Systemic Inflammation in the First 2 Weeks After Birth as a Determinant of Physical Growth Outcomes in Hospitalized Infants With Extremely Low Gestational Age

Abstract

Objective:

To examine associations of systemic inflammation with growth outcomes at neonatal intensive care unit (NICU) discharge/transfer among infants with extremely low gestational ages.

Study design:

We studied 850 infants at born 23–27 weeks of gestation. We defined inflammatory protein elevation as the highest quartile of c-reactive protein (CRP), interleukin 6 (IL-6), tumor necrosis factor-alpha (TNF-∝), or interleukin 8 (IL-8) on postnatal days 1, 7, and 14. We compared z-scores of weight, length, and head circumference at NICU discharge/transfer between infants with vs without inflammatory protein elevation, adjusting in linear regression for birth size z-score, sex, gestational age, diet, co-morbidities, medications, and length of hospitalization.

Results:

Mean gestational age was 25 weeks (range, 23–27) and birth weight z-score 0.14 (range, −2.73, 3.28). Infants with CRP elevation on day 7 had lower weights at discharge/transfer (−0.17 z-scores, 95% CI −0.27, −0.06) than infants without CRP elevation; with similar results on day 14. Infants with CRP elevation on day 14 were also shorter (−0.21 length z-scores, 95% CI −0.38, −0.04), and had smaller head circumferences (−0.18 z-scores, 95% CI −0.33, −0.04) at discharge/transfer. IL-6 elevation on day 14 was associated with lower weight (−0.12, 95% CI −0.22, −0.02); IL-6 elevation on day 7 was associated with shorter length (−0.27, 95% CI −0.43, −0.12). TNF-∝ and IL-8 elevation on day 14 were associated with lower weight at discharge/transfer.

Conclusion:

Postnatal systemic inflammation may contribute to impaired nutrient accretion during a critical period in development in infants with extremely low gestational ages.

Despite existing clinical interventions to enhance nutrient delivery,^1–3 newborn infants with extremely low gestational ages struggle to grow during the neonatal intensive care unit (NICU) hospitalization. In the United States, the typical infant born very preterm leaves the NICU with a bodyweight nearly a standard deviation below expected for postmenstrual age.⁴ Further, average body length falls by more than a standard deviation from birth to NICU discharge, indicating linear growth stunting.⁵ Impairments in physical growth reflect the diminished accretion of nutrients during a critical period in development,⁶ and are associated with adverse long-term neurodevelopmental outcomes among infants born extremely preterm.^{7, 8}

In addition to nutrient intake, non-nutritional factors may impact physical growth in the NICU. In rodent models,^9,10 systemic inflammation impairs linear growth via insulin-like growth factor-1 (IGF-1) mediated pathways, independent of nutrient intake. Systemic inflammation is implicated in infant linear growth stunting globally.^11–13 Among hospitalized infants born preterm, systemic inflammation arises from such common NICU exposures as mechanical ventilation, infection, and supplemental oxygen.^14–16 In one U.S. study, higher levels of the pro-inflammatory cytokine interleukin 6 (IL-6) were associated with slower weight gain through 36 weeks of postmenstrual age.¹⁷ Understanding the contribution of systemic inflammation to physical growth impairment is necessary to design effective NICU-based strategies to promote healthy growth and improve long-term outcomes. We aimed to investigate associations of circulating inflammation-related proteins with physical size at NICU discharge. We hypothesized that systemic inflammation, indicated by higher c-reactive protein [CRP], IL-6, tumor necrosis factor-alpha [TNF-α], and interleukin 8 [IL-8]), would be associated with lower body weight, length, BMI, and/or head circumference at NICU discharge or transfer, independent of nutrient intake.

Methods

The Extremely Low Gestational Age Newborn (ELGAN) Study was originally designed to examine antecedents of brain structural and functional disorders in infants born between 23 0/7 and 27 6/7 weeks of gestation. This observational cohort study prospectively enrolled 1506 infants at 14 centers from March 2002 to August 2004. Women were approached for consent either antenatally or shortly after delivery; 83% of eligible infants enrolled. Data were collected from medical records using standardized forms and by maternal interview, with quality control procedures in place. Details about enrollment and data collection have been reported.^{18, 19} For this analysis, we included 1222 infants who survived to discharge or transfer and excluded infants with missing inflammatory protein data (n=114) or with discharge or transfer at chronologic age <28 days or postmenstrual age >42 weeks (n=258). The final sample size for this analysis was 850.

Measures

Inflammatory proteins

Drops of blood remaining from a clinically indicated specimen were collected on filter paper and stored at −80°C in sealed bags with a desiccant until processing. Collection time points during the NICU hospitalization included days 1 (range, 1–3), 7 (range, 5–8), and 28 (range, 26–29). Laboratory work was performed in the Genital Tract Biology Laboratory at Brigham and Women’s Hospital. Proteins were eluted from the stored blood spots as previously described.²⁰ The total protein concentration in each eluted sample was determined by bicinchoninic acid assay (Thermo Scientific, Rockford, IL) using a multi-label Victor 2 counter (Perkin Elmer, Boston, MA). The Meso Scale Discovery platform was used to measure individual proteins in duplicate. The measurement of each protein was normalized to the milligrams of total protein. Of the 27 proteins that were previously assayed for this cohort,²¹ we chose to focus this analysis on CRP, a liver-derived acute-phase protein that, when elevated in the first postnatal month, was linked to white matter damage^{22, 23} and adverse neurodevelopmental outcomes in the ELGAN study;^{24, 25} primary pro-inflammatory cytokines that are linked with linear growth stunting in human pediatric populations or animal models (IL-6, TNF-α)^{9, 11, 13, 26, 27} and with poor neurodevelopmental outcomes in the ELGAN study;²⁴ and IL-8, a proinflammatory chemokine which has been associated with adverse neurodevelopmental outcomes in this cohort.^{24, 28, 29} As we have done previously,³⁰ we defined elevation for each protein and each time point as being in the top quartile, and sustained elevation as being in the top quartile for at least 2 time points.

Infant weight, length, and head circumference

During the NICU hospitalization, infants were weighed and measured according to local clinical practice. ELGAN study staff abstracted from the medical record each participant’s weight, length, and head circumference at birth and discharge or transfer. We derived z-scores using the Olsen intrauterine growth reference.^{31, 32} One z-score corresponds to the number of standard deviations by which the infant’s weight, length, or head circumference is above or below the median for infants of the same postmenstrual age. We excluded implausible length or head circumference measures (z-score >4 or <−4) from analyses.

Covariates

Gestational age was estimated based on a hierarchy of the quality of available information, including 1) dates of embryo retrieval or intrauterine insemination or fetal ultrasound before the 14^th week (60%); 2) fetal ultrasound at 14 or more weeks (29%); 3) last menstrual period without fetal ultrasound (7%); and 4) gestational age recorded in the NICU log (1%).³³ Maternal chorioamnionitis, antenatal steroids, infant sex, length of stay, bronchopulmonary dysplasia (supplemental oxygen requirement at 36 weeks of postmenstrual age), necrotizing enterocolitis (Stage 2 or higher), and medications (postnatal steroids, non-steroidal anti-inflammatory drugs [NSAIDs]) were abstracted from the medical record. Daily protein and energy intakes in the first postnatal month were estimated from parenteral and enteral nutrition recorded on days 7, 14, 21, and 28 of life as previously reported.³⁴ Protein and energy intakes in this cohort were almost exclusively parenteral on day 7 and transitioned to almost exclusively enteral by day 28. In this analysis, we used the mean total value for each participant.

Statistical analyses

Our overall approach was to compare growth outcomes (weight, length, head circumference z-scores for postmenstrual age at discharge/transfer) between infants with vs. without protein elevation. Because baseline size is a determinant of subsequent growth, we adjusted all models for the weight, length, or head circumference z-score for gestational age, corresponding to the outcome in the model. We conceptualized potential confounders as variables that were determinants of both exposure (systemic inflammation) and outcome (size at NICU discharge or transfer), but not in the causal pathway. We adjusted for these variables using a sequential approach. In Model 1, we adjusted for gestational age, sex, and NICU length of stay. In Model 2, we additionally adjusted for estimated mean protein and energy intakes, which are important determinants of growth outcomes.³⁵ In Model 3, we additionally adjusted for maternal chorioamnionitis, antenatal steroids, infant bronchopulmonary dysplasia, postnatal steroid, and non-steroidal anti-inflammatory drug exposure; we excluded infants with necrotizing enterocolitis. We also performed a sensitivity analysis excluding 37 infants with congenital anomalies. All models used generalized estimating equations linear regression and accounted for clustering due to multiple gestation, as recommended.³⁶ We used SAS version 9.4 for all analyses.

Results

Infant characteristics are shown in Table I for the overall cohort and stratified by CRP elevation status. At birth, the mean gestational age was 25 completed weeks (range, 23 to 27) and the mean weight was 859 grams (range, 420 to 1450). The mean birth weight for gestational age z-score was 0.14 (range, −2.73 to 3.28). At NICU discharge or transfer, the mean weight was 2389 grams (range, 971 to 4120), with a corresponding mean weight z-score for postmenstrual age of −1.21 (range, −3.88 to 1.30). The mean length z-score for postmenstrual age was −1.63 (range, −6.14 to 1.59). In the first month, the mean energy intake was 93 kcal/kg/day, and the mean protein intake was 3.6 g/kg/day. Infants with CRP elevation had lower mean energy intake (89 vs. 94 kcal/kg/day, P < .001) compared with infants without CRP elevation. Mean protein intake did not differ between groups (3.7 vs. 3.6 g/kg/day, p=0.2). Forty-two percent of infants had bronchopulmonary dysplasia and 5.1% had necrotizing enterocolitis.

Table 1.

Participant characteristics at birth and NICU discharge or transfer

Measure	N	Mean or number (percent)	Range	Mean or number (percent)	Range	Mean or number (percent)	Range	p-value *
		All (n=850)		With CRP elevation (n=152)		Without CRP elevation (n=698)
Prenatal / maternal
Antenatal steroids	837	744 (88.9%)		132 (88.0%)		612 (89.1%)		0.702
Chorioamnionitis	850	293 (34.5%)		73 (48.0%)		220 (31.5%)		<0.001
Infant
At birth
Gestational age (completed weeks)	850	25.8	23, 27	25.7	23, 27	25.8	23, 27	0.776
Male sex	850	423 (49.8%)		82 (53.9%)		341 (48.9%)		0.225
Singleton	850	568 (66.8%)		117 (77.0%)		451 (64.6%)		0.003
Size
Weight (grams)	850	859	420, 1450	809	420, 1355	869	435, 1450	<0.001
Weight z-score	850	0.14	−2.73, 3.28	−0.21	−2.73, 2.30	0.21	−2.70, 3.28	<0.001
Length (cm)	791	34.0	21, 45	33.5	26, 40	34.1	21, 45	0.038
Length z-score	791	0.18	−4.89, 5.70	−0.02	−2.94, 3.11	0.22	−4.89, 5.70	0.012
Head circumference (cm)	817	23.8	19, 29	23.4	19, 28	23.9	19, 29	<0.001
Head circumference z-score	817	0.15	−3.00, 3.56	−0.20	−2.46, 2.99	0.23	−3.00, 3.56	<0.001
Nutrient provision (parenteral + enteral)
Energy (kcal/kg/day)
Day 7	844	81	0, 146	75	0, 146	82	28, 144	<0.001
Day 14	844	91	0, 182	87	0, 132	91	0, 182	0.027
Day 21	844	98	0, 197	94	0, 149	99	0, 197	0.023
Day 28	844	102	0, 169	99	0, 168	103	5, 169	0.027
Mean	844	93	46, 146	89	51, 126	94	46, 146	<0.001
Protein (g/kg/day)
Day 7	844	4.2	0, 12.9	4.1	0, 8.1	4.2	0, 12.9	0.541
Day 14	844	3.7	0, 8.7	3.8	0, 8.3	3.6	0, 8.7	0.215
Day 21	844	3.4	0, 12.0	3.7	0, 8.2	3.4	0, 12.0	0.025
Day 28	844	3.3	0, 8.9	3.3	0, 7.5	3.3	0, 8.9	0.532
Mean	844	3.6	0, 8.2	3.7	0.1, 7.6	3.6	0, 8.2	0.237
Duration of parenteral nutrition or IV fluids
<28 days	850	491 (57.58%)		60 (39.5%)		431 (61.8%)		<0.001
≥28 days	850	359 (42.2%)		92 (60.5%)		267 (38.2%)
Postnatal steroids	850	141 (16.6%)		24 (15.8%)		117 (16.8%)		0.770
Postnatal NSAIDs	828	62 (7.5%)		13 (8.7%)		49 (7.2%)		0.526
At discharge or transfer
Postmenstrual age (weeks)	850	37.1	30, 41	37.5	31, 41	37.0	30, 41	0.039
Length of stay (days)	850	79.7	29, 129	82.5	30, 125	79.1	29, 129	0.070
Size
Weight (grams)	850	2389	971, 4120	2318	1200, 3716	2405	971, 4120	0.060
Weight z-score	850	−1.21	−3.88, 1.30	−1.54	−3.88, 0.60	−1.14	−3.73, 1.30	<0.001
Length (cm)	635	44.8	31, 54	44.6	33, 51	44.9	31, 54	0.416
Length z-score	635	−1.63	−6.14, 1.59	−1.98	−5.45, −0.09	−1.55	−6.14, 1.59	<0.001
Head circumference (cm)	545	32.6	23, 45	32.7	29, 44	32.6	23, 45	0.806
Head circumference z-score	545	−0.60	−5.50, 8.23	−0.79	−2.80, 7.01	−0.56	−5.50, 8.23	0.049
Diagnosis
Bronchopulmonary dysplasia	845	354 (41.9%)		78 (52.0%)		276 (39.7%)		0.006
Necrotizing enterocolitis	850	43 (5.1%)		17 (11.2%)		26 (3.7%)		<0.001
Change in size, birth to discharge (z-scores)
Weight	850	−1.35	−5.43, 1.55	−1.33	−3.19, 0.40	−1.35	−5.43, 1.55	0.845
Length	582	−1.78	−5.58, 1.84	−1.92	−4.51, 0.35	−1.75	−5.58, 1.84	0.160
Head circumference	628	−0.77	−4.68, 4.30	−0.71	−4.42, 2.05	−0.79	−4.68, 4.30	0.472

N for individual variables may be <850 due to missing data. Z-scores for age and sex are based on the Olsen reference. Energy and protein provision shown are intakes from all parenteral and enteral sources. CRP elevation was defined as being in the top quartile on 2 or more of postnatal days 1, 7, and 14.

^*p-values comparing those with elevated vs. not elevated CRP from 2-sample t-test when comparing means and chi-square test when comparing percentages.

Table 2 shows the adjusted associations of CRP, IL-6, TNF-α, and IL-8 elevation with weight z-score at NICU discharge/transfer. After adjustment for birth weight z-score, gestational age at birth, sex, and NICU length of stay (Model 1), infants with CRP elevation on days 7 and 14, and infants with sustained CRP elevation, had lower weight z-scores at NICU discharge. Additional adjustment for protein and energy intakes (Model 2) slightly attenuated these estimates, whereas further adjustment for potential confounders and exclusion of infants with necrotizing enterocolitis (Model 3) had little effect. Findings were similar for IL-6 elevation, although only elevation on day 14 and sustained elevation remained statistically significant (confidence intervals excluded the null hypothesis) in Models 2 and 3. Elevation of TNF-α and IL-8 on day 14 were associated with lower weight z-score at discharge in all models. Results changed minimally after excluding 37 infants with congenital anomalies (data not shown).

Table 2.

Inflammatory protein elevation and weight z-score at NICU discharge in infants with extremely low gestational ages (<28 weeks) (n=850)

	Model 1			Model 2			Model 3
	Beta	95% CI	P-value	Beta	95% CI	P-value	Beta	95% CI	P-value
CRP
Day 1	0.03	−0.07, 0.13	0.59	0.05	−0.05, 0.14	0.36	0.01	−0.10, 0.11	0.92
Day 7	−0.22	−0.33, −0.12	<0.001	−0.15	−0.25, −0.05	0.004	−0.17	−0.27, −0.06	0.002
Day 14	−0.25	−0.36, −0.15	<0.001	−0.19	−0.30, −0.08	0.001	−0.21	−0.32, −0.10	<0.001
Sustained	−0.22	−0.33, −0.11	<0.001	−0.14	−0.25, −0.03	0.011	−0.19	−0.30, −0.08	<0.001
IL-6
Day 1	−0.03	−0.13, 0.07	0.56	−0.04	−0.14, 0.05	0.37	−0.03	−0.13, 0.07	0.50
Day 7	−0.13	−0.22, −0.03	0.012	−0.07	−0.13, 0.03	0.18	−0.09	−0.20, 0.01	0.09
Day 14	−0.15	−0.26, −0.05	0.003	−0.13	−0.23, −0.04	0.006	−0.12	−0.22, −0.02	0.016
Sustained	−0.21	−0.32, −0.11	<0.001	−0.15	−0.25, −0.05	0.003	−0.15	−0.26, −0.04	0.007
TNF-α
Day 1	0.05	−0.05, 0.16	0.32	0.01	−0.08, 0.11	0.79	0.03	−0.08, 0.13	0.64
Day 7	0.02	−0.08, 0.12	0.74	0.03	−0.06, 0.13	0.48	0.04	−0.06, 0.14	0.45
Day 14	−0.17	−0.27, −0.06	0.002	−0.14	−0.23, −0.04	0.006	−0.12	−0.22, −0.02	0.016
Sustained	−0.05	−0.16, 0.06	0.36	−0.04	−0.14, 0.06	0.45	−0.01	−0.10, 0.09	0.90
IL-8
Day 1	0.14	0.04, 0.24	0.007	0.12	0.02, 0.22	0.019	0.10	−0.00, 0.20	0.056
Day 7	−0.05	−0.15, 0.05	0.33	−0.01	−0.11, 0.09	0.83	−0.03	−0.13, 0.07	0.56
Day 14	−0.24	−0.35, −0.13	<0.001	−0.19	−0.30, −0.08	0.001	−0.17	−0.28, −0.06	0.002
Sustained	−0.07	−0.19, 0.06	0.28	−0.03	−0.15, 0.09	0.60	−0.03	−0.15, 0.09	0.58

Beta coefficients represent estimated mean differences in weight z-score at discharge/transfer for infants in the top quartile for each inflammation-related protein vs. not in the top quartile. ‘Sustained’ indicates being in the top quartile on at least 2 days. Estimates in Model 1 are adjusted for birth weight z-score, gestational age at birth, sex, and NICU length of stay. Model 2 is additionally adjusted for mean protein and energy intakes over the first month. Model 3 is additionally adjusted for bronchopulmonary dysplasia, antenatal steroids, chorioamnionitis, postnatal steroids, and postnatal NSAIDs, and excludes infants with NEC (n=43). All models account for clustering due to multiple gestation. Bold face type indicates confidence limits that exclude the null hypothesis.

Table 3 shows the adjusted associations of inflammatory protein elevation with body length z-score at NICU discharge/transfer. CRP elevation on day 14, IL-6 elevation on day 7, and sustained IL-6 elevation were associated with lower body length in all 3 models. For head circumference at NICU discharge/transfer, (Table 4; available at www.jpeds.com), CRP elevation on days 7 and 14 was associated with a smaller head z-score at discharge in all 3 models.

Table 3.

Inflammatory protein elevation and length z-score at NICU discharge in infants with extremely low gestational ages (<28 weeks) (n=618)

	Model 1			Model 2			Model 3
	Beta	95% CI	P-value	Beta	95% CI	P-value	Beta	95% CI	P-value
CRP
Day 1	−0.03	−0.21, 0.14	0.70	−0.03	−0.19, 0.14	0.76	−0.06	−0.25, 0.13	0.55
Day 7	−0.20	−0.37, −0.03	0.022	−0.13	−0.29, 0.04	0.13	−0.15	−0.32, 0.03	0.10
Day 14	−0.28	−0.44, −0.12	<0.001	−0.21	−0.38, −0.05	0.013	−0.21	−0.38, −0.04	0.014
Sustained	−0.21	−0.40, −0.02	0.027	−0.14	−0.33, 0.04	0.12	−0.18	−0.37, 0.02	0.072
IL-6
Day 1	−0.13	−0.28, 0.02	0.08	−0.15	−0.29, −0.00	0.046	−0.19	−0.34, −0.03	0.019
Day 7	−0.30	−0.44, −0.15	<0.001	−0.24	−0.39, −0.10	0.001	−0.27	−0.43, −0.12	<0.001
Day 14	−0.12	−0.31, 0.06	0.19	−0.08	−0.27, 0.10	0.38	−0.09	−0.28, 0.11	0.38
Sustained	−0.38	−0.55, −0.21	<0.001	−0.33	−0.50, −0.15	<0.001	−0.34	−0.54, −0.15	<0.001
TNF-α
Day 1	0.00	−0.19, 0.19	0.99	−0.02	−0.20, 0.16	0.82	−0.05	−0.24, 0.15	0.64
Day 7	0.01	−0.17, 0.16	0.92	−0.00	−0.16, 0.16	0.97	−0.04	−0.20, 0.13	0.65
Day 14	−0.00	−0.19, 0.18	0.96	0.01	−0.16, 0.19	0.88	0.05	−0.14, 0.23	0.63
Sustained	−0.01	−0.19, 0.17	0.93	−0.01	−0.18, 0.17	0.95	−0.01	−0.20, 0.17	0.90
IL-8
Day 1	0.07	−0.24, 0.09	0.40	−0.09	−0.24, 0.07	0.27	−0.12	−0.29, 0.05	0.17
Day 7	−0.12	−0.31, 0.07	0.20	−0.10	−0.29, 0.08	0.28	−0.12	−0.31, 0.07	0.21
Day 14	−0.18	−0.37, 0.01	0.064	−0.15	−0.34, 0.12	0.12	−0.13	−0.33, 0.07	0.19
Sustained	−0.19	−0.39, 0.02	0.073	−0.16	−0.35, 0.04	0.12	−0.12	−0.33, 0.08	0.24

Beta coefficients represent estimated mean differences in length z-score at discharge/transfer for infants in the top quartile for each inflammation-related protein vs. not in the top quartile. ‘Sustained’ indicates being in the top quartile on at least 2 days. Estimates in Model 1 are adjusted for birth length z-score, gestational age at birth, sex, and NICU length of stay. Model 2 is additionally adjusted for mean protein and energy intakes over the first month. Model 3 is additionally adjusted for bronchopulmonary dysplasia, antenatal steroids, chorioamnionitis, postnatal steroids, and postnatal NSAIDs, and excludes infants with NEC (n=43). All models account for clustering due to multiple gestation. Bold face type indicates confidence limits that exclude the null hypothesis.

Table 4 (online only).

Inflammatory protein elevation and head circumference z-score at NICU discharge in infants with extremely low gestational ages (<28 weeks) (n=651).

	Model 1			Model 2			Model 3
	Beta	95% CI	P-value	Beta	95% CI	P-value	Beta	95% CI	P-value
CRP
Day 1	0.14	−0.00, 0.29	0.056	0.14	−0.01, 0.28	0.066	0.09	−0.06, 0.24	0.24
Day 7	−0.25	−0.41, −0.08	0.004	−0.18	−0.34, −0.01	0.040	−0.20	−0.38, 0.02	0.03
Day 14	−0.27	−0.41, −0.12	<0.001	−0.17	−0.31, −0.03	0.018	−0.18	−0.33, −0.04	0.012
Sustained	−0.17	−0.34, −0.01	0.042	−0.10	−0.27, 0.07	0.23	−0.17	−0.35, 0.01	0.07
IL-6
Day 1	−0.07	−0.21, 0.07	0.33	−0.08	−0.23, 0.05	0.23	−0.10	−0.24, 0.05	0.19
Day 7	−0.07	−0.23, 0.10	0.42	0.01	−0.15, 0.17	0.91	0.04	−0.13, 0.22	0.64
Day 14	−0.08	−0.25, 0.09	0.38	−0.02	−0.18, 0.15	0.84	0.01	−0.17, 0.18	0.95
Sustained	−0.06	−0.23, 0.11	0.47	0.01	−0.15, 0.17	0.92	0.04	−0.13, 0.22	0.61
TNF-α
Day 1	0.10	−0.06, 0.26	0.21	0.08	−0.08, 0.23	0.34	0.07	−0.10, 0.24	0.41
Day 7	0.05	−0.10, 0.19	0.51	0.07	−0.07, 0.21	0.31	0.11	−0.03, 0.25	0.13
Day 14	0.05	−0.13, 0.23	0.60	0.09	−0.09, 0.27	0.33	0.11	−0.08, 0.30	0.25
Sustained	0.08	−0.09, 0.25	0.36	0.08	−0.09, 0.25	0.35	0.10	−0.07, 0.28	0.24
IL-8
Day 1	0.05	−0.11, 0.22	0.53	0.02	−0.14, 0.18	0.81	−0.01	−0.19 0.16	0.88
Day 7	0.01	−0.14, 0.16	0.87	0.05	−0.10, 0.20	0.53	0.04	−0.12, 0.19	0.64
Day 14	−0.08	−0.29, 0.12	0.42	−0.02	−0.22, 0.18	0.83	0.01	−0.20, 0.23	0.90
Sustained	−0.06	−0.25, 0.14	0.58	−0.02	−0.22, 0.18	0.84	−0.04	−0.25, 0.16	0.68

Beta coefficients represent estimated mean differences in head circumference z-score at discharge/transfer for infants in the top quartile for each inflammation-related protein vs. not in the top quartile. ‘Sustained’ indicates being in the top quartile on at least 2 days. Estimates in Model 1 are adjusted for birth head circumference z-score, gestational age at birth, sex, and NICU length of stay. Model 2 is additionally adjusted for mean protein and energy intakes over the first month. Model 3 is additionally adjusted for bronchopulmonary dysplasia, antenatal steroids, chorioamnionitis, postnatal steroids, and postnatal NSAIDs, and excludes infants with NEC (n=43). All models account for clustering due to multiple gestation. Bold face type indicates confidence limits that exclude the null hypothesis.

The Figure summarizes the results from Model 3 across all pro-inflammatory proteins and growth outcomes and illustrates that CRP elevation on day 14 was associated with lower weight, length, and head circumference z-scores at discharge/transfer.

Figure.

Inflammatory protein elevation and infant size at NICU discharge or transfer in infants with extremely low gestational ages (<28 weeks). Weight is shown in Panel A, length in Panel B, and head circumference in Panel C. Beta coefficients from Model 3 (indicated by squares for weight, triangles for length, diamonds for head circumference) represent the estimated mean differences in weight, length, or head circumference z-score at discharge/transfer for infants in the top quartile for each inflammatory protein compared with a reference group of infants not in the top quartile. ‘Sustained’ elevation indicates being in the top quartile on at least 2 of 3 available days (1, 7, and/or 14).

Discussion

In this longitudinal cohort study of over 800 infants born <28 weeks of gestation, we found that elevations of CRP and IL-6 in the first 2 weeks after birth were associated with lower weight and length at NICU discharge/transfer, independent of protein and energy intakes. Additionally, CRP elevation was associated with smaller head circumference, and TNF-α and IL-8 elevation were associated with lower weight. The magnitude of the effect sizes was clinically important; for example, body length was 0.21 z-scores shorter at NICU discharge/transfer among infants with vs. without CRP elevation on day 14.

Our results confirm and extend the results from 2 other cohort studies^{17, 37} that examined relationships of systemic inflammation, indexed by circulating proteins, with physical growth outcomes in infants born preterm. Consistent with our findings, Denson et al reported an association of higher IL-6 at day 21 with slower weight gain velocity through 36 weeks of postmenstrual age or NICU discharge in a U.S. cohort of 768 infants born <1000 grams.¹⁷ An Argentinian study reported that the presence of a systemic inflammatory response, characterized by elevations in CRP and procalcitonin during the first 28 days of life, was associated with impairments in weight gain, linear growth, and head growth that were present at NICU discharge and—importantly—persisted to 12 months of corrected age. Overall, the limited available evidence is consistent in indicating that among hospitalized infants born preterm, systemic inflammation early in life is associated with impaired nutrition accretion, as measured by lower weight, length, and/or head size through NICU discharge. Further, these impairments may persist well beyond NICU discharge.

It is important to consider the role of lower nutrient intake because sicker infants receive less nutritional support than less sick infants.³⁸ In our study, infants with CRP elevation received less energy than infants without CRP elevation, although protein intakes did not differ between groups. It is also important to consider that infants with similar nutrient intakes accrete nutrients differently depending on their underlying inflammation-related illness.³⁹ This latter concept is supported by both our study and the Denson study because adjustment for energy and/or protein intake attenuated but did not fully eliminate the observed associations of inflammatory biomarkers with growth outcomes.

We were particularly interested in examining systemic inflammation in relation to linear growth stunting given the relevance of linear growth in early life to long-term neurodevelopmental outcomes in preterm and other populations.^{8, 40} Complementing our study, cohort studies from other contexts–including pediatric patients with inflammatory bowel disease⁴¹ and infants in low resource settings^11–13–support the concept that systemic inflammation is part of the “nutritional ecology” of linear growth stunting.⁴² Additionally, a causal link is supported by our understanding of the biological mechanisms connecting protein and energy intakes with linear growth of long bones and lean tissues via the insulin-growth hormone–insulin-like growth factor 1 (IGF-1) axis, and inhibition of these pathways by pro-inflammatory cytokines, including IL-6.^{11, 43} Consistent with this proposed mechanism, in a separate analysis of data from the ELGAN cohort, CRP elevation was associated with lower IGF-1.⁴⁴ Additionally, in a small cohort of infants born preterm, we previously linked clinical markers of inflammatory-based comorbidities (supplemental oxygen requirement, antibiotics) with linear growth suppression.⁴⁵ Overall, a causal role of systemic inflammation in inhibiting linear growth via cytokines inhibiting IGF-1 is supported by mounting evidence from both mechanistic animal studies and human studies across several populations.

The timing of inflammatory protein elevation and growth outcomes is also of interest. In our study, elevations of CRP and pro-inflammatory cytokines on days 7 and/or 14 were associated with poorer growth outcomes, whereas elevations on day 1 were generally not. Our results suggest that early, transient inflammation may not be sufficient to measurably impair physical growth. Similarly, day 1 inflammatory protein elevations were largely not associated with adverse cognitive and neurologic outcomes in this cohort.^46–48 It is plausible that inflammation present on day 1 reflects different (e.g. intrauterine) exposures compared with subsequent inflammation which reflects postnatal exposures.

Strengths of our study include a large, prospective, 14 NICU ELGAN cohort with repeatedly measured serum inflammatory markers. We expect relatively high generalizability of our findings to other infants with extremely low gestational ages, but our findings may not generalize to infants who are more mature at birth. Although clinical measurements of infant length are subject to error and may differ systematically between centers, we were nonetheless able to detect associations. We examined linear growth as a proxy for lean tissue accretion but lacked data on body composition. Although we estimated and adjusted for protein and energy intakes averaged over the first month of life, residual confounding by diet is still possible, especially because we lacked nutritional data after 1 month of age. Energy intake was relatively low in this cohort (mean 93 kcal/kg/day vs. recommended 90–115 parenteral and 110–150 enteral kcal/kg/day),⁴⁹ likely reflecting typical nutritional practices nearly 20 years ago when these data were collected. We included infants across the full range of birth weight-for-gestational age. Associations of inflammation with discharge size could differ for small- compared with appropriate-for-gestational age infants, but we did not have adequate power to rigorously assess this interaction. We addressed the non-independence of multiple gestations by accounting for clustering in our statistical models, as recommended.³⁶ Like all observational studies, the potential for residual confounding by unmeasured variables is also a limitation.

In conclusion, we found that protein biomarkers of systemic inflammation–most prominently CRP and IL-6–were associated with impairments in weight gain and linear growth among infants with extremely low gestational age during the NICU hospitalization. These results imply that in addition to ensuring adequate nutrient provision, strategies to reduce systemic inflammation may promote greater nutrient accretion in the NICU, and ultimately translate to improved neurodevelopmental and other growth-related long-term outcomes.

Abbreviations:

ELGAN

Extremely low gestational age

CRP

C-reactive protein

IL-6

Interleukin 6

TNF-α

Tumor necrosis factor-alpha

IL-8

Interleukin 8

IGF-1

Insulin-like growth factor 1

NICU

Neonatal intensive care unit