Abstract
Background:
Alternative BMI metrics are superior to BMI z score (BMIz) in tracking obesity but have not been evaluated in patients with nonalcoholic fatty liver disease (NAFLD). Our objective was to evaluate whether BMI-adjusted z score (BMIaz) or BMI expressed as a percentage of the 95th percentile (%BMIp95) are better predictors of degree of alanine aminotransferase (ALT) elevation, a surrogate for NAFLD severity, compared with BMIz in patients with NAFLD.
Methods:
A retrospective study of 776 subjects aged 2–18 years with BMIz > 1.0 followed in a NAFLD subspecialty clinic was conducted. Regression analysis was used to determine predictors of elevated ALT.
Results:
There was no association between BMIz, BMIaz, or %BMIp95 and degree of ALT elevation using linear or logistic regression.
Conclusion:
These results do not support the use of alternative BMI metrics for evaluating NAFLD severity. Future studies should investigate longitudinal assessments and correlation with histology.
Keywords: BMI, BMI z score, NAFLD, NASH, obesity
Introduction
Nonalcoholic fatty liver disease (NAFLD) severity ranges from steatosis to steatohepatitis with or without fibrosis, known as nonalcoholic steatohepatitis (NASH). NASH is becoming the leading indication for liver transplant in young adults.1 Obesity severity is linked to liver disease severity in pediatric NAFLD.2 Conversely, a BMI z score (BMIz) reduction of ≥0.25 in patients with NASH is associated with significantly increased odds of NASH resolution.3
BMIz, also known as BMI standard deviation (SD) score, is a BMI metric adjusted for age and sex.4 Pediatric literature has shown BMIz to be an imprecise metric in severe obesity. Freedman et al. showed BMIz to be a poor predictor of adiposity in children with severe obesity.5 Another study showed that BMIz tracking over time in children with severe obesity is weak, especially in patients younger than 10 years. They found that two alternative BMI metrics, BMI expressed as a percentage of the 95th percentile (%BMIp95) and BMI-adjusted z score (BMIaz), which is calculated by dividing the difference between the median BMI and observed BMI by a fixed SD,6 are superior to BMIz for tracking obesity severity over time.7 These alternative BMI metrics have not been evaluated in patients with NAFLD. Defining a superior obesity status measure in predicting NAFLD severity would be helpful to both clinicians and researchers.
The objective of this study was, therefore, to evaluate whether BMIaz or %BMIp95 are better predictors of the degree of alanine aminotransferase (ALT) elevation, a surrogate for NAFLD severity, compared with BMIz in patients with NAFLD.
Methods
A retrospective cohort study was performed in a single institution. Data were collected from children aged 2–18 years with BMIz > 1.0 followed for presumed or histologically confirmed NAFLD from January 1, 2010, to June 30, 2021. Presumed NAFLD was defined by radiological evidence of steatosis and/or elevated ALT in the context of a negative workup for other liver diseases. Patients were excluded if they had secondary NAFLD (e.g., secondary to medications), liver disease other than NAFLD, extrahepatic diagnoses that could affect aminotransferases, history of bariatric surgery or liver transplantation, history of type 1 diabetes mellitus, or current vitamin E supplementation. Transgender patients were excluded due to inability to accurately interpret their BMI metrics, which are dependent on genetic sex. Patients were excluded due to lack of documented height/weight and/or ALT level. Our institutional review board had approved this study before the initiation of study-related activities.
Data Collection
Data collected from the electronic health record included demographics (age, sex, ethnicity) and anthropometrics (height, weight, BMI) within 3 months of the first clinic visit. Histologically confirmed versus presumed NAFLD was recorded. Available fasting laboratory investigations were collected within 3 months of the visit and included: ALT, aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alkaline phosphatase, glucose, insulin, blood urea nitrogen, creatinine, uric acid, triglyceride, low-density lipoprotein (LDL), high-density lipoprotein, total cholesterol, and glycated hemoglobin (HbA1c).
Statistical Analyses
Descriptive statistics were used to determine the mean and SD or median and interquartile range of continuous variables, as indicated by distribution. Categorical variables are reported in absolute values and percentages. All laboratory values were log-transformed for regression analysis except for HbA1c, which was converted to a categorical variable (presence or absence of type 2 diabetes mellitus [T2DM]). Subjects were considered to have T2DM if HbA1c was ≥6.5%. Generalized linear regression was used to assess the association between log-transformed ALT and predictors.
Multiple linear regression analyses with backward variable selection were performed to identify predictors associated with response variables. Subanalysis was performed on subjects younger than 10 years. Logistic regression was used to assess association between ALT (binary outcome ALT >80 U/L or ALT ≤80 U/L) and predictors. Nominal two-sided p-values <0.05 were considered statistically significant. Statistical analyses were performed by using SAS® statistical software version 9.4 (SAS Institute, Cary, NC) and R (R Core Team, 2021).8
Results
There were 863 subjects aged 2–18 years with BMIz >1.0 followed for presumed or histologically confirmed NAFLD from January 1, 2010, to June 30, 2021. Of those, 49 subjects were excluded due to lack of ALT data, 15 due to liver disease other than NAFLD, 5 due to history of bariatric surgery, 8 due to vitamin E use, and 3 due to secondary NAFLD. Four transgender patients were excluded. Three additional patients were excluded, one each due to lack of height/weight measurements, history of liver transplant, and history of type 1 diabetes mellitus. There were 776 subjects remaining for analysis.
Baseline characteristics of the 776 subjects are presented in Table 1. The majority was non-Hispanic (76%) and white (86%), with a mean age of 12.6 years and a male predominance (66%); 36% (n = 277) had histologically confirmed NAFLD. Only 3% (n = 27) of subjects had a normal ALT level (≤22 U/L in females and ≤26 U/L in males). There were significantly more Hispanic patients in the younger subgroup (<10 years; 42% vs. 20%, p < 0.001). All subjects with T2DM (n = 26) were in the older subgroup. The median BMIz was 2.4, whereas the median BMIaz and %BMIp95 were 3.2 and 133%, respectively. The medians of all three BMI metrics were significantly higher in the younger subgroup compared with that in the older subgroup. The median ALT was 67 U/L and did not differ between subgroups.
Table 1.
Demographic and Clinical Characteristics in Children with Nonalcoholic Fatty Liver Disease by Age
| Characteristic | Totala children (n = 776) | Age group |
p b | |
|---|---|---|---|---|
| 2–9 Years (n = 136) | 10–18 Years (n = 640) | |||
| Age (years), mean (SD) | 12.6 (3.2) | 7.8 (1.5) | 13.7 (2.4) | n/a |
| Male sex, n (%) | 515 (66) | 83 (61) | 432 (68) | 0.16 |
| Hispanic ethnicity, n (%) | 57 (24) | 57 (42) | 128 (20) | <0.001 |
| Race: white, n (%) | 571 (86) | 81 (86) | 490 (86) | 1.0 |
| Comorbidity, n (%) | ||||
| Type 2 diabetes | 26 (3.7) | 0 (0) | 26 (3.7) | 0.008 |
| Anthropometrics | ||||
| BMI, kg/m2 | 32.8 (28.6, 37.7) | 26.9 (25.0, 30.8) | 34.0 (30.1, 38.6) | <0.001 |
| BMI z score | 2.4 (2.2, 2.6) | 2.5 (2.3, 2.8) | 2.4 (2.1, 2.6) | <0.001 |
| %BMIp95 | 133 (118, 150) | 136 (120, 155) | 132 (117, 149) | 0.04 |
| BMIaz | 3.2 (2.4, 4.1) | 3.6 (2.6, 5.3) | 3.1 (2.3, 4.0) | <0.001 |
| Laboratory | ||||
| ALT (U/L) | 67 (47, 97) | 67 (48, 89) | 67 (47, 99) | 0.79 |
| AST (U/L) | 38 (28, 55) | 39 (31, 55) | 37 (28, 54) | 0.11 |
| GGT (U/L) | 35 (23, 52) | 28 (20, 41) | 37 (25, 55) | <0.001 |
| Cholesterol, total (mg/dL) | 157 (135, 181) | 158 (133, 179) | 157 (136, 181) | 0.58 |
| Cholesterol, HDL (mg/dL) | 38 (32, 45) | 39 (33, 47) | 37 (32, 44) | 0.02 |
| Cholesterol, LDL (mg/dL) | 88 (70, 109) | 91 (71, 108) | 88 (70, 109) | 0.77 |
| Triglycerides (mg/dL) | 138 (92, 190) | 121 (86, 164) | 141 (95, 197) | <0.001 |
| Uric acid (mg/dL) | 5.8 (4.6, 6.9) | 4.5 (3.7, 5.2) | 6.1 (5.0, 7.0) | <0.001 |
| Glucose (mg/dL) | 90 (84, 97) | 88 (83, 93) | 91 (85, 97) | <0.001 |
| HOMA-IR (mg/dL × μU/mL/405) | 7.2 (2.8, 8.0) | 3.0 (1.9, 4.9) | 5.2 (3.2, 8.7) | <0.001 |
| HbA1c (%) | 5.3 (5.0, 5.5) | 5.2 (5.0, 5.4) | 5.3 (5.1, 5.5) | 0.004 |
Numbers in bold highlight statistical p-values <0.05.
Data are median (interquartile range) unless otherwise specified.
The significance of difference in categorical variables between age groups was tested with Fisher's exact test; continuous variables were analyzed with Mann–Whitney U test. All p-values are two-sided.
%BMIp95, BMI expressed as a percentage of the 95th percentile; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMIaz, BMI-adjusted z score; GGT, gamma-glutamyl transferase; HbA1c, glycated hemoglobin; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment for insulin resistance; LDL, low-density lipoprotein; n/a, not applicable; SD, standard deviation.
Univariate analysis found that T2DM in addition to log-transformed AST, GGT, glucose, insulin, total cholesterol, LDL, triglycerides, and uric acid were all significantly associated with log-transformed ALT. For subjects younger than 10 years, log-transformed AST, GGT, and total cholesterol were significantly associated with log-transformed ALT. None of the BMI metrics were significantly associated with log-transformed ALT in the entire cohort or in the age <10 subgroup. Figure 1 shows scatter plots of each BMI metric against ALT in the entire cohort (Fig. 1A–C) and the age <10 subgroup (Fig. 1D–F).
Figure 1.
Scatter plots of each BMI metric (BMIz, BMIaz, and %BMIp95) and log-transformed ALT level in the entire cohort (A–C) and in the subgroup of subjects <10 years (D–F). %BMIp95, BMI expressed as a percentage of the 95th percentile; ALT, alanine aminotransferase; BMIaz, BMI-adjusted z score; BMIz, BMI z score.
Multivariable analysis for both the entire cohort and the age <10 years subgroup showed that none of the BMI metrics were significantly associated with log_ALT. In each model containing a different BMI metric, T2DM and log_triglyceride level were significantly associated with log_ALT in the entire cohort. Estimates for T2DM (estimate 0.108, standard error [SE] 0.050, p = 0.030 in BMIz model) and log_triglyceride level (estimate 0.088, SE 0.039, p = 0.025 in BMIz model) were similar across models. For the age <10 subgroup, log_AST (estimate 0.900, SE 0.066, p < 0.001 in BMIz model) and GGT (estimate 0.106, SE 0.044, p = 0.018) were significantly associated with log_ALT to a similar degree in all three models (one for each BMI metric). Logistic regression using a binary outcome of ALT >80 U/L or ≤80 U/L also showed no significant association with any of the BMI metrics (data not shown).
Discussion
In this study, as expected, T2DM in addition to log_AST, GGT, glucose, insulin, total cholesterol, LDL, triglycerides, and uric acid were associated with elevated ALT. However, no association was found between the variables of interest, BMIz, BMIaz, or %BMIp95, and degree of ALT elevation in a large cohort of youth with NAFLD. Subanalysis of subjects younger than 10 years also yielded similar results. Logistic regression using a binary outcome of ALT >80 U/L or ALT ≤80 U/L also showed no association with BMI metrics.
The literature has shown BMIaz and %BMIp95 to be superior metrics to BMIz in the assessment of children with obesity. Freedman and Berenson compared the same three BMI metrics for tracking of obesity between examinations in a large cohort (n = 6994) and found both BMIaz and %BMIp95 to be superior to BMIz. BMIz was even weaker than the other two metrics in children younger than 10 years.7 In another study, in children with severe obesity, BMIz was a poor predictor of adiposity, as measured by anthropometry and dual X-ray absorptiometry (DXA). They found %BMIp95 and change in BMIp95 generally had stronger correlations with most measures of adiposity compared with BMIz.5 Vanderwall et al. evaluated the relationship of BMIz to change in percent body fat assessed using DXA in children with obesity over time, and despite good sensitivity (70%), found it to be an inadequate predictor due to poor specificity (42%).9
We used serum ALT levels to determine liver disease severity in children with NAFLD. While ALT is used in clinical practice to assess disease severity, to decide about proceeding with liver biopsy, and to evaluate response to treatment, it has also limitations as a biomarker. Previous literature has shown that no laboratory marker has sufficient sensitivity or specificity to replace liver biopsy in evaluating severity of pediatric NAFLD.10 However, ALT remains the most validated noninvasive biomarker for this purpose in pediatric NAFLD.3,11–13 In our study, even when using a higher threshold of ALT >80 U/L, which doubles the odds of NASH,14 regression analysis did not show an association with any BMI metrics.
This study was strengthened by its large sample size. Limitations include its retrospective design, resulting in missing data. Additionally, the study by Freedman and Berenson evaluated tracking of BMI metrics over time,7 whereas our study was cross-sectional. It is possible that in following changes in these metrics over time, one may be superior in terms of correlation to NAFLD severity and this would be important to investigate. While these results are not generalizable to patients with obesity without NAFLD and/or with NAFLD and normal ALT levels, they are still reflective of the majority of patients referred to subspecialty clinics for further evaluation for suspected NAFLD.
All patients followed at the Steatohepatitis Program undergo abdominal ultrasound to rule out structural liver disease; however, given that we do not rely on ultrasonographic evidence of steatosis for the diagnosis of NAFLD, as per the pediatric guidelines, we did not include ultrasound findings in the results.15,16 Although biopsy-confirmed NAFLD was not an inclusion criteria, this was by design, as limiting the study population only to those with histologically confirmed NAFLD would have created selection bias for those with the most severe disease and would not accurately represent most patients followed in subspecialty clinics for NAFLD.
In this large cohort of children with NAFLD, we found no association between BMIz, BMIaz, or %BMIp95 and degree of ALT elevation, a marker for NAFLD severity. While the recent literature has suggested moving away from the use of BMIz in favor of the other metrics.5,7,9,17 Our results suggest that no one metric of weight status is superior in predicting NAFLD severity. Future studies should investigate other potential clinical indicators of NAFLD severity. These include, but are not limited to, body composition analysis, as sarcopenia has been associated with pediatric NAFLD,18 as well as dietary factors such as fructose intake, which has also been linked to NAFLD presence and severity.19 It would also be useful to investigate whether there are correlations between imaging findings from magnetic resonance imaging or transient elastography with BMI metrics.
Acknowledgment
The authors thank Syeda Meryum for assistance with data collection.
Impact Statement
Alternative BMI metrics are superior to BMI z score (BMIz) in tracking obesity over time; however, this study demonstrates that BMIz, BMI-adjusted z score, and BMI expressed as a percentage of the 95th percentile are equally poor predictors of elevated alanine aminotransferase, a marker of nonalcoholic fatty liver disease severity, at a cross-sectional level.
Authors' Contributions
All authors approved the final article as submitted and agree to be accountable for all aspects of the work.
Funding Information
This work was supported by the National Institutes of Health grant P30 DK078392 (Clinical Component) of the Digestive Diseases Research Core Center in Cincinnati. The funding agency had no role in the design or conduct of the study; the collection, management, analysis, or interpretation of the data; the preparation, review, or approval of the article; or the decision to submit the article for publication.
Author Disclosure Statement
R.S. has received funds from Vivus Inc. The remaining authors have no conflicts of interest relevant to this article to disclose.
References
- 1. Doycheva I, Issa D, Watt KD, et al. Nonalcoholic steatohepatitis is the most rapidly increasing indication for liver transplantation in young adults in the United States. J Clin Gastroenterol 2018;52:339–346. [DOI] [PubMed] [Google Scholar]
- 2. Moran-Lev H, Cohen S, Webb M, et al. Higher BMI predicts liver fibrosis among obese children and adolescents with NAFLD—An interventional pilot study. BMC Pediatr 2021;21:385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Xanthakos SA, Lavine JE, Yates KP, et al. Progression of fatty liver disease in children receiving standard of care lifestyle advice. Gastroenterology 2020;159:1731..e10–1751.e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Must A, Anderson SE. Body mass index in children and adolescents: Considerations for population-based applications. Int J Obes 2006;30:590–594. [DOI] [PubMed] [Google Scholar]
- 5. Freedman DS, Butte NF, Taveras EM, et al. BMI z-scores are a poor indicator of adiposity among 2- to 19-year-olds with very high BMIs, NHANES 1999–2000 to 2013–2014. Obesity 2017;25:739–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Centers for Disease Control and Prevention (CDC). Modified z-scores in the CDC growth charts. https://www.cdc.gov/nccdphp/dnpao/growthcharts/resources/biv-cutoffs.pdf (Last accessed November 19, 2021).
- 7. Freedman DS, Berenson GS. Tracking of BMI z scores for severe obesity. Pediatrics 2017;140:e20171072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, 2021. https://www.R-project.org/ (Last accessed December 27, 2021).
- 9. Vanderwall C, Eickhoff J, Randall Clark R, et al. BMI z-score in obese children is a poor predictor of adiposity changes over time. BMC Pediatr 2018;18:187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Patton HM, Lavine JE, Van Natta ML, et al. Clinical correlates of histopathology in pediatric nonalcoholic steatohepatitis. Gastroenterology 2008;135:1961..e2–1971.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Arsik I, Frediani JK, Frezza D, et al. Alanine aminotransferase as a monitoring biomarker in children with nonalcoholic fatty liver disease: A secondary analysis using TONIC trial data. Children 2018;5:64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Newton KP, Lavine JE, Wilson L, et al. Alanine aminotransferase and gamma-glutamyl transpeptidase predict histologic improvement in pediatric nonalcoholic steatohepatitis. Hepatology 2021;73:937–951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Vuppalanchi R, Jain AK, Deppe R, et al. Relationship between changes in serum levels of keratin 18 and changes in liver histology in children and adults with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol 2014;12:2121..e2–2130.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Schwimmer JB, Newton KP, Awai HI, et al. Paediatric gastroenterology evaluation of overweight and obese children referred from primary care for suspected non-alcoholic fatty liver disease. Aliment Pharmacol Ther 2013;38:1267–1277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Vos MB, Abrams SH, Barlow SE, et al. NASPGHAN clinical practice guideline for the diagnosis and treatment of nonalcoholic fatty liver disease in children: Recommendations from the expert committee on NAFLD (ECON) and the North American Society of Pediatric Gastroenterology, Hepatology and Nutrition (NASPGHAN). J Pediatr Gastroenterol Nutr 2017;64:319–334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Draijer LG, Feddouli S, Bohte AE, et al. Comparison of diagnostic accuracy of screening tests ALT and ultrasound for pediatric non-alcoholic fatty liver disease. Eur J Pediatr 2019;178:863–870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kelly AS, Daniels SR. Rethinking the use of body mass index z-score in children and adolescents with severe obesity: Time to kick it to the curb? J Pediatr 2017;188:7–8. [DOI] [PubMed] [Google Scholar]
- 18. Pacifico L, Perla FM, Andreoli G, et al. Nonalcoholic fatty liver disease is associated with low skeletal muscle mass in overweight/obese youths. Front Pediatr 2020;8:158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. DiStefano JK, Shaibi GQ. The relationship between excessive dietary fructose consumption and paediatric fatty liver disease. Pediatr Obes 2021;16:e12759. [DOI] [PMC free article] [PubMed] [Google Scholar]