The Tri-ponderal Mass Index is associated with adiposity in adolescent type 2 diabetes mellitus: a cross-sectional analysis

Pediatric type 2 diabetes mellitus (T2DM) patients are often overweight or obese, yet there are no validated clinical measures of adiposity to stratify cardiometabolic risk in this population. The tri-ponderal mass index (TMI, kg/m3) has recently been reported as a measure of adiposity in children, but there has been no validation of the association of TMI with adiposity in pediatric T2DM. We hypothesized that in children with T2DM, the TMI can serve as a more accurate measure of adiposity when compared to BMI z-score, and that it is associated with components of the metabolic syndrome. This is a cross-sectional secondary data analysis from the Improving Renal Complications in Adolescents with Type 2 Diabetes Through REsearch (iCARE) study (n = 116, age 10.20–17.90 years). Spearman’s correlations and multivariable regression were used in the analyses. When compared to DXA, TMI demonstrated significant correlation with total adiposity versus BMI z-score (TMI r = 0.74, p-value < 0.0001; BMI z-score r = − 0.08, p-value 0.403). In regression analyses, TMI was associated with WHtR (B = 35.54, 95% CI 28.81, 42.27, p-value < 0.0001), MAP dipping (B = 1.73, 95% CI 0.12, 3.33, p-value = 0.035), and HDL (B = − 5.83, 95% CI − 10.13, − 1.54, p-value = 0.008). In conclusion, TMI is associated with adiposity and components of the metabolic syndrome in pediatric T2DM patients.

Pediatric type 2 diabetes mellitus (T2DM) rates are rising around the world, and its emergence has mirrored the global rise in childhood obesity [1][2][3] . Certain pediatric populations are impacted disproportionately by T2DM rates; for example, in Canada, Indigenous youth account for almost half of all newly diagnosed T2DM cases annually 4 .
Pediatric T2DM is a more aggressive disease than adult T2DM, and may present with several concomitant comorbidities including dyslipidemia, hypertension, non-alcoholic fatty liver disease, and with complications such as early nephropathy [5][6][7][8] . Importantly, these patients may have increased carotid intima-media thickness, an early risk marker of future cardiovascular disease [9][10][11] .
In adult population-based studies, adipose tissue expansion in obesity is an important risk factor in the development of diabetes-related cardiovascular disease via several mechanisms including atherosclerosis, inflammation, and insulin resistance [12][13][14][15] . In particular, the expansion of the visceral adipose compartment has been proposed as a stronger predictor of adverse cardiometabolic outcomes when compared to body mass index (BMI) and total adiposity [15][16][17][18] . While both waist-to-hip ratio (WHR) and waist-to-height ratio (WHtR) are used as clinical measures of central adiposity, the latter has emerged as a more robust predictor of adiposity and cardiometabolic risk factors in children when compared to BMI 19,20 .
There is a limited set of measures to quantify adiposity in children. The body mass index (BMI) z-score, a clinical measure of obesity in children, often misclassifies children as obese 21 . It is also a poor predictor of change in adiposity over time 22 . Furthermore, the use of dual-energy x-ray absorptiometry (DXA) scans, which Table 1. Participants' descriptive characteristics. BMI body mass index, TMI tri-ponderal mass index, WHR waist-to-hip ratio, WHtR waist-to-height ratio, BMC bone mineral content, FM% fat mass percentage, bpm beats per minute, BP blood pressure, mmHg millimeters of mercury, MAP mean arterial pressure, SD standard deviation.  All participants received lifestyle modification intervention. The majority of participants were on insulin therapy (n = 103), while 10 were on metformin, two on glyburide, and one on gliclazide. The treatment choice reflected the glycemic control status.
The association of TMI and BMI z-score with DXA-measured adiposity. To assess the strength of the relationship between the clinical and DXA-based measures of adiposity, we performed correlation analyses ( Table 3, Fig. 1). TMI correlated strongly with total adiposity (FM%; r = 0.74, p-value < 0.001). In addition, TMI correlated strongly with WHtR (r = 0.85, p-value < 0.001) and to a lesser degree with WHR (r = 0.26, p = 0.005).
The BMI z-score did not significantly correlate with the FM% (r = − 0.08, p-value 0.403) (Fig. 2), and had a weak but positive correlation with WHR (r = 0.19, p-value = 0.050). The BMI z-score did not correlate with the WHtR (r = 0.11, p-value = 0.267). Of note, TMI did not correlate with BMI z-score (r = 0.09, p-value 0.378). In summary, the TMI demonstrated a stronger correlation with total and central adiposity measures than BMI z-score.
The association of TMI with components of the metabolic syndrome. To assess the association of TMI with components of the metabolic syndrome, we performed correlation analyses. The TMI had a small negative correlation with HDL levels (r = − 0.26, p-value = 0.005), but did not correlate with other components of the lipid profile including total cholesterol, LDL, triglycerides, and ApoB:Apo A ( Table 3).
Taken together, these data demonstrate that TMI has stronger association with measures of total and central adiposity when compared to BMI z-score. In addition, TMI correlated with components of the metabolic syndrome frequently noted in T2DM patients.

Discussion
The emergence of T2DM in youth is a global phenomenon that has been accelerating over the past few years. As diabetes is a major driver of adverse cardiometabolic outcomes in the general population, and youth with T2DM are expected to live longer with their diagnosis than their adult counterparts, there is an urgent need to define www.nature.com/scientificreports/  www.nature.com/scientificreports/ the drivers of cardiometabolic outcomes in this population. While adiposity is a major driver of cardiometabolic risk in the general population, its measurement has relied on using technologies that are not widely accessible. Compared to DXA scan-based adiposity measurement, TMI has emerged as a novel clinical tool to measure adiposity in adolescents with T2DM. TMI has already been validated against DXA and bioelectrical impedance as an adiposity measurement tool in healthy pediatric populations, and our results corroborate its value as an adiposity measure in youth with T2DM [27][28][29] .
The use of TMI to measure adiposity in pediatric populations offers several advantages. While the BMI z-score is the most widely used surrogate marker of adiposity in children, this approach was adopted from BMI use as a marker of adiposity in adult populations. Adult BMI relies on the regression of weight on a constant height squared, which has some caveats yet is the most widely used measure to define obesity in population studies and clinical care settings 30 . However, the three dimensional nature of growth in children, including height gain, makes the regression of weight to height cubed, as with TMI calculation, a more accurate measure of adiposity 31 . In addition, TMI cut-offs are age-independent and sex-specific. Having a constant value in children and adolescents, although its value can be population-specific to define normal and excess adiposity, is an important criterion to assess adiposity during childhood 27,29,32 .
One of the important benefits of using TMI as a measure of adiposity is that its components are generated using stadiometers and weight scales, devices that are already part of routine pediatric clinical practice. This ease of calculation provides a powerful practical measure of adiposity that is based on routine anthropometric testing in the clinical setting and helps avoid the need for specialized and costly methodologies. In addition, the estimation of adiposity using TMI improves at higher fat mass levels 27 , which makes it an effective measure in the T2DM population as they typically have significant adiposity 33 .
TMI correlated with WHR and WHtR, important central adiposity measures, and with other metabolic syndrome components including blood pressure and HDL. The WHtR is one of the most reliable clinical measures of central adiposity 34,35 , which is associated with cardiometabolic disorders 12,15,17,18,36 .  www.nature.com/scientificreports/ The association of TMI with measures of central adiposity is an important finding, as previous evidence linked TMI to total adiposity assessments only. This may potentially allow the use of TMI as a prediction tool for metabolic syndrome and cardiometabolic comorbidities in T2DM patients 37,38 ; however, this requires further validation.
A limitation of this study is that the use of WHtR ratio, a surrogate marker of central adiposity, was not validated against more accurate methods such as magnetic resonance imaging (MRI). The cost of such modalities is an important consideration in determining the feasibility of their measurement, yet it would be an important question to address in future studies.
TMI cut offs for determining overweight and obesity are calculated from the specific population under study. In our sample that is primarily composed of Indigenous youth in Canada, the generalizability of the cut offs proposed in our study may not be applicable to other populations, and further studies are required to determine the appropriate cut off for overweight and obesity for different ethnic groups. In addition, the cross-sectional nature of the study limits the determination of TMI as a tool to predict future cardiometabolic outcomes in this population.
Longitudinal data from a healthy pediatric population assessed whether childhood TMI can predict adult cardiometabolic risk. The TMI was associated with adult obesity, T2DM, high low-density lipoprotein, and increased carotid intima-media thickness 39 . However, the TMI performed equally well to BMI. It is uncertain whether the use of TMI in a population that already has T2DM during childhood may have higher predictive ability of future cardiometabolic diseases. This will require longitudinal follow-up data for the children with T2DM.
In conclusion, TMI is associated with total and central adiposity as well as markers of the metabolic syndrome in pediatric T2DM. TMI facilitates the measurement of adiposity in the clinical setting and, with further validation, may also be a useful longitudinal measure of future cardiometabolic risk prediction in pediatric T2DM patients.

Methods
This is a cross-sectional secondary data analysis from the Improving Renal Complications in Adolescents with Type 2 Diabetes Through REsearch (iCARE) cohort Study, a national study that is assessing renal outcomes in children with T2DM in Canada. The data in this analysis are limited to the original site in Manitoba. The published study protocol reports further details regarding study procedures 40 .
The study has been approved by the Health Research Ethics Board, University of Manitoba and follows the relevant national and international regulations of human research studies. All participants and/or their guardians provided written informed consent and assent.
Study participants. Patients with T2DM aged 10.20-17.90 years were included in this secondary data analysis (n = 116). The participants were recruited from the diabetes and nephrology clinics in Winnipeg, Manitoba, Canada.
The diagnosis of diabetes was based on the Diabetes Canada diagnostic criteria and supported by clinical criteria and the absence of diabetes associated autoantibodies 41 .
The study excluded patients with medication-or surgery-related diabetes, genetic forms of diabetes, and the presence of type 1 diabetes-related autoantibodies. In addition, those with a diagnosis of cancer and alcohol or drug abuse were excluded, as well as cases where either the patient or their caregiver were unable or unwilling to provide voluntary informed assent/consent, respectively. Data collection. Demographic data collected included age at study visit, age at diagnosis, sex, and duration of diabetes. Anthropometric data collected included height, weight, waist circumference, and hip circumference. Height was measured using 'Health o meter Professional" model # 500LK. Waist-to hip ratio and waist-to-height ratio were calculated from the primary data. Dual-energy x-ray absorptiometry (DXA) scans (Hologic, Bedford, MA) were performed to quantify percent body fat, total fat mass, trunk fat mass, and fat-free mass.
Blood pressure was assessed using 24-h ambulatory blood pressure monitors (SpaceLabs, Washington, USA). The mean arterial pressure (MAP) dipping was the parameter chosen from the ambulatory blood pressure profile for our analysis. The loss of the physiologic drop in blood pressure during sleep, or non-dipping, is associated with increased cardiovascular risk 42 , increased urinary albumin excretion which is a surrogate marker of microvascular disease 43 , and increased arterial wall stiffness 44 . Additionally, individuals with a higher BMI are more likely to have a smaller decrease in overnight blood pressure readings 45 .
Glycated hemoglobin A1c (HbA1c) levels were analyzed on a Roche Cobas Integra 800 CTS at local Diagnostic Services Manitoba (DSM) laboratory (assay referenced to the Diabetes Control and Complications Trial standard), and poor glycemic control was defined as an HbA1c level > 9.00% 40 .
Additionally, total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides (TG), apolipoprotein A (ApoA) and apolipoprotein B (ApoB) were measured to evaluate for dyslipidemia. Alanine aminotransferase (ALT), aspartate transaminase (AST) and gamma-glutamyl transferase (GGT) were measured to determine liver health and the potential presence of fatty liver disease. Samples were collected in the fasting state if possible; 20% of samples were collected in a non-fasting state.
Statistical analysis. Continuous variables are reported as means (standard deviation), and categorical variables are reported as numbers (percentage). Data were tested for normality of distribution using the Shapiro-Wilk test, and the data were log-transformed if not normally distributed, which included BMI percentile, fat mass percentage, waist-to-hip ratio, MAP dipping, triglycerides, and HDL. Spearman's correlation test was used www.nature.com/scientificreports/ to determine the relationship between the different variables including TMI, BMI z-score, fat mass percentage, waist-to-hip ratio, waist-to-height ratio, MAP dipping, and lipids. Multivariable linear regression analysis was performed to examine the association between TMI with MAP dipping, WHtR, and HDL, with age and sex added to the model. TMI was set as the dependent variable, with age, sex, MAP dipping, WHtR, and HDL as independent variables. We excluded LDL and total cholesterol due to the collinearity with HDL detected using the variance inflation factor analysis, and this analysis was also applied to assess the collinearity between WHR and WHtR. Unstandardized coefficient (B) and their p-values were reported. Statistical significance was set at alpha of 0.05.