Immune thrombocytopenic purpura increased risk of subsequent pancreatitis: A Nationwide population cohort study

Immune thrombocytopenic purpura (ITP) is characterized by thrombocytopenia and bleeding diathesis. Pancreatitis is a very rare complication but may be fatal. We analyzed data of newly diagnosed ITP patients, excluding those with a history of splenectomy, unknown sex or date of birth, or preexisting pancreatitis at the time of ITP diagnosis, and compared these with selected age-, gender-, and index-year-matched controls, using the Taiwan National Health Insurance Research Database from 1996 to 2013. The study enrolled 100,177 ITP patients and 100,177 controls. We found that pancreatitis risk was higher in secondary ITP patients, regardless of age group, gender, baseline Charlson comorbidity index (CCI) score, history of biliary stone, hyperlipidemia, or alcoholism, than in the control population. Primary ITP patients with CCI score 1 and without biliary tract stone history also showed a higher pancreatitis risk than the controls. The incidence rate and cumulative incidence of pancreatitis were increased in primary, secondary, and unspecified ITP cases. These phenomena may be related to the presence of autoantibodies against glycoprotein IIb/IIIa, or to IgG4, microparticle obstruction, or sclerosis. We noted a direct association between ITP and the development of pancreatitis in Taiwan population.

ITP increases pancreatitis risk. We evaluated the risk of pancreatitis in different stratifications in Tables 2,   3 and 4. Disease diagnoses were identified using the International Classification of Diseases, 9 th Revision, Clinical  Modification (ICD-9-CM). We divided ITP patients into 3 subgroups according to the ICD-9-CM codes: primary (ICD9 code 287.3), secondary (ICD9 code 287.4), unspecified (ICD9 code 287.5). Overall ITP patients had significant higher risk of pancreatitis than the control group (adjusted HR = 1.22 with P = 0.0154 in primary ITP, Table 2; adjusted HR = 1.86 with P < 0.0001 in secondary ITP, Table 3; adjusted HR = 2.00 with P < 0.0001 in thrombocytopenia, unspecify, Table 4). When stratifying all patients by age, gender, and CCI score, secondary ITP patients and thrombocytopenia, unspecific patients had significant higher risks of pancreatitis than the control group in all stratifications. Primary ITP patients with baseline CCI score 1 showed a higher risk of pancreatitis. Among subjects without biliary stone, hyperlipidemia or alcoholism, ITP patients had significantly higher risk of pancreatitis than the controls. However, it showed the reverse result among subjects with biliary stone in unspecific thrombocytopenia patients (adjusted HR = 0.79 with 95% CI = 0.67-0.93, Table 4). A trend of lower pancreatitis risk in all three groups with alcoholism was also noted.
We also compared the risk of pancreatitis among all subgroups with that of the control group. All ITP subgroups had significant higher risk of pancreatitis. Among ITP group, primary ITP patients had lower risk of pancreatitis than unspecified ITP patients (adjusted HR = 0.62 with 95% CI = 0.48-0.81. Table 5).

ITP increases pancreatitis risk with time.
We analyzed the cumulative incidence of pancreatitis in ITP group, ITP subgroups and the control group. There was a significantly higher incidence of pancreatitis in ITP group/ITP subgroups, especial in secondary and unspecified groups, than in the control group (Fig. 2, Log-rank p < 0.0001).

Discussion
Idiopathic thrombocytopenic purpura (ITP) is an acquired, immune-mediated disease of adults and children, with increasing of bleeding risk, that caused by transient or persistent decrease of the platelet count 23 . In addition, ITP patients were with increased susceptibility to bleeding because of antibody-and cell-mediated destruction of platelets and the suppression of platelet production. Moreover, the destruction and suppression of platelets are due to the autoimmune reactions of B cell (and sometimes CD8+ T cell) that directed against circulating platelets and megakaryocytes 1 .
ITP complicated by pancreatitis is very rare 10 . We used ICD-9 code including primary ITP, secondary ITP, and thrombocytopenia, unspecific to analysis if ITP increasing pancreatitis risk. The reason we chose those 3 codes is that we want to screen not only primary, but also secondary ITP effect on pancreatitis risk. We also pick up code 287.5, defined as thrombocytopenia, unspecific, because some clinician would use this broad define code, instead of more specific code, like 287.3 (primary ITP) or 287.4 (secondary ITP). Many patients putting in  Table 2. Risk of pancreatitis in ITP patients compared with the comparison group @ . † Per 1000 person-years. *p < 0.05; **p < 0.01; ***p < 0.001. @ Stratified by age, sex, and baseline characteristics. Variables listed in Table 1 were considered as potential risk factors and were all put in the adjusted model.
www.nature.com/scientificreports www.nature.com/scientificreports/ thrombocytopenia, unspecific item may be primary ITP ones, but less secondary ITP ones. Based on nationwide data in Taiwan, we found that the risk of pancreatitis increased in ITP patients, regardless of age, gender, and base line CCI. Primary ITP patients had higher pancreatitis risk in baseline CCI score 1 populations, but not in other CCI score ones. Biliary tract stones, hyperlipidemia, and alcoholism have been reported to increase pancreatitis risk [24][25][26] . In the Taiwanese population, primary ITP patients without biliary tract stones, and secondary ITP ones without biliary tract stones, hyperlipidemia, or alcoholism had a higher risk of pancreatitis than controls. A trend toward increasing pancreatitis risk was observed in ITP patients with hyperlipidemia, compared with that in controls, but the difference was only significant in secondary ITP group. However, primary ITP patients with a biliary stone history and alcoholism had a lower risk of pancreatitis than non-ITP patients. This suggests that biliary stones and alcoholism are stronger risk factors for pancreatitis than ITP. In addition, we aimed to determine whether the risk was different between primary and secondary ITP; the latter may be associated with other autoimmune diseases (eg, systemic lupus erythematosus [SLE], antiphospholipid antibody syndrome [APS], immune thyroid disease, or Evans syndrome), a lymphoproliferative disease (eg, chronic lymphocytic leukemia [CLL] or large granular T-lymphocyte lymphocytic leukemia), or chronic infection, eg, with Helicobacter pylori, human immunodeficiency virus (HIV), or hepatitis C virus (HCV) 27 . Our data showed that pancreatitis risk was higher in both primary and secondary ITP patients than in controls. However, the strength of association was less in the primary ITP group than in secondary or unspecified ITP groups. We may see stronger association in primary ITP group after expansion the collection patient number and years. In daily practice, we observed that ITP patients with obvious risk factors, including other autoimmune diseases, drugs, and infections, tended to develop secondary ITP. Most ITP patients without obvious risk factors were recorded as unspecified. The majority of unspecified ITP patients probably had primary ITP.
One possible cause of induced pancreatitis in ITP is splenectomy. Qu et al. reported a higher risk of pancreatitis in 604 post-splenectomy patients, possibly due to pancreatic injury or a fistula caused by surgery 28 . Therefore, in order to avoid this bias, we have excluded post-splenectomy patients in the current series. In addition, corticosteroids, the primary medication used in ITP, can also induce pancreatitis, and there were reported steroid-induced pancreatitis [29][30][31] . However, a meta-analysis and other reports showed a protective role of steroids for pancreatitis [32][33][34] . This inconsistency may indicate further investigations were needed for the role of steroid in this issue.
Other possible mechanisms of increased pancreatitis risk in ITP patients include antibody-associated pathways. One example is the IgG4-associated disease. Patients with autoimmune pancreatitis and ITP may share IgG4 antibodies. Patients with concurrent autoimmune pancreatitis and ITP reportedly had increased IgG4 levels 16,20 . In another mechanism, autoantibodies may target platelet membrane proteins. Autoantibodies against  Table 3. Risk of pancreatitis in ITP patients compared with the comparison group @ . † Per 1000 person-years. *p < 0.05; **p < 0.01; ***p < 0.001. @ Stratified by age, sex, and baseline characteristics. Variables listed in Table 1 were considered as potential risk factors and were all put in the adjusted model. www.nature.com/scientificreports www.nature.com/scientificreports/ platelet glycoprotein (GP) IIb/IIIa and/or GPIb-IX are found in the majority of ITP patients, which could be detected using antigen-specific assays. In addition, multiple antibodies could be produced by many patients, which might be attributed to the phenomenon of epitope spreading. Once produced, autoantibodies may either bind to platelets or megakaryocytes, which cause destruction by either phagocytosis or complement activation and lysis in the former, and decreased thrombopoiesis in the latter 35 . In mouse studies, anti-GPIbα monoclonal antibodies induced severe and irreversible thrombocytopenia (<3% of normal) through Fc-independent mechanisms. Monoclonal antibodies against conformational epitopes in GPIIb/IIIa-or anti-GPIIb/IIIa-specific inflammatory reactions could induce irreversible thrombocytopenia, hematocrits decrease, hypothermia, acute systemic reactions, and a paradoxical loss of surface GPIIb/IIIa on platelets in vivo. The loss of surface GPIIb/IIIa implied the formation of platelet-derived microparticles 36 . These microparticles may obstruct small vessels and cause ischemic changes in terminal organs, including the pancreas, leading to pancreatitis. We analyzed pancreatitis risk after a diagnosis of ITP. We noted a time-dependent, higher pancreatitis incidence rate and cumulative incidence in all ITP patients, including primary, secondary, and unspecified groups, than in controls. The incidence of acute pancreatitis increased with age 37 . In our study, we showed higher accumulation incidence of pancreatitis in ITP patients. This indicates that increasing pancreatitis risk in ITP patients may be related to platelet-derived microparticles and the loss of platelet surface GPIIb/IIIa due to an anti-GPIIb/IIIa antibody effect.
The possible mechanism of developing pancreatitis in secondary ITP is more complex and may relate to underline disease. For examples, HCV infects pancreas acinar cells and pancreatic duct epithelial cells directly and leads to pancreatitis 38,39 . Immunosuppression in HIV infection patients is related to increase risk of pancreatitis 40 . Autoimmune disease like SLE related pancreatitis may induced by vascular damage (including vasculitis,   www.nature.com/scientificreports www.nature.com/scientificreports/ intimal thickening, immune complex deposition, occlusion of arteries, and arterioles), autoantibody production, abnormal cellular immune response, and drug toxicity 41 . Leukemic cells infiltration in the pancreas in CLL patients, showed pancrease parenchymal destruction and fibrous scars, thereby resembling chronic pancreatitis 42 . And drug related pancreatitis, eg asparaginase 43 for acute lymphoblastic leukemia and Highly Active Anti-Retroviral Therapy (HAART) for HIV infection 40 .
The limitation of this study describe below. In the NHIRD, the disease was defined based on the International Classification of Diseases, Ninth Revision, Clinical Modification code. The disease code(s) of the patients is according to the diagnosis of specialists. Therefore, in the current study, it is hard to obtain precise diagnosis of ITP due to databank limitation and only diagnosis codes were available. For example, only 4 numbers code for ITP in our database makes it difficult to identify congenital and hereditary thrombocytopenia purpura. But most likely, congenital disease can be excluded by age stratification. However, all the insurance claims in Taiwan were scrutinized and coded by medical reimbursement specialists and peer reviewed according to the standard diagnosed criteria. If these doctors or hospitals commit errors in diagnoses or coding, they will be punished with a lot of penalties. Thus, medical staffs were very concerned about the correct diagnosis codes, and we think the codes regarding diagnoses were highly reliable.
In conclusion, adult ITP patients, especially secondary ITP ones without biliary stones, hyperlipidemia, or history of alcoholism, have a higher risk of pancreatitis than controls. Primary adult ITP patients also show higher risk of pancreatitis, especial in patients without biliary stone history., than controls. That may be related to immune mechanisms.

patients and Methods
Data source. This study used inpatient admissions data from the Taiwan National Health Insurance Research Database from 1996 to 2013. Established in 1995, the Taiwan National Health Insurance (NHI) program covers over 99% of the population and contracts with 93% of the medical institutions in Taiwan 44 . Disease diagnoses were identified using the International Classification of Diseases, 9 th Revision, Clinical Modification (ICD-9-CM). A disease diagnosis without valid supporting clinical findings may be considered a fraudulent claim by the NHI, with a penalty 100-fold greater than the payment claimed by the treating physician or hospital.
Patients aged 18 year-old or more, with newly diagnosed ITP (ICD-9 codes: 287.3-287.5) between 2000 and 2013, were identified. The date of ITP diagnosis was considered the study index date. Patients with the following criteria before the index date were excluded: (1) history of splenectomy; (2) unknown sex or birthday; (3) diagnosis of pancreatitis; or (4) deceased or withdrawn from the NHI program. We randomly assigned index dates for subjects who had never been diagnosed with ITP and selected age-, gender-, and index-year-matched subjects in our control group, using a control/case ratio of 1. We included 100,177 patients in the ITP group and 100,177 in the control group (Fig. 1). All patients were followed up from the index date to the first occurrence of one of the following: pancreatitis, withdrawal from the NHI program, or the last day of 2013. Baseline comorbidities, including biliary stones (ICD-9 code: 574), hyperlipidemia (ICD-9 code: 272), and alcoholism (ICD-9 codes: 291, 303, 305.00, 305.01, 305.02, 305.03, 571.0-571.3, 790.3, and V11.3) diagnosed before the index date were also considered. The Charlson comorbidity index (CCI) score was calculated for each subject.
The details of ICD-9 codes of ITP show as below.