- Some previous studies, including the ORAL Surveillance randomized trial, have shown moderately greater cancer risks with JAK inhibitors versus TNF inhibitors.
- But some other analyses of real-world data found no difference in these safety issues.
- This German registry study did show increased cancer rates overall with JAK inhibitors relative to biologic agents, and the risk was even greater in certain subgroups.
Use of Janus kinase (JAK) inhibitors for rheumatoid arthritis (RA) was associated with significantly higher rates of malignancies compared with biologic agents, data from a large German registry indicated.
Based on more than 6,400 treatment episodes with the two drug classes, crude cancer incidence rates were 11.6 per 1,000 patient-years for JAK inhibitors versus 9.3 per 1,000 for biologic disease-modifying anti-rheumatic drugs (bDMARDs), according to Martin Schaefer, PhD, of the German Rheumatology Research Center in Berlin, and colleagues.
That worked out to a hazard ratio of 1.40 (95% CI 1.09-1.80) after adjusting for a host of demographic and clinical parameters, the researchers reported in Annals of the Rheumatic Diseasesopens in a new tab or window.
This appears to be the first registry study to support increased cancer risks with JAK inhibitors. But the German data are very much in line with results from the landmark ORAL Surveillance studyopens in a new tab or window, a randomized trial conducted to evaluate this very issue. It found a hazard ratio of 1.48 for incident cancers (as well as a 33% increase in cardiovascular event rates) among selected RA patients. ORAL Surveillance was required by the FDA, moreover, because data from clinical trials with the pioneering JAK inhibitor tofacitinib (Xeljanz) had suggested higher cancer and cardiovascular event risks compared with tumor necrosis factor (TNF) blockers.
But precisely because these prospective trials enrolled patients meeting certain criteria (e.g., age >50 and one additional cardiovascular risk factor in ORAL Surveillance), manufacturers and clinicians wanted corroboration from real-world data -- which heretofore had not been forthcoming. As Schaefer and colleagues noted, registry studies from around the world had either failed to find even a hint of increase in cancer risk or were underpowered to detect an effect. "It remains largely unclear what the differences in the results of the individual studies can be attributed to," the researchers observed.
As it happens, Schaefer and co-author Anja Strangfeld, MD, PhD, also of the German Rheumatology Research Center, lead a German registry called RABBITopens in a new tab or window that tracks RA patients receiving biologics and other advanced therapies, including JAK inhibitors. It supplied the data for the current analysis. Participants and their clinicians submit data periodically, aiming to cover at least 5 years and up to 10 years of treatment.
The researchers looked at treatment episodes, spanning the period after a particular drug was given; consequently, an individual patient who changes from a bDMARD to a JAK inhibitor would have two treatment episodes. Only those incident cancers diagnosed at least 6 months after starting a therapy (the "latency period") were counted.
In total, 2,285 JAK inhibitor episodes and 4,259 episodes with biologics were included. Tofacitinib, baricitinib (Olumiant), upadacitinib (Rinvoq), and filgotinib (Jyseleca) were the JAK inhibitors tracked; biologics included the five most common TNF inhibitors, two interleukin-6 blockers, abatacept (Orencia), and rituximab (Rituxan). Median follow-up was 47 months for JAK inhibitors and 50 months for bDMARDs.
Median patient age was just under 60, and three-quarters were women. Roughly half were also taking conventional DMARDs such as methotrexate. Some 60% were current or former smokers. In half the biologic episodes, the drug was the first targeted agent that the patient had received; this was true only for one-quarter of the JAK inhibitor episodes.
It should be emphasized that incident malignancies were rare. Overall, there were 88 during JAK inhibitor episodes and 135 following a bDMARD episode. Moreover, the extra risk found for JAK inhibitors was relatively modest: Schaefer and colleagues calculated a "number needed to harm" of 368, meaning that one would need to treat that number of patients with JAK inhibitors to see one extra cancer.
The researchers also looked at certain subgroups of patients, and some of these appeared to be more at risk from JAK inhibitors, at least numerically. For example, patients meeting the main ORAL Surveillance criteria -- age >50 with one additional cardiovascular risk factor -- faced a 61% higher risk of cancer with JAK inhibitors relative to bDMARDs (HR 1.61, 95% CI 1.22-2.13). This relatively broad confidence interval meant that it wasn't significantly greater than the 1.40 seen in the total cohort.
Also, cancer risk increments with JAK inhibitors increased as treatment continued. During the first 16 months, with the 6-month latency period, there was no difference in cancer incidence between drug types (HR 0.94, P=0.868). But with a 16-month latency period, a significant difference did emerge (HR 1.54, P=0.002).
When comparing JAK inhibitors with non-TNF-targeting agents -- used in 40% of the bDMARD episodes -- there remained a numerical difference between those drugs and JAK inhibitors, but the hazard ratio shrank to 1.30 and fell short of statistical significance. By the same token, when looking only at anti-TNF drugs in comparison with JAK inhibitors, the latter were associated with greater risk (HR 1.54, P=0.007).
Other factors that seemed to drive greater cancer risks with JAK inhibitors included age >60, history of smoking, RA in remission at baseline (but also high RA disease activity), and longer disease duration.
So does all this mean that JAK inhibitors should be avoided? Schaefer and colleagues said no -- people with an uncontrolled inflammatory disease also face increased risks for cancer. It's unclear how this risk compares with the risk from JAK inhibitors.
"Neither the [ORAL Surveillance] trial nor our study were designed to assess whether patients with high disease activity lacking alternative treatment options are at lower malignancy risk with or without a [JAK inhibitor] (and such assessments are difficult to perform for practical and ethical reasons)," the researchers wrote.
If it appears that a JAK inhibitor is the best option to bring a patient's disease under control, it might well represent a good tradeoff in risks. Schaefer and colleagues also pointed out that their number needed to harm with JAK inhibitors, 368, is massively higher than estimates of the number needed to treat to bring RA under control (≤10, they indicated).
Furthermore, no mortality difference between drug types was seen (HR 0.95, P=0.673). "In the absence of differing competing risks of death, this suggests a reasonable overall safety profile," the investigators concluded.
Limitations to the analysis included the lack of data on disease activity during follow-up and the potential for RABBIT participants to be unrepresentative of the population at large. Also, the study focused on RA, although JAK inhibitors may be used for a variety of other conditions including inflammatory bowel disease, making it unclear whether the results can be generalized to these other patient groups. No analyses of individual drugs were performed.
Disclosures
The RABBIT registry is supported by a joint, unconditional grant from AbbVie, Alfasigma, Amgen, Biocon, Bristol Myers Squibb, Fresenius Kabi, Hexal, Eli Lilly, Pfizer, Samsung Bioepis, Sanofi Aventis, and UCB, and previously by Celltrion, Merck Sharpe & Dohme, and Roche.
Authors reported relationships with these and multiple other commercial entities.
