Only individuals with psoriasis and DM were included in this analysis.

Patients were considered exposed to study drugs from treatment initiation until the last administered dose, November 2024, or the censoring date for patients lost to follow-up. For these individuals, the censoring date corresponded to their last dermatology visit. Pharmacologic groups included classic systemic therapies, such as acitretin (ACT), methotrexate (MTX), and cyclosporine (CYA); phosphodiesterase-4 (PDE4) inhibitors, such as apremilast (APR); tumor necrosis factor alpha (TNF-α) inhibitors, including adalimumab, certolizumab, etanercept, infliximab, and corresponding biosimilars; interleukin (IL)-12/23 inhibitors, such as ustekinumab and biosimilars; IL-23 inhibitors, including guselkumab, risankizumab, and tildrakizumab; and IL-17 inhibitors, including brodalumab, ixekizumab, bimekizumab, and secukinumab.

Patients could contribute more than 1 treatment cycle during follow-up and, therefore, could be included in more than 1 treatment group over time. Treatment groups were defined at the treatment-cycle level, independently of outcome occurrence. DM-AEs were attributed to a given treatment cycle if they occurred between treatment initiation and the last dose or within 90 days after the last dose.11 Patients receiving combination therapy were excluded from this analysis.

Based on previous literature, we defined DM-AEs as events involving the development or progression of clinically recognized diabetic complications, including macrovascular outcomes (e.g., MACE4), microvascular complications (e.g., diabetic nephropathy, retinopathy, neuropathy, and diabetic foot disorders, including ulcers and osteomyelitis), and acute metabolic disturbances, such as hypoglycemia, hyperosmolar hyperglycemic state, and diabetic ketoacidosis.2,12,13 We also prespecified new-onset or clinically meaningful worsening of hyperglycemia and hypertriglyceridemia as DM-AEs because these drug-modifiable metabolic abnormalities are diabetes related, prompt clinical intervention, and are established intermediates associated with acute metabolic decompensation and downstream micro- and macrovascular complications.

To ensure clinical relevance, DM-AEs were restricted to specific complications with established associations with diabetes. Nonspecific conditions, such as frequent infections (e.g., candidiasis), which are more common in patients with DM but are also associated with immunosuppressive treatments and lack specific coding and causality, were excluded. Inclusion of these conditions could introduce heterogeneity and dilute results, particularly in the context of real-world evidence.

The classification of an AE as DM related was predefined exclusively through review of MedDRA codes in the database before any statistical analysis was performed.

The primary outcome measure was the adjusted incidence rate ratio (aIRR) for DM-AEs and DM-SAEs among patients receiving MTX compared with those receiving other systemic therapies.

Categorical variables are presented as No. (%), and continuous variables as mean (SD) or median (IQR), as appropriate. Descriptive data were compared between groups using the Student t test or Mann–Whitney U test and the Pearson's chi-square test (or Fisher exact test when appropriate), depending on variable distribution.

Missing data were assumed to be missing at random and were handled using multiple imputation by chained equations (20 datasets), with estimates pooled according to Rubin rules after convergence.

Incidence rates (IRs) were calculated as the number of new DM-related complications per 1000 person-years (PYs) of exposure.

We estimated adjusted incidence rate ratios (aIRRs) and 95% CIs for DM-AEs and DM-SAEs by comparing IRs in the MTX group with those of each treatment group using robust Poisson regression models. MTX was selected as the comparator for patients with DM despite not being recommended in some clinical guidelines for this population. This decision was based on its widespread use in routine clinical practice, where MTX remains one of the most frequently prescribed conventional systemic therapies for psoriasis and is often used as first-line systemic treatment, including in patients with metabolic comorbidities. Its use is further driven by structural and organizational factors, including low cost, broad accessibility, and frequent requirement by hospital pharmacy protocols before initiation of biologic therapies. Therefore, using MTX as the reference treatment reflects routine clinical decision-making in public healthcare settings rather than guideline-based recommendations alone.14

To mitigate confounding bias resulting from nonrandomized treatment assignment, we calculated a propensity score (PS) for the indication of each drug relative to MTX. The PS was calculated using variables associated with both exposure and outcome15: sex, age, smoking status, alcohol consumption, disease duration, presence of psoriatic arthritis, body mass index (BMI), number of previous cardiovascular comorbidities, Psoriasis Area and Severity Index (PASI), and prior classic systemic treatments. These variables were selected based on clinical expertise and previously identified significant confounders.16 All confounders were assessed at baseline. Propensity scores were incorporated into the Poisson regression models as inverse probability of treatment weights (IPTW).

All analyses were performed using Stata Statistical Software, release 17 (StataCorp LLC). A two-sided P<.05 was considered statistically significant.

Patient characteristics are summarized by treatment group in Table 1.

The mean age was 67.5 years (SD, 13.2); 41% of patients were women, and plaque psoriasis was the most common clinical subtype (84%). Mean baseline PASI was 11.3 (SD, 7.8). Mean BMI was 31kg/m2 (SD, 5.6).

Cardiovascular comorbidities other than DM, such as hypertension (69%) and hypercholesterolemia (70%), were the most prevalent comorbidities. Only 17% of patients had no previous cardiovascular comorbidities other than DM. Noncardiovascular comorbidities were less frequent, and 63% of patients had none. Fifty-six percent of patients were current or former smokers.

A total of 298 DM-AEs and 83 DM-SAEs were reported during 2865 PYs of exposure.

The most frequent DM-AEs included hypertriglyceridemia (n=31), hyperglycemia (n=24), hypertension (n=15), increased creatinine levels (n=11), and heart failure (n=10).

The overall IR of DM-AEs ranged from 51.9 (95% CI, 28.7–93.7) per 1000 PYs in patients treated with APR to 457.9 (95% CI, 260.0–806.2) per 1000 PYs among those receiving CYA. Regarding DM-SAEs, the lowest IR was also observed in the APR group at 9.4 (95% CI, 2.4–37.7), whereas the highest DM-SAE rate occurred in the CYA-treated group, with 76.3 (95% CI, 19.1–305.1) per 1000 PYs. The IR of DM-AEs in the MTX group was 110.7 (95% CI, 77.9–157.5), with a DM-SAE rate of 32.1 (95% CI, 16.7–61.8). Detailed IRs for each treatment group are presented in Table 2.

Compared with MTX, CYA showed the highest crude IRR for all adverse events (IRR, 4.1; 95% CI, 2.1–8.1; P<.001), followed by ACT (IRR, 1.7; 95% CI, 1.1–2.6; P=.020). Conversely, APR demonstrated a significantly lower crude risk (IRR, 0.5; 95% CI, 0.2–0.9; P=.031).

After adjustment for confounders, CYA remained associated with the highest aIRR (aIRR, 7.5; 95% CI, 3.3–17.3; P<.001), and ACT also continued to show a significantly increased adjusted risk (aIRR, 2.1; 95% CI, 1.2–3.7; P=.014). APR remained significantly associated with a reduced adjusted risk of DM-AEs compared with MTX (aIRR, 0.3; 95% CI, 0.1–0.6; P=.003).

In subgroup analyses by sex, female patients treated with IL-12/23 inhibitors showed a significantly lower crude risk of DM-related adverse events compared with MTX (IRR, 0.3; 95% CI, 0.1–0.8; P=.019); however, this association was not retained after adjustment (aIRR, 0.4; 95% CI, 0.1–1.0; P=.054).

The remaining systemic treatments did not show statistically significant increases or reductions in the risk of DM-AEs compared with MTX.

Regarding DM-SAEs, no statistically significant differences were observed between treatment groups and MTX in either crude or adjusted analyses (all P>.05), although APR (aIRR, 0.1; 95% CI, 0.0–1.4; P=.092) and CYA (aIRR, 5.6; 95% CI, 0.8–37.7; P=.075) remained close to statistical significance.

Fig. 1.

Adjusted incidence rate ratios (aIRR) for diabetes-related adverse events by treatment group, referenced to methotrexate. (A) All diabetes-related adverse events and (B) serious diabetes-related adverse events. Points indicate adjusted aIRR and horizontal bars the 95% confidence intervals; the red line marks IRR=1 (MTX reference). Estimates come from IPTW-weighted models with robust standard errors, adjusted for age, sex, body mass index, psoriatic arthritis, PASI, number of cardiovascular comorbidities, and center. Treatment groups: TNFi (anti-TNF), IL-12/23i, IL-17i, IL-23i, APR (apremilast), ACT (acitretin), CYA (cyclosporine).

MTX was selected as the reference treatment in our study because of its widespread use and role as first-line systemic therapy in psoriasis management. However, MTX has been associated with a significantly increased risk of hepatic fibrosis in patients with DM (adjusted HR, 2.40; 95% CI, 1.05–5.51).17 This elevated risk has been linked to reduced renal clearance in patients with diabetic nephropathy, potentially increasing hepatotoxicity.3 Although short-term studies have not demonstrated direct hyperglycemic effects of MTX,18 current clinical guidelines recommend lower starting doses, appropriate dose adjustments, and rigorous monitoring in patients with DM to minimize potential safety risks.19

CYA has been associated with hyperlipidemia, hypertension, nephrotoxicity, and an increased risk of MACE in patients with psoriasis.4,20,21 Additionally, CYA exhibits diabetogenic effects through inhibition of insulin secretion from pancreatic islet cells.3,22 These findings are consistent with the elevated risk of DM-related complications observed in our cohort. Given these data, CYA is generally not recommended for the treatment of psoriasis in patients with DM.3

ACT therapy may induce hyperlipidemia, representing an additive cardiovascular risk factor in patients with DM.22 Its effects on glycemic control in patients with DM remain insufficiently established.23,24 Nevertheless, our findings suggest that ACT use in patients with DM should be limited because it was associated with an increased risk of DM-related AEs compared with MTX.

APR has demonstrated favorable effects on metabolic parameters in patients with psoriasis, including those with DM,25 as well as reductions in MACE incidence.4 APR has also been reported to modestly reduce body weight in a subset of patients, independently of gastrointestinal adverse effects such as nausea and diarrhea. Additionally, APR may positively influence glucose metabolism and potentially enhance the efficacy of metformin.22,26 These observations are consistent with the findings of our cohort.

In our cohort, biologic therapies were not associated with statistically significant increases or reductions in the risk of DM-related AEs or SAEs compared with MTX. Available data regarding DM-related complications in patients with psoriasis remain limited.

TNF inhibitors may improve insulin sensitivity in patients with psoriasis by counteracting inflammation-driven insulin resistance.27,28 However, in a retrospective registry study, Kalb et al. reported that the presence of DM was a significant predictor of serious infection in patients treated with adalimumab, etanercept, or ustekinumab (HR, 1.7; 95% CI, 1.25–2.23).29

IL-17 inhibitors have demonstrated neutral30,31 or beneficial 32 effects on cardiovascular parameters and have been associated with lower cardiovascular disease risk compared with MTX.4 However, these agents may further increase the risk of candidiasis, a well-recognized class AE, particularly in patients with DM.32–35

The metabolic profile of IL-23 inhibitors in patients with psoriasis and DM appears favorable. These therapies have been associated with improvements in fasting glucose, insulin levels, triglycerides, and inflammatory markers.32 IL-23 inhibitors have not been associated with increased risks of metabolic complications, such as hyperglycemia, diabetic ketoacidosis, cardiovascular events, or serious infections, compared with the overall psoriasis population.33–35

This study has several limitations. As an observational analysis conducted in a high-risk population, residual and unmeasured confounding factors (e.g., physical activity, diet, baseline diabetes control, and time-varying metabolic control) cannot be excluded despite IPTW adjustment. Therefore, findings should be interpreted as associations rather than causal effects. Diabetes status was recorded only at baseline, and information regarding antidiabetic therapies (e.g., insulin, glucagon-like peptide-1 receptor agonists, sodium-glucose cotransporter 2 inhibitors) and changes in metabolic control during follow-up was unavailable. These concomitant treatments may have modified the observed associations.36

MTX was selected as a pragmatic reference treatment instead of placebo, topical therapy, or alternative systemic therapies because it remains widely prescribed in routine psoriasis care. Nevertheless, several clinical guidelines do not recommend MTX in patients with DM. DM-AEs were predefined using MedDRA codes to reduce adjudication bias; however, misclassification remains possible, and many reported signals were laboratory defined (e.g., dyslipidemia and hyperglycemia). Although statistically significant, some findings may have limited clinical relevance and should primarily guide monitoring strategies and optimization of cardiovascular risk factors.

Certain exposure groups, particularly CYA, were relatively small, resulting in wide confidence intervals, and follow-up duration may have been insufficient to fully assess rare or long-latency outcomes. Differential monitoring across therapies and channeling by indication may have persisted despite propensity weighting, and exclusion of combination therapy limits generalizability to monotherapy regimens. Because patients could contribute multiple treatment cycles, observations were not fully independent. Although robust variance estimation was used to mitigate within-patient correlation, residual correlation and treatment-sequencing effects cannot be excluded.

This study also has several strengths. Its prospective design and use of the BIOBADADERM registry, which includes continuous pharmacovigilance procedures, provided high-quality, systematically collected real-world evidence that is uncommon in observational studies. The large multicenter sample across Spain improves generalizability. Inclusion of all systemic therapies routinely used in clinical practice enabled comprehensive treatment comparisons. Major confounders, including age, lifestyle factors (e.g., smoking and alcohol consumption), and cardiovascular and noncardiovascular comorbidities, were addressed using robust adjustment methods. Classification of DM-related adverse events was performed before analysis and without knowledge of study results, reducing adjudication bias. Finally, consistency between our findings, established mechanistic pathways, and previous evidence supports the validity of the observed associations.