Clinical characteristics and risk factors of organ failure and death in necrotizing pancreatitis

Acute pancreatitis (AP) remains a common gastrointestinal disease, and the number of hospitalizations due to AP has been on the rise over the past decade [1]. According to the 2012 revised Atlanta classification criteria, AP can be classified as mild, moderate and severe [2]. Although most cases of pancreatitis are mild, around 20% will progress to severe pancreatitis with a mortality rate of 25%, characterized by persistent OF beyond 48 h and local complications including peripancreatic or pancreatic necrosis [2, 3]. Cases with necrosis exceeding 30% are classified as necrotizing pancreatitis, which accounts for 5–10% of acute pancreatitis cases [2].

Patients with sterile necrosis generally require symptomatic treatment, and the mortality is approximately 15% [4, 5]. By contrast, patients with infected necrosis usually have significantly increased mortality due to sepsis and multiple organ failure (MOP) [6]. Infected necrotizing pancreatitis (IPN) with surgical intervention has a mortality rate as high as 30%, while without any intervention, the mortality is close to 100% [7, 8]. Although the standard regimen for acute necrotizing pancreatitis (ANP) has gradually changed from early conventional surgery to endoscopic surgery and minimally invasive surgery (non-surgical treatment) over the past 10 years, with the continuous focus on being less invasive, patients with ANP still require considerable healthcare resources and have high mortality [9‐15]. Hence, extensive further research is necessary to determine the clinical traits of this large population affected by necrotizing pancreatitis (NP) and to recognize the risk factors associated with worse outcomes.

Organ failure (OF), is thought to be a significant risk in AP which plays a key role in mortality [16‐18]. A meta-analysis of 1478 patients with AP in 2010 revealed a total of 600 patients who developed OF with mortality as high as 30%, which is close to 10 times higher than in patients without OF [8, 19]. Therefore, an in-depth understanding of OF is crucial for the management of AP patients. Studies showed that specific characteristics of OF affect the clinical course and outcome. In clinical practice, renal, cardiovascular and respiratory failure are most commonly considered in AP [20]. Among these individual types of OF, respiratory failure is the most frequent, and cardiovascular failure leads to the worst outcomes, while the prognosis is significantly worse in case of multi-organ failure [10, 21, 22]. More importantly, the combination of OF and IPN doubles the risk of mortality, which suggests a close link between OF and necrosis [8, 23, 24]. This conjecture has been demonstrated in previous studies as both types of injury depend on the inflammatory response [25]. A prospective study of 104 patients with pancreatic necrosis indicated that the extent of necrosis and infected pancreatic necrosis were the two most significant factors associated with the development of OF [25]. The same conclusion was drawn by Isenmann et al. [26], but other studies reported discrepant observations [27, 28]. Accordingly, the effect of necrosis on OF remains uncertain, and there are also few comprehensive studies on other risk factors for OF in NP.

Taken together, improved insight into the details of OF and mortality will be helpful to the management of NP. Thus, in our research, we addressed the risk factors of OF and death in NP patients.

NP is a potentially life-threatening disease, with an especially poor prognosis in cases of infection and OF secondary to peripancreatic necrosis [8, 23, 24]. According to the determinant-based classification (DBC) of AP, critical acute pancreatitis (CAP), infected pancreatic necrosis (IPN) and persistent organ failure (POF) beyond 48 h are hallmark events of the newly-defined category [33]. Prospective observational studies from different centers reported that CAP may occur in only 2.2–6.6% of AP, while being associated with significant mortality ranging from 17.7 to as high as 87.5% [34]. Although management strategies have changed from early surgery to conservative treatment including internal medicine, endoscopic surgery and minimally invasive surgery, which achieved a better prognosis in NP, a high incidence still remains [11, 13]. For these reasons, increasing knowledge on risk factors for worse outcome of NP is important for clinical practice.

To our knowledge, few studies conducted a detailed analysis of the effects of different factors on OF and outcomes in NP to date. Within this context, data from 432 patients with NP were retrospectively collected from the AP database at the First Affiliated Hospital of Nanchang University. We investigated the clinical outcomes, and assessed the risk factors associated with high mortality and OF in NP patients.

As described in the results section, the present study found that older age was a significant risk factor for death in NP. This is in line with a previous study that confirmed the connection between increasing age and higher AP-related mortality in detail [35]. A possible explanation would be the decreasing resistance to various physiological stresses such as inflammation, infection, and oxidative damage in the elderly [36]. In addition, elderly NP patients are deficient in protective pancreatitis-associated proteins and have a higher production of pro-inflammatory factors in case of systemic inflammatory response syndrome (SIRS) and sepsis compared to younger patients [37‐39]. According to the World Health Organization (WHO), over 13% of the global adult population were considered “obese” (BMI ≥ 30) in 2016, and the growing global obesity rate is alarming due to numerous comorbidities such as hyperlipidemia or osteoarthritis, which are associated with increasing medical expenses [40]. Several previous studies reported that a high BMI was directly linked to increased risk of MOF, local complications and poor prognosis in AP [41‐44]. In our multivariate logistic analysis, higher BMI was independently associated with OF (OR 0.87, 95% CI 0.79–0.95, P = 0.003) which is probably related to the unsaturated fatty acids (UFAs) generated by the lipolysis of visceral fat [45]. However, no significant association was found between higher BMI and death. This was different from a recent study indicating that morbid obesity (BMI > 30 kg/m²) is associated with an increased risk of in-hospital mortality (OR 2.58, 95% CI 1.21–5.53, P = 0.02). We speculated that this discrepancy may be attributable to the different populations used to calculate the mortality, since the former included automatically discharged patients into the death group, while the latter only calculated in-hospital mortality. However, these results for NP were based on only 2 studies, and further studies are necessary for firm conclusions.

As the primary tool used to estimate the severity of pancreatitis, the Acute Physiology and Chronic Health Evaluation II (APACHE II) score consists of age index, acute physiology and chronic health evaluation [46]. A previous study revealed that the APACHE-II score was correlated with the prognosis of SAP patients [47]. In another retrospective study, Kazuhiro Minami et al. compared the clinical features between death and surviving patients with a diagnosis of NP, and determined the role of higher APACHE-II score as independent risk factor of death [36]. Similarly, we also found that the APACHE-II score is an independent risk factor for mortality and OF (OR 1.11, 95% CI 1.00–1.23, P = 0.050) (OR 0.91, 95% CI 0.83–0.99, P = 0.022). In addition, SAP was a robust risk factor for OF in our study, which is in agreement with the fact that OF is one of the crucial characteristics of SAP [2].

In terms of etiology, biliary-induced pancreatitis is considered to be the most significant etiological factor of AP in China, especially in the elderly who have a much larger diameter of the bile duct and higher incidence of gallstones [29, 38, 48‐50]. Despite such a wide base of biliary causes, only one-sixth of the elderly in our NP cohort made the proportion of cases with a biliary etiology rank second, while the most common cause in our study was hyperlipidemia. Several observations in AP patients indicated that hypertriglyceridemia-induced pancreatitis had generally a more severe clinical course, including an increased incidence of OF and pancreatic necrosis [51‐53]. A retrospective cohort study even suggested that elevated serum TGs in AP patients were independently correlated with OF [54]. The more severe outcomes observed in hyperlipidemia are probably caused by the lipolysis of serum triglycerides, which leads to an excess of free fatty acids (FFAs), resulting in acinar injury, endothelial dysfunction and activation of the inflammatory cascade [55‐57]. It therefore stands to reason that the highest proportion of NP etiology in our dataset was HTG (38.0%), as HTG is certainly associated with profound damage. However, although our analysis also noted a higher rate of OF and higher mortality regardless of HTG status, there was no significant difference in terms of NP clinical course between different etiologies according to our data. This finding was consistent with a recent large, single-institution study of 676 patients with NP [58]. One potential explanation for this may be that once a patient with AP develops NP, worst clinical outcomes are mainly explained by other risk factors (such as hemorrhage [59], colon involvement [60], abdominal free fluid [61], low skeletal muscle density [62], etc.) rather than etiology. For this reason, we still do believe that hypertriglyceridemia is an aggravating risk factor for death and OF in AP as well as NP patients, which should be confirmed in the future by more detailed studies.

It used to be generally believed that high mortality was mainly related to concomitant surgical or medical diseases, rather than complications owing to the pathological process of AP [63]. However, our data did not recognize differences in preexisting chronic diseases between NP outcomes, nor did preexisting chronic diseases act as independent risk factors for OF or mortality. This conclusion was consistent with a previous study by Quero et al., who found no strict relation between the Charlson score (an index used to predict the risk of death resulting from comorbid disease) and mortality [64, 65].

Pancreatic necrosis is a common complication which may be categorized as acute necrotic collection (ANC) or walled-off necrosis (WON). The former usually develops within 4 weeks of the pancreatitis course, consisting of peripancreatic effusion and accumulation of pancreatic or peripancreatic necrotic material [66]. Nearly 50% of ANC cases proceed to develop walled-off necrosis (WON) [2, 67]. The collected pancreatic fluid and necrotic material are sterile in the early stages, but it becomes infected in approximately one-third of patients during the course of disease [68]. Our study also found a similar infection rate of 19.5%. Moreover, it seems that the incidence of IPN is generally correlated with the prevalence of OF and even death in acute pancreatitis [8, 26, 69]. In comparison with the sterile group in our study, there was also a higher proportion of infected necrosis both in the OF and death group, with significant differences. Notably, a German prospective cohort study from 1999 comprising 300 NP patients identified infected necrosis as the most significant risk factor for OF [26]. However, multivariate logistic regression analysis failed to confirm infected necrosis as an independent prognostic factor for OF in our study, although there was a statistically significant difference between the two groups. We suspect that the different conclusions between the two studies might mainly stem from changes of diagnostic criteria and treatment strategies over the past 20 years, as well as the uncontrollable nature of prospective studies. On the other hand, we also observed that infection was not independently associated with mortality. The results of a large prospective study support our findings after adjusting for OF [70].

Here, we mainly focused on OF as one of the most important complications of pancreatitis. Consistent with the original hypothesis, the occurrence of OF (i.e. respiratory failure, renal failure and circulatory failure) was associated with an increased risk of death, with respiratory failure ranking first among the proximal causes of mortality. The obviously higher rate of respiratory failure (69.0%) was the major cause. This is in line with a Scottish report, which found that pulmonary dysfunction was most prevalent in AP [21]. Another explanation is likely to involve AP-induced systemic inflammation. A large number of inflammatory factors reach the pulmonary tissue through systemic circulation, leading to impaired alveolar gas exchange, reduced ventilation and increased plasma protein permeability [71‐73]. One consequence of the massive damage to the air-filled aerated lung tissue is the development of severe hypoxemia, which exacerbates systemic dysfunction and further increases the risk of death [74]. Broadly comparable to these earlier studies, MOF was identified as a determinant linked to high mortality in NP patients by Trikudanathan et al. [62]. In contrast with their findings, OF was not identified as an independent risk factor for mortality in our multivariate logistic analysis. However, the included population (only patients aged 18–80 years were selected) and the endpoint (default definition of patients who were automatically discharged from the hospital as dead) in our research may have resulted in selection bias and potential confounding. Thus, there is still reason to believe that OF is a risk factor for death. According to our analysis, increased mortality was related to the number of failed organs, with a ratio of 9/217(4.1%) for one, 30/79(38.0%) for two and 5/6(83.3%) for three. This observation was in agreement with a cohort study of patients with NP from 21 Dutch hospitals [10]. The Dutch study revealed an impact of OF on mortality in necrotizing pancreatitis, but there was no connection between mortality and the duration of OF. However, two other studies reported findings in contrast with the Dutch data [75, 76], as there was a significant relationship between mortality and the duration of OF in these two studies. A possible explanation for this divergence could be the study design, as the latter studies were mostly performed on retrospective data, while the Dutch team conducted a prospective study that may have correctly deduced a causal link between the duration of OF and death in contrast to retrospective studies. Another cause could be data collection itself, which is a long and tedious process including the recording of the start, end and type of OF on a daily basis. It is conceivable that errors in this complex task can easily lead to inconsistencies in the results of analysis. Unfortunately, we did not document the dynamics of OF in detail to examine these different results. Further multicenter prospective studies should be performed to provide more precise data to make up for these shortcomings.

There are several strengths in this study. Firstly, we focused on the worse prognosis related to NP. Broader insights were obtained provided into the clinical characteristics and risk factors of OF and death in necrotizing pancreatitis, and no similar studies have been published to date. We found that MOF, age and shock could potentially be used as simple parameters to identify high-risk mortality. BMI and SAP were identified independent factors of OF in patients with necrotizing pancreatitis, and NP patients with higher APACHE II scores had more severe outcomes. The identification of patient characteristics related to OF and mortality may help clinicians recognize individuals at greatest risk of worse prognosis during hospitalization. Although this was a retrospective study based on data derived from a single center, given that all NP patients over a 5-year duration were consecutively enrolled in our study, it is likely that our observation represents the majority of current clinical experience. However, we do acknowledge some limitations of our study. Firstly, a large number of severe cases were referred to our tertiary hospital, which may have resulted in potential confounding factors related to different initial treatment. Another limitation is that we did not documented the dynamics of OF in detail to solve disputes already raised in the existing literature, such as the impact of the duration of OF on mortality.

Our results suggest that BMI, SAP and APACHE II score are independent risk factors for OF in patients with NP. Furthermore, MOF, age, shock and APACHE II score could potentially be used as a simple index to identify patients at high risk of mortality. However, sex, etiology and past history (pre-existing hypertension, pre-existing diabetes, smoking and drinking) were not associated with differences in mortality or OF. Clinicians should focus on these features to effectively aid in risk stratification of hospitalized patients and allocate intensive care resources accordingly.