Background/Purpose: The
long-term lymphoma risk in patients with rheumatoid arthritis (RA) treated with
biologic disease modifying anti-rheumatic drugs (bDMARDs)
remains a concern. The aim of this study was to
extend assessments of overall and subtype-specific lymphoma risks with bDMARD therapy in RA.  

Methods:
 By linking data from the nationwide Swedish
Biologics (ARTIS), Patient, and Cancer Registers, we assembled a cohort of 13,240
RA patients starting a first bDMARD 1998-2012. As
comparators we identified a national cohort of bio-naïve RA patients (n=46,568),
and an age and gender matched general population referent cohort (n=458,846). Patients
were followed until the first of lymphoma, emigration, death, end of study
period (Dec 31, 2012) or start of bDMARD (for
bio-naïve patients). We estimated hazard ratios (HR) using Cox models for
lymphoma overall and for specific lymphoma subtypes according to ICD10-codes,
adjusted for age, sex, calendar time, and selected co-morbidities. A
histopathological review confirmed agreement between ICD-codes and the clinical
subtype diagnosis in >80% of cases.  

Results:
We found
69 lymphomas during 75,661 person-years (py) in the bDMARD treated patients, 241 lymphomas (237,789 py) in the bio-naïve patients, and 1413 lymphomas
(2,890,131 py) in the general population referent
cohort. All of the bDMARD treated lymphoma patients
were exposed to TNFi, 3 also to abatacept, 2 to anakinra and 2 to rituximab. Compared to the general
population both bDMARD treated (HR=2.5; 95% CI 2.0-3.3) and bio-naïve patients
(HR=2.2; 95% CI 1.9-2.5) were at increased risk for lymphoma. The lymphoma risk
following start of a first TNFi (mean follow-up 5.9 years, maximum follow-up 15
years) or start of a first bDMARD were similar and
not significantly increased compared to bio-naïve RA patients (HR bDMARD=1.2; 95% CI 0.9-1.6).  There were no significant differences in HRs
with age and calendar period at bDMARD start, time
since treatment start or time on active bDMARD treatment
in bDMARD treated vs. bio-naive RA. We noted a higher
point estimate of lymphoma risk in men (HR=1.5) vs. women (HR=1.1); this difference
was borderline significant (p=0.04).

There were no clear differences in
lymphoma risks for different TNFi drugs (infliximab, etanercept,
adalimumab). Due to few events, we abstained from
assessing lymphoma risks for specific other bDMARDs.

Compared to the general population, bio-naive
and bDMARD-treated RA patients were both at increased
risks for most common lymphoma subtypes (Table).

Conclusion:
Overall,
the lymphoma risk in RA remains
increased compared to the general population. TNFi-treatment with a mean follow-up
for 5.9 years or overall bDMARD therapy does not substantially
influence this risk. The distribution of lymphoma subtypes warrants further
assessment.