The incidence of SAP with hemorrhage complications has been reported in the range of 1.2%~14.5% in the existing literature[6]. A domestic study found that the bleeding rate could be as high as 15% in patients with acute necrotizing pancreatitis[17]; while a foreign study showed that 6.2% (28/449) patients had hemorrhages, and the average bleeding time was day 27 of the course of disease[4]. Due to the diversity of clinical manifestations of AP-related bleeding, its incidence shows significant differences; for example, a bleeding rate of 11.5% was reported in 539 cases of AP [5], and another study on SAP indicated that there were 24 (13%) cases of bleeding complications, and the average duration of pancreatitis before bleeding was 27 days [16]. In this study, from 804 cases of SAP, 97 (12.1%) cases developed hemorrhages, and the median bleeding time was 30 days, most of which occurred at later stages of the pancreatitis course, which is consistent with previous studies.
The in-hospital mortality rate of hemorrhage was 46.6% in this study, previous studies have shown that all the hemorrhages were associated with significant mortality rates, regardless of the type of bleeding, and the estimated in-hospital mortality rate for hospitalized patients was 34–52%[18, 19]; however, in a multicenter prospective cohort study[7], after controlling for related confounding factors, it was found that hemorrhagic events did not significantly increase mortality rates, and similarly similar studies concluded that hemorrhage complications themselves had little relationship with mortality rates[4]; while recent studies reported that SAP combined with hemorrhage was associated with poor prognosis, and bleeding increased in-hospital mortality rates[16]. In this study, when we analyzed the effect of hemorrhage on in-hospital mortality rates, we found that the mortality rate of patients with hemorrhage was significantly higher than that of non-bleeding patients (46.6% VS 6.5%, P < 0.001), we believe that hemorrhagic events significantly increase the in-hospital mortality rate of SAP patients. The CTSI score and APACHE II score can judge the outcome of AP at an early stage and predict prognosis[11, 20], in our results, the CTSI score and APACHE II score and complication incidence rate of the bleeding group were significantly higher than those of the non-bleeding group, suggesting that patients in the bleeding group had more severe conditions and worse prognosis, which is similar to the results of a domestic study[21]. Hemorrhagic events are the second hit on patients, especially SAP with organ dysfunction, and hemorrhage itself exacerbates the condition, prolongs hospitalization time, and increases mortality rates. Notably, 41 (46.6%) of these 88 patients with hemorrhage eventually died, compared with the 23 (6.5%) death rate found in this study without the bleeding group, this huge difference in mortality rates indicates that hemorrhage will lead to a bad prognosis, which also indirectly confirms that the more severe the condition of pancreatitis, the more likely it is to cause hemorrhage.
In recent years, with the popularization of minimally invasive step-up strategy, the bleeding complications related to necrotic drainage and endoscopic debridement have increased, making more attention paid to such complications[22, 23]. Our retrospective study analyzed the correlation between endoscopic debridement and bleeding risk, and further explored the relationship between the number and time of debridement with bleeding, previous studies considered multiple necrotic drainage and endoscopic debridement may increase bleeding risk[16, 24], but in our study, although the number of endoscopic debridements was higher in the bleeding group than in the non-bleeding group (20.4%VS15%), however, the number of endoscopic debridements did not affect bleeding risk (P = 0.477), similarly, there was no significant difference in the intervention time of endoscopic debridement between two groups (P = 0.233), we believed that the intervention time of debridement is not an independent risk factor, that is, the timing of debridement intervention does not affect bleeding risk, this conclusion is also consistent with a recent high-quality multicenter study[25]. About endoscopic necrosectomy, although the proportion of endoscopic necrosectomy in the bleeding group was more than that in the no-bleeding group (21.6%VS15.1%), the difference between the two groups was not statistically significant (P = 0.138); moreover, comparing two groups on different entry points of transluminal endoscopic necrosectomy and percutaneous endoscopic necrosectomy did not show significant differences (P > 0.05), which was consistent with the findings of postoperative bleeding after different entry point debridement compared with previous studies in our center[24]; data from this study found that endoscopic debridement and multiple debridements would not increase bleeding risk, both endoscopic debridement and multiple debridements had no obvious correlation with bleeding, which conformed to our clinical experience that multiple debridements do not cause bleeding, which was also consistent with the recent research findings in our center[26].
In the present study, we explored the association between pancreatic necrosis and bleeding risk, and the results indicated that there was a significant difference in pancreatic necrosis between the bleeding group and the nonbleeding group (P < 0.001); especially, the larger the area of pancreatic necrosis in the bleeding group, the higher incidence of bleeding risk increased accordingly, which seemed to indicate that an increase in necrotic area was associated with an increase in bleeding risk (11.4% VS 25.0% VS 59.1%); however, no pancreatic necrotic area was found to be a risk factor for bleeding in univariate and multivariate analyses, which was consistent with the findings of Gupta et al. [16], but different from those of Elhence et al., who considered that extensive necrosis including pancreatic necrosis was a risk factor for bleeding[27], although this study did not confirm a direct correlation between pancreatic necrotic area and bleeding, it indirectly indicated that patients with bleeding had a larger area of pancreatic necrosis and more severe disease. It should be noted that this study mainly focused on SAP patients, which may have affected the interpretation of the results. In addition, Shen et al. found[28] that IPN was associated with increased bleeding risk, and Labarca et al.’s study also pointed out that IPN was more likely to be complicated by bleeding[7]. In this retrospective study, the incidence of IPN in the bleeding group was significantly higher than that in the non-bleeding group (80.7%VS18.2%, P < 0.001); multivariate analysis showed that IPN was an independent risk factor for bleeding (OR: 4.407, 95%CI: 1.854–10.476, P < 0.001). Among IPN, bleeding events were more common, probably because infection necrosis increased the possibility of causing vascular injury, and infection and bacterial metalloproteinase-induced local tissue and vessel wall degradation in infected necrotic areas might further increase the risk of bleeding[21, 27].
The scoring system reflects the severity of the disease, and a higher score indicates the severity of the patient's condition, reflecting severe damage to the pancreatic structure, with increased bleeding risk from the disorder of neo-granulation tissue and blood vessels[4, 6]. Previous studies have pointed out that the development of hemorrhage in SAP is related to the severity of the disease[5, 6]; In this study, the CTSI score was significantly higher in the bleeding group than in the non-bleeding group (10VS6, P < 0.001), and also the proportion of APACHE II score ≥ 15 in the bleeding group was higher than that in the no bleeding group (80.7%VS14.8%, P < 0.001), suggesting that patients in the bleeding group had more severe conditions than those in the non-bleeding group, and worse prognosis; Subsequent multivariate analysis showed that CTSI score and APACHE II score ≥ 15 were independent risk factors for hemorrhage in SAP patients (OR: 1.771, 95% CI: 1.434–2.187, P < 0.001), (OR: 12.323, 95% CI: 5.015–30.282, P < 0.001). When investigating the risk factors for SAP combined with gastrointestinal bleeding, it was also found that APACHE II score and CTSI score were independent risk factors[13], in addition, relevant studies confirmed that CTSI score was significantly correlated with abdominal bleeding[12], and similar significant correlation between APACHE II score and SAP combined with intra-abdominal bleeding has also been reported[21]. Therefore, when the CTSI score and APACHE II score are elevated, indicating that SAP patients have severe conditions and increased bleeding risk, which requires special attention from clinicians.
Reviewing the past studies of our center on 145 cases of endoscopic debridement, it was pointed out that bleeding risk during endoscopic debridement is related to renal failure and it was observed that renal failure may increase the risk of bleeding[24]; however, in our present study, although univariate logistic analysis considered a renal failure and CRRT as influencing factors, multivariate logistic regression analysis did not consider renal failure as an independent influencing factor, but CRRT was identified as an independent influencing factor for bleeding (OR: 0.251, 95% CI: 0.094–0.670, P = 0.006). Interestingly, although CRRT may be considered a potential risk factor for bleeding, our combined prediction model showed that CRRT was negatively correlated with bleeding (OR: 0.251 < 1), indicating that it is a protective factor for reducing the risk of bleeding. When SAP is complicated by renal dysfunction, patients often have coagulation abnormalities[8], traditionally we believe that bleeding is related to coagulation abnormalities, also in this study, there were significant differences in PT, and APTT between the bleeding group and non-bleeding group, but coagulation abnormalities were not found to be influencing factors for bleeding in the following univariate and multivariate analyses (P > 0.05), considering that coagulation abnormalities are common clinical features of SAP, it may be that these factors are all confused by the severity of pancreatitis, the more severe the pancreatitis, the more likely to have coagulation abnormalities. There are studies reported that thrombocytopenia significantly increases the risk of bleeding[29], but in this study, there was no significant difference between the two groups of PLT (P > 0.05), and our results did not confirm this, instead, the PLT value of the bleeding group was higher than that of the non-bleeding group (223VS190), we believe it may be related to the bleeding that makes PLT compensate for elevation. When SAP is complicated by multiple organ dysfunction, deposition of substances such as urea nitrogen, and creatinine, and release of various inflammatory mediators on the microvascular wall will cause endothelial injury and increased vascular permeability, while coagulation abnormalities are complicated with microcirculatory disorders, thereby increasing the risk of bleeding in SAP patients[8]; application of CRRT can effectively clear these damaged blood vessels from substances, alleviate inflammation, improve microcirculatory disorders and thus reduce the risk of bleeding. In addition, in AP caused by hypertriglyceridemia, CRRT can rapidly reduce blood lipid levels and prevent deterioration[30]. Therefore, CRRT can not only improve the condition of SAP but also reduce the risk of bleeding.
ACS is one of the serious complications in SAP patients. Due to the extensive inflammatory necrosis, edema swelling, and massive exudation of the pancreas in SAP patients, which caused the increase of intra-abdominal pressure, when the intra-abdominal pressure was persistent > 20 mmHg with organ dysfunction, it led to ACS [15]. Previous studies have confirmed that ACS was associated with higher APACHE II score and multiple organ dysfunction syndrome (MODS), and MODS not only included respiratory or renal system but also hemodynamic instability, etc., and its bleeding risk increased accordingly with the development of disease[31, 32]; in this study, compared with the non-bleeding group, the incidence of ACS was significantly increased in the bleeding group (34.1%VS7.4%, P < 0.001), multivariate analysis indicated that ACS was an independent risk factor for bleeding combined with SAP (OR: 3.685, 95% CI: 1.462–9.287, P = 0.006); although there were few related kinds of literature, this finding emphasized the importance of ACS combined with SAP because it was significantly associated with poor prognosis[33]. Therefore, for SAP combined with ACS, appropriate decompression treatment such as PCD should be adopted to reduce intra-abdominal pressure and improve the condition[15]. The analysis of this study pointed out that the bleeding group received more PCD than the nonbleeding group before bleeding (52.3%VS25.9%, P < 0.001), but the results of multivariate analysis showed that PCD had no direct association with bleeding, which illustrated that PCD might be an effective decompression treatment but did not directly affect the bleeding risk[34].
The results of subgroup analysis showed that there were no significant differences in baseline including comorbidities, mortality rate between the two groups (P > 0.05); while further comparing the details of pre-bleeding intervention and prognosis between the two groups, it was found that the abdominal bleeding group had more endoscopic debridement intervention (23.6%VS18.2%) than the gastrointestinal bleeding group, the number of people receiving PCD was significantly higher than that of gastrointestinal bleeding group (63.6%VS33.3%, P = 0.006), the number of people requiring surgical intervention was significantly higher than that of gastrointestinal bleeding group (29.1%VS9.1%, P = 0.027), which is similar to a foreign study reported [16], we believe that more surgical intervention is needed for hemostasis after intra-abdominal bleeding; The interval time between rebleeding in abdominal bleeding group was significantly shorter than that in gastrointestinal bleeding group (4VS24, P = 0.031), indicating that abdominal bleeding is more likely to rebleed again in a short time than gastrointestinal bleeding; Considering the own condition of SAP patients with abdominal bleeding, there are more invasive interventions, reflecting the destruction of the intact structure of the pancreas, and the confusion of new granulation tissue and blood vessels may increase the risk of rebleeding.
The strength of this study is that it is the largest retrospective analysis to date containing a large number of cases with SAP bleeding complications; however, we are also aware of some limitations of the study: First, there may be selection bias due to the single-center retrospective nature of the study. Second, because not all bleeding cases were included, it may not have fully captured all clinically relevant factors and did not consider the influence of medications given during treatment, which may have affected our findings. Third, although we used propensity score matching methods to balance baseline differences between the bleeding group and no bleeding group and reduce confounding effects, it was still impossible to completely exclude the potential impact of all unmeasurable or unconsidered factors[35]. Finally, the results of this study may not apply to all AP patients because the study only involved SAP patients and did not consider recurrent and mild AP.