Lymphoma in rheumatoid arthritis: An immune system set up for failure


Despite its name, rheumatoid arthritis (RA) is much more than a joint disease. Extraarticular manifestations may occur in almost any organ system, with nodule formation, interstitial lung disease, and leukocytoclastic vasculitis the more frequent indicators of the multisystemic character of the syndrome (1). Indeed, inflammatory blood vessel disease could be considered to be one of the common denominators connecting chronic inflammatory tissue injury in diverse organ systems. Even the increased risk of cardiovascular complications in patients with RA could be interpreted as a variant of arterial wall inflammation (2).

Comprehensive epidemiologic studies have shown that a state of smoldering inflammatory stress is associated with increased risk of cardiovascular disease, even in individuals who do not have an autoimmune disease (3). The site of that smoldering inflammation remains a matter of debate, but the atherosclerotic lesions in arterial vessels are obvious candidates. Atherosclerotic lesions that are prone to rupture and give rise to acute coronary syndromes are known to typically have an inflammatory infiltrate of activated T cells and macrophages (4). It is plausible that inflammation is more intense in the atherosclerotic lesions of patients with RA, accelerating the process of instability and its complications. Essentially, one could conclude that all manifestations of RA, located inside or outside of the synovium, are ultimately caused by unremitting inflammation, initiated and sustained by a complex interplay of genetic susceptibility factors, environmental stressors, and sheer “bad luck.”

However, some manifestations of RA seem not to follow this rule. It is not straightforward to explain Felty's syndrome as a consequence of chronic inflammation. And how would chronic inflammation, elicited by any instigator, increase the risk of infection acquisition in the host (5, 6)? Finally, there is now good evidence that patients with RA have a 2-fold increased risk of developing lymphoma (7, 8). Specifically, patients are alerted to an increased risk of lymphoma when considering treatment with a tumor necrosis factor α (TNFα) blockade agent, but it is difficult to dissect the spontaneous risk from treatment-induced risk (9).

Increased susceptibility to infections and malignancies are typical features in an immunocompromised host (10–13), posing the question of whether RA itself, the therapeutic interventions, or a combination of both is associated with fundamental immunodeficiency. Is it possible that patients with RA, clearly in a much more subtle way than patients with primary immunodeficiencies, lack immunocompetence, which renders them susceptible to infection and lymphoma? Immunosuppression is the standard of care for RA, and, thus, it would seem logical to suggest that the therapy itself, and not the disease, is responsible for weakening the patients' immune response. The discussions on whether infectious complications and lymphomagenesis in RA patients are a consequence of too little or too much immunosuppression have posed quite a dilemma. Are patients paying a price for more effective antiinflammatory control in that immunosuppressive interventions impair their tumor surveillance? Or is the disease itself a risk factor for lymphoma development, with the possibility that some treatments may even reduce the occurrence of lymphoid neoplasm?

This question of whether lymphoma in RA may result from insufficient or too-aggressive immunosuppressive therapy has been addressed in an elegant study by Baecklund and colleagues, reported elsewhere in this issue of Arthritis & Rheumatism (14). By making use of unique patient and tissue registries in Sweden and tracking large cohorts of RA patients over decades, those researchers identified 378 consecutive cases with RA and subsequent malignant lymphoma and compared them with 378 control patients from a population-based cohort of 74,651 Swedes followed up for the diagnosis of RA. The study cohort covered the time period between 1964 and 1995.

Baecklund et al should be applauded for their care in constructing case definitions and for their relentlessness in tracing every possible case of lymphoma associated with RA. Lymphoma tissue blocks were secured, and all specimens were reexamined to account for shifts in diagnostic criteria operational over the 3 decades included in the study. Importantly, tissues were examined for Epstein-Barr virus (EBV), since lymphomas seen in immunodeficient hosts are typically EBV related. Finally, the researchers invested considerable effort into developing a system for estimating disease activity from information extracted from medical records.

To address the question of inflammation versus immunosuppressive therapy as a causative factor for lymphomagenesis, they recorded treatments given to cases and controls. The investigators found that RA patients with high cumulative disease activity had a dramatically increased risk of developing lymphoma. Approximately half of the lymphomas were categorized as diffuse large B cell lymphoma (DLBCL), and within this subset, EBV infection was detected in 19 of 160 tumors (12%). Management of RA in Sweden prior to 1995 included nonsteroidal antiinflammatory drugs in almost 90% of patients and aspirin in ∼60–70%. Approximately half of the patients in Baecklund and colleagues' study received corticosteroids, and intraarticular steroids were used in 44% of the lymphoma cohort and 63% of the nonlymphoma controls. Just over 70% of the patients were treated with disease-modifying agents, with antimalarial agents being by far the most frequently used of these drugs. Few patients received azathioprine (44 of 756) or methotrexate (43 of 756). Obviously, none of the patients treated prior to 1995 was exposed to TNFα blockade therapy.

Comparison of drug use in the cases and controls revealed increased of lymphoma risk in the small group of azathioprine-treated patients. Intramuscular gold, methotrexate, and sulfasalazine did not affect the risk of lymphoma. Most interestingly, oral steroids reduced the risk of lymphoma, with an odds ratio of 0.6. This benefit was even more impressive with intraarticular steroids, reducing the odds ratio to almost one-third. It is somewhat surprising that patients in the RA-plus-lymphoma category were essentially treated very similarly to the controls, although they had more active and more severe disease, with 23% of patients having high inflammatory activity, 28% categorized as being in Steinbrocker functional class III (15), and 14% assigned to functional class IV. One would expect that rheumatologists would make an effort to suppress aggressive disease activity and utilize available disease-modifying antirheumatic drugs (DMARDs) to the fullest; however, only 71% of the cases (compared with 74% of patients without lymphoma) had received DMARD treatment. The higher disease activity in the lymphoma group would have easily explained a bias for more immunosuppressive therapy, but this was not the case. It is therefore the more remarkable that only 48% of the lymphoma patients received oral steroids and only 44% of that patient cohort was treated with intraarticular steroids.

Baecklund et al's investigation builds upon the superb resources of Sweden in biobanking tissues, maintaining medical records, and carefully assembling patient registries. This study will probably be the last in which the impact of RA therapy before powerful immunosuppressive agents became available can be tested with confidence, and it thus provides a valuable set of data.

EBV is an oncogenic virus and has been implicated in the development of certain types of lymphoma (16, 17). It is typically found in immunodeficiency-associated DLBCL, which raises a question regarding the extent to which the lymphoma cases assessed in the study by Baecklund and coworkers can be considered a manifestation of failing antiviral immunity. Although low in absolute numbers, the EBV infection rate of 12% among DLBCLs is actually quite high since DLBCLs in immunocompetent hosts are consistently negative for EBV (18).

RA patients diagnosed as having lymphoma had a mean age of 70 years (range 32–91); therefore, a significant proportion of individuals in the eighth, ninth, and tenth decades of life were included. Even under the best conditions, immunocompetence in such individuals is reduced, due to the process of immunosenescence (19). Typical risk factors for EBV-related lymphomagenesis include old age, organ transplantation, and immunodeficiency disorders (12, 20). We have to assume that the lymphomas found among the 378 RA cases were related to a variety of risk factors and indeed represent a mixture of malignant tumors. Is it possible that the 12% of DLBCLs that were EBV positive are the excess malignancies attributable to the underlying disease process in these patients? The report by Baecklund et al does not include information on lymphomas in Swedes without RA, a comparison that could address some aspects of this issue.

Acquired immunodeficiency syndrome (AIDS)–related non-Hodgkin's lymphoma (NHL) has emerged as an instructive biologic model to investigate the development and progression of high-grade NHLs, particularly NHLs that arise in immunodeficient hosts (21). Several of the typical features described for NHLs in human immunodeficiency virus (HIV)–infected hosts may be informative when addressing how the chronic inflammatory disease RA can confer risk of B cell malignancy. In HIV-infected individuals, the B cell neoplasms are clinically aggressive, 80% arising in nodal or extranodal sites and 20% arising as primary central nervous system lymphomas. The report by Baecklund and colleagues does not provide data on the course of the neoplastic disease in the RA patients. Did it resemble the clinical aggressiveness of lymphomas encountered in the setting of severely immunocompromised individuals? Did the EBV-positive tumors occur in elderly patients who had accumulated multiple risk factors?

Interestingly, recent reports have described polymorphic lymphoproliferative disorders associated with HIV infection, stressing the multiplicity of pathogenic pathways involved. Careful studies of AIDS-related NHLs have led to the conclusion that several factors contribute to lymphomagenesis by supporting induction of oligoclonal B cell expansions. The malignant B cell clones are then characterized by a spectrum of genetic lesions, again emphasizing the multiplicity of promalignant conditions encountered in such hosts. Specifically, AIDS-related NHLs are known to show EBV infection, rearrangement of the myc gene, mutations and deletions of p53, ras gene mutations, and gene rearrangement of bcl-6. There is evidence that the molecular lesions in DLBCLs in immunocompetent individuals may have a different pattern, but also include activation of the protooncogenes rel, muc-1, bcl-8, and c-myc (18). Principal disease mechanisms generating permissive conditions for the outgrowth of malignant B cell clones include virus-induced immunosuppression, failing immunosurveillance, profuse production of growth factors and cytokines, and ongoing antigenic stimulation in the chronically infected host.

Which of these conditions could apply to RA? RA patients have increased levels of B cell growth and survival factors, such as B lymphocyte stimulator (BLyS; trademark of Human Genome Sciences, Rockville, MD) and APRIL (22). If present in overabundance, such TNF-like cytokines could drive B cell expansion. Recent studies demonstrate that BLyS and APRIL are produced in synovial lesions and that different types of rheumatoid synovitis can be distinguished according to the functional activity of these TNF-like cytokines. However, blocking studies targeting BLyS and APRIL have revealed that both pro- and antiinflammatory mechanisms depend on these cytokines (22). Also, it is unlikely that TNFα itself is a lymphoma-promoting factor, since the risk of lymphoma development appears to be higher, not lower, in patients treated with TNFα-blocking agents (23,24).

Alternatively, it could be suggested that RA is associated with chronic viral infection, such as with the oncogenic herpes virus EBV, which generates permissive conditions for the development of EBV-related NHL (25, 26). Reports of spontaneously remitting NHL in RA patients after discontinuation of methotrexate therapy would support this notion, although numbers of such cases are low, and no sufficiently powered studies are available (27).

Finally, one needs to consider the question of whether RA patients, particularly those with high inflammatory activity as described by Baecklund et al, are genuinely immunoincompetent (Figure 1). Studies focusing on a hallmark of a competent immune system, the diversity of the T cell pool, have indicated that RA is associated with a marked contraction in T cell diversity (28). The entire T cell pool in RA patients has been under excessive proliferative stress, as is demonstrated by the premature erosion of telomeres (29, 30), the protective ends of chromosomes that are not replicated during cell division and that can be used as surrogate markers of the true biologic age of lymphocytes. Intriguingly, premature cellular aging not only affects lymphocytes, but is also present in the granulocytes of RA patients, implicating accelerated replicative stress of the hematopoietic progenitor cell as a potential disease mechanism (31). Most importantly, age-inappropriate losses of telomeres in both myeloid and lymphoid cells occur in healthy individuals with the RA-associated HLA–DR4 haplotype (31).

Figure 1.

Model of immunoincompetence in rheumatoid arthritis (RA). Endogenous and exogenous factors contribute to the state of immunodeficiency in RA. The model proposes that chronic synovial inflammation is a manifestation of immune failure, with the immune system losing protective tolerance mechanisms. Other consequences of an impaired immune system include increased risk of malignancy, particularly B cell neoplasm.

These findings have led to the model that RA is a disease of premature aging (32). Premature immunosenescence results in a spectrum of immunologic deviations, specifically the accumulation of senescent lymphocytes that release high amounts of proinflammatory cytokines (33). By virtue of an altered gene expression program, senescent T lymphocytes become responsive to a new set of environmental signals and prosper in the synovial microenvironment. The premature aging of B cells as well as T cells needs to be considered as one of the risk factors predisposing an individual with RA to neoplasm development.

The study by Baecklund et al holds an important lesson for RA patients and for treating rheumatologists. Inflammatory activity may be the not cause of, but an epiphenomenon of, immune failure (Figure 1). The inability to control unwanted inflammation, often referred to as loss of tolerance, may be directly connected to the inability to survey lymphoid and extralymphoid organs for neoplasms and suppress the outgrowth of monoclonal B cell clones. Consideration of this model, however, does not make it any easier to design the best possible treatment strategy for each patient. Highly effective suppression of common inflammation pathways would not be likely to restore immunocompetence. On the contrary, it may be worthwhile to keep in mind that it is easy to aggravate immunodeficiency in a borderline-competent host. The work by Baecklund and colleagues, though, provides fascinating insight into the unpredictable effects of different therapeutic reagents used to manage RA. Antimalarial drugs, sulfasalazine, and intramuscular gold appeared to be neutral with regard to lymphoma risk. Azathioprine slightly increased the risk, but case numbers were small. Methotrexate was given to few patients but did not seem to increase risk. Intriguingly, oral, as well as intraarticular, steroids conferred significant protection. This provides reassurance that we can “have our cake and eat it, too.” We can simultaneously suppress rheumatoid inflammation and reduce the risk for lymphoma development.


The authors thank Tamela Yeargin for manuscript editing.