Malignancy and the Rheumatic Diseases

Deborah PM Symmons

Professor of Rheumatology and Musculoskeletal Epidemiology

arc Epidemiology Unit, University of Manchester

Reports on the Rheumatic Diseases Series 5 : Topical Reviews

• Most inflammatory rheumatic diseases are associated with an increased risk of non-Hodgkin's lymphoma (NHL) and skin cancer
• This risk may be further increased by treatment with immunosuppressive or anti-tumour necrosis factor (TNF) agents
• Rheumatic diseases which may involve the lung – e.g. scleroderma, rheumatoid arthritis (RA), systemic lupus erythematosus (SLE) – are associated with an increased risk of lung cancer
• Dermatomyositis, scleroderma and systemic vasculitis may present as paraneoplastic syndromes
• Women with SLE and early scleroderma should be encouraged to have regular mammograms. Women with SLE should be encouraged to have regular cervical smears, especially if they are on immunosuppressives

Introduction

There are many links between malignancy and musculoskeletal conditions. Both primary and secondary tumours may occur in bones and the soft tissues. Musculoskeletal symptoms may occur as part of a paraneoplastic syndrome. Musculoskeletal symptoms may also occur as a side-effect of cancer therapy. However, the purpose of this paper is to review the evidence that malignancy may develop as a complication of individual rheumatic diseases or their treatment.

Much of the literature focuses on the recognised link between autoimmune diseases and lymphoproliferative and haematological malignancies (LHM). The term lymphoproliferative malignancy includes non-Hodgkin's lymphoma (NHL), Hodgkin's disease and multiple myeloma. Some rheumatic diseases are also linked with an excess of malignancies at specific sites.

The most informative studies are those that compare the occurrence of malignancy in a large cohort of patients with well-characterised disease with the expected occurrence in an age–sex matched cohort of the (local) general population. The results are then expressed as standardised incidence ratios (SIRs) if the occurrence of malignancy is studied, or standardised mortality ratios (SMRs) if mortality from malignancy is studied. The Scandinavian countries are well placed to conduct such studies because of their ability to link records (e.g. of hospital admissions, or drugs prescribed for specific diseases, and the national cancer or death register) using an individual's unique identifying number.

Sjögren's syndrome

Patients with primary Sjögren's syndrome have the highest risk of all the rheumatologic autoimmune diseases of developing NHL. The largest study to date showed that patients with primary Sjögren's syndrome had an SIR of 8.7 (95% confidence interval (CI) 4.3, 15.5) of developing NHL; patients with secondary Sjögren's syndrome had an SIR of 4.5 (95% CI 1.5, 10.6).1 The NHL usually develops within the salivary gland but it can be extraglandular. The evolution from a benign polyclonal lesion through an intermediate premalignant 'pseudolymphoma' to an overt monoclonal malignant B-cell NHL is gradual.2 Many of the NHL which occur in primary Sjögren's syndrome are marginal-zone B-cell lymphomas of the mucosa-associated lymphoid tissue (MALT)-type, which are low-grade indolent tumours.3 Risk factors for the development of NHL in patients with Sjögren's syndrome include young age at onset, parotid swelling, lymphadenopathy, splenomegaly, palpable purpura and low C4 levels at baseline.1,4

The pre-malignant stage in Sjögren's syndrome may also be associated with monoclonal paraprotein, free light chains in the urine, or Type II cryoglobulinaemia.

Rheumatoid arthritis

There have been four major record linkage studies in rheumatoid arthritis (RA) which have examined the occurrence of site-specific cancers.5-8 All have shown around a 2-fold increase in the risk of NHL. The most common histological type is diffuse large B-cell lymphoma.9 In the general population the lifetime risk of lymphoma (NHL plus Hodgkin's disease) is around 1%. The incidence has increased quite significantly over recent decades but does seem to have plateaued since around 1990.

Although immunosuppressive therapy may further increase the risk of NHL (see below), there is good evidence that patients with RA have an increased risk of NHL that is independent of treatment. The first reports of the association between RA and NHL occurred in the 1970s before the widespread use of immunosuppressives. A study of patients with Felty's syndrome found a 13-fold increased risk of developing lymphoma which was independent of treatment.10 Nested case-control studies of RA patients with NHL have found evidence of a link with cumulative disease activity and damage11 and co-morbidity12 but not with specific therapies. It is not clear to what extent secondary Sjögren's syndrome contributes to the excess risk of NHL in RA or other rheumatic diseases.

In the general population the lifetime incidence of lung cancer is 5% for men and 2% for women. It is thus of concern that all four record linkage studies report a 25–50% increased risk of lung cancer in RA patients.5-8 Much of this excess may be attributable to the higher rates of, in particular past, but also current smoking among RA patients. Rheumatoid lung disease (interstitial fibrosis) may also play a part in the aetiology of the excess lung cancers. There are also reports of an increased incidence of non-melanoma skin cancers in RA. On the other hand, patients with RA have a reduced incidence of colorectal cancer. This may be due to a protective effect of non-steroidal anti-inflammatory drugs (NSAIDs).13

Systemic lupus erythematosus

There have been several large cohort or record linkage studies of the association between SLE and cancer.14-21 The most recent was a multinational collaborative study of 9547 patients from 23 centres in 6 countries. All the studies have found an increased risk of NHL. In the most recent study, the SIR was 3.6 (95% CI 2.6, 4.9). As in RA, the predominant NHL type is diffuse large B-cell lymphoma.22

SLE also appears to be associated with an increased risk of lung cancer. For example, two record linkage studies of patients with a hospital discharge diagnosis of SLE found SIRs of 1.9 (95% CI 1.1, 3.1)17 and 2.7 (95% CI 2.1, 3.4).18 There has been some debate as to whether SLE is associated with an increased incidence of breast cancer. The most recent large studies have not found an association when compared with the general population. However, Bernatsky et al studied 871 women with SLE from three centres.23 They estimated the incidence of breast cancer based on an established tool called the Gail model (which includes current age, age at menarche, age at first live birth, parity, personal history of benign breast disease and family history). Comparing the actual occurrence (15 cases) with the estimated number they obtained an SIR of 2.1 (95% CI 1.1, 3.5). The risk of breast cancer was not influenced by use of the oral contraceptive pill or hormone replacement therapy (HRT).23 Two studies have reported an association with hepatobiliary cancer.17,21

There have been a number of reports of SLE being associated with an increased risk of cervical dysplasia24,25 and in situ cervical neoplasms.26 In a nested case control study, cervical dysplasia was associated with a prior history of sexually transmitted disease and oral contraceptive use. There was also evidence that immunosuppressive use further increased the risk but did not explain the excess entirely.27

Systemic sclerosis

Patients with systemic sclerosis have an increased risk of malignancy at a number of sites.28-30 Studies are, however, small because of the rarity of the condition. The risk of NHL is increased around 3-fold. There is an increased risk of breast cancer in the early years of the disease. In fact, scleroderma may be viewed as a rare paraneoplastic syndrome in some women with breast cancer. Lung cancer is a late complication of systemic sclerosis. It is related to underlying pulmonary fibrosis. Association between pulmonary fibrosis and lung cancer has been recognised for many years. Persistent inflammation is believed to lead to repeated cycles of damage and repair of respiratory endothelium and eventual malignant transformation. The increase in lung cancer is independent of smoking. The cancers develop predominantly in the peripheral lung zones and in the lower lobes where fibrotic changes predominate. The cancers may be multicentric. Reports differ as to the predominant cell type – small cell, squamous cell or adenocarcinoma.31 Patients who smoke have a further additional risk, as in the general population. Other malignancies reported to be associated with scleroderma include non-melanotic skin cancer30 and carcinoma of the tongue.32

The risk of cancer is substantially higher in patients with diffuse scleroderma than in those with limited scleroderma. No other risk factors have been identified.

Dermatomyositis and polymyositis

The first link between dermatomyositis and malignancy was published in 1916. Since then, a number of studies have shown that patients with dermatomyositis (DM) and polymyositis (PM) have an increased incidence of malignancy at the time of diagnosis and in the next 2 years. A meta-analysis of four case-control and cohort studies found that cancer risk was high in the 4 years before and after the diagnosis of DM. For PM, the risk was increased only in the 5 years after diagnosis. This suggests that DM may be a paraneoplastic syndrome in some cases.33 A recent record linkage study from Scotland found that patients with DM had an almost 8-fold increase in concurrent or subsequent malignancy compared with the general population (SIR 7.7; 95% CI 5.7, 10.1) and patients with PM a 2-fold increased risk (SIR 2.1; 95% CI 1.5, 2.9).34 A study from Australia found very similar results, with patients with DM having an SIR of 6.2 (95% CI 3.9, 10.0) compared with the general population and patients with PM having an SIR of 2.0 (95% CI 1.4, 2.7). They also found an increased risk associated with inclusion body myositis (SIR 2.4; 95% CI 1.2, 4.9).35 The risk of malignancy is particularly high in patients aged over 65. In one report, almost 50% of patients aged over 65 with DM were found to have an underlying malignancy.36 Evidence of capillary damage on muscle biopsy, cutaneous necrosis on the trunk and leucocytoclastic vasculitis have also been reported to be associated with an increased risk of underlying malignancy in DM.

In a study of 198 cases of DM with malignancy, the risk of ovarian (11-fold), lung (6-fold), pancreatic (4-fold), stomach (4-fold) and colorectal (3-fold) cancers and NHL (4-fold) were increased compared with the general population.37 The Scottish study also found an association with lung, stomach, ovary and colon cancers and NHL – as well as cervical cancer. In 137 cases of PM with malignancy, the risks of NHL (4-fold) and lung (3-fold) and bladder cancers (2-fold) were increased.34

The presence of malignancy does not influence the clinical manifestations or response to treatment of the myositis.

Ankylosing spondylitis

During the 1940s, radiotherapy of the spine was widely used in the treatment of ankylosing spondylitis (AS). Court Brown and Doll identified all 14,554 AS patients who had received radiotherapy between 1935 and 1954 in one of the 87 radiotherapy centres in the UK.38 They obtained adequate follow up of 98% of patients up to the beginning of 1960. These AS patients had 9.5 times the expected rate of leukaemia and 1.4 times the expected rate of cancer at other heavily irradiated sites. When the follow up was extended to January 1992, the SMR for cancer remained elevated at 1.3 (95% CI 1.2, 1.4).39 The follow-up study also included 1021 AS patients diagnosed at the same centres between 1935 and 1957 who did not have radiotherapy. These patients presumably had milder disease. They did not have an excess cancer mortality.

A more recent study linked data from 6918 patients who were discharged from hospital in Sweden between 1965 and 1995 with a diagnosis of ankylosing spondylitis and the national cancer and death registers. The SIR for all cancers was not increased (SIR 1.05; 95% CI 0.95, 1.17). As in RA, there was a reduced incidence of rectal cancer (SIR 0.41; 95% CI 0.15, 0.89), which may be attributable to a protective effect of NSAIDs. The risk of renal cancer was increased (SIR 5.90; 95% CI 1.61, 15.1). The risk of LHM was not significantly increased (SIR 1.34; 95% CI 0.93, 1.89).40

Psoriatic arthritis

There have been no studies of malignancy in psoriatic arthritis (PsA). A longitudinal study of PsA from a tertiary referral centre in Toronto did show an increased mortality rate in both men (standardised mortality ratio (SMR) 1.65; 95% CI 1.09, 2.40) and women (SMR 1.59; 95% CI 1.04, 2.33).41 There was no increase in the rate of deaths from cancer (SMR 0.73; 95% CI 0.31, 1.46).

There have been a few relevant studies in psoriasis. In a record linkage study from Finland data from 5687 patients discharged from hospital with psoriasis from 1973–1984 were linked to the national cancer registry.42 The investigators found an overall excess in cancer incidence (SIR 1.3; 95% CI 1.2, 1.4). The estimated relative risks were highest for Hodgkin's disease (SIR 3.3; 95% CI 1.4, 6.4), squamous cell skin cancer (SIR 3.2; 95% CI 2.3, 4.4), NHL (SIR 2.2; 95% CI 1.4, 3.4) and laryngeal cancer (SIR 2.9; 95% CI 1.5, 5.0). There was also an increase in lung cancer (SIR 1.4; 95% CI 1.2, 1.8) and liver cancer in men (SIR 2.7; 95% CI 1.3, 5.0). A similar record linkage study involving 6910 patients with a hospital discharge diagnosis of psoriasis in Denmark found an excess of lung cancers (SIR 1.4 in men and 1.6 in women), laryngeal and pharyngeal cancers in men, and colon and kidney cancer in women,43 as well as a 2.5-fold increased risk of non-melanotic skin cancers in men and women. A study using the UK General Practice Research Database (GPRD) found an excess of lymphomas in a cohort of 2718 patients with psoriasis aged over 65 (adjusted relative risk 2.94; 95% CI 1.82, 4.74).44

Laryngeal, lung and pharyngeal cancers are all related to smoking, and liver cancer is related to alcohol. A nested case-control study in the Finnish cohort showed no evidence that the cases of NHL were related to treatment of the psoriasis.42 Similarly, in the UK GPRD the lymphoma cases did not appear to be related to treatment with either ciclosporin or methotrexate.44

These observations suggest that psoriasis is associated with an increased risk of lymphoma independent of treatment. It is very likely that patients with PsA share this risk. Similarly there is evidence of excess malignancy related to smoking and alcohol in patients with psoriasis which again is likely to be present in patients with PsA.

Systemic vasculitis

There has been a suggestion that vasculitis may occasionally occur as a paraneoplastic phenomenon. Kurzrock et al reviewed over 200 patients who had simultaneous cancer and vasculitis and found that the majority had haematological malignancies. Only 36 of these patients had solid tumours.45 Tatsis et al later studied 477 patients with Wegener's granulomatosis and compared them with 479 control patients with RA in order to establish the frequency of malignancy occurring before or simultaneously with the diagnosis of the Wegener's.46 Although there was no significant difference in the frequency of malignancy between the Wegener's (48%) and the RA groups (38%), 7 of the 23 malignancies in the Wegener's group were renal cell carcinomas.

An analysis of 1065 Swedish patients with a hospital discharge diagnosis of Wegener's granulomatosis from 1969 onwards found a 2-fold overall increased risk of cancer after diagnosis compared with the general population. The risk was most pronounced for bladder cancer (SIR 4.8; 95% CI 2.6, 8.1), squamous cell skin cancer (SIR 7.3; 95% CI 4.4, 12), lymphoma (SIR 4.2; 95% CI 1.8, 8.3) and leukaemia (SIR 5.7; 95% CI 2.3, 12).47 The increased risk of bladder cancer is likely to be related to treatment with cyclophosphamide.

Immunosuppressive therapy and malignancy

A link between immunosuppressive therapy and malignancy (in particular LHM) was first reported in 1979.48 Silman et al49 reported an excess of LHM in RA patients treated with high dose azathioprine compared with non-azathioprine treated RA controls. The European League Against Rheumatism (EULAR) Immunosuppressive Registry reported an adjusted incidence rate ratio of 3.7 (95% CI 1.5, 9.5) of developing an LHM, skin or bladder cancer in patients with a rheumatic disease treated with an immunosuppressive drug for 6 or more years, compared with those treated for less than 1 year.50

The risk of bladder cancer is particularly high following cyclophosphamide treatment and persists for at least 17 years after stopping the drug.51 The use of chlorambucil is associated with leukaemia in patients with RA, SLE and juvenile arthritis. There are reports of reversible lymphomas occurring in patients treated with methotrexate. Most of these lymphomas were Epstein–Barr virus positive. Overall, no excess risk of NHL was found in a 3-year prospective study of around 27,000 RA patients treated with methotrexate in France52 or in patients treated with methotrexate in the US National Databank for Rheumatic Diseases.12

The risk of immunosuppressive therapy is in addition to that conferred by the underlying disease. The attributable risk of azathioprine (i.e. the risk over and above the disease that is being treated) is around 6 per 1000 patient years.53

Lymphoma risk in anti-TNFα treated patients

Tumour necrosis factor alpha (TNFα) promotes a cytotoxic T-cell response against malignant B cells.54 There is, therefore, a concern that treatment with TNFα antagonists may further enhance the risk of NHL in patients with inflammatory musculoskeletal conditions. Conversely, given the strong link between the risk of NHL and cumulative disease activity in RA11 it is possible that, by improved disease control, TNFα antagonists might actually reduce the risk of NHL.

By 2002, 26 cases of NHL following treatment with infliximab or etanercept had been reported to the US Food and Drug Administration.55 Interestingly, several of these occurred very soon after starting treatment. However, there were no denominator data so it is impossible to know whether this number of cases is more than would be expected in a cohort of RA patients. A study conducted by the US National Databank for Rheumatic Diseases followed 18,572 patients with RA, 8614 of whom had been treated at some time with a biologic agent. SIR for NHL was almost trebled compared with the general US population.12

A recent study from Sweden56 found a hazard ratio of developing NHL in 757 RA patients treated with etanercept or infliximab (and followed for an average of 2 years) compared with 800 biologic-naive RA patients (followed for an average of 5 years) of 4.9 (95% CI 0.9, 26.2). This ratio persisted after adjusting for difference in baseline Health Assessment Questionnaire (HAQ) score, as a marker of disease severity. However, this study detected only five cases of NHL. It remains to be seen whether the relative risk of NHL will rise or fall with prolonged exposure and/or duration of follow up. There was also a 2-fold increase in smoking-related cancers compared with the general population in both the anti-TNF treated and the comparison RA cohort.

Implications for clinical practice

Thus it is clear that most autoimmune inflammatory rheumatic disorders are associated with an increased risk of NHL (Table 1). This risk is further enhanced by the use of immunosuppressants and biologic agents. The average rheumatologist might expect to see one case of NHL in his/her RA patients every 1–2 years. The clinician needs to be alert to the fact that features such as fever, weight loss, night sweats and lymphadenopathy may indicate the development of lymphoma rather than simply a flare in the underlying disease.

TABLE 1. Association between non-Hodgkin's lymphoma and rheumatologic autoimmune disease. Disease Typical SIR 95% CI Reference Risk factors 1° Sjögren's syndrome 8.7 (4.3, 15.5) Kauppi et al 19971 Young age at onset Parotid swelling Lymphadenopathy Splenomegaly Irradiation of parotid 2° Sjögren's syndrome 4.5 (1.5, 10.6) Kauppi et al 19971   Wegener's granulomatosis* 4.2 (1.8, 8.3) Knight et al 200247   Polymyositis 3.7 (1.7, 8.2) Buchbinder et al 200135   Dermatomyositis 3.6 (1.2, 11.1) Buchbinder et al 200135 Capillary damage on muscle biopsy Cutaneous necrosis on trunk Leukocytoclastic vasculitis SLE 3.6 (2.6, 4.9) Bernatsky et al 200521   Systemic sclerosis 2.9 (0.8, 7.4) Rosenthal et al 199530   Rheumatoid arthritis 2.1 (1.7, 2.6) Thomas et al 20008 Longer disease duration Felty's syndrome * 7 of 8 lymphomas were non-Hodgkin's CI confidence interval; SIR standardised incidence ratio; SLE systemic lupus erythematosus

The link between autoimmune rheumatic diseases and non-melanotic skin cancers is also clear but much less well publicised. Because of their psoriasis, patients with PsA are also at increased risk of skin cancer (Table 2). Again, the risk of skin cancer is magnified by immunosuppressive use. A high index of suspicion needs to be maintained when new skin lesions develop.

Autoimmune rheumatic disorders which may be complicated by interstitial lung disease are associated with an increased risk of lung cancer whether or not the patient smokes (Table 2). Smokers, of course – as in the general population – have a further increased risk. A case can be made for such patients to have regular chest x-rays.57

TABLE 2. Cancers other than non-Hodgkin's lymphoma associated with rheumatologic conditions. Disease Lung Skin Others Rheumatoid arthritis ↑ ↑   SLE ↑ ↑↑ Hepatobiliary Scleroderma ↑↑ ↑ Tongue Dermatomyositis ↑↑ – Stomach, ovary, pancreas, colon, cervix Wegener's granulomatosis ↑ ↑↑ Bladder, liver, brain ↑ = SIR between 1 and 2 ↑↑= SIR between 2 and 10 SIR standardised incidence ratio; SLE systemic lupus erythematosus

It is recognised that patients with chronic rheumatic diseases often fail to attend for regular cancer screening offered at the population level.58 There may be a number of reasons for this. Patients may be too unwell to attend. They may feel falsely reassured that regular hospital attendance means that any cancers will be picked up anyway. As part of comprehensive care of the patient with autoimmune rheumatic disease the clinician should emphasise the importance of responding to invitations to attend for cancer screening (Table 3). In particular, women with SLE should be encouraged to attend for regular cervical smears and mammograms. It has been suggested that women with SLE on immunosuppressives should have annual smears.27 Women with scleroderma should also be offered regular mammograms in the early years of disease.

TABLE 3. Screening for malignancy in patients with autoimmune rheumatic disorders. Disease Screening/investigation All High index of suspicion for NHL and skin cancer Pulmonary fibrosis Regular chest x-ray Women with SLE Regular cervical smears and mammograms Women with scleroderma Regular mammograms in early disease NHL non-Hodgkin's lymphoma; SLE systemic lupus erythematosus

It is important to consider the possibility of an underlying malignancy in patients who present with DM or PM – particularly older patients. Investigations should be guided by the patient's clinical symptoms and signs. All patients should have a full blood count, ESR, biochemical profile, urinalysis and chest x-ray. Women should also undergo mammography, pelvic ultrasound and serum cancer antigen (CA)-125 evaluation. There may also be a case for computerised tomography (CT) scanning of the chest, abdomen and pelvis. Malignancy is likely to be detected in 15–25% of cases. Vigilance needs to be maintained for at least 5 years after diagnosis, particularly if the myositis relapses after a period in remission.

Other Titles in this series

• June 2008 No 14
  Sports-Related Musculoskeletal Disorders
• October 2007 No 13
  Glucocorticoids in Rheumatic Diseases
• June 2007 No 12
  Immunisation of Immunosuppressed Patients with Rheumatic Diseases
• February 2007 No 11
  Rheumatological Manifestations of Malignant Haematological Conditions
• October 2006 No 10
  Work-Related Disorders of the Upper Limb
• June 2006 No 9
  Qualitative Research: Its Relevance and Use in Musculoskeletal Medicine
• February 2006 No 8
  Rheumatic Diseases: Serological Aids to Early Diagnosis
• June 2005 No 6
  Systemic Amyloidosis in the Rheumatic Diseases
• February 2005 No 5
  Heritable Collagen Disorders
• October 2004 No 4
  An Update on Gout
• June 2004 No 3
  Surgery for Arthritis: Total Hip and Knee Joint Replacement
• February 2004 No 2
  Idiopathic Inflammatory Myopathies
• September 2003 No 1
  Low Back Pain

download a pdf of this publication

6627.pdf
(188Kb)