Predictive value of serum apolipoprotein A-I in the organ failure of acute pancreatitis: a retrospective cohort study

Abstract Background Organ failure (OF) largely governs the outcomes and mortality in patients with acute pancreatitis (AP), but there is a lack of optimal prognostic biomarker for OF. This study is designed to investigate whether the serum apolipoprotein A-I (Apo A-I) level can predict OF in patients with AP. Methods A total of 424 patients with AP were reviewed in the study, and we finally got 228 patients eligible for analysis. Patients were divided into two groups based on serum Apo A-I level. Demographic information and clinical materials were retrospectively collected. The primary outcome was the occurrence of OF. Univariate and multivariate binary logistic regression were conducted to analyze the relationship between Apo A-I and OF. Additionally, we used receiver operating characteristic analysis to clarify the predictive value of serum Apo A-I level for OF and mortality. Results Ninety-two patients and 136 patients were included in Apo A-I low and non-low groups, respectively. The occurrence of OF was significantly different in the two groups (35.9 vs. 9.6%, p < 0.001). Moreover, serum Apo A-I level markedly decreased across disease severity based on the 2012 Revised Atlanta Classification of AP. The decrease of serum apolipoprotein A-I was an independent risk factor for organ failure (OR: 6.216, 95% CI: 2.610, 14.806, p < 0.001). The area under the curve of serum Apo A-I was 0.828 and 0.889 for OF and mortality of AP, respectively. Conclusions Serum Apo A-I level in the early stage of the disease has a high predictive value for OF of AP.


Introduction
As one of the most prevalent digestive emergencies, acute pancreatitis (AP), is a disease involving acute inflammation of the pancreas. Most patients have a self-limiting course, but also about 20% of patients are susceptible to developing systemic inflammatory response syndrome (SIRS) from localized pancreas inflammation followed by organ failure (OF), with a mortality rate of up to 15-35% [1]. OF largely governs the outcomes and mortality in patients with AP [2]. Therefore, early prediction of the onset of OF in AP is very important. To assess disease severity in AP, there are numerous common multifactorial scoring systems and biochemical markers currently used in clinical practice -which are both inefficient and cumbersome [3].
Apolipoprotein A-I (Apo A-I), a 243 amino acid polypeptide of molecular mass 28,300 Da, accounting for roughly 70% of the total high-density lipoproteins (HDLs) protein content [4]. Apo A-I is the most common apolipoproteins in human plasma, predominantly produced by the liver (80%) and the gut (20%). It plays a critical role in maintaining cholesterol homeostasis and reverse cholesterol transport into the liver [5].
Recent studies have shown that acute and critical disease state can lead to early HDL level changes [6]. It was reported that a substantial reduction in HDL levels in the early serum of sepsis patients has a high correlation with the emergence of multi-organ failure and 28-day mortality [7]. Apo A-I overexpression in animal models attenuates the inflammatory response and improves survival in septic mice [8]. Moreover, the previous studies showed that the low level of HDL was associated with the severity of AP [9,10]. However, the relationship between serum Apo A-I and AP is incompletely defined. Therefore, this study's aim was to determine if the serum Apo A-I level might predict the AP outcomes.

Study design
The study was a retrospective observational cohort study. Patients with AP were hospitalized at the Department of Gastroenterology and Department of Critical Care Medicine of Shanghai Hospital between July 2019 and June 2022. This hospital was affiliated with the Shanghai Jiao Tong University School of Medicine. AP patients had all of their medical records collected, and patients were retrospectively analyzed during hospitalization. The Declaration of Helsinki was followed to conduct this study and our hospital's ethics committee granted its approval for this study. Informed consent was waived due to the retrospective chart review study design. Our research strictly follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement of observational study.

Participants
At least two of the following criteria must be met for AP to be diagnosed: (1) characteristic abdominal pain; (2) serum amylase and/or lipase levels more than the typical upper limit of 3; (3) characteristic AP imaging results. AP was classified into mild acute pancreatitis (MAP), moderately severe acute pancreatitis (MSAP), and severe acute pancreatitis (SAP) based on Revised Atlanta Classification (RAC) [11].

Variables
The primary outcome was the occurrence of OFs, including respiratory, renal, and cardiovascular [12]. The secondary outcomes comprised the occurrence of SIRS, in-hospital death, pancreatic necrosis, and pleural effusion. OF was defined as having a score of at least two for the respiratory (PaO 2 / FiO 2 ≤ 300 mmHg), renal (Scr ≥ 170 μmol/L), and cardiovascular (systolic blood pressure < 90 mmHg, not fluid responsive) organ systems while being hospitalized, in accordance with the modified Marshall score [12]. No fewer than two of the following characteristics were used to define SIRS [13]: white blood cell count <4 × 10 9 /L or >12 × 10 9 /L; temperature <36 or >38 °C; PaCO 2 < 32 mmHg or respiratory rate >20/min; heart rate > 90/min. In our hospital, the normal range of detection values for apolipoprotein A-I is 1.04-2.25 g/L, so we defined 1.04 and 2.25 g/L as the cut-off value. The Apo A-I low group referred to patients with Apo A-I value < 1.04 g/L while the Apo A-I non-low group referred to those with Apo A-I value ≥ 1.04 g/L (including normal level: 1.04 ≤ Apo A-I ≤ 2.25 g/L, high level: Apo A-I > 2.25 g/L).

Data collection
For each patient, age, gender, etiologies, alcohol usage, categorization of disease severity, and body mass index (BMI) at admission was collected as demographic and clinical baseline information. During the first 24 h of hospitalization, a sample of serum lipids was collected and immediate analysis was done. In addition, various biochemical test findings at admission, such as C-reactive protein (CRP), blood urea nitrogen (BUN), serum creatinine (Scr), and calcium were also obtained. In the clinical laboratory's department, commercial kits were utilized to analyze all the routine laboratory tests following manufacturers' guidelines on a 7600 automatic analyzer (Hitachi, Tokyo, Japan). The worst results obtained during the first 24 h of hospitalization were used to calculate the bedside index for severity in AP (BISAP). Contrast-improved CT was performed within 7 days of admission to assess local complications. Senior radiologists reviewed all images after junior radiologists had read them.

Statistical analysis
All continuous variables' normal distributions were examined using the Kolmogorov-Smirnov test. Subsequently, for the variables' skewness, they were summarized utilizing median and (P25, P75). For continuous variables, depending on the variable distribution, the Mann-Whitney U test was utilized. Percentages and frequencies were utilized for describing categorical variables and computed utilizing the chi-square (χ 2 ) test. Logistic regression analysis was conducted to evaluate the risk factors for OF. Multivariate logistic regression only involved the variables that showed statistical significance in univariate analysis (p < 0.10) with forced entry of any factors that were considered to be clinically imbalanced between the two groups at baseline. Employing SPSS 26.0 (SPSS, Chicago, IL, USA), variables were analyzed in addition to receiver operating characteristic (ROC) analysis. The predictive power of lipids, laboratory parameters, and scoring systems for OF and mortality were investigated utilizing the area under the curve (AUC). ROC curves were utilized for investigating the negative likelihood ratio (NLR), cut-off value, positive likelihood ratio (PLR), specificity, and sensitivity. Also, the ideal cut-off point to predict OF or mortality was investigated utilizing the best Youden index. A two-tailed p < 0.05 determined the statistical significance of the results. To calculate the contrasts between ROC curves, Hanley and McNeil's approach was employed with the MedCalc software bvba (18.11.3).

Results
A total of 424 AP patients were initially reviewed, and 228 patients were included in the analysis. All included patients were divided into two groups based on the values of serum Apo A-I. Finally, we got 92 patients in Apo A-I low group and 136 patients in Apo A-I non-low group ( Figure 1). Table 1 displays the demographics and clinical features. This study comprised a total of 228 patients with AP who met our criteria with a mean age of 52.4 ± 14.23 years old, comprising 118 (51.8%) males and 110 (48.2%) females. AP of gallstones-originated was the most common etiology (60.5%), followed by HTG (31.6%), with alcoholic and idiopathic accounting for the remaining 7.9%. Serum Apo A-I level showed no difference among different etiologies of AP ( Figure  S1). In addition, analysis of variance showed an increase in CRP and Calcium in the Apo A-I low group (p < 0.05).

Clinical outcomes in Apo A-I low group and non-low group
The primary and secondary outcomes are shown in Table 1. A total of 46 patients complicated with organ failure, including 33 in Apo A-I low group and one in non-low group, indicating a significant organ failure difference between the two groups (35.9 vs. 9.6%, p < 0.001). For the specific organ system affected, the proportion of respiratory (32.6 vs. 5.9%,  p < 0.001), renal (21.7 vs. 6.6%, p = 0.001), cardiovascular (14.1 vs. 2.9%, p = 0.002) system in Apo A-I low group was significantly higher than that in the non-low group.
In addition, we divided serum Apo A-I level into three ranges (low level, normal level, and high level), the low level group had the worst clinical outcomes, followed by the normal level group (Table 2).

Serum Apo A-I level in patients with AP of different severity
The association between serum Apo A-I level and disease severity of AP are shown in Figure 2. Serum Apo A-I level decreased progressively with the number of OFs (Figure 2(A)). Moreover, serum Apo A-I level was significantly lower in patients who died during hospitalization than in survivors (Figure 2(B)). Serum Apo A-I level markedly decreased across disease severity based on the 2012 Revised Atlanta Classification of AP.

Analysis of the logistic regression and receiver operating characteristic curve
Logistic regression analysis was performed to evaluate the risk factors for OF ( Table 3). The decrease of serum apolipoprotein A-I level is significantly associated with OF (OR: 5.292, 95% CI: 2.594, 10.795, p < 0.001). After adjustment for age, gender, BMI, etiology, CRP, Calcium, the decrease of serum apolipoprotein A-I was an independent risk factor for OF (OR: 6.216, 95% CI: 2.610, 14.806, p < 0.001).
Receiver operating characteristic curve analysis showed a superior predictive ability of serum Apo A-I level for OF with the area under curve (AUC) of 0.828. The optimum cut-off for predicting the OF development was 0.89 g/L. Furthermore, we compared serum Apo A-I level with some other parameters, including CRP, calcium, BISAP score, and found Apo A-I had a better predictive value for OF ( Figure  3(A), Table 4). Similarly, the predictive ability of serum Apo A-I level for mortality could be also identified with an AUC of 0.889, and related results were shown in Figure 3(B) and Table S1.   or: odds ratio; ci: confidence interval; M: male; f: female; apo a-i: apolipoprotein a-i; BMi: body mass index; HtG: hypertriglyceridemia; crP: c-reactive protein; low apo a-i refers to that the concentration of serum apo a-i is lower than 1.04 g/l.

Discussion
In this study, we found that AP patients with low serum apolipoprotein A-I had poor clinical outcomes. Furthermore, the decrease of serum apolipoprotein A-I was an independent risk factor for organ failure in AP. And serum Apo A-I was a sensitive marker to predict and reflect OF in AP. When AP progresses to a severe state, the clinical prognosis of patients becomes worse and the mortality increases [2]. Early risk assessment and prediction of severe disease in patients with AP can improve the clinical prognosis [1]. There are some clinical laboratory indicators and scoring systems, such as Scr, Bun, CRP, and calcium utilized for predicting the AP severity, but more or fewer deficiencies, such as low predictive efficacy and cumbersome clinical application are exhibited by them [14].
Apo A-I is a member of the apoliprotein family of proteins. It's a vital protein that helps in the production of HDL particles. Apo A-I plays an essential role in cholesterol homeostasis by transporting excessive cholesterol from extra-hepatic tissues to the liver [15]. Not only that, Apo A-I has pleiotropic properties, including antioxidant, antiapoptotic, antithrombotic, anti-inflammatory, or anti-infective functions [16]. Apo A-I's anti-inflammatory effects have attracted improving research interest. Chenaud et al. [17] found that a low serum level of Apo A-I at intensive care unit (ICU) admission was associated with an increase in the number of SIRS criteria during the ICU stay. Tsai et al. [18] found that in patients with cirrhosis combined with severe sepsis, serum HDL, and Apo A-I levels had a negative association with a liver reserve and disease severity, and lower Apo A-I levels had an association with significantly impaired effective arterial volume, multiple organ dysfunction syndrome (MODS), and bad prognosis. Murphy et al. [19] developed in vitro and in vivo models of inflammation in which Apo A-I inhibits neutrophil-platelet adhesion, as well as neutrophil spreading and migration and Apo A-I inhibited leukocyte recruitment to endothelial cells.
Our research revealed that AP patients with OF had poorer serum levels of Apo A-I. The following is an explanation of the mechanism: when AP progresses to a severe state, the inflammatory response breaks through the local pancreas, and numerous inflammatory factors including TNF-α and IL-6 are produced in vivo not only to induce SIRS and mediate distant OF [20]; but also to inhibit hepatic production of Apo A-I [21], and to induce hepatic production of certain acute phase proteins. For example, acute-phase protein-secreted phospholipase A2 (Lp-PLA2) is associated with lipoproteins in human plasma and can expedite the breakdown of Apo A-I [22]. Thus, elevated levels of inflammatory factors in AP patients disrupt the synthesis of Apo A-I and increase the breakdown of lipoproteins. In addition, Apo A-I seems to be self-depleted in exerting their anti-inflammatory impacts by inhibiting immune cells' transendothelial transfer, suppressing cytokine production by T cells, regulating monocyte function and differentiation,  interfering with the expression of innate immunity receptors, and inhibiting lipid peroxidation [10]. In recent years, several researchers have conducted studies related to ours. Prospective research by Peng et al. [23] involved 66 patients who were hospitalized in the ICU with predicted SAP. HDL and Apo A-I levels at ICU admission had a negative correlation with disease severity scores and had better discriminatory power than CRP in predicting the occurrence of SAP (AUC 0.904 vs. 0.895). However, the study has limitations because its sample size was only 66 cases and included patients with predicted SAP with acute physiology, age, and chronic health evaluation (APACHE II) score ≥8 admitted to the ICU, but in clinical practice, a certain percentage of patients with SAP or MSAP present with an APACHE II score <8 on arrival; therefore, the findings might not be applied to all patients with the mild disease on arrival and subsequent acute exacerbation. Zhou et al. [24] executed a retrospective examination of 102 AP patients who developed OF during hospitalization and found that HDL and Apo A-I had high predictive values for persistent OF with optimal threshold values of 0.8 mmol/L and 0.8 g/L, respectively. However, they did not record lipids in MAP patients and therefore, the findings could not be applied to MAP patients. Prospective research by Huh et al. [25] involved 191 patients with AP. The Apo B/A-I ratio was positively correlated with the Revised Atlanta Classification, computed tomography severity index, and BISAP score. The ApoB/A-I ratio showed the highest predictive value (AUC 0.706, 95% CI: 0.538, 0.874, p = 0.008) for SAP compared with Apo B or Apo A-I alone. But the specific value for the predictive power of Apo-I was not given.
Compared with previous studies, the present study has certain advantages: (1) a larger and more complete group of patients was included, with AP patients in both ICUs and gastroenterology general wards, reducing selection bias; (2) we focused on lipid levels in patients with AP at the early stage of the disease and observed an association between decreased Apo A-I level and adverse outcomes in AP, particularly OF and death; (3) Lipids are easily obtained in emergency biochemical tests and can be a good supplement to the scoring system to assist clinicians in the early detection of AP patients with a tendency to severe disease. Based on the predictive information, active monitoring, and treatment are used to improve patient prognosis. However, this study has some limitations: (1) we only analyzed lipid levels obtained within 24 h after admission and did not dynamically test whether their subsequent changes affect the clinical outcome of AP; (2) we only quantified Apo A-I and did not study the changes in the structure and function; (3) we did not examine the metabolic enzyme profiles linked to Apo A-I, and the current research could not explain well the reasons for the decline of Apo A-I. We will further investigate this issue in depth to remedy the deficiencies and solve the doubts; (4) The retrospective nature of the study so that the causal relationship between low Apo A-I levels and negative outcomes of AP is unclear. And we excluded 35 patients because lipid values were unavailable. Future prospective, multi-center studies are needed to validate the obtained findings and to further explore the clinical significance of Apo A-I in AP.

Conclusions
In summary, the level of serum Apo A-I in the early stage of the disease has a high predictive value for OF of AP. The lipid metabolism's function in the pathogenesis of pancreatitis is worthy of further study.