ABSTRACT
Aims
To investigate the association between serum ALP levels and the development of AKI in paediatric patients with sepsis.
Methods
Retrospective screening of a computerized database of 4 Israeli regional hospitals for children (0–18 years) hospitalized between 1.01.2000–1.06.2024 with ICD‐9 sepsis‐related diagnoses (study group) and age‐ and sex‐matched non‐septic febrile patients (controls). Serum ALP levels were sex‐ and age‐adjusted and described as standard deviation scores (SDS).
Results
Of 809 952 screened hospitalizations, 701 records comprised the study group. Their median (IQR) age was 17 (10–38) months, 53% were males. Nadir ALP‐SDS was lower in the sepsis group compared to controls (−1.65 ± 1.11 vs. −0.97 ± 1.36, p < 0.001) and further lower in those with AKI, which occurred in 23% of the sepsis cohort (−1.95 ± 1.16 vs. −1.56 ± 1.09, p < 0.001). Hospital stay was longer in the AKI compared to non‐AKI group [median (IQR): 7 (4, 12) vs. 4 (3, 5) days, p < 0.001]. Lower ALP‐SDS was independently associated with AKI [OR (95% CI): 0.7 (0.6–0.8), p < 0.001], after controlling for CRP and leucocyte count.
Conclusions
ALP was lower in sepsis patients compared to controls, and significantly lower in AKI vs. non‐AKI group, suggesting it as an additional marker of infection severity.
Keywords: acute kidney injury, alkaline phosphatase, sepsis
Key Notes
Alkaline phosphatase (ALP) plays several roles, including bacterial endotoxin detoxification
In this retrospective multicenter study we could identify that paediatric patients with sepsis had lower sex and age adjusted ALP values compared to febrile non‐septic matched controls, and significantly lower values in AKI versus non‐AKI group.
These findings suggest that ALP may serve as an additional marker of infection severity and a possible contributor to AKI pathogenesis.
Abbreviations
- AKI
acute kidney injury
- ALP
alkaline phosphatase
- CRP
C‐reactive protein
- SDS
standard deviation score
1. Introduction
Alkaline phosphatase (ALP) is a membrane‐bound glycoprotein that catalyses the hydrolysis of phosphate monoesters at basic pH values. It is encoded by distinct genes as several tissue‐specific isoenzymes exist. Tissue nonspecific ALP (TNSAP) is expressed in the liver, kidney and bones and is encoded by the TNSAP gene on chromosome 1 [1]. The TNSAP bone isoform is expressed in osteoclasts and plays a crucial role in bone mineralization [2]. Total serum ALP activity changes with age. Since ALP is a marker for osteoblastic activity, growing children have higher levels than adults, with the highest levels in infancy and puberty [3]. Elevated ALP levels are seen in states of increased bone turnover (intensive growth, rickets, malignancy) and liver disease. Decreased ALP levels are seen in malnutrition, including zinc and magnesium deficiency (as these anions are crucial for enzyme activity), celiac and Wilson disease and the genetic disease hypophosphatasia [4].
Sepsis and septic shock are potentially life‐threatening conditions of infective origin, which may lead to multiorgan dysfunction with high morbidity and mortality [5]. Sepsis is defined as systemic inflammatory response syndrome (SIRS) with organ dysfunction. Diagnosis of septic shock is accompanied by hemodynamic instability [6]. Acute kidney injury (AKI) is a frequent and severe complication of paediatric sepsis, occurring in approximately 20% to 50% of critically ill children with sepsis. The development of AKI in this population is associated with increased mortality, longer hospital stays and a higher risk of long‐term kidney dysfunction [7, 8].
AKI in children is currently diagnosed based on the dynamics of increase in serum creatinine levels and decrease in urine output. These two clinical parameters form the basis of widely used criteria, such as the KDIGO classification, which defines AKI and stratifies its severity into three stages [9]. While these measures are considered the standard in both research and clinical settings, ongoing efforts aim to identify additional biomarkers that could improve risk assessment and early detection of AKI in paediatric sepsis.
ALP also plays a role in systemic infections via bacterial endotoxin detoxification through dephosphorylation of bacterial lipopolysaccharides (LPS) [10]. Dephosphorylation of LPS represents a crucial detoxification reaction, which may provide a new strategy to treat LPS‐induced diseases like sepsis [11]. These effects of ALP are being intensively studied in the context of sepsis‐associated acute kidney injury (AKI) [12]. Local upregulation of ALP activity following stimulation with LPS and increased ALP mRNA were shown in kidney glomeruli [13]. Kidney ALP depletion during ischemia was associated with the development of AKI in an animal model [14]. Apart from local effects in the kidney, ALP may attenuate the innate immune response, as dephosphorylation of endotoxin abolishes its biological activity and induces tolerance to subsequent endotoxin exposure [15]. In animal models of sepsis, ALP administration attenuated the inflammatory response and reduced mortality [16]. ALP blood levels were evaluated in an adult cohort of patients with sepsis and septic shock. Recent retrospective work showed a modest increase in ALP activity during sepsis‐associated AKI, mainly of the liver and bone isoforms, with a weak positive correlation between the 3‐day follow‐up ALP activity and hospitalization length. However, neither baseline nor follow‐up ALP activity was associated with survival [17]. Moreover, treatment with exogenous ALP improved renal function in patients with severe sepsis and septic shock with AKI [12].
Contrary to adults, serum ALP activity in paediatric patients with sepsis has not yet been investigated. Therefore, the aim of this study was to investigate the association between serum ALP levels and the development of AKI in paediatric patients with sepsis.
2. Methods
The study protocol has been submitted to the Human experimentation committee and approved (reference number: BRZ‐24‐0052, from date 10/2024). We performed a retrospective multicenter study using the “Kineret” database, which aggregates unified data from four Israeli government regional hospitals (Barzilai, Padeh‐Poriah, Hillel Yaffe, Bnai Zion) and its usefulness for research has been shown [18].
Inclusion criteria: Study group was comprised of paediatric patients (age 0–18 years) hospitalized in these hospitals between January 2000 to June 2024, who had different ICD‐9 diagnoses related to sepsis in their electronic medical records, including: 38.0–38.9 (septicemia), 785.52 (septic shock), 995.91–995.94 (SIRS), 998, 998.02 (postprocedural sepsis/shock), 790.7 (bacteremia), 40.82 (toxic shock syndrome). Age‐ and sex‐matched patients hospitalized in one of those regional hospitals (H1) during the same study period with an ICD‐9 diagnosis of “fever” but without sepsis‐related diagnoses comprised the control group. Matching was performed at the group level (frequency matching) to achieve comparable age and sex distributions, rather than at the individual 1:1 pair level. Because eligible febrile controls from H1 did not always yield an exact numerical match, the final control group (n = 673) is slightly smaller than the sepsis cohort (n = 701). Independent‐samples statistical tests (t‐test, Mann–Whitney U, chi‐square) were therefore used for all sepsis vs. control comparisons. For multiple group comparisons a two‐step statistical approach was used: Kruskal–Wallis first to test for an overall difference, followed by pairwise Mann–Whitney U tests with Bonferroni correction to identify which specific group pairs differed significantly.
Exclusion criteria: Patients without available serum ALP and creatinine measurements during hospitalization or those with a diagnosis of chronic kidney disease (CKD) on admission or discharge were excluded. To ensure data quality and avoid skewing by extreme values, patients with absolute ALP SDS values > 3 were excluded to avoid inclusion of hepatic injury or transient hyperphosphatasia cases.
Data Collection and Definitions: Demographic, clinical, laboratory and microbiological data were retrieved from electronic medical records recorded into the Kineret electronic system. AKI was defined according to KDIGO criteria, confined only to the dynamics in serum creatinine, as follows: diagnosis was established if there was an increase in serum creatinine of > 0.3 mg/dL between the minimum and maximum values recorded during hospitalization, or if a serum creatinine level exceeded the normal age‐specific threshold. This threshold was defined based on the estimated glomerular filtration rate (eGFR) using the Schwartz formula and the 50th percentile for height, categorized as > 0.4 mg/dL for ages 0–6 months, > 0.5 mg/dL for 6 months–4 years, > 0.6 mg/dL for 4–8 years, > 0.7 mg/dL for 8–12 years and > 0.9 mg/dL or > 1.1 mg/dL for males and females aged 12 years and older, respectively. AKI was analysed both as a binary outcome and by severity stage, where Stage 1 was defined as an increase in serum creatinine to 1.5–1.9 times baseline or an increase of > 0.3 mg/dL within 48 h; Stage 2 as an increase to 2.0–2.9 times the baseline; and Stage 3 as an increase to ≥ 3 times the baseline or a serum creatinine level of > 4 mg/dL [19].
ALP Normalization: Serum ALP levels were adjusted to sex‐ and age‐specific normal values and described, for unification between age groups and sexes, as standard deviation scores (SDS), z‐scores, as previously described [20].
2.1. Statistical Analysis
Continuous variables were summarized using medians with interquartile ranges (IQR) or means ± standard deviations, while categorical variables were summarized using percentages. Two‐tailed t‐tests and Mann–Whitney U tests were used for the comparison of continuous parameters, and the Chi‐square test was used for categorical variables. Comparisons were made between Sepsis vs. Non‐Sepsis control group and between Sepsis patients with AKI versus Sepsis patients without AKI.
To identify independent risk factors for AKI, a multivariate logistic regression analysis was performed. The model included variables that showed statistical significance in the univariate analysis (such as CRP, leukocyte count and phosphate levels). Age and sex were not included as covariates in the regression model, as they are inherently accounted for within the ALP SDS calculation (which is adjusted for age and sex norms), thereby avoiding multicollinearity.
3. Results
During the study period there were 809 952 recorded hospitalizations of paediatric patients in the included hospitals. Among them, 1504 had ICD‐9 diagnoses related to sepsis, and in 715 of them concomitant data related to their ALP levels during that hospitalization were available, 14 of them had ALP SDS > 3 and were excluded. The final study group comprised of 701 patients (Figure 1). The study cohort included patients from all four participating hospitals: 299 patients (43%) from Hospital H1, 237 patients (34%) from Hospital H2, 158 patients (22%) from Hospital H3 and 7 patients (1%) from Hospital H4. The study group was compared to a matched control group of febrile patients without sepsis‐related diagnoses (n = 673). Demographic, clinical and laboratory characteristics of both groups are presented in Table 1. The matching process was successful, as evidenced by no significant differences in age (median (IQR) 17 (10, 38) vs. 19 (7, 40) months, p = 0.3) or sex distribution (males: 53% vs. 52%, p = 0.7) between the groups. As expected, mean maximal CRP levels were markedly higher in the sepsis group compared to controls (113 ± 100 vs. 9 ± 5.4 mg/L, p < 0.001), as were mean maximal WBC counts (20 ± 10 vs. 12 ± 6 × 109/L, p < 0.001). While leukocytosis (WBC > 5 × 109/L) was significantly more common in sepsis, leukopenia (WBC < 5 × 109/L) was more frequent in the control group (12.7% vs. 6%, p < 0.001). Biochemical markers also reflected the systemic burden of sepsis: patients with sepsis had significantly lower mean minimal albumin levels (3.8 ± 0.6 vs. 4.2 ± 0.4 g/dL, p < 0.001) and lower mean minimal serum phosphorus levels (4.1 ± 1.2 vs. 4.9 ± 1 mg/dL, p < 0.001) compared to the non‐sepsis group. The distribution of minimal ALP values in the control, sepsis‐AKI and sepsis non‐AKI groups with overall comparison across all three groups (H = 105.4, p < 0.001) and pairwise post hoc comparisons between the control group and sepsis non‐AKI and between the control group and sepsis AKI (both ***p < 0.001) is depicted in Figure 2.
FIGURE 1.
Population flow chart.
TABLE 1.
Comparison of the demographic and laboratory characteristics of the study (sepsis) versus control (febrile non‐septic) groups.
| Parameter | Sepsis N = 701 | Non‐sepsis N = 673 | p |
|---|---|---|---|
| Age (months), median (IQR) | 17 (10, 38) | 19 (7, 40) | 0.3 |
| Males, n (%) | 375 (53%) | 352 (52%) | 0.7 |
| Bacteremia, n (%) | 122 (17%) | 17 (2.5%) | < 0.001 |
| Bacteriuria, n (%) | 37 (5.2%) | 36 (5.3%) | 1 |
| Max‐CRP (mg/L), mean ± SD | 113 ± 100 | 9 ± 5.4 | < 0.001 |
| Max‐WBC (×109/L), mean ± SD | 20 ± 10 | 12 ± 6 | < 0.001 |
| WBC < 5 × 109/L, n (%) | 41 (6%) | 86 (12.7%) | < 0.001 |
| WBC > 15 × 109/L, n (%) | 466 (66%) | 184 (27%) | < 0.001 |
| Platelets < 150 (×109/L), n (%) | 57 (8%) | 46 (6.8%) | 0.4 |
| Minimal Albumin (g/dL), mean ± SD | 3.8 ± 0.6 | 4.2 ± 0.4 | < 0.001 |
| Albumin < 3.5 g/dL, n (%) | 103/435 (24%) | 6/228 (2.6%) | < 0.001 |
| Minimal Phosphorus (mg/dL), mean ± SD | 4.1 ± 1.2 | 4.9 ± 1 | < 0.001 |
| Phosphorus < 3.5 mg/dL, n (%) | 95/352 (27%) | 6/90 (6.6%) | < 0.001 |
| Minimal Sodium (mEq/L), mean ± SD | 135 ± 3.6 | 135 ± 2.4 | 0.02 |
| Sodium < 135 mEq/L, n (%) | 302 (43%) | 240 (35%) | 0.006 |
| Minimal ALP (U/L), mean ± SD | 172.4 ± 59.5 | 215 ± 79 | < 0.001 |
| Minimal ALP SDS, mean ± SD | −1.65 ± 1.11 | −0.97 ± 1.36 | < 0.001 |
| Hospital duration, days, median (IQR) | 4 (3, 6) | 2 (1, 3) | < 0.001 |
| Mortality, n (%) | 7 (1%) | 0 (0%) | 0.02 |
Abbreviations: ALP, alkaline phosphatase; CRP, C‐reactive protein; SDS, standard deviation score; WBC, blood leucocytes.
FIGURE 2.
Whisker plots of nadir ALP‐SDS values in febrile controls (n = 673) and septic children without (n = 546) and with (n = 164) AKI. The reported medians in this figure are different from the means presented in Tables 1 and 2. ***p < 0.001.
Minimal ALP levels during hospitalization were significantly reduced in the sepsis group compared to the non‐sepsis controls, both in absolute values (172.4 ± 59.5 U/L vs. 215 ± 79 U/L, p < 0.001) as well as in ALP‐SDS (−1.65 ± 1.11 vs. −0.97 ± 1.36, p < 0.001). Regarding clinical outcomes, mortality occurred only in the sepsis group (1% vs. 0%, p = 0.02) and the median length of hospital stay was significantly longer [median (IQR): 4 (3, 6) vs. 2 (1, 3) days, p < 0.001] in the sepsis group vs. controls.
In the study group, 161 patients (23%) developed acute kidney injury (AKI). A significant variability in AKI incidence was observed between the participating hospitals. The number of AKI cases was 67 (22.4%) in Hospital H1, 39 (16.4%) in Hospital H2, 51 (32.2%) in Hospital H3 and 4 (57.1%) in Hospital H4 (p < 0.001). Most cases of AKI showed mild kidney injury: 108 patients (66%) met criteria for Stage 1 AKI, while 53 patients (33%) had more advanced kidney injury, including 32 (20%) patients with Stage 2 and 21 (13%) patients with Stage 3 AKI. Compared to non‐AKI patients, AKI patients were slightly older [median (IQR) age: 18.7 (9.4, 50.4) vs. 16.6 (10.5, 33.5) months, p = 0.004], without a significant difference in sex between the groups (males 55.9% vs. 52.7%, p = 0.5). Regarding microbiological findings, bacteremia was significantly more frequent in the AKI group (34.7% vs. 12.2%, p < 0.001), as well as the rate of bacteriuria (11.8% vs. 3.3%, p < 0.001). Inflammatory and hematologic markers indicated a more severe disease course in the AKI group. Mean maximal CRP levels were significantly higher (146.5 ± 127.1 vs. 102.3 ± 87.1 mg/L, p < 0.001) in the AKI‐group, as well as mean maximal total leucocyte count (22.1 ± 10.3 vs. 19.4 ± 8.7 × 109/L, p = 0.03), although the prevalence of leukocytosis (WBC > 15 × 109/L) was similar between the groups (71.4% vs. 65%, p = 0.155). Leukopenia (WBC < 5 × 109/L) was significantly more common in AKI patients (14.9% vs. 3.1%, p < 0.001). Thrombocytopenia (platelet count < 150 × 109/L) was also markedly more prevalent in the AKI group (27.9% vs. 2.2%, p < 0.001). Mean minimal albumin level during hospitalization was significantly lower in the AKI group (3.35 ± 0.74 vs. 3.97 ± 0.42 g/dL, p < 0.001), with a higher prevalence of hypoalbuminemia (< 3.5 g/dL) (54.5% vs. 10.2%, p < 0.001). Similarly, mean minimal serum phosphorus was lower in the AKI group (3.56 ± 1.44 vs. 4.45 ± 0.99 mg/dL, p < 0.001), and hypophosphatemia (< 3.5 mg/dL mg/dL) was more frequent (47.8% vs. 16.6%, p < 0.001). Mean minimal serum sodium was lower in AKI patients (133.5 ± 4.7 vs. 135.5 ± 3.1 mEq/L, p < 0.001), and hyponatremia (< 135 mEq/L) was more common (61% vs. 38%, p < 0.001). Minimal ALP levels were significantly lower in the AKI group (150.4 ± 62.1 U/L vs. 179 ± 57.2 U/L, p < 0.001). When adjusted to age‐ and sex‐specific normal values, minimal ALP‐SDS was still significantly lower in AKI patients [mean (SD): −1.95 ± 1.16 vs. −1.56 ± 1.09, p < 0.001]. Finally, hospital stay length was significantly longer [median (IQR): 7 (4, 12) vs. 4 (3, 5) days, p < 0.001], and mortality was significantly higher in the AKI group (4.3% vs. 0%, p < 0.001) (Table 2).
TABLE 2.
Comparison of the demographic and laboratory characteristics of septic children with (AKI) and without (non‐AKI) acute kidney injury.
| AKI (n = 161) | Non‐AKI (n = 540) | p | |
|---|---|---|---|
| Age (months), median (IQR) | 18.7 (9.4, 50.4) | 16.6 (10.5, 33.5) | 0.004 |
| Males, n (%) | 90 (55.9%) | 285 (52.7%) | 0.5 |
| Bacteremia, n (%) | 56 (34.7%) | 66 (12.2%) | < 0.001 |
| Bacteriuria, n (%) | 19 (11.8%) | 18 (3.33%) | < 0.001 |
| Max‐CRP (mg/l), mean ± SD | 146.5 ± 127.1 | 102.3 ± 87.1 | < 0.001 |
| Max‐WBC (×109/L), mean ± SD | 22.1 ± 10.23 | 19.4 ± 8.7 | 0.03 |
| WBC < 5 × 109/L, n (%) | 24 (14.9%) | 17 (3.1%) | < 0.001 |
| WBC > 15 × 109/L, n (%) | 115 (71.4%) | 351 (65%) | 0.2 |
| Platelets < 150 × 109/L, n (%) | 45 (27.9%) | 12 (2.2%) | < 0.001 |
| Minimal Albumin, g/dl, mean ± SD | 3.35 ± 0.74 | 3.97 ± 0.42 | < 0.001 |
| Albumin < 3.5 g/dL, n (%) | 72/132 (54.5%) | 31/303 (10.2%) | < 0.001 |
| Minimal Phosphorus (mg/dL), mean ± SD | 3.56 ± 1.44 | 4.45 ± 0.99 | < 0.001 |
| Phosphorus < 3.5 mg/dL, n (%) | 56/117 (47.8%) | 39/235 (16.6%) | < 0.001 |
| Minimal Sodium (mEq/L), mean ± SD | 133.53 ± 4.71 | 135.47 ± 3.12 | < 0.001 |
| Sodium < 135 mEq/L, n (%) | 98 (60.9%) | 204 (37.8%) | < 0.001 |
| Minimal ALP (U/L), mean ± SD | 150.4 ± 62.1 | 179.0 ± 57.2 | < 0.001 |
| Minimal ALP SDS, mean ± SD | −1.95 ± 1.16 | −1.56 ± 1.09 | < 0.001 |
| Hospital stay (days), median (IQR) | 7 (4, 12) | 4 (3, 5) | < 0.001 |
| Mortality, n (%) | 7 (4.3%) | 0 (0%) | < 0.001 |
Abbreviations: ALP, alkaline phosphatase; CRP, C‐reactive protein; SDS, standard deviation score; WBC, blood leucocytes.
To assess the independent association between ALP levels and the development of AKI in sepsis, a multivariate logistic regression analysis was performed. AKI was associated with lower ALP SDS (OR = 0.7, 95% CI: 0.6–0.8, p < 0.001), even after controlling for CRP (OR = 1.004, 95% CI: 1.002–1.006, p < 0.001) and leucocyte count (OR = 1.029, 95% CI: 1.007–1.051, p = 0.009).
A receiver operating characteristic (ROC) analysis was performed to evaluate ALP‐SDS discriminatory performance for predicting kidney injury. ALP‐SDS demonstrates an inverse association with AKI (AUC 0.39, 95% CI 0.34–0.44; inverted AUC ~0.61), supporting its role as one biomarker within a multi‐variable predictive framework for sepsis‐associated AKI. In a combined model, adding ALP‐SDS to a base model of CRP + leukocytosis (WBC > 5 × 109/L) produced a statistically significant improvement (ΔAUC = 0.065, 95% CI: 0.02−0.11, p = 0.005).
4. Discussion
In this large multicenter retrospective study, we investigated serum ALP levels in a cohort of 701 paediatric patients hospitalized with sepsis and compared them with a group of febrile, but non‐sepsis, age‐ and sex‐matched children. Clinical and laboratory parameters indicated a significantly more severe disease course in the sepsis group compared to febrile non septic controls, as expected. It is important to acknowledge that patient identification relied on ICD‐9 coding rather than clinical adjudication, which likely produced a heterogeneous cohort spanning mild systemic infections to severe sepsis with organ dysfunction. In addition, the study period (2000–2024) spans three eras of evolving sepsis definitions (Sepsis‐2, Sepsis‐3 and the paediatric Phoenix criteria), and some included patients—particularly those coded solely with SIRS or bacteremia diagnoses—may not fulfil current sepsis criteria. The febrile control group similarly may include many patients with viral infections, limiting direct comparison. These caveats should be considered when interpreting our findings. Our analysis yielded a major novel finding of lower ALP levels in the sepsis group compared to reported normal values which was reflected in a negative (and not a zero) mean SDS ALP value in our cohort, contrary to close to zero values in the control group of febrile non septic children. Moreover, inside the sepsis group, ALP levels were significantly lower in patients who developed AKI compared to those who maintained normal kidney function, independent of age, sex and inflammatory markers. Acute kidney injury is a marker of sepsis severity, as its occurrence and progression are strongly associated with increased disease severity, organ dysfunction and mortality [21]. This is reflected in this study too, by the comparison of laboratory markers of disease severity, length of hospitalization and mortality rate between AKI vs. non‐AKI groups.
The lower nadir ALP levels in the sepsis group compared to controls, and even lower in the AKI group compared to the non‐AKI group, suggest that lower ALP levels reflect a more severe disease. Considering its function in bacterial detoxification, low ALP levels may reflect ALP depletion in severe bacterial disease, as shown in animal studies [16]. Our data contrasts with data from adult studies, where ALP levels were shown to be elevated in sepsis and higher levels have been observed in patients with bacteremia and septic AKI, although the prognostic significance was variable. Notably, Baek et al. reported that ALP activity in adult ICU patients with septic AKI increased progressively over several days, interpreted as part of a host adaptive or organ‐release response. Whether children display a similar rising pattern later in the disease course could not be assessed in our study, as we analysed nadir ALP values rather than serial measurements. Longitudinal ALP trajectories in paediatric sepsis remain an important topic for future prospective investigation [17, 22]. This can be probably explained by a significantly higher baseline ALP levels in children due to their rapid bone turnover and higher ALP levels variability. The findings also raise the possibility that AKI in sepsis may arise through at least two distinct pathophysiological pathways. In one pathway, severe systemic inflammation—reflected by the constellation of thrombocytopenia, hyponatremia and hypoalbuminemia observed in our AKI group—drives multi‐organ dysfunction including kidney injury. In a second, potentially overlapping pathway, depletion of ALP via consumption during endotoxin detoxification and extracellular ATP dephosphorylation may directly impair renal protection, independent of the degree of systemic inflammatory response. The independent association between ALP‐SDS and AKI after adjustment for CRP and leukocyte count supports this second mechanism, as it suggests ALP depletion contributes to AKI risk even when controlling for the overall inflammatory burden. This is particularly relevant given that gram‐positive bacteria—which predominate in paediatric sepsis—do not express LPS. Still the ability of ALP to dephosphorylate molecules in the ATP‐adenosine pathway may be an important LPS‐independent mechanism through which ALP exerts renoprotection across bacterial subtypes [23]. We were unable to formally compare ALP levels in gram‐negative versus gram‐positive sepsis in our cohort due to incomplete microbiological data; this represents an important direction for future research.
Recent studies show that ALP plays a protective and potentially therapeutic role in sepsis. Its primary mechanism involves the dephosphorylation and detoxification of bacterial lipopolysaccharide (LPS), a key driver of the inflammatory response in sepsis. By dephosphorylating LPS and other toll‐like receptor agonists, ALP reduces their pro‐inflammatory activity, thereby attenuating systemic inflammation and organ injury, particularly in the kidney [23, 24]. The enzyme also converts extracellular ATP released during cellular stress into adenosine, which has anti‐inflammatory and tissue‐protective effects [25]. These actions collectively contribute to reduced cytokine production, improved organ function and, in some studies, improved survival [26]. Our findings support the concept of ALP serving not just as a marker, but also as a potential biological contributor to AKI pathogenesis, raising the hypothesis that ALP supplementation could be a therapeutic strategy. This hypothesis is supported by prior preclinical and early clinical data but is not directly supported by the present observational study and should be interpreted with caution. In animal models and early‐phase clinical trials, administration of exogenous ALP improved renal function and reduced markers of tubular injury in patients with sepsis‐associated AKI [27, 28].
This study has several limitations. First, the retrospective design limits causal inference; therefore, we could not establish a direct causal link between serum ALP levels and AKI. To address this, our analysis focused on identifying associations rather than causal relationships, while adjusting for key confounding variables such as age, sex and illness severity (based on inflammatory markers), to reduce bias. Furthermore, because we used the nadir (minimum) ALP value recorded during the entire hospitalization, we cannot exclude the possibility of reverse causation—that is, that AKI itself contributed to ALP reduction rather than low ALP predating kidney injury. Future prospective studies with serial ALP measurements are needed to establish temporal directionality. Second, regarding potential variability in ALP measurements across hospitals, verification was made to ensure that the measurements were performed using comparable analytical methods across all participating centres, ensuring consistency and data reliability. Third, missing data on ALP levels resulted in the exclusion of cases from the analysis: of 1504 patients with sepsis‐related ICD‐9 codes, 789 (52%) were excluded due to absent ALP measurements, likely reflecting that ALP is not a routine emergency panel test in children. We cannot exclude the possibility that differential ALP ordering rates introduced a selection bias, and this should be considered when interpreting the findings. However, the inclusion of data from four diverse medical centres enhances the generalizability of our findings and mitigates potential single‐centre bias. In addition, excluding patients with nadir ALP SDS > 3 could remove patients with hepatic involvement secondary to sepsis, or other relevant pathology, which would confound ALP interpretation. We acknowledge that some of these patients may have had hepatic involvement secondary to sepsis, which could represent a more severe subgroup. However, only 14 (2%) of the original cohort had ALP SDS > 3 and were excluded. Fourth, patient inclusion relied exclusively on ICD‐9 sepsis‐related codes without clinical re‐adjudication. ICD‐9‐based definitions of paediatric sepsis are known to have variable precision, and our cohort likely encompasses a wide clinical spectrum, including patients who would not fulfil modern sepsis criteria (Sepsis‐3 or Phoenix). Additionally, children with severe infections coded under organ‐specific diagnoses (e.g., pneumonia, pyelonephritis) may have been missed, meaning our cohort may not fully represent the broader population of paediatric sepsis. The study also spans over two decades during which sepsis definitions evolved considerably (from Sepsis‐2 through Sepsis‐3 to the paediatric Phoenix criteria), potentially introducing temporal heterogeneity in case composition. Fifth, our AKI definition was restricted to serum creatinine dynamics, as urine output data were not reliably available in the database. Omission of the urine output criterion is consistent with most retrospective studies but leads to underestimation of AKI incidence, meaning the true association between ALP and AKI may be stronger than reported. Sixth, residual confounding cannot be excluded: important unmeasured variables such as fluid balance, antibiotic timing, nutritional status and zinc and magnesium levels (known cofactors of ALP activity) may independently influence both ALP levels and AKI risk. Finally, we were unable to perform subgroup analyses by bacterial aetiology (gram‐negative vs. gram‐positive) due to incomplete microbiological data, which limits evaluation of the LPS‐mediated mechanistic hypothesis.
In conclusion, reduced ALP levels are strongly associated with the presence of sepsis and the development of AKI in paediatric patients, possibly reflecting its depletion in the fight against infection. Further prospective studies are needed to confirm the predictive value of ALP for severe infection and to evaluate the therapeutic potential of exogenous ALP administration in paediatric sepsis.
Author Contributions
Ido Levi: data curation, investigation, writing – original draft. Yehonatan Marziano: funding acquisition. Michael Kalinin: conceptualization. Evgenia Gurevich: conceptualization, investigation, methodology, data curation, writing – original draft. Daniel Landau: conceptualization, supervision, writing – review and editing, methodology. Natalya Bilenko: formal analysis.
Funding
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Sharma U., Pal D., and Prasad R., “Alkaline Phosphatase: An Overview,” Indian Journal of Clinical Biochemistry 29, no. 3 (2014): 269–278, 10.1007/s12291-013-0408-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Vimalraj S., “Alkaline Phosphatase: Structure, Expression and Its Function in Bone Mineralization,” Gene 754 (2020): 144855, 10.1016/j.gene.2020.144855. [DOI] [PubMed] [Google Scholar]
- 3. Crofton P. M., “Wheat‐Germ Lectin Affinity Electrophoresis for Alkaline Phosphatase Isoforms in Children: Age‐Dependent Reference Ranges and Changes in Liver and Bone Disease,” Clinical Chemistry 38, no. 5 (1992): 663–670. [PubMed] [Google Scholar]
- 4. Riancho J. A., “Diagnostic Approach to Patients With Low Serum Alkaline Phosphatase,” Calcified Tissue International 112, no. 3 (2023): 289–296, 10.1007/s00223-022-01039-y. [DOI] [PubMed] [Google Scholar]
- 5. Rudd K. E., Johnson S. C., Agesa K. M., et al., “Global, Regional, and National Sepsis Incidence and Mortality, 1990‐2017: Analysis for the Global Burden of Disease Study,” Lancet 395, no. 10219 (2020): 200–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Levy M. M., Fink M. P., Marshall J. C., et al., “SCCM/ESICM/ACCP/ATS/SIS. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference,” Critical Care Medicine 31, no. 4 (2003): 1250–1256. [DOI] [PubMed] [Google Scholar]
- 7. Fitzgerald J. C., Ross M. E., Thomas N. J., Weiss S. L., Balamuth F., and Anderson A. H., “Risk Factors and Inpatient Outcomes Associated With Acute Kidney Injury at Pediatric Severe Sepsis Presentation,” Pediatric Nephrology 33, no. 10 (2018): 1781–1790. [DOI] [PubMed] [Google Scholar]
- 8. Khatana J., Thavamani A., Umapathi K. K., Sankararaman S., and Roy A., “Increasing Incidence of Acute Kidney Injury in Pediatric Severe Sepsis and Related Adverse Hospital Outcomes,” Pediatric Nephrology 38, no. 8 (2023): 2809–2815. [DOI] [PubMed] [Google Scholar]
- 9. Kuai Y., Li M., Chen J., et al., “Comparison of Diagnostic Criteria for Acute Kidney Injury in Critically Ill Children: A Multicenter Cohort Study,” Critical Care 26, no. 1 (2022): 207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Koyama I., Matsunaga T., Harada T., Hokari S., and Komoda T., “Alkaline Phosphatases Reduce Toxicity of Lipopolysaccharides In Vivo and In Vitro Through Dephosphorylation,” Clinical Biochemistry 35, no. 6 (2002): 455–461, 10.1016/s0009-9120(02)00330-2. [DOI] [PubMed] [Google Scholar]
- 11. Bentala H., Verweij W. R., Huizinga‐Van der Vlag A., et al., “Removal of Phosphate From Lipid A as a Strategy to Detoxify Lipopolysaccharide,” Shock 18, no. 6 (2002): 561–566, 10.1097/00024382-200212000-00013. [DOI] [PubMed] [Google Scholar]
- 12. Pickkers P., Heemskerk S., Schouten J., et al., “Alkaline Phosphatase for Treatment of Sepsis‐Induced Acute Kidney Injury: A Prospective Randomized Double‐Blind Placebo‐Controlled Trial,” Critical Care 16, no. 1 (2012): R14, 10.1186/cc11159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kapojos J. J., Poelstra K., Borghuis T., et al., “Induction of Glomerular Alkaline Phosphatase After Challenge With Lipopolysaccharide,” International Journal of Experimental Pathology 84, no. 3 (2003): 135–144, 10.1046/j.1365-2613.2003.00345.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Khundmiri S. J., Asghar M., Khan F., et al., “Effect of Reversible and Irreversible Ischemia on Marker Enzymes of BBM From Renal Cortical PT Subpopulations,” American Journal of Physiology. Renal Physiology 273 (1997): F849–F856. [DOI] [PubMed] [Google Scholar]
- 15. Wy C. A., Goto M., Young R. I., Myers T. F., and Muraskas J., “Prophylactic Treatment of Endotoxic Shock With Monophosphoryl Lipid A in Newborn Rats,” Biology of the Neonate 77, no. 3 (2000): 191–195, 10.1159/000014215. [DOI] [PubMed] [Google Scholar]
- 16. Beumer C., Wulferink M., Raaben W., Fiechter D., Brands R., and Seinen W., “Calf Intestinal Alkaline Phosphatase, a Novel Therapeutic Drug for Lipopolysaccharide (LPS)‐Mediated Diseases, Attenuates LPS Toxicity in Mice and Piglets,” Journal of Pharmacology and Experimental Therapeutics 307, no. 2 (2003): 737–744, 10.1124/jpet.103.056606. [DOI] [PubMed] [Google Scholar]
- 17. Baek S. D., Kang J. Y., Yu H., et al., “Change in Alkaline Phosphatase Activity Associated With Intensive Care Unit and Hospital Length of Stay in Patients With Septic Acute Kidney Injury on Continuous Renal Replacement Therapy,” BMC Nephrology 19 (2018): 243, 10.1186/s12882-018-1028-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Rappoport N., Livne G., Perry Cohen N., et al., “Kineret: Israel's Largest Hospital Network Transformed Into the OMOP Common Data Model for Collaborative Research,” PLoS One 20, no. 10 (2025): e0334848, 10.1371/journal.pone.0334848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Khwaja A., “KDIGO Clinical Practice Guidelines for Acute Kidney Injury,” Nephron. Clinical Practice 120, no. 4 (2012): c179–c184, 10.1159/000339789. [DOI] [PubMed] [Google Scholar]
- 20. Strauch J. M., Vogel M., Meigen C., et al., “Pediatric Reference Values of Alkaline Phosphatase: Analysis From a German Population‐Based Cohort and Influence of Anthropometric and Blood Parameters,” Bone 174 (2023): 116809, 10.1016/j.bone.2023.116809. [DOI] [PubMed] [Google Scholar]
- 21. Patanwala A. E. and Erstad B. L., “Epidemiology of Septic Shock Associated Acute Kidney Injury: A National Retrospective Cohort Study,” Critical Care Medicine 53 (2025): e1601–e1609, 10.1097/CCM.0000000000006735. [DOI] [PubMed] [Google Scholar]
- 22. Liu S., Zhao K., Shao C., Xu L., Cui X., and Wang Y., “Association Between Alkaline Phosphatase to Albumin Ratio and Mortality Among Patients With Sepsis,” Scientific Reports 14, no. 1 (2024): 3170, 10.1038/s41598-024-53384-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Peters E., Geraci S., Heemskerk S., et al., “Alkaline Phosphatase Protects Against Renal Inflammation Through Dephosphorylation of Lipopolysaccharide and Adenosine Triphosphate,” British Journal of Pharmacology 172, no. 20 (2015): 4932–4945, 10.1111/bph.13261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Pettengill M., Matute J. D., Tresenriter M., et al., “Human Alkaline Phosphatase Dephosphorylates Microbial Products and Is Elevated in Preterm Neonates With a History of Late‐Onset Sepsis,” PLoS One 12, no. 4 (2017): e0175936, 10.1371/journal.pone.0175936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Peters E., van Elsas A., Heemskerk S., et al., “Alkaline Phosphatase as a Treatment of Sepsis‐Associated Acute Kidney Injury,” Journal of Pharmacology and Experimental Therapeutics 344, no. 1 (2013): 2–7, 10.1124/jpet.112.198226. [DOI] [PubMed] [Google Scholar]
- 26. Heemskerk S., Masereeuw R., Moesker O., et al., “Alkaline Phosphatase Treatment Improves Renal Function in Severe Sepsis or Septic Shock Patients,” Critical Care Medicine 37, no. 2 (2009): 417–423, 10.1097/CCM.0b013e31819598af. [DOI] [PubMed] [Google Scholar]
- 27. Hümmeke‐Oppers F., Hemelaar P., and Pickkers P., “Innovative Drugs to Target Renal Inflammation in Sepsis: Alkaline Phosphatase,” Frontiers in Pharmacology 10 (2019): 919, 10.3389/fphar.2019.00919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Pickkers P., Mehta R. L., Murray P. T., et al., “Effect of Human Recombinant Alkaline Phosphatase on 7‐Day Creatinine Clearance in Patients With Sepsis‐Associated Acute Kidney Injury: A Randomized Clinical Trial,” Journal of the American Medical Association 320, no. 19 (2018): 1998–2009, 10.1001/jama.2018.14283. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.