Urine complement activation fragments are increased in patients with kidney injury after cardiac surgery

Abstract

Experiments in mouse models have shown that the complement cascade is activated within the kidney after ischemia-reperfusion and that complement activation contributes to tubular injury in this setting. Less is known, however, about complement activation in human kidneys after ischemia or whether complement activation in the tubulointerstitium can be detected by measurement of complement fragments in the urine. We hypothesized that urine biomarkers of complement activation would rapidly increase in patients who develop ischemic acute kidney injury, signaling complement activation within the kidney. We confirmed that the alternative pathway of complement is activated in the kidneys of mice after ischemia-reperfusion, and we found that levels of factor B fragments (generated during alternative pathway activation) rapidly increase in the urine. We next performed a case-control study in which we measured complement fragments in human urine samples from patients undergoing cardiac surgery using ELISAs. The level of Ba increased after cardiac surgery and was significantly higher in patients who developed acute kidney injury. The increase in Ba also correlated with magnitude of the subsequent rise in serum creatinine and with the need for hemodialysis during the hospitalization. These findings demonstrate that the alternative pathway of complement is activated in patients who develop acute kidney injury after cardiac surgery and that increases in the level of urine Ba may be a predictive and functional biomarker of severe kidney injury.

INTRODUCTION

Acute kidney injury (AKI) may affect up to 20% of hospitalized patients worldwide (26) and is the most common complication of cardiac surgery (1). AKI prolongs hospital stays, and it is associated with an increased risk of chronic kidney disease and death (4–7, 12, 23). Unfortunately, there are currently no specific therapies of proven efficacy for preventing or shortening AKI, and care of patients with this disease is primarily supportive. Several obstacles have hampered the development of new therapies for AKI. The diagnosis is based on an increase in the serum creatinine level and urine output. These are not injury markers and may reflect transient underperfusion of the kidneys. Furthermore, these indexes may not change until days after the kidney has been damaged. In addition, AKI is caused by a large number of heterogeneous etiologies, and changes in serum creatinine or urine output do not distinguish among these causes. Because of these challenges, there has been great interest in identifying new biomarkers of AKI (14, 38). In particular, mechanistic biomarkers may be useful for identifying patients most likely to respond to a particular therapy.

Experimental work has shown that AKI is associated with an inflammatory response in the kidney, even when the primary kidney insult is “nonimmune,” such as after ischemia-reperfusion or toxic injury (3, 28, 31). This inflammatory response contributes to kidney injury (20), and it also alters the immune response systemically (10). Systemic inflammation may contribute to the extrarenal manifestations of AKI, including risk of infection and cardiovascular disease. Several different mechanisms may translate cellular injury within the kidney into systemic immune activation. The release of damage-associated molecular patterns can activate Toll-like receptors (16, 37). Murine models have also shown that the complement system is activated in the kidney after ischemia-reperfusion (31, 39). Kidney ischemia is associated with local production of complement proteins by tubular cells as well as impaired complement regulation by the cells, thereby linking aseptic kidney injury with conditions that promote complement activation (8, 31, 39). Immunostaining of biopsy tissue from human patients with tubular injury confirmed that complement is activated in the damaged tubulointerstitium (33).

Previous work has also shown that the alternative pathway of complement is activated when blood comes in contact with artificial materials, such as during cardiopulmonary bypass (CPB) (9, 15). Activation in the bypass circuit may contribute to systemic inflammation. Interestingly, extrarenal complement activation by artificial surfaces may prolong ischemic AKI (24), potentially linking extracorporeal therapies to more severe kidney injury.

In addition to triggering an inflammatory response, the complement system generates several soluble and tissue-bound protein fragments that are useful biomarkers (34). For example, C3a and C5a (anaphylatoxins) are generated during complement activation (36). Ba is produced by activation of the alternative pathway of complement, and C4a is generated during activation of the classical and lectin pathways. Complement activation fragments are short lived, so their detection indicates ongoing complement activation (19). Detection of elevated complement fragments could identify patients most likely to benefit from complement inhibitory drugs and also monitor the response of patients to treatment. Complement activation fragments have been measured in the urine of patients with glomerular disease (35), although it is not known whether measurement of complement fragments in the urine can be used to detect tubulointerstitial complement activation.

Based on the extensive evidence of alternative pathway complement activation in preclinical models and human AKI (29, 31, 33, 39), we hypothesized that an alternative pathway fragment, Ba, would increase in the urine of patients with AKI. To test this hypothesis, we measured activation fragments in urine samples from patients who underwent cardiac surgery, and we correlated the results with the subsequent development of AKI.

MATERIALS AND METHODS

Patients.

The Translational Research Investigating Biomarker Endpoints in AKI (TRIBE-AKI) study is a prospective study of adults at high risk for AKI who underwent cardiac surgery (coronary artery bypass graft or valve surgery) at six academic medical centers in North American from July 2007 to December 2009. Samples were collected preoperatively and daily postoperatively for up to 5 days. Detailed methods have been previously described (18). From the 1,219 enrolled participants, an initial analysis was performed on 20 randomly selected patients and 20 randomly selected controls. For a more detailed analysis of urine Ba and soluble C5b-9 (sC5b-9), 105 participants were randomly selected by stratified sampling on case/control sample. Cases were defined as participants receiving acute dialysis or a doubling in serum creatinine from preoperative value (Acute Kidney Injury Network stage 2 or higher) (17). Controls were defined as participants without AKI by Acute Kidney Injury Network stages. Thirty-five cases and seventy matched controls were selected, matching on preoperative estimated glomerular filtration rate (eGFR) category, age (within 5 yr), and sex. All participants had preoperative, day 1 (0–6 h), day 2, and day 3 urine samples available.

All pre- and postoperative serum creatinine levels were measured in the same laboratory for each patient at all centers. Preoperative characteristics, operative details, and postoperative complications were defined as outlined by The Society of Thoracic Surgeons (2). Preoperative eGFR was measured using the Chronic Kidney Disease Epidemiology Collaboration equation.

Antibodies.

The following primary antibodies were used in these studies: FITC-conjugated goat IgG to mouse complement C3 (MP Biomedicals, Santa Ana, CA), a monoclonal antibody (mAb) that only recognizes the iC3b and C3d activation fragments of C3 (mAb 3d29) (30), an antibody to C4 (Hycult Biotech, Uden, The Netherlands), goat anti-human properdin factor B (DiaSorin/Bio-Rad, Hercules, CA), and a biotinylated mAb to mouse factor B that binds to the Ba fragment (29). Secondary antibodies include Cy-5 goat anti-rat IgG, horseradish peroxidase (HRP-conjugated donkey anti-rat IgG, and HRP-streptavadin (all from Jackson ImmunoResearch, West Grove, PA). Isotype controls included rat IgG2a, κ (R&D Systems, Minneapolis, MN), and FITC-conjugated goat IgG, whole molecule (Jackson ImmunoResearch).

Murine model of kidney ischemia-reperfusion injury.

Ischemia-reperfusion injury (IRI) was induced as previously described (21). Mice were anesthetized with ketamine and xylazine, and their body temperature was maintained by placing them on a heating pad during surgery. The left and right renal pedicles were clamped for 24 min using surgical clips (Miltex Instrument, Dedham, MA). The clamps were then removed, and the kidneys were observed to ensure reperfusion. The abdominal fascia and skin were then sutured with 4-0 silk (United States Surgical, Norwalk, CT), and normal saline (0.5 ml) was injected subcutaneously. Mice were placed in an incubator at 29°C for 2 h to maintain body temperature while recovering from anesthesia. Two urine samples [“Pre” (3 h before IRI) and “3 h” (3 h after reperfusion)] were collected from each animal and immediately snap frozen in liquid nitrogen. Mice were euthanized after 24 h of reperfusion.

Fluorescence microscopy.

Kidney samples were snap frozen in OCT (Sakura Finetek USA, Torrance, CA) and stored at −80°C. Sagittal sections (5-μm sections) were cut using a cryostat maintained at −20°C, briefly allowed to dry to room temperature, and immediately fixed with absolute acetone. Nonspecific binding was blocked for 1 h at room temperature with 5% heat-inactivated goat serum with 1% BSA in PBS. Primary antibodies were diluted in 2% heat-inactivated goat serum with 1% BSA in PBS and incubated overnight at 4°C. Autofluorescence was blocked with 0.05% Sudan Black B in 70% ethanol for 20 min at room temperature followed by two 10-min washes in deionized water (25). Fluorescence images were obtained with a Zeiss Axio Observer D1 inverted fluorescence microscope (Carl Zeiss Microscopy, Thornwood, NY).

Complement measurements.

To measure Ba in mouse urine samples, Immulon 4HBX ELISA plates (ThermoFisher Scientific, Waltham, MA) were coated with anti-human properdin factor B diluted in 50 mM carbonate/bicarbonate buffer (pH 9.6), incubated overnight at 4°C, washed with 0.05% PBS-Tween 20, and then blocked with ELISA Ultrablock (Bio-Rad) for 1 h at room temperature. Urine collected 3 h before and after IRI was diluted 1:40, and purified mouse factor B was serially diluted 9.4–1,000 ng/ml, all in 0.05% PBS-Tween2 0 with 2% BSA. Diluted samples were plated and left to incubate for 1 h at room temperature. Factor B was detected using a biotinylated mAb to factor B (5 µg/ml) followed by HRP-streptavadin (1 µg/ml) and developed with 3,3′,5,5′-tetramethylbenzidine (ThermoFisher Scientific) diluted 1:1 with Milli-Q water. The reaction was stopped with 0.2 N sulfuric acid, and plates were read in triplicate at 450 nm.

C4 in mouse urine was detected by Western blot analysis. Briefly, equal volumes of urine collected from each animal 3 h before and after IRI were applied to a 10% gel, in which urine proteins were separated using SDS-PAGE under nonreducing conditions and then transferred to a polyvinylidene fluoride membrane. The membrane was first blocked with 5% nonfat dry milk in 0.05% PBS-Tween 20 followed by primary and secondary antibody incubation, both of which were diluted in 2.5% nonfat dry milk in 0.05% PBS-Tween 20. Proteins were visualized with Pierce enhanced chemiluminescence Western blot substrate (ThermoFisher Scientific).

To measure complement activation fragments (Ba, C4d, C3a, and sC5b-9) in human samples, analytes were detected using commercial ELISAs according to the manufacturer’s instructions (Quidel, San Diego, CA). Freshly voided urine was processed as previously described (18), and samples were stored at −80°C until use. Urine samples were diluted as follows: Ba, 1:200 (samples requiring further dilution were diluted 1:800 or 1:1,000); C4d, 1:5; C3a, 1:10; and sC5b-9, 1:12.

Measurement of other biomarkers.

Other urine and plasma biomarkers were measured as previously described, including urine albumin, urine creatinine, IL-18, and neutrophil gelatinase-associated lipocalin (NGAL) (18).

Statistical methods.

Continuous variables were compared using Wilcoxon rank-sum tests and dichotomous variables were compared with a χ²-test or Fisher’s exact test. P values of <0.05 were considered significant.

RESULTS

Complement Ba levels increase in mouse urine after renal IRI.

We have previously found that complement fragments are deposited in the kidneys of mice with IRI (32). In mice, the classical pathway of complement is activated in the glomeruli after ischemia, but the activation is well controlled at this location and does not contribute to AKI (11). The alternative pathway of complement is activated in the tubulointerstitium, however, and contributes to tubular injury (31). We induced IRI in wild-type mice. We confirmed alternative pathway activation in the tubulointerstitium by dual staining kidney tissue for C4 and C3 fragments (Fig. 1A). C3 fragments were deposited on the proximal tubules in mice with IRI. As previously reported, C4 deposits were seen in the glomeruli but were not seen in the tubulointerstitium (21).

Fig. 1.

The alternative pathway of complement is activated in mice after kidney ischemia-reperfusion (IR). C57BL/6 mice were subjected to 24 min of kidney ischemia or to sham surgery. A: kidneys were examined by immunofluorescence microscopy for C3b (green) and C4 (red), and nuclei were stained with DAPI (blue). C4 was seen in the glomeruli (arrowheads), and C3b was seen along the tubules (arrows) of sham-treated animals. C3b and C4 deposition was more intense in mice after ischemia and 24 h of reperfusion (T24 IR). C3 staining was still not seen in the glomeruli, however, and C4 staining was not seen in the tubulointerstitium. Control staining with isotype-matched antibodies was negative. Original magnification ×200. Scale bar = 50 μm. B: urine Ba fragments were measured by ELISA in urine samples collected before IR (pre) or after 3 h of reperfusion (3 h). Levels were significantly higher after IR. ***P < 0.001. C: Western blot analysis of urine for C4 showed that levels of intact C4 as well as C4 fragments were higher in samples collected after 3 h of reperfusion. WT, wild-type.

Blood urea nitrogen and histological injury scores are significantly elevated after 24 h of reperfusion in this model (32). We developed an ELISA for measuring murine factor B fragments, and we measured factor B levels in urine samples collected before IRI and 3 h post-IRI (before blood urea nitrogen levels are significantly elevated). Factor B levels were significantly increased in the urine after renal IRI (Fig. 1B). We also measured C4 levels by Western blot analysis in urine collected before and after IRI. Both intact C4 and C4 fragments were increased in the urine after IRI (Fig. 1C), likely reflecting classical pathway activation in the glomerulus.

Patient characteristics.

To determine whether similar increases in urine complement occur in patients with ischemic AKI, we analyzed samples from the TRIBE cardiac surgery cohort. Patients selected for the present study were matched for sex, preoperative eGFR, and CPB utilization (Table 1). Compared with control patients, those who developed AKI were more likely to have had an emergent procedure, a longer perfusion time, and a longer cross-clamp time. The presence of congestive heart failure and diabetes mellitus was lower in the AKI group than in the control group.

Table 1.

Participant characteristics

Label	Value	All	Control	Case	P Value
n		105	70	35
Age at the time of surgery, yr [median (IQR)]		72 (62, 76)	72 (62, 76)	72 (61, 76)	0.14
Sex, n (%)	Female	21 (20)	14 (20)	7 (20)	1.000
	Male	84 (80)	56 (80)	28 (80)
White race, n (%)	Not white	6 (6)	5 (7)	1 (3)	0.40
	White race	99 (94)	65 (93)	34 (97)
Diabetes, n (%)	No	67 (64)	49 (70)	18 (51)	0.04*
	Yes	38 (36)	21 (30)	17 (49)
Hypertension, n (%)	No	23 (22)	19 (27)	4 (11)	0.09
	Yes	82 (78)	51 (73)	31 (89)
Congestive heart failure, n (%)	No	81 (77)	64 (91)	17 (49)	<0.001*
	Yes	24 (23)	6 (9)	18 (51)
Status of procedure, n (%)	Urgent or emergent	15 (14)	6 (9)	9 (26)	0.018*
	elective	90 (86)	64 (91)	26 (74)
Cardiac catheterization in the last 72 h, n (%)	No	99 (94)	67 (96)	32 (91)	0.40
	Yes	6 (6)	3 (4)	3 (9)
Surgery, n (%)	CABG and valve	23 (22)	15 (21)	8 (23)	0.86
	CABG or valve	82 (78)	55 (79)	27 (77)
Perfusion time, min [median (IQR)]		110 (77, 148)	102 (69, 128)	146 (107, 266)	<0.001*
Cardiopulmonary bypass utilization, n (%)	Comb	3 (3)	1 (1)	2 (6)	0.72
	Full	84 (80)	56 (80)	28 (80)
	None	18 (17)	13 (19)	5 (14)
Cross-clamp time, min [median (IQR)]		79 (46, 110)	65.5 (35, 94)	100 (74.5, 156.5)	<0.001*
Preoperative serum creatinine, mg/dl [median (IQR)]		1.1 (0.9, 1.3)	1.1 (0.9, 1.3)	1.1 (0.9, 1.3)	0.90
Preoperative estimated glomerular filtration rate, ml·min−1/1.73 m−2 [median (IQR)]		66 (54, 85)	65 (54, 86)	68 (50, 83)	0.79
Preoperative glomerular filtration rate preoperatiuve categories, n (%)	>60	66 (63)	44 (63)	22 (63)	0.590
	30–60	37 (35)	24 (34)	13 (37)
	≤30	2 (2)	2 (3)

n, number of participants. IQR, interquartile range; CABG, coronary artery bypass graft.

^*P < 0.05.

Ba levels increase in patients who develop AKI after cardiac surgery.

To test whether we could detect complement fragments in the urine of human patients with AKI, we measured several complement fragments in urine samples from an initial cohort of 20 patients. C4d and C3a levels did not increase after cardiac surgery and were not significantly different between case and control subjects (Supplemental Figure S1, available online at https://doi.org/10.6084/m9.figshare.7841318.v1). These analytes were not examined further.

We observed a trend toward increased urine Ba levels in both cases and controls after cardiac surgery in the initial cohort, so we analyzed a larger cohort of samples (Fig. 2 and Supplemental Table S1, available online at https://doi.org/10.6084/m9.figshare.7841372.v1). Urine Ba levels were higher in urine samples from patients who developed AKI compared with controls, although this was not statistically significant until 2 days after cardiac surgery (Fig. 2). Ba levels in patients with AKI 2 days after cardiac surgery were still significantly higher than in controls after correction for urine creatinine, but they were not significantly higher in patients with AKI when corrected for urine albumin (Supplemental Table S1). Urine sC5b-9 levels were increased in both groups 1 day after surgery but did not differ between the two groups of patients (Fig. 2 and Supplemental Table S1).

Fig. 2.

Urine Ba and soluble C5b-9 (sC5b-9) levels after cardiac surgery. Urine samples were collected before cardiac surgery and for the first 3 days after surgery. Ba and sC5b-9 were measured by ELISAs. Ba levels increased after cardiac surgery, and levels were significantly higher in patients who developed acute kidney injury than in controls in day 2 samples (*P < 0.05). Urine sC5b-9 levels also increased after cardiac surgery, but levels in patients with acute kidney injury were not significantly higher than in controls.

Complement proteins can activate on artificial surfaces, and there is evidence that the alternative pathway is activated during CPB (15). Of the 105 participants in this study, 21 participants had off-pump surgeries, and 7 of those patients developed AKI. When we examined the effects of CPB on urine Ba levels, we found that the uncorrected urine Ba levels were not higher in the CPB group compared with the non-CPB group at any of the time points (Table 2). Ba corrected for urine creatinine increased in both groups (CPB and non-CPB) 1 day after cardiac surgery, and levels were significantly higher in patients who underwent CPB on day 1. Among those who received CPB (n = 84), Ba levels on day 2 were significantly higher in patients who developed AKI than in controls (Supplemental Table S2, available online at https://doi.org/10.6084/m9.figshare.8248937.v1). In non-CPB patients, urine Ba levels were not significantly higher in patients with AKI than in controls, although the number of patients in the non-CPB AKI subset was small (n = 7).

Table 2.

Urine Ba levels by CPB utilization and dialysis

		CPB Utilization			Dialysis
Time Point	All	No CPB (n = 21)	Yes CPB (n = 84)	P value	No dialysis (n = 95)	Yes dialysis (n = 10)	P value
Ba, ng/ml
Preoperative	113 (104, 128)	111 (104, 126)	113 (104, 131)	0.74	113 (105, 130)	94.5 (57, 123)	0.25
Day 1	152 (111, 339)	369.29 (744.58)	477.42 (834.85)	0.19	138 (109, 235)	938 (347, 1333)	0.002†
Day 2	186 (129, 358)	223 (122, 346)	183 (129, 373)	0.82	175 (125, 317)	874.5 (433, 1861)	<0.001†
Day 3	312 (159, 805)	358 (155, 1220)	302 (162.5, 736)	0.80	309 (156, 726)	644 (181, 2970)	0.20
Urine creatinine-corrected Ba, ng/mg
Preoperative	1.3 (0.7, 1.9)	1.3 (0.7, 1.8)	1.3 (0.7, 1.9)	0.89	1.3 (0.7, 1.9)	1.1 (0.6, 1.8)	0.30
Day 1	7.0 (2.9, 14.1)	3.2 (1.5, 6.3)	7.5 (3.7, 14.1)	0.02*	6.3 (2.6, 12.3)	21.7 (7.0, 45.5)	0.02†
Day 2	1.8 (1.0, 4.2)	1.9 (1.2, 3.7)	1.8 (1.0, 4.4)	0.86	1.8 (0.9, 3.4)	11.8 (4.3, 15.7)	<0.001†
Day 3	3.0 (1.5, 6.3)	4.4 (1.6, 5.9)	2.6 (1.4, 6.7)	0.68	2.7 (1.4, 5.8)	5.0 (2.3, 20.7)	0.12

Results are presented as medians (interquartile ranges); n, number of participants. CPB, cardiopulmonary bypass.

^*P < 0.05 for CPB utilization P values;

^†P < 0.05 for dialysis P values.

Increases in urine Ba correlate with AKI severity after cardiac surgery.

When we compared the magnitude of the change in Ba with the change in serum creatinine, we found that the Ba level on day 2 was positively correlated with peak serum creatinine within the first 7 days after surgery and with the magnitude of the change in serum creatinine from baseline (Supplemental Table S3, available online at https://doi.org/10.6084/m9.figshare.8248922.v1). We also found that the change in Ba level from baseline to day 2 (ΔBa) correlated with these changes in serum creatinine. When the cases and controls were analyzed separately, the change in Ba levels on day 2 significantly correlated with peak serum creatinine and the change in serum creatinine in both groups of patients (Fig. 3).

Fig. 3.

Urine Ba levels on day 2 correlate with the rise in serum creatinine after cardiac surgery. Urine Ba levels on day 2 after cardiac surgery were compared with the peak increase in serum creatinine from baseline for each patient (delta peak serum creatinine). The Ba level was significantly correlated with the magnitude in the peak change in serum creatinine for both groups of patients (Spearman correlation: 0.35, P < 0.01).

Ten patients in this study required hemodialysis during their hospitalization. Ba levels in the patients on dialysis were much higher within the first 6 h after cardiac surgery than in the AKI group as a whole [median (interquartile range): 938 (347, 1,333) vs. 138 (109, 235) ng/nl, P = 0.002; Table 2]. Ba levels in patients requiring hemodialysis were also significantly higher than in patients with AKI who did not require hemodialysis at the day 2 time point (Fig. 4).

Fig. 4.

Urine Ba levels in patients based on the need for dialysis. Urine Ba levels were compared in patients who required dialysis, patients who developed acute kidney injury but did not require dialysis, and in control patients. Ba levels were significantly higher at all three time points after cardiac surgery in patients who required dialysis during the hospitalization. *P < 0.05 for the comparison between the three groups of patients.

Correlation of urine Ba with other biomarkers of AKI after cardiac surgery.

To analyze the relationship between urine Ba levels and other biomarkers of AKI, we examined the correlation between urine Ba concentrations and levels of urine NGAL and IL-18. Urine Ba levels on day 1 after surgery correlated with urine NGAL levels and IL-18 levels on day 2 (Table 3).

Table 3.

Spearman correlations

	Urine neutrophil gelatinase-associated lipocalin				Urine IL-18
Ba	Preoperative	Day 1	Day 2	Day 3	Preoperative	Day 1	Day 2	Day 3
Preoperative	0.30*	−0.03	0.30*	0.10	0.35*	−0.06	0.12	−0.02
Day 1	0.10	0.75*	0.37*	0.17	0.13	0.70*	0.28*	0.11
Day 2	0.16	0.27	0.75*	0.44*	0.06	0.27	0.55*	0.27
Day 3	0.14	0.12	0.49*	0.56*	0.01	0.11	0.40*	0.46*

^*P < 0.01.

DISCUSSION

Preclinical studies have shown that the complement system is activated in the kidney after ischemia-reperfusion (31, 39). Evaluation of kidney biopsy tissue has also revealed that the alternative pathway is activated in the tubulointerstitium of patients with tubular injury (33). In the present study, we measured the levels of complement activation fragments in urine collected from patients at various time points after cardiac surgery. Although we measured multiple different fragments (Ba, C4d, C3a, and sC5b-9), urine Ba levels showed the greatest change after surgery and was the only one of these complement fragments significantly associated with the development of AKI. The magnitude of the increase in urine Ba at early time points correlated with the severity of AKI as assessed by the magnitude of the increase in serum creatinine or the need for hemodialysis.

Ba is generated during alternative pathway activation. Increased levels of urine Ba in patients with AKI support the observation that the alternative pathway is activated in the kidney after ischemia-reperfusion (31, 33) and is consistent with the pattern of complement fragments detected in the tubulointerstitium and urine of mice with IRI (Fig. 1). Complement activation is also expected to generate C3a and sC5b-9, although we did not see significant increases in these analytes. This may be due to shorter half-lives for these fragments, lower stability in urine, or decreased sensitivity of the detection ELISAs. Urine Ba may be useful as an early biomarker of AKI. The increase in urine Ba was also greater in patients who underwent CPB, suggesting that some of the increase may have been due to activation within the extracorporeal circuit and that the urine Ba may have come from filtration of these fragments. Among patients who received CPB, however, urine Ba levels on day 2 were significantly higher in patients who developed AKI than in controls. Furthermore, the magnitude of Ba increase for all patients with AKI correlated with the degree of subsequent AKI, and the increase in Ba was very large (over 14-fold) in patients who required hemodialysis during the hospitalization.

Changes in urine Ba levels may be useful in several ways. Complement activation in the renal tubulointerstitium probably occurs as a result of cellular injury. Since changes in urine Ba precede a rise in serum creatinine, this may provide an early biomarker of tubulointerstitial renal injury. Furthermore, complement activation contributes to kidney injury after ischemia (31, 39). Complement inhibitors may therefore be useful for preventing or treating AKI, and specific alternative pathway inhibitors have been developed (27, 29). Detection of elevations in urine Ba could be useful as a means of stratifying patients for treatment with a therapeutic complement inhibitor and for monitoring the response to treatment.

Limitations of this study include the nonspecific nature of complement activation and the relatively small number of patients. The complement system is activated in a wide range of inflammatory and infectious diseases (22). Furthermore, preexisting chronic kidney disease and proteinuric kidney diseases may affect baseline urine Ba levels (13, 35). In patients undergoing cardiovascular surgery, CPB and tissue ischemia may both contribute to complement activation and Ba generation. Measurement of urine Ba levels in patients with AKI who have not undergone CPB, such as renal transplant recipients with delayed graft function, may help confirm that CPB does not account for all of the increase in urine Ba seen in these patients. Further studies will be needed to assess the specificity of changes in urine Ba for AKI and to determine whether changes in urine Ba associate with AKI in patients who have preexisting autoimmune diseases or proteinuric kidney diseases, conditions associated with higher baseline urine Ba levels. Additional studies will also be needed to determine whether complement activation fragments increase in the urine of patients with AKI from other causes, such as infections or toxins.

In conclusion, we found that urine Ba levels increase after cardiac surgery and are associated with the development of AKI. Very high levels of urine Ba may identify patients at risk for severe AKI and hemodialysis. Urine Ba may be a useful functional biomarker of complement activation in the damaged kidney, and it may identify those patients most likely to benefit from therapeutic complement inhibition to ameliorate AKI.

GRANTS

This work was supported by National Institutes of Health Grants R01-DK-076690 and R01-DK-113586 (to J. M. Thurman) and RO1-HL-085757 and U01-DK-082185 (to C. R. Parikh).

DISCLOSURES

J. M. Thurman receives royalties from Alexion Pharmaceuticals. J. M. Thurman is also a consultant for AdMIRx, a company developing complement inhibitors. He holds stocks and will receive royalty income from AdMIRx. C. R. Parikh is on the Advisory Board of RenalytixAI and owns equity in the company.

AUTHOR CONTRIBUTIONS

C.R.P. and J.M.T. conceived and designed research; J.L. performed experiments; J.L., H.T.P., C.R.P., and J.M.T. analyzed and interpreted the data; H.T.P. and J.M.T. prepared figures; J.M.T. drafted manuscript; J.L., H.T.P., C.R.P., and J.M.T. edited and revised manuscript; J.L., H.T.P., C.R.P., and J.M.T. approved final version of manuscript.