Acute kidney injury biomarker olfactomedin 4 predicts furosemide responsiveness

Abstract

Background: Acute kidney injury (AKI) is associated with increased morbidity and mortality in critically ill patients. Olfactomedin 4 (OLFM4), a secreted glycoprotein expressed in neutrophils and stressed epithelial cells, is upregulated in loop of Henle (LOH) cells following AKI. We hypothesized that urine OLFM4 (uOLFM4) will increase in patients with AKI and may predict furosemide responsiveness.

Methods: Urine from critically ill children was collected prospectively and tested for uOLFM4 concentrations with a Luminex immunoassay. Severe AKI was defined by KDIGO (stage 2/3) serum creatinine criteria. Furosemide responsiveness was defined as > 3 mL/kg/h of urine output in the 4 h after a 1 mg/kg IV furosemide dose administered as part of standard of care.

Results: Fifty-seven patients contributed 178 urine samples. Irrespective of sepsis status or AKI cause, uOLFM4 concentrations were higher in patients with AKI (221 ng/mL [IQR 93-425] vs. 36 ng/mL [IQR 15-115], p = 0.007). uOLFM4 concentrations were higher in patients unresponsive to furosemide (230 ng/mL [IQR 102-534] vs. 42 ng/mL [IQR 21-161], p = 0.04). Area under the receiver operating curve for association with furosemide responsiveness was 0.75 (95% CI, 0.60-0.90).

Conclusions: AKI is associated with increased uOLFM4. Higher uOLFM4 is associated with a lack of response to furosemide. Further testing is warranted to determine whether uOLFM4 could identify patients most likely to benefit from earlier escalation from diuretics to kidney replacement therapy to maintain fluid balance. A higher resolution version of the Graphical abstract is available as Supplementary information.

Keywords: Acute kidney injury; Diuretic responsiveness; Furosemide stress test; Olfactomedin 4; Urine biomarker.

Fig. 1

Urine OLFM4 and NGAL concentrations in patients with and without sepsis and severe AKI. a uOLFM4 concentrations in patients in four different groups. Linear mixed modeling was used for testing on log transformed data. For each of the groups, n = 33, 15, 67, and 63, in the same order as presented in the figure. *p < 0.001. b uNGAL concentrations in patients in four different groups. For each of the groups, n = 33, 14, 67, and 63, in the same order as presented in the figure. *p < 0.001

Fig. 2

Pairwise comparisons of urine OLFM4 and NGAL concentrations in subgroups of patients. Linear mixed modeling was used for testing on log transformed data was used for all pairwise comparisons. a uOLFM4 concentrations in all patients with and without severe AKI. n = 48, 130; p < 0.001. b uNGAL concentrations in all patients with and without severe AKI. n = 48, 130, p < 0.001. Assessing c uOLFM4 and d uNGAL concentrations in only the no severe AKI subset of patients, comparing patients without and with sepsis. n = 33, 15; p = 0.69, and n = 33, 14; p <0.001, respectively. Assessing e uOLFM4 and f uNGAL concentrations in only the sepsis subset of patients, comparing patients without and with severe AKI. n = 15, 63; p ≤ 0.001 and n = 14, 63; p = 0.006, respectively *p < 0.05. uOLFM4, urine olfactomedin 4; uNGAL, urine neutrophil gelatinase-associated lipocalin

Fig. 3

Urine OLFM4 but not NGAL concentration predicts furosemide responsiveness. Linear mixed modeling was used for testing on log transformed data for pairwise comparisons. a uOLFM4 concentration is higher in those patients who failed to respond to furosemide n = 19, 27; p = 0.04. b uNGAL concentration is not different in patients who responded to furosemide vs. those who did not. n = 19, 26; p = 0.11. c Receiver operating curve of the ability for uOLFM4 concentration to predict furosemide responsiveness. *p < 0.05. uOLFM4, urine olfactomedin 4; uNGAL, urine neutrophil gelatinase associated lipocalin; AUC, area under the receiver operating curve

Fig. 4

Urine uromodulin/OLFM4 ratio predicts severe AKI and failure to respond to furosemide. For these comparisons, log transformed data was used with a Mann–Whitney test rather than linear mixed model due to smaller numbers. a Uromodulin concentrations alone are not different between patients with vs. without severe AKI n = 25, 76; p = 0.16. b The uromodulin/OLFM4 ratio is lower in those patients with severe AKI n = 25, 76; p ≤ 0.0001. *p < 0.05. OLFM4, olfactomedin 4; Umod, uromodulin