Tri-Ponderal Mass Index vs. Fat Mass/Height3 as a Screening Tool for Metabolic Syndrome Prediction in Colombian Children and Young People

Tri-ponderal mass index (TMI) and fat mass index (FMI) have been proposed as alternative approaches for assessing body fat since BMI does not ensure an accurate screening for obesity and overweight status in children and adolescents. This study proposes thresholds of the TMI and FMI for the prediction of metabolic syndrome (MetS) in children and young people. For this purpose, a cross-sectional study was conducted on 4673 participants (57.1% females), who were 9–25 years of age. As part of the study, measurements of the subjects’ weight, waist circumference, serum lipid indices, blood pressure and fasting plasma glucose were taken. Body composition was measured by bioelectrical impedance analysis (BIA). The TMI and FMI were calculated as weight (kg)/height (m3) and fat mass (kg)/height (m3), respectively. Following the International Diabetes Federation (IDF) definition, MetS is defined as including three or more metabolic abnormalities. Cohort-specific thresholds were established to identify Colombian children and young people at high risk of MetS. The thresholds were applied to the following groups: (i) a cohort of children where the girls’ TMI ≥ 12.13 kg/m3 and the boys’ TMI ≥ 12.10 kg/m3; (ii) a cohort of adolescents where the girls’ TMI ≥ 12.48 kg/m3 and the boys’ TMI ≥ 11.19 kg/m3; (iii) a cohort of young adults where the women’s TMI ≥ 13.21 kg/m3 and the men’s TMI ≥ 12.19 kg/m3. The FMI reference cut-off values used for the different groups were as follows: (i) a cohort of children where the girls’ FMI ≥ 2.59 fat mass/m3 and the boys’ FMI ≥ 1.98 fat mass/m3; (ii) a cohort of adolescents where the girls’ FMI ≥ 3.12 fat mass/m3 and the boys’ FMI ≥ 1.46 fat mass/m3; (iii) a cohort of adults where the women’s FMI ≥ 3.27 kg/m3 and the men’s FMI ≥ 1.65 kg/m3. Our results showed that the FMI and TMI had a moderate discriminatory power to detect MetS in Colombian children, adolescents, and young adults.

mass/(height) 2 ) to predict MetS, with results confirming its close relationship to the components of MetS [21]. However, there is no consensus as to the cut-off value that can be used to define excess adiposity based on TMI (kg)/height 3

) and FMI (fat mass)/height 3 ).
As MetS is acknowledged to be a global public health problem and given the increasing prevalence of METs in the Colombian population, there is a clear need to establish gender-specific TMI and FMI cut-off values for estimating the fat mass and body adiposity dysfunction associated with MetS. Accordingly, the objective of this study was to explore thresholds of TMI and FMI (fat mass)/height 3 ) for the prediction of MetS in a large population of Colombian children and youth.

Study Design and Sample Population
This research is based on a two-part relatively large-scale study (FUPRECOL) [22] of over 1000 children/adolescents and young adults [23]. We used the data from these two projects, which totaled 4673 participants that were 9-25 years old (n = 1047 children 9-12 years; n = 1830 adolescents 13-17 years; and n = 1796 young adults 18-25 years). All participants and their parents/guardians provided their written, informed consent. Each study was approved by the authorized institutional review boards (UMB N • 01-1802-2013 and UR N • CEI-ABN026-000262) and complied with the Declaration of Helsinki (as revised in Hong Kong in 1989 and in Edinburgh, Scotland, in 2000). The studies also complied with Colombian laws regulating clinical research on human subjects (Resolution 008430/1993 of the Ministry of Health). Exclusion factors included a clinical diagnosis of CVD, DM-1 and 2, pregnancy, the use of alcohol or drugs and in general, the presence of any disease not directly associated with nutrition. Volunteers received no compensation for their participation.

Data Collection
Body weight (kg) was measured using an electric scale (Model Tanita ® BC-418 ® or BF-689 ® , Tokyo, Japan) and height (cm) with a portable stadiometer (Seca ® 216, Hamburg, Germany). BMI was calculated as weight (kg)/height (m 2 ) [24]. WC (cm) was measured as the narrowest point between the lower costal border and the iliac crest using a metal tape measure (Lufkin W606PM ® , Parsippany, NJ, USA) in accordance with the guidelines of the International Society for the Advancement of Kinanthropometry [25]. BMI and TMI were calculated as weight (kg)/height (m 2 ) and weight (kg)/height (m 3 ), respectively, while the BMI z-score was estimated using the World Health Organization Growth Reference from 2007 [24]. For all anthropometric variables, a low technical error measurement was reported (less than 2%).
We determined body fat percentage (BF%) and FMI using bioelectrical impedance analyses (BIA) by whole-body impedance (Tanita Model BC-418 ® , Tokyo, Japan). Detailed information about the BIA technique has been provided in previous studies [26,27]. The FMI was then calculated by dividing each subject's fat mass (kg) by the cubed value of his/her height (m), as described in Burton [18].

Metabolic Syndrome Diagnosis
Between 6:00 and 9:00 a.m. after 10-12 h of an overnight fast, blood samples were extracted from capillary sampling. Participants were asked not to engage in prolonged exercise 24 h prior to testing. Using enzymatic colorimetric methods, we measured (i) high density lipoprotein cholesterol (HDL-c); (ii) triglycerides; (iii) total cholesterol; and (iv) fasting blood glucose. Over a period of 15 days, the inter-assay reproducibility, estimated by the coefficient of variation, was determined from 16 replicate analyses of eight plasma pools. For all blood samples, we obtained errors of less than 4%.
We measured blood pressure levels on the left arm at the heart level using an automatic device Omron M6 Comfort (Omron ® Healthcare Europe B.V., Hoofddorp, The Netherlands). Individuals were seated in a semi-reclined position with arms relaxed and supported, while the midpoint of the arm was positioned at the level of the heart. In children and adolescents, MetS score was defined in accordance with the updated harmonized criteria of de Ferranti et al. [28] and Magge et al. [29] based on age/gender/height criteria. In young adults, MetS was defined in accordance with the updated harmonized criteria of the IDF [30]. According to the criteria of the IDF, participants were considered to have MetS if they showed three or more of the following: (1) abdominal obesity (WC ≥80 cm in women and ≥90 cm in men); (2) hypertriglyceridemia (≥150 g/dL); (3) low HDL-C (<50 mg/dL in women and <40 mg/dL in men); (4) high blood pressure (systolic blood pressure (SBP) ≥130 mmHg or diastolic blood pressure (DBP) ≥85 mmHg); (5) high fasting glucose (≥100 mg/dL).
We calculated a MetS score that reflected a continuous score of the five MetS risk factors. The MetS score was calculated from the data collected for each participant, based on the IDF [30], Ferranti et al. [28], and Magge et al. [29], with standard deviations using data from the total cohort at baseline. The mean of this continuously distributed MetS was thus zero by definition.

Statistical Analysis
The characteristics of the participants were given as mean values and standard deviation (SD). Independent two-tailed t-tests for continuous variables and chi-square (χ 2 ) tests for categorical variables were used to examine sex differences and age groupings. The relationships between BMI, FMI, TMI, and MetS were tested by means of partial correlation coefficients. This analysis was adjusted by age and sex. Locally weighted scatterplot smoothing curves were used to illustrate the shape of the relationship between TMI, FMI, and MetS. The results are given for the lineal form. A higher fraction of the explained variance (i.e., R 2 values) in MetS indicates greater accuracy of the anthropometric indexes.
The area under the curve (AUC), ranging from 0 (worthless test) to 1 (perfect test), was used to predict MetS with the TMI and FMI. AUC has been reported to be a global indicator of diagnostic performance since it represents the ability of the test to correctly classify participants with a high risk of MetS (p-values < 0.01 and an AUC > 0.80) [31]. In addition, the positive likelihood ratio LR (+) and the negative likelihood ratio LR (−) were also determined. Cutoff points were chosen based on the highest J-Youden index, which uses the point on the receiver operating characteristic (ROC) curve that is farthest from the line of equality [32]. The analysis of the data was performed with the SPSS statistical software package, version 24.0 (IBM, Chicago, IL, USA) for Windows. Table 1 shows the results obtained for the anthropometric variables, body composition, blood pressure, and metabolic biomarker variables. The final sample comprised a total of 4673 children and young people. It was composed of 2001 males (42.9%) and 2672 females (57.1%) with a mean age of 17 ± 4.0 (a minimum age of 9 years and a maximum age of 25 years). In the cohort of children, girls were found to have a significantly lower WC, HDL, and glucose level as well as a greater quantity of body fat (%), FMI, and triglycerides compared to boys (p < 0.05).

Study Participants
In the cohort of adolescents, the weight, height, WC, SBP, mean arterial pressure, and glucose level of the girls were significantly lower in comparison to the boys (p < 0.05). In contrast, their BMI, TMI, body fat, FMI, total cholesterol, triglycerides, LDL, and HDL were significantly higher (p < 0.001).
In the cohort of young adults, women were found to have a lower weight, height, WC, SBP, DBP, MAP, and triglycerides than men (p < 0.05). However, they had a higher TMI, body fat, FMI, total cholesterol, LDL, HDL, and glucose level in comparison to the men (p < 0.001). It is worth highlighting that, among young adults, the prevalence of MetS was greater in women than in men (p = 0.001). Table 1. Characteristics among a sample from Colombia (mean (standard deviation (SD)) or frequency (%)).

Characteristic
Children 9 Table 2 shows the partial correlation between BMI, FMI, TMI, and MetS score. Overall, we showed an acceptable to moderate positive correlation with MetS score (all p < 0.01).  Figures 1 and 2 show plots of the association between FMI, TMI, and MetS score for sex and age groups. As observed, the locally weighted scatterplot smoothing curves provide evidence of a well-defined threshold effect of high/low value indexes on MetS score, which is provided as a continuous outcome.

Optimal Cut-Off Value in the Screening of MetS
The ROC curve analyses for the diagnostic performance of TMI and FMI in identifying a high risk of MetS are shown in Tables 3 and 4, respectively. The ROC analyses showed that TMI and FMI parameters could be used to detect MetS in Colombian children, adolescents, and young adults. Table 3. Parameters of the receiver operating characteristic (ROC) curves analysis for the diagnostic performance of Tri-ponderal Mass (TMI) vs. fat mass index (FMI) in identifying high risk of metabolic syndrome (MetS) according to the Ferranti and Maggie criteria in children and adolescents.

Parameter TMI (kg/m 3 ) FMI (Fat Mass)/Height 3 )
Girls ( For the girls in the cohort of children, the TMI cut-off value of 12.13 kg/m 3 provided a sensitivity of 80%, an LR (+) value of 2.04, a specificity of 61%, and an LR (−) value of 0.33. Regarding the FMI, the cut-off value of 2.59 fat mass/m 3 provided a sensitivity of 85%, an LR (+) value of 2.05, a specificity of 59%, and an LR (−) value of 0.26. For the boys in the same cohort, the TMI cut-off value of 12.10 kg/m 3 provided a sensitivity of 85%, an LR (+) value of 2.05, a specificity of 59%, and an LR (−) value of 0.26. The ROC curve for the FMI was also obtained using a cut-off value of 1.98 fat mass/m 3 , which achieved a sensitivity of 82%, an LR (+) of 2.04, a specificity of 60%, and an LR (−) of 0.31.
For the girls in the cohort of adolescents, the TMI cut-off value of 12.48 kg/m 3 provided a sensitivity of 86%, an LR (+) value of 2.87, a specificity of 70%, and an LR (−) value of 0.20. For the FMI, the cut-off value of 3.12 fat mass/m 3 provided a sensitivity of 87%, an LR (+) value of 2.55, a specificity of 66%, and an LR (−) value of 0.19. For the adolescent boys in this cohort, the TMI cut-off value of 11.19 kg/m 3 provided a sensitivity of 93%, an LR (+) value of 3.09, a specificity of 70%, and an LR (−) value of 0.10. The ROC curve for FMI was also obtained using a cut-off value of 1.46 fat mass/m 3 , which achieved a sensitivity of 84%, an LR (+) of 2.10, a specificity of 60%, and an LR (−) of 0.27.
For the women in the young adult cohort, the TMI cut-off value of 13.21 kg/m 3 provided a sensitivity of 94%, an LR (+) value of 2.81, a specificity of 67%, and an LR (−) value of 0.09. For FMI, the cut-off value of 3.27 fat mass/m 3 provided a sensitivity of 97%, an LR (+) value of 2.52, a specificity of 62%, and an LR (−) value of 0.08. Regarding the TMI of the men in this cohort, the cut-off value of 12.19 kg/m 3 provided a sensitivity of 94%, an LR (+) value of 3.11, a specificity of 70%, and an LR (−) value of 0.09. The ROC curve for FMI was also obtained using a cut-off value of 1.65 fat mass/m 3 , which achieved a sensitivity of 93%, an LR (+) of 2.14, a specificity of 57%, and an LR (−) of 0.13.

Discussion
MetS is considered a worldwide public health problem that increases the risk of cardiovascular morbidity and mortality [10]. In Colombia, the prevalence of MetS was found to be 7.8% and 11.0%, according to the definitions proposed by the Ford et al. and de Ferranti et al., respectively in a population-based sample of schoolchildren [11]. Our study found that the MetS prevalence was higher in girls in the cohorts of children and adolescents (14.6% and 8.1%), while the prevalence was higher in males for young adults. These findings differ from the results of Ruano-Nieto et al. [33], who found that the estimated prevalence of MetS was 8.4% for women and 6.1% for men in a population of Ecuadorian university students. These differences could be explained by the MetS cluster used, the design method, and the target population. Thus, the identification of useful screening tools to detect MetS early in life is of particular interest since this will facilitate more timely and effective interventions in those subjects who are at greater risk.
This study provides gender-specific TMI and FMI reference thresholds as a way of estimating the fat mass and body adiposity dysfunction associated with MetS during childhood and early adulthood. To the best of our knowledge, this is the first evaluation of the predictive capacity of TMI and FMI for MetS. As reflected in the data obtained in our study, both TMI and FMI have a moderate discriminating power for the detection of MetS in Colombian children and youth. This finding supports the diagnostic capabilities of both indices to identify children and young people at a high risk of MetS. This is clinically relevant, since it is known that MetS increases the risk of CVD.
The FMI (fat mass/m 2 ) has been established as a valid predictor of fatness, which is based on fat mass and height in children and adolescents [34][35][36][37]. However, since previous research has indicated that the FMI does not predict adiposity, this research study focused on the prediction capacity of the FMI (fat mass/m 3 ) as proposed by Burton [18]. The results of the MetS ROC analysis demonstrated that FMI showed the greatest AUC compared with the TMI for predicting MetS risk in all groups.
These results are consistent with those of a previous study that compared the BMI and BF% of a population of 1698 Chinese adults and their accuracy in predicting MetS [21]. Their conclusion was that FMI (fat mass/(height) 2 ) was the best screening tool for the prediction of MetS [21]. Nevertheless, differences in ethnicity, genetic susceptibility, lifestyle, and age range of the two study populations signify that the data are not directly comparable [38]. Thus, there is a growing interest in proposing cut-off points for body fat levels for the early detection of CVD risk. Future longitudinal studies should be conducted that will lead to a deeper understanding of the role of different levels of adiposity in regard to CVD risk stratification in children, adolescents, and young adults.
Furthermore, it has been found that in adolescents, TMI estimates body fat levels more accurately than BMI [14]. As shown in our study, in addition to being an effective obesity screening tool, TMI can be used to detect MetS among Colombian children and young adults. The thresholds established to detect a high risk of MetS in Colombian children and young people were as follows: (i) children cohort: TMI ≥ 12.13 kg/m 3 (girls); TMI ≥ 12.10 kg/m 3 (boys); (ii) adolescent cohort: TMI ≥ 12.48 kg/m 3 (girls); TMI ≥ 11.19 kg/m 3 (boys); (iii) young adult cohort: TMI ≥ 13.21 kg/m 3 (women); TMI ≥ 12.19 kg/m 3 (men). These results indicate that TMI is a useful indicator for predicting MetS at early ages. However, the differences observed between boys and adolescent boys justify the need to consider sex-and age-specific thresholds as such thresholds are considered in BMI assessments for children and adolescents [39].
Interestingly, among the study cohorts (children, adolescents, and young adults), the greatest AUCs were observed for TMIs of 0.854 for females and 0.814 for males and for FMIs of 0.882 for females and 0.848 for males in the cohort of young adults. This highlights the fact that TMI and FMI are valuable tools that can be used to combat MetS risk in early adulthood. In this same line, previous studies have reported that age-and sex-associated variations of body fat stem from development during growth [40]. According to Rolland-Cachera [41], a developing maturational process and, thus, the absence of a well-defined fat pattern among the children and adolescents in this study could explain the differences in predictive capacity observed between cohorts.
The present study has certain limitations. Firstly, our analysis was of a cross-sectional sample and did not have a longitudinal design. Because the children and adolescents were still growing, the percentiles in height or BF% could vary markedly. Secondly, our study was limited to a well characterized sample of Colombian children and young people. Given that children and adolescents of different races or ethnic groups show important differences in body composition, which mainly occurs in body fat, our reported cut-off values may not be extrapolated to other populations. Nevertheless, this fact does not limit the interesting results obtained. Future population-based studies conducted in other ethnic or racial groups are required to provide TMI and FMI cut-off values for the detection of MetS during childhood and early adulthood. Finally, due to its cross-sectional design, it is not possible to examine whether TMI and FMI are predictors for future MetS. Therefore, further studies investigating their power longitudinally are required.
It should be highlighted that the main strength of our research is that it is the first study to investigate the predictive power of FMI and TMI in relation to MetS in a large population of 4673 subjects and to provide cut-off values for the identification of MetS by FMI and TMI in a population of children and young people. Accordingly, this research has relevant clinical implications in that it permits the early detection of increased cardiometabolic risk at young ages.

Conclusions
In summary, FMI and TMI were found to have a moderate discriminatory power to detect MetS in a Colombian population of children and young adults. For the first time, reference cut-off values were provided for the detection of MetS by FMI and TMI in children and young people. The reported thresholds could be used in clinical practice to identify Colombian children and youth at high risk of MetS. Furthermore, these results justify the need to incorporate FMI and TMI in daily clinical practice as indicators for predicting MetS and avoiding its associated complications. However, since most clinicians and researchers globally utilize BMI, there would have to be an overall proven increase in usefulness of TMI and FMI in order to convince the healthcare community to adopt these alternate measures. Future studies of different races and ethnic groups are needed to obtain reference values applicable to populations in countries throughout the world.

Abbreviations
The following abbreviations are used in this manuscript: