Systemic Human Neutrophil Lipocalin Associates with Severe Acute Kidney Injury in SARS-CoV-2 Pneumonia

Abstract

Neutrophils have been suggested mediators of organ dysfunction in COVID-19. The current study investigated if systemic neutrophil activity, estimated by human neutrophil lipocalin (HNL) concentration in peripheral blood, is associated with acute kidney injury (AKI) development. A total of 103 adult patients admitted to intensive care, with PCR-confirmed SARS-CoV-2 infection, were prospectively included (Clinical Trials ID: NCT04316884). HNL was analyzed in plasma (P-HNL Dimer) and in whole blood (B-HNL). The latter after ex vivo activation with N-formyl-methionine-leucine-phenylalanine. All patients developed respiratory dysfunction and 62 (60%) were treated with invasive ventilation. Sixty-seven patients (65%) developed AKI, 18 (17%) progressed to AKI stage 3, and 14 (14%) were treated with continuous renal replacement therapy (CRRT). P-HNL Dimer was higher in patients with invasive ventilation, vasopressors, AKI, AKI stage 3, dialysis, and 30-day mortality (p < 0.001-0.046). B-HNL performed similarly with the exception of mild AKI and mortality (p < 0.001-0.004). The cohort was dichotomized by ROC estimated cutoff concentrations of 13.2 µg/L and 190 µg/L for P-HNL Dimer and B-HNL respectively. Increased cumulative risks for AKI, AKI stage 3, and death were observed if above the P-HNL cutoff and for AKI stage 3 if above the B-HNL cutoff. The relative risk of developing AKI stage 3 was nine and 39 times greater if above the cutoffs in plasma and whole blood, respectively, for CRRT eight times greater for both. In conclusion, systemically elevated neutrophil lipocalin, interpreted as increased neutrophil activity, was shown to be associated with an increased risk of severe AKI, renal replacement therapy, and mortality in COVID-19 patients with respiratory failure.

Keywords: COVID-19; HNL; acute kidney injury; intensive care; neutrophils.

Conflict of interest statement

Per Venge is the inventor and holds patents of measuring neutrophil-specific dimeric HNL by unique antibodies and HNL after activation in whole blood. None of the other authors declare any conflict of interest.

Figures

Figure 1

Western blot of plasma samples during the first week of ICU treatment. Monomeric (22 kD), homodimeric (45 kD), and heterodimeric (>90 kD) HNL all present in plasma samples collected at admission (1) and on day three (3) during the first week of intensive care. The letters A–K represent individual patients in the cohort. HNL, human neutrophil lipocalin.

Figure 2

Human neutrophil lipocalin in plasma at admission and peak concentrations in whole blood during intensive care vs. acute kidney injury stage and plasma creatinine. A one-way ANOVA (A) demonstrating that P-HNL Dimer at admission differed between AKI stages (p = 0.010); however, post-hoc analysis indicated the main difference to be between no AKI and AKI stage 3 * (p = 0.006, Bonferroni-adjusted). The plasma creatinine concentration increased as P-HNL Dimer increased significantly (B). The biomarker could explain 13% of the variation in creatinine. A one-way ANOVA of AKI stage and B-HNL (C) was also significant (p < 0.001). Post-hoc analysis revealed a difference between no AKI, AKI stage 1, and AKI stage 2 separately with AKI stage 3 * (p = 0.002, p < 0.001, and p = 0.003, Bonferroni-adjusted). Plasma creatinine increased with peak B-HNL concentrations, (D) and the biomarker could explain 31% of the variation in plasma creatinine.

Figure 3

Probability of admission to intensive care, acute kidney injury of any stage, and stage 3 development, as well as mortality vs. human neutrophil lipocalin in plasma at admission (A–D) and vs. peak concentrations of human neutrophil lipocalin in whole blood (E–H) during intensive care. ROC in conjunction with Youden’s J statistic were calculated to find suitable cutoff values for HNL in plasma and whole blood. Since the primary focus was severe AKI development, a cutoff close to that calculated from AKI stage 3 was applied to all four outcomes in the figure. The concentrations used were 13.2 µg/L and 190 µg/L for P-HNL Dimer and peak B-HNL, respectively. Kaplan–Meyer curves were than calculated for ICU admission (A,E), survival (B,F), AKI (C,G), and AKI stage 3 (D,H). Time to ICU admission was not different in the groups above and below the cutoffs for P-HNL Dimer or B-HNL (p = 0.92 and p = 0.87). The cumulative risk of death, developing AKI of any stage and AKI stage 3 were higher if above 13.2 µg/L in plasma at admission (p = 0.0057, p = 0.035, and p < 0.0001, respectively). The cumulative risk of death or developing AKI did not differ between those above or below 190.0 µg/L B-HNL (p = 0.18 and p = 0.36, respectively); however, the cumulative risk of developing AKI stage 3 was greater in the patients above the cutoff (p < 0.001).

Figure 4

Log odds for outcomes dependent on the concentration of dimeric human neutrophil lipocalin in plasma at admission (A) and peak concentration of human neutrophil lipocalin in whole blood (B) during intensive care. For each outcome, a logistic regression was performed for P-HNL Dimer at admission and peak B-HNL during intensive care. The biomarker was used as a categorical variable, either above or below the cutoff of 13.2 µg/L and 190 µg/L, respectively, and the regression was adjusted for age. Patients with P-HNL Dimer and peak B-HNL concentrations above said cutoffs demonstrated increased risks of developing AKI stage 3 (8.99 and 39.2 respectively) and renal replacement therapy (7.77 for both).

Figure 5

Human neutrophil lipocalin in plasma at admission and peak concentrations in whole blood during intensive care vs. invasive ventilation (A,B), 30-day mortality (C,D), and SOFA score (E,F). P-HNL Dimer at admission differentiated between survivors and non-survivors at 30 days, but peak B-HNL concentration did not (p = 0.034 and p = 0.087). Those that were treated with invasive ventilation demonstrated higher mean concentrations of P-HNL Dimer and B-HNL (p = 0.004 and p = 0.002). P-HNL Dimer and B-HNL could both separate between patients with high and low SOFA scores with increasing HNL concentrations associated with higher SOFA scores during ICU stay (p = 0.002 and p = 0.011).