Cardiovascular Risk Management in Patients Treated with JAK Inhibitors

Abstract

The Janus kinase (JAK)-signal transducer and activator of transcription (STAT) pathway plays a critical role in the pathogenesis of many immune-mediated inflammatory diseases (IMIDs). Although Janus kinase inhibitors (JAKi) are an effective treatment for several IMIDs, they have come under scrutiny as a class due to a potential risk of venous thromboembolism (VTE) and cardiovascular (CV) events, specifically noted with the oral JAKi, tofacitinib, as reported in the ORAL Surveillance Trial of a high CV risk rheumatoid arthritis population. This trial resulted in a black box warning from the Food and Drug Administration and European Medicines Agency regarding risk of VTE and CV events that was extended across several types of JAKi (including topical ruxolitinib) when treating IMIDs, leading to considerable controversy. Included is an up-to-date review of the current and rapidly evolving literature on CV risk in patients with IMIDs on JAKi therapy, including identification of potential risk factors for future VTE and CV events on JAKi therapy. We suggest a comprehensive, multimodal, and systematic approach for evaluation of CV risk in patients considering taking JAKi and emphasize that cardiologists play an important role in risk stratification and mitigation for patients with high CV risk factors or on long-term JAKi therapies.

INTRODUCTION

Over the last decade, Janus kinase inhibitors (JAKi) have rapidly emerged as an effective treatment option for a wide breadth of diseases. JAKi initially proved efficacious in the treatment of myeloproliferative disorders, gaining US Food and Drug Administration (FDA) approval for their use for these indications in 2011. However, since then, JAKi have garnered much attention for their success in the treatment of immune-mediated inflammatory diseases (IMIDs), with numerous approvals by the US FDA and European Medicines Agency (EMA) for alopecia areata (AA), ankylosing spondylitis (AS), atopic dermatitis (AD), non-radiographic axial spondyloarthritis (axSpA), nonsegmental vitiligo, polyarticular course juvenile idiopathic arthritis (JIA), psoriatic arthritis (PsA), rheumatoid arthritis (RA), and ulcerative colitis (UC). ^1–6 The list of approved JAKi indications is rapidly growing. Despite their promise, there has been significant concern regarding use of FDA-approved JAKi in IMIDs due to an attached black box warning regarding the risk of “serious heart-related events, cancer, blood clots, and death.” ⁷ In this review, we take an in-depth look at the literature on cardiovascular (CV) risk in patients with IMIDs who may require JAKi and provide recommendations regarding CV risk stratification and management. We will also introduce a novel subclass of JAKi, a selective Tyrosine Kinase 2 (TYK2) inhibitor FDA-approved for plaque psoriasis (PsO), in the context of CV risk.

THE ADVENT OF JAK INHIBITORS

The molecular components of the Janus kinase (JAK)-signal transducer and activator of transcription (STAT) pathway were identified 30 years ago. ⁸ The JAK family consists of four different receptor-associated tyrosine kinases (JAK1, JAK2, JAK3, and TYK2) and the STAT family consists of 7 different intracellular transcription factors. ⁹ The JAK-STAT pathway is initiated by the binding of a cytokine to its cognate type I and/or type II cytokine receptor, resulting in activation of a pair of associated JAKs, usually heterodimers (Figure 1). ¹⁰ The JAKs, in turn, mediate the recruitment and phosphorylation of STATs. ^8,10–12 Subsequently, the activated STATs dimerize, translocate to the nucleus, and bind target DNA sequences. ^10,11 Stimulation of the JAK-STAT pathway thus allows for regulation of gene expression, resulting in increased production of proinflammatory cytokines and growth factors. ⁹

Figure 1.

Diagram of the JAK-STAT pathway. Cytokines bind to cognate receptors, resulting in activation of a pair of JAKs through phosphorylation (P). This pair usually consists of heterodimers from the JAK family, consisting of JAK1, JAK2, JAK3, and TYK2. The activated JAKs mediate the activation of STATs, which dimerize, translocate to the nucleus, and regulate gene expression of proinflammatory cytokines and growth factors. JAK inhibitors inhibit this pathway through targeting specific members of the JAK family. This figure was created with BioRender.com

We now know that the JAK-STAT pathway promotes nearly 60 soluble factors (including interleukins, interferons, colony stimulating factors, hormone-like cytokines, and growth factors) which subsequently modulate gene expression and pathways causal in many pro-inflammatory IMIDs. ¹⁰ Consequently, interference with its signal transduction through JAKi has the potential to modulate a broad set of diseases characterized by immune system dysregulation. ^8,11 For example, IL-6, interferons, and granulocyte-macrophage colony-stimulating factor are implicated in the pathogenesis of RA and are all directly promoted by multiple JAK isoforms within the JAK-STAT pathway. ¹² As another example, the IL-23/IL-17 axis plays a prominent role in the pathogenesis of PsA, and signaling of IL-23 is directly modulated by the JAK-STAT pathway through JAK2/TYK2 heterodimer signaling. ¹³

Significant strides have been made in the past several decades in the treatment of IMIDs with biologic disease-modifying antirheumatic drugs (bDMARDs); however, a significant portion of patients remain refractory to therapy and suffer from disease-related morbidity and possible mortality. In RA, for example, it is estimated that only 60% of patients achieve an American College of Rheumatology 20% response with TNF-inhibitors. ¹⁴ JAKi have therefore become essential in the treatment of autoimmunity because they provide an alternative pro-inflammatory pathway to target to promote disease remission and have the potential to minimize glucocorticoid use. ¹² JAKi have shown robust and rapid inhibition of bone erosion in RA, comparable to effects observed with TNF-inhibitors. ¹² Guidelines from the American College of Rheumatology have therefore elevated the recommendation for JAKi to the same level as bDMARDs. ¹⁵ Another example of the success story of JAKi is with severe AA, a disease that has historically been very difficult to treat. In phase 3 trials of patients with severe AA, 35–39% of patients on baricitinib 4mg versus 3–6% of patients on placebo achieved adequate scalp hair coverage after 36 weeks of treatment. ¹⁶ Shortly thereafter, baricitinib became the first FDA-approved systemic therapy for moderate to severe AA. ¹⁷

There are several distinct JAKi in oral or topical formulations that have already been FDA-approved for various IMIDs including: oral abrocitinib, baricitinib, tofacitinib, upadacitinib, and topical ruxolitinib. These JAKi that target JAK1, JAK2, and/or JAK3 isoforms are approved across a range of IMIDs including: AA, AS, AD, non-radiographic axSpA, nonsegmental vitiligo, polyarticular course JIA, PsA, RA, and UC. ^1–6 Additionally, oral deucravacitinib is a new FDA-approved JAKi for plaque PsO that targets TYK2. With over 310 registered clinical trials involving JAKi on ClinicalTrials.gov, we expect this list of approved JAKi and indications to swiftly grow. ¹⁸

As small molecule enzyme inhibitors, JAKi are advantageous to biologics such as TNF-inhibitors given oral administration, no anti-drug antibodies, and short half-life allowing easy discontinuation in the case of infection or surgery. ¹⁹ However, unlike biologics which target specific cytokine pathways, JAKi are less specific towards their target. ¹¹ The aforementioned FDA-approved JAKi that target JAK1-JAK3 bind to active sites within the enzyme’s catalytic domain. However, this is a highly conserved domain across tyrosine kinases, and thus these JAKi can be selective towards one or more of the four different JAK isoforms but they are not 100% specific for these pathways and have the potential to produce off-target effects against non-JAK tyrosine kinases. ^9,11 To date, there are no clear off-target drug effects that explain the potential increased risk of thrombosis with JAKi. ²⁰ JAKi selectivity furthermore appears dose-dependent with greater potential for off-target effects at higher doses. ¹¹ In contrast, deucravacitinib is an allosteric inhibitor that binds to the regulatory domain of TYK2 in a highly selective manner with little to no activity against JAK1-JAK3. ⁹ Deucravacitinib’s high selectivity for TYK2 in contrast with other JAKi’s lack of specificity has important implications for potential safety concerns discussed later.

THE STORY BEHIND THE BLACK BOX WARNINGS

In 2017, the US FDA issued its first black box warning on a JAKi, baricitinib, recommending caution on its use in patients who may be at increased risk of thrombosis. ²¹ This black box warning was based on a numerical imbalance of venous thromboembolism (VTE) in a small number of randomized controlled trials (RCTs) between baricitinib 4mg and placebo in those with RA. ^22,23 However, the individual trials did not have enough power to either confirm or exclude a significant difference in VTE risk due to the rarity of events.

Tofacitinib also came under early scrutiny by the FDA during drug development for increases in serum lipid levels and the incidence of cancers in the early 2010s. ²⁴ In the ORAL (Oral Rheumatoid Arthritis trial) phase 3 trials, increases were observed in HDL cholesterol, LDL cholesterol, total cholesterol, apolipoprotein A-I, and apolipoprotein B levels. ^25,26 These increases generally occurred during the first four weeks of treatment and stabilized after three months of treatment without lipid-specific intervention. ^26,27

The implications of these changes on patient safety were initially unclear but concerning, especially when considering that patients with RA, and IMIDs more broadly, may have an increased CV risk at baseline. For context, two large meta-analyses showed that RA was associated with a 48% increased risk of CV events and 50% higher incidence of CV disease related mortality in comparison to the general population. ²⁸ Furthermore, a large population-based study found an increased risk of CV disease in all nineteen different autoimmune diseases that were tested. ²⁹ As a result, with the approval of tofacitinib at doses of 5mg twice daily (BID) or 11mg once daily for RA, the FDA mandated a randomized, post-authorization trial to assess the risk of major cardiovascular events (MACE) and cancers in tofacitinib. ²⁴

This post-authorization, head-to-head safety trial was termed “ORAL Surveillance” and commenced in March 2014. It was an open-label, randomized, noninferiority trial that served to assess the risk of MACE or cancers through comparison of tofacitinib, either 5mg BID or 10mg BID, with TNF-inhibitors. ²⁴ As a noninferiority trial, its purpose was to determine whether the new treatment, tofacitinib, was not worse in terms of safety than the control treatment, TNF-inhibitors, by an acceptably small amount. ³⁰ The trial recruited patients with RA who were at least 50 years of age and had at least one additional CV risk factor (current cigarette smoker, hypertension, HDL cholesterol <40 mg/dL, diabetes mellitus, family history of premature coronary heart disease [CAD], extraarticular RA, or history of CAD) between 2014 and 2020. ²⁴ The coprimary endpoints were adjudicated MACE (death due to a CV event, non-fatal myocardial infarction [MI], or non-fatal stroke) and malignancy (excluding nonmelanoma skin cancers).

The first signal to harm regarding tofacitinib’s increased CV risk occurred before the trial concluded in February 2019, when the trial’s data and safety monitoring board recommended a dose reduction from tofacitinib 10mg BID to 5mg BID. ²⁴ They recommended this dose reduction after noting a higher frequency of pulmonary embolism (PE) and all-cause death among patients receiving tofacitinib 10mg BID compared to those receiving a TNF-inhibitor. ²⁴ The FDA promptly issued a black box warning regarding the risk of PE and death with tofacitinib 10mg BID in arthritis or UC. ³¹

In January 2022, the results of this cardinal head-to-head trial were released and did not show noninferiority for the combined doses of tofacitinib compared to TNF-inhibitors for adjudicated MACE or adjudicated cancers. ²⁴ For MACE, the hazard ratio was 1.33 with a 95% confidence interval of 0.91 to 1.91 with the noninferiority endpoint not met because the confidence interval extended beyond 1.8. ²⁴ In other words, in this population, combined doses of tofacitinib were shown to be worse than TNF-inhibitors by exceeding the prespecified margin that denoted an acceptable difference in safety risk. Noteworthy among secondary safety end points was adjudicated VTE, which was more likely in patients treated with tofacitinib 10mg BID compared to TNF-inhibitors and likely driven by PE events, although confidence intervals were wide (VTE: HR 3.52, 95% CI 1.74–7.12; PE: HR 8.26, 95% CI 2.49–27.43). ²⁴ There were no significant differences in adjudicated VTE risk (or PE risk) between patients treated with tofacitinib 5mg BID and TNF-inhibitors. ²⁴ There was also no significant difference in adjudicated DVT risk when comparing either tofacitinib dose with TNF-inhibitors. ²⁴

In response to the results of ORAL Surveillance, the FDA updated their black box warnings on JAKi in 2021 to reflect “the risks of serious heart-related events, cancer, blood clots, and death”. ⁷ Tofacitinib is selective towards JAK1, JAK2, and JAK3, but not TYK2; thus, this warning was extended to JAKi with overlaps in selectivity of JAK isoforms, including: abrocitinib, baricitinib, upadacitinib, and even topical ruxolitinib (not deucravacitinib because of its selective TYK2 inhibition). ¹¹ Even though this warning resulted from a trial exclusively studying RA patients greater than 50 years of age with at least one additional CV risk factor, based on FDA guidance, providers are expected to discuss this black box warning with all patients prescribed these medications regardless of demographics, baseline CV risk, or specific IMID indication. The extension of the black box warning to all medications with an overlapping mechanism of action is a common FDA practice. ³²

Official FDA guidance urges healthcare professionals to weigh benefits and risks prior to initiating or continuing therapies with these medications in patients who are current or former smokers, those with other CV risk factors, and those with malignancy. ⁷ With the exception of moderate to severe AA which does not have other FDA-approved therapies, they have recommended reserving oral JAKi for patients after trialing other systemic therapies (specifically TNF-inhibitors in patients with AS, RA, non-radiographic axSpA, polyarticular course JIA, PsA, or UC). ^3–7 The EMA has similar recommendations that were finalized in January 2023. They have urged that JAKi should be used only if no suitable treatment alternative is available in patients over 65 years of age, those at increased risk of major CV problems, current and former long-term smokers, and those at increased risk of cancer. They have advised for caution in all patients at increased risk for VTE and suggested dose reductions, if possible, in patients at increased risk of VTE, cancer, or major CV problems. This safety warning applies to all IMID-approved uses of JAKi. ³³

THE UNANSWERED QUESTIONS ABOUT THE CARDIOVASCULAR SAFETY PROFILE OF JAK INHIBITORS

Considerable controversy has emerged on the topic of CV risk in patients treated with JAKi given the extension of the black box warning to multiple JAKi and to all patients prescribed these medications regardless of IMID indication or patient characteristics such as demographics or baseline risk factors. The current black box warning even applies to topical ruxolitinib, despite achieving a mean steady-state plasma concentrate significantly below the level expected for systemic effects. ³⁴ Furthermore, the mechanism by which JAKi might increase risk of MACE and VTE is unclear. Many have speculated that greater modulation of JAK2 may specifically contribute to adverse thrombotic events. ³⁵ However, evidence has been inconclusive, and additional studies are needed. ^35–38 Given that real-world data has shown TNF-inhibitors to be protective against MACE, others have instead suggested that ORAL Surveillance might demonstrate TNF-inhibitors to be relatively more protective than tofacitinib. Evidently, additional basic and translational studies are needed to validate these hypotheses and ultimately elucidate how JAK inhibition might increase risk of MACE and VTE.

Delineation of Cardiovascular Risk in Studies Beyond Oral Surveillance

Despite FDA and EMA warnings on CV risk with JAKi use, several systematic reviews and meta-analyses utilizing RCT data from broader populations have shown CV risk to be similar between JAKi and control users. For instance, prior to ORAL Surveillance, a systematic review and meta-analysis of 25 RCTs including only RA patients with 2,193 JAKi patient exposure years (PEYs) and 962 placebo PEYs showed no significant difference regarding risk of MACE (OR 0.80, 95% CI 0.36–1.75) or VTE (OR 1.16, 95% CI 0.48–2.81) in JAKi versus placebo. ³⁹ Notably, a dose-dependent impact on incidence of all CV events (inclusive of MACE and VTE) was not seen in tofacitinib or upadacitinib, but was seen in baricitinib (2 mg safer than 4 mg; OR 0.19, 95% CI 0.04–0.88). ³⁹

Another pooled analysis prior to ORAL Surveillance with 42 RCTs including AS, inflammatory bowel disease (IBD), PsA, PsO, and RA patients with 6,542 JAKi PEYs and 1,578 placebo PEYs found the pooled incidence rate ratio of VTE in JAKi vs placebo to be 0.68 (95% CI 0.36–1.29). ²¹ After ORAL Surveillance, an updated analysis was conducted on 66 RCTs including 38,574 patients with RA, PsA, IBD, and AD with a mean follow-up of 10.5 months; the authors found a numerically higher but non-significant difference regarding occurrence of MACE (1.19, 95% CI 0.86–1.64) or VTE (1.65, 95% CI 0.97–2.79). ⁴⁰ Notably, a large portion of their results (>50%) were driven by the ORAL Surveillance trial; therefore, their population was skewed towards individuals with RA and thus an enriched CV risk at baseline. ⁴⁰

These data, predominantly driven by the patient population in ORAL Surveillance, suggest that the increased risk of MACE or VTE with JAKi may disproportionately affect individuals with increased CV disease risk factors. For example, the STAR-RA trial used claims data to investigate CV risk in two separate RA populations: one real-world population including all RA patients greater than 18 years of age and the other population mimicking the strict inclusion/exclusion criteria of ORAL Surveillance (age ≥50 with an additional CV risk factor). ⁴¹ In the real-world population cohort including 102,263 RA patients, the pooled weighted hazard ratio that compared tofacitinib with TNF-inhibitors for the composite outcome of MI and stroke was 1.01 (95% CI 0.83 to 1.23). ⁴¹ In the cohort mimicking ORAL Surveillance consisting of older patients with CV risk factors or disease, the hazard ratio was 1.24 (95% CI 0.90 to 1.69), similar to the results of ORAL Surveillance (HR 1.33; 95% CI 0.91–1.94). ⁴¹ Of note, as a retrospective, observational study, there may be additional unaccounted for differences between the real-world population and the ORAL Surveillance mimicked population beyond baseline CV risk alone. For example, confounding by indication is a concern. Although the STAR-RA trial has limitations, to summarize, this large population-based study did not show an increased risk of adverse CV events in a broad cohort of RA patients while the ORAL Surveillance mimicked cohort, enriched with CV risk factors, showed a numerically but not statistically significantly increased CV risk in patients receiving tofacitinib, consistent with the original trial.

Stratification by Atherosclerotic Cardiovascular Disease History and Risk

Several post-hoc analyses have thus attempted to delineate CV risk more precisely through stratifying patients within ORAL Surveillance by baseline CV disease risk factors. These analyses attempt to understand whether CV risk is higher in certain subpopulations within this already higher-risk cohort.

A post-hoc analysis subdivided the ORAL Surveillance population into patients with and without a history of atherosclerotic CV disease (ASCVD). ⁴² Approximately 15% of the ORAL Surveillance cohort had a history of ASCVD, defined based on events, diagnoses, and procedures associated with atherosclerosis such as CAD, MI, unstable angina, ischemic stroke, and peripheral artery disease. ⁴². The analysis showed a numerically higher risk of MACE with tofacitinib versus TNF-inhibitors that was greater in magnitude in patients with a history of ASCVD (HR 1.98, 95% CI 0.95–4.14) compared to those without a history of ASCVD (HR 1.14, 95% CI 0.73–1.79) with an interaction p-value of 0.059. ⁴² This analysis supports the notion of a stronger signal to harm in populations with a previous ASCVD history.

This post-hoc analysis also examined CV risk in the context of the 10-year risk of an ASCVD event in patients without a history of ASCVD. ⁴² The 10-year risk of an ASCVD event was determined with the ASCVD-pooled cohort equations calculator with a 1.5 multiplier applied. There was no difference in MACE risk with combined tofacitinib doses versus TNF-inhibitors in patients at high (≥20% risk; HR 1.03, 95% CI 0.56–1.92) or intermediate risk (≥7.5 to <20% risk; HR 1.07, 95% CI 0.52–2.18). ⁴² Additional post-hoc analyses presented at the European Alliance of Associations for Rheumatology (EULAR) 2022 meeting similarly performed stratified analyses by 10-year ASCVD event risk. While these post-hoc analyses should be considered preliminary and exploratory, these results contrastingly did show a trend of increased MACE and VTE risk estimates with a history of ASCVD or higher 10-year ASCVD event risk (particularly ≥20% risk). VTE risk estimates additionally trended higher with tofacitinib dosing above 10mg daily. ^43,44

Conservative interpretation of these analyses suggests patients on JAKi with a history of ASCVD, high 10-year ASCVD event risk, or on higher dosing regimens may have a greater likelihood of experiencing an adverse CV or VTE event.

Identification of Other Risk Factors for MACE and VTE

Another post-hoc analysis presented at the EULAR 2022 meeting took a broader approach to identifying risk factors for MACE in ORAL Surveillance using univariable and multivariable Cox proportional hazards models. ⁴⁵ While all 95% CIs for HRs crossed 1 indicating lack of statistical significance, they identified current smoking, aspirin use, history of chronic lung disease, history of diabetes, male sex, and age ≥65 years to be potential baseline risk factors for MACE in the ORAL Surveillance cohort. ⁴⁵ Of these risk factors, increased risk of MACE was clearest in patients who were current/past smokers (versus never smoked), aspirin users (versus non-users), and patients aged ≥65 years who had ever smoked (versus patients ≥50-<65 years who had never smoked). ⁴⁵ Furthermore, while this analysis did not look at statin usage, a different post-hoc analysis presented at EULAR 2022 showed suboptimal baseline statin usage in ORAL Surveillance. Across CV risk score categories for all treatment groups, fewer than 50% of patients received statins at baseline. However, patients in the higher CV risk categories or patients with diabetes that did receive statins at baseline had lower MACE incidence rates (IRs) compared to those who were not on statins, emphasizing the well-established role of statins in CV risk mitigation. ⁴⁶

To identify risk factors for VTE in ORAL Surveillance, a similar approach was taken using univariable and multivariable Cox proportional hazards models. ⁴⁷ In this post-hoc analysis presented at the EULAR 2022 meeting, prior history of VTE, antidepressant use, body mass index (BMI) ≥30 kg/m², corticosteroid use, male sex, age ≥65 years, oral contraceptives/hormone-replacement therapy were identified as significant baseline risk factors for PE. ⁴⁷ VTE, DVT, and PE IRs were numerically higher with tofacitinib 10mg BID versus 5mg BID. Furthermore, this analysis did not find evidence that VTE risk accumulates with increased duration of tofacitinib use given that VTE, DVT, and PE IRs were consistent over time.

These post-hoc analyses suggest that that the combined use of JAKi and traditional CV risk factors may work together to increase the risk of MACE and VTE. Conventional strategies to lower CV risk, such as lipid-lower therapies, continue to be vital.

Evidence Across Indications

JAKi are approved for several IMIDs that are diverse with regards to their epidemiology and baseline CV risk. ^48–57 While RA, PsA, AS, UC, and axial spondylarthritis are clearly associated with CV disease, research is ongoing regarding CV risk in AA, AD, polyarticular course juvenile idiopathic arthritis, and vitiligo. ^48–57 The heterogeneity in patient characteristics of those utilizing JAKi in the real world is highlighted by Table 1 which summarizes demographics in seminal phase 3 trials. Table 1 highlights the great variability in mean/median patient age, ranging from 11.9 to 57.1 years. While only patients aged 50 or older were included in Oral Surveillance, the mean/median patient age is below 50 in all phase 3 trials, with the exception of some trials for psoriatic arthritis and rheumatoid arthritis. Thus, making a definitive statement regarding the relationship between JAKi and increased VTE or MACE risk may not yet be possible in these lower risk populations based on the results of ORAL Surveillance. Furthermore, while many studies have attempted to extrapolate risk of CV events from these phase 3 trials, follow-up time is limiting. The vast majority of trials have followed patients for one year or less with follow-up times as short as 8 weeks. The long-term risk of CV events therefore remains unknown for many populations utilizing JAKi. In summary, more long-term data is needed on CV event rates in patients utilizing JAKi for different indications and with different baseline CV risk profiles.

Table 1.

Patient characteristics of phase 3 trials for US FDA-approved JAKi and FDA-approved indications

Condition	JAK inhibitor	Trials (Trial Duration)	Mean or Median Age Range	Males (%) Range
Alopecia Areata	Baricitinib	NCT03570749 (36 weeks) NCT03899259 (36 weeks)	36.3–39.0	36.8–41.4
Ankylosing Spondylitis	Tofacitinib	NCT03502616 (48 weeks)	41.1	83.3
	Upadacitinib	NCT03178487 (14 weeks) NCT04169373 (14 weeks)	42.4–45	70.6–74.0
Atopic Dermatitis	Abrocitinib	NCT03349060 (3 months) NCT03575871 (3 months) NCT03720470 (4 months) NCT03796676 (3 months)	14.9–37.7	48.9–58.6
	Topical Ruxolitinib	NCT03745638 (8 weeks) NCT03745651 (8 weeks)	35.2–36.4	38.0–38.5
	Upadacitinib	NCT03569293 (16 weeks) NCT03607422 (16 weeks) NCT03568318 (16 weeks)	32.1–32.8	52.9–59.9
Non-radiographic Axial Spondyloarthritis	Upadacitinib	NCT04169373 (1 year)	42.1	41.5
Nonsegmental Vitiligo	Topical Ruxolitinib	NCT04052425 (52 weeks) NCT04057573 (52 weeks)	38.9–40.2	43.6–49.9
Polyarticular Course Juvenile Idiopathic Arthritis	Tofacitinib	NCT02592434 (44 weeks)	11.9	24.9
Psoriatic Arthritis	Tofacitinib	NCT01877668 (12 months) NCT01882439 (6 months)	47.9–50.0	44.7–46.7
	Upadacitinib	NCT03104400 (24 weeks) NCT03104374 (24 weeks)	50.8–53.4	45.7–46.8
Rheumatoid Arthritis	Baricitinib	NCT01721057 (24 weeks) NCT01721044 (24 weeks)	51.8–55.7	18.1–18.2
	Tofacitinib	NCT00814307 (6 months) NCT00856544 (12 months) NCT00853385 (12 months) NCT00847613 (2 years) NCT00960440 (6 months) NCT01039688 (2 years)	49.6–55.0	13.4–20.7
	Upadacitinib	NCT02706873 (24 weeks) NCT02706951 (14 weeks) NCT02675426 (12 weeks) NCT02629159 (48 weeks) NCT02706847 (12 weeks)	53.7–57.1	16.1–24.3
Ulcerative Colitis	Tofacitinib	NCT01465763 (8 weeks) NCT01458951 (8 weeks) NCT01458574 (52 weeks) NCT01470612 (52 weeks)	41.0–42.9	55.5–59.1
	Upadacitinib	NCT02819635 (52 weeks) NCT03653026 (8 weeks)	42.2–42.9	61.6–62.5

SUGGESTIONS FOR CARDIOVASCULAR RISK STRATIFICATION AND MANAGEMENT

The analyses discussed in this review underscore the importance of CV risk stratification prior to initiating JAKi therapies. Unfortunately, the search for biomarkers to predict CV risk has not yet been successful. One group found D-dimer levels to be associated with subsequent VTE risk in a post-hoc analysis of ORAL Surveillance; however, its low specificity and variability in inflammatory states deem it unreliable as a marker for VTE in patients on JAKi who are already in a hyperinflammatory state. ⁵⁸

Most of the evidence thus far supports a multiple-hit hypothesis wherein CV risk increases with a combination of traditional CV risk factors and use of JAKi. Thus, we recommend that patients undergo an initial comprehensive CV history and laboratory evaluation to identify patients that would benefit from serious consideration of alternative treatment modalities, if available, as well as CV risk mitigation strategies if JAKi therapy is initiated (Figure 2). For patients with a high pre-test probability of CVD, very high CV risk, or who have had an MI or stroke prior to JAKi initiation, we suggest early referral to a cardiologist.

Figure 2.

Suggested workflow for cardiovascular risk stratification and management in patients identified as potential candidates for JAKi therapy

Although most IMIDs have several options for systemic therapies outside of the JAKi class, a subset of patients will have a scarcity of FDA-approved treatment options. For example, patients with moderate to severe AA have only one US FDA-approved medication currently: oral baricitinib. Other patients may have already failed most other therapeutic options. Given the uncertainty regarding CV risk with JAKi, many rheumatologists prefer to first try other biologic therapies, such as TNF-inhibitors, with more extensive, real-world safety data available for reference. Similarly, in PsA, rheumatologists may prefer to trial TNF, IL-17 and/or IL-12/23 inhibitors before trialing a JAKi. As a result, some patients with high-risk criteria for CV events may not have alternative therapeutic options at the time a JAKi is being seriously considered. Shared decision-making that involves a thorough discussion of the risks and benefits of foregoing a JAKi is essential. If trialing a JAKi, an aggressive approach to mitigating CV risk may be warranted, involving initiating/intensifying medication therapies (ex. aspirin, statin) as well as optimizing lifestyle factors (ex. diet, exercise, weight loss).

Once the decision is made to initiate a JAKi, lipid levels should be re-checked in all patients approximately 3 months after therapy initiation. ¹⁹ A network meta-analysis of patients with RA showed that JAKi lead to a mean increase of 8.11 mg/dL in HDL cholesterol and 11.37 mg/dL in LDL cholesterol. ⁵⁹ If baseline lipid levels are elevated or if an increase in lipid levels is observed during therapy which pushes the patient into a higher calculated 10-year CV risk (Figure 2), providers should discuss the need to add or escalate statin therapy or consider other lipid-lowering agents and strategies. A continued multidisciplinary approach that integrates cardiology is recommended in all patients that are maintained on JAKi therapy longitudinally.

A NOVEL CLASS OF JAK INHIBITORS: TYK2 INHIBITORS

TYK2 is a member of the JAK-STAT superfamily that mediates signaling downstream of IL-12, IL-23, and type I IFN receptors. ⁹ A novel subclass of JAKi, selective TYK2 inhibitors, has been developed for PsO and potentially other IMIDs, where IL-12 and IL-23 play central roles in initiating and maintaining chronic inflammation. ⁹

To date, there is one selective TYK2 inhibitor, deucravacitinib, that is FDA approved for moderate-to-severe plaque PsO. ² To contrast with the previously discussed JAKi targeting JAK1, JAK2, and/or JAK3 that bind to active sites in the catalytic domain of the targeted enzymes, deucravacitinib binds to the regulatory domain of TYK2 in a highly selective manner. ⁹ Deucravacitinib has little to no activity against JAK1, JAK2, and JAK3 at clinically relevant doses. ^9,60 This high selectivity and unique mechanism of action is believed to limit the side effect profile of deucravacitinib.

While the FDA’s black box warning was applied to other JAKi beyond tofacitinib on the basis of shared mechanisms of action, deucravacitinib with its novel and unique mechanism of action did not automatically receive a black box warning. The two phase 3 trials that led to deucravacitinib’s FDA approval, POETYK PSO-1 and PSO-2, assessed efficacy in 1,684 subjects randomized to deucravacitinib (6mg orally once daily), placebo, or apremilast (30mg orally BID; a PDE-4 inhibitor FDA-approved for plaque PsO across all severities). In these trials, there were no apparent differences in rates of MACE or VTE between deucravacitinib and comparator arms (Table 2). The two VTE cases in the deucravacitinib arms were not considered drug related. Deucravacitinib was not associated with significant changes in hematologic parameters, HDL and LDL cholesterol, creatinine, immunoglobulins, or liver enzymes. ³⁵ Elevations in triglyceride levels (~10 mg/dL) and creatine phosphokinase levels were the most common laboratory abnormalities. ^61,62 Serum lipid levels, preferably fasting, should thus be monitored routinely upon initiation of deucravacitinib with abnormalities prompting referral to the patient’s primary care provider or cardiologist if the patient has additional CV risk factors.

Table 2.

Cardiovascular Event Rates in POETYK PSO-1 and PSO-2 (Deucravacitinib Phase 3 Trials)

Trial Name	Mean Age	Males (%)	Study Duration	Trial Arm	Na	Adjudicated MACE (EAIR/100 PY)	VTE (EAIR/100 PY)
PSO-163	46.1	68.0	52 weeks	Deucravacitinib	531	1 (0.2)	1 (0.2)b
				Placebo	165	2 (4.1)	0 (0)
				Apremilast	168	2 (1.7)	0 (0)
PSO-264	46.9	66.1	52 weeks	Deucravacitinib	833	2 (0.4)	1 (0.2)c
				Placebo	501	0 (0)	0 (0)
				Apremilast	254	1 (0.9)	0 (0)

EAIR, exposure-adjusted incidence rate; PY, person-years

^aIncludes patients who crossed over between trial arms in PSO-1 and PSO-2

^b48-year-old male patient with a history of smoking and hypertension who experienced an acute ascending aortic dissection complicated by a pulmonary artery thrombosis at Week 48. The event was not considered to be related to the study drug. The study drug was withheld during the event and resumed after recovery without subsequent venous thrombotic events.

^c19-year-old patient who developed rash, headache, and fever, and stopped deucravacitinib on Day 4. The patient developed radial vein thrombosis after intravenous cannulation for antibiotic treatment on Day 16, which was reported as not drug related.

While there has not yet been evidence suggesting deucravacitinib increases CV or thrombotic risk, long-term extension studies and post-marking surveillance data are needed for further risk assessment in plaque PsO. Deucravacitinib is also under current study investigation for use in other diseases including AA, cutaneous lupus, systemic lupus, IBD, and PsA. Notably, it is likely that deucravacitinib would not be as effective in the treatment of RA given that several key cytokines implicated in its pathogenesis signal through other JAK isoforms. ¹² As the FDA potentially approves deucravacitinib for additional indications, more investigations will be needed to determine whether risk differs by indication, similarly to what is hypothesized with other JAKi.

CONCLUSION

JAKi, excluding selective TYK2 inhibitors, have a significant side effect profile, as summarized in Table 3. While MACE and VTE and not been definitively proven to be increased in those taking JAKi with nonsignificant hazard ratios, there is sufficient evidence for concern given the wide confidence intervals, consistent hazard ratio point estimates above 1, and failure to meet the noninferiority endpoint for MACE in ORAL Surveillance. Emerging data has furthermore indicated that CV events occur disproportionately in certain individuals. Older patients with RA and pre-existing CV disease risk factors are at increased risk, especially those with a prior history of ASCVD or VTE. Further studies are needed to confirm risk factors for MACE and VTE identified through exploratory analyses. However, they begin to pave the way for an approach to risk stratification and management of patients who may be candidates for longitudinal JAKi therapy. We suggest a comprehensive, multimodal approach for evaluation of CV risk in these patients. This approach is based on an extensive and up-to-date review of the current literature that is rapidly evolving.

Table 3.

Summary of notable adverse side effects of JAKi

Parameter	Effect of JAKi	Source
Cancer (excluding nonmelanoma skin cancer)	Tofacitinib vs. TNF-inhibitor RA patients >50 with 1 additional CV risk factor HR (95% CI): 1.48 (1.04–2.09)	Ytterberg-2022 (Oral Surveillance Trial) 24
Infection	Tofacitinib 5mg BID vs. TNF-inhibitor RA patients >50 with 1 additional CV risk factor Serious Infection, HR (95% CI): 1.17 (0.92–1.50) Opportunistic, HR (95% CI): 1.82 (1.07–3.09) Herpes Zoster, HR (95% CI): 3.28 (2.44–4.41)	Ytterberg-2022 (Oral Surveillance Trial) 24
Lipid Levels	Low-Density Lipoprotein Cholesterol Increase by 11.37 mg/dL (95% CI 7.84–14.91) High-Density Lipoprotein Cholesterol Increase by 8.11 mg/dL (95% CI 6.65–9.58) These increases are generally observed between 1–3 months of treatment and levels remain elevated thereafter.	Li et al-2022 (Network Meta-Analysis) 59 Charles-Schoeman et al-2016 (Review)
MACE	Tofacitinib vs. TNF-inhibitor RA patients >50 with 1 additional CV risk factor HR (95% CI): 1.33 (0.91–1.94)	Ytterberg-2022 (Oral Surveillance Trial) 24
	JAKi vs. Placebo/Comparator All IMID patients HR (95% CI): 1.19 (0.86–1.64)	Maqsood-2022 (Meta-Analysis of RCTs) 40
	Tofacitinib vs. TNF-inhibitor RA patients (real-world cohort) HR (95% CI): 1.01 (0.83 to 1.23)	Khosrow-Khavar-2022 (STAR-RA study) 41
VTE	Tofacitinib 5mg BID vs. TNF-inhibitor RA patients >50 with 1 additional CV risk factor HR (95% CI): 1.66 (0.76–3.63)	Ytterberg-2022 (Oral Surveillance Trial) 24
	JAKi vs. Placebo/Comparator All IMID patients HR (95% CI): 1.65 (0.97–2.79)	Maqsood-2022 (Meta-Analysis of RCTs) 40

MACE: Major Adverse Cardiovascular Events; VTE: Venous Thromboembolism

The state of current literature on the association of CV risk and JAKi therapy has several gaps. Basic and translational studies are necessary to clarify the mechanisms of how JAKi increase CV risk. Most evidence on CV risk has focused on tofacitinib, and it remains unclear if there are differences in safety or efficacy between other selective JAK1-JAK3 inhibitors. As the FDA considers approving other JAKi with different mechanisms of actions, such as ritlecitinib (JAK3/TEC kinase inhibitor under investigation for AA), understanding how specific mechanisms contribute to CV risk will become even more critical. Most evidence has also focused on RA; however, patients taking JAKi for other indications may be at a significantly lower baseline CV risk which may affect CV risk on JAKi maintenance therapy. Furthermore, additional, longer-term data is needed to better understand whether CV risk is cumulative over time. Emerging RCT and ongoing post-marketing data will help us answer these questions.

While JAKi therapies carry risk profiles that are higher compared to other indicated therapies, they remain an important tool in the arsenal of rheumatologic and dermatologic treatments to manage chronic, oftentimes refractory IMIDs that have a significant burden of morbidity and/or mortality. JAKi, with their risk profile, may not be the appropriate therapy for all patients. As such, a precision medicine approach to utilizing JAKi is essential, where clinicians consider individual patient characteristics and risk factors to determine the appropriateness of JAKi use and, in those who may benefit from JAKi use, tailored plans of care for mitigation and management of CV risk.