APIP is a rare but critical disease, which could lead to adverse outcome both in pregnant women and fetus [10]. In our study, 66.7% of NMAP patients were admitted into ICU and 61.1% NMAP patients end the pregnancy, with longer hospital stays than MAP patients. As a special type of AP, APIP is differ from other types of AP, owing to anatomical and physiological changes in pregnancy. Despite several literatures having reported the risk factors of APIP patients [11,12,13], there is still lack of study to develop a practical tool to predict the prognosis for APIP patients.
To our knowledge, the unique physiological state of pregnancy could affect bile flow, bile composition, gallbladder contractility, gallbladder postprandially emptying and lipid metabolism, might lead to gallstone formation, cholestasis, hyperlipidemia and so on, which contribute to occurrence of APIP [14]. Our study showed that the etiology of APIP was mainly biliary (37.8%) and hypertriglyceridemia (15.6%). In Europe and America, gallstones are the most etiology of APIP consistent with our study [15]. However, the most common etiology of APIP in several Chinese research cohorts was hypertriglyceridemia [16, 17]. Obviously, though our data primarily originate from China, the etiological composition may still vary across different regions within the same country. In the future, it will be necessary to include populations from more regions and country for analysis. Interestingly, in our cohort, APIP patients with hypertriglyceridemia exhibited a higher proportion of NMAP group than MAP group (85.7% VS 14.3%), but biliary APIP was more common in MAP group than in NMAP group (82.4% VS 17.6%), which might suggest the risk of hypertriglyceridemia in APIP patients. The underlying mechanisms for this observed disparity are likely multifactorial. Pregnancy is characterized by physiological hyperlipidemia and a state of heightened inflammatory readiness [18, 19]. In hypertriglyceridemia-induced APIP, the massive release of serum fatty acid may induce severe pancreatic injury and a potent systemic inflammatory response, resulting in respiratory, kidney, and cardiovascular failure in AP patient [20]. On the other hand, the management of biliary APIP, often involving timely endoscopic intervention to relieve obstruction, may lead to the termination of disease process. It is recommended to pay more attention to the management of hypertriglyceridemia-induced APIP patients.
To date, there are no standardized and special scoring systems to evolute the severity and prognosis of APIP. Computed tomography scan is main technique for AP prognostic estimation, while is unsuitable for pregnant woman. Risk stratification of APIP patients still rely on clinical experience. Previous studies showed that routine laboratory tests were useful predictors in the early assessment of the severity of AP [7, 21]. Several prognostic models have been constructed based on clinical laboratory tests for APIP patients. Tang’s team established a nomogram model for predicting the risk factors of APIP, which contained five indicators including diabetes, triglyceride, Body Mass Index, white blood cell, and C-reactive protein [22]. Yang et al. also constructed a predictive model based on four indicators including lactate dehydrogenase, triglyceride, cholesterol, and albumin [23]. However, these prediction models of APIP require more indicators, simpler and practical tools are still needed. Our nomogram incorporates only two readily available variables including ALB and BUN based on stepwise logistic regression and LASSO regression. This makes our model more accessible for rapid clinical decision-making. Besides, in terms of predictive accuracy, our model achieved an AUC of 0.920, which is competitive with the high AUC of 0.942 reported by Tang et al. and superior to the model by Yang et al. (AUC: 0.865). What’s more, as the ROC curves and calibration curves showed, the model also could effectively predict the probability of pregnant woman admitted ICU (AUC: 0.819). Notably, previous models have rarely been developed to predict ICU admission of APIP patients. In a word, our model not only maintains high predictive accuracy but also excels in simplicity and clinical usability.
There were several scoring systems utilized for the assessment of severity in AP patients, such as Ranson and APACHE-II scoring systems. However, these scoring systems incorporate clinical, laboratory and radiographic data, usually demand at least 48 h to evaluate the severity. And the items of these scoring systems were too complex to be inconvenient for clinicians to use. Besides, BISAP and SIRS score were also used to evaluate severity in AP patients in the first 24 h. In this study, the model only contained two items in assessment of APIP severity and was visualized as nomogram, with robustness and accuracy. Figure 3A also showed that the AUC of nomogram was higher than BISAP score and SIRS score, which indicated nomogram may be more suitable for APIP patients.
It is known that ALB is a plasma protein synthesized by the liver, which plays an important role in maintaining plasma colloid osmotic pressure, transporting substances, etc. Numerous studies have demonstrated that ALB and other serum nutritional biomarkers played a significant role in the disease prognosis prediction including cancer, abdominal sepsis and so on [24,25,26]. Studies reported that the synthesis of ALB usually decreases while patients suffer from AP [27]. Meanwhile, the inflammatory response leads to the rise of capillary permeability, resulting in a large loss of ALB and a decrease in serum ALB levels [28]. In this study, ALB obtained within the first 24 h after admission was found to be an independent risk factor of the severity of APIP. Previous studies also showed that AP patients with low level ALB usually had poor prognosis [29]. Our study showed that ALB exhibited moderate diagnosis values to predict APIP prognosis. Previous studies found that combination of ALB and other laboratory indicators could effectively enhance predictive performance [30, 31]. Present study also showed that the AUC of nomogram incorporating ALB and BUN was higher than single indicator. By integrating these predictors, the model could offer more reliable prediction results. Additionally, ALB could be easily detected from peripheral blood at a low cost, which could contribute to clinical evaluation for APIP. However, it is necessary to recognize that various factors could influence the levels of ALB, such as nutritional status, other complications (liver or kidney) and exogenous ALB [32].
In addition, BUN was also selected in our prognostic model. In general, BUN is related to glomerular filtration and volume status. At the onset of AP, BUN is observed to be ascending because of the decrease of the intravascular volume, fluid loss in body and acute renal injury [33]. The level of BUN has been deemed to be one of the most valuable single routine laboratory tests for predicting mortality in AP, as well as included in BISAP and RANSON scoring systems [33]. Remarkably, BUN is also disturbed by various factors, including protein intake, gastrointestinal bleeding, corticosteroid use and so on, which might lead to interference in disease evaluation [34]. In this study, we identified BUN obtained within the first 24 h after admission as an independent risk factor after multivariate analysis. It implies that BUN could be an effective predictor of APIP, but the influence of other factors should be taken into consideration.
There were several limitations to the present study. First, due to the rarity of APIP, the sample size of present study was small. More clinical centers should participate in statistics in the future, and the model still needs to verify in an external and larger cohort. Second, as a retrospective study, some clinical data was not available. Thus, comparison of other scoring systems cannot be achieved, such as Ranson and APACHE-II score. It is necessary to collect more data in next research.