This study showed that the median onset of NEC was 15 days, and the median AKI onset in infants with NEC was two days prior to the onset of NEC (interquartile range -5.75 to 0 days) i.e. 13 days (IQR 9 to 15days). Vasopressor use was significantly higher in infants with NEC associated with AKI 82.1% vs. 54.2% (p = 0.009; Table 1). Moderate to severe AKI more prevalent than the milder form in infants with NEC and have an increased mortality rate.

Previous studies [11,12,13,14] have shown the association of AKI that develops after NEC onset. Various pathophysiologic explanations were given for this association, such as a significant inflammatory cascade caused by NEC that can lead to microcirculatory disturbance, sepsis causing hypotension or direct tubular damage, nephrotoxic medications used after NEC diagnosis [18, 19]. Accordingly, it is presumed that AKI develops after NEC onset, and previous studies have focused on AKI onset on or after NEC onset. In contrast to these studies, this study shows that the onset of AKI precedes the week before the onset of NEC. This critical finding questions our understanding of the pathophysiology involved in the association of NEC with AKI.

This study also showed that a significant proportion of infants with NEC who have AKI required vasopressors indicating that these infants were hemodynamically unstable. This finding invokes an exciting hypothesis that impaired perfusion due to hemodynamic instability could be a common factor in the development of AKI and NEC. Studies have shown that intestinal microcirculatory dysfunction plays a significant role in the pathogenesis of NEC [20]. Animal models show similar pathophysiology in both AKI and NEC. The cecal ligation and puncture AKI animal model has many features similar to NEC. Studies also suggest a causal relationship in which AKI appears to drive other organ dysfunction and vice-versa, referred to as “crosstalk.” [21,22,23]. Thus, systemic inflammation associated with AKI could affect the intestines and cause NEC.

AKI incidence in infants with NEC in this study was found to be 70% which is higher than the reported incidence of 21–54% in the previous studies. This could be due to the increased recognition of AKI that was prevalent the week before the onset of NEC and by evaluating both SCr and UOP for AKI diagnosis. The other possibility for higher incidence could be from previous episodes of AKI, which predisposes to recurrent AKI. In a retrospective report from the AWAKEN database, in infants with late-onset AKI ( > 7 days of age), 28% had an earlier episode of AKI [24]. We explored this option and found that in our cohort, we had 18% with previous AKI, which is less likely to explain the higher incidence in this cohort.

This study also showed that most infants with NEC and AKI progressed to moderate-severe AKI. The incidence of severe AKI (stages 2 and 3) in this cohort was 84%, and 76% had surgical NEC. This is a significant finding compared to the study by Garg et al. [14], who reported 32.6% incidence of severe AKI; 58.7% after surgical NEC, despite a similar incidence rate of surgical NEC in both studies (52% vs. 51.5%) respectively. Again, this is possibly due to increased recognition of AKI and a quicker progression to severe stages before NEC onset.

Most previous studies have utilized SCr only for AKI diagnosis, except for Garg et al., who monitored SCr and UOP per KDIGO criteria. In our study, 10.7% of infants with AKI were diagnosed by UOP criteria alone. Moreover, among infants diagnosed with AKI based on both SCr and UOP, 48% of infants had higher staging with UOP criteria. The progression of AKI severity is also quicker when UOP criteria are followed for AKI diagnosis, thus stressing the importance of including UOP measurements also in AKI diagnosis along with SCr.

This study is in concordance with the other studies [11,12,13,14], which showed an increase in mortality rate among infants with NEC associated with AKI. We found a two-fold increase in mortality rate in infants with NEC with AKI compared to those without AKI, confirming the findings reported by Criss et al. [11]. The mortality rate was even higher, up to 75.8% in infants with surgical NEC with AKI. We did not find any statistically significant difference in the LOS among the survivors, between the two groups.

In a recent study, Han et al. investigated the potential use of changes in SCr, reflecting the development of AKI as a surrogate biomarker for the impending development of NEC [15]. They evaluated AKI in preterm infants born between 23- and 32 weeks GA who did and did not develop NEC and the incidence of AKI in infants one week before the diagnosis of medical and surgical NEC. AKI was screened using SCr measurements only from the DOL 8 onwards. They reported that one of the 13 with AKI had AKI within 7 days before NEC diagnosis (7.7%). Our study is similar to the above in evaluating AKI onset one week before NEC diagnosis but differs in several other ways. Our cohort consists of infants with NEC, and a comparison was made between those with or without AKI. Both SCr and UOP criteria were used for AKI diagnosis, and we also included SCr measurements from the first week of life. Out of 56 infants with AKI, 47 (84%) had AKI within 7 days before NEC onset. We speculate that this significant difference in results from Han et al. is due to the difference in AKI diagnostic criteria and patient population.

The mean GA in this cohort was 27 weeks, with a mean BW of ~840 g, and the median age at NEC onset was 15 days. According to a study by Yee et al., preterm infants with BW > 1000 grams present with NEC at a mean of 7 days (early onset) versus a mean of 32 days (late onset) in infants with BW < 1000 g [25]. In comparison to this study, NEC onset seems to be earlier in our study. It is unclear whether this earlier onset of NEC affects the association with AKI.

We found similar results in the two centers where we conducted this study, such as median NEC and AKI onset, prevalence, and severity of AKI, which emphasizes the credibility of this finding.

Since this is a retrospective study, cause-and-effect relationship claims cannot be made. The small sample size makes it difficult to generalize the study results. SCr is a marker of renal function and not renal injury, so elevated SCr merely reflects a decline in renal function. SCr varies with gestational and postnatal age. In preterm infants, the decline in SCr after birth is slower. It reaches nadir values in the first 1–2 months, so identifying a baseline SCr value is challenging in the first month of life. Accurate UOP measurements were not always possible as these measurements are obtained by weighing diapers. However, this was true across both sites and study cohorts. Despite this, we used the KDIGO definition because it is the most widely accepted definition used by previous studies on NEC with AKI. It also allowed us to compare our study results.

This study is the first to evaluate the prevalence and timing of AKI one week before and after NEC onset, using SCr and UOP measurements. Regardless of the possible mechanisms involved in the association of NEC and AKI, as discussed above, the detection of AKI seems to precede NEC. In our cohort, the median age of NEC onset was 15 days, and infants had frequent electrolyte measurements, including SCr, in this period after birth and so we had adequate SCr measurements to trend and screen for AKI. This vital knowledge could help us with the early identification and management of NEC and prevent complications. Thus, close monitoring of SCr and UOP measurements is imperative in premature infants.