Cancer and TNF inhibitors Save

Cancer risk minimisation is a high priority for people with rheumatic diseases, as it is for the general population. Tumor necrosis factor inhibitors (TNFi) have a long history of association with cancer risk discussions. I mean, the clue is in the name, tell anyone that you are going to give them a drug to block tumor necrosis factor and you are going to get a second take when they google necrosis.

Evaluation of cancer risk in rheumatoid arthritis is complicated by the relatively strong cancer promoting effect of the disease itself, a risk which is essentially dose (or disease activity) dependent. Channelling bias can then make it appear as if our most effective medications, given to the most severe disease, are associated with negative cancer outcomes. We now have excellent evidence that TNFi are not associated with solid organ cancers. We have ACR guidelines which tell us to treat the RA patient with a solid organ malignancy as if they don’t have one, and use whichever treatment we normally would. Despite this in real world practice, we still frequently see biologic agents being permanently stopped following a cancer and a reluctance to initiate them in people with a history of cancer. This is so prevalent that I honestly have to recheck the cancer data myself on a semi-regular to make sure I haven’t missed something. 

An oral presentation by Suarez-Almazor et al (Abstract 1675) in Monday’s RA – Treatments II: RA Treatment Safety session provided more supportive data on using TNFi in this population. Their study focuses on probably the most common cancer scenario I see in clinical practice, early stage breast cancer. They evaluated RA patients with early stage breast cancer who received TNFi in the year after diagnosis, compared to those who received csDMARDs or no DMARDs. They utilised two large databases, Optum’s Clinformatics and SEER/TCR-Medicare. They included 970 patients in Clinformatics and 1246 in SEER/TCR-Medicare. Only 17% and 16% respectively received TNFi, again confirming the variance of real-world practice from the guidelines. They calculated propensity scores for use of TNFi through a logistic regression model and performed multivariate Cox proportional hazards regression, controlling for covariates and the propensity score. Survival was evaluated out to 5 years. There was no significant difference in overall survival for TNFi - Clinformatics HR=0.75 (95% CI 0.41-1.37), SEER/TCR-Medicare HR=0.86 (95%CI 0.55-1.34). 

There are two further interesting side notes to this study. The first is that often the go to in these patients, when TNFi are avoided, is glucocorticoids. This data suggests that this is a bad idea, Patients receiving prednisone-equivalent doses of 7.5mg/day or more had dramatically worse survival than those who did not receive glucocorticoids in both databases, Clinformatics HR=2.51, (95%CI 1.32-4.76), SEER/TCR-Medicare HR=1.63 (95%CI 0.93-2.87). The second point is that breast cancer-specific survival was significantly better in TNFi treated patients HR=0.29 (95%CI 0.09-0.98). This is clearly encouraging and in keeping with the hypothesis that better disease activity control could lead to lower cancer risk. It is also intriguing with regard to the hypothesis suggested to explain the ORAL Surveillance cancer association with JAKi, that maybe it’s the TNFi that are preventing cancers better than the JAKi, rather than a negative association with either agent.

It makes sense to be cautious over cancer risk, but we are doing our patients a disservice and placing them at higher risk if we withhold our effective treatment options through unwarranted fear. My hope is that the steady accretion of data demonstrating the cancer safety of TNFi will ultimately result in a sea-change in the totality of clinical practice.