Reply to: The Severity of Neuronal Damage in Neonatal Hypoxic-Ischemic Encephalopathy: Does Vitamin D Status Matter?

The Severity of Neuronal Damage in Neonatal Hypoxic–Ischemic Encephalopathy: Does Vitamin-D Status Matter?

We thank Drs. Panda and Sharawat for their interest in our study assessing the role of vitamin D in brain injury after neonatal hypoxic–ischemic encephalopathy (HIE). We hope that the following will provide some additional context regarding the thoughtful points that they raised in their letter.

With regard to vitamin-D dosage, our neonatal intensive care unit (NICU) protocol during the study period was to provide higher vitamin-D supplementation to infants with low vitamin-D levels (<30 mg/dL). This protocol was for all infants admitted to the NICU and was not directed by our study design. Based on this prescribing practice, the trend toward improved respiratory outcomes in those infants receiving higher doses of vitamin D could be interpreted as an association between lower initial 25-hydroxyvitamin D (25OHD) levels and decreased ventilator days. As Drs. Panda and Sharawat mentioned, this assumption would be inverse to our hypothesis. As a response to this concern, we have rerun the regression model between vitamin-D dose and ventilator days, adding initial 25OHD level as a covariate. In the new multivariate model, we found that the relationship between cumulative vitamin-D dose in the first week of life and the number of days on a ventilator is actually stronger when adjusting for initial 25OHD level (p = 0.012) than it was uncorrected. This result gives additional support to our hypothesis that low vitamin-D levels are related to adverse outcomes and vitamin-D dosing may attenuate the phenotypic severity in this population.

Additionally, we agree that there are important potential confounding variables that should be considered when discussing outcomes in neonatal HIE. Sex and gestational age are not confounders in our analysis since neither was associated with 25OHD levels (p = 0.31 and 0.17, respectively). Although low 25OHD levels have previously been associated with prematurity (<32 weeks of gestation at birth),[1] we did not see an association between gestational age and 25OHD because our study only included term and late preterm (>35 weeks of gestation) infants with HIE. Correcting for birth weight, which was positively associated with 25OHD levels (R = 0.49, p = 0.02), does decrease the magnitude and significance of the association between 25OHD levels and magnetic resonance imaging (MRI) injury (β = −0.21, p = 0.09), so birth weight should certainly be corrected in future prospective trials. Given the small sample size and exploratory nature of our pilot study, however, the decreased significance does not change our overall conclusions from the study, and we look forward to future prospective studies that will be able to analyze these questions with more statistical certainty.

We also appreciate the additional discussion regarding possible unmeasured comorbidities. This is a potential limitation to our study and should have been mentioned as such in the discussion. During the early part of our retrospective study period, it was not yet standard for all infants with HIE to receive continuous encephalography in this NICU setting. As such, we were not able to record electrographic seizures nor amplitude-integrated EEG background patterns as part of this study. We also did not collect data on neonatal sepsis. Although the negative effects of sepsis could certainly worsen outcomes for this population, we would expect between 0 and 1 infant in our study to have sepsis (rate of 2.3% in infants with HIE in U.S. children's hospitals),[2] so even if we had collected infectious information, we would not be able to feasibly perform a subanalysis given the low prevalence.

While several other electrolytes, such as calcium, phosphorus, and alkaline phosphorus are important endocrinological outcomes to assess after vitamin-D supplementation, this study was intended to specifically assess neurological outcomes, so we did not record those data. Similarly, although there may be differences in the amount of active unbound 25OHD between the patients in our study, the clinical chemiluminescent immunoassay (CLIA) that was performed on our patients measures total (bound and unbound) 25OHD, thus the 25OHD levels in our study would not be affected by albumin levels.[3] Measuring the amount of unbound 25OHD would certainly be a valuable addition for future studies, but since only approximately 10% of bound 25OHD is bound to albumin, with most of the remaining being bound to vitamin-D binding protein,[4] we cannot estimate free 25OHD using albumin levels alone.

We agree that the levels of deficiency/insufficiency seen in our unit will differ from other units, limiting the direct comparison of rates across studies due to the significant effects of location on population 25OHD levels.[5] This is in part why we analyzed 25OHD as a continuous variable to improve the interunit applicability, as well as avoiding issues due to the controversy surrounding exact cut-offs to diagnose deficiency and insufficiency. Although we do not have historical healthy controls of CLIA-measured 25OHD levels to compare with, we do know that CLIA tends to overestimate 25OHD levels by an average of 13.5 ng/mL.[6] Thus, if we subtract this from each of our patients' levels to provide a rough comparison to the previous study by Lowe et al,[7] our adjusted rates of deficiency/insufficiency would be very similar to theirs, with 56% deficient (<20 ng/mL) and 28% insufficient (20–30 ng/mL).

Lastly, we unfortunately cannot correlate the neonatal and maternal values. Our study was performed in a referral unit where all infants are outborn, so we did not have direct access to maternal records. In addition, the regional obstetric practices at the time did not include regular monitoring of maternal 25OHD levels. We do know from previous studies, however, that neonatal levels are closely associated with maternal levels at delivery,[8] so we would expect a close correlation in our patients as well. As such, we completely agree that, if increased 25OHD levels are indeed proven to be protective for HIE, maternal supplementation would be a possible avenue to ensure appropriate early levels after birth.

Publication History

Received: 08 August 2020

Accepted: 08 November 2020

Article published online:
28 January 2021

• References

• 1 Burris HH, Van Marter LJ, McElrath TF. et al. Vitamin D status among preterm and full-term infants at birth. Pediatr Res 2014; 75 (1-1): 75-80
• 2 Rao R, Lee KS, Zaniletti I. et al. Antimicrobial therapy utilization in neonates with hypoxic-ischemic encephalopathy (HIE): a report from the Children's Hospital Neonatal Database (CHND). J Perinatol 2020; 40 (01) 70-78
• 3 Arneson WL, Arneson DL. Current methods for routine clinical laboratory testing of vitamin D levels. Lab Med 2013; 44 (01) e38-e42
• 4 Chun RF, Peercy BE, Orwoll ES, Nielson CM, Adams JS, Hewison M. Vitamin D and DBP: the free hormone hypothesis revisited. J Steroid Biochem Mol Biol 2014; 144 (Pt A) 132-137
• 5 Arantes HP, Kulak CA, Fernandes CE. et al. Correlation between 25-hydroxyvitamin D levels and latitude in Brazilian postmenopausal women: from the Arzoxifene Generations Trial. Osteoporos Int 2013; 24 (10) 2707-2712
• 6 Lu M, Hollis BW, Carey VJ. et al. Determinants and measurement of neonatal vitamin D: overestimation of 25(OH)D in cord blood using CLIA assay technology. J Clin Endocrinol Metab 2020; 105 (04) e1085-e1092
• 7 Lowe DW, Hollis BW, Wagner CL. et al. Vitamin D insufficiency in neonatal hypoxic-ischemic encephalopathy. Pediatr Res 2017; 82 (01) 55-62
• 8 Marshall I, Mehta R, Petrova A. Vitamin D in the maternal-fetal-neonatal interface: clinical implications and requirements for supplementation. J Matern Fetal Neonatal Med 2013; 26 (07) 633-638