In this retrospective study the applicability of the Pottel method - obtaining a rescaled Scr marker by dividing Scr with the mean Scr value of the age and sex specific healthy population (Qcrea) - to detect nephrotoxicity following exposure to nephrotoxic agents was assessed in a cohort of 201 ELBW neonates. An existing cohort was updated, previously used to elucidate Scr patterns in ELBW neonates during the use of ibuprofen, amikacin or vancomyin [16, 17]. A statistically significant increase in Pottel scores during ibuprofen treatment was demonstrated starting from PNA day 4, possibly explained by the fact that this drug is rarely administered during the first days of life in the Leuven unit. Furthermore, there is a progressive increase of the Pottel score with consecutive days of ibuprofen exposure. These findings align with our earlier findings demonstrating substantial increases in Scr and concurrent absolute reductions in creatinine clearance during ibuprofen treatment, manifesting as shifts of approximately 1 SD in ibuprofen-exposed ELBW neonates [14, 16]. In contrast, a significant signal when evaluating the Pottel score during the use of amikacin and/or vancomycin was not identified. The quantification of an AKI threshold such as the previously reported value of 1.33 in older populations was not possible, as well as defining grades of AKI, as we rather wanted to illustrate the feasibility [15].
AKI is a common morbidity in neonates, strongly correlated with lower GA or BW, with reported incidences of 28–48% in extremely premature neonates (GA < 28 weeks) and up to 60% in ELBW neonates [4, 18–20]. Accurate detection of AKI remains challenging. Several definitions are available, such as the current predominantly used mKDIGO as well as pRIFLE (pediatric risk, injury, failure, loss, end-stage renal disease) and AKIN (Acute Kidney Injury Network), showing similar AKI incidences in ELBW neonates [20]. These definitions all rely on defined changes in urine output (UOP) or Scr assuming a steady-state situation and needing at least 2 observations. Their use is therefore limited in preterm neonates due to their time-dependent Scr physiology as well as challenges in accurately measuring UOP [11, 14, 21]. ELBW neonates exhibit a postnatal Scr rise from the day of birth (day 1) to approximately day 3 with a subsequent decline, more delayed with increasing immaturity [12]. This physiological rise over the first days often already is in line with the definition of AKI stage 1 (0.3 mg/dl changes) and the absence of an expected decline afterwards is not captured by current definitions [14].
Consequently, strategies to optimize the bedside use of Scr in ELBW neonates are indicated to improve AKI diagnosis and treatment. Of interest is the Pottel method 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 [15, 22]. Previously defined p50 values per PNA day in an ELBW cohort enabled us to calculate Pottel scores for every available Scr value, adjusting for the postnatal time-dependent Scr changes [14].
Medication stewardship during nephrotoxic drug exposure, frequent in ELBW neonates, was shown to significantly reduce the AKI rates in NICUs by improving medication adjustments, altering nephrotoxic medication pharmacologic monitoring or discontinuation where appropriate. [7, 8, 21]. Furthermore, in preterm neonates exposure was also associated with elevated markers of kidney dysfunction at a median age of 5 years, irrespective of the development of AKI according to the classical definitions, and thus suggesting subclinical kidney damage [23]. We hypothesized that the Pottel method could be useful for pharmacovigilance in ELBW neonates, exploring drug-related nephrotoxicity signals.
In ibuprofen-exposed neonates a clear signal was observed, with the difference in scores increasing with advancing age, suggesting a more robust nephrotoxic AKI signal during later neonatal life. However for amikacin and vancomycin, two of the more frequently used antibiotics in our unit, a significant signal in Pottel scores beyond PNA day 5 could not be shown, suggesting either a less pronounced nephrotoxic effect of these antibiotics within our study population or a suboptimal timing for capturing their effects [8]. Notably in our prior analysis there was already a more modest change in Scr dynamics during amikacin and vancomycin treatment compared to ibuprofen, with Scr increases typically remaining below the 0.3 mg/dL threshold for mKDIGO criteria [17]. Exploring the cumulative effect during treatment with ibuprofen 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 as early measurements might not fully capture the complete nephrotoxic effect.
Scr reflects impaired kidney function due to decreased filtration, where urinary and serum biomarkers reflect kidney damage due to an aberrant expression in response to kidney injury. Several biomarkers have been explored in the pediatric and adult literature, but data on the premature neonatal population are limited. In preterm neonates serum cystatin C demonstrates high diagnostic accuracy surpassing Scr but urinary levels yield conflicting results [24–27]. Neutrophil gelatinase-associated lipocalin (NGAL) shows promise in preterm neonates for both serum and urine concentrations but the specificity appears to be limited by strong associations with inflammatory parameters [24, 26, 28]. A scala of other potential biomarkers, such as urinary kidney injury molecule-1 (uKIM-1) have been investigated in limited studies in premature neonates [24, 26, 29, 30]. Recently various urinary biomarkers were found to be statistically significantly elevated in extremely premature neonates treated with or without nephrotoxic drugs, without a significant difference in AKI diagnosis showing their potential to detect subclinical kidney (tubular) damage [31]. However most urinary biomarkers also show variation with GA, in part determined by the degree of prematurity at birth, emphasizing the need to establish GA-and PNA-adjusted reference values [32, 33]. Despite the promising potential, also due to their non-invasive way of diagnosing and reducing iatrogenic anaemia, the current lack of comprehensive data in premature neonates indicates that they are not yet ready for integration into clinical practice.
Several limitations of our study warrant acknowledgment. The p50 values used for calculating the Pottel scores were derived from the same cohort as this study, influencing statistical outcomes. However our main intention was to introduce the concept rather than conducting a comprehensive analysis. A next obvious step would be to determine p50 values for Scr for a diversity of subpopulations commonly admitted in neonatal intensive care units, to validate the approach used. Our limited sample size could have impacted the statistical power in detecting antibiotic-related signals, possibly requiring a larger population to detect significant differences. The current study is single-centre and focuses solely on ELBW neonates, while still displaying considerable heterogeneity in terms of BW (500-1000g) and GA (23–34 weeks), resulting in substantial variability regarding disease severity and limiting possible generalization of the results. The relationship between Scr and GFR in ELBW neonates remains unclarified resulting in uncertainty to what degree the Pottel score accurately predicts kidney damage. We did not account for prolonged medication effects after cessation of treatment nor did we study the impact of piperacillin-tazobactam as well as other nephrotoxic drugs, often co-administered alongside amikacin in NEC. The impact on the Pottel score could be different based on the type and duration of drug exposure. For instance, the signal for ibuprofen becomes pronounced after 1–3 days of treatment where a milder nephrotoxic effect with antibiotics could occur after a longer duration of treatment. However, our main aim was to illustrate the utility of a bedside applicable tool to diagnose and quantify AKI.
In summary and taking these limitations into account, our findings underscore the potential value of the Pottel method in Scr-based bedside pharmacovigilance for ELBW neonates. In addition, it also provides a way to standardize AKI definitions in clinical trials, improving the comparability in this population. While other prospective biomarkers display promising results, their integration into clinical practice necessitates rigorous further investigation. 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. It’s important to note that the established upper limit of 1.33, validated as a dependable threshold in other populations, does not seem to be applicable in ELBW neonates. Our initial findings hint towards a considerably lower threshold. Comprehensive observational large-scale multicentre studies are necessary to validate and effectively develop this tool as well as ascertain solid p50 values and the appropriate cutoff values in this unique and vulnerable patient population.