A Sustained Reduction in Serum Cholinesterase Enzyme Activity Predicts Patient Outcome following Sepsis

Early sepsis identification is of paramount importance for an effective therapy and the patient outcome; however, a suitable prognostic biomarker is lacking. Anti-inflammatory nonneuronal cholinergic signaling modulates the magnitude of an immune response. Serum cholinesterase (BChE), an enzyme that hydrolyzes acetylcholine, plays an important role during inflammatory response and serves as an accurate index of cholinergic activity. BChE activity was measured in septic patients using a point-of-care system, and levels of conventional inflammatory markers and the disease severity scores were obtained. We observed a strong, sustained reduction in BChE activity in patients who died within a 90-day observation period, as compared to survivors. Reduced BChE activity when measured at the ICU admission effectively differentiated between the 90-day survivor and the nonsurvivor patient groups. We estimated a critical BChE level of 1.661 kU/L (CI 0.5–0.8, 94% sensitivity, 48% specificity, AUC 0.7) to best predict patient outcome providing a benchmark criterion for early detection of potentially fatal sepsis measured at the admission. This finding suggests that the BChE activity, used in combination with the laboratory tests, clinical examination, and the disease severity scoring, could serve to identify high-risk patients at the ICU admission, the most critical time point in the sepsis treatment.


Introduction
Sepsis, an infection complicated by acute organ dysfunction, is the major cause of death in the intensive care units, estimated to be far over 19 million cases per year worldwide [1]. Various factors, including the primary site of infection, the type of pathogen involved, the severity of the acute organ dysfunction, the primary health status, or the delay until the septic patient received the initial therapy, are decisive factors in the clinical appearance, severity, and prognosis of sepsis.
Despite improved modern critical care therapy including an early treatment initiation, support of failing organs, and treatment of the underlying infection, mortality lies between 20 and 30% in the last decade [2]. Early detection of the disease onset is therefore critical for the recognition of complex cases and high-risk patients who require immediate intensive medical care and appropriate therapy.
Various prognostic and diagnostic biomarkers for sepsis have been proposed in the literature [3]. Nevertheless, none of them show sufficient predictive power to be routinely employed in the clinical practice. Sepsis is an inflammatory response to multiple and extremely diverse factors. Therefore, identifying a single sepsis biomarker remains a challenge for critical care medicine.
Disease severity scoring systems (e.g., APACHE II, SOFA, and SAPS II) represent a quick and powerful complement to the laboratory biomarkers routinely used in the clinical setting [4,5]. Clinical scores are a quick, noninvasive, and reliable assay used for patient stratification and outcome assessment in the ICU [6]. However, most of the scoring systems require a great amount of diverse patient data including laboratory and organ function tests, as well as a detailed patient history. In some cases, not all data is readily available at the most critical time point: the admission of a patient to the ICU with suspected septic onset.
The cholinergic system plays an important role in maintaining and modulating an adequate immune response upon an inflammatory episode. Cholinergic activity, acting extrasynaptically (nonneuronal acetylcholine), has been proposed to be a major mediator of the neuroimmune response to inflammation in a classical feedback reflex manner [7,8]. Cholinergic activity is able to modulate the appropriate magnitude of an immune response upon an emerging inflammatory challenge [9][10][11][12]. The immune response might therefore be affected by the cholinergic system, acting anti-inflammatory.
Serum cholinesterase (butyrylcholinesterase (BChE)) is an enzyme that hydrolyzes acetylcholine [13]. BChE is synthetized in the liver and has conventionally been used as a liver function test. Our previous work, and that of others, has proposed that the enzymatic activity of the BChE might accurately reflect the nonneuronal cholinergic activity during systemic inflammation [14][15][16][17][18][19][20]. By using a point-of-care-testing (POCT) system, we have previously shown a rapid and strong reduction in the BChE activity following an inflammatory challenge [21]. Here, we tested whether the measured BChE activity, obtained from septic patients at the admission to the ICU, might predict the 90-day survival and therefore be used as an early assay for the septic patient outcome.

Materials and Methods
2.1. Study Design. This work represents a secondary analysis of the prospective observational study, powered for the detection of mycoses in patients with septic shock [22]. The study was performed at the intensive care unit of the Heidelberg University Hospital. The Ethics Committee of the Medical Faculty at the Heidelberg University approved this study (file number: S-097/2013). All patients or their legal designees gave informed consent. Patients included in this study were diagnosed septic according to the criteria described in the Surviving Sepsis Campaign: International Guidelines for Managements of Severe Sepsis and Septic Shock 2012 [23]. Exclusion criterion for this study was an extensively disrupted liver function with positive results in three or more of the following tests: aspartate aminotransferase (ASAT) > 100 U/L, alanine aminotransferase (ALAT) > 100 U/L, gamma glutamyl transferase (GGT) > 100 U/L, alkaline phosphatase (AP) > 200 U/L, total bilirubin > 2 mg/dL, and international normalized ratio (INR) > 1.3. Out of 50 recruited patients, 9 patients had documented extensive liver function disruption and were therefore excluded from the study. This resulted in a study population of 41 patients. Basic demographic data, site of infection, and the patient outcome are listed in Table 1 [14]. Enzyme activity is measured in kU/L. The analysis of conventional inflammation biomarkers white blood cell count (WBCC), C-reactive protein (CRP), and procalcitonin (PCT) was conducted according to the standardized protocols of the central laboratory of the clinic. Plasma concentrations of interleukins IL-4, IL-6, and IL-10, tumor necrosis factor alpha (TNF-α), and midrange proadrenomedullin (MR-proADM) were measured according to the manufacturer's instructions as previously described [22]. Disease severity of septic patients was assessed at the ICU by obtaining APACHE II (Acute Physiology and Chronic Health Evaluation II), SOFA (Sequential Organ Failure Assessment), and SAPS II (Simplified Acute Physiology Score) scores. Patients were routinely scored at the above described time points for blood samples.

Statistical Analysis.
The data were electronically gathered and stored by using Excel (Microsoft Corp., Redmond, WA). Data analysis was performed by using GraphPad Prism 6 for Mac (GraphPad Software, La Jolla California, USA, http://www.graphpad.com) and SPSS (version 21.0; SPSS Inc., Chicago, IL, USA). The optimal cutoff value estimation was performed by using Youden index. Data are presented as median with interquartile range (IQR). D'Agostino and Pearson omnibus normality test was used to verify the Gaussian distribution of the study groups. Statistical significance between the patient groups was tested using Mann-Whitney U test. A p value < 0.05 indicated statistical significance.

Results
We set the 90-day survival as a measure of the patient outcome following sepsis detection. Septic patients were divided in two groups: those who survived 90 days (survivors, 25 patients) and patients who did not survive 90 days following sepsis (nonsurvivors, 16 patients). Using the POCT, BChE enzyme activity was measured at the starting time point and on days 1, 2, 7, 14, 21, and 28 following the clinical onset of sepsis. As compared to the previously reported normal range [14], the BChE enzyme activity was markedly reduced in all septic patients during the observation period (p = 0 0001, Mann-Whitney U test), and the measured BChE activity in nonsurvivors was notably lower than that of the surviving patients for the duration of 28 days (p < 0 01, Mann-Whitney U test,  , showed an initial increase in activity, peaking two days after the clinical onset of the sepsis. No statistical difference between the surviving and the nonsurviving patients was observed. Furthermore, we tested the activity pattern of PCT, a sepsis biomarker showing elevated activity during bacterial and fungal but not viral infections. Although increased PCT activity was initially observed in nonsurvivors, no differences were seen at the later time points (Figures 2(a) and 2(b), Supplementary Table 1). Significantly higher activity levels of MR-proADM, a novel sepsis biomarker, were observed in nonsurvivors in later time points starting 21 days following clinical onset of sepsis (p < 0 05, Mann-Whitney U test). However, MR-proADM activity levels could not discriminate between the surviving and the nonsurviving patients, when measured during the initial observation period (Figures 2(c) and 2(d), Supplementary Table 1). The activity pattern of the proinflammatory cytokines IL-6 and TNF-α as well as the anti-inflammatory cytokines IL-4 and IL-10 showed comparable activity patterns without a significant difference between the two patient groups in the initial observation period. The difference in the cytokine activities could be observed in the later time points after the clinical onset of the disease (after 28 days for IL-6, p < 0 05,  Table 2; Mann-Whitney U test). Thus, none of these conventional biomarkers could early predict patient survival at the 90-day period.
Next, we analyzed how well the measured BChE activity predicts the outcome of the septic patients. The 90-day survival of 61% following the clinical onset of the sepsis served as an independent long-term outcome parameter (Figure 3(a)). We observed that measured BChE enzyme activity could predict the survival of the septic patients. The receiver operator characteristic (ROC) curve analysis revealed an area under the curve (AUC) of 0.7. Further analysis rendered an optimal cutoff value of 1.661 kU/L for the measured BChE enzyme activity with 94% sensitivity and 48% specificity (Figure 3(b)). This finding was verified by plotting the Kaplan-Meier survival curves from septic patients with the initial BChE activity greater than the cutoff value and those with the initial BChE activity below the optimal cutoff value (Figure 3(c)). In addition, the ROC curve analysis of the conventional and novel sepsis biomarkers could poorly predict the outcome of the septic patients when measured at the initial time point (Supplementary Figure 3). We further examined the relationship between the patient outcome and the initially measured inflammation parameters (BChE, CRP, WBCC, PCT, MR-proADM, TNF-alpha, IL-4, IL-6, and IL-10) by conducting the U test and the full stepwise logistic regression analysis. Results of the U test (p = 0 04) and the stepwise logistic regression analysis (p = 0 03, likelihood ratio test) identified BChE as a sole independent prognostic factor for the patient outcome. Furthermore, we performed separate stepwise logistic regression analyses for three parameter subgroups. Separate analysis of the first subgroup (BChE, PCT, and MR-proADM) identified BChE enzyme activity as an independent prognostic factor for patient outcome (p = 0 02; likelihood ratio test). IL-10 was identified in the second subgroup (TNF-alpha, IL-10, and CRP) as an independent prognostic factor for patient outcome (p = 0 04; likelihood ratio test). In the third subgroup of factors (IL-4, IL-6, and WBCC), both IL-4 (p = 0 02, likelihood ratio test) and IL-6 (p = 0 045, likelihood ratio test) could be identified as independent prognostic factors for patient outcome.
BChE is synthetized in the liver, so impaired liver function would result in reduced BChE activity. To verify that reduced BChE activity is not caused due to liver dysfunction, we analyzed the liver function and found no impairment in septic patients included in the study (Supplementary  Table 3).
To assess the patient outcome following sepsis, we concurrently calculated the APACHE II and SOFA as well as SAPS II scores (Figure 4

Discussion
The present study highlights the validity and prognostic power of BChE measurements for early detection of high-risk, life-threatening sepsis. Reduced activity of the BChE, when measured at the clinical onset of sepsis, mirrors the activity changes of both conventional (CRP, WBCC, PCT) and novel (IL-4, IL-6, IL-10, TNF-α, MR-proADM) inflammation biomarkers and was validated as a biomarker in the early detection of sepsis by its correlation with several disease severity scores used for patient outcome analysis (APACHE  II, SOFA, and SAPS II). Surprisingly, the BChE assessment at the onset of sepsis proved more effective than these disease severity scores in discriminating between the surviving and the nonsurviving patient groups making it a valuable biomarker for the early detection of high-risk sepsis patients. In accordance with our previous studies, the reduction in BChE activity was observed as early as 1-2 hours following the inflammatory onset [21]. Moreover, a sustained BChE activity reduction, observed in nonsurviving patients, represents a novel red flag warning indicator in the critical care setting, allowing simple and rapid detection of the high-risk septic patients.
Septic patients included in the study showed elevated CRP and WBCC levels, which corresponded to the welldescribed pathophysiology following an immune response [24][25][26][27][28][29][30]. However, the initial measurements of these inflammatory biomarkers failed to distinguish between the survivor and the nonsurvivor patient groups in this study.
The levels of proinflammatory cytokines IL-6 and TNFα, suggested to be effective in sepsis detection and mortality prediction in severe sepsis [31][32][33][34][35][36], showed altered, abnormal levels as expected throughout the observation period; however, these measurements during the initial 48 hours of observations could not discriminate the survivor and the nonsurvivor patient groups. We interpret this finding to be either due to the delayed abnormalities in cytokine peak activity (72 hours) [37], as compared to the earlier change in BChE activity, or due to the rather high variability of the obtained cytokine measurements. Moreover, separate subgroup stepwise logistic regression analysis identified IL-4, IL-6, and IL-10 as independent predictors for patient outcome, when measured at the initial time point. However, this finding could not be observed in other concurrently performed tests (U test, full stepwise regression analysis, Mann-Whitney U test). We interpret this observation to be presumably due to a low number of study patients and a high variability of the obtained interleukin measurements. Conducting a study with a larger population would presumably help overcome the observed discrepancy; however, such high variability suggests that cytokine tests will not provide a reliable biomarker assay for individual patient prognosis.
Our ROC analysis, identifying the optimal BChE cutoff value of the measured BChE enzyme activity at the time point corresponding to the clinical onset of sepsis, suggests that a single bedside BChE measurement could early identify a high-risk patient. Bedside measurement of BChE activity might help reduce the time delay between the initial clinical presentation of the patient with the suspected sepsis and the first therapeutic decisions, when conventional laboratory and diagnostic tests (e.g., chest X-ray and microbiology results) and clinical scores are not readily available. This is highly relevant for ICU clinicians who are often confronted with a fulminant course of the disease, requiring immediate action.
The AUC testing showed a reasonable accuracy for the BChE activity (0.7), which could be explained by the small group size in this study. The optimal cutoff value for the 90-day survival compares to the previously described BChE levels, observed upon the onset of the severe systemic inflammation [14]. Our previous study showed even stronger reduction of the normalized BChE activity upon systemic inflammation than nonnormalised data [21]; however, we did not observe this effect in the present study. Probable reason could be that the exact time point of the clinical diagnosis of the sepsis (and the time point of the initial BChE measurement) differed from the actual time point of the primary inflammation. The missed time interval between the genuine inflammation and the clinical onset of sepsis presumably led to a failed initial BChE measurement and the resulting initial activity reduction. This hypothesis could be further supported by the finding that the initially measured levels of the conventional inflammatory biomarkers (CRP, WBCC, PCT, IL-6, and TNF-α) showed markedly elevated activity, which are regularly described to peak at the earliest 24 hours following the inflammatory onset. This observation might therefore suggest that the proposed optimal cutoff value might be somewhat higher. Further study would be needed to answer this question.
MR-proADM, a peptide with immune modulating, metabolic, and vasoactive properties [38], has been shown to be an effective prognostic marker in sepsis [39] and septic shock [40] with a comparable and even superior prognostic value for sepsis relative to the conventional inflammatory biomarkers (CRP, PCT, and IL-6), as well as to the disease severity scores (APACHE II and SAPS) [39]. We observed a comparable pattern in BChE and MR-proADM activity when compared between the survivor and the nonsurvivor patient groups, further validating the efficacy of the BChE in predicting survival of patients with sepsis.
Disease severity scores are well-established clinical assays for the patient outcome analysis [6,[41][42][43]. In the present study, the difference between the two patient groups was revealed by the score calculations; however, these scores failed to discriminate the survivors from the nonsurvivors at the starting time point. These clinical scores are efficient tools in predicting outcome; however, they require documenting multiple and diverse datasets. The datasets are in most cases readily available; however, in some cases, a particular set of data might not be accessible, delaying or making the scoring impossible. By using a POCT system for a single BChE measurement, the results of an equally efficient outcome assessment tool are readily available at the bedside.
Why do nonsurviving septic patients show continuous and markedly lower levels of the serum cholinesterase enzymatic activity as compared to their surviving counterparts? Nonneuronal cholinergic activity is known to participate in the immune response [44]. The central nervous system (CNS) receives afferents from the immune system via both the humoral and the neural pathways [45,46]. The humoral route uses the active transport of the proinflammatory cytokines through the blood-brain barrier, carrying the information about the commencing inflammatory process in the periphery. The neural immune pathway to the CNS from affected organs prevails when cytokine levels are low, providing early alert information of mild to moderate inflammation. This neural inflammatory signaling, together with pain and injury signals from the periphery, operates at a much lower detection threshold and reaches the CNS much quicker than the humoral pathway [12,47]. The vagus nerve, a cholinergic neuron innervating most of the organs, mediates the neuroimmune response. The extrasynaptic cholinergic activity might therefore play a crucial role in both phases of the neuroimmune response: a quick and effective detection of an inflammatory challenge in the early phase but also in the process of maintaining the equilibrium of the neuroimmune response in the later phase of the severe inflammation. Therefore, the extrasynaptic activity of the cholinergic system might be of great importance during both early and late phases of the inflammatory response.
How does the enzyme activity of the serum cholinesterase relate to the cholinergic activity during an inflammatory episode? Several hypotheses might explain the observed phenomenon. Firstly, cholinergic neurotransmission is continuously active in a feedback response manner, maintaining physiological homeostasis. An inflammatory challenge would immediately cause cholinergic activation, causing a counteracting immune reaction [8]. The neuroimmune loop might act much quicker than the conventional immune reaction [48,49]. The role of the reduced serum cholinesterase might mirror a neuroimmune setting where increased cholinergic activity is required. In the later time point, when the inflammation recedes and the organism recovers, cholinergic activity and the resulting levels of the serum cholinesterase will slowly increase, returning to the baseline. Such homeostatic regulation of BChE would presumably not take place in the continuously challenged and decompensated system of the nonsurvivor, where constantly low levels of the BChE activity mirror a decompensated organism which eventually leads to severe organ failure and death. The question whether low levels of BChE are caused by low levels of the extrasynaptically available enzyme substrate, acetylcholine, or a separate yet unknown mechanism reduces the BChE levels remains an open question. In the presence of low BChE activity, elevated levels of the extrasynaptic cholinergic activity would ensue, producing an antiinflammatory response needed during the severe systemic inflammation. At this moment, it is not possible to provide a plausible answer whether reduced BChE activity results from the presumably reduced level of the nonneuronal acetylcholine or is actually an underlying cause. Further studies would be needed to address these questions.
Since serum cholinesterase is an enzyme synthetized in the liver [50,51], the activity of the BChE is affected by disrupted liver function [52]. A clear limitation for using BChE enzyme activity as a biomarker for potentially fatal sepsis is patients with dysfunctional livers. The results obtained from these patients could not reliably be interpreted. Thus, the authors recommend against using this assay in patients with severe dysfunction of the liver.
A further limitation of this study is the low number of included patients. However, even with the low study sample, the described test demonstrated high sensitivity, particularly in the initial time period, as compared to the benchmark methods, suggesting a rapid, effective, and simple patient outcome assay. A larger, possibly multicenter, study would certainly be needed to validate our findings.
In summary, a single measurement of the BChE enzyme activity may serve as the simplest way to best predict patient outcome when performed at the initial time point. Moreover, the BChE activity correlates with the activity pattern of the conventional inflammatory biomarkers. The observed difference between the 90-day survivors and the nonsurvivors obtained by calculating the disease severity scores, a benchmark tool for the outcome analysis in the clinical setting, further verified the advantage of using the POCT method for a rapid patient outcome screening.

Conclusions
A single bedside measurement of the serum cholinesterase activity might be used to assess the 90-day patient outcome following sepsis. A suggested optimal cutoff value 1.661 kU/ L, when obtained at the time point of the clinical onset of sepsis, might predict patient survival with 94% sensitivity and 48% specificity. This assay provides a rapid and simple bedside test which might be used for an early, simple, and effective identification of high-risk septic patients.