Urine YKL-40 is associated with progressive acute kidney injury or death in hospitalized patients

Acute kidney injury (AKI) is common and appears to be increasing among hospitalized patients [1‐3]. Observational studies demonstrate that even transient AKI is associated with adverse events ranging from increased hospital length of stay, to the development of end-stage renal disease and death [4‐6]. However, the biological mechanisms that underlie the relationship between episodes of transient AKI and outcomes are not well understood.

In patients traditionally categorized as having “pre-renal” AKI, temporary increases in serum creatinine (SCr) concentrations are caused by changes in fluid status and/or renal perfusion. When there is no structural kidney damage in these cases, the diagnosis of AKI may not be causally associated with adverse outcomes. Conversely, a subgroup of patients within the current AKI classification scheme have structural kidney injury and are likely to be at higher risk of significant outcomes directly attributable to their renal diagnosis. Inability to discriminate such a group may have contributed to the lack of success in developing treatments for AKI [7].

We have evaluated the potential diagnostic and prognostic roles of several protein biomarkers measured from blood and urine in different clinical AKI settings [8‐11]. One of the most promising biomarkers from our work and others’ has been neutrophil gelatinase-associated lipocalin (NGAL), a 25 kDa protein expressed at low levels by several cell types but at particularly high levels by cells in the distal nephron during periods of structural kidney injury [12]. While actively examining biomarkers like NGAL, we have also continued to investigate the potential clinical utility for other proteins not previously described in the context of AKI. As such, our group recently showed that the products of the chitinase 3-like-1 gene (called BRP-39 in the mouse and YKL-40 in man) modulate renal repair mechanisms after ischemic kidney injury in mice and can act as effective markers of renal injury in the setting of kidney transplantation in man [13]. By this model, YKL-40 levels should be low in pre-renal AKI but relatively high when overt structural kidney injury triggers repair mechanisms and contributes to clinical outcomes.

Pepe and colleagues described 5 phases of biomarker development for detecting cancer [14]; however, the same categories can be applied to biomarker studies of kidney injury and repair. Accordingly, in this phase 3 biomarker development and proof-of-concept study, we investigated whether urine YKL-40 on the day of AKI diagnosis from any cause can help predict outcome in hospitalized patients. In mouse models, BRP-39 does not rise until 24 hours after a renal insult and peaks on day 3. The delay from renal insult to clinical AKI diagnosis by SCr criteria is typically more than 24 hours, so we postulated that YKL-40 levels could already be discriminatory in urine collected on the day of AKI diagnosis. We also hypothesized that further prognostic information would be provided by a model that incorporates urine levels of both YKL-40 and NGAL, the most predictive AKI biomarker previously measured in this mixed cohort of hospitalized patients.

This is an ancillary study to the previously described hospitalized AKI cohort [8]. Briefly, we prospectively screened all patients aged at least 18 years at Yale-New Haven Hospital between 2008 and 2009 for AKI Network SCr criteria (urine output not used) [15]. Patients were eligible if admitted with at least stage 1 AKI (using outpatient SCr as baseline) or developed at least stage 1 AKI during the hospitalization. We excluded patients with end-stage kidney disease or kidney transplant, those discharged within 24 hours of enrollment, and those with stage 3 AKI at enrollment (as we wanted to detect progression from mild/moderate AKI to more severe outcomes). AKI stage on the day of diagnosis and peak AKI stage during the admission were determined relative to baseline SCr: stage 1, increase in SCr by ≥0.3 mg/dl or 0.5 to <2-fold increase; stage 2, 2 to <3-fold increase; and stage 3, ≥3-fold increase, or SCr ≥4.0 mg/dl after a rise of at least 0.5 mg/dl, or acute dialysis requirement.

Baseline glomerular filtration rate (GFR) was estimated from baseline SCr using the 4-variable Modification of Diet in Renal Disease study equation [16]. Other patient characteristics (including comorbidities) were recorded from the medical histories obtained by admitting/consulting physicians. For descriptive purposes, AKI type was determined by retrospective chart adjudication as previously described [8]. Study physicians considered all available notes and hospital data (e.g., nephrology consult notes and urine electrolytes when available, changes in creatinine and urine output, timing and amount of fluids given, patient disposition, etc.) to classify AKI as acute tubular necrosis (ATN), pre-renal azotemia, or “other”. The primary outcome was a composite of worsened AKI Network stage or in-hospital death. For further details, see our previous publication from this cohort.

We adhered to the Declaration of Helsinki in conducting this study, which was approved by the Yale Institutional Review Board (IRB). Waiver of written consent allowed for the immediate collection of urine (that would have otherwise been discarded) in real time from any patient that met inclusion criteria along with their de-identified hospital information and pre-admission baseline SCr.

This is the first investigation of YKL-40 in urine from injured native human kidneys. Our findings support the hypothesis that urine YKL-40, a “repair-phase” protein, may have utility as a marker of severe kidney injury. The current cohort was more heterogeneous and likely had lower overall structural AKI severity than the cohort of deceased-donor kidney transplants we previously evaluated [13]. In keeping with this, we found urine YKL-40 in hospitalized patients with AKI of any cause to be detectable, but with levels that were markedly lower than those seen in peri-operative kidney transplant recipients. Our analyses also show the potential clinical benefit of considering a biomarker cutoff based on prior experience in complementary settings. We found a modest association between high levels of YKL-40 and AKI progression or death—patients with YKL-40 ≥ 5 ng/ml had a three-fold increase in the odds of this outcome.In addition, adding YKL-40 ≥ 5 ng/ml to our baseline clinical model significantly improved risk reclassification primarily for patients who would not develop the primary outcome. This improvement is clearly seen in Figure 2B, where overwhelmingly more patients without the outcome were correctly reclassified as lower predicted risk rather than higher predicted risk after adding YKL-40 to the clinical model. Furthermore, when combined in a simple step-wise approach with NGAL, the best injury biomarker previously measured in this cohort, YKL-40 ≥ 5 ng/ml significantly partitioned the high-risk group (defined by an NGAL value in the upper quartile) into moderate-risk and very high-risk groups. Taken together, these findings illustrate the concept that incorporating an additional biomarker, which has complementary mechanisms of action, can enhance the utility of other biomarkers or even improve upon clinical prediction models.

BRP-39/YKL-40 is a 39 kDa protein that lacks true chitinolytic activity, like all chitinase-like proteins, and is produced by several cell types including epithelial cells in the airway and colon, chondrocytes, hepatic stellate cells, vascular smooth muscle cells, fibroblasts and differentiated macrophages [21]. Our group recently provided preclinical and translational data regarding BRP-39/YKL-40 expression in the kidney, detectable urine levels, and the physiologic role it plays in limiting tubular cell apoptosis during the repair phase of AKI [13]. This work is an example of ongoing efforts to develop effective biomarker panels to assess renal health and prognosis in different clinical settings. As a proof of concept, the current study suggests that YKL-40 can be non-invasively measured in urine at the first clinical sign of AKI in general hospitalized patients (e.g., a rise in SCr or drop in urine output) and the results used in combination with other biomarkers (e.g., NGAL levels to quantitate distal nephron damage) in order to assess renal injury-repair processes in real-time and potentially improve outcome prediction.

While much AKI biomarker research has appropriately focused on diagnosing AKI earlier than SCr [22‐25], there have been fewer advances in identifying novel biomarkers that can effectively classify patients with AKI as more likely versus less likely to progress [26, 27]. A key application for a tool of this kind may lie in excluding patients at low risk of progression from enrollment into trials that evaluate management strategies initiated immediately after the diagnosis of AKI. Clinicians frequently use the term “pre-renal” AKI to describe these low-risk individuals, but the pre-renal state is typically assigned in retrospect after observing the patient’s response to a fluid challenge. Our current findings suggest that a straightforward combination of kidney injury-repair biomarkers may be useful for prospectively risk-stratifying potential trial subjects. To most effectively assess different treatment strategies in the context of a controlled trial, only participants with significant structural renal injury and at high risk for progression (e.g., high NGAL and high YKL-40 values) should potentially be randomized. As an aside but of particular interest to the AKI field, the misclassification of “pre-renal” patients as having structural AKI may also be diluting or otherwise complicating studies of novel biomarkers to diagnose AKI earlier than SCr [28].

There are some limitations to consider. First, this was an ancillary study to a prospectively collected observational cohort. Nonetheless, the cohort was assembled for the express purpose of evaluating the performance of traditional and novel biomarkers, like YKL-40, as they relate to in-hospital outcomes (i.e., to perform appropriate phase 3 biomarker development studies) [14]. In addition, we followed proper biorepository procedures including validated and blinded biomarker measurements from sample aliquots that had not undergone prior freeze-thaw cycles. The use of a dichotomous cutoff value for a continuous measurement can be considered a limitation; however, the value of 5 ng/ml for YKL-40 was pre-specified based on our previous findings in a separate patient cohort. In addition, sample size limited our ability to evaluate multiple biomarker combinations and comparisons. We therefore controlled for the same confounders and evaluated a simple, representative combination of YKL-40 with the best injury biomarker based on our prior analyses. Lastly, this study was limited by a lack of post-hospitalization information and outcomes given IRB approval with waiver of consent to study the effects of AKI on inpatient outcomes only.

In summary, YKL-40 is a repair-phase protein that is detectable in urine on the first day of clinically apparent AKI and provides only modest prognostic potential on its own. Our findings suggest that there may be utility in combining YKL-40 with other biomarkers like NGAL to refine AKI prognosis and better determine the relationship between injury severity and the degree of repair activation. Further studies are needed to validate the performance of YKL-40 in a large population of patients with AKI and, in particular, to follow trends in this marker of renal injury and repair over the course of disease progression and/or recovery. The kidney is clearly an intricate organ with several cell types that are responsible for multiple homeostatic functions. As we continue to learn more about the complex mechanisms that maintain and repair the renal-urinary system following periods of stress and injury, we envision developing better clinical AKI biomarker panels that provide useful diagnostic and prognostic information relative to those mechanisms. With further insights into the renal responses to acute injury, novel biomarkers like YKL-40 may also begin to give us better information about overall kidney health and could suggest new therapeutic targets for study in humans.

Isaac E. Hall, MD, MS and Edward P. Stern, MBBS are co-first author.