JAK inhibitors are safe for most people when prescribed with the same vigilance used for any potent immunosuppressant, but they are not risk-free. Roughly 1 in 100 to 1 in 300 treated patients per year develops a serious infection, and the drug class carries boxed warnings for serious infections, malignancy, major cardiovascular events, and blood clots. Safety depends less on the molecule than on the patient: age, smoking status, kidney and liver function, and prior cardiovascular or cancer history determine who faces a two- to three-fold elevated risk and who faces almost none.
How JAK inhibitors work and why safety follows from the mechanism
Janus kinases (JAK1, JAK2, JAK3, TYK2) sit inside immune cells and transmit signals from cytokines. When a cytokine docks on a receptor, JAK enzymes phosphorylate STAT proteins, and the signal reaches the nucleus to turn on inflammation genes. JAK inhibitors block that phosphorylation step, so the cytokine message never arrives. This is why they calm rheumatoid arthritis, psoriatic arthritis, atopic dermatitis, ulcerative colitis, and myeloproliferative disorders.
The safety profile is a direct consequence of where the enzymes live. JAK1 and JAK3 drive lymphocyte activation and inflammatory cytokine signaling, so blocking them suppresses infection defense and raises viral reactivation risk. JAK2 carries the erythropoietin and thrombopoietin signal, so blocking it lowers red cell and platelet production. TYK2 handles interferon and interleukin-23 signaling, which is why TYK2-selective drugs spare more of the infection machinery. In plain terms: the drug does what it is supposed to do, and the side effects are the same mechanism landing on tissue you need.
Selectivity matters more than class membership. A pan-JAK inhibitor such as tofacitinib touches four enzymes; a JAK1-preferential drug such as upadacitinib leans one way; a TYK2-selective drug such as deucravacitinib leans another. The 2022 ORAL Surveillance trial (Ytterberg et al., NEJM) showed the pan-JAK signal at high doses in older patients, and regulators extended warnings across the class while acknowledging that selectivity changes the absolute numbers.
The boxed warnings in plain numbers
The FDA requires boxed warnings on all approved JAK inhibitors for serious infections, mortality, malignancy, major adverse cardiovascular events (MACE), and thrombosis. These are class-level warnings, and the numbers underneath them come mostly from rheumatoid arthritis trials in patients aged 50 and older with at least one cardiovascular risk factor.
Serious infection rates in that population run about 3 to 4 per 100 patient-years for tofacitinib at 5 mg twice daily, versus roughly 2.5 per 100 patient-years on a TNF inhibitor. Herpes zoster is notably more frequent with JAK inhibitors than with biologics: about 3 to 4 per 100 patient-years versus 1 to 2. MACE incidence in ORAL Surveillance was 3.4 percent on tofacitinib versus 2.5 percent on a TNF inhibitor over the trial, a difference that reached statistical significance in the combined-dose analysis. Malignancy excluding non-melanoma skin cancer was 4.2 percent versus 2.9 percent. This is why the FDA in 2021 limited initial use of tofacitinib and later applied labeling changes across the class, reserving them for patients who have failed one or more TNF inhibitors when alternatives exist.
Those absolute numbers are the part most explainers skip. A 3 to 4 per 100 patient-year infection rate means roughly 96 to 97 of every 100 treated patients do not have a serious infection in a given year.
Who faces elevated risk: age, smoking, and cardiometabolic status
The risk gradient is steep, and it is driven by host factors more than by which JAK inhibitor is chosen. The ORAL Surveillance population was specifically enriched for risk: median age 61, about 80 percent with a history of cardiovascular disease risk factors, and roughly half current or former smokers.
| Risk factor | Effect on serious infection or MACE | Practical threshold |
|---|---|---|
| Age 65 or older | MACE and malignancy rates roughly double versus under 50 | Discuss alternatives before starting |
| Current smoker | Adds cardiovascular and malignancy risk; FDA labeling flags it | Cessation counseling before initiation |
| Prior MI or stroke | MACE risk climbs into the 4 to 5 per 100 patient-year range | Cardiology input advised |
| eGFR under 60 | Requires dose reduction for some agents | Check creatinine before dosing |
| Prior malignancy | Recurrence signal in ORAL Surveillance subgroup | Oncology history review |
| Chronic HBV or HCV | Reactivation risk without prophylaxis | Screen and treat first |
The reconciliation of these thresholds matters: a 45-year-old with moderate atopic dermatitis and no cardiovascular history faces a different absolute risk than a 68-year-old smoker with rheumatoid arthritis, even on the identical molecule. Age 65, smoking, and prior cardiovascular events are the three factors that most consistently move the risk from "low" to "discuss with a specialist."
Serious infections, shingles, and opportunistic pathogens
Bacterial pneumonia, cellulitis, urinary tract infection, and sepsis dominate the serious infection list. Tuberculosis reactivation is uncommon in the United States but real, and every patient should have latent TB screening (interferon-gamma release assay) before starting. Hepatitis B reactivation follows the same pattern as with biologics and requires antiviral prophylaxis when surface antigen is positive.
Herpes zoster is the signature infection of the class, occurring in roughly 3 to 4 per 100 patient-years, and it often presents as multidermatomal or disseminated disease rather than the classic single-dermatome rash. The CDC's Advisory Committee on Immunization Practices recommends recombinant zoster vaccine (Shingrix) for adults 50 and older, and giving it before starting a JAK inhibitor reduces that burden substantially. Live vaccines are contraindicated during therapy; inactivated and recombinant vaccines are safe.
Cancer, cardiovascular events, and clots: separating signal from noise
The malignancy signal in ORAL Surveillance was real but concentrated in lung cancer and lymphoma among older smokers. Non-melanoma skin cancer is elevated across the class and warrants annual dermatologic skin checks. For patients under 50 without smoking history, the absolute excess is small enough that many rheumatologists and dermatologists consider it acceptable against the benefit of disease control.
Major cardiovascular events and venous thromboembolism are the warnings that changed prescribing most. The FDA's 2019 addition of a thrombosis warning followed postmarketing reports of pulmonary embolism and deep vein thrombosis, particularly at the 10 mg twice-daily tofacitinib dose. Current labeling advises avoiding JAK inhibitors in patients with a history of thrombosis and using the lowest effective dose. Compared with TNF inhibitors, the MACE difference in ORAL Surveillance was about 0.9 percentage points over the trial, a figure that is meaningful at population scale and modest at the individual level for a low-risk patient.
JAK inhibitors compared with alternatives
The safety calculus only makes sense against the comparator. Here is how the class stacks up against the biologics most often used in the same diseases.
| Feature | JAK inhibitors | TNF inhibitors | IL-17 / IL-23 biologics | Methotrexate |
|---|---|---|---|---|
| Route | Oral pill | Injection or infusion | Injection | Oral or injection |
| Serious infection (per 100 patient-years) | 3 to 4 | 2 to 3 | 1 to 2 | 1 to 2 |
| Herpes zoster | 3 to 4 | 1 to 2 | Low | Low |
| TB reactivation | Low but screen first | Higher; screen first | Very low | Very low |
| MACE signal | Present in high-risk RA | Neutral | Neutral to date | Neutral |
| Malignancy signal | Elevated in older smokers | Neutral | Neutral to date | Slight for skin |
| Monitoring burden | Labs at baseline and periodically | Labs and TB screen | Labs and TB screen | Liver and marrow labs |
| Time to effect | 1 to 2 weeks | 2 to 12 weeks | 2 to 12 weeks | 6 to 12 weeks |
The comparison is not "safe versus unsafe." It is "different risk shapes." JAK inhibitors trade injection burden and immunogenicity for a cardiovascular and zoster signal; TNF inhibitors trade a higher TB risk and injection for a cleaner cardiovascular profile. For a patient who cannot tolerate injections or has failed biologics, the oral option with a managed risk profile is often the better choice.
Practical safety actions before and during treatment
Concrete steps reduce the absolute risk materially. These are the actions that separate a well-managed patient from a statistic.
- Screen for latent TB, hepatitis B and C, and HIV before the first dose; treat positives before starting.
- Update recombinant zoster vaccine and other inactivated vaccines at least two weeks before initiation.
- Get baseline complete blood count, liver panel, creatinine, and lipid panel; recheck at 4 to 8 weeks and then every 3 to 6 months.
- Stop smoking and control blood pressure and lipids before starting if age is 65 or older.
- Report fever, persistent cough, new leg swelling, or chest pain immediately rather than waiting for the next visit.
- Avoid live vaccines and use contraception if pregnancy is possible, since data in pregnancy are limited.
Red-flag symptoms that warrant urgent evaluation while on therapy:
- Fever above 101°F or shaking chills
- New shortness of breath or pleuritic chest pain
- Unilateral leg swelling or calf pain
- A painful blistering rash, especially across more than one dermatome
- Unexplained weight loss or a new persistent cough
- Sudden vision change or severe headache
What the first year looks like: a monitoring timeline
Risk is front-loaded. Most serious infections and zoster events occur in the first 6 to 12 months, and laboratory shifts appear within weeks. The timeline below reflects standard rheumatology and dermatology practice.
| Time point | What happens | Why it matters |
|---|---|---|
| Week -4 to 0 | TB, hepatitis, HIV screening; vaccines; baseline labs | Prevents reactivation and establishes reference values |
| Week 2 | Symptom check; early infection vigilance | Drug reaches steady state in about 5 to 7 days |
| Week 4 to 8 | Repeat CBC, liver, lipids | Catches cytopenias and lipid shifts early |
| Month 3 | First full safety review; dose optimization | Lowest effective dose reduces long-term exposure |
| Month 6 | Skin check if risk factors present | Non-melanoma skin cancer surveillance |
| Month 12 | Annual review of benefit versus risk | Guides continuation, switching, or tapering |
| Every visit | Blood pressure, weight, new symptoms | Cardiovascular risk accrues silently |
Limitations of the safety evidence
Most long-term safety data come from rheumatoid arthritis populations aged 50 and older with cardiovascular risk factors. That population overstates risk for a 30-year-old with atopic dermatitis and understates it for an 80-year-old with multiple comorbidities. Pediatric data remain thin, pregnancy registries are still accumulating, and head-to-head trials comparing individual JAK inhibitors on hard cardiovascular endpoints do not yet exist. The ORAL Surveillance findings apply most directly to tofacitinib at studied doses; extrapolating the exact percentages to upadacitinib, baricitinib, or TYK2-selective agents requires caution. Postmarketing surveillance through the FDA Adverse Event Reporting System and long-term registries continues to refine the picture, and labels have been revised multiple times since 2019.
The honest summary of the evidence: the class is safe enough for most patients when the prescriber screens, vaccinates, monitors, and chooses the lowest effective dose, and it is genuinely risky for older smokers with prior cardiovascular disease who have safer alternatives available. The question is not whether JAK inhibitors are safe in the abstract. It is whether the specific patient in front of you has been screened for the specific risks the class carries.