From Fragmented Data to Integrated Drug Development in Asphyxiated Neonates Undergoing Therapeutic Hypothermia

Abstract

Methylxanthines are very commonly administered to preterm (gestational age [GA] < 37 weeks) neonates in whom they are associated with improved respiratory, renal, and neurodevelopmental outcomes. More recently, these drugs are also considered for several indications in full-term (GA 37-42 weeks) or near-term (GA 34/35-36 weeks) neonates. Suggested mechanisms of the observed beneficial effects relate to respiratory stimulation, as well as anti-inflammatory, diuretic, and renal protective effects, in part by interaction with the adenosine A3 receptor (ADORA3). Therefore, we read with great interest the recent paper in the journal on caffeine pharmacokinetics in asphyxiated neonates undergoing therapeutic hypothermia following moderate to severe encephalopathy.¹ It is our understanding that this paper is part of a planned drug development program to explore the potential add-on value of caffeine on neurological outcomes after perinatal asphyxia.^{2, 3}

This commentary likes to reflect on how to improve the workflow, the feasibility, and ways to reduce the uncertainties of drug development plans (e.g., sampling strategy and dose/exposure relationship). The ultimate goal of such a structured approach is to evolve toward personalized pharmacotherapy (e.g., precision dosing) in this specific population. We already earlier suggested that a subpopulation-specific physiologically based pharmacokinetic (PBPK) model is probably such a strategy.⁴

PBPK modeling and simulation imply that we collect data on both drug-specific characteristics (what is known about the drug?), as well as on population-relevant physiology characteristics (what is known about the targeted population relevant to the planned PK study?) (Figure 1). Preclinical findings (animal research, in vitro data) can further inform such efforts. This approach might also have informed the caffeine research team on how to conduct their trial, for example, sampling strategy or targeted exposure.⁵

Related to drug-specific aspects, this covers, for example, physicochemical characteristics of the drug or reported absorption, distribution, metabolism, and excretion (ADME) properties in other populations, including the formation of metabolites. Related to the targeted population (in this case, asphyxiated neonates undergoing therapeutic hypothermia), this includes aspects like weight or gestational age range, enzyme ontogeny, cardiac output and organ-specific blood flow, glomerular filtration rate (GFR), or specific (patho)physiological characteristics.⁴ Because of the notorious time-dependent physiology of this subpopulation, such data are at best longitudinal over postnatal age, and weighted to similar data in non-asphyxiated cohorts, as very recently reported for albumin values in this journal.⁶

The elimination half-life of caffeine is reported to be about 100 h in the term newborn. This means that the sampling strategy applied (up to 24 h) a priori has its limitations in generating robust clearance estimates, while, for example, modifications of the loading dose practices from intravenous bolus injection to prolonged infusion might have been a useful option to avoid high peak concentrations. Furthermore, since therapeutic hypothermia is provided for 72 h as standard practice and if exposure during therapeutic hypothermia is the therapeutic targeted window, loading dose practices are far more relevant compared to maintenance doses. The importance of a loading dose relates to achieving early targeted exposure, in a setting of increased volume of distribution for hydrophilic drugs. Consequently, clinical trials should focus on this aspect by collecting data to facilitate robust estimations of distribution in this scenario. PBPK-based simulations prior to conducting clinical trials could help optimize the dosing regimen and blood sampling schedule.

Asphyxia and therapeutic hypothermia affect the (patho)physiology of (near)term neonates, impacting both drug distribution and elimination. Besides the earlier mentioned albumin patterns, other aspects affecting distribution may relate to, for example, weight characteristics, since these can be different from “average” in specific pathologies, like asphyxia. Compared to a reference cohort, the median and interquartile birth weights (3.35, 2.93-3.74 kg) are very similar in neonates undergoing therapeutic hypothermia. However, the portion of neonates <2.5 kg or >4 kg was 8.8% (instead of 6.3%) and 15.2% (instead of 8.3%), respectively. When calculated based on being small for gestational age (birth weight <10th centile) or large for gestational age (birth weight >90th centile), these were 18.4% and 14.8%.⁷ Consequently, this suggests that there are relatively more “weight outliers” in asphyxia cohorts, which is relevant for study design and simulations.

Related to clearance, caffeine is exclusively eliminated by GFR in neonates, which is quite different from the metabolic clearance in children and adults. It is known that GFR is also affected by asphyxia and therapeutic hypothermia, while the exact pattern of postnatal GFR changes in neonates undergoing therapeutic hypothermia is still poorly described.⁷ Krzyzanski et al and Deferm et al reported on a population model of time-dependent changes in serum creatinine in (near)term neonates with hypoxic-ischemic encephalopathy during and after therapeutic hypothermia and on mannitol pharmacokinetic data, one of the markers used to measure GFR (mGFR).^{8, 9}

The current caffeine data therefore add to the list of drug-specific reports with fragmented data sets on the pharmacokinetics of antibiotics (gentamicin, amikacin, amoxicillin, ampicillin, benzylpenicillin, and vancomycin) during or shortly following therapeutic hypothermia (commonly 72 h, to be initiated <6 h of postnatal life) that consistently provide evidence for a reduced GFR (20-60% reduction).^10-13 We still qualify these data as fragmented, since—except for amikacin—data were in “isolation,” only describing pharmacokinetics in this specific subpopulation, not compared to findings in a non-asphyxiated reference cohort.¹²

In line with the impact of GFR on caffeine clearance, the authors mentioned that their PK estimates were similar to estimates reported in neonates immediately following cardiac bypass surgery, a setting also well known to be associated with acute kidney injury.¹⁴ Identification of high-risk AKI cases could therefore become an integrated part of a future drug development plan for this specific drug in this population. Alternatively, we highly recommend, at least, collecting data on kidney function in future studies on this specific drug.

In conclusion, we welcome the drug-specific observations on caffeine pharmacokinetics in neonates undergoing therapeutic hypothermia. This is even more relevant, since the current outcome of neonates after therapeutic hypothermia remains suboptimal. Still, about 45% (instead of 55%, the number needed to treat 7-9) display poor outcomes (mortality, major morbidity), indicating that therapeutic interventions added to therapeutic hypothermia are urgently needed.⁴ However, an integrated drug development, informed by the specific characteristics of this population, by pre-clinical observations (animals, in vitro data) is very likely warranted to make this and other development programs of drugs considered as add-on therapy to therapeutic hypothermia more effective. This includes applying PBPK modeling approaches or model-informed drug development tools in general. The patients, their families, and the clinical communities involved in this intensive care deserve such an integrated approach.¹⁵

Karel Allegaert is an editorial board member of the Journal of Clinical Pharmacology. He was not included in the editorial handling of this paper.

This research was funded by a Senior Research Grant from the Research Scientific Foundation-Flanders (FWO)—G0D0520N, I-PREDICT: Innovative Physiology-based pharmacokinetic model to pREdict Drug exposure In neonates undergoing Cooling Therapy, and by a KU CELSA research project (Central Europe Leuven Strategic Alliance, CELSA/24/022). The research activities of A. Smits are supported by a Senior Clinical Investigatorship of the Research Foundation—Flanders (FWO) (18E2H24N).