Association between perinatal growth restriction and subsequent body build in very-low-birth-weight infants

Abstract.

The present study assessed the influence of perinatal growth restriction on subsequent body build until nine years old in very-low-birth-weight infants (VLBWIs). Growth restriction was defined as being born small for gestational age (SGA) or extrauterine growth restriction (EUGR) if born at term. Two hundred and eighty-three VLBWIs were divided into four groups according to their history of SGA or EUGR. Variations in z-scores of body height (zHT) and obesity index (OI), as well as the prevalence of short stature, underweight, and overweight were evaluated from six to nine years old. As a result, zHT increased significantly, but OI did not change. In contrast, the prevalence of underweight and overweight significantly increased, but short stature did not change. A comparison of the four groups demonstrated that EUGR may be associated with short stature and underweight regardless of SGA status. SGA, but not EUGR, may be a risk factor for being overweight. The prevalence of abnormal body builds without SGA or EUGR seemed to be similar with standard prevalence of Japanese children. In conclusion, perinatal growth restriction, especially EUGR had a significant influence on subsequent body build until nine years old.

Highlights

● Variations in anthropometric values and prevalence of abnormal body build were evaluated in 283 VLBWIs of 6–9 yr old.

● From six to nine years old, the Z-score of body height significantly increased, but the prevalence of short stature did not change. In contrast, the obesity index did not change, but the prevalence of underweight and overweight significantly increased.

● Perinatal growth restriction, especially EUGR had a significant influence on subsequent body build until nine years old.

Introduction

The characteristics of growth in very-low-birth-weight infants (VLBWIs) during neonatal intensive care unit (NICU) admission are influenced by two critical phases: intrauterine growth until birth and extrauterine growth from birth to term. Recently, a new Japanese neonatal anthropometric chart was established, and the z-score of weight at any gestational age until term was obtained (1). Using this chart (2), allows for a quantitative index of intrauterine growth (z-score at birth) and growth at term (z-score at term), to be individually obtained. We have previously evaluated the z-score of body height (zHT) and the percentile of body mass index (BMI) of VLBWIs prior to elementary school entrance (nearly 6 yr old), and the relationship between small for gestational age (SGA) or extrauterine growth restriction (EUGR) and the subsequent incidence of short stature, thinness, and obesity was evaluated (3). However, the Japan Society for the Study of Obesity and Pediatric Endocrinology has recommended that the evaluation of thinness and obesity using the percentile of BMI is not appropriate, and that the obesity index should be used. Obesity index < –20% and > 20% were proposed as definitions for underweight (tendency for thinness) and overweight (tendency for obesity) (4).

In the present study, a retrospective evaluation was conducted to determine the variations in anthropometric values in VLBWIs of six to nine years old. In addition, the prevalence of abnormal body build was compared among the four groups, according to the presence or absence of perinatal growth restriction at nine years old.

Subjects and Methods

Participants

Between April 1999 and March 2015, 780 VLBWIs were discharged from our NICU, and 322 VLBWIs underwent a comprehensive medical check-up before elementary school entrance (almost 6 yr old) and the 3rd year of elementary school (almost 9 yr old), including physical and laboratory examinations between April 2008 and November 2024. Among the 322 VLBWIs, 20 subjects received human GH formulation due to GH deficiency or small for gestational age, 5 subjects whose Gross Motor Function Classification System – Expanded and Revised (5) was level > 2, and 14 subjects whose perinatal data could not be completely obtained were excluded. Ultimately, 283 patients were enrolled in the present study, none of whom required any respiratory or feeding support, including home oxygen therapy or tube feeding and also observed any apparent congenital anomaly.

Body weight (BW), height (HT), head circumference (HC), and waist circumference (WC) were measured during each period. WC was measured at the umbilicus level by nursing staff while the subject was standing with the measuring tape horizontally around their body at the end of normal expiration. However, we could not measure WC data in 77 subjects at 6 yr old or in 11 subjects at 9 yr old, because cooperation could not be obtained. In addition, the reproducibility of the WC value at 6 yr old was not sufficient. Therefore, we evaluated WC and the incidence of central obesity at 9 yr old (n = 272).

Methods

Using anthropometric data, we determined the z-score of body height (zHT) and obesity index (OI) at each period. The OI was calculated using the following formula: OI (%) = (actual BW – reference BW) / reference BW ×100. The Japanese reference BW for individuals of 5–17 yr old can be obtained automatically from a computing equation that contains the actual HT, age, and sex (6). BMI was calculated by dividing weight (kg) by the square of height (m²), and the z-score of BMI (zBMI) at the corresponding age was obtained using files approved by the Japan Society for the Study of Obesity and Pediatric Endocrinology (7).

Growth restriction at birth was defined as a z-score for body weight or body length of < –2.0, which was the same as the definition for SGA. Although the definition of growth restriction after birth is not yet established, whether the evaluation is cross-sectional or longitudinal (8,9,10), EUGR in the present study was defined as a z-score for body weight or length of < –2.0 on the expected date of confinement or at a corresponding age if the subjects were discharged before the expected date of confinement, as we have previously reported (3). The classifications of the 283 VLBWIs infants in the study population were as follows: SGA and EUGR (group 1, n = 63), non-SGA but EUGR (group 2, n = 77), SGA but non-EUGR (group 3, n = 13), and non-SGA and non-EUGR (group 4, n = 130).

To clarify the variations in anthropometric values and body build from six to nine years old, the values of zHT, zBMI and OI as well as the prevalence of short stature (zHT< –2.0), underweight (OI < –20%), and overweight (OI > 20%) were compared in the four groups. In addition, the prevalence of central obesity (waist-to-height ratio [WtHR] > 0.5) at nine years old was compared in the four groups. This study was approved by the Institutional Ethics Committee of the hospital.

Statistical analyses

Statistical analyses were performed using IBM SPSS Statistics Desktop for Japan (ver. 26). The Kruskal-Wallis test was used to compare the values of zHT, zBMI, OI, and WtHR in the four groups at the same age. A paired Wilcoxon signed-rank test and paired t-test were used to compare the values of zHT, zBMI, and OI between six and nine years old in the overall cohort and each group. Multiple comparisons of the prevalence of short stature, underweight, overweight, and central obesity among the four groups at identical ages were conducted using the z-test with Bonferroni correction. A paired McNemar’s test was used to compare the prevalence of short stature, underweight, and overweight between six and nine years old in the overall cohort and each group. We also compared the prevalence of overweight and central obesity in 273 VLBWIs among four groups using the z-test with Bonferroni correction, and the paired McNemar’s test was used to compare the prevalence of overweight and central obesity in the overall cohort and in each group at nine years old.

In order to know the perinatal factors associated with such morbidites as a short stature, being underweight, overweight and having central obesity at nine years of age, we conducted the following analysis. In addition to having a history of SGA and EUGR, sex difference and gestational age were adopted as independent variables since they may influence an individual’s subsequent body build. First, the subjects were divided into two groups based on whether they were male, SGA infants, or EUGR infants, and the proportions of a short stature, being underweight, being overweight, and having central obesity in each group were compared using Fisher’s exact test. Gestational age was also compared between the two groups, whether or not each morbidity was complicated, using an unpaired t-test. Next, a logistic regression analysis with the force injection method was used to extract the significant factors associated with each morbidity among the significant independent variables obtained by a univariate analysis. Statistical significance was set at p < 0.05.

Results

The gestational age and anthropometric data in 283 subjects were as follows: gestational age, 28.3 ± 2.9 wk; body weight at birth (z score), 989 ± 310 g (–1.02 ± 1.33); body weight at term (z score), 2473 ± 414 g (–1.71 ± 1.42); body length at birth (z score), 34.9 ± 3.9 cm (–0.75 ± 1.26); body length at term (z score), 45.6 ± 2.7 cm (–1.92 ± 1.41), respectively. At 6 yr old, the median (IQR) values for HT, zHT, BW, OI, and BMI were 109.5 (106.2, 112.8) cm, –0.70 (–1.30, 0.00), 17.3 (15.7, 19.4) kg, –3.0 (–9.0, 3.3), and 14.6 (13.8, 15.7), respectively. At 9 yr old, the median (IQR) values for HT, zHT, BW, OI, and BMI were 128.0 (124.8, 132.5) cm, –0.60 (–1.20, 0.20), 25.1 (22.5, 29.7) kg, –3.7 (–11.4, 7.5), and 15.3 (14.2, 17.4), respectively. The zHT at nine years was significantly higher than at six years old (p < 0.001), but no significant difference was observed for the zBMI (p = 0.407) or OI (p = 0.568). At six years old, the prevalence of short stature, underweight, and overweight were 22/283 (7.8%), 5/283 (1.8%), and 8/283 (2.8%), respectively. At nine years old, the prevalence of short stature, underweight, and overweight were 20/283 (7.1%), 14/283 (4.9%), and 31/283 (11.0%), respectively. A significant difference was observed for underweight (p = 0.035) and overweight (p < 0.001), but not for short stature (p = 0.774) (Table 1).

Table 1. Anthropometric data 283 VLBWIs during the perinatal period, before entrance to elementary school, and in 3^rd grade of elementary school.

Perinatal growth and subsequent anthropometric data at six and nine years old in the four groups are shown in Table 2. The zHT of group 1 was significantly lower than that of the other three groups, and the zHT of group 2 was significantly lower than that of group 4 at six years old; similar differences between the four groups continued at nine years old (Fig. 1a). The zBMI of groups 1 and 2 were significantly lower than those of groups 3 and 4 at six years old, whereas significant differences between groups 1, 3, and 4 were only observed at nine years old (Fig. 1b). The OI values of groups 1 and 2 were significantly lower than those of groups 3 and 4 at six years old, while significant differences between groups 1 and groups 3 and 4 were only observed at nine years old (Fig. 1c), the relationship between the four groups was identical to that of zBMI at both ages. The waist-to-height ratio (WtHR) was evaluated only at nine years old. The WtHR in group 3 seemed to be higher than that in the other three groups, but the difference was not statistically significant (Fig. 1d). Comparison with the corresponding group between six and nine years old demonstrated that the zHT at nine years was significantly higher than at six years old in groups 1 (p = 0.026), 2 (p = 0.003), and 4 (p < 0.001) (Fig. 1a) and zBMI at 9 yr old was significantly higher than at 6 yr old in group 2 (p = 0.030) (Fig. 1b). In contrast, no significant difference was observed in the OI of the corresponding group between the two ages (Fig. 1c).

Table 2. Characteristics of anthropometric values before entrance to elementary school, and in 3^rd grade of elementary school in groups according to history of SGA or EUGR.

Fig. 1.

The box-plot for z score of body height (zHT), z score of body mass index (zBMI), obesity index (OI) and waist to height ratio (WtHR) in each group. white box indicates at six years and dark box indicates at nine years old. a: Z score of body height (zHT): At six years old, the zHT of group 1 was significantly lower than that of groups 2 (p = 0.022*), 3 (p = 0.002**), and 4 (p < 0.001***), and the zHT of group 2 was significantly lower than that of group 4 (p = 0.013*). A similar significant difference was observed among the four groups for the zHT at nine years old: group 1 vs. 2 (p = 0.036*), 1 vs. 3 (p = 0.001**), 1 vs. 4 (p < 0.001***), and 2 vs. 4 (p = 0.016*). The zHT at nine years old was significantly higher than that at six years old in groups 1 (p = 0.026^#), 2 (p = 0.002^##), and 4 (p < 0.001^###). b: The Z score of the body mass index (zBMI): At 6 yr old, the zBMI of group 1 was significantly lower than that of groups 3 (p < 0.001***) and 4 (p < 0.001***), and the zBMI of group 2 was significantly lower than that of groups 3 (p = 0.002**) and 4 (p < 0.001***). At 9 yr old, significant differences were seen between group 1 and groups 3 (p = 0.002**) and 4 (p < 0.001***). The zBMI at 9 yr old was significantly higher than that at 6 yr old in group 2 (p = 0.030^#), c: Obesity index (OI): At six years old, the OI of group 1 was significantly lower than that of groups 3 (p = 0.002**) and 4 (p < 0.001***), and the OI of group 2 was significantly lower than that of groups 3 (p = 0.002**) and 4 (p = 0.049*). At nine years old, the only significant differences were between group 1 and groups 3 (p = 0.040*) and 4 (p = 0.019*). No significant difference was observed in the corresponding OI between the two age groups. d: Waist-to-height ratio (WtHR): Data were only obtained at nine years old. The WtHR in group 3 appeared to be higher than that in the other three groups, but no statistically significant differences were observed.

The prevalence of abnormal body build in each group is shown in Table 3. By comparing each morbidity rate in the four groups, the rates of children with a short stature and those that were underweight seemed to be higher in groups 1 and 2 than in groups 3 and 4, while the rate of overweight in group 3 was higher than in the other three groups at both ages. Significant differences in children with a short stature were observed between groups 2 and 4 at six years old and between groups 1 and 4 at nine years old. A significant difference was also observed for children that were overweight between group 3 and groups 1 and 2 at six years old, and between group 3 and the other three groups at nine years old. Comparing the rate of each morbidity between both ages, the prevalence of children being overweight in groups 2 and 4 and of those being underweight in group 2 at nine years old was significantly higher than that observed at six years old.

Table 3. Cross-tabulation for the prevalence of abnormal body builds in each group according to the history of SGA or EUGR before entrance to elementary school and in the 3^rd grade of elementary school (n = 283).

The prevalence of children with a short stature and those that were underweight in groups 1 and 2 and overweight in group 3 seemed to be high at both ages in comparison to standard Japanese children at the corresponding age. Unlike the other three groups, the prevalence of a short stature, being underweight, and being overweight in group 4 was similar to that in standard Japanese children until nine years old (Table 3).

The prevalence of central obesity in group 3 was also higher than that in the three other groups; however, a significant difference was only observed between groups 3 and 1. The prevalence of central obesity in VLBWIs was 40/273 (14.7%), which was significantly higher (p = 0.027) than in overweight individuals 29/273 (10.6%). Comparing the prevalence of overweight and central obesity in the four groups, a significant difference was only observed in group 2 (12/73 (16.4%) vs. 6/73 (8.2%), p = 0.031) (Table 4).

Table 4. Cross-tabulation for the prevalence of general and central obesity in each group according to the history of SGA or EUGR in the 3^rd grade of elementary school (n = 272).

Significant influences were observed for the gestational age and EUGR on subsequent short stature and underweight at nine years old; however, a significant difference was not observed for sex difference and SGA on any morbidities at nine years old in the univariate analysis (Table 5). A logistic regression analysis revealed that EUGR was the only significant factor for short stature (odds ratio [OR], 5.58; 95% confidence interval [CI], 1.56–19.94; p = 0.008) and underweight (OR, 5.54; 95% CI, 1.16–25.52, p= 0.032), however, gestational age was not selected as a significant factor for a short stature (odds ratio [OR], 0.89; 95% confidence interval [CI], 0.16–1.05; p= 0.175) or being underweight (OR, 0.87; 95% CI, 0.71–1.05, p = 0.150) at 9 yr old (Table 6).

Table 5. A comparison for the prevalence of short stature, underweight, overweight and central obesity between two groups whether or not the presence of male sex, SGA, EUGR and gestational age in 283 VLBWIs at 9 yr old.

Table 6. A logistic regression analysis with force injection method for perinatal variables associating with abnormal body build in VLBWIs at 9 yr old.

Discussion

The fetus frequently encounters various types of stress, including, but not limited to, inflammation and hypoxia. Consequently, some fetuses are born as VLBWIs. VLBWIs often suffer from intrauterine growth restriction, and their body function is promptly forced to adapt to extrauterine life with insufficient nutritional supply, which usually results in growth restriction after birth. Such an insult during the perinatal period may cause epigenetic changes and later onset of metabolic disorder (11, 12). SGA or EUGR may provide the model of Bakar theory (13) for infants who have received recent neonatal intensive care, and it is interesting to know the characteristics of body build according to the type of perinatal growth restriction.

We have previously reported the anthropometric values for 322 VLBWIs at six years old, and the prevalence of short stature, thinness, and obesity were 11.8%, 11.8%, and 1.9%, respectively (3). However, thinness and obesity were defined by the < 3^rd percentile and > 97^th percentile of BMI, respectively, which was not appropriate for the evaluation of Japanese children. In addition, patients who received GH replacement therapy were included. In the present study, the definition of short stature was identical to that in a previous study, but subjects receiving GH replacement therapy were excluded, and the tendency for thinness and obesity was redefined using the obesity index as underweight (OI < –20%) and overweight (OI >20%). Although 196 (69%) subjects were enrolled in a previous study, the prevalence of short stature, underweight, and overweight at six years old was 7.8%, 1.8%, and 2.8%, respectively. Using the present definition, we can compare the prevalence of underweight and overweight in the present study subjects with age corresponding to the Japanese public data (14). Conversely, the coefficient correlation between zBMI and OI was extremely high at r = 0.92 at 6 yr old and r = 0.90 at 9 yr old (data not shown), suggesting that zBMI may be a viable substitute for OI.

Although a significant difference was not observed in the prevalence of short stature between two ages, it is interesting that the zHT at nine years old was significantly larger than at six years old. Knops et al. reported catch-up growth up to 10 yr old in 510 children born very preterm or with very low birth weight, and the body height of some SGA preterm babies, especially male SGA infants, still showed short stature until 10 yr old but was expected to exhibit catch-up growth from 5 to 10 yr old; however, most very preterm AGA infants were able to outgrow their short stature by 5 yr old, and apparent catch-up growth was usually not observed thereafter (15). In contrast, regarding catch-up growth of body height, even in AGA (not SGA) children, zHT increased significantly from the age of 6 to 9 yr. This finding was observed not only in group 1 but also in groups 2 and 4, who were not SGA infants. The reason for this conflicting result is unclear, as we were unable to evaluate hormonal, radiological, and biochemical markers in the study subjects; however, familial background, the impact of secondary sexual characteristics, and lower follow-up rates may have been involved.

In contrast, the zBMI and OI at nine years old was not significantly different from at six years old; however, the prevalence of underweight and overweight at nine years old was significantly higher than at six years old, which may reflect the bipolarization of body build during early elementary school. A similar tendency was observed in standard Japanese children of six to nine years old (14), but the degree of increase in the prevalence of underweight and overweight seemed to be higher in VLBWIs.

The characteristics of the body build in VLBWIs are influenced by perinatal growth restriction, especially EUGR. Regardless of whether they are classified as SGA, VLBWIs with a history of EUGR continued to tend to be small and underweight until the 3^rd grade of elementary school. A multivariate analysis also revealed that EUGR was the only significant factor for short stature and underweight; however, gestational age, sex, and SGA were not. In contrast, no significant perinatal factors were involved in the subsequent onset of overweight and central obesity at nine years old. This result may suggest the clinical significance of growth restriction at term on the subsequent occurrence of a short stature and thinness at nine years of old, however, the 95% confidence interval for the odds ratio is wide, thus indicating a high degree of uncertainty regarding this estimation and further examinations are therefore required to evaluate this problem.

VLBWIs with SGA but not those with EUGR show a tendency toward being overweight with standard height. It is also interesting that the prevalence of abnormal body builds without perinatal growth restriction (i.e., VLBWIs without SGA or EUGR) seemed to resemble the prevalence of standard Japanese children until nine years old. This result suggests that preterm birth itself is not directly linked to abnormal body build during childhood. However, a number of studies have identified preterm birth itself as a risk factor for non-communicable diseases in later life (16,17,18) and further observation is necessary.

We could not evaluate the details of body composition, such as body fat or muscle quantity (lean body mass), which is a major limitation of the present study. Instead, we measured WC and obtained the WtHR. In clinical practice, it is difficult to measure WC for VLBWIs before elementary school entrance, so we adopted the data at nine years old only. Recent reports demonstrated that central fat distribution is associated with an increased risk of cardiovascular and metabolic diseases in both adults and children, and that WC and WtHR appear to be correlated with the degree of intra-abdominal adiposity, as measured by dual-energy X-ray absorptiometry (DXA) (19), magnetic resonance imaging (MRI) (20), and computed tomography (CT) (21). Thomas et al. reviewed the characteristics of fat deposition in preterm infants from birth to adulthood and found that abdominal fat deposition in preterm infants was larger than that in term infants at term-equivalent age according to MRI findings (11). Johnson et al. also reported a difference in body composition of preterm infants at term-equivalent age, and the percentage of total body fat was greater; however, lean body mass and absolute fat volume were smaller in preterm infants than in term infants (22). Whether or not this difference was persistent during the early infantile period is unclear; fat deposition in preterm birth infants should be dominant in the trunk versus the limbs during childhood according to DXA, although the total fat mass (absolute volume or percentage of body weight) does not increase (23). In contrast, adults born prematurely had elevated total adipose tissue and increased abdominal adipose tissue (subcutaneous and internal), and this finding was most pronounced in male subjects (24). The reason for the difference in fat distribution between adults born prematurely and adults born at term until adulthood has not been clarified; however, forced adaptation in the face of an insufficient nutritional environment until the early infantile period may induce an economical constitution, with subsequent sufficient nutritional supply potentially resulting in excessive fat deposition in some preterm infants. In addition, treatment with antenatal glucocorticoids in preterm adults with the 363S variant of the glucocorticoid receptor gene may tend to cause abdominal fat accumulation (25). In the present study, a significant correlation was observed between OI and WtHR (r = 0.83, p < 0.001, data not shown), and the prevalence of central obesity was significantly higher than that of overweight obesity. Although the validity of WtHR for evaluating intra-abdominal adiposity in children remains controversial (26, 27) and an age-specific relationship between obesity index and WtHR has not been established in Japan, this result may suggest that the characteristics of obesity in VLBWIs might be truncal rather than general, and it may also be a risk factor for non-communicable diseases in later life.

Another limitation of this study was the discordance in the gestational age of each group. However, it is difficult to enroll a sufficient number of VLBW infants of identical gestational ages in each category in clinical practice. A multivariate analysis was conducted to correct this problem, and gestational age was not identified as a significant variable in the present study. However, enrollment of gestational age-matched subjects may be preferable when performing such evaluations. In addition, we could not evaluate the dietary habits of the study subjects and the body build of parents and siblings. In contrast, the strengths of this study were that it was a cohort study conducted at a single facility and that all data were treated in a structured and uniform manner.

In conclusion, the characteristics of the body build for VLBWIs may be influenced by perinatal growth restriction, and EUGR is mainly associated with short stature and underweight. However, the prevalence of overweight increase from six to nine years old. In addition, the potential risk of non-communicable diseases may be concealed in the absence of an abnormal body build during childhood (28); thus, further follow-up with laboratory tests is necessary to evaluate this problem.

Conflict of interests

The authors declare no conflicts of interest in association with the present study.