Differential risks of venous thromboembolic events in pemphigus and pemphigoid diseases

Venous thromboembolic events (VTE) are severe and potentially fatal complications of bullous pemphigoid (BP) and pemphigus.1-3 In contrast, the risk of VTE in other autoimmune skin blistering diseases (AIBD), such as mucous membrane pemphigoid (MMP) and epidermolysis bullosa acquisita (EBA) remains unknown. To address this clinically relevant knowledge gap, we used a large-scale, federated electronic health record database and performed retrospective cohort studies as described.4, 5 Specifically, we used the Global or US Collaborative Network of TriNetX to identify electronic health records (EHRs) from patients with pemphigus or pemphigoid diseases, as well as non-AIBD controls. Case groups were defined by ICD-10-CM-codes: Pemphigus (ICD10-CM:L10.0, L10.1, L10.2 or L10.4), BP (ICD10-CM:L12.0), MMP (ICD10-CM:L12.1) or EBA (ICD10-CM:L12.3). For each case group, all other AIBD were excluded. The non-AIBD control group was defined by ICD-10-CM:Z00 'Encounter for general adult medical examination without abnormal findings' and absence of any AIBD ICD-10-CM code. For cases, the index event was defined as the first recorded documentation of the respective AIBD. For controls, it was determined by the healthcare encounter. VTE was defined as an outcome based on ICD-10-CM codes for pulmonary embolism (ICD-10-CM:I26) or acute embolism and thrombosis of the deep veins of the lower extremity (ICD-10-CM:I82.4). Outcomes prior to the index event were excluded using the default method within the TriNetX analytical platform or at data retrieval.4 VTE risks between cases and controls were contrasted using Kaplan–Meier. Propensity-score matching (PSM) with the following covariates was applied to improve comparability: Age at index (years, continuous), female sex (binary), neoplasms (binary) and documented surgery (binary). Five sensitivity analyses (S1–S5) were performed: For some of the sensitivity analyses, different and/or additional covariates were used for PSM that included VTE risk factors.6, 7 The primary analysis considered outcomes from 1 day to 3 years after the index event. S1 assessed outcomes from 1 month to 3 years. S2 excluded any outcomes before the index event at data retrieval. In S3, parameters for PSM were modified. S4 mandated a minimum baseline time window. S5 applied extensive PSM. All analyses were performed from 3 March to 20 June 2025. Details on the cohorts are shown in Mendeley Tables S1–S4 in Data S1 (https://data.mendeley.com/datasets/y2dbh2rp3s/1). Following PSM, the cohorts included 971 (EBA) to 14,220 (BP) EHRs, with equally sized, corresponding non-AIBD control groups. In line with previous reports, we found increased VTE risks in patients diagnosed with pemphigus or BP1-3 with hazard ratios (HR) ranging from 1.46 to 2.13 across the primary and all sensitivity analyses (Figure 1). Results also persisted when we included exposure to high-dose intravenous immunoglobulin G treatment and coagulation defects for PSM, as well as when mandating a second documentation of pemphigus or BP (not shown). In the primary analysis of MMP cases and controls, VTE was documented in 0.74% of cases and 1.27% of controls. However, this risk difference was not statistically significant (HR 0.67, 95% CI: 0.43–1.11, p = 0.08). Again, these results persisted across all sensitivity analyses (Figure 1). In EBA, the primary analysis indicated an elevated VTE risk (HR 3.19, 95% CI: 1.35–7.49, p = 0.005). This risk difference was validated in 2/5 sensitivity analyses. This study is limited by potential misclassification from ICD-10 coding and differences in healthcare access. Compared to controls, a shorter median follow-up was noted in BP cases. This bias was addressed by the time-to-event analysis. The retrospective design does not allow causal inference. We confirm increased VTE risk in patients with pemphigus, BP and potentially EBA, but not MMP. Although the risk increase is moderate, it involves a serious outcome in a vulnerable population. VTE monitoring may be warranted in selected, but not all, AIBD patients. We thank Friederike Uebing (UKSH, Campus Kiel) for her exceptional support in facilitating TriNetX access. This project was developed and conducted during the Master's specialization course in Molecular Life Sciences at the University of Lübeck. We also acknowledge the TriNetX Help Center for their continued and excellent support. Cluster of Excellence Precision Medicine in Chronic Inflammation (EXC 2167), Collaborative Research Center PANTAU (SFB 1562), Individual Research Grant LU 877/25-1, all by the Deutsche Forschungsgemeinschaft and the Schleswig-Holstein Excellence-Chair Programme from the State of Schleswig-Holstein. RJL has received honoraria for speaking or consulting, or has obtained research grants from Novartis, Lilly, Bayer, Dompe, Synthon, Alpine Immune Sciences, 32QBio, CSL, Pharmaxis, QIMA Monasterium and TriNetX. All other authors declare no conflict of interest. Our study did not require Institutional Review Board approval. The data reviewed is a secondary analysis of existing data, does not involve intervention or interaction with human subjects and is de-identified per the de-identification standard defined in Section §164.514(a) of the HIPAA Privacy Rule. All data are included in the manuscript and the Mendeley Tables S1–S4 (https://data.mendeley.com/datasets/y2dbh2rp3s/1).