This cohort study analyzed data from 2 large commercial insurance claims databases to characterize the incidence rate of bullous pemphigoid associated with the use of dipeptidyl peptidase–4 inhibitors.
Key Points
Question
What is the absolute risk of bullous pemphigoid in patients who initiate use of dipeptidyl peptidase–4 inhibitors and its comparative risk against sulfonylurea use?
Findings
In this cohort study of 1 664 880 patients with type 2 diabetes, the incidence rate of bullous pemphigoid per 1000 person-years among patients who initiated dipeptidyl peptidase–4 inhibitor therapy was 0.42 compared with 0.31 for sulfonylurea. The risk was 42% higher with dipeptidyl peptidase–4 inhibitors than with sulfonylurea and 62% to 70% higher in subgroups of patients who were 65 years or older, white, and linagliptin users.
Meaning
These findings suggest that use of dipeptidyl peptidase–4 inhibitors is associated with higher risk of bullous pemphigoid compared with sulfonylurea, and that clinicians should be aware of this rare adverse effect in subgroups with higher risk.
Abstract
Importance
Despite several recent reports on the elevated risk of bullous pemphigoid in patients with type 2 diabetes treated with dipeptidyl peptidase–4 (DPP-4) inhibitors, evidence on the absolute risk and comparative safety against other antidiabetics is limited.
Objective
To characterize the incidence rate of bullous pemphigoid associated with DPP-4 inhibitor use compared with second-generation sulfonylureas.
Design, Setting, and Participants
This cohort study used data from 2 large commercial insurance claims databases (Optum Clinformatics Data Mart from October 17, 2006, to December 31, 2018, and IBM MarketScan Research Database from October 17, 2006, to December 31, 2017) and Medicare data from January 1, 2006, to December 31, 2016. Patients with type 2 diabetes who initiated treatment with DPP-4 inhibitors or second-generation sulfonylurea were included.
Main Outcomes and Measures
The primary outcome of the study was bullous pemphigoid, identified using diagnosis codes. After 1:1 propensity score matching, the incidence rates of bullous pemphigoid and the hazard ratios (HRs) with 95% CIs comparing patients who initiated DPP-4 inhibitor and second-generation sulfonylurea therapy were estimated. Subgroup analyses by age, sex, race, and individual DPP-4 agents were performed. The results from each database were pooled using inverse-variance fixed-effects meta-analysis.
Results
A total of 1 664 880 patients who initiated DPP-4 inhibitors (51.0% female; mean [SD] age, 63.9 [9.7] years) and sulfonylurea (50.4% female; mean [SD] age, 63.9 [9.9] years) were included. The incidence rate of bullous pemphigoid per 1000 person-years was 0.42 in the DPP-4 inhibitor group vs 0.31 in the sulfonylurea group (HR, 1.42; 95% CI, 1.17-1.72). Higher rates per 1000 person-years for DPP-4 inhibitor vs sulfonylurea groups were seen in those who were 65 years or older (0.79 vs 0.49; HR, 1.62; 95% CI, 1.32-1.99), white (0.93 vs 0.54; HR, 1.70; 95% CI, 1.30-2.24), and treated with linagliptin (1.20 vs 0.55; HR, 1.68; 95% CI, 1.16-2.43).
Conclusions and Relevance
This study found that patients who initiated DPP-4 inhibitor therapy had higher risk of bullous pemphigoid than those who initiated second-generation sulfonylurea therapy. Clinicians should be aware of this rare adverse effect of DPP-4 inhibitors in subgroups of patients who are older, white, and linagliptin users.
Introduction
Dipeptidyl peptidase–4 (DPP-4) inhibitors are widely used as a second- or third-line treatment in the management of type 2 diabetes.1 Over time, concerns have been raised about potential adverse events associated with DPP-4 inhibitors, including pancreatitis,2 heart failure,3 and severe joint pain,4,5 along with cutaneous eruptions.6,7 Bullous pemphigoid is a rare autoimmune blistering disease characterized by tense blisters and urticarial plaques.8 Since 2012, an increased risk of bullous pemphigoid among DPP-4 inhibitor users has been reported in case series,9 pharmacovigilance reports,10,11 and case-control studies.12,13,14,15,16 The reported odds ratios varied greatly, ranging from 1.5812 to 3.16.15 However, these studies do not provide absolute risk of bullous pemphigoid and can be subject to several methodological limitations, such as confounding (lack of adjustment for factors that are potentially associated with both the exposure and the outcome), overadjustment (covariates used for adjustment are measured after or concurrently with treatment decision), selection bias (differential surveillance between the 2 comparator groups), and time-window bias (controls chosen at the end of available data serve more time in the data and thereby have a differential exposure assessment period).17
To date, only 1 published cohort study18 is available on the risk of bullous pemphigoid in users of DPP-4 inhibitors. The study was based on the UK Clinical Practice Research Datalink and explored the risk of bullous pemphigoid in patients with type 2 diabetes initiating use of either DPP-4 inhibitors or other antidiabetics.18 The absolute risk of bullous pemphigoid was 21.1 cases per 100 000 person-years. As in case-control studies, a higher risk of bullous pemphigoid was seen in users of DPP-4 inhibitors (hazard ratio [HR], 2.21; 95% CI, 1.45-3.38). However, the total number of identified bullous pemphigoid events was small (150 patients with a new diagnosis during 711 311 person-years), and lumping all second- to third-line antidiabetics may not be ideal compared with a single drug class comparison in clinical settings.
With these limitations from previous literature in mind, we designed a retrospective cohort study to compare the risk of bullous pemphigoid in initiators of DPP-4 inhibitors vs second-generation sulfonylureas by using 2 US commercial claims and Medicare databases. We chose sulfonylureas as the active comparator because they remain the most commonly used second-line agents in diabetes treatment and are not known to cause bullous pemphigoid.19
Methods
Data Source
We conducted a cohort study using data from 3 large US health care databases: Optum Clinformatics Data Mart (Optum) from October 17, 2006, to December 31, 2018; IBM MarketScan Research Database (MarketScan) from October 17, 2006, to December 31, 2017; and Medicare Parts A, B, and D (Medicare) from October 17, 2006, to December 31, 2016. Optum and MarketScan databases contain longitudinal medical and pharmacy claims of the commercially insured population in the United States. Medicare is a federally funded insurance program that covers legal residents in the United States 65 years and older and individuals with certain disabilities. This study was approved by the institutional review board of the Brigham and Women’s Hospital, Boston, Massachusetts. The requirement for patient informed consent was waived because personal identifiers were removed from the data sets to protect patient confidentiality. Signed data license agreements were in place for all data sources. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.20
Study Design and Population
We created a new-user cohort of patients with a prescription claim for either a DPP-4 inhibitor or a second-generation sulfonylurea from the date of first DPP-4 inhibitor approval (October 17, 2006) to the end of available data in each database. We applied a washout period of 180 days for both drugs, and the date of the first prescription corresponded to the date of cohort entry. We included individuals with at least 180 days of continuous enrollment in commercial insurance and fee-for-service Medicare and at least 1 diagnosis of type 2 diabetes based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), code 250.xx and International Statistical Classification of Diseases, Tenth Revision (ICD-10), code E11.xx in both the inpatient and outpatient settings. We excluded individuals younger than 40 years because bullous pemphigoid is a condition usually seen in older individuals. Furthermore, we excluded individuals with history of bullous pemphigoid, type 1 diabetes, secondary diabetes, malignant neoplasm, pregnancy, organ transplant, HIV, and AIDS based on diagnosis codes 180 days before the cohort entry date.
Study Outcomes
The primary outcome of interest was bullous pemphigoid. The occurrence of bullous pemphigoid was defined based on ICD-9 (694.5) or ICD-10 (L12.0) diagnosis codes in any position and any care setting. The follow-up began 1 day after cohort entry and continued until the earliest of the following events: occurrence of the study outcome, end of health care or pharmacy eligibility, end of study period, switch to a comparator drug, and discontinuation of the index drug. We used a prespecified grace period of 30 days and exposure risk window of 30 days. We conducted the following subgroup analyses: (1) those 65 years or older vs younger than 65 years, (2) male vs female, (3) white vs nonwhite (in Medicare only) to assess effect modification, and (4) individual DPP-4 inhibitors (sitagliptin phosphate, saxagliptin hydrochloride, and linagliptin) to quantify associations between individual agents and bullous pemphigoid. Alogliptin benzoate was excluded from the subgroup analysis owing to a low number of users.
Covariates
All covariates were measured during the baseline period of 180 days before the initiation of DPP-4 inhibitor or second-generation sulfonylurea use. The predefined covariates included age, sex, comorbidities (such as cardiovascular, endocrine, or psychiatric diseases based on ICD-9 or ICD-10 diagnosis), prescription drug use (including other diabetic and nondiabetic medications), and use of health care services (such as a recent hospitalization and emergency department visit).
Statistical Analysis
The analyses were conducted in each database, and results were later combined to provide pooled estimates of the association. We first compared the baseline patient characteristics of the DPP-4 inhibitor and the second-generation sulfonylurea groups in each database. To reduce the imbalance in the clinical characteristics of each group, we used 1:1 propensity score matching. Propensity score matching was performed in each database, and results were pooled across data sources. Propensity score was defined as the estimated probability of a patient to start a DPP-4 inhibitor vs a second-generation sulfonylurea, estimated from a logistic regression model that included more than 100 predefined baseline covariates. The 1:1 propensity score–matched cohort was created using a 1:1 nearest-neighbor matching algorithm (with maximum caliper width of 0.01), and the balance between covariates after propensity score matching was evaluated through standardized mean differences. We determined a standardized mean difference of less than 0.10 as achieving adequate balance.21
In the 1:1 propensity score–matched cohort of each database, we calculated the number of events, incidence rates (IRs), and HRs with 95% CI. We confirmed that the proportional hazards assumption had not been violated by examining the interaction between exposure and time. As mentioned above, the analyses were conducted separately in each database and then combined using inverse-variance fixed-effects meta-analysis. Kaplan-Meier curves were generated to assess the risk of outcome over time.
Additional predefined sensitivity analysis was conducted in the Medicare database only with a stricter bullous pemphigoid outcome definition of 2 or more diagnoses of bullous pemphigoid separated by 7 to 365 days in both inpatient and outpatient settings. All analyses were conducted using the Aetion Evidence Platform (Aetion, Inc) and SAS, version 9.4 (SAS Institute Inc).
Results
Characteristics of Study Population
Our cohort consisted of 870 709 new users of DPP-4 inhibitors (49.0% male and 51.0% female; mean [SD] age, 63.7 [9.7] years) (205 703 in the Optum database; 381 168 in the MarketScan database; and 283 838 in the Medicare database) and 1 910 304 new users of second-generation sulfonylureas (51.8% male and 48.2% female; mean [SD] age, 64.8 [10.2] years) (524 557 in the Optum database; 759 463 in the MarketScan database; and 626 284 the Medicare database) (eFigure in the Supplement). Compared with patients with commercial insurance, those with Medicare were older and had a higher proportion of comorbidities and recent hospitalization (eTable 1 in the Supplement). Before propensity score matching, sulfonylurea users were older than the DPP-4–inhibitor users (mean [SD] age, 64.8 [10.2] vs 63.7 [9.7] years) and had a higher number of antidiabetic drugs prescribed at the index date (mean [SD], 1.26 [0.49] vs 1.16 [0.42]). After 1:1 propensity score matching, we had 199 780 pairs of DPP-4 inhibitor or second-generation sulfonylurea users in Optum (mean [SD] age, 61.5 [11.6] years; median follow-up, 119 [IQR, 58-305] days), 354 759 pairs in MarketScan (mean [SD] age, 57.3 [10.3] years; median follow-up, 130 [IQR, 58-320] days), and 277 901 pairs in Medicare (mean [SD] age, 74.1 [7.3] years; median follow-up, 135 [IQR, 68-332] days). All measured clinical characteristics were well balanced after the propensity score matching, as indicated by standardized mean difference of less than 0.10 (Table 1).
Table 1. Baseline Characteristics of Patients With Type 2 Diabetes Before and After Propensity Score Matchinga.
| Characteristic | User group | |||||
|---|---|---|---|---|---|---|
| Before matching | After matching | |||||
| Second-generation sulfonylurea (n = 1 910 304) | DPP-4 inhibitors (n = 870 709) | SMD | Second-generation sulfonylurea (n = 832 440) | DPP-4 inhibitors (n = 832 440) | SMD | |
| Sociodemographic | ||||||
| Age, mean (SD), y | 64.8 (10.2) | 63.7 (9.7) | 0.11 | 63.9 (9.9) | 63.9 (9.7) | <0.01 |
| Male | 51.8 | 49.0 | 0.06 | 49.6 | 49.0 | 0.01 |
| Raceb | ||||||
| White | 77.2 | 74.4 | 0.07 | 75.0 | 74.7 | 0.01 |
| Black | 11.8 | 11.7 | <0.01 | 11.7 | 11.7 | <0.01 |
| Asian | 3.3 | 4.9 | 0.08 | 4.5 | 4.7 | 0.01 |
| Hispanic | 4.1 | 4.8 | 0.03 | 4.6 | 4.7 | <0.01 |
| Native American | 0.7 | 0.6 | 0.01 | 0.6 | 0.6 | <0.01 |
| Other/unknown | 3.0 | 3.7 | 0.04 | 3.5 | 3.6 | 0.01 |
| Comorbidities | ||||||
| Myocardial infarction | ||||||
| Acute | 1.6 | 1.5 | 0.01 | 1.5 | 1.5 | <0.01 |
| Old | 2.6 | 2.4 | 0.01 | 2.4 | 2.4 | <0.01 |
| Heart failure | 8.7 | 8.2 | 0.02 | 8.4 | 8.4 | <0.01 |
| Hypertension | 67.7 | 69.9 | 0.05 | 70.2 | 69.9 | 0.01 |
| Ischemic heart disease | 19.4 | 19.1 | 0.01 | 19.4 | 19.2 | 0.01 |
| Ischemic stroke | 6.1 | 6.1 | <0.01 | 6.2 | 6.1 | <0.01 |
| Dementia | 4.6 | 4.7 | <0.01 | 4.8 | 4.7 | <0.01 |
| Delirium | 1.7 | 1.7 | <0.01 | 1.8 | 1.7 | 0.01 |
| Alcohol abuse | 0.9 | 0.7 | 0.02 | 0.8 | 0.7 | 0.01 |
| Smoking | 8.7 | 8.1 | 0.02 | 8.3 | 8.3 | <0.01 |
| Falls | 2.7 | 2.7 | <0.01 | 2.6 | 2.7 | 0.01 |
| Combined Comorbidity Index, mean (SD)c | 2.58 (2.32) | 2.61 (2.31) | 0.01 | 2.64 (2.30) | 2.62 (2.33) | 0.01 |
| Medications | <0.01 | |||||
| Angiotensin-converting enzyme inhibitors | 44.5 | 39.7 | 0.10 | 40.2 | 40.3 | <0.01 |
| Angiotensin receptor blockers | 21.5 | 28.1 | 0.15 | 27.5 | 27.3 | <0.01 |
| β-Blockers | 35.2 | 34.4 | 0.02 | 34.8 | 34.6 | <0.01 |
| Loop diuretics | 14.2 | 13.2 | 0.03 | 13.5 | 13.4 | <0.01 |
| Statins | 57.6 | 61.9 | 0.09 | 61.8 | 61.6 | <0.01 |
| Anticonvulsants | 12.2 | 13.0 | 0.02 | 13.1 | 13.0 | <0.01 |
| Antidepressants | 21.7 | 24.0 | 0.05 | 24.0 | 23.8 | <0.01 |
| Antipsychotics | 2.6 | 2.9 | 0.02 | 2.9 | 2.9 | <0.01 |
| Benzodiazepines | 9.4 | 10.3 | 0.03 | 10.3 | 10.3 | <0.01 |
| Corticosteroids, oral | 14.9 | 15.4 | 0.01 | 15.5 | 15.4 | <0.01 |
| Dementia drugs | 2.6 | 2.9 | 0.02 | 2.9 | 2.9 | 0.01 |
| Opioids | 25.1 | 24.7 | 0.01 | 25.0 | 24.8 | <0.01 |
| Parkinsonism drugs | 2.3 | 2.5 | 0.01 | 2.6 | 2.5 | <0.01 |
| Insulin | 11.7 | 15.8 | 0.12 | 15.4 | 15.1 | 0.01 |
| Metformin | 65.1 | 71.9 | 0.15 | 70.9 | 71.0 | <0.01 |
| No. of antidiabetic drugs at index date, mean (SD) | 1.26 (0.49) | 1.16 (0.42) | 0.22 | 1.16 (0.40) | 1.16 (0.42) | <0.01 |
| Use of health care services | <0.01 | |||||
| Emergency department visits, mean (SD) | 0.43 (1.48) | 0.37 (1.41) | 0.04 | 0.39 (1.34) | 0.38 (1.44) | 0.01 |
| Any hospitalizations | 12.9 | 11.0 | 0.06 | 11.4 | 11.2 | 0.01 |
Abbreviations: DPP-4, dipeptidyl peptidase–4; SMD, standardized mean difference.
Unless otherwise indicated, data are expressed as percentage of patients.
Available in Medicare only.
Scores range from −2 to 26, with higher scores indicating a higher burden of comorbidity.
Risk of Bullous Pemphigoid in DPP-4 Inhibitor vs Sulfonylurea Users
We estimated database-specific and pooled results for the risk of bullous pemphigoid after propensity score matching (Table 2). The incidence rates (IRs) (per 1000 person-years) of bullous pemphigoid for DPP-4 inhibitor vs sulfonylurea groups in each database were 0.36 vs 0.27 cases (HR, 1.34; 95% CI, 0.88-2.04) in Optum; 0.19 vs 0.19 cases (HR, 1.00; 95% CI, 0.68-1.49) in MarketScan; and 0.81 vs 0.48 cases (HR, 1.68; 95% CI, 1.30-2.17) in Medicare. The pooled number of events in the 3 databases was 247 cases (IR, 0.42 per 1000 person-years) in the DPP-4 inhibitor group and 184 cases (IR, 0.31 per 1000 person-years) in the sulfonylurea group. The pooled HR from the 3 databases was 1.42 (95% CI, 1.17-1.72). The Figure illustrates the cumulative incidence of bullous pemphigoid in the 1:1 propensity score–matched cohort in each database. The 2 groups diverged after approximately 6 months of drug use.
Table 2. Database-Specific and Pooled IRs and HRs of Bullous Pemphigoid in Initiation of Second-generation Sulfonylurea and DPP-4 Inhibitor Use.
| Databasea | Sulfonylurea | DPP-4 inhibitors | HR (95% CI) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. of patients | No. of events | Person-years | IR per 1000 person-years | No. of patients | No. of events | Person-years | IR per 1000 person-years | ||
| Optum | 199 780 | 39 | 143 072 | 0.27 | 199 780 | 49 | 134 667 | 0.36 | 1.34 (0.88-2.04) |
| MarketScan | 354 759 | 48 | 251 202 | 0.19 | 354 759 | 51 | 266 978 | 0.19 | 1.00 (0.68-1.49) |
| Medicare | 277 901 | 97 | 200 707 | 0.48 | 277 901 | 147 | 181 786 | 0.81 | 1.68 (1.30-2.17) |
| Combined | 832 440 | 184 | 594 981 | 0.31 | 832 440 | 247 | 583 431 | 0.42 | 1.42 (1.17-1.72) |
Abbreviations: DPP-4, dipeptidyl peptidase–4; HR, hazard ratio; IR, incidence rate.
Includes Optum Clinformatics Data Mart (Optum), IBM MarketScan Research Database (MarketScan), and Medicare Parts A, B, and D (Medicare).
Figure. Cumulative Incidence of Bullous Pemphigoid Among Groups Initiating Use of Dipeptidyl Peptidase–4 (DPP-4) Inhibitors and Second-generation Sulfonylureas.
Databases include Optum Clinformatics Data Mart (Optum), IBM MarketScan Research Database (MarketScan), and Medicare Parts A, B, and D (Medicare).
Subgroup and Sensitivity Analyses
The IRs of bullous pemphigoid in the DPP-4 inhibitor and sulfonylurea groups were examined by the subgroups in each database (eTable 2 in the Supplement) and then as pooled results (Table 3). The increased risk was more evident in those who were 65 years or older (pooled HR, 1.62; 95% CI, 1.32-1.99) and white (pooled HR, 1.70; 95% CI, 1.30-2.24). Among individual DPP-4 inhibitor agents, linagliptin demonstrated the highest risk (pooled HR, 1.68; 95% CI, 1.16-2.43), whereas the risk was less evident in sitagliptin and saxagliptin compared with sulfonylurea use. Higher risk of bullous pemphigoid in the DPP-4 inhibitors group was consistent between men (pooled HR, 1.38; 95% CI, 1.06-1.79) and women (pooled HR, 1.34; 95% CI, 1.03-1.75).
Table 3. Pooled Results of Subgroup Analyses of Bullous Pemphigoid in Initiation of Second-generation Sulfonylurea and DPP-4 Inhibitor Use.
| Characteristic | Sulfonylurea | DPP-4 inhibitor | HR (95% CI) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. of patients | No. of events | Person-years | IR per 1000 person-years | No. of patients | No. of events | Person-years | IR per 1000 person-years | ||
| Age, y | |||||||||
| ≥65 | 423 862 | 155 | 316 764 | 0.49 | 423 862 | 229 | 290 423 | 0.79 | 1.62 (1.32-1.99) |
| <65 | 399 989 | 26 | 269 975 | 0.10 | 399 989 | 26 | 286 052 | 0.09 | 0.94 (0.55-1.62) |
| Sex | |||||||||
| Male | 406 720 | 99 | 303 820 | 0.33 | 406 720 | 122 | 295 157 | 0.41 | 1.38 (1.06-1.79) |
| Female | 424 850 | 96 | 290 818 | 0.33 | 424 850 | 124 | 287 548 | 0.41 | 1.34 (1.03-1.75) |
| Racea | |||||||||
| White | 208 960 | 86 | 158 099 | 0.54 | 208 960 | 128 | 138 181 | 0.93 | 1.70 (1.30-2.24) |
| Nonwhite | 68 499 | 28 | 41 988 | 0.67 | 68 499 | 22 | 43 411 | 0.51 | 0.77 (0.44-1.34) |
| Individual drugs | |||||||||
| Sitagliptin phosphate | 654 923 | 136 | 464 464 | 0.29 | 654 923 | 156 | 448 105 | 0.35 | 1.15 (0.91-1.45) |
| Saxagliptin hydrochloride | 123 216 | 20 | 93 256 | 0.21 | 123 216 | 25 | 86 928 | 0.29 | 0.97 (0.56-1.67) |
| Linagliptin | 119 316 | 42 | 76 012 | 0.55 | 119 316 | 85 | 71 103 | 1.20 | 1.68 (1.16-2.43) |
Abbreviations: DPP-4, dipeptidyl peptidase-4; HR, hazard ratio; IR, incidence rate.
Available in Medicare only.
In the sensitivity analysis of bullous pemphigoid outcome definition based on 2 or more diagnosis codes (conducted in Medicare only), the numbers of events in the propensity score–matched population for DPP-4 inhibitor vs sulfonylurea use were 20 and 18, respectively, and the IRs (per 1000 person-years) were 0.11 and 0.09, respectively. Numerically elevated risk of bullous pemphigoid was seen in the DPP-4 inhibitor compared with the sulfonylurea group, although it was not statistically significant (HR, 1.25; 95% CI, 0.66-2.36).
Discussion
Our results are consistent with those of several previous studies that have reported a positive association between DPP-4 inhibitors and bullous pemphigoid. However, the absolute risk of bullous pemphigoid has not been well characterized, and the magnitude of the association varied widely among published studies. Our study directly compared DPP-4 inhibitors against the most commonly prescribed second-line antidiabetic agent, second-generation sulfonylureas, using 2 large US commercial claims and Medicare databases. We found that the overall IR of bullous pemphigoid was low in the DPP-4 inhibitor group (0.42 per 1000 person-years) and in the sulfonylurea group (0.31 per 1000 person-years). These IRs were nevertheless higher than the IRs identified in the general population of the United Kingdom (0.043 per 1000 person-years)22 and Sweden (0.071 per 1000 person-years).23 Depending on the study protocol, inclusion of diagnosis codes for ocular, cicatricial, juvenile, and not otherwise classified pemphigoid subtypes may lead to differing IRs.18,23 In our study—especially for the ICD-10 era—we included only the diagnosis code for bullous pemphigoid (L12.0) and no other subtypes.
The Medicare and Optum databases in the study signaled the direction of elevated risk of bullous pemphigoid in those who initiated DPP-4 inhibitor use, but it was most evident in the Medicare database. Because the Medicare database mostly consists of beneficiaries 65 years and older and bullous pemphigoid risk peaks around 70 years of age,23 the HR in our meta-analysis appears to be mainly driven by the mean age of the population, as suggested by the age subgroup analysis.
In the subgroup analyses, we noted a higher risk of bullous pemphigoid in patients 65 years or older, those who were white, and those exposed to linagliptin. The risk was elevated in both men and women. Previous studies showed that vildagliptin was frequently associated with the highest risk of bullous pemphigoid.12,13,14,15 However, vildagliptin is currently unavailable in the US market; among the available agents, linagliptin was the only agent with a statistically significant association. The possible link between linagliptin and bullous pemphigoid was also suggested in a recent randomized clinical trial—Cardiovascular and Renal Microvascular Outcome Study With Linagliptin in Patients With Type 2 Diabetes Mellitus—in which 7 bullous pemphigoid events (0.2%) were reported vs 0 in the placebo arm.24 Our study using 3 large databases provides additional evidence of this association.
Despite all the signals, the mechanism behind an increased risk of bullous pemphigoid with DPP-4 inhibitors remains to be elucidated. One recent study that assessed a pharmacodynamic and pharmacokinetic association between DPP-4 inhibitors and bullous pemphigoid25 reported significant linear correlation between bullous pemphigoid and affinity at the DPP-4 signal but not with other similar target enzymes, including DPP-2, DPP-8, and DPP-9. The DPP-4 enzyme, also known as CD26, cleaves various incretin hormones, such as glucagonlike peptide-1 and glucose-dependent insulinotropic polypeptide.26 In patients with type 2 diabetes, the DPP-4 inhibitor’s mechanism of action can be attributed to prolonged activity of glucagonlike peptide-1– or glucose-dependent insulinotropic polypeptide–dependent secretion of insulin induced by increased blood glucose level.27,28 However, DPP-4 is also able to activate plasminogen and form plasmin on the cell surface to act as a serine protease in cleaving the bullous pemphigoid antigen 180. Inhibition of plasmin by DPP-4 inhibitors has been linked with a potential change in the proper cleavage of bullous pemphigoid antigen 180 and alteration of its antigenicity.29 In addition, inhibition of DPP-4 has been postulated to enhance the activity of proinflammatory chemokines, such as Eotaxin-1/CCL11, which can activate eosinophils in the skin and contribute to tissue damage and blister formation.27,30 However, phenotypically and histologically, no clear differences were found in DPP-4 inhibitor–associated bullous pemphigoid compared with non–drug-induced subtype.27,31 More research is warranted on the mechanism behind the development of autoimmune dermatoses.
Strengths and Limitations
The strengths of our study include analyses of 3 large databases that encompass a commercially insured population as well as an older population covered by Medicare in the United States. We used a new-user, active comparator design32,33 and propensity score matching to control for a large number of confounders. Several subgroup analyses were also attempted to explore possible effect modifications. Among these, race-stratified results deserve attention, because racial diversity is a potential strength of the Medicare data. No racial/ethnic predilection has been previously reported in bullous diseases, but future studies may explore possible differences in the disease risk depending on race/ethnicity.
The following limitations of our findings should be acknowledged. This study is subject to residual confounding despite propensity score matching, because the data used could not provide information on variables such as laboratory values (serum albumin or erythrocyte sedimentation rate), which may be associated with diabetes and bullous pemphigoid.34,35 Outcome misclassification is another possibility because a single occurrence of ICD-based code may be insufficient in differentiating different bullous diseases.36 However, given the uniqueness of the clinical presentation of bullous diseases, even a single diagnosis code of bullous pemphigoid can indicate a clinically meaningful dermatologic adverse reaction. Furthermore, numerically elevated risk of DPP-4 inhibitors was still seen in the sensitivity analysis with the requirement of 2 or more diagnoses of bullous pemphigoid. The sensitivity analysis conducted during our short follow-up time in the Medicare data (median, 135 days) resulted in fewer cases, which may be responsible for the wide 95% CIs. Also, some patients with stable disease who did not need frequent encounters might have been missed. In general, the low number of total events in our and all previous bullous pemphigoid studies demonstrates the challenges in conducting cohort studies with a rare outcome. We tried to overcome this limitation by analyzing multiple databases. Another limitation is the lack of specificity of the ICD-9 coding system for bullous pemphigoid (694.5, which indicates pemphigoid and may include other bullous disease). Future work on the validation of claims-based algorithms for bullous pemphigoid will be helpful to assess our outcome definition. Last, as with the usual challenges of conducting a real-world database study in the US claims data, we could not observe outcomes in disenrolled patients and could not account for a potential overlap among the 3 databases. However, such overlap would only affect the precision of the point estimates and not their magnitude.
Conclusions
Our study shows that there is an increased risk of bullous pemphigoid among DPP-4 inhibitor users compared with second-generation sulfonylurea users. However, DPP-4 inhibitor–associated bullous pemphigoid is rare and may remit after discontinuing the culprit drug. Moreover, not all bullous pemphigoid cases are associated with serious morbidity or mortality. Therefore, although clinicians should not avoid DPP-4 inhibitors entirely in patients with type 2 diabetes who may be otherwise good candidates for the therapy, they should be aware of the potential risk of bullous pemphigoid in DPP-4 inhibitor use, especially in subgroups of older and white patients and linagliptin users.
eFigure. Flowchart of the Study Population Selection
eTable 1. Baseline Characteristics of Study Population Before and After Propensity Score-Matching in Optum, MarketScan, and Medicare
eTable 2. Database-Specific and Combined Incidence Rates and Hazard Ratios of Bullous Pemphigoid in Second-generation Sulfonylurea and DPP-4 Inhibitor Initiators for Subgroup Analyses
References
- 1.Karagiannis T, Paschos P, Paletas K, Matthews DR, Tsapas A. Dipeptidyl peptidase-4 inhibitors for treatment of type 2 diabetes mellitus in the clinical setting: systematic review and meta-analysis. BMJ. 2012;344:e1369. doi: 10.1136/bmj.e1369 [DOI] [PubMed] [Google Scholar]
- 2.DeVries JH, Rosenstock J. DPP-4 inhibitor–related pancreatitis: rare but real! Diabetes Care. 2017;40(2):161-163. doi: 10.2337/dci16-0035 [DOI] [PubMed] [Google Scholar]
- 3.Packer M. Do DPP-4 inhibitors cause heart failure events by promoting adrenergically mediated cardiotoxicity? clues from laboratory models and clinical trials. Circ Res. 2018;122(7):928-932. doi: 10.1161/CIRCRESAHA.118.312673 [DOI] [PubMed] [Google Scholar]
- 4.Mascolo A, Rafaniello C, Sportiello L, et al. Dipeptidyl peptidase (DPP)-4 inhibitor–induced arthritis/arthralgia: a review of clinical cases. Drug Saf. 2016;39(5):401-407. doi: 10.1007/s40264-016-0399-8 [DOI] [PubMed] [Google Scholar]
- 5.US Food and Drug Administration. FDA drug safety communication: FDA warns that DPP-4 inhibitors for type 2 diabetes may cause severe joint pain. Published August 28, 2015. Accessed September 29, 2019. https://www.fda.gov/drugs/drug-safety-and-availability/fda-drug-safety-communication-fda-warns-dpp-4-inhibitors-type-2-diabetes-may-cause-severe-joint-pain
- 6.Desai S, Brinker A, Swann J, Iyasu S. Sitagliptin-associated drug allergy: review of spontaneous adverse event reports. Arch Intern Med. 2010;170(13):1169-1171. doi: 10.1001/archinternmed.2010.188 [DOI] [PubMed] [Google Scholar]
- 7.Dicker D. DPP-4 inhibitors: impact on glycemic control and cardiovascular risk factors. Diabetes Care. 2011;34(suppl 2):S276-S278. doi: 10.2337/dc11-s229 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Schmidt E, Zillikens D. Pemphigoid diseases. Lancet. 2013;381(9863):320-332. doi: 10.1016/S0140-6736(12)61140-4 [DOI] [PubMed] [Google Scholar]
- 9.Skandalis K, Spirova M, Gaitanis G, Tsartsarakis A, Bassukas ID. Drug-induced bullous pemphigoid in diabetes mellitus patients receiving dipeptidyl peptidase-IV inhibitors plus metformin. J Eur Acad Dermatol Venereol. 2012;26(2):249-253. doi: 10.1111/j.1468-3083.2011.04062.x [DOI] [PubMed] [Google Scholar]
- 10.Béné J, Moulis G, Bennani I, et al. ; French Association of Regional PharmacoVigilance Centres . Bullous pemphigoid and dipeptidyl peptidase IV inhibitors: a case-noncase study in the French Pharmacovigilance Database. Br J Dermatol. 2016;175(2):296-301. doi: 10.1111/bjd.14601 [DOI] [PubMed] [Google Scholar]
- 11.García M, Aranburu MA, Palacios-Zabalza I, Lertxundi U, Aguirre C. Dipeptidyl peptidase-IV inhibitors induced bullous pemphigoid: a case report and analysis of cases reported in the European pharmacovigilance database. J Clin Pharm Ther. 2016;41(3):368-370. doi: 10.1111/jcpt.12397 [DOI] [PubMed] [Google Scholar]
- 12.Lee SG, Lee HJ, Yoon MS, Kim DH. Association of dipeptidyl peptidase 4 inhibitor use with risk of bullous pemphigoid in patients with diabetes. JAMA Dermatol. 2019;155(2):172-177. doi: 10.1001/jamadermatol.2018.4556 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Varpuluoma O, Försti AK, Jokelainen J, et al. Vildagliptin significantly increases the risk of bullous pemphigoid: a Finnish nationwide registry study. J Invest Dermatol. 2018;138(7):1659-1661. doi: 10.1016/j.jid.2018.01.027 [DOI] [PubMed] [Google Scholar]
- 14.Benzaquen M, Borradori L, Berbis P, et al. Dipeptidyl peptidase IV inhibitors, a risk factor for bullous pemphigoid: retrospective multicenter case-control study from France and Switzerland. J Am Acad Dermatol. 2018;78(6):1090-1096. doi: 10.1016/j.jaad.2017.12.038 [DOI] [PubMed] [Google Scholar]
- 15.Kridin K, Bergman R. Association of bullous pemphigoid with dipeptidyl-peptidase 4 inhibitors in patients with diabetes: estimating the risk of the new agents and characterizing the patients. JAMA Dermatol. 2018;154(10):1152-1158. doi: 10.1001/jamadermatol.2018.2352 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Schaffer C, Buclin T, Jornayvaz FR, et al. Use of dipeptidyl-peptidase IV inhibitors and bullous pemphigoid. Dermatology. 2017;233(5):401-403. doi: 10.1159/000480498 [DOI] [PubMed] [Google Scholar]
- 17.Suissa S, Azoulay L. Metformin and the risk of cancer: time-related biases in observational studies. Diabetes Care. 2012;35(12):2665-2673. doi: 10.2337/dc12-0788 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Douros A, Rouette J, Yin H, Yu OHY, Filion KB, Azoulay L. Dipeptidyl peptidase 4 inhibitors and the risk of bullous pemphigoid among patients with type 2 diabetes. Diabetes Care. 2019;42(8):1496-1503. doi: 10.2337/dc19-0409 [DOI] [PubMed] [Google Scholar]
- 19.Montvida O, Shaw J, Atherton JJ, Stringer F, Paul SK. Long-term trends in antidiabetes drug usage in the US: real-world evidence in patients newly diagnosed with type 2 diabetes. Diabetes Care. 2018;41(1):69-78. doi: 10.2337/dc17-1414 [DOI] [PubMed] [Google Scholar]
- 20.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 21.Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083-3107. doi: 10.1002/sim.3697 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Langan SM, Smeeth L, Hubbard R, Fleming KM, Smith CJP, West J. Bullous pemphigoid and pemphigus vulgaris—incidence and mortality in the UK: population based cohort study. BMJ. 2008;337(7662):a180-a180. doi: 10.1136/bmj.a180 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Thorslund K, Seifert O, Nilzén K, Grönhagen C. Incidence of bullous pemphigoid in Sweden 2005-2012: a nationwide population-based cohort study of 3761 patients. Arch Dermatol Res. 2017;309(9):721-727. doi: 10.1007/s00403-017-1778-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rosenstock J, Perkovic V, Johansen OE, et al. ; CARMELINA Investigators . Effect of linagliptin vs placebo on major cardiovascular events in adults with type 2 diabetes and high cardiovascular and renal risk: the CARMELINA randomized clinical trial. JAMA. 2019;321(1):69-79. doi: 10.1001/jama.2018.18269 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Carnovale C, Mazhar F, Arzenton E, et al. Bullous pemphigoid induced by dipeptidyl peptidase-4 (DPP-4) inhibitors: a pharmacovigilance-pharmacodynamic/pharmacokinetic assessment through an analysis of the Vigibase®. Expert Opin Drug Saf. 2019;18(11):1099-1108. doi: 10.1080/14740338.2019.1668373 [DOI] [PubMed] [Google Scholar]
- 26.Tasanen K, Varpuluoma O, Nishie W. Dipeptidyl peptidase-4 inhibitor–associated bullous pemphigoid. Front Immunol. 2019;10:1238-1238. doi: 10.3389/fimmu.2019.01238 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Remm F, Franz WM, Brenner C. Gliptins and their target dipeptidyl peptidase 4: implications for the treatment of vascular disease. Eur Heart J Cardiovasc Pharmacother. 2016;2(3):185-193. doi: 10.1093/ehjcvp/pvv044 [DOI] [PubMed] [Google Scholar]
- 28.Yazbeck R, Howarth GS, Abbott CA. Dipeptidyl peptidase inhibitors, an emerging drug class for inflammatory disease? Trends Pharmacol Sci. 2009;30(11):600-607. doi: 10.1016/j.tips.2009.08.003 [DOI] [PubMed] [Google Scholar]
- 29.Izumi K, Nishie W, Mai Y, et al. Autoantibody profile differentiates between inflammatory and noninflammatory bullous pemphigoid. J Invest Dermatol. 2016;136(11):2201-2210. doi: 10.1016/j.jid.2016.06.622 [DOI] [PubMed] [Google Scholar]
- 30.Forssmann U, Stoetzer C, Stephan M, et al. Inhibition of CD26/dipeptidyl peptidase IV enhances CCL11/eotaxin-mediated recruitment of eosinophils in vivo. J Immunol. 2008;181(2):1120-1127. doi: 10.4049/jimmunol.181.2.1120 [DOI] [PubMed] [Google Scholar]
- 31.Plaquevent M, Tétart F, Fardet L, et al. ; French Investigators for Skin Adverse Reaction to Drugs; French Study Group on Autoimmune Bullous Skin Diseases . Higher frequency of dipeptidyl peptidase-4 inhibitor intake in bullous pemphigoid patients than in the French general population. J Invest Dermatol. 2019;139(4):835-841. doi: 10.1016/j.jid.2018.10.045 [DOI] [PubMed] [Google Scholar]
- 32.Schneeweiss S. A basic study design for expedited safety signal evaluation based on electronic healthcare data. Pharmacoepidemiol Drug Saf. 2010;19(8):858-868. doi: 10.1002/pds.1926 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yoshida K, Solomon DH, Kim SC. Active-comparator design and new-user design in observational studies. Nat Rev Rheumatol. 2015;11(7):437-441. doi: 10.1038/nrrheum.2015.30 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Rzany B, Partscht K, Jung M, et al. Risk factors for lethal outcome in patients with bullous pemphigoid: low serum albumin level, high dosage of glucocorticosteroids, and old age. Arch Dermatol. 2002;138(7):903-908. doi: 10.1001/archderm.138.7.903 [DOI] [PubMed] [Google Scholar]
- 35.Jun JE, Lee SE, Lee YB, et al. Increase in serum albumin concentration is associated with prediabetes development and progression to overt diabetes independently of metabolic syndrome. PLoS One. 2017;12(4):e0176209. doi: 10.1371/journal.pone.0176209 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hsu D, Brieva J, Nardone B, Silverberg JI. Validation of database search strategies for the epidemiological study of pemphigus and pemphigoid. Br J Dermatol. 2016;174(3):645-648. doi: 10.1111/bjd.14172 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure. Flowchart of the Study Population Selection
eTable 1. Baseline Characteristics of Study Population Before and After Propensity Score-Matching in Optum, MarketScan, and Medicare
eTable 2. Database-Specific and Combined Incidence Rates and Hazard Ratios of Bullous Pemphigoid in Second-generation Sulfonylurea and DPP-4 Inhibitor Initiators for Subgroup Analyses