EDITORIAL - Lymphoma and RA - again

[Guest guest]

Rheumatology Advance Access originally published online on August 27, 2006

Rheumatology 2007 46(1):1-2; doi:10.1093/rheumatology/kel303

--------------------------------------------------------------------------------

EDITORIALS

Lymphoma and rheumatoid arthritis—again

D. P. M. Symmons

Professor of Rheumatology and Musculoskeletal Epidemiology, University

of Manchester, Manchester M13 9PT, UK

The link between lymphoma and rheumatoid arthritis (RA) continues to

intrigue. It was first noted in a record linkage study from Finland

published in 1978 [1]. Since then, there have been numerous reports of

an increased incidence of lymphoma—in particular, non-Hodgkin's

lymphoma (NHL) and more particularly diffuse large B-cell lymphoma

(DLBCL) [2]—in patients with RA from around the world [3–6]. It can

now be considered beyond dispute that RA and, indeed, many other

autoimmune diseases are associated with an increased risk of NHL.

However, the question remains as to the relative contribution of the

underlying disease and the drugs used to treat it to the aetiology of

the malignancy. The advent of each new treatment for RA seems to

re-open the debate.

There is now considerable evidence that rheumatoid disease itself

plays an important role in the aetiology of the NHL. The increased

risk is manifested fairly early in the disease course. In a study from

the Norfolk Arthritis Register (NOAR), the increased risk was apparent

within the first 5 yrs [7]. At a median follow-up of 8.4 yrs, the

standardized incidence ratio (SIR) of lymphoma compared with the

general population was 2.4 (95% CI 1.2, 4.2). The risk was highest

amongst those who were rheumatoid factor (RF) positive [sIR = 3.6 (95%

CI 1.3, 7.8)]. Other studies have suggested that the risk of NHL is

related to disease duration. However, this may simply be a function of

allowing cumulative disease activity to accrue. A recent meticulous

study from Sweden identified 378 patients with RA and lymphoma by

linking the in-patient registry and the cancer registry [8]. The

lymphomas were diagnosed between 1964 and 1995. Three control RA

patients matched for gender, year of birth, year of first discharge

with RA and county of residence were selected for each case.

Information on disease activity was extracted for every visit for each

case and each control. A relationship was found both with the average

disease activity [highest vs lowest quartile unadjusted odds ratio

(OR) 71.3, 95% CI 24.1, 211.4] and cumulative disease activity (10th

decile vs 1st decile OR 61.6, 95% CI 21.0, 181.0).

Baecklund et al. [2] found that 22 of the 35 NHL (67%) that they were

able to characterize were DLBCL. A similar proportion was found in the

Norfolk study [7]. This compares with 30–40% of sporadic NHL in the

general population. Immunodeficiency, in particular, impaired T-cell

function such as has been reported in RA, is the main risk factor for

NHL [9]. Such T-cell impairment may permit the proliferation of the

oncogenic herpes virus Epstein–Barr virus (EBV). Up to 90% of

post-transplant NHLs are EBV positive [9]. In the 30 NHL tested by

Baecklund et al. [2], 5 (17%) were EBV+ including 4 out of 22 (18%) of

the DLBCL. This compares with <5% EBV positivity in sporadic DLBCL

[10].

Azathioprine was the first drug to be investigated with regards to

enhancing lymphoma risk in RA. An association between azathioprine

therapy and NHL had first been noted in transplant patients [11]. In

1988, Silman et al. [12] reported an excess of NHL in RA patients

treated with high dose azathioprine compared with non-azathioprine

treated RA controls. The Canadian registry of azathioprine use

reported an 8-fold increased risk of lymphoproliferative disorders

based on four cases in 530 patients [13]. The European League against

Rheumatism (EULAR) Immunosuppressive Registry found an adjusted

incidence rate ratio of 2.9 (95% CI 0.7, 12.7) of developing NHL in

patients with rheumatic diseases who had been treated with

azathioprine for 6 yrs compared with those treated for <1 yr [14]. In

the recent Swedish case control study, azathioprine was the only

disease-modifying anti-rheumatic drug (DMARD) associated with an

increased risk of NHL (adjusted OR for ever vs never azathioprine 4.3;

95% CI 1.6, 12.0). The risk appeared to rise with increasing length of

exposure to azathioprine [8].

Then the spotlight fell on methotrexate (MTX). The first report of a

lymphoma developing in a rheumatoid patient treated with MTX appeared

in 1991 [15]. However, three large epidemiological studies have failed

to find an association between MTX exposure and the development of NHL

in RA patients [16–18]. In the most recent study, Wolfe and Michaud

[18] reported an SIR of 1.5 (95% CI 0.8, 2.7) based on 10 cases of

lymphoma occurring in 6396 RA patients exposed to MTX. The exception

is the Norfolk study that found a SIR of 4.9 (95% CI 1.88, 10.6) in

MTX-exposed patients with inflammatory polyarthritis but, as the

authors acknowledge, this is an unadjusted figure and may be due to

confounding by indication [7]. There are around 50 cases in the

literature of regression of NHL following withdrawal of MTX in RA

patients [19]. This is a phenomenon also seen in post-transplant

patients associated with EBV positivity. Around 16–44% of NHL

associated with MTX are EBV + [17, 20, 21]. It has been shown that MTX

can re-activate latent EBV and this may contribute to the development

of NHL in some cases [22].

Now the introduction of the anti-TNF agents once again brings

attention to the link between lymphoma and RA. TNF promotes a T-cell

cytotoxic response against B-cell malignancies, and so right from the

beginning there has been anxiety that anti-TNF therapy might be

associated with an increased risk of NHL. In 2002, Brown et al. [23]

published details of 26 cases of lymphoma that had been spontaneously

reported to the US Food and Drug Administration up to December 2000.

There were no denominator data so it was impossible to be sure whether

this was more or less than would be expected. Wolfe and Michaud [18]

reported a SIR of 2.9 (95% CI 1.7, 4.9) of NHL developing in 8614 RA

patients exposed to either infliximab or etanercept [18]. Based on

five cases occurring in 757 patients from southern Sweden exposed to

either infliximab or etanercept, Geborek et al. [24] reported a SIR of

11.5 (95% CI 3.5, 26.9). The SIR in their comparison group of

non-biologically exposed RA patients (based on two cases) was 1.4 (95%

CI 1.1, 1.8). Thus, the rate in the biologically treated group was

increased 5-fold compared with the comparison cohort [24]. By

comparison Askling et al. [25] investigated the incidence of lymphoma

in the whole Swedish Biologics Register (n = 4160) exposed to

etanercept, infliximab or adalimumab. Based on nine cases, they

estimated a SIR of 2.9 (95% CI 1.3, 5.5). However, the risk was not

increased compared with that in a prevalent cohort of RA in-patients:

risk ratio (RR) 1.1 (95% CI 0.6, 2.1). Clearly, these reports are all

based on small numbers of cases and so are somewhat unstable. They

also represent experience in the early months of treatment with

anti-TNF therapy. It remains to be seen whether long-term anti-TNF

therapy will increase or decrease the risk of NHL in RA.

A recently published meta-analysis examined the malignancy risk in

nine published randomized controlled trials (RCTs) of infliximab and

adalimumab [26]. They identified four lymphomas which occurred during

the trials and an additional six, which occurred after the trials were

over. All were in the biologically treated arm of the trial. The

authors did not present an odds ratio based on these data-focusing,

instead on the overall risk of malignancy. While observations of

increased risk coming from the Biologics Registers may be criticized

for confounding by indication (control group not adequately matched

for disease severity) or protopathic bias (patients with latent NHL

having increased joint symptoms so likely to have their treatment

changed), this does not apply to RCTs. It is possible, however, that

the biologically treated patients may have been subjected to greater

surveillance than the placebo arm in the unblinded follow-up phase of

the RCT and also that the drop-out rate may have been higher in the

placebo arm leading to less person-time of observation.

It seems likely that there is the potential for RA treatment both to

increase and to reduce lymphoma risk in RA. Lymphoma risk is greatest

in RA patients with the highest cumulative disease activity [8]. Yet

of the 35 RA patients with NHL in Baecklund's study, 17 had never

received any DMARD treatment and a further 10 had received it for <1

yr [2]. Despite the impressive efficacy of the anti-TNF agents, most

patients enrolled in the RCTs still satisfied the entry criteria for

the trial at the end of the study (i.e. they still had significant

disease activity). Perhaps the way to abolish the excess risk of NHL

in patients with RA is to treat the disease adequately but not to

overtreat.

http://rheumatology.oxfordjournals.org/cgi/content/full/46/1/1

--

Not an MD