Blood pressure and kidney function in neonates and young infants with intrauterine growth restriction

IUGR neonates admitted to our neonatal intensive care unit (NICU) or to our newborn nursery at the University Hospital Bonn between April 2019 and September 2020 were eligible for study participation if they remained hospitalised in our NICU and/or newborn nursery for at least 72 h after birth and prenatal findings met the German, Austrian and Swiss 2017 consensus guideline for IUGR (“fetal estimated weight < 10th percentile and/or non-percentile growth during the progress and pathological Doppler sonography of the umbilical or uterine artery or oligohydramnion”) [4]. The study period ended with discharge from the hospital. All neonates admitted to our NICU and newborn nursery during the same period, who did not meet the criteria for IUGR with a birth weight between the 10th and the 90th percentile and who remained hospitalised in our NICU and/or newborn nursery for at least 72 h after birth, were eligible to be enrolled as controls. Exclusion criteria for cases and controls were chromosomal anomalies, congenital malformations of the kidneys or urinary tract, haemodynamically relevant congenital heart defects, congenital diaphragmatic hernia (CDH), the need for extracorporeal membrane oxygenation (ECMO) therapy, unstable condition according to the assessment of the treating physicians, perinatal asphyxia or death during hospitalisation. Informed written consent was obtained from parents or legal representatives before enrolment in the study. This study was approved by the local Ethics Committee of the University of Bonn Medical School (22.05.2020/No. 079/19).

We retrospectively extracted the following information from the mothers’ and children’s patient records: sex, birth weight, gestational age, length at birth, APGAR scores, pH from the umbilical artery, occurrence of birth complications, birth mode, maternal age, gravida, body mass index, maternal morbidity, medication during pregnancy, drug abuse in pregnancy, single or multiple pregnancy, pregnancy complications, blood pressure levels, protein intake, medication, drug levels in serum when taking nephrotoxic drugs (vancomycin, tobramycin), body weight and length on discharge and duration of hospitalisation.

Both invasive and non-invasive blood pressure measurements were considered. Systolic, diastolic and mean arterial pressures were calculated as means of individual measurements during a predefined period of time (BP1: between 36th and 72nd hour of life, BP2: 8th day of life, BP3: last five measurements before discharge). Since the time of discharge significantly differed between cases, there is a large variability in timing of BP3 (between 3rd and 161st day of life). In cases where children were discharged before the 8th day of life (10/99), no BP2 could be registered and the blood pressure before discharge was registered (BP3). In order to control for patient’s age and gestational age, we compared the blood pressure measurements to the patients’ individual reference values obtained from Pejovic et al. [21].

Serum creatinine (SCr) and blood urea (BUN) levels were measured only in the context of medically necessary blood sampling. The first measurement was between the 30th and 120th hour of life (serum 1), mostly at the time of the newborn screening (between 36th and 72nd hour of life). Also, a second blood sample was obtained only when a blood draw was medically indicated (e.g. repeat newborn screening in preterm neonates or infants) and therefore shows a high variability with regard to timing (between 5th and 53rd day of life) (serum 2). We compared the measured SCr and BUN values with corresponding reference values [22, 23] to investigate whether one group (IUGR or controls) exceeded the limits more frequently than the other. Urine was collected regardless of clinical indication with consent of the legal guardians between the 30th and 240th hour of life. Non-invasive measures such as urine bags or insertion of compresses into the diaper were used to obtain spontaneous urine unless a urinary catheter was medically indicated. To detect proteinuria, we measured albumin, α-1-microglobulin, transferrin, immunoglobulin G and total protein in spontaneous urine and normalized the measured values by urine creatinine level. We compared the measured urine excretion of the listed parameters with respective specific limits [24‐26], except for transferrin, for which no neonatal reference values are available, to again investigate whether one group (IUGR or controls) exceeded the limits more frequently than the other. All collected blood and urine samples were processed immediately at the central laboratory facility of the University Hospital Bonn. SCr, urine creatinine and BUN levels were measured using VIS photometry on a cobas c702 analyser with the reagent CREJ2 for creatinine and UREAL for urea (Roche Diagnostics). Albumin, immunoglobulin G and total protein were measured using turbidimetry with the specific reagent on a cobas c702. Turbidimetry was also used to measure α-1-microglobulin levels but on a cobas c502. To determine the transferrin content, immunnephelometry was used on a BN ProSpec. If a value was below the respective detection limit, calculations were performed using a constant value below that of the lower limit. The respective limits of detection (LODs) in the urine and values used for further calculation are listed here: creatinine: LOD 4.2 mg/dl, used constant value 4.1 mg/dl; albumin: LOD 3 mg/l, used constant value 2.9 mg/l; α-1-microglobulin: LOD 5 mg/l, used constant value 4.9 mg/l; transferrin: LOD 2.3 mg/l, used constant value 2.2 mg/l; and total protein: LOD 40 mg/l, used constant value 39.9 mg/l. In one case, the upper detection limit of α-1-microglobulin (LOD 400 mg/l) was exceeded; here, we calculated with a constant value (401 mg/l) above the detectable range. An overview of the recorded laboratory values is shown in Fig. 1.

During inpatient treatment, the protein intake of all patients in our department is automatically calculated by our medical documentation system. To exclude protein intake as a major confounder influence of BUN concentration, we extracted the respective intakes at both times of serum sampling from the individual patient’s medical record and compared it between cases and controls.

Statistical analyses were performed using R version 3.6.2 [27]. Comparisons of the discrete data of both groups, IUGR and non-IUGR neonates, were conducted using Pearson’s chi-squared test or Fisher’s exact test in case of expected cell counts below 5 in the respective contingency table. Student’s t test and Mann–Whitney U test for normally distributed and non-normally distributed data, respectively, were applied to the collected metric data. The Shapiro–Wilk test with an error level of p = 0.01 was used to check whether there was a normal distribution for the collected data. Starting from an alpha error of 5% for a two-sided test, Bonferroni correction was applied to neutralise the accumulation of alpha errors in multiple comparisons. Due to the variable timing of the second serum analysis (together with a medically indicated blood sampling) and BP3 (prior to discharge), we performed an analysis of covariance (ANCOVA) to control for timing as a confounder. Percentile and z scores were adapted from Voigt et al. [28].

During the period of our study, 130 neonates were born with IUGR in the University Hospital Bonn. Of these 130 neonates, 62 were not eligible for inclusion because they met at least one of the exclusion criteria. An additional 29 IUGR neonates could not be included as there was no parental consent for participation in the study (Fig. 1). These neonates did not differ significantly from participants in gestational age, birth weight or sex.

Of the neonates participating in the study, 39 were enrolled in the IUGR case group with an average gestational age of 31.90 weeks (standard deviation (SD) 3.46 weeks). Sixty neonates were enrolled in the control group, with an average gestational age of 32.45 weeks (SD 2.91 weeks). As entailed by the definition of IUGR, birth weight was significantly lower in the IUGR group 1.38 kg (SD 0.52 kg) than in the control group 1.96 kg (SD 0.58 kg) (mean z scores: IUGR group − 1.43, SD 0.58; control group − 0.08, SD 0.58). Consecutively, birth length (39.23 cm, SD 5.12 cm; mean z score − 1.33, SD 0.74) in the IUGR group was significantly lower than in the control group (43.83 cm, SD 4.48 cm; mean z score − 0.02, SD 0.88). The mean weight-for-length z scores were − 3.01 (SD 1.92) in the IUGR group and 0.23 (SD 2.36) in the control group. Expectedly, body weight at discharge was still lower in IUGR patients (2.08 kg, SD 0.28 kg; mean z score − 2.53, SD 0.66) than in control patients (2.39 kg, SD 0.32 kg; mean z score − 1.40, SD 0.60, p < 0.001). The same was true for body length at discharge (IUGR 43.64 cm, SD 1.97 cm; mean z score − 2.96, SD 0.93; controls 46.74 cm, SD 2.85 cm; mean z score − 1.22, SD 1.30; corrected p < 0.001). Z scores at discharge are based on the corrected gestational age. Other demographic data did not differ (see Table 1). Neonates in the IUGR group were more likely to be treated with blood pressure-increasing drugs like catecholamines and glucocorticoids at the time of the first measurement (BP1) than neonates in the control group. This difference was not significant after Bonferroni correction for multiple comparisons (p = 0.039). The application of the first line antibiotic tobramycin was comparable in both groups, whereas the IUGR cases received second-line antibiotic treatment with vancomycin non-significantly (p = 0.039) more frequently during hospitalisation. In every patient, vancomycin and tobramycin were monitored by serum-level controls, and none of the levels measured reached the nephrotoxic range (tobramycin > 12 µg/ml, vancomycin > 30 µg/ml). There was no acute kidney injury (AKI, based on neonatal AKI KDIGO classification [29]) observed in the participants of the IUGR or the control group. The length of hospitalisation and hence the study period differed between cases and controls, with a median hospital stay of 31 days for IUGR cases compared to 19 days for controls.

Figure 1 displays the data available for each time point of blood pressure measurement. There was no significant difference in blood pressure between cases and controls at any time (Online Resource 1). Compared with their respective normal values, IUGR patients did not exceed (χ² = 0.19, p = 0.66) or fall below (χ² = 0, p = 1) their normal range obtained from Pejovic et al. [21] more often than control patients did. To account for the large variability in the timing of BP3 (prior to discharge), we performed an analysis of covariance controlling for measurement timing as a confounder. Again, we found no significant effect of IUGR on systolic (p = 0.82), mean (p = 0.24) or diastolic (p = 0.16) blood pressure values in BP3. Blood pressure measurements for both groups are comparatively visualized in Fig. 2.

In total, 97/99 (38/39 IUGR, 59/60 control) neonates had measurement data available at the first measuring time (serum 1). In 81 of these 97 neonates (29 IUGR, 52 control), a serum sample could be obtained between the 36th and 72nd hour of life. The second blood sampling (serum 2) could be obtained at 16.5 days of life (median) (interquartile range (IQR) 11.00 − 25.75 days, min 5 days, max 53 days) in 28/39 IUGR cases and 42/60 controls. In 34 study participants (14/39 IUGR, 20/60 control), the second measurement (serum 2) was taken as part of the repeat newborn screening with a corrected gestational age of 32.0 weeks. At both times of blood sampling, SCr levels were lower in the IUGR group (serum 1 0.71 [IQR 0.66–0.95] mg/dl, serum 2 0.40 [IQR 0.33–0.46] mg/dl) than in the control group (serum 1 0.85 [IQR 0.69 − 0.99] mg/dl, serum 2 0.48 [IQR 0.38 − 0.59] mg/dl) (Fig. 3). The same finding was observed for BUN levels in the IUGR group (serum 1 30.6 [IQR 21.6 − 41.6] mg/dl, serum 2 11.9 [IQR 8.4–15.6] mg/dl) compared to the control group (serum 1 47.2 [IQR 33.9–64.4] mg/dl, serum 2 19.8 [IQR 12.6–26.6] mg/dl) (Fig. 3). After Bonferroni correction, the results remained significant (p < 0.0125) for SCr and BUN collected between the 30th and 120th hour of life as well as for SCr collected between the 5th and 53rd day of life. This was confirmed by an analysis of covariance controlling for the highly variable timing of the second blood sampling as a confounder. Compared by χ² testing with Bonferroni correction, participants with IUGR did not exceed normal values for SCr and BUN more often than the control group (Online Resource 2). Urine collection was successful in 87/99 participants (34/39 IUGR, 53/60 control). In 32 of these 87 neonates (8 IUGR, 24 control), a urine sample could be obtained between the 36th and 72nd hour of life. There was no significant difference in protein/creatinine ratio between cases and controls for albumin (p = 1), total protein (p = 0.332), α-1-microglobulin (p = 1), immunoglobulin G (p = 1) and transferrin (p = 1). Except for transferrin, for which no neonatal reference values are available, the respective specific limits for urinary excretion of the above parameters were not exceeded significantly more often by the IUGR neonates compared with the control group (Online Resource 2).

There was no significant difference (p = 0.504) in protein supply between the groups during the first 30 to 120 h of life (IUGR 1.9 [IQR 1.4–2.5] g/kg body weight; controls 1.8 [IQR 1.2–2.4] g/kg body weight). At the time of the second serum measurement (serum 2), the protein intake was significantly higher (p = 0.01) in the IUGR group (3.8 [IQR 2.8–4.7] g/kg body weight) compared to controls (3.3 [IQR 2.6–3.9] g/kg body weight).