Lymphoma Risk in Rheumatoid Arthritis in the US: No Association with AntiTNF Therapy Frederick Wolfe

1
Lymphoma Risk in Rheumatoid Arthritis in the US
No Association with Anti-TNF Therapy  Frederick
Wolfe1 Kaleb Michaud1,2 1National Data Bank for
Rheumatic Diseases, Wichita, KS 2Stanford
University, Stanford, CA
Background. It is generally acknowledged that the
risk of lymphoma is increased in persons with RA,
and population based studies from Sweden and
other European countries suggest that the risk is
increased by approximately 1.9 times (Askling,
2005). Although no lymphoma increase was found in
patients treated with anti-TNF therapy in the
Askling study and other studies (Baecklund,
2004), treatment risk remains a major concern of
regulatory authorities throughout the world and
is the subject of warnings by regulators,
pharmaceutical companies and physicians.
Objectives. To determine the risk of lymphoma
in a large RA population in which treatment with
anti-TNF therapy was common, and to determine
whether anti-TNF therapy was associated with
increased risk. Methods. Participants were
members of the U.S National Data Bank for
Rheumatic Diseases longitudinal study of
rheumatic disease outcomes. Between 1998 and June
2005, we followed 22,151 RA patients and 6,302
patients with non-inflammatory musculoskeletal
disorders (MSD) with semiannual questionnaire
assessments. All lymphoma reports were followed
up on by contact with the patients for specific
details, and then followed up by a request for
hospital records or physician confirmation. We
also searched the National Death Index annually
and received reports of deaths from family and
friends. To determine expected rates of
lymphoma, we used the MSD group and the U.S. SEER
data bank as comparison populations. The MSD
group was assessed by the same methods used to
assess RA and therefore may be considered an
ideal comparison group. The SEER (Surveillance,
Epidemiology, and End Results) Program of the
National Cancer Institute (NCI) is an
authoritative source of information on cancer
incidence and survival in the United States. SEER
currently collects and publishes cancer incidence
and survival data from population-based cancer
registries covering approximately 26 percent of
the US population. We used age and sex categories
from the SEER database to estimate the
Standardized Incidence Ratio (SIR) for lymphoma
in the RA study population compared with the US
population.
Methods (continued). Lymphomas were categorized
as Hodgkins lymphoma, non-Hodgkins lymphoma
(NHL) and chronic lymphocytic leukemia (CLL).
Current SEER analyses do not include CLL as a
part of NHL and we did not analyze CLL cases
unless the CLL case also was diagnosed separately
as NHL. We used life table methods, Cox
regression and conditional logistic regression
(CLR) to compare groups and categorize patients
and groups. CLR, conditioning on semiannual entry
and exit periods, was employed to control for
unobserved heterogeneity. The standardized
incidence ratios (SIR) were based on age and sex
adjusted comparisons to the SEER data bank.
Exposure time for biologic treatments was
calculated as time on treatment during study
observation.
Results. There were 22,151 RA patients and 6,302
MSD patients. Biologic therapy was used by 51.3
of patients (Table 1). Infliximab was used by
36.7. This large number is found because the NDB
is the site of an infliximab safety registry. At
the time of first assessment methotrexate (MTX)
was used by 56.4 of patients. 122 lymphomas
occurred during 112,543 patient years of
follow-up, including 104 in RA (89,659
patient-years) and 18 (22,860 patient-years) in
MSD (Table 2). The incidence rate for lymphoma in
RA was 116 (95 C.I. 95 to 141) per 100,000
patient years of observation. Compared with MSD,
the hazard ratio (HR) for RA was 1.9 (95 C.I.
1.1 to 3.1) (Table 3). Using the SEER database
for comparison, the SIR for lymphoma in RA
patients was 2.0 (95 C.I. 1.7 to 2.5). We
examined the association of biologic therapy and
lymphoma in RA in Table 4 using conditional
logistic regression. Model I shows no association
between biologic therapy and lymphoma (p 0.38).
When use of prednisone at first assessment and
the patient activity score (PAS) were added as
severity covariates, there was almost no change
in the odds ratio or p-value. Analysis specific
to biologics in Model I and Model II showed no
significant association between any biologic and
lymphoma. We tested whether the odds ratio of
infliximab and etanercept differed in Model II.
The chi2 was 0.85 and the p value for the
difference was 0.356. In data not shown we also
tested whether duration of exposure for each
biologic was associated with the risk of
lymphoma. No significant association was
found. Discussion. The results of this study,
using two control populations, show the risk of
lymphoma is increased by approximately two times
in RA. There were no statistically significant
increases in the risk of lymphoma associated with
specific RA treatments, including treatment with
biologics. These data are in accord with the
results of other long-term registries (Askling,
2005) and do not suggest that anti-TNF therapy is
a risk factor for lymphoma development.
Table 4. Lymphoma Risk and Biologic Therapy in RA

Table 1. Characteristics of 22,151 RA Patients



Observational studies can provide useful
real-life experience of treatment risk. However
such studies can be biased by non-random
selection and unobservable differences between
patients prescribed various treatments
(heterogeneity). In addition, the risk of
treatment and the severity of patients receiving
treatment changes with time. In these analyses we
used conditional logistic regression to assess
risk. In the CLR model, patients are compared
within groups, entry and exit time forming the
139 groups of this study. This method controls
not only for entry time, but for study drop-out
as well as severity of patients entering at
different times. Although we also controlled for
disease severity using the prednisone
prescription at entry and the combined HAQ, pain
and patient global (PAS score) (Table 4), these
measures had no effect on risk estimates. In
summary, although lymphoma is increased in RA,
this very large study found no association with
biologic therapy.
Table 2.The Rate of Lymphoma Among RA and MSD
Patients
Table 3. The Risk of Lymphoma Among Patient with
RA
References 1) Askling, J, et al. Ann.Rheum.Dis
64(10)1414-1420, 2005. 2) Baecklund, E, et al.
Current Opinion in Rheumatology. 16(3)254-261,
2004
The NDB has received research grant support from
the following companies Centocor,
Bristol-Meyers-Squibb, Amgen, Abbott, Sanofi,
Merck and Pfizer.