PDF
Citation
Print
INTRODUCTION
Organ failure is a hallmark of sepsis [1]. Kidney failure frequently occurs in sepsis, and sepsis is the most common etiology of acute kidney injury (AKI) [2]. AKI is not a single disease but a loose collection of syndromes as diverse as sepsis, cardiorenal syndrome, and urinary tract obstruction [3]. Sepsis encompass a spectrum of disorders involving both the heart and kidneys, in which acute or chronic dysfunction in one organ can induce acute or chronic dysfunction in another [4]. Even though the entire phenomenon of sepsis-associated AKI is not fully understood, its long-term adverse outcomes are related to multiple organ failures [5, 6]. Because both sepsis and AKI are independently associated with adverse outcomes, the early detection of AKI is critical to provide opportunities for successful intervention and resuscitation in septic patients.
The early detection of kidney failure by monitoring kidney function using injury and/or stress biomarkers would be crucial to decrease hospital mortality and the duration of hospital stays [7]. However, the current clinical diagnosis of kidney failure relies on markers such as glomerular filtration rate (GFR) based on increased serum creatinine (sCr) levels and decreased urine output [8, 9]. In sepsis, the sequential (sepsis-related) organ failure assessment (SOFA) scoring system has been used to assess organ functions, and the sCr level is used to evaluate kidney function [10]. Changes in sCr levels or urine output are neither sensitive nor specific for AKI; the sCr level is also affected by numerous factors, and its solitary use might not reflect the actual degree of kidney function [3]. Moreover, the inverse relationship between GFR and sCr levels is nonlinear in patients with near-normal kidney function [8]. Therefore, early and reliable biomarkers are required for critically ill septic patients at risk of developing AKI or likely to need clinical treatments, such as renal replacement therapy (RRT) [11].
Proenkephalin A 119–159 (PENK) is a stable surrogate marker for endogenous enkephalins, and it has been suggested to be a functional kidney marker closely related to the iohexol-determined GFR, which is considered the gold standard [12–14]. PENK has been investigated as a novel biomarker for AKI in various clinical settings, including sepsis and heart failure [14–22]. Further, few recent studies showed that high PENK levels are associated with the deterioration of kidney function and adverse clinical outcomes for critically ill patients. However, no study has explored whether PENK levels can substitute for or augment the current role of the SOFA scoring system in septic patients. We explored the clinical utility of PENK for sepsis patients in terms of sepsis severity, organ failure, and mortality. We hypothesized that PENK levels would be related to the renal subscore of the SOFA score, estimated GFR (eGFR), RRT requirement, and short-term mortality in sepsis.
METHODS
Study population and design
A total of 215 patients were enrolled at Konkuk University Medical Center (KUMC), Seoul, Korea, from August 2016 to August 2017. They were diagnosed as having sepsis (N=109, 50.7%) or septic shock (N=106, 49.3%) according to the Sepsis-3 criteria [1]. For all patients, standard-of-care biomarkers, including white blood cells, C-reactive protein, lactate, procalcitonin, and sCr levels, were measured on the day of diagnosis; on the same day, the SOFA score was assessed in six different organ systems with an assigned value from 0 (normal) to 4 (high degree of failure) [10]. All patients received proper standard-of-care treatments. Their medical records were reviewed retrospectively to retrieve demographic, clinical, and laboratory data; the baseline characteristics of the study population are summarized in Table 1.
This study protocol was reviewed and exempt from approval by the Institution Review Board of KUMC (KUH1200085). This registry study required neither study-intended blood sampling nor additional interventions; therefore, the requirement to obtain written informed consent from the patients was waived. Residual EDTA plasma samples that were available on the same day of sepsis diagnosis were collected, split into small aliquots, and then stored at −70°C until use.
Assay
Samples that had been collected from August 2016 to August 2017 were analyzed in September 2017. PENK is known to be stable for at least 2.5 years at −70°C (unpublished data) and for up to three freeze–thaw cycles [23]. The long-term stability of PENK has been tested in a set of patients EDTA plasma samples covering the measurement interval; frozen samples were thawed at room temperature and gently mixed immediately before measuring PENK levels. PENK levels were measured using the sphingotest penKid assay (Sphingotec GmbH, Hennigsdorf, Germany). This is a chemiluminescence sandwich immunoassay using two monoclonal antibodies directed against the middle portion of PENK (anti-PENK 129–144 mAb as a tracer antibody) and the C-terminus of PENK (anti-PENK 152–159 mAb as a capture antibody) [23]. The assay was calibrated using dilutions of synthetic PENK 119–159. Samples/calibrators (50 μL) were pipetted into white polystyrene coated 96-well microtiter plates (Greiner Bio-One International AG, Austria). After adding labeled anti-PENK 129–144 mAb (150 μL), the microtiter plates were incubated for 18 hours at 22°C without agitation. Unbound tracer was removed using washing solution (350 μL per well, five times), and remaining chemiluminescence was measured for 1 second per well using the Centro LB 960 microtiter plate luminescence reader (Berthold Technologies GmbH & Co. KG, Germany). The PENK level was determined using a five-point calibration curve (27.4–2,223 pmol/L). Calibrators and samples were run in duplicate with a required <20% coefficient of variation (CV) between the duplicates. The mean value of duplicates of each sample was used for statistical analysis. The analytical sensitivity (limit of detection) was 17.3 pmol/L (CV=2.7%), and the measurable range was 17.3–2,223 pmol/L. The manufacturer’s recommended reference range at the 99th percentile is 24.6–80 pmol/L, and 80 pmol/L is suggested as a clinical cut-off [24].
Statistical analysis
All continuous variables showed non-parametric distribution; accordingly, the data were expressed as median (interquartile range) or number (percentage). Mann–Whitney test or chi-squared test was used to compare clinical and laboratory variables between the two groups, namely sepsis and septic shock. Kruskal–Wallis test with a post-hoc test was used to compare PENK levels among the groups according to the SOFA renal score (from 0 to 4), the chronic kidney disease epidemiology collaboration (CKD-EPI) eGFR categories (from G1 to G5), and the RRT requirement (no RRT, RRT on day 1, RRT later during hospitalization) [25].
The PENK levels were divided into quartiles, and a chi-squared test or Fisher’s exact test was used to compare the eGFR (eGFR <60 mL/min/kg/1.73 m2), RRT requirement, and the number of organ failures based on the SOFA scoring system (≥2) in each PENK quartile and SOFA renal subscore group; the Cochran-Armitage test for trends was used for trend analysis. Kaplan–Meier survival curves and hazard ratios (HRs) with the 95% confidence interval (CI) were used to compare 30-day mortality in each PENK quartile and for each SOFA renal subscore group; the log-rank test for trends was used for trend analysis. The HR was used to estimate the relative risk of the survival event (survival vs non-survival) in each group and was considered significant when the CI did not include the value 1. A chi-squared test or Fisher’s exact test was also used to compare survival events. These analyses were conducted for all patients, as well as in each group (sepsis and septic shock). MedCalc Statistical Software version 19.1.7 (MedCalc Software Bvba, Ostend, Belgium) and Analyse-it for Microsoft Excel 5.30.2 (Analyse-It Software, Ltd., Leeds, United Kingdom) were used for statistical analyses. Rounding rules were applied to summary statistics, and two-sided P<0.05 were considered statistically significant [26].
RESULTS
Distribution of PENK levels
The PENK levels were significantly higher in the septic shock group than in the sepsis group (118.7 pmol/L vs 75.7 pmol/L, P=0.02; Table 1). The PENK levels were also significantly higher in patients with vasopressor use than in those without vasopressor use (116.9 pmol/L vs 72.7 pmol/L, P=0.007) and in the non-survivors than in the survivors (171.5 pmol/L vs 79.8 pmol/L, P<0.001; data not shown). PENK levels gradually increased according to the increased SOFA renal subscores and CKD-EPI eGFR categories (Fig. 1A and 1B, both P<0.001); this significance was constantly observed across each group (P<0.05, post-hoc test), except for eGFR categories G3a and G3b. Regarding RRT requirements, the PENK levels were significantly higher in patients who required RRT on day 1 (N=3) than in those who did not require RRT (N=193) or required RRT later during hospitalization (N=19) (1,026.8 pmol/L vs. 284.3 pmol/L vs. 88.4 pmol/L, P<0.001, P for trend <0.001); a similar trend was noted for sCr, CKD-EPI eGFR, and SOFA renal scores (all P<0.001; data not shown).
PENK quartiles and SOFA renal subscores for predicting clinical outcomes
The PENK levels were divided into quartiles as follows: Q1<52.5 pmol/L (N=54); 52.5 pmol/L≤Q2<103.0 pmol/L (N=53); 103.0 pmol/L≤Q3<207.5 pmol/L (N=54); Q4 ≥ 207.5 pmol/L (N=54). Tables 2 and 3 show the PENK quartiles and SOFA renal subscores, respectively, to predict clinical outcomes.
The distribution of lower eGFR (<60 mL/min/1.73 m2), RRT requirement, SOFA renal subscore, and the number of organ failures differed significantly according to the PENK quartile. The 30-day mortality rate also differed significantly according to the PENK quartile (P for trend <0.001). Compared with the PENK Q1, higher PENK groups (Q2–Q4) were associated with poor survival outcomes [HR (95% CI) ranging from 4.5 (1.3–15.6) to 12.5 (5.2–29.7)]; this significant finding was consistently observed in each group, namely sepsis and septic shock, as well as in total patients (Table 2). When the clinical cut-off for PENK (80 pmol/L) was applied, the patients with increased PENK levels showed poor survival outcomes than the patients with normal PENK levels [HR (95% CI)=3.0 (1.8–4.9), P<0.001; data not shown].
Regarding SOFA renal subscores, although RRT requirements were significantly different according to each subscore, the number of organ failures did not reach statistical significance. Although no patient received RRT during hospitalization in the PENK Q1 group, four of the 72 patients (5.6%) with a SOFA renal subscore of 0 received RRT during hospitalization. The 30-day mortality rate differed significantly according to the SOFA renal subscore (P for trend=0.005, for event). Compared with the SOFA renal subscore 0 group, the other subscore groups (1–4) were associated with poor survival outcomes based on total patients [HR (95% CI) ranging from 2.1 (0.8–5.3) to 3.5 (1.5–8.6)]. When the patients were divided into sepsis and septic shock groups, however, this significance was observed only in septic patients with SOFA renal subscores of 2 and 3 [HR (95% CI), 3.8 (1.1–12.9) and 8.4 (1.1–66.7)] (Table 3).
DISCUSSION
We demonstrated the clinical utility of PENK to assess organ failure and predict mortality in septic patients. The PENK level was significantly associated with septic shock, vasopressor use, and 30-day mortality; these findings are in line with those of some recent studies [16, 19, 27]. In our data, PENK levels showed significant associations with SOFA renal subscores and CKD-EPI eGFR categories (Fig. 1). It was noteworthy that the PENK level showed more prognostic efficacy than the SOFA renal subscore for predicting the RRT requirement; RRT was required for some patients even with a SOFA renal subscore of 0, although RRT was not required for patients in the PENK Q1 group. A recent study also showed that PENK is an effective predictor of AKI development particularly in septic patients presenting to the emergency department with normal sCr levels [28]. PENK quartiles were also associated with an increasing trend in the number of organ failures (Table 2); this implies that the PENK level might reflect sepsis-induced organ failure simply and objectively and has the clinical potential to substitute for or augment SOFA scores.
Another noticeable finding in our study was the significant association between the PENK quartile and 30-day mortality rate. In all patients, both the PENK quartile and SOFA renal subscore showed significant associations with the 30-day mortality rate (Tables 2 and 3). However, when patients were divided into sepsis and septic shock groups, only the PENK quartile predicted and stratified the 30-day mortality rate consistently, whereas the SOFA renal subscore showed limitations. Compared with PENK Q1, high levels of PENK (Q2–Q4) were related to a higher mortality risk. Additionally, high levels of PENK (Q2–Q4) during sepsis identified patients with a high mortality risk (despite the fact that they did not present with shock), whereas very low levels of PENK (Q1) during septic shock identified patients with a low mortality risk. This finding implies that PENK levels could be useful to stratify patients further even in the same stage of sepsis or septic shock.
The present data also support the use of a clinical cut-off for PENK (80 pmol/L) [15–17]. In the groups with poor clinical status or outcomes, including septic shock, vasopressor use, and non-survivors, the median value of PENK levels was higher than the clinical cut-off. This clinical cut-off was useful to identify patients with increased mortality risk, and likewise for risk stratification by PENK quartile (Q1 vs Q2–Q4). Although its prognostic validity should be supported by further studies, applying a single clinical cut-off value for PENK or an array of such values, including PENK quartiles, would be one of important and interesting topic.
This study has several limitations. This was a small, single-center registry study; accordingly, the severity and mortality of sepsis in this study might be different from those of other population cohorts. Second, we could not keep a strict sampling time and could not define the exact time delay from sepsis diagnosis to sample collection, although it was performed within 24 hours. The in vivo half-life and clearance of PENK are still unknown, and a recent study showed that 24-hour changes in PENK levels are associated with subsequent AKI at 48 hours and seven days [17]. In contrast, sCr cannot reflect real-time GFR. Specifically, sCr peaks on day 4 with a rise of approximately 115% after the AKI episode, showing a considerable time lag between the changes in GFR and the subsequent increase in sCr and recovery [29]. As a biomarker of kidney function in the acute stage, PENK could show significant changes during the first 24 hours after sepsis diagnosis; therefore, observations of delta changes in PENK levels would increase its clinical significance.
In conclusion, we demonstrated that the PENK level is associated with sepsis severity, organ failure, RRT requirement, and short-term mortality in patients with sepsis and septic shock. The PENK quartile seems to be superior to the SOFA renal subscore for predicting and stratifying 30-day mortality in each sepsis and septic shock patient group. PENK could thus be an objective and reliable marker that has the potential to substitute for or augment the current role of the SOFA scoring system in critically ill septic patients.
XML Download