In this retrospective study, the applicability of the Pottel method—a rescaled Scr marker by dividing Scr with the mean Scr value of the age- and sex-specific healthy population (Qcrea)—to detect drug-induced nephrotoxicity in ELBW neonates was assessed.

AKI is common in ELBW neonates, with incidences up to 60% [3, 14,15,16]. ELBW neonates already have a decreased kidney reserve due to their disrupted kidney development, with evidence showing smaller kidney size and abnormal kidney-related outcomes in childhood compared to healthy controls [1, 2, 17]. Experiencing neonatal AKI likely further imposes an increased risk of long-term kidney dysfunction [17, 18]. These patients are often exposed to nephrotoxic drugs, which co-occurs with several known risk factors such as hemodynamic instability, LOS, NEC, or PDA [3,4,5,6]. Consequently, timely detection of AKI during the use of nephrotoxic drugs could reduce AKI rates in ELBW neonates and improve long-term kidney outcomes. However, accurate detection of AKI remains challenging. Several definitions are available, such as the predominantly used mKDIGO, the pRIFLE (pediatric Risk, Injury, Failure, Loss, End-stage kidney disease), or AKIN (Acute Kidney Injury Network), all showing similar AKI incidences in ELBW neonates [16]. These definitions are all based on defined changes in urine output (UOP) or Scr, assuming a steady-state situation and needing at least two observations. ELBW neonates, however, exhibit a physiologic postnatal Scr rise from the day of birth (day 1) to approximately day 3 with a subsequent decline, delayed more with increasing immaturity [8]. The Scr rise over the first days of life noted in ELBW in many instances crosses the thresholds to diagnose stage 1 AKI by the modified, neonatal KDIGO definition (Figure S1). However, in these neonates, this may be consistent with a normal, physiologic phenomenon rather than a pathophysiologic state. The use of current definitions is therefore limited in preterm neonates due to their time-dependent Scr physiology as well as challenges in accurately measuring UOP [19].

This scenario serves as an example of the utility of a rescaled Scr marker, such as the Pottel method, which may be better for detection of truly abnormal Scr than the present KDIGO criteria in circumstances such as these. Pottel et al. analyzed the use of their rescaled Scr marker with a multicohort study in a population > 2 years of age demonstrating a distribution of Scr/Qcrea around 1, with the 2.5th and 97.5th percentile at 0.67 and 1.33, respectively, showing specificity and sensitivity for impaired kidney function close to 90% over all age groups when using 1.33 as the cut-off value [10, 20].

We hypothesized that the Pottel method could be useful to explore (drug-related) nephrotoxicity in ELBW cases. An existing cohort, previously used to describe Scr patterns in ELBW neonates during ibuprofen, amikacin, or vancomycin exposure, was used as we knew this cohort showed an association between drug exposure and kidney impairment [12, 13].

In ibuprofen-exposed neonates, a clear association was observed with a statistically significant increase in Pottel scores during ibuprofen treatment starting from PNA day 4. There was a progressive difference in scores, increasing with advancing age, suggesting a more robust nephrotoxic AKI effect, confirming previous analyses [11, 12]. Exploring the cumulative effect during ibuprofen exposure revealed a rising Pottel score until day 3 of exposure followed by a relatively stable value thereafter. This emphasizes the importance of well-timed assessments during drug treatment before drawing definitive conclusions about potential kidney damage. Early measurements might not fully capture the full nephrotoxic effect.

In contrast, we could not show a significant association beyond PNA day 5 when evaluating the Pottel score during the use of amikacin and/or vancomycin. This seems to reflect a less pronounced nephrotoxic effect of these antibiotics in our study population. In our prior analysis, there was also a more modest change in Scr dynamics during amikacin and vancomycin treatment compared to ibuprofen [13]. Unexpectedly, we even noticed a decreased Pottel score in the latter half of the first month of life. As this study was explorative, we describe this finding, but fail to explain it. Among other causes, this could be due to less robust Q reference values in the third and fourth week in the currently available dataset as one of the limitations of the current analysis.

There are obvious limitations. Likely most relevant, the p50 values used for calculating the Pottel scores were derived from the same cohort in a previous analysis, influencing statistical outcomes [11]. However, our main intention was to introduce the concept. We do not claim that the p50 Scr values applied (Supplementary Table 1, column 2) to convert absolute Scr values to Pottel scores are ready reference values as they were derived from a single-center, small dataset (Table 1) [11]. Our primary goal was to illustrate the potential strength of this approach, as Scr displays extensive time-dependent variability in the first weeks of postnatal age, not yet at steady state. We do not claim that this approach is already a robust substitute for the currently used AKI definitions. A next obvious step would be to determine p50 values for Scr for a diversity of subpopulations commonly admitted to NICUs and to validate the approach used. Preferably, these datasets are generated from different units to reflect diversity in practices and settings.

We did not account for prolonged medication effects after cessation of treatment, nor did we study the impact of other drugs as we did not have sufficient granularity in our current dataset. The AWAKEN study showed that diuretics, vasopressors, and methylxanthines were associated with a lower risk for early onset AKI (postnatal days 2–7), whereas some medications were associated with increased odds for late AKI (> 7 days after birth) in neonates [4, 21]. The impact on the Pottel score could be different based on the type and duration of drug exposure. We neither accounted for other known variables in ELBW neonates that may have impacted kidney health, e.g., prenatal factors such as maternal hypertension or certain drug exposure or postnatal factors such as hypotension or elevated mean airway pressure, as we did not have sufficient data available in our retrospective dataset [22].

Considering these limitations, our findings underscore the potential value of the Pottel method in Scr-based bedside pharmacovigilance for ELBW neonates as an alternative technique. This approach enables bedside assessment based on a single Scr value and captures the time-dependent pattern. In addition, it could provide a way to standardize AKI definitions in clinical trials, improving comparability in this population. It is important to note that the established upper limit of 1.33 (for chronic kidney disease and in children > 2 years) does not seem to be applicable in ELBW neonates. Our initial findings hint towards a considerably lower threshold. Scr remains a widely recognized and measured biomarker, and the application of the Pottel method moderates certain inherent limitations linked to its use in this population. Additional steps, with comprehensive observational large-scale multicentre studies, are needed to further explore and develop this approach, to ascertain solid p50 values, to make the current findings more generalizable, and to compare this tool to the currently used AKI definitions.