第一部分 1型自身免疫性胰腺炎胰周血管受累及合并糖尿病的临床特征及危险因素分析

背景与目的：自身免疫性胰腺炎（autoimmune pancreatitis, AIP）是一种自身免疫机制参与特殊类型的胰腺炎，约占慢性胰腺炎的5%到6%，1型AIP是IgG4相关疾病（immunoglobulin G4-related disease, IgG4-RD）的胰腺表现。胰腺周围血管受累（peripancreatic vascular involvement, PVI）是1型AIP特征性的临床表现之一，由于恶性肿瘤也具有血管侵袭的生物学行为，这增加了临床上在鉴别AIP与胰腺恶性肿瘤的困难程度。然而，既往关注PVI的报道很少。本研究希望通过回顾分析1型AIP患者的临床特征，寻找与PVI相关的重要因素。同时，由于胰腺为本病受累器官，临床上也常发现1型AIP患者伴糖耐量受损或糖尿病（diabetes mellitus, DM），糖尿病可以是AIP的首发也是唯一的临床表现；且对AIP的激素治疗的反应不一，预后亦存在差异。然而，关于针对AIP的激素治疗是否能改善或恶化DM的证据仍不充分。在这种背景下，本研究希望通过临床特征的描述和分析，探究可区分AIP诊断时伴有新发DM与既往固有DM的特征性差异，并初步评估激素治疗对糖尿病预后的影响。

方法：本研究回顾性了2013年1月至2018年12月期间在北京协和医院消化内科住院治疗的1型AIP患者共101例，从病历系统中提取患者的详细信息，统计人口学、临床表现、实验室检查及影像学检查等临床资料。将所有患者分为非PVI组和PVI组，通过临床资料分析探究AIP患者的PVI的潜在危险因素。同时又根据有无合并DM、AIP和DM诊断的时间顺序将患者分为三组：既往固有DM（pre-existing DM, pDM）、新发DM（concurrent DM, cDM）和非DM（non-DM, nDM），通过三组间的临床资料特征分析，寻找与AIP患者新发DM相关的临床因素。另外，通过评估DM对激素治疗的反应，初步探究AIP合并的DM在治疗后改善与否的预测因素。

结果：在101例1型AIP患者中，52例（51.5%）伴有PVI，男女比例为5.5:1，平均年龄58.37±8.68岁。与非PVI组相比，PVI组胰尾部受累的比例更高（80.8% vs 57.1%，P = 0.010），脾大发生率更高（28.8% vs 4.1%，P = 0.001），外周血白细胞（WBC）计数水平更低（5.96±1.54×10⁹ vs 6.80±2.01×10⁹，P = 0.020）。进一步多因素分析发现，胰尾部受累（OR, 3.06; 95%置信区间, 1.17-8.02；P = 0.023）和脾大（OR, 8.23；95%置信区间, 1.67-40.51；P = 0.010）均为1型AIP患者出现PVI的危险因素。PVI组中有25例患者接受了激素治疗并进行了不少于6个月的放射学随访，其中18例患者PVI较病初改善。

101例1型AIP患者中，共计52例（51.5%）患者在诊断AIP时合并有糖尿病，其中，cDM组有36例患者，占DM组主要部分（69.2%）。cDM胰腺弥漫肿大（弥漫型）的发生率显著高于pDM组（72.2% vs. 37.5%，P = 0.018）和nDM组（72.2% vs. 51.0%，P = 0.049）。同时，cDM组胰腺体/尾部受累的比例也明显高于pDM组（91.7% vs 56.3%，P = 0.009）和nDM组（91.7% vs 69.4%，P = 0.013）。48/52的AIP合并DM患者接受了激素治疗。通过对比DM改善组与未改善组的临床特征，并进行单因素及进一步的多因素分析发现，血清GGT水平升高（OR, 0.032; 95%置信区间, 0.003-0.412；P = 0.008）和激素治疗后胰腺萎缩（OR, 0.027; 95%置信区间, 0.003-0.295；P = 0.003）两个因素均为降低激素治疗后DM改善可能性的独立预测因素。

结论：1型AIP患者合并PVI和DM的发生率均较高，且与部分临床特征具有相关性。临床上应关注1型AIP患者是否出现以上并发症及其预后。部分患者受累的胰周血管病变经过激素治疗后可以好转甚至消失，故需注意血管受累在AIP与恶性肿瘤之间的不同预后。激素对部分诊断时合并DM的1型AIP患者的糖尿病血糖控制情况产生了有益影响，也在部分患者身上引起了类固醇相关DM，故治疗上应注意评估激素治疗的双重影响，选择有针对性的个体化治疗。

第二部分 1型自身免疫性胰腺炎相关趋化因子谱初步描述

背景与目的：IgG4-RD是一种免疫介导性疾病，以纤维炎性病变为特征的疾病，其病变可出现在唾液/泪腺、眶、胰腺和胆道、主动脉、肾脏、脑膜和甲状腺等器官。1型AIP是IgG4-RD中胰腺受累的表现形式。血清IgG4升高可对诊断1型AIP及IgG4-RD提供重要的提示，但仅凭借单一指标难以进行临床诊断或适当的疾病分类，IgG4-RD的诊断仍需要包括组织学、影像学和血清学等全面的检查，并排除潜在的恶性肿瘤、感染或其他免疫介导性疾病可能。与此同时，随着对疾病病理生理学理解的深入和高通量诊断平台的可用性，过去几年中已经调查并描述了许多新的血液生物标志物在IgG4-RD中的应用。趋化因子(chemotactic factors)是一类小分子细胞因子或信号蛋白，既往研究在包括类风湿性关节炎、系统性红斑狼疮、系统性硬化症等许多人类自身免疫性疾病中均观察到高水平的趋化因子表达，趋化因子及其受体被认为能诱导免疫细胞进入这些疾病的受累器官。CCL17、CCL26、CXCL12、CXCL13等都曾被报道过与IgG4-RD相关。然而，既往研究多局限于一个或几个细胞因子在IgG4-RD或AIP患者中的特异表达及其临床的相关性，且得出结论不一，由于适应性免疫的调控机制复杂，各细胞及细胞因子之间的相互作用和调控是多维度多方位的。因此，本研究旨在利用多因子检测这一高效的技术手段，从全景范围里初步描述1型AIP的趋化因子谱区别于健康群体及胰腺肿瘤的特征，为后续的机制探究和临床应用提供理论参考和数据基础。

方法：前瞻性纳入36例1型AIP患者，18例胰腺癌患者以及18例性别、年龄与AIP患者匹配的健康对照者，收集其基线状态外周血，使用Luminex检测技术和多因子试剂盒，定量测定40种细胞因子/趋化因子的血浆表达水平，利用统计学检验进行组间的差异比较，筛选出1型AIP患者表达升高的细胞因子/趋化因子，并进一步与其IgG4水平进行相关性分析。利用ROC曲线分析检测的趋化因子作为AIP诊断指标的诊断效能，初步评估其生物标志物的应用潜能。利用热图初步描绘AIP、胰腺癌和健康对照组各组之间的细胞因子及趋化因子相对水平。

结果：AIP组、PC和HC三组比较时，AIP血浆CCL1，CCL3，CCL19，CCL23，CCL26，CCL27，CXCL2，CXCL6，CXCL9，CXCL13，IL-1β，IL-4，IL-8/CXCL8，IL-10，GM-CSF（共15个因子）水平显著高于HC组（P＜0.05）。AIP与PC组的血浆CCL2，CCL13，CCL15，CCL17，CCL21，CXCL1，CXCL12，CXCL2，IL-4（共9个因子）水平的差异存在显著性（P＜0.05）。其中，AIP组的血浆CCL13，CCL17，IL-4，CXCL1，CXCL2水平显著高于PC组；而PC组的血浆CCL2，CCL15，CCL21，CXCL12则显著高于AIP组。1型AIP患者的血IgG4水平与血CCL19水平（r = 0.5418，P = 0.0006）、血CXCL13水平（r = 0.6220，P ＜0.0001）之间分别呈显著正相关性。ROC曲线提示CXCL13、CCL3、IL-4、IL-8和CCL19 的AUC＞0.8，且均具有统计学意义（P＜0.001）；其中，CCL3（截断值2.64pg/ml）的诊断灵敏度最高，可达到89%；而IL-4（截断值48.67pg/ml）的特异度最高，为100%。

结论：1型AIP患者外周血细胞因子及趋化因子的表达谱具有区分于健康人和胰腺癌的特征性。AIP患者外周血存在部分趋化因子与IgG4水平具有显著相关性。在AIP患者中差异表达的趋化因子具有作为诊断及疾病监测的生物标志物的前景，并为疾病的机制研究提供了线索和数据基础。

Part 1: Clinical features and risk factors analysis of peripancreatic vascular involvement and diabetes mellitus in patients with type 1 autoimmune pancreatitis

Background: Autoimmune pancreatitis (AIP) is a type of pancreatitis involving autoimmune mechanisms, accounting for 5% to 6% of chronic pancreatitis. Type 1 AIP is a pancreatic manifestation of immunoglobulin G4-related disease (IgG4-RD). Peripancreatic vascular involvement (PVI) is a highly characteristic clinical feature of type 1 AIP. However, the presence of PVI can make it difficult to distinguish AIP from pancreatic malignancy, as malignant tumors also have biological behaviors that can invade vessels. Despite this, there have been few reports focusing on PVI. This study aimed to retrospectively review and analyze the clinical characteristics of type 1 AIP patients and identify important factors associated with PVI. Additionally, as the pancreas is affected in this disease, impaired glucose tolerance or diabetes mellitus (DM) is frequently observed in type 1 AIP patients. DM can be the first even the only clinical manifestation of AIP at diagnosis, and the response to corticosteroid treatment (CST) for AIP varies, with differences in prognosis. However, evidence regarding whether CST for AIP can improve or worsen DM is still insufficient. In this context, this study aims to explore the characteristic differences between newly developed DM and pre-existing DM in AIP patients, and to preliminarily evaluate the effect of CST on the prognosis of DM through the demonstration and analysis of clinical features.

Methods: This retrospective study included 101 cases of type 1 AIP patients who were hospitalized in the Department of Gastroenterology, Peking Union Medical College Hospital between January 2013 and December 2018. Detailed information was extracted from the medical record system, including demographic, clinical, laboratory, and imaging data. All patients were divided into non-PVI and PVI groups, and potential risk factors for PVI in AIP patients were analyzed through clinical data. Patients were also divided into three groups based on the chronology in which AIP and DM were diagnosed: pre-existing DM (pDM), concurrent DM (cDM), and non-DM (nDM). The factors associated with cDM in AIP patients were analyzed through the analysis of clinical data among the three groups. In addition, the predictive factors for the improvement or worsening of DM after steroid therapy for AIP were explored.

Results: Among 101 cases of type 1 AIP patients, 52 (51.5%) had PVI, with a male-female ratio of 5.5:1 and a mean age of 58.37±8.68 years. Compared with the non-PVI group, the PVI group had a higher proportion of pancreatic tail involvement (80.8% vs. 57.1%, P = 0.010), a higher incidence of splenomegaly (28.8% vs. 4.1%, P = 0.001), and lower levels of peripheral blood white blood cell (WBC) count (5.96±1.54×10⁹ vs. 6.80±2.01×10⁹, P = 0.020). Multivariate logistic analysis revealed that pancreatic tail involvement (OR, 3.06; 95% CI, 1.17-8.02; P = 0.023) and splenomegaly (OR, 8.23; 95% CI, 1.67-40.51; P = 0.010) were both risk factors for PVI in type 1 AIP patients. A total of 18 out of 25 patients in PVI group who underwent corticosteroid treatment and no less than 6 months radiological follow-up showed improvement in vascular lesions, and no case exhibited exacerbation of PVI lesions during follow-up.

Among 101 patients with type 1 AIP, a total of 52 patients (51.5%) were diagnosed with DM at the time of AIP diagnosis, among which the cDM group had 36 patients, accounting for the majority of the DM group (69.2%). The incidence of diffuse pancreatic swelling (diffuse type) in the cDM group was significantly higher than that in the pDM group (72.2% vs. 37.5%, P = 0.018) and the nDM group (72.2% vs. 51.0%, P = 0.049). At the same time, the proportion of pancreatic body/tail involvement in the cDM group was also significantly higher than that in the pDM group (91.7% vs. 56.3%, P = 0.009) and the nDM group (91.7% vs. 69.4%, P = 0.013). Of the 48/52 AIP patients with DM, CST was administered. By comparing the clinical characteristics of the DM improvement group with those of the non-improvement group and conducting univariate and multivariate analysis, it was found that elevated serum GGT levels at diagnosis (OR, 0.032; 95% CI, 0.003-0.412; P = 0.008) and pancreatic atrophy after steroid treatment (OR, 0.027; 95% CI, 0.003-0.295; P = 0.003) were both independent predictors of lower likelihood of DM improvement after steroid treatment.

Conclusion: The incidences of PVI and DM in type 1 AIP patients are relatively high, and their occurrence is correlated with some clinical characteristics. Clinicians should pay attention to the occurrence of complications and their prognosis in type 1 AIP patients. PVI in some patients can improve after steroid treatment. CST has a beneficial effect on DM control in some type 1 AIP patients with cDM at the time of diagnosis, but they can also cause steroid-related DM in some patients. Therefore, when treating type 1 AIP patients, it is necessary to evaluate the pros and cons of CST and individualize the treatment accordingly.

Part 2: Chemokine patterns associated with type 1 autoimmune pancreatitis

Background: IgG4-RD is an immune-mediated disease characterized by fibro-inflammatory lesions that can occur in various organs, including the salivary/lacrimal glands, orbit, pancreas and biliary tract, aorta, kidneys, meninges, and thyroid. Type 1 AIP is a manifestation of IgG4-RD. Elevated serum IgG4 levels can provide important clues for diagnosing type 1 AIP and IgG4-RD, but clinical diagnosis or appropriate disease classification based on a single indicator is difficult. The diagnosis of IgG4-RD still requires comprehensive examinations including histology, imaging, and serology, to complete the exclusion of potential malignancies, infections, or other immune-mediated diseases. Meanwhile, with the further understanding of disease pathophysiology and the availability of high-throughput diagnostic platforms, many new blood biomarkers have been investigated and described in IgG4-RD in recent years. Chemokines are a class of small molecule cytokines or signaling proteins, and high levels of chemotactic factor expression have been observed in many human autoimmune diseases, such as rheumatoid arthritis, systemic lupus erythematosus, and systemic sclerosis. Chemokines and their receptors are believed to induce immune cell infiltration into affected organs in these diseases. CCL17, CCL26, CXCL12, CXCL13 have been reported to be associated with IgG4-RD. However, previous studies have mostly focused on the specific expression of one or a few chemokines in IgG4-RD or AIP patients and their clinical correlations, with conflicting conclusions. Due to the complex regulatory mechanisms of adaptive immunity, the interaction and regulation between various cells and cytokines are multidimensional and multi-directional. Therefore, this study aims to use multi-factor detection as an efficient technique to preliminarily describe the chemokines profile of type 1 AIP compared to healthy controls and pancreatic cancer in a panoramic range, providing a reference and data basis for subsequent mechanistic exploration and clinical application.

Methods: Thirty-six patients with type 1 AIP, 18 patients with pancreatic cancer, and 18 healthy controls matched in gender and age with type 1 AIP patients were prospectively enrolled. Peripheral blood was collected before initial treatment, and the plasma expression levels of 40 kind of cytokines/chemokines were quantitatively measured using Luminex technology and multi-factor reagent kit. Statistical tests were used to compare differences between groups and screen for cytokines/chemokines with elevated expression in type 1 AIP patients and further analyze their correlation with IgG4 levels. ROC curve analysis was used to assess the diagnostic efficiency of the detected chemotactic factors as AIP diagnostic indicators and to preliminarily evaluate their potential as biomarkers. Heat maps were used to preliminarily depict the relative levels of cytokines and chemotactic factors among the type 1 AIP, pancreatic cancer, and healthy control groups.

Results: When comparing the AIP group with the PC and HC groups, the levels of 15 kind of factors, including CCL1, CCL3, CCL19, CCL23, CCL26, CCL27, CXCL2, CXCL6, CXCL9, CXCL13, IL-1β, IL-4, IL-8/CXCL8, IL-10, and GM-CSF, in AIP plasma were significantly higher than those in the HC group (P＜0.05). The differences in the levels of 9 kind of factors, including CCL2, CCL13, CCL15, CCL17, CCL21, CXCL1, CXCL12, CXCL2, and IL-4 in AIP plasma compared to PC plasma were significant (P＜0.05). Specifically, the levels of CCL13, CCL17, IL-4, CXCL1, and CXCL2 in AIP plasma were significantly higher than those in PC plasma, whereas the levels of CCL2, CCL15, CCL21, and CXCL12 in PC plasma were significantly higher than those in AIP plasma. The levels of IgG4 in type 1 AIP patients were significantly positively correlated with the levels of CCL19 (r = 0.5418, P = 0.0006) and CXCL13 (r = 0.6220, P＜0.0001). The ROC curves indicated that the AUC of CXCL13, CCL3, IL-4, IL-8, and CCL19 were all more than 0.8 and statistically significant (P＜0.001). Among these factors, CCL3 had the highest diagnostic sensitivity (89%) with a cut-off value of 2.64 pg/ml, while IL-4 had the highest specificity (100%) with a cut-off value of 48.67 pg/ml.

Conclusion: The expression profiles of peripheral blood chemokines and cytokines in type 1 AIP patients exhibit characteristics that are distinct from those of healthy individuals and pancreatic cancer patients. Some chemokines in the peripheral blood of AIP patients are significantly correlated with serum IgG4 levels. The differentially expressed chemokines in AIP patients may serve as promising biomarkers for diagnosis and disease monitoring and provide important clues and background for the mechanistic study of the disease.