AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 154:42–51 (2014)
Significant Sex Difference in the Association Between C-Reactive Protein Concentration and Anthropometry Among 13- to 19-Year Olds, but not 6- to 12-Year Olds in Nepal Shoko Konishi,1,2* Rajendra Prasad Parajuli,1 Erica Takane,1 Makhan Maharjan,3 Ken’ichi Tachibana,4 Hong-Wei Jiang,5 Krishna Pahari,6 Yosuke Inoue,1 Masahiro Umezaki,1 and Chiho Watanabe1 1
Department of Human Ecology, Graduate School of Medicine, The University of Tokyo, Tokyo 113-0033, Japan Department of Anthropology, University of Washington, Seattle WA, 98195 3 Environment and Public Health Organization (ENPHO), Kathmandu Nepal 4 College of Social Sciences, Ritsumeikan University, Kyoto 603–8577, Japan 5 Research Institute for Humanity and Nature, Kyoto 603–8047, Japan 6 Nepal Development Research Institute (NDRI), Kathmandu Nepal 2
KEY WORDS
BMI; dried blood spot; height; inflammation
ABSTRACT Life history theory predicts a trade-off between immunostimulation and growth. Using a crosssectional study design, this study aims to test the hypothesis that C-reactive protein (CRP) is negatively associated with height-for-age z-scores (HAZ scores) and BMI-for-age z-scores (BAZ scores) among 6- to 19-year olds (N 5 426) residing in five Nepalese communities. Dried blood spot (DBS) samples were collected and assayed for CRP using an in-house enzyme immunoassay (EIA). Sex- and age-group-specific CRP quartiles were used to examine its association with growth in linear mixed-effects (LME) models. A significant difference was found in the proportion of elevated CRP (>2 mg/L, equivalent to 3.2 mg/L serum CRP) between 13- and 19-year-old boys (12%) and girls (4%). Concentrations of CRP were positively associated with HAZ score among adolescent (13–19 years) boys, which may indicate that
individuals with greater energy resources have better growth and a better response to infections, thus eliminating the expected trade-off between body maintenance (immunostimulation) and growth. Adolescent boys with low BAZ and HAZ scores had low CRP values, suggesting that those who do not have enough energy for growth cannot increase their CRP level even when infected with pathogens. Among adolescent girls a positive association was observed between CRP and BAZ scores suggesting the possible effects of chronic lowgrade inflammation due to body fat rather than infection. The association between CRP and growth was less evident among children (6–12 years) compared with adolescents, indicating that the elevated energy requirement needed for the adolescent growth spurt and puberty may play some role. Am J Phys Anthropol 154:42–51, 2014. VC 2014 Wiley Periodicals, Inc.
Life history theory predicts a trade-off between immunostimulation and growth (McDade, 2005). While an acute phase response is necessary to protect the body from pathogens, this immunological response requires much energy, which may limit the amount of energy available for growth. Therefore, a negative association is expected between growth and immune function, assuming that the increased amount of energy allocated to immune activation would decrease the amount of energy allocated for growth (McDade, 2005; McDade et al., 2008). Such an immunological trade-off has been observed in a prospective follow-up study that examined the association between inflammation status at baseline and the amount of growth in a subsequent 3-month follow-up period (McDade et al., 2008). Among 2- to 10-year-old children in Lowland Bolivia, elevated C-reactive protein (CRP), as defined by dried blood spot (DBS) CRP >2 mg/ L (3 mg/L serum CRP), at baseline was associated with smaller gains in height over the subsequent 3 months (McDade et al., 2008). The association was stronger for 2- to 4-year olds and for those with a low skinfold thickness at baseline suggesting that younger age and low energy stores at the time of immunostimulation further
increases the negative impact on growth. Other longitudinal studies have taken repeated measurements of immunostimulation over the follow-up period and examined its association with growth. A 2-year follow-up study targeting infants aged 3 to 24 months in rural Bangladesh in which plasma levels of acute phase protein, a-1-acid glycoprotein (AGP), were measured every 12 weeks, found that a higher concentration of AGP was associated with poorer height-for-age z-scores (HAZ
Ó 2014 WILEY PERIODICALS, INC.
Grant sponsor: Global Environment Research Fund, Ministry of Environment, Japan; Grant number: H-063; Grant sponsor: KAKENHI; Grant number: 21406021. *Correspondence to: Shoko Konishi, Department of Human Ecology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-0033, Japan. E-mail: [email protected] Received 19 June 2013; accepted 30 December 2013 DOI: 10.1002/ajpa.22470 Published online 16 January 2014 in Wiley Online Library (wileyonlinelibrary.com).
CRP AND ANTHROPOMETRY IN NEPALI CHILDREN scores) and weight-for-age z-scores (WAZ scores), but not weight-for-height z-scores (WHZ scores), indicating chronic malnutrition (Goto et al., 2009). While these studies have indicated that a negative association exists between immunostimulation and anthropometric characteristics, other studies have found no association: among 3–18–month-old children in Nepal there was no significant association between immunostimulation as measured by plasma AGP and growth variables, including HAZ, WAZ, WHZ scores, or the amount of increase in each variable (Panter-Brick et al., 2009). Among Ghanaian children aged 12 SD, respectively (WHO, 1997). The sum of three skinfold thicknesses (SUMS), i.e. biceps, triceps, and subscapular, were calculated and used in the analysis. Age, the anthropometric variables, and Hgb and CRP concentrations were compared between boys and girls in each age category using the Wilcoxon test. Bivariate correlations between age, CRP, the BAZ, HAZ, and SUMS scores, and Hgb were calculated for each sex and age group. In previous studies different CRP cut-off values have been used to define elevated CRP values reflecting the different infectious burdens of each environment as well as the different analytical methods used to quantify CRP concentrations. Since we do not have any basic data to define a specific CRP cut-off for the present population, we decided to use CRP quartiles for each sex and age group (6–12 or 13–19 years) for the analyses. Then age and anthropometric measures were compared between the CRP quartile categories for each subgroup by using American Journal of Physical Anthropology
46
S. KONISHI ET AL.
TABLE 2. Age, anthropometric parameters, and hemoglobin and CRP concentrations shown as mean (SD) or median (first quartilethird quartile), stratified by sex and age group (6–12 or 13–19 years) 6–12 yr
Age (y) HAZ Stunting (%) BAZ Underweight (%) Percent BF WC (cm) MUAC (cm) SUMS (mm) Hgb (g/dL) CRP (mg/L)
13–19 yr
Boys (N 5 109)a
Girls (N 5 105)b
Boys (N 5 100)c
Girls (N 5 112)d
9.4 (2.0) 21.18 (1.00) 21 21.33 (1.07) 26 7.8 (5.5)** 56.3 (6.1) 16.6 (2.1) 16.5 (9.9)** 12.1 (1.1) 0.15 (0.07–0.75)
9.6 (2.0) 21.24 (1.01) 27 21.19 (1.04) 24 13.6 (5.2) 56.8 (7.1) 17.0 (2.2) 21.5 (12.2) 12.3 (1.2) 0.18 (0.08–0.52)
15.7 (1.9) 21.20 (1.03) 20 21.53 (1.14)** 33 11.1 (5.0)** 69.2 (6.9) 21.2 (2.6)* 18.9 (9.5)** 13.8 (1.3)** 0.21 (0.06–0.58)
15.6 (2.4) 21.24 (0.91) 21 20.64 (1.02) 9 26.4 (7.2) 70.6 (7.7) 22.0 (2.5) 38.8 (15.9) 12.6 (1.3) 0.19 (0.05–0.37)
Stunting and underweight were defined as HAZ or BAZ less than 22 SD, respectively. HAZ, height-for-age z-scores; BAZ, BMI-for-age z-scores; BF, body fat; WC, waist circumference; MUAC, mid-upper arm circumference; SUMS, sum of skinfold thicknesses at biceps, triceps, and subscapular; Hgb, hemoglobin; CRP, C-reactive protein. a N 5 103 for Hgb. b N 5 101 for Hgb. c N 5 97 for Hgb. d N 5 103 for Hgb. *P < 0.001, **P < 0.05; boys versus girls for each age category, by Wilcoxon test.
Fig. 2. Histogram showing the frequencies of the CRP values for (a) boys (N 5 209) and (b) girls (N 5 217). Individuals with CRP concentrations >3.0 mg/L are shown as 3.0 mg/L.
ANOVA. For the purpose of describing the sex difference in the distribution of CRP concentration, the proportion of elevated CRP (arbitrarily defined as >2 mg/L DBS CRP, equivalent to 3.2 mg/L serum CRP) was calculated for each sex. The concentration of CRP is shown as a concentration in DBS unless stated otherwise. The association between CRP and the growth parameters (BAZ and HAZ) was examined using linear mixed-effects (LME) models (Pinheiro and Bates, 2000) which were adjusted for age (a fixed effect) and community (a random effect) in order to account for between-community heterogeneity in anthropometric and immunological parameters (e.g. Panter-Brick et al., 2009; Konishi et al., 2011). Separate LME models were run for each sex and age group.
RESULTS Girls aged 13- to 19-years old had a significantly higher BAZ score, percent body fat, MUAC, and SUMS, compared with the boys in the same age category, American Journal of Physical Anthropology
whereas in the younger age group, when compared with boys, girls had a higher percentage of body fat and SUMS, although there were no differences in any of the other anthropometric variables (Table 2). There were only four boys and nine girls who were categorized as overweight (BAZ score >1) among whom two boys and one girl were obese (BAZ score >2). The median (first quartile 2 third quartile) CRP value was 0.15 (0.07–0.75) mg/L for 6- to 12-year-old boys and 0.21 (0.06–0.58) mg/ L for 13- to 19-year old boys and 0.18 (0.08–0.52) and 0.19 (0.05–0.37) mg/L for girls in the respective age groups (Table 2). While there was no significant sex difference in Hgb concentration in the younger age group, it was significantly higher among boys compared with girls in the older age group (Table 2). The distribution of CRP concentration was highly skewed and 73% of boys and 74% of girls had CRP 2 mg/L (3.2 mg/L CRP in serum) was 12% for boys in both age groups but only 9% and 4% for girls in the younger and older age categories, respectively, reflecting the sex difference in the distribution of the CRP concentration. Age had a significantly negative or positive correlation with CRP among boys aged 6–12 and 13–19 years, respectively, whereas there was no significant correlation among girls of either age group (Table 3). Height-for-age z-scores were positively correlated with CRP among 13- to 19-yearold boys (r 5 0.275; P < 0.01), but not among 6- to 12year-old boys or girls in either of the age groups (Table 3). The sum of skinfold thicknesses were positively correlated with CRP among girls of both age groups (r 5 0.208, P < 0.05 for the younger and r 5 0.270, P < 0.01 for the older girls) and boys aged 6- to 12-years old (r 5 0.213, P < 0.05), but not among boys aged 13- to 19-years old (Table 3). Hgb concentration was not significantly correlated with CRP in any of the sex and age groups (Table 3). While an increasing trend in BAZ and SUMS scores was observed with each increased CRP quartile among
47
CRP AND ANTHROPOMETRY IN NEPALI CHILDREN TABLE 3. Bivariate correlations between age, logCRP (common logarithm), BAZ and HAZ scores, SUMS, and Hgb for each sex and age group (6–12 or 13–19 years) Boys (6–12 yr; N 5 109)a 1 1. 2. 3. 4. 5. 6.
Age (yr) – logCRP (mg/L) 20.270* BAZ 0.072 HAZ 0.089 SUMS (mm) 0.113 Hgb (g/dL) 0.144
2
3
– 0.908 0.075 0.213* 20.152
Girls (6–12 yr; N 5 105)b 4
– 0.138 0.640*** 0.069
5
– 0.149 0.064
1
– 20.006
2
– 20.114 0.027 20.159 0.163 0.239**
Boys (13–19 yr; N 5 100)c 1. 2. 3. 4. 5. 6.
Age (yr) logCRP (mg/L) BAZ HAZ SUMS (mm) Hgb (g/dL)
– 0.212* 20.055 20.026 0.161 0.347***
– 0.015 0.275** 0.149 20.060
– 0.212* 0.696*** 0.142
– 0.083 20.057
– 0.090 0.037 0.208** 20.007
3
– 0.238* 0.678*** 0.066
4
5
– 0.304** 0.007
– 0.010
Girls (13–19 yr; N 5 112)d – 0.058 20.039 20.253 0.192* 0.117
– 0.182 20.048 0.270** 0.078
– 0.346*** 0.717*** 0.274**
– 0.267** 20.006
– 0.191
CRP, C-reactive protein; BAZ, BMI-for-age z-scores; HAZ, height-for-age z-scores; SUMS, sum of skinfold thicknesses at biceps, triceps, and subscapular; Hgb, hemoglobin. a N 5 103 for correlation between Hgb and other measures. b N 5 101 for correlation between Hgb and other measures. c N 5 97 for correlation between Hgb and other measures. d N 5 103 for correlation between Hgb and other measures. *P < 0.01, **P < 0.05, ***P < 0.001.
boys aged 6- to 12-years old, there was no such trend among girls in the same age category (Table 4). On the other hand, among the 13- to 19-year olds, girls had an increasing trend of SUMS with higher CRP quartiles, while an unexpected trend was observed among boys: in the highest CRP quartile, HAZ was highest [20.91 (0.88)], but the BAZ score was second lowest [21.78 (1.26)] across the CRP quartile groups (Table 4). In other words, adolescent boys with elevated CRP values tended to be taller, but thinner. It should be noted that the median concentration in the highest CRP quartile was higher among boys (1.89 mg/L) than girls (0.70 mg/L), suggesting that the top CRP quartile may reflect the effect of infection more than body fat among boys, whereas the opposite seems to be the case among girls. When age and place of residence were further adjusted for using LME models (Table 5), the overall associations between the CRP quartiles and BAZ and HAZ scores were similar to those seen in Table 4. There was no significant association between the CRP quartiles and growth (HAZ nor BAZ score) in boys and girls aged 6- to 12-year-old, whereas in the older age group the highest CRP quartile was associated with a high HAZ score among boys and a high BAZ score among girls. Among adolescent boys, compared with the lowest CRP quartile, the second and the third, but not the highest CRP quartiles were associated with higher BAZ scores (Table 5). These results again indicate that adolescent boys with elevated CRP (in the top CRP quartile) tend to be taller, but thinner, compared with those with lower CRP values.
DISCUSSION Sex difference in CRP-anthropometry relationships The present study is the first community-based study to report cross-sectional associations between growth
and CRP concentrations among 6- to 19-year olds in Nepal. We observed a significant sex difference in the associations: while CRP was positively associated with the BAZ score among 13- to 19-year-old girls, it was positively associated with the HAZ, but not BAZ score, among 13- to 19-year-old boys. In addition, this sex difference was only evident among 13- to 19-year olds, but not 6- to 12-year olds, suggesting that puberty and the adolescent growth spurt may be related to the observed sex difference in the association between anthropometry and immunostimulation. Our original hypothesis on the negative association between growth and CRP was rejected, probably because low HAZ and BAZ scores in this population indicate that fewer resources (energy and nutrients) were available for both immunostimulation and growth, which may have eliminated the trade-off between the two. It is possible that only those who have enough resources for growth will have elevated acute phase protein once infected with pathogens. In other words, it can be hypothesized that those with insufficient energy (as reflected by a low HAZ score) will not be able to increase CRP concentrations even when infected with pathogens. In fact, in the current study we observed both higher growth and higher CRP among adolescent boys. This might indicate that in our cross-sectional study the difference in CRP concentrations is more likely to reflect different responses once infected with pathogens, rather than because of elevated CRP concentrations limiting resources for growth. Although it was not possible to differentiate elevated CRP due to infections or body fatness, the present results may indicate that chronic CRP elevation due to body fatness is evident among adolescent girls, but not in boys or children of either sex, as further discussed below. The sex difference we observed in the association between anthropometry and CRP may be attributable to American Journal of Physical Anthropology
48
S. KONISHI ET AL. TABLE 4. CRP, age, BAZ, and HAZ scores, SUMS, and Hgb by CRP quartiles shown as median (first quartile-third quartile) or mean (SD) for each sex and age group (6–12 or 13–19 years) CRP quartiles
Boys, 6–12 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)a Boys, 13–19 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)b Girls, 6–12 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)c Girls, 13–19 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)d
Q1
Q2
Q3
Q4
P
N 5 27 0.04 (0.03–0.06) 10.3 (1.6) 21.57 (1.01) 21.37 (1.00) 13.5 (3.0) 12.1 (0.9) N 5 25 0.04 (0.03–0.06) 15.1 (2.0) 22.11 (0.90) 21.72 (0.99) 14.9 (3.5) 13.7 (1.6) N 5 26 0.06 (0.03–0.08) 10.0 (2.1) 21.29 (1.01) 20.98 (1.18) 19.9 (8.1) 12.2 (1.2) N 5 28 0.04 (0.03–0.05) 15.3 (1.9) 20.83 (0.93) 21.11 (0.93) 34.1 (12.6) 12.8 (1.3)
N 5 27 0.09 (0.07–0.15) 9.3 (2.0) 21.43 (0.70) 21.15 (0.96) 15.5 (4.1) 12.3 (1.2) N 5 25 0.15 (0.06–0.21) 15.6 (1.9) 21.21 (0.84) 21.17 (1.08) 17.4 (5.0) 13.9 (1.4) N 5 26 0.13 (0.08–0.17) 9.7 (1.8) 21.12 (0.93) 21.45 (0.92) 19.0 (6.0) 12.5 (1.1) N 5 28 0.12 (0.05–0.19) 15.8 (2.1) 20.93 (0.92) 21.49 (0.97) 35.5 (13.8) 12.4 (1.8)
N 5 27 0.30 (0.15–0.45) 9.2 (2.2) 21.21 (0.78) 21.16 (1.02) 15.8 (4.6) 11.7 (1.3) N 5 25 0.29 (0.21–0.57) 16.1 (1.7) 21.03 (1.21) 21.02 (1.04) 23.8 (14.0) 14.1 (1.1) N 5 26 0.29 (0.18–0.49) 9.6 (2.0) 21.25 (1.00) 21.29 (0.95) 22.9 (12.7) 12.2 (1.4) N 5 28 0.24 (0.19–0.37) 15.6 (2.1) 20.56 (0.91) 21.05 (0.81) 40.0 (13.7) 12.3 (0.8)
N 5 28 1.75 (0.51–9.56) 9.0 (1.9) 21.13 (1.55) 21.07 (1.04) 21.0 (17.8) 12.1 (0.9) N 5 25 1.89 (0.60–8.33) 16.2 (2.0) 21.78 (1.26) 20.91 (0.88) 19.2 (9.8) 13.5 (1.2) N 5 27 0.89 (0.52–28.7) 9.1 (1.9) 21.11 (1.24) 21.24 (0.99) 24.2 (18.1) 12.3 (1.2) N 5 28 0.70 (0.38–8.92) 15.8 (2.1) 20.23 (1.21) 21.30 (0.90) 45.8 (20.4) 12.9 (1.3)
Significant Sex Difference in the Association Between C-Reactive Protein Concentration and Anthropometry Among 13- to 19-Year Olds, but not 6- to 12-Year Olds in Nepal Shoko Konishi,1,2* Rajendra Prasad Parajuli,1 Erica Takane,1 Makhan Maharjan,3 Ken’ichi Tachibana,4 Hong-Wei Jiang,5 Krishna Pahari,6 Yosuke Inoue,1 Masahiro Umezaki,1 and Chiho Watanabe1 1
Department of Human Ecology, Graduate School of Medicine, The University of Tokyo, Tokyo 113-0033, Japan Department of Anthropology, University of Washington, Seattle WA, 98195 3 Environment and Public Health Organization (ENPHO), Kathmandu Nepal 4 College of Social Sciences, Ritsumeikan University, Kyoto 603–8577, Japan 5 Research Institute for Humanity and Nature, Kyoto 603–8047, Japan 6 Nepal Development Research Institute (NDRI), Kathmandu Nepal 2
KEY WORDS
BMI; dried blood spot; height; inflammation
ABSTRACT Life history theory predicts a trade-off between immunostimulation and growth. Using a crosssectional study design, this study aims to test the hypothesis that C-reactive protein (CRP) is negatively associated with height-for-age z-scores (HAZ scores) and BMI-for-age z-scores (BAZ scores) among 6- to 19-year olds (N 5 426) residing in five Nepalese communities. Dried blood spot (DBS) samples were collected and assayed for CRP using an in-house enzyme immunoassay (EIA). Sex- and age-group-specific CRP quartiles were used to examine its association with growth in linear mixed-effects (LME) models. A significant difference was found in the proportion of elevated CRP (>2 mg/L, equivalent to 3.2 mg/L serum CRP) between 13- and 19-year-old boys (12%) and girls (4%). Concentrations of CRP were positively associated with HAZ score among adolescent (13–19 years) boys, which may indicate that
individuals with greater energy resources have better growth and a better response to infections, thus eliminating the expected trade-off between body maintenance (immunostimulation) and growth. Adolescent boys with low BAZ and HAZ scores had low CRP values, suggesting that those who do not have enough energy for growth cannot increase their CRP level even when infected with pathogens. Among adolescent girls a positive association was observed between CRP and BAZ scores suggesting the possible effects of chronic lowgrade inflammation due to body fat rather than infection. The association between CRP and growth was less evident among children (6–12 years) compared with adolescents, indicating that the elevated energy requirement needed for the adolescent growth spurt and puberty may play some role. Am J Phys Anthropol 154:42–51, 2014. VC 2014 Wiley Periodicals, Inc.
Life history theory predicts a trade-off between immunostimulation and growth (McDade, 2005). While an acute phase response is necessary to protect the body from pathogens, this immunological response requires much energy, which may limit the amount of energy available for growth. Therefore, a negative association is expected between growth and immune function, assuming that the increased amount of energy allocated to immune activation would decrease the amount of energy allocated for growth (McDade, 2005; McDade et al., 2008). Such an immunological trade-off has been observed in a prospective follow-up study that examined the association between inflammation status at baseline and the amount of growth in a subsequent 3-month follow-up period (McDade et al., 2008). Among 2- to 10-year-old children in Lowland Bolivia, elevated C-reactive protein (CRP), as defined by dried blood spot (DBS) CRP >2 mg/ L (3 mg/L serum CRP), at baseline was associated with smaller gains in height over the subsequent 3 months (McDade et al., 2008). The association was stronger for 2- to 4-year olds and for those with a low skinfold thickness at baseline suggesting that younger age and low energy stores at the time of immunostimulation further
increases the negative impact on growth. Other longitudinal studies have taken repeated measurements of immunostimulation over the follow-up period and examined its association with growth. A 2-year follow-up study targeting infants aged 3 to 24 months in rural Bangladesh in which plasma levels of acute phase protein, a-1-acid glycoprotein (AGP), were measured every 12 weeks, found that a higher concentration of AGP was associated with poorer height-for-age z-scores (HAZ
Ó 2014 WILEY PERIODICALS, INC.
Grant sponsor: Global Environment Research Fund, Ministry of Environment, Japan; Grant number: H-063; Grant sponsor: KAKENHI; Grant number: 21406021. *Correspondence to: Shoko Konishi, Department of Human Ecology, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-0033, Japan. E-mail: [email protected] Received 19 June 2013; accepted 30 December 2013 DOI: 10.1002/ajpa.22470 Published online 16 January 2014 in Wiley Online Library (wileyonlinelibrary.com).
CRP AND ANTHROPOMETRY IN NEPALI CHILDREN scores) and weight-for-age z-scores (WAZ scores), but not weight-for-height z-scores (WHZ scores), indicating chronic malnutrition (Goto et al., 2009). While these studies have indicated that a negative association exists between immunostimulation and anthropometric characteristics, other studies have found no association: among 3–18–month-old children in Nepal there was no significant association between immunostimulation as measured by plasma AGP and growth variables, including HAZ, WAZ, WHZ scores, or the amount of increase in each variable (Panter-Brick et al., 2009). Among Ghanaian children aged 12 SD, respectively (WHO, 1997). The sum of three skinfold thicknesses (SUMS), i.e. biceps, triceps, and subscapular, were calculated and used in the analysis. Age, the anthropometric variables, and Hgb and CRP concentrations were compared between boys and girls in each age category using the Wilcoxon test. Bivariate correlations between age, CRP, the BAZ, HAZ, and SUMS scores, and Hgb were calculated for each sex and age group. In previous studies different CRP cut-off values have been used to define elevated CRP values reflecting the different infectious burdens of each environment as well as the different analytical methods used to quantify CRP concentrations. Since we do not have any basic data to define a specific CRP cut-off for the present population, we decided to use CRP quartiles for each sex and age group (6–12 or 13–19 years) for the analyses. Then age and anthropometric measures were compared between the CRP quartile categories for each subgroup by using American Journal of Physical Anthropology
46
S. KONISHI ET AL.
TABLE 2. Age, anthropometric parameters, and hemoglobin and CRP concentrations shown as mean (SD) or median (first quartilethird quartile), stratified by sex and age group (6–12 or 13–19 years) 6–12 yr
Age (y) HAZ Stunting (%) BAZ Underweight (%) Percent BF WC (cm) MUAC (cm) SUMS (mm) Hgb (g/dL) CRP (mg/L)
13–19 yr
Boys (N 5 109)a
Girls (N 5 105)b
Boys (N 5 100)c
Girls (N 5 112)d
9.4 (2.0) 21.18 (1.00) 21 21.33 (1.07) 26 7.8 (5.5)** 56.3 (6.1) 16.6 (2.1) 16.5 (9.9)** 12.1 (1.1) 0.15 (0.07–0.75)
9.6 (2.0) 21.24 (1.01) 27 21.19 (1.04) 24 13.6 (5.2) 56.8 (7.1) 17.0 (2.2) 21.5 (12.2) 12.3 (1.2) 0.18 (0.08–0.52)
15.7 (1.9) 21.20 (1.03) 20 21.53 (1.14)** 33 11.1 (5.0)** 69.2 (6.9) 21.2 (2.6)* 18.9 (9.5)** 13.8 (1.3)** 0.21 (0.06–0.58)
15.6 (2.4) 21.24 (0.91) 21 20.64 (1.02) 9 26.4 (7.2) 70.6 (7.7) 22.0 (2.5) 38.8 (15.9) 12.6 (1.3) 0.19 (0.05–0.37)
Stunting and underweight were defined as HAZ or BAZ less than 22 SD, respectively. HAZ, height-for-age z-scores; BAZ, BMI-for-age z-scores; BF, body fat; WC, waist circumference; MUAC, mid-upper arm circumference; SUMS, sum of skinfold thicknesses at biceps, triceps, and subscapular; Hgb, hemoglobin; CRP, C-reactive protein. a N 5 103 for Hgb. b N 5 101 for Hgb. c N 5 97 for Hgb. d N 5 103 for Hgb. *P < 0.001, **P < 0.05; boys versus girls for each age category, by Wilcoxon test.
Fig. 2. Histogram showing the frequencies of the CRP values for (a) boys (N 5 209) and (b) girls (N 5 217). Individuals with CRP concentrations >3.0 mg/L are shown as 3.0 mg/L.
ANOVA. For the purpose of describing the sex difference in the distribution of CRP concentration, the proportion of elevated CRP (arbitrarily defined as >2 mg/L DBS CRP, equivalent to 3.2 mg/L serum CRP) was calculated for each sex. The concentration of CRP is shown as a concentration in DBS unless stated otherwise. The association between CRP and the growth parameters (BAZ and HAZ) was examined using linear mixed-effects (LME) models (Pinheiro and Bates, 2000) which were adjusted for age (a fixed effect) and community (a random effect) in order to account for between-community heterogeneity in anthropometric and immunological parameters (e.g. Panter-Brick et al., 2009; Konishi et al., 2011). Separate LME models were run for each sex and age group.
RESULTS Girls aged 13- to 19-years old had a significantly higher BAZ score, percent body fat, MUAC, and SUMS, compared with the boys in the same age category, American Journal of Physical Anthropology
whereas in the younger age group, when compared with boys, girls had a higher percentage of body fat and SUMS, although there were no differences in any of the other anthropometric variables (Table 2). There were only four boys and nine girls who were categorized as overweight (BAZ score >1) among whom two boys and one girl were obese (BAZ score >2). The median (first quartile 2 third quartile) CRP value was 0.15 (0.07–0.75) mg/L for 6- to 12-year-old boys and 0.21 (0.06–0.58) mg/ L for 13- to 19-year old boys and 0.18 (0.08–0.52) and 0.19 (0.05–0.37) mg/L for girls in the respective age groups (Table 2). While there was no significant sex difference in Hgb concentration in the younger age group, it was significantly higher among boys compared with girls in the older age group (Table 2). The distribution of CRP concentration was highly skewed and 73% of boys and 74% of girls had CRP 2 mg/L (3.2 mg/L CRP in serum) was 12% for boys in both age groups but only 9% and 4% for girls in the younger and older age categories, respectively, reflecting the sex difference in the distribution of the CRP concentration. Age had a significantly negative or positive correlation with CRP among boys aged 6–12 and 13–19 years, respectively, whereas there was no significant correlation among girls of either age group (Table 3). Height-for-age z-scores were positively correlated with CRP among 13- to 19-yearold boys (r 5 0.275; P < 0.01), but not among 6- to 12year-old boys or girls in either of the age groups (Table 3). The sum of skinfold thicknesses were positively correlated with CRP among girls of both age groups (r 5 0.208, P < 0.05 for the younger and r 5 0.270, P < 0.01 for the older girls) and boys aged 6- to 12-years old (r 5 0.213, P < 0.05), but not among boys aged 13- to 19-years old (Table 3). Hgb concentration was not significantly correlated with CRP in any of the sex and age groups (Table 3). While an increasing trend in BAZ and SUMS scores was observed with each increased CRP quartile among
47
CRP AND ANTHROPOMETRY IN NEPALI CHILDREN TABLE 3. Bivariate correlations between age, logCRP (common logarithm), BAZ and HAZ scores, SUMS, and Hgb for each sex and age group (6–12 or 13–19 years) Boys (6–12 yr; N 5 109)a 1 1. 2. 3. 4. 5. 6.
Age (yr) – logCRP (mg/L) 20.270* BAZ 0.072 HAZ 0.089 SUMS (mm) 0.113 Hgb (g/dL) 0.144
2
3
– 0.908 0.075 0.213* 20.152
Girls (6–12 yr; N 5 105)b 4
– 0.138 0.640*** 0.069
5
– 0.149 0.064
1
– 20.006
2
– 20.114 0.027 20.159 0.163 0.239**
Boys (13–19 yr; N 5 100)c 1. 2. 3. 4. 5. 6.
Age (yr) logCRP (mg/L) BAZ HAZ SUMS (mm) Hgb (g/dL)
– 0.212* 20.055 20.026 0.161 0.347***
– 0.015 0.275** 0.149 20.060
– 0.212* 0.696*** 0.142
– 0.083 20.057
– 0.090 0.037 0.208** 20.007
3
– 0.238* 0.678*** 0.066
4
5
– 0.304** 0.007
– 0.010
Girls (13–19 yr; N 5 112)d – 0.058 20.039 20.253 0.192* 0.117
– 0.182 20.048 0.270** 0.078
– 0.346*** 0.717*** 0.274**
– 0.267** 20.006
– 0.191
CRP, C-reactive protein; BAZ, BMI-for-age z-scores; HAZ, height-for-age z-scores; SUMS, sum of skinfold thicknesses at biceps, triceps, and subscapular; Hgb, hemoglobin. a N 5 103 for correlation between Hgb and other measures. b N 5 101 for correlation between Hgb and other measures. c N 5 97 for correlation between Hgb and other measures. d N 5 103 for correlation between Hgb and other measures. *P < 0.01, **P < 0.05, ***P < 0.001.
boys aged 6- to 12-years old, there was no such trend among girls in the same age category (Table 4). On the other hand, among the 13- to 19-year olds, girls had an increasing trend of SUMS with higher CRP quartiles, while an unexpected trend was observed among boys: in the highest CRP quartile, HAZ was highest [20.91 (0.88)], but the BAZ score was second lowest [21.78 (1.26)] across the CRP quartile groups (Table 4). In other words, adolescent boys with elevated CRP values tended to be taller, but thinner. It should be noted that the median concentration in the highest CRP quartile was higher among boys (1.89 mg/L) than girls (0.70 mg/L), suggesting that the top CRP quartile may reflect the effect of infection more than body fat among boys, whereas the opposite seems to be the case among girls. When age and place of residence were further adjusted for using LME models (Table 5), the overall associations between the CRP quartiles and BAZ and HAZ scores were similar to those seen in Table 4. There was no significant association between the CRP quartiles and growth (HAZ nor BAZ score) in boys and girls aged 6- to 12-year-old, whereas in the older age group the highest CRP quartile was associated with a high HAZ score among boys and a high BAZ score among girls. Among adolescent boys, compared with the lowest CRP quartile, the second and the third, but not the highest CRP quartiles were associated with higher BAZ scores (Table 5). These results again indicate that adolescent boys with elevated CRP (in the top CRP quartile) tend to be taller, but thinner, compared with those with lower CRP values.
DISCUSSION Sex difference in CRP-anthropometry relationships The present study is the first community-based study to report cross-sectional associations between growth
and CRP concentrations among 6- to 19-year olds in Nepal. We observed a significant sex difference in the associations: while CRP was positively associated with the BAZ score among 13- to 19-year-old girls, it was positively associated with the HAZ, but not BAZ score, among 13- to 19-year-old boys. In addition, this sex difference was only evident among 13- to 19-year olds, but not 6- to 12-year olds, suggesting that puberty and the adolescent growth spurt may be related to the observed sex difference in the association between anthropometry and immunostimulation. Our original hypothesis on the negative association between growth and CRP was rejected, probably because low HAZ and BAZ scores in this population indicate that fewer resources (energy and nutrients) were available for both immunostimulation and growth, which may have eliminated the trade-off between the two. It is possible that only those who have enough resources for growth will have elevated acute phase protein once infected with pathogens. In other words, it can be hypothesized that those with insufficient energy (as reflected by a low HAZ score) will not be able to increase CRP concentrations even when infected with pathogens. In fact, in the current study we observed both higher growth and higher CRP among adolescent boys. This might indicate that in our cross-sectional study the difference in CRP concentrations is more likely to reflect different responses once infected with pathogens, rather than because of elevated CRP concentrations limiting resources for growth. Although it was not possible to differentiate elevated CRP due to infections or body fatness, the present results may indicate that chronic CRP elevation due to body fatness is evident among adolescent girls, but not in boys or children of either sex, as further discussed below. The sex difference we observed in the association between anthropometry and CRP may be attributable to American Journal of Physical Anthropology
48
S. KONISHI ET AL. TABLE 4. CRP, age, BAZ, and HAZ scores, SUMS, and Hgb by CRP quartiles shown as median (first quartile-third quartile) or mean (SD) for each sex and age group (6–12 or 13–19 years) CRP quartiles
Boys, 6–12 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)a Boys, 13–19 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)b Girls, 6–12 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)c Girls, 13–19 yr CRP (mg/L) Age (yr) BAZ HAZ SUMS (mm) Hgb (g/dL)d
Q1
Q2
Q3
Q4
P
N 5 27 0.04 (0.03–0.06) 10.3 (1.6) 21.57 (1.01) 21.37 (1.00) 13.5 (3.0) 12.1 (0.9) N 5 25 0.04 (0.03–0.06) 15.1 (2.0) 22.11 (0.90) 21.72 (0.99) 14.9 (3.5) 13.7 (1.6) N 5 26 0.06 (0.03–0.08) 10.0 (2.1) 21.29 (1.01) 20.98 (1.18) 19.9 (8.1) 12.2 (1.2) N 5 28 0.04 (0.03–0.05) 15.3 (1.9) 20.83 (0.93) 21.11 (0.93) 34.1 (12.6) 12.8 (1.3)
N 5 27 0.09 (0.07–0.15) 9.3 (2.0) 21.43 (0.70) 21.15 (0.96) 15.5 (4.1) 12.3 (1.2) N 5 25 0.15 (0.06–0.21) 15.6 (1.9) 21.21 (0.84) 21.17 (1.08) 17.4 (5.0) 13.9 (1.4) N 5 26 0.13 (0.08–0.17) 9.7 (1.8) 21.12 (0.93) 21.45 (0.92) 19.0 (6.0) 12.5 (1.1) N 5 28 0.12 (0.05–0.19) 15.8 (2.1) 20.93 (0.92) 21.49 (0.97) 35.5 (13.8) 12.4 (1.8)
N 5 27 0.30 (0.15–0.45) 9.2 (2.2) 21.21 (0.78) 21.16 (1.02) 15.8 (4.6) 11.7 (1.3) N 5 25 0.29 (0.21–0.57) 16.1 (1.7) 21.03 (1.21) 21.02 (1.04) 23.8 (14.0) 14.1 (1.1) N 5 26 0.29 (0.18–0.49) 9.6 (2.0) 21.25 (1.00) 21.29 (0.95) 22.9 (12.7) 12.2 (1.4) N 5 28 0.24 (0.19–0.37) 15.6 (2.1) 20.56 (0.91) 21.05 (0.81) 40.0 (13.7) 12.3 (0.8)
N 5 28 1.75 (0.51–9.56) 9.0 (1.9) 21.13 (1.55) 21.07 (1.04) 21.0 (17.8) 12.1 (0.9) N 5 25 1.89 (0.60–8.33) 16.2 (2.0) 21.78 (1.26) 20.91 (0.88) 19.2 (9.8) 13.5 (1.2) N 5 27 0.89 (0.52–28.7) 9.1 (1.9) 21.11 (1.24) 21.24 (0.99) 24.2 (18.1) 12.3 (1.2) N 5 28 0.70 (0.38–8.92) 15.8 (2.1) 20.23 (1.21) 21.30 (0.90) 45.8 (20.4) 12.9 (1.3)
