Baricitinib Safety for Events of Special Interest in Populations at Risk: Analysis from Randomised Trial Data Across Rheumatologic and Dermatologic Indications

Safety data from nine randomised clinical trials [Phase 1b: I4V-MC-JADB; Phase 2: NCT01185353, NCT00902486; NCT01469013 (Japan); Phase 3: NCT01710358 (RA-BEAM), NCT01721044 (RA-BEACON), NCT01721057 (RA-BUILD), NCT01711359 (RA-BEGIN), NCT02265705 (RA-BALANCE)] and one completed long-term extension (LTE) trial [NCT01885078 (RA-BEYOND)] running from May 2009–November 2020 were pooled. The all-bari-RA analysis set includes data of all patients who received at least one dose of baricitinib (doses ranged from 1 to 15 mg daily, with 2 mg and 4 mg daily doses used in the phase 3 and long-term extension trials) using all available data after the first dose without censoring for rescue or dose change.

Safety data are included from six double-blinded, randomised clinical studies [Phase 2: NCT02576938; Phase 3: NCT03334396 (BREEZE-AD1), NCT03334422 (BREEZE-AD2), NCT03428100 (BREEZE-AD4), NCT03435081 (BREEZE-AD5), NCT03733301 (BREEZE-AD7)], one double-blinded, randomised LTE [NCT03334435 (BREEZE-AD3)], and one open-label LTE [NCT03559270 (BREEZE-AD6)], with completed studies running from November 2017–August 2019 and data cut-offs of 15 December 2021 for BREEZE-AD4, and 3 November 2021 (BREEZE-AD3) and 21 December 2021 (BREEZE-AD6) in the ongoing LTE studies. The all-bari-AD analysis set includes data for all patients who received at least one dose of baricitinib (1 mg, 2 mg, or 4 mg) from any of the eight clinical trials at any time.

Safety data are included from Phase 2 and Phase 3 cohorts of BRAVE-AA1 (NCT03570749) and Phase 3 BRAVE-AA2 (NCT03899259), with data cut-offs of 11 November 2021 and 5 November 2021, respectively. The all-bari-AA analysis set combines all patients exposed to any baricitinib dose (1 mg, 2 mg, or 4 mg) at any time up to the data cut-off.

All studies were conducted in accordance with ethical principles of the Declaration of Helsinki and Good Clinical Practice guidelines, and the research protocols were approved by each centre’s institutional review board or ethics committee. All patients provided written informed consent.

Each analysis set was reviewed for AESI, defined as MACE [positively adjudicated events of myocardial infarction (MI), stroke and cardiovascular deaths combined), MI separately, malignancies (excluding NMSC), VTE (deep vein thrombosis (DVT) or pulmonary embolism (PE)], serious infections, and mortality.

For AESI, the incidence rate (IR) was calculated as standard (number of patients with an event per 100 patient-years at risk (PYR)], including follow-up time censored at event onset date. Poisson distribution was used to calculate 95% CI.

The IR for AESI was evaluated in subgroups of patients aged over 65 years or with at least one of eight specified risk factors [history of ASCVD, hypertension, diabetes mellitus, current smoking, HDL cholesterol less than 40 mg/dL, history of malignancy, BMI 30 kg/m² or greater, and severe mobility impairment on EQ-5D (i.e., responding either ‘I have severe problems in walking about’ or ‘I am unable to walk about’)] and patients aged below 65 years without any of the specified risk factors. Past smoking was not systematically documented in all studies but was checked for in case histories of patients with an event. This IR was calculated as 100 times the number of patients experiencing the event divided by PYR (exposure time up to the event for patients with the AESI and exposure time up to the end of the exposure and 1-month follow-up for patients without the AESI) in years in the subgroup of the specific factor. To account for ageing of the cohort, the number of expected malignancies (excluding NMSC) were calculated using age-specific malignancy data from the Surveillance, Epidemiology, and End Results 17 (SEER17), 2013–2017 US population cancer rates [32] and compared to the number of observed malignancies in the baricitinib clinical programme.

The baricitinib clinical trials for RA did not exclude patients with any of the above-mentioned risk factors, except for recent comorbidities such as MI or stroke within 12 weeks, active or recent (within 30 days) serious infection, or malignant disease [33]. Clinical trials studying baricitinib in AD and AA also excluded patients with MI, stroke, or VTE events within 12 weeks of screening, active or recent (within 30 days) serious infection, or malignant disease, as well as those who had history of, or are considered at high risk for, VTE [34, 35].

In the baricitinib long-term clinical trial dataset, over 60% of patients with RA treated with baricitinib who recorded AESI had at least two risk factors, the most common of which were hypertension and BMI of at least 30 kg/m². These risk factors are not only easily identifiable in this patient population but can be managed to minimise the risk of AESI in these patients. The proportion of patients with RA at risk, specifically those with long-established disease, was higher than for AD and AA, where the AESI were much lower for both the low-risk and at-risk populations.

The IR of MACE in the baricitinib clinical data for the at-risk RA subpopulation of ≥ 65 years old with at least one risk factor was 0.70 per 100 PY and much lower for the RA low-risk population, 0.05 per 100 PY. Both fall within the expected IR of MACE in patients with RA reported in real-world settings (0.27–3.2 per 100 PY) [28]. A retrospective observational study pooling data from disease registries and claims databases, with the French SNDS and Swedish ARTIS data sources contributing the most patients, has investigated the safety of baricitinib compared with TNFi in a matched RA population based on propensity scores for confounding factors. The study showed a numerically, but not significantly, greater risk of MACE in patients receiving baricitinib (incidence rate ratio 1.54, 95% CI 0.93–2.54; incidence rate difference 0.22, 95% CI − 0.07 to 0.52 per 100 PY) [43]. Recently, Hoisnard et al., using data from SNDS after inverse propensity score treatment weighting, showed that there was no difference in risk of MACE (or VTE) between patients initiating a JAKi (60% baricitinib, 40% tofacitinib) and initiating the TNFi adalimumab. A subgroup analysis of patients aged 65 years or older, and with at least one risk factor for cardiovascular disease, also found no evidence of a difference in risk between the JAKi and adalimumab cohorts [44]. Further contextualization of these observational cohort studies with respect to ORAL Surveillance is hard to elucidate due to different study designs and endpoints (Cox model and time to event vs. incidence rates, as well as length of follow-up and methods of addressing confounding factors through propensity scores vs. inverse probabilities).

For the baricitinib clinical programme in patients with AD, IR for MACE in the at-risk subpopulation (0.25 per 100 PY) lies within the expected rates for the AD patient population reported in real-world settings (IR of 0.15–0.63 per 100 PY), with the low-risk AD baricitinib clinical population having a much lower IR (0.04) [28]. The IRs for patients with AA in the baricitinib clinical studies (subpopulation at-risk and low-risk, 0.10 and 0 per 100 PY, respectively) are the lowest of those observed in the baricitinib studies, and reflect that the general population of patients with AA do not have an increase in risk of MACE.

For VTEs, the IR in the baricitinib clinical programme for the at-risk RA subgroup (0.66 per 100 PY) falls within the expected IR reported in the literature for patients with RA of 0.33–0.79 per 100 PY, with a minimal IR in the low-risk patient population from the baricitinib studies (0.09 per 100 PY). However, a recent study using data from the Swedish Rheumatology Quality Register found a higher incidence of VTE in patients receiving JAKi (IR = 11.33 per 1000 PY) compared with both TNFi (IR = 5.15) and the general RA population (IR = 5.86) [45]. Similarly, Salinas et al. found increased risk of VTE with baricitinib versus TNFi (incidence rate ratio 1.51, 95% CI 1.10–2.08; incidence rate difference 0.26, 95% CI − 0.04 to 0.57 per 100 PY) [43]. However, Salinas et al. made no adjustment based on previous treatment failures or disease activity, nor carried out any analysis on patient subpopulations with or without relevant risk factors. As highlighted before, Hoisnard et al. did not show differences for VTEs in patients treated with JAKi compared with those who received adalimumab using SNDS data, including in a subpopulation similar to the at-risk population of the baricitinib studies presented here (aged 65 years or older with at least one risk factor for cardiovascular disease) [44]. The occurrence of VTE in at-risk patients who received baricitinib treatment in the RA clinical programme is thus associated with previously identified independent VTE risk factors, such as previous VTE, older age, obesity, malignancy, and immobility [46]. The risk of VTE is reflected in the current SmPC for baricitinib due to an imbalance in observed events during the placebo-controlled period of RA Phase 3 studies [29].

In the baricitinib AD and AA clinical trial programmes, the IRs of VTE in at-risk baricitinib treatment groups were low (IR of 0.08 and 0.10 per 100 PY, respectively), as expected due to the younger age and lower vulnerability of these study populations, both in the clinical studies and the general populations of patients with these conditions. These are in line with, or below, the background IR of 0.18–0.24 per 100 PY in patients with AD and 0.09 per 100 PY in patients with AA. Although 2 of the patients with VTE in the AD population were receiving hormonal contraception, overall approximately 30% of female patients were taking hormonal contraceptives during these studies. A risk factor analysis performed on data from the baricitinib RA clinical programme did not identify oral contraceptive use as an independent risk factor in RA [46]. This analysis was not performed for the AD baricitinib data due to the low number of events.

In this analysis of the baricitinib clinical trial programmes, the IR of malignancies excluding NMSC in the LTE at-risk RA population was 1.23 per 100 PY, which is slightly elevated compared to the all-bari-RA population (IR = 0.92 per 100 PY). When comparing the all-bari-RA population to the expected number of malignancies based on SEER17 data, the age-adjusted standardised IR was 1.07 (95% CI 0.90–1.26) [29], suggesting a modest increase in incidence for these at-risk patients. The IR of malignancies excluding NMSC in the at-risk group in AD (0.45) and AA (0.31) baricitinib clinical trials appear to be at the lower end of the reported IR in these disease populations (AD: 0.33–0.85 per 100 PY [47, 48]; and AA: 0.37–0.43 per 100 PY [20]).

The mortality rate for the baricitinib LTE at-risk RA subpopulation (0.78) is lower than expected when compared with the general population calculated from the SEER incidence distribution (2013–2017) (83 observed deaths vs. 105.9 expected) [32]. IRs of mortality for baricitinib at-risk groups in AD and AA clinical studies are low (0.16 and 0, respectively) and in line with the younger age and lower vulnerability of these study populations.

The risk factors presented in this analysis are those relevant for the adverse events of special interest investigated, as determined by the recommendation of the safety committee of the European Medicines Agency (EMA) following the results of the ORAL Surveillance study to use JAK inhibitors in the following patients only if no suitable treatment alternatives are available: those aged 65 years or above, those at increased risk of major cardiovascular problems (such as heart attack or stroke), those who smoke or have done so for a long time in the past, and those at increased risk of cancer [49]. The Committee also recommended using JAK inhibitors with caution in patients with risk factors for blood clots in the lungs and in deep veins (VTE). The cardiovascular risk factors included in our analysis were chosen to reflect the entry criteria for the ORAL Surveillance study [27]: risk of VTE is linked to older age, increased BMI [50], and immobility, represented by the surrogate factor of poor mobility on EQ-5D, while risk of cancer was determined by age, history of malignancy, and smoking status as the most reliable predictive factors for cancer risk.

Per recent recommendations from committees of the EMA [Pharmacovigilance Risk Assessment Committee (PRAC) and Committee for Medicinal Products for Human Use (CHMP)], the benefit–risk profile of baricitinib for patients with no risk factors remains positive for patients with RA, AD, and AA who have responded inadequately to, or who are intolerant to, cDMARDs [51, 52] or topical treatments [53]. For patients at risk for AESI, benefit–risk shared decision-making discussions should be taken into consideration by the prescriber to evaluate use of baricitinib if no other suitable alternatives are available. Further, PRAC and CHMP recommend the doses of JAK inhibitors should, where possible, be reduced in patient groups who are at risk of VTE, cancer, or major cardiovascular problems [49]. Based on available data across indications for baricitinib, including randomised controlled trials, LTE, and real-world evidence data, a clinically relevant or consistent dose-dependent response with a higher risk for the 4 mg versus the 2 mg dose has not been observed for AESI, except for infections, including herpes zoster, nor for VTE, for which an imbalance between doses was observed during the placebo-controlled period of RA trials [46].

EULAR have recently proposed an updated approach for the management of RA with targeted synthetic disease-modifying antirheumatic drugs (tsDMARDs) and biological DMARDs (bDMARDs), aligned with previous EULAR recommendations in which JAKi remain on the same level as bDMARDs and can be used after conventional DMARD failure in patients who do not have risk factors for cardiovascular disease, VTE, or malignant diseases [54]. EULAR emphasises the importance of shared decision-making, and that, when making recommendations, clinicians give careful consideration to specific risk factors of relevance for the individual patient before initiating JAKi. There are many precedents for the success of mitigation strategies to prevent treatment-associated risks in the case of biologic therapies: for example, screening for latent tuberculosis and hepatitis, cardiac failure, and demyelination risks [55‐57]. Furthermore, regardless of medication, RA management recommendations emphasise the importance of careful screening and management of cardiovascular, infection, and malignancy risks [40]. The contribution of specific risk factors to each of these AESIs remains to be further elucidated. However, all of these known risk factors can be identified in routine clinical practice, and therefore health-care providers should follow established guidelines to manage cardiovascular risk, screen for malignancies, and vaccinate patients, as for any other patients with immune-mediated inflammatory conditions.

For AD, it is recommended to limit systemic corticosteroids to use only for 1–2 weeks for severe acute exacerbations, due to the largely unfavourable risk–benefit ratio [58, 59]. Treatment duration with cyclosporine, the first-line systemic option for patients with severe disease, usually varies from 3 months to 1 year due to poor drug survival [58, 59]. Long-term use of cyclosporine is associated with multiple risks, including hyperlipidaemia, hypertension, kidney toxicity, and malignancy. Therefore, it is important to provide access to treatments with a better long-term benefit–risk balance, including dupilumab and JAKi indicated for AD [60, 61]. Baricitinib has been shown in clinical trials to have fast onset of action, especially on the most limiting AD symptom itch, which is also recognised in the recently published EuroGuiDerm 2022 guideline on atopic eczema (atopic dermatitis) [59], and has been shown to be effective on AD signs and other symptoms, such as skin pain and sleep, leading to an overall improved quality of life [62‐64]. This treatment algorithm is recommended by the European Guidelines for Treatment of AD, who recommend dupilumab and baricitinib with the same level of evidence “in atopic eczema patients who are candidates for systemic treatment”, while recommending against the long-term use of systemic glucocorticosteroids [59].

In most geographies, for patients with severe AA, there is no licensed treatment alternative to baricitinib. In the baricitinib AA clinical trials, baricitinib treatment demonstrated significant hair regrowth of scalp, eyebrows, and eyelashes compared to placebo up to 36 weeks [35]. Approximately half of the population in the baricitinib AA clinical programme are younger than 65 years, have no cardiovascular or thrombotic risk factors, and have limited risk of malignancy or infection. Published population risks for AA differ from the published background risks for the RA population. For example, in a retrospective cohort study, AA was not associated with an increased risk of heart disease [18]. Additionally, a study involving a heterogeneous AA population in the US found that AA was not associated with an increased risk of stroke and MI [65].