Association between bone mineral density and content and physical growth parameters among children and adolescents diagnosed with HIV: a cross-sectional study

ABSTRACT BACKGROUND: During childhood and adolescence, there are significant increases in bone mineral content (BMC) and bone mineral density (BMD). OBJECTIVE: To investigate physical growth parameters associated with BMD and BMC among children and adolescents diagnosed with human immunodeficiency virus (HIV). DESIGN AND SETTING: Cross-sectional study conducted in Florianópolis, Brazil, among 63 children and adolescents (aged 8-15 years) diagnosed with HIV. METHOD: BMD, BMC and fat percentage z score were evaluated using dual X-ray absorptiometry. Age/height z score and body mass index (BMI)/age z score were obtained in accordance with international recommendations, and bone age was obtained through hand-wrist radiography. Sex, family income, information on HIV infection (T CD4+ lymphocyte count, viral load and type of antiretroviral therapy, moderate-vigorous physical activity and sedentary behavior) were used as adjustment variables in the analyses. Simple and multiple linear regression analyses were performed, with a significance level of P ≤ 0.05. RESULTS: Subtotal BMD (without the head region) was directly associated with bone age, BMI/age z score and fat percentage z score, even after adjusting for covariates. Subtotal BMC/height was directly associated with bone age, height/age z score, BMI/age z score and fat percentage z score, even after adjusting for covariates. CONCLUSION: Subtotal BMD and subtotal BMC/height were directly associated with physical growth indicators among children and adolescents diagnosed with HIV.


INTRODUCTION
During childhood and adolescence, significant increases in bone mineral content (BMC) and bone mineral density (BMD) occur in association with the growth spurt and a high mineralization rate. 1 These factors are considered to be determinants for protection against osteopenia, and later against osteoporosis, during adult life. 2 Among children and adolescents diagnosed with human immunodeficiency virus (HIV), lower bone mass concentrations and mineralization rates may occur due to impairments of adequate physical growth and consequent pubertal delay associated with HIV infection and prolonged use of antiretroviral therapy (ART). 3,4 In the growth process, regardless of sex, BMD and BMC concentrations do not develop linearly with chronological age. Their growth rate has been shown to be higher after the growth spurt. 5 In a study conducted among children and adolescents diagnosed with HIV (aged 8-16 years), it was found that body mass, body mass index (BMI) and body fat were directly associated with BMD. 6 The mechanisms through which HIV can lead to changes in BMD are not fully understood, but are probably multifactorial, including factors associated with HIV (ART and disease stage) 6 and modifiable risk factors (changes in physical growth, low physical activity and highly sedentary behavior). 6 Use of ART can decrease the BMC and BMD of patients diagnosed with HIV. 7 In a previous study, a direct association between moderate-vigorous physical activity (MVPA) and high gravitational or muscular load and bone mass was identified among children and adolescents without any diagnosis of HIV. 8 This highlights the importance of factors relating to disease/treatment and lifestyle, which can impact physical growth parameters.
Studies carried out among children and adolescents diagnosed with HIV showed that height/ age z score was a bone mass predictor. 6,9 Cross-sectional studies have only investigated associations between BMD and physical growth parameters, with no data regarding BMC. 6,9,10 Other studies did not consider lifestyle variables, such as MVPA and sedentary behavior, in correlating BMD and BMC with physical growth parameters. [11][12][13] OBJECTIVE Thus, the aim of this study was to investigate physical growth parameters associated with BMD and BMC among children and adolescents diagnosed with HIV.

Research characteristics
This was a cross-sectional study that formed part of the "Saúde were added for losses and refusals. Thus, for multiple linear regression analysis, the sample of 63 children and adolescents made it possible to find associations with an average effect size of 0.50.

Eligibility criteria
The inclusion criteria were that the subjects needed to: 1) present HIV infection through vertical transmission in the medical record; 2) be aged 8-15 years; 3) have clinical and laboratory information available in the medical record; and 4) be able to stand and communicate.
The exclusion criteria were the following situations: 1) motor impairment or contraindication of vigorous exercise; 2) speech, hearing and/or cognitive impairment; 3) diseases that change body composition, except for those related to HIV infection; and 4) regular use of diuretic drugs or immunotherapies not related to ART.

Dependent variables
BMD and BMC analyses were carried out by means of dual emission X-ray absorptiometry (DXA). DXA measurements were per- Internal quality control was obtained through a daily automatic calibration process that preceded evaluations and which was done in accordance with the manufacturer's instructions. One researcher who had previously been trained to make these measurements, was responsible for all evaluations and followed all the procedures standardized by the equipment manufacturer. During the evaluations, the participants were instructed to wear appropriate clothing (top or bikini and swimming trunks) and be barefoot, without use of earrings and/or finger rings or any metallic adornments.
The evaluations performed through DXA produced measurements of subtotal BMD (without the head region) and subtotal BMC (without the head region), with corrections for height and body fat percentage. These subtotal measurements (BMD and BMC) are considered to be parameters with good accuracy and reproducibility, compared with total BMD and BMC, which can dilute changes in bone mass. 14

Independent variables
The physical growth variables evaluated were the age/height z score, BMI/age z score and fat percentage z score. Z scores allow specific values to be compared with the population, taking into account typical values and dispersion, and considering age and sex. 15 To calculate the z score standardized according to age and height, height was obtained using a stadiometer (Altura Exata, Belo Horizonte, Brazil), with measurement capacity from 115 cm to 210 cm and resolution of 0.1 cm, in accordance with the recommendations proposed by the International Society for the Advancement of Kinanthropometry (ISAK). 16 Chronological age was collected through an interview.
To calculate the z score standardized according to age and BMI, body mass was obtained using a portable digital scale (Tanita; model BF-683W) with 0.1 kg accuracy and capacity for up to 150 kg. BMI was calculated from body mass and height measurements. For body mass measurements, the recommendations of the International Society for the Advancement of Kinanthropometry (ISAK) 16 were followed.
To calculate the z score standardized according to fat percentage, the fat percentage was obtained through DXA. Test-retest reproducibility for body fat was examined in an independent sample of the present study (n = 10) that was similar in age and sex distribution: 11.6 ± 5.8 and 11.5 ± 5.9 kg; intraclass correlation coefficient (ICC) = 1.00; 95% confidence interval (CI) = 0.99-1.00. 17 Bone age was assessed by means of left-limb hand-wrist radiography, following standardized procedures, 18  The forearm, palm and fingers were in contact with the film plate and the x-ray beam was centered in the distal region of metacarpal III. Bone age was determined by comparing the radiography obtained with a series of standard radiographs that represent the skeletal maturation process in healthy subjects. 18 The radiographs were read by a radiology specialist and all procedures were performed at the Department of Radiology of HIJG.

Covariates
The variables of sex (male or female) and family income were collected using a questionnaire developed for this study. Family income was determined as the number of minimum monthly wages (1 minimum wage at the time of the study was R$ 724.00) and was categorized in the following ranges: ≤ 2 minimum wages; > 2 to 5 minimum wages; or > 5 minimum wages.
Information on HIV infection was obtained through analysis of the medical records, from which information was obtained regarding CD4+ lymphocyte count, viral load and type of ART.
The type of ART was categorized as follows: use of ART with protease inhibitors (PI); use of ART without PI; or no use of ART.
Data on moderate to vigorous physical activity (MVPA) and sedentary behavior were collected by means of triaxial Actigraph accelerometers (3.8 x 3.7 x 1.8 cm), model GT3XE-Plus (Manufacturing Technology Inc., Fort Walton Beach, United States). These allowed measurement of acceleration produced through body movement. 19 This equipment was used continuously over a period of 7 to 14 days, including weekends. 19,20 The participants were instructed to attach the equipment to the right hemibody at the waistline early in the morning and use it until the end of the day, and only to remove it for activities in water and for sleeping. 20 The motion sensor was calibrated and the stored information was downloaded to the Actilife 6.0 software (ActiGraph, Pensacola, Florida, United States) in a process that took 15 seconds. The subject's usual physical activity was represented by at least four valid days, comprising three weekdays and one weekend day containing at least 10 hours of information (600 minutes), after removing nonuse periods of at least 60 consecutive zeros. 21 Contacts through phone calls and messages, at least once a week, were used to ensure regular use of the device. Verbal and written instructions were provided to subjects and their guardians before the device was used. 20 For subjects who did not present valid days through previous measurements, up to two further attempts to use the device were made.
The numbers of minutes at different physical activity intensities were proportionally adjusted to a 14-hour period, considering that this is the average time agreed for this population. This was done through the following formula: adjusted minutes = (registered minutes/time of use) * (14 * 60). 22 The number of minutes of MVPA was obtained considering the cutoff point proposed by Evenson et al. 23 Uninterrupted blocks (bouts) of at least five and ten minutes of MVPA were derived. 22 The test-retest reproducibility of this protocol was examined using a subsample of the pres- During the test, the dynamometer or the hand did not touch any other objects. Right handgrip (D) and left handgrip (E) were alternately assessed, with two attempts per assessment. The best left and right scores obtained in each test were added together to obtain the general score (D + E, in kilograms). The standardization of the Canadian Society for Exercise Physiology 25 was adopted.

Statistical analysis
Descriptive analysis was firstly performed (mean and standard deviation; and median and interquartile range) and observations with missing data were excluded. Kurtosis and asymmetry were assessed in order to ascertain whether the data had normal distribution (range from -2 to + 2), and histograms were also analyzed to identify data distribution normality. Pearson or Spearman linear correlation and simple and multiple linear regression were used to verify correlations and associations between dependent and independent variables, respectively. For multiple linear regression analysis, three models were built. Hierarchical adjusted analysis was performed, 26

Sixty-three children and adolescents who had been diagnosed
with HIV participated in the study (males = 28; females = 35).
In the full sample, the mean age was 12.14 years (± 1.95), the mean height was 147.3 cm (± 13.08), the mean body mass was 39.9 kg (± 11.04) and the mean bone age was 12.02 years (± 2.68) (data not shown in tables/figures). The sample characteristics are shown in Table 1. Table 2 shows the correlation matrix among all the variables.
Subtotal BMC/height showed a significant positive correlation with subtotal BMD, height/age z score, BMI/age z score, fat percentage z score and bone age. Subtotal BMD showed a significant positive correlation with BMI/age z score, fat percentage z score and bone age and showed a negative correlation with CD4+ lymphocytes ( Table 2).
Simple linear regression analysis between subtotal BMD and physical growth among these children and adolescents diagnosed with HIV demonstrated that subtotal BMD was directly associated with bone age, which explained 64% of BMD variability (P < 0.001). After adding the covariates of sex, family income, viral load, CD4+ lymphocytes, ART, MVPA, sedentary behavior and muscle strength, subtotal BMD was found to be directly associated with bone age and the model explained 79% of the subtotal BMD variability (P < 0.001) ( Table 3).
Subtotal BMD was directly associated with BMI/age z score, which explained 14% of subtotal BMD variability (P < 0.001) in the crude model. After adding the covariates of sex, family income, viral load, CD4, use of ART, MVPA, sedentary behavior and muscle strength (model III), subtotal BMD remained directly associated with BMI/age z score and the model explained 47% of subtotal BMD variability (P < 0.001). Subtotal BMD was directly associated with fat percentage z score, which explained 4% of its variability (p = 0.005) in the crude model. The variables of sex, income, viral load, CD4+ lymphocytes, use of ART, MVPA, sedentary behavior and muscle strength (model III) were directly associated with fat percentage z score, which explained 36% of subtotal BMD variability (P < 0.001) ( Table 3).
Simple linear regression analysis demonstrated that subtotal BMC/height was directly associated with bone age, which explained 69% of BMC variability (P < 0.001). The variables of sex, income, viral load, CD4+ lymphocytes, ART, MVPA, sedentary behavior and muscle strength (model III) were directly associated with bone age, which explained 76% of BMC variability (P < 0.001). Subtotal BMC/height was directly associated with height/ age z score, which explained 3% of BMD variability (P = 0.005) in  Table 4).

DISCUSSION
The main findings of this study were: 1) there were direct associations between subtotal BMD and bone age, BMI/age z score and fat percentage z score, even after adjusting for covariates; and 2) subtotal BMC/height was directly associated with bone age, height/age z score, BMI/age z score and fat percentage z score even after adjusting for covariates.
In the present study, a direct association between subtotal BMD and bone age was observed. A previous study carried out among 1,218 children aged 6-18 years of both sexes, without any HIV diagnosis, also found a direct correlation between BMD and bone age. 28 One possible explanation for the association between BMD and bone age may be the correlation between the beginning, peak and end of the puberty growth spurt and specific bone development states. 10 The period from the beginning to the end of the puberty growth spurt lasts approximately two years, and the peak growth rate occurs around one year after the start of the growth spurt. 29 Linear growth in adolescence is greatest during pubertal development, with no net gain in bone mass after peak bone mass is reached. 30 Thus, peak bone mass is considered to form the bone reservoir for future life and is the factor that determines BMD in adulthood and, consequently, the risk of osteoporotic fractures. 30 It has thus been observed that linear growth is closely linked to skeletal development. However, because HIV infection and ART can affect bone development in children and adolescents diagnosed    37 In addition, BMC assists in bone modeling regulation. 38 In a study that compared children and adolescents diagnosed with HIV with children and adolescents without an HIV diagnosis, it was reported that exposure to HIV and use of ART since birth were associated with lower bone mass. 7 In children and adolescents diagnosed with HIV, lower bone concentrations and mineralization rates may occur due to impaired adequate physical development and subsequently delayed puberty, which can be related to HIV infection. 3,4 The present study had limitations that need to be considered when interpreting its data: the cross-sectional design did not allow inferences regarding cause-effect relationships. The sample used, which had a wide age range that included pre and post-pubertal children and adolescents, can also be considered to have been a limitation: because of the sample size, it was not possible to perform stratifications according to maturation stage.
Some strengths of this study include its use of highly reliable methods, such as DXA to evaluate BMD and BMC, accelerometers to measure MVPA and sedentary behavior and hand-wrist radiography analyzed by a radiologist to obtain bone age.

CONCLUSIONS
Through the findings of the present study, it could be concluded that subtotal BMD and subtotal BMC/height were directly associated with physical growth indicators (BMD: bone age, BMI/age z score and fat percentage z score; and BMC: bone age, height/age z score, BMI/age z score and fat percentage z score, respectively), even after adjusting for sex, family income, viral load, CD4+ lymphocytes, use of ART, MVPA, behavior sedentary and muscle strength.
Assessing and improving bone health in children and adolescents diagnosed with HIV is crucial, in order to minimize the risk of bone complications over the long term. It needs to be emphasized that with prolonged use of ART, there may be significant bone mass reductions. Thus, monitoring of physical growth indicators that are simple to assess, such as age/height z score, BMI/age z score and fat percentage z score can be useful for identifying possible bone mass deficits in children and adolescents diagnosed with HIV.