Serum Creatinine: Beyond Renal Function as a Biomarker of Adverse Outcomes in Premature Neonates

Acute kidney injury (AKI) in premature infants is a common life-threatening complication with approximately 10% to 63% of neonates affected.¹ AKI has also been found to be an independent risk factor for in-hospital mortality. Also, AKI and prematurity have additional adverse kidney sequelae such as chronic kidney disease.^2,3 Therefore, recognizing the early stages of AKI in infants in order to prevent its progression and clinical sequelae is key. Unfortunately, the timely identification of AKI in premature infants remains a big challenge and an important topic of ongoing investigation.

The criteria for the diagnosis of AKI established by The Kidney Disease: Improving Global Outcomes (KDIGO) for adults and adapted for neonates, have been widely adopted by the research community.⁴ However, it has become clear that modifying definitions of AKI for adults and applying them to premature neonates has limitations. Therefore, the nephrology and neonatology communities have struggled with the best approaches to recognize neonates undergoing early AKI in this unique patient population. Neonates born at 22 weeks are much earlier in the process of nephrogenesis with much smaller nephron endowment compared to neonates born premature at 36 weeks. In addition, the 22 weeks premature neonate has a much smaller muscle mass and higher total body water compared to the 36 weeks premature neonate.⁵ These are only a few of the many physiologic reasons why serum creatinine (SCr) measurements should be interpreted with great caution in premature infants.

In the current issue of Pediatric Research, Chen et al.⁶ highlights the limitations in using the neonatal modified KDIGO criteria to recognize premature infants undergoing early kidney function impairment and they further build a case that the limitations in neonatal modified KDIGO criteria need to be addressed in order to recognize more premature neonates undergoing kidney injury and optimize their clinical care to improve their in-hospital outcomes and long-term kidney function. They⁶ highlight a group of premature patients that do not meet neonatal modified KDIGO SCr criteria for AKI but still have elevated SCr >95^th percentile compared to their gestational age peers. This elevated SCr group has worse clinical outcomes (i.e., increased in-hospital mortality and decreased growth) than their gestational age equivalent peers with SCr levels <95th percentile. The >95th percentile SCr group also had a higher proportion of small for gestational age infants.

These findings are both clinically relevant and provocative. If SCr levels in premature infants function not as a specific biomarker of AKI, but rather as a broader indicator of adverse outcomes, then their use as the cornerstone for defining AKI in this population warrants serious reconsideration. Such an interpretation (using outcomes to define AKI like mortality or others) would challenge the validity of the current KDIGO-based diagnostic criteria, which rely heavily on SCr changes to identify AKI, and raises concern that existing definitions may misclassify or fail to capture the unique pathophysiology of kidney dysfunction in premature neonates. In turn, these findings may even have clinical implications because delaying the diagnosis of AKI in infants based on a given definition may limit opportunities for timely interventions and can also confound the design of interventional trials aimed at preventing or treating AKI. Both factors, not delaying the diagnosis of AKI, and developing a more homogenous and early AKI definition for clinical trials, were two of the main reasons used to replace the term acute renal failure for AKI 25 years ago. Notably, this decision was made without selecting a gold standard to define kidney injury and was based on the assumption that the SCr levels and/or urine output (both classic markers of renal function) could be used as proxies of kidney injury until new and more sensitive biomarkers of AKI were developed.

Previous studies also showed that SCr levels have limitations to diagnose early AKI in preterm infants <30 weeks’ gestational age.¹ SCr levels fluctuate in response to changes in renal maturation, and extracellular fluid volume, including dehydration, fluid overload, insensible losses, and poor tissue perfusion. Furthermore, at birth, SCr levels reflect the maternal concentrations, while immature tubular handling further limits its reliability as a measure of renal clearance. In addition, several commonly used medications (e.g., diuretics, aminoglycosides such as gentamicin) can influence SCr levels independently of kidney injury. Urine output could be unreliable as well, as many premature infants with AKI maintain relatively high urine output despite significant renal dysfunction. Additionally, diuretics, fluid management plans, and insensible losses affect the urine output.

The findings of Chen et al.^6–8 add to the growing body of literature that the neonatal modified KDIGO SCr criteria may fail to recognize newborns undergoing significant kidney injury.^7–8 Other promising approaches to identify neonates with impaired kidney function and/or injury associated with worse clinical outcomes include: i.) a delayed rate of SCr decline in the first week of life; ii.) renal near infrared spectroscopy (NIRS); and iii.) urine biomarkers. First, a delayed rate of serum creatinine decline in the first week of life has been associated with worse clinical outcomes in premature neonates and neonates with hypoxic ischemic encephalopathy.^7,8 Second, NIRS have shown promise in identifying renal impairment associated with worse clinical outcomes in premature neonates and neonates with hypoxic ischemic encephalopathy.⁹ Third, urine biomarkers in unique subpopulations of infants show promise in the literature.¹⁰

There are limitations in the study by Chen et al.⁶ that should be highlighted. First, temporal shifts in clinical practice may have occurred during the long study period (2007–2023). Second, two different laboratory methods to measure SCr probably introduced a certain degree of variability. Third, the study included a small single Taiwanese cohort, and a relatively small number of postnatal SCr values were available in many neonates, thus it is hard to assess the trend of the SCr changes in these patients and the resulting SCr reference distribution values may have limited generalizability. Overall, these constraints underscore the need to perform prospective, multicenter studies involving a larger number of infants and employing standardized SCr measurement protocols to establish more robust and widely applicable reference values with long term follow up. More studies are needed to define normative values of SCr in premature neonates at birth and their trajectory during the first months of life in large multicenter cohorts. Additional research is needed on the utility of renal NIRs and urine biomarkers as non-invasive tools to identify clinically significant renal impairment in neonates.

In conclusion, the results of Chen et al.⁶ highlight the urgent need to refine AKI definitions and identify more specific and reliable biomarkers of AKI in premature infants. We advocate for the incorporation of gestational age into refinement of AKI definitions for premature neonates. Also, we feel that there is promise in complementing SCr and urine output criteria with renal NIRS and urine biomarkers to identify those infants at higher risk of developing AKI.¹⁰

Figure 1.

Preterm infants (< 30 weeks of GA) showing SCr values above the 95^th percentile have higher mortality and adverse clinical outcomes compared to similar preterm infants showing normal SCr values (Abbreviations: SCr: serum Creatinine; KDIGO: The Kidney Disease: Improving Global Outcomes; SGA: small for gestational age).