## Introduction

Acute pancreatitis (AP) is an acute inflammatory process in which the pancreatic injury may remain localized, spread to nearby tissues, or lead to systemic inflammation through activation of cytokine cascades. The underlying pathophysiology behind the progression of local pancreatic injury to systemic inflammation has not been fully elucidated1. Although most cases of AP are self-limiting, approximately 15% of patients develop serious conditions. Despite improvements in knowledge regarding AP, severe AP (SAP) still has a mortality rate of 2–9%2,3.

Recent studies have demonstrated that metabolic abnormalities, such as diabetes, hypertriglyceridemia, morbid obesity, vitamin D deficiency and the apolipoprotein B to A-I ratio, are closely related to the severity and prognosis of AP4,5,6,7,8. In line with these findings, blood glucose level, which an indicator of ongoing metabolic dysfunction, is used as a criterion in several AP-severity scoring systems such as the Ranson score and the Glasgow-Imrie criteria.

Insulin resistance is defined clinically in terms of the failure of insulin to maintain glucose homeostasis9 and it plays an essential role in the pathogenesis of chronic metabolic disease10,11,12. There has been ample evidence indicating an association between insulin resistance and diabetes, dyslipidemia, and metabolic syndrome (MS), all of which are well-known contributors to the development and severity of acute pancreatitis4,13,14,15. Since insulin resistance is a chronic, low-grade inflammatory status16, insulin resistance has been postulated to play a pathogenic role in other inflammatory diseases such as acute pancreatitis. This hypothesis is supported by reports of pre-existing diabetes increasing the risk of AP development and progression to SAP, as well as the risk of local and systemic complications in AP17,18. However, although one study reported increased insulin resistance in patients with AP19, little is known regarding the association between insulin resistance and the severity of AP. Therefore, the aim of this study was to investigate whether insulin resistance is associated with the prognosis of AP. In the present study, we assessed insulin resistance using the homeostatic model assessment of insulin resistance (HOMA-IR), which is the most widely validated surrogate measure of general insulin resistance20.

## Results

### Patient characteristics

Table 1 shows the baseline characteristics of all patients. A total of 269 patients were enrolled with a mean age of 57.6 ± 18.6 years, and 179 (66.5%) of the patients were male. The etiologies for AP included gallstones (47.6%), alcohol (34.6%), hypertriglyceridemia (5.9%), and idiopathic (11.9%). At the initial evaluation, 99 (36.8%) patients had hypertension, 72 (26.8%) had diabetes mellitus, and the mean body mass index (BMI) was 24.2 ± 4.8 kg/m2. According to the Atlanta classification, 173 (64.3%) patients were classified into the mild group, 79 (29.4%) into the moderately severe group, and 17 (6.3%) into the severe group. The median Ranson, CTSI, and BISAP scores of the patients were 2, 2 and 1, respectively, and the median hospital stay for the patients was 5 days. Among all patients, 34 (12.6%) were admitted to the ICU, and 4 (1.5%) died during treatment.

In the initial laboratory findings, the mean CRP and procalcitonin levels of the patients were 4.5 ± 6.4 mg/dL and 7.3 ± 24.2 ng/mL, respectively. The mean TG level was 242.9 ± 687.7 mg/dL. The mean HbA1c was 6.1 ± 1.3%, and the mean HOMA-IR and HOMA-ß were 4.5 ± 8.8 and 79.6 ± 107.1, respectively.

### Comparison of parameters according to insulin resistance

We divided the patients into the non-IR group (patients with HOMA-IR scores <2.5) and the IR group (patients with HOMA-IR scores ≥ 2.5) and compared the clinical data between the two groups (Table 2). More females were included in the IR group and the mean BMI was higher in the IR group. Mean TG level and CRP after 72 hours were higher in the IR group compared to the non-IR group. The proportion of severe AP according to Atlanta classification was also higher in the IR group compared to the non-IR group (P = 0.021). Additionally, the percentage of ICU admissions and mortality were higher in the IR group.

### HOMA-IR for predicting severe acute pancreatitis

We calculated the area under the curves (AUCs) of HOMA-IR, CTSI, Ranson, and BISAP scores for predicting severe AP using receiver operating characteristic analysis (Table 3). The Ranson score showed the greatest accuracy for prediction of severe AP (AUC = 0.848). The AUC of HOMA-IR for predicting severe AP was 0.719 (95% CI 0.59–0.85, P = 0.003). This value was not notable different from the AUCs of the other scoring systems (Fig. 1). We performed a logistic regression analysis to find risk factors predicting severe AP or ICU admission in patients with AP. IR (HOMA-IR ≥2.5) was the only independent factor for ICU admission (OR 5.95, 95% CI 1.95–18.15, P = 0.002) or severe AP (OR 6.72, 95% CI 1.34–33.62, P = 0.020) (Tables 4 and 5).

## Discussion

We demonstrated that insulin resistance assessed using HOMA-IR was significantly associated with AP severity and ICU admission. We found that this significant association between insulin resistance and severe AP was independent of the presence of diabetes, the body mass index, and the levels of inflammation markers. We also demonstrated that the ability of insulin resistance to predict severe AP was as good as other traditional scoring systems for AP. These findings indicate that insulin resistance might be critical to the pathogenesis of AP.

An important part of the pathophysiology of AP is inflammation of the pancreatic adipose tissue13. MS is a chronic, low-grade inflammatory status characterized by high circulating levels of pro-inflammatory cytokines16. The inflammatory changes that accompany MS may intensify both the immune and non-immune responses that can trigger and exacerbate AP21, which has been confirmed in several investigations showing an increased incidence and severity of AP in patients with MS13,14. The associations among the various components of MS and IR are well documented11, and insulin resistance plays an important role in the development of MS22,23. However, little is known regarding the association between insulin resistance and the prognosis of AP. Therefore, this study aimed to determine the relationship between insulin resistance and the severity of AP. Insulin resistance is defined as a clinical state of decreased sensitivity or responsiveness to insulin24. HOMA-IR has a strong linear correlation with glucose clamp estimates of IR and has been widely used in various prospective clinical trials and clinical research studies25,26. A HOMA-IR score ≥2.5 is the accepted cutoff value as an indicator of IR24,27. To date, there has been a single study demonstrating IR as a risk factor for post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis28. HOMA-IR was an independent predictor of post-ERCP pancreatitis and was used as a considerable factor in predicting the risk of post-ERCP pancreatitis and in decreasing related morbidity28. However, the authors did not demonstrate an association between IR and the severity of the pancreatitis.

Our study illustrates several important and novel findings. First, more females were included in the IR group, and the mean TG and BMI level were higher, compared to the non-IR group. Hypertriglyceridemia is well known in the etiology of AP, and elevated serum TG is independently and proportionally correlated with persistent organ failure, regardless of etiology5. Also, obesity induces a low-grade pro-inflammatory state and is linked with the development of complications in cases of AP6. The number of cases with severe AP according to the Atlanta classification was higher in the IR group. Also, the number of ICU admissions and the mortality rate were higher in the IR group compared to the non-IR group. Second, our ROC analysis found that for predicting severe AP, the AUC of HOMA-IR was not significantly different from that of other scoring systems. This implies that a simple measurement of serum chemistries at the clinical baseline may be able to reliably replace traditional prognostic indices that require multiple clinical measurements. Third, IR (HOMA-IR ≥2.5) was the only independent factor for ICU admission or severe AP in our study, but other factors, including DM, BMI, and CRP, were not significantly associated with either the development of severe AP or ICU admission. This result strongly supports the prognostic value of HOMA-IR in patients with AP and is also in line with the results of previous studies that failed to demonstrate an association between DM and the severity of AP29,30.

Insulin resistance and hyperglycemia, which are hallmarks of DM, are important factors linked to the susceptibility of diabetics to AP31. The existence of links between IR and several pro-inflammatory molecules, such as nuclear factor kB32, tumor necrosis factor α33, amylin34, calcitonin gene-related peptide35, leptin36 and interleukin-637, has been postulated, and these molecules may play a critical role in the pathogenesis of AP in patients with IR. Also, several in vitro studies found that insulin played a protective role against palmitoleic acid-induced AP in rat acinar cells by inhibiting cytosolic calcium overload response38,39 and in L-arginine-induced AP rat models by protecting against oxidative stress as well as contributing to acinar cell regeneration40. Thus, impaired pancreatic β-cell responsiveness and decreases in circulating insulin caused by pancreatic acinar cell exposure to hyperglycemia, which results in oxidative stress, may play important roles in the susceptibility of diabetics to AP. However, the exact mechanism of the association between IR, DM, and AP has not been fully elucidated. Our findings suggest a possible common pathophysiologic pathway for AP in patients with IR and/or DM. Further investigation into this question may explain the apparently conflicting results regarding a correlation between DM and the incidence and severity of AP presented in past studies.

This study has several limitations. First, the number of patients enrolled in this study was small, and this study was performed in a tertiary care center, which could have resulted in the disproportional inclusion of patients with a severe disease status. Such selection bias might have overestimated the predictive value of HOMA-IR. Second, whether the patients had IR before the diagnosis of AP or IR was a result of AP could not be fully examined. If the patients included in this study would have undergone serum sampling prior to the diagnosis of AP, a more complete explanation as to which is the cause and which is the result may have been available. Third, we did not take into account changes in the HOMA-IR score during treatment, which could have varied according to the progression of AP. Despite these limitations, this study also has a number of strengths. This is the first prospective study investigating the predictive value of HOMA-IR in AP and suggesting HOMA-IR as a possible parameter in improving on previously established severity-scoring systems. Replacing the blood glucose level with HOMA-IR in traditional prognostic scoring systems could improve their performance.

## Conclusions

HOMA-IR, a surrogate marker of insulin resistance, was the only independent prognostic factor for predicting either severe AP or ICU admission in patients with AP. This finding suggests that insulin resistance might influence the risk of SAP irrespective of the cause of the pancreatitis. Therapy targeted at decreasing insulin resistance may have promising role in improving overall outcomes of SAP. Further investigation is needed to confirm the possible deleterious effect of insulin resistance on AP prognosis and to investigate the underlying biological mechanisms for this association in greater detail.

## Methods

### Patients

This was a prospective study of patients with AP in Yonsei University Wonju College of Medicine from March 2015 to April 2018. The study protocol was approved by the International Review Board for Human Research (CR315005-002) of Yonsei University Wonju College of Medicine. This study was performed in accordance with relevant guidelines and regulations. Written informed consent was obtained from all patients.

AP was diagnosed based on the presence of two of the following three features41: (1) typical abdominal pain, (2) serum amylase and/or lipase ≥3 times the upper normal limit, and (3) radiologic findings. Laboratory tests on peripheral blood samples, such as hemoglobin, hematocrit, white blood cell count, calcium, phosphorus, blood urea nitrogen, creatinine, lactate dehydrogenase, aspartate aminotransferase, C-reactive protein (CRP, normal range <0.5 mg/dL), and arterial blood gas analysis, were performed at the time of admission. An abdominal computed tomography (CT) scan was performed on all patients upon admission to differentiate AP from other diseases. Once AP was diagnosed, the levels of fasting insulin, glucose, and triglyceride (TG) were verified. The HOMA-IR score was calculated as $$\frac{Glucose\times Insulin}{405}$$ with glucose in mg/dL, and a HOMA-IR score ≥2.5 was taken as an indicator of IR24,27. Additional scoring systems, such as the Ranson score, CT scoring index (CTSI), and BISAP, were applied. The severity of AP was assessed according to the Atlanta 2012 criteria and classified as mild, moderately severe, or severe41. Mild AP was defined by the absence of organ failure (OF) and local or systemic complications. Moderately severe AP was defined as transient OF that resolved within 48 hours and was accompanied by local or systemic complications. Severe AP was defined as persistent OF.

### Statistical analysis

All statistical analyses were performed using SPSS software, version 21.0 (SPSS Inc., Chicago, IL, USA). Categorical variables are presented as the frequency and percentage. Continuous variables are presented as either the mean (±standard deviation) or median with range. The paired t-test was used to compare continuous variables, and the chi-square test was used to compare categorical variables. The odds ratios (ORs) and confidence intervals (CIs) for having severe AP or an intensive care unit (ICU) admission were calculated using multivariable logistic regression analysis after adjustment for confounding variables. Receiver operating characteristic curves were generated to assess the predictive ability of HOMA-IR for severe AP. A P-value < 0.05 was considered statistically significant.