Diagnostic and Prognostic Performance of Plasma Albumin and Cholinesterase in Patients with Sepsis and Septic Shock

• Data regarding the prognostic role of plasma albumin and cholinesterase in the sepsis-3 era is limited.

• In 239 patients with sepsis and septic shock, cholinesterase was associated with moderate diagnostic accuracy for septic shock and risk of 30-day all-cause mortality.

• In contrast, albumin did not predict all-cause mortality.

Sepsis affects approximately 49 million patients every year worldwide and is estimated to account for up to 20% of all deaths [1], especially in patients with septic shock, which is characterized by a profound circulatory and metabolic dysregulation, in-hospital mortality reaches up to 60% [2]. Pathophysiologically, sepsis leads to an imbalance of inflammatory and anti-inflammatory processes [3]. The incidence of sepsis has been shown to increase with age. A significant increase in the incidence of sepsis is postulated as a result of an aging population in Europe and the USA [4].

Patients with sepsis are known to be admitted with abnormal levels of serum albumin and serum cholinesterase (ChE) levels. Serum albumin and ChE were recently shown to be reduced in multi-morbid patients, such as patients with cancer, chronic liver disease or in patients with severe infection [5]. While most studies investigating the impact of the ChE in patients with a sepsis were published prior to the publication of the sepsis-3 definition [6], Peng et al. [7] suggested an increased risk of 30-day mortality in 166 patients with sepsis or septic shock with low ChE levels on admission (i.e., <4000 U/L). However, changes of the ChE during the course of sepsis were beyond the scope of their study. Similarly, low levels of albumin – especially a decline in albumin levels – were shown to be associated with an increase in sepsis-related mortality [8]. However, the identification of high-risk patients with hypalbuminemia needs further investigation, as randomized controlled trials did not demonstrate improved outcomes in septic patients receiving albumin replacement therapy [9].

Although albumin and ChE are easy to measure and less cost-expensive than the investigation of novel biomarkers, their diagnostic and prognostic value in patients with sepsis and septic shock are not clarified yet [10]. The establishment of frequently measured biomarkers may contribute to improved diagnosis and prognosis in patients with sepsis and septic shock. The present study investigates the diagnostic value of the albumin and the ChE for the diagnosis of septic shock, their prognostic value with regard to the 30-day all-cause mortality, as well as the dynamic changes in levels of albumin and the ChE during the course of intensive care unit (ICU) hospitalization.

The present study prospectively included all consecutive patients presenting with sepsis or septic shock on admission. The study is derived from the “Mannheim Registry for Sepsis and Septic Shock” (MARSS-registry), as recently published [11]. For the present study, all consecutive patients with sepsis and septic shock were included. Patients with no evidence of albumin or ChE on sepsis day 1 were excluded from the present study. No further exclusion criteria were applied.

After adequate clotting of blood samples during 1 h at room temperature, samples were centrifuged at 2,000 g for 10 min at 18°C, according to the manufacturer’s recommendation. ChE in serum or LiHep Plasma was determined by fully automated extinction measurement (Atellica CH 930; Siemens Healthineers, Erlangen, Germany). The assay is based on ChE-dependent hydrolysis of butyrylthiocholine into butyrate and thiocholine. The reference range for ChE activity in adults is 7,000–19,000 U/mL, and the detection limit (LoD) of the applied assay method is 1,500 U/mL providing a linear measurement range up to 30,000 U/mL. Albumin measurements were performed fully automatically by extinction measurement based on the color binding method according to Carter and Louderback. The reference range for albumin in adults has been reported to be 34–50 g/L. The LoD of the method is reported by the manufacturer as 6 g/L, with a linearity range of 5–80 g/L. All investigations were carried out in an accredited laboratory under DIN-EN ISO 15189 conditions.

Quantitative data are presented as mean ± standard error of mean, median and interquartile range (IQR), and ranges depending on the distribution of the data. They were compared using the Student’s t test for normally distributed data or the Mann-Whitney U test for nonparametric data. Deviations from a Gaussian distribution were tested by the Kolmogorov-Smirnov test. Qualitative data are presented as absolute and relative frequencies and were compared using the ?² test or the Fisher’s exact test, as appropriate. Box plots for albumin and ChE were created for the comparisons of patients with sepsis and septic shock during the first week of sepsis on days 1, 3, 5, and 7. Spearman’s rank correlation for nonparametric data was used to test the association of albumin and ChE with medical and laboratory parameters.

For C-statistics, receiver operating characteristic analyses were performed within the entire cohort to assess the diagnostic performance of albumin and ChE for diagnosis of septic shock compared to sepsis during the first week of sepsis onset on days 1, 3, 5, and 7. Areas under the curve (AUCs) for diagnostic performance were compared by the method of Hanley and McNeil [12].

C-statistics were applied with calculation of receiver operating characteristic and the corresponding AUC within the entire cohort for 30-day all-cause mortality on days 1, 3, 5, and 7. AUCs for prognostic performance were compared by the method of Hanley and McNeil [12].

Concentrations of albumin and ChE over time (days 1, 2, 3, 5, and 7) were estimated using mixed factorial analyses of variance to estimate the effects of the two factors, time and survival, on biomarker levels. The sphericity was tested by using the Mauchly test and a Huynh-Feldt correction was applied to the mixed analyses of variance results in case of not fulfilling the spherical assumption. The Huynh-Feldt correction was preferred over the Greenhouse-Geisser correction, as the latter tends to increase the risk for Type II error by being more conservative [13]. Kaplan-Meier analyses according to albumin and ChE quartiles were performed within the entire study cohort, as well as separated by patients with sepsis and septic shock on admission. Univariable hazard ratios (HRs) were given together with 95% confidence intervals. Thereafter, multivariable Cox regression models were developed using the “forward selection” option. Results of all statistical tests were considered significant for p = 0.05. SPSS (Version 25; IBM, Armonk, NY, USA) and GraphPad Prism (Version 9; GraphPad Software, San Diego, CA, USA) were used for statistics.

From a total of 361 patients with sepsis or septic shock, 122 were excluded due to lack of data on albumin and/or ChE on day 1. The final study cohort comprised of 239 patients, 58% presented with sepsis and 42% with septic shock. Patients were admitted with a median albumin level of 21.4 g/dL (lowest value 9.2 g/dL) and a median ChE of 5004 U/L (lowest value: 506 U/L). As seen in Table 1 (left panel), patients had a median age of 69 years and most patients were males. The distribution of cardiovascular risk factors (such as arterial hypertension, diabetes mellitus, and hyperlipidemia) did not differ among patients with sepsis or septic shock. In line with this, there were no significant differences between the groups in terms of coronary artery disease, congestive heart failure, atrial fibrillation, and chronic kidney disease. Finally, the rates of patients with LVEF <35% and cardiopulmonary were higher in patients with septic shock.

As illustrated in Table 2, patients with septic shock had higher acute physiology score, acute physiology and chronic health evaluation II (APACHE II) score and SOFA score as compared to patients with sepsis. Specifically, vasopressor support and invasive mechanical ventilation were more common in patients with septic shock. Patients with septic shock presented with higher lactate levels, alongside with higher D-dimer levels and higher international normalized ratios. In contrast, albumin and ChE did not differ in patients with sepsis and septic shock.

Online supplementary Table 1 illustrates the correlation of albumin and ChE on day 1 with clinical and laboratory data. Albumin significantly correlated with hemoglobin (r = 0.245; p = 0.001), platelet count (r = 0.162; p = 0.012), activated partial thromboplastin time (r = -0.253; p = 0.001), C-reactive protein (r = -0.222; p = 0.001), and ChE (r = 0.405; p = 0.001). In line, ChE correlated with bilirubin (r = -0.265; p = 0.001), hemoglobin (r = 0.391; p = 0.001), and the activated partial thromboplastin time (r = -0.329; p = 0.001). In contrast, only ChE correlated with renal replacement days (r = -0.191; p = 0.003), whereas both albumin and ChE did not correlate with catecholamine use (r = -0.018; p = 0.786; r = -0.109; p = 0.092). Finally, both albumin (r = 0.185; p = 0.005) and ChE (r = 0.252; p = 0.001) correlated with body mass index (BMI). Of note, no correlation with either albumin or ChE with sepsis-related scores was observed.

Box plots presenting the distribution of albumin and ChE in the presence of sepsis or septic shock at day 1, 3, 5, and 7 are presented in online supplementary Figure 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000530631). Albumin levels did not differ among patients with sepsis and septic shock on day 1 (median 21.7 g/L vs. 20.5 g/L; p = 0.123), as well as during the first week of ICU treatment until day 7 (p = 0.233). In contrast, ChE levels were significantly lower in patients with septic shock as compared to sepsis on day 3 (median 3,081 U/L vs. 4,492 U/L; p = 0.040) and day 5 (median 2,015 U/L vs. 4,153 U/L; p = 0.010). C-statistics revealed a significant diagnostic AUC of albumin (AUC = 0.575) on day 1, whereas ChE (AUC = 0.530) was not able to discriminate patients with septic shock from those with sepsis on day 1. In contrast, ChE revealed a moderate discriminative value of septic shock on days 3 and 5 (range of AUCs for ChE 0.641 to 0.762; range of AUCs for albumin 0.561 to 0.583; p for AUC differences >0.05) (online suppl. Table 2).

Overall risk of 30-day all-cause mortality was 54%. Median albumin on day 1 was significantly lower in non-survivors as compared to survivors (20.7 g/L [IQR 16.3–24.8 g/L] versus 21.6 g/L [IQR 18.4–26 g/L]; p = 0.033). In contrast, albumin did not differ among 30-day non-survivors and survivors on day 3–7 (p = 0.276) (online suppl. Fig. 2). ChE was lower in 30-day non-survivors on day 1 (median 4,136 U/L [2,817–6,789 U/L] versus 5873 U/L [4,053–7,340 U/L]; p = 0.005), which was still observed on sepsis day 3 (median 3,120 U/L vs. 5,082 U/L; p = 0.001), day 5 (median 3,204 U/L vs. 4,553 U/L; p = 0.004), and day 7 (median 3,176 U/L vs. 4,484 U/L; p = 0.024).

As illustrated in online supplementary Figure 3 (left panel), survival alone had no statistically significant effect on albumin (p = 0.772). An interaction of time with survival was not detected (p = 0.142). In contrast, both survival (p = 0.036) and time (p = 0.001) alone had an effect on ChE during the first 7 days of ICU treatment. The prognostic AUCs of ChE were statistically significant at all time points to predict all-cause mortality at 30 days (range of AUC 0.612–0.686). Of note, prognostic AUCs for albumin were not as consistent as those of the ChE on the evaluated treatment days. Specifically on day 3 to day 7, the prognostic AUCs for ChE were significantly better than those of the albumin (p for AUC differences <0.05) (online suppl. Table 3).

The risk of 30-day all-cause mortality occurred in 58% of the patients with albumin below the median versus 49% of patients with albumin above the median. Accordingly, risk of all-cause mortality was not affected by albumin levels below the median of 21.4 g/L (log rank p = 0.124; HR = 1.307; 95% CI = 0.922–1.853; p = 0.132) (Fig. 1, left panel). No prognostic impact of albumin on 30-day all-cause mortality was observed in both patients presenting with sepsis (48 vs. 39%, log rank p = 0.191; HR = 1.396; 95% CI = 0.841–2.316; p = 0.197) and septic shock (69 vs. 65%, log rank p = 0.987; HR = 0.996; 95% CI = 0.614–1.615; p = 0.988) (Fig. 1, middle and right panel). In contrast, ChE levels below the median were associated with an increased risk of 30-day all-cause mortality (65 vs. 42%, log rank p = 0.001; HR = 1.820; 95% CI = 1.273–2.601; p = 0.001) (Fig. 2, left panel), which was specifically seen in patients with sepsis (59 vs. 31%, log rank p = 0.001; HR = 2.355; 95% CI = 1.397–3.969; p = 0.001), whereas ChE had no prognostic impact on the risk of 30-day all-cause mortality in patients with septic shock (71 vs. 61%, log rank p = 0.399; HR = 1.224; 95% CI = 0.751–1.996; p = 0.418) (Fig. 2, middle and right panel).

After multivariable adjustment, ChE below the median was associated with increased risk of all-cause mortality at 30 days (HR = 1.912; 95% CI = 1.265–2.890; p = 0.002), whereas mortality was not affected by albumin (HR = 1.136; 95% CI = 0.l753 – 1.714; p = 0.545) (Table 3). Furthermore, especially increasing age (HR = 1.018; p = 0.024) was associated with increased risk of 30-day all-cause mortality.

The present study investigates the prognostic impact of albumin and ChE in patients with sepsis or septic shock according to the “sepsis-3” criteria. The main findings can be summarized as follows: the diagnostic performance of ChE has been suggested to be superior to that of albumin in terms of discriminating between patients with septic shock and sepsis. Furthermore, ChE was shown to be associated with moderate predictive accuracy with regard to 30-day all-cause mortality, whereas albumin was not shown to be predictive of the 30-day all-cause mortality. Finally, lower ChE values were associated with an increased risk of 30-day all-cause mortality, which was still observed within a multivariable Cox regression analysis.

In the presence of sepsis or septic shock, both albumin and ChE are inappropriate to reflect the nutritional status of patients, as they may be significantly lower as a consequence of decreased protein synthesis [14, 15]. However, albumin and ChE have been shown to predict short-term outcomes in various clinical conditions, including critically ill patients. While most of these studies were not restricted to patients with sepsis or septic shock [16], they did investigate the prognostic role of ChE and albumin in sepsis predominantly on patients prior to the “sepsis-3” criteria [15]. Albumin has been shown to be a good predictor of short- and long-term outcomes in various settings, such as patients with acute coronary syndrome, acute heart failure, chronic kidney disease, and in older patients [17]. Kim et al. [18] demonstrated that initial levels of albumin were significantly lower among in-hospital non-survivors as compared to survivors including 248 patients with aspiration pneumonia, whereas no further stratification was performed according to dynamic changes of albumin during the course of pneumonia. Fageria and Sharma [8] reported a great decline in albumin levels during the first 10 days of ICU treatment (i.e., 1.07 g/dL vs. 0.75 g/dL) in non-survivors than in survivors among critically ill patients. However, their study was not restricted to patients with sepsis or septic shock. Low albumin levels (i.e., <2.6 g/dL) were recently shown to increase the risk of 30-day all-cause mortality in elderly patients (>65 years of age) with sepsis or septic shock [19]. In the present study which was not restricted to elderly patients, albumin levels did not differ among 30-day non-survivors and survivors during the first week of ICU treatment and albumin had no prognostic impact on 30-day all-cause mortality.

Bahloul et al. [20] demonstrated that the ChE reflects a reliable predictor for the identification of patients with septic shock with a sensitivity of 78% and specificity of 89%, which was even superior to C-reactive protein and procalcitonin. However, in their study, ChE was not found to be associated with all-cause death. The present study confirms these findings with regard to the diagnostic value of the ChE, suggesting a better diagnostic value for the diagnosis of septic shock than the albumin. Furthermore, the current study identifies the ChE as an appropriate tool for the prediction of 30-day all-cause mortality, which was consistent even after multivariable Cox regression analysis.

In patients with sepsis and especially septic shock, multiple mechanisms may contribute to both decrease of albumin and ChE. As a consequence of increased capillary permeability, transcapillary loss is one of the major mechanisms accounting for low albumin and ChE in septic patients, which may be aggravated by dilution effects due to fluid administration [21]. On the other hand, albumin and ChE may be decreased related to impaired synthesis rates in patients with sepsis-induced liver failure due to release of proteases and cytokines by neutrophils [22, 23]. In the present study, the ratio of aspartate to alanine aminotransferase ratio was shown to be increased in non-survivors and associated with higher risk of 30-day all-cause mortality, which may underline the increased prevalence of sepsis-induced liver failure in patients with more advanced stages of septic shock [11]. Furthermore, albumin and ChE levels may be decreased due to catabolism in patients with hypermetabolic and hypercatabolic states [24]. The “cholinergic anti-inflammatory pathway” (CAP) plays an important role in maintaining and modulating the immune response in patients with sepsis and septic shock, and alpha-7 nicotinic acetylcholine receptors have been identified as the key receptor of the CAP [25].

ChE has been shown to reflect non-neuronal cholinergic activity during inflammatory processes [26, 27]. Decrease in levels of ChE was observed after 1–2 h following inflammatory onset [28]. It was shown that the activation of the CAP prevents both the synthesis and release of pro-inflammatory cytokines. We suggest that decrease in levels of ChE seen during the course of ICU treatment in our study may indicate a shift toward pro-inflammatory responses which may be predominantly observed in 30-day non-survivors.

Finally, data focusing on the dynamic changes in ChE levels is scarce. 42–50% lower ChE levels were found among 81 patients with trauma on days 1, 3, and 7 as compared to healthy controls [29]. In line with this, Takegawa et al. [15] investigated the prognostic value of several biomarkers (such as albumin, total cholesterol, and ChE) in 136 patients with sepsis, specifically focusing on patients treated at the ICU for more than 7 days. They demonstrated that decreasing levels of biomarkers were associated with an increased risk of all-cause mortality, but patients were enrolled according to the sepsis-2 criteria, and patients with ICU treatment =7 days were excluded. However, within the present study overall risk of mortality within the first week of sepsis onset was 18% in patients with sepsis and 51% in patients with septic shock suggesting very high risk of early sepsis-related death.

This study has several limitations. Due to the single-center and observational study design, results may be influenced by unmeasured confounding factors that may still be present although we adjusted for potential confounders using multivariable Cox regression. Albumin and ChE were not measured in 34% of the initial study population. However, the baseline characteristics of the patients with evidence of the albumin and the ChE at day 1 did not differ from the excluded patients. Therefore, generalizability may not be significantly affected. The stages of concomitant liver disease, such as Child-Pugh and model for end-stage liver disease score were only available for a minor proportion of the study population and therefore beyond the scope of the present study. Administration of exogenous human albumin was beyond the scope of the present study. Furthermore, transient confounders during course of ICU treatment may specifically affect ChE levels, such as the loss of ascitic or pleural fluid, bleeding, whereas ChE levels may increase by transfusion of fresh frozen plasma, which was not assessed within the present study [30]. With regard to the diagnostic value of albumin and ChE, no control group with healthy individuals was inducted. Finally, the effect of albumin and ChE on long-term outcomes was beyond the scope of the present study.

ChE may be considered as an appropriate tool for the diagnosis of septic shock. It is also able to discriminate between survivors and non-survivors at the 30-day time interval. An increased risk of 30-day all-cause mortality was observed in patients with low ChE, based on multivariable Cox regression analyses. On the other hand, albumin did not have any prognostic impact on patients with sepsis or septic shock.

This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Ethics Committee of Mannheim.

The authors declare that they do not have any conflict of interest.

This study did not receive any funding.

Conceptualization: Tobias Schupp, Katrhin Weidner, Michael Behnes, and Ibrahim Akin; methodology: Tobias Schupp, Julian Müller, Schanas Jawhar, Katrhin Weidner, Jonas Rusnak, Lea Marie Brück, Maximilian Kittel, and Floriana Dulatahu; software: Tobias Schupp, Floriana Dulatahu, Schanas Jawhar, Jan Forner, Jonas Rusnak, Lea Marie Brück, and Maximilian Kittel; validation: Tobias Schupp, Jan Forner, Michael Behnes, Julian Müller, Katrhin Weidner, Jonas Rusnak, and Thomas Bertsch; formal analysis: Floriana Dulatahu, Tobias Schupp, Schanas Jawhar, and Thomas Bertsch; investigation: Tobias Schupp, Katrhin Weidner, Michael Behnes, and Ibrahim Akin; resources: Ursula Hoffmann, Michael Behnes, and Ibrahim Akin; data curation: Katrhin Weidner, Jonas Rusnak, and Thomas Bertsch; preparation of draft: Tobias Schupp; review and editing: Katrhin Weidner, Jonas Rusnak, Lea Marie Brück, Maximilian Kittel, and Ursula Hoffmann; supervision: Michael Behnes, Thomas Bertsch, and Ibrahim Akin; all authors have read and agreed to the published version of the manuscript.

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.