorical variables are reported as counts and percentages. Between-group

7

comparisons of medians were performed using quantile regression. Multiple linear regression was

8

used to evaluate the association between vitamin D (continuous, ng/ml) and the four predictors of

9

interest (PTH, NAS, NASH and fibrosis) after correction for age (continuous, years), BMI

10

(continuous, SDS) and gender (discrete, 0 = female; 1 = male). In separate regression models we

11

also addressed the independent effects of GFR, HOMA, waist circumference and the metabolic

12

syndrome. Among predictors, PTH was modeled as continuous (ng/dl), NAS as continuous

13

(arbitrary units), NASH as discrete (0 = no; 1 = yes) and fibrosis as discrete (0 = F0; 1 = F1; 2 =

14

F2). Multivariable fractional polynomials were used to test whether the relationships of vitamin D

15

with continuous predictors were linear (21). We found that all relationships were linear and we

16

modeled them as such. Statistical analysis was performed using Stata version 12.1 (Stata Corp,

17

College Station, TX, USA). Statistical significance was set to a p-value < 0.05.

18

Page 10 of 25

10

1 2

RESULTS

3

Seventy-three children (64% males) aged 8 to 18 years were consecutively enrolled into the study.

4

The children were mostly males (n= 47, 64%) and their measurements are given in Table 1 with

5

and without stratification for vitamin D status.

6

34/73 (47%) children with NAFLD showed low vitamin D levels whereas 39/73 (53%) had normal

7

vitamin D concentrations.

8

Table 1 here

9

Not surprisingly, the children with low vitamin D levels had higher levels of PTH (p < 0.001) than

10

those with normal vitamin D but PTH values were within the normal range in all cases. Although

11

children with low vitamin D levels had lower levels of ALT (p < 0.001), this is not clinically

12

relevant as we have shown the lack of association between ALT and liver histopathology (22).

13

Vitamin D levels in both control groups were within the normal range (Group A, mean: 29 ± 3.3

14

ng/ml; Group B, 29.1 ± 8.3).

15

53 (73%) children had a large waist circumference, 2 (3%) high blood pressure, 3 (4%) high

16

glucose, 22 (30%) high triglycerides and 26 (36%) low HDL and 10 children (14%) had 3 or more

17

of the above, i.e. MS (17). The histopathological features of the children are given in Table 2. Table 2 here

18

19

NASH was present in 67% of children and fibrosis of any degree (F1 or F2) in 67% of cases. All

20

cases of fibrosis occurred in NASH patients.

21

The associations between vitamin D and PTH, NAS, NASH and fibrosis after correction for

22

gender, age, waist circumference and BMI are depicted in Figure 1. The corresponding regression

23

models are given in Supplemental Data 1.

Page 11 of 25

11

Figure 1 here

1

2

3

An increase of 1 pg/mL of vitamin D was associated with a mean decrease of PTH of 0.2 ng/ml (p