How to use biologic agents in patients with rheumatoid arthritis who have comorbid disease
BMJ 2015; 351 doi: https://doi.org/10.1136/bmj.h3658 (Published 17 August 2015) Cite this as: BMJ 2015;351:h3658
Chinese translation
该文章的中文翻译
- John S Richards, staff rheumatologist12,
- Sharon M Dowell, assistant professor of medicine3,
- Mercedes E Quinones, assistant professor of medicine3,
- Gail S Kerr, chief of rheumatology123
- 1Veterans Affairs Medical Center, Washington DC 20422, USA
- 2Georgetown University, Washington, DC, USA
- 3Howard University, Washington, DC, USA
- Correspondence to J S Richards john.richards1{at}va.gov
Abstract
Although biologic disease modifying anti-rheumatic drugs (bDMARDs) have improved the quality of life of patients with rheumatoid arthritis, side effects remain a problem, especially for patients with pre-existing comorbidities. Randomized controlled trials of such drugs in rheumatoid arthritis typically exclude patients with comorbidities and are of short duration, so they do not provide data on adverse events in these people. Other data sources such as national bDMARD registries are limited by the preselection of patients for both conventional synthetic chemical compound disease modifying anti-rheumatic drugs (csDMARDs) and bDMARDs, and although these databases document comorbidity, they rarely report its severity. This too limits definitive statements on the progression or resolution of pre-existing disease. Although bDMARDs may not be contraindicated in patients with rheumatoid arthritis and certain comorbid diseases, additional assessments or precautions are recommended. This review summarizes current data on the use of bDMARDs in people with rheumatoid arthritis and common comorbid diseases. It provides an evidence base for doctors and their patients when discussing and selecting the appropriate bDMARDs.
Introduction
The global prevalence of rheumatoid arthritis is about 0.24%.1 The standardized mortality rate of patients with rheumatoid arthritis is estimated at 1.5 times that of the general population, although the incident mortality rate has decreased by 2.3% per year over the past four decades.2 3 The major causes of death include cardiovascular disease, cancer, and infections.4 Furthermore, the prevalence of comorbid conditions in people with rheumatoid arthritis can be as high as 60%.5
Although the decline in the incident mortality rate of rheumatoid arthritis cannot be attributed solely to biologic disease modifying anti-rheumatic drugs (bDMARDs), multiple studies confirm their efficacy in reducing disability and improving the quality of life in people with rheumatoid arthritis.6 But the design of randomized controlled trials (RCTs) required for the approval of bDMARDs restricts the recruitment of people with more severe comorbid diseases, raising questions about the efficacy and safety of these drugs in people with chronic diseases.
Similarly, the duration of clinical trials may be insufficient to assess the effect of these drugs on comorbid diseases. Strategies that channel younger healthier patients with rheumatoid arthritis to bDMARDs and those with chronic medical conditions to conventional synthetic chemical compound disease modifying anti-rheumatic drugs (csDMARDs) limit the interpretation of data from observational studies and registries. Such algorithms also discount the potential benefit of bDMARDs on chronic illnesses, such as the reduction of serum insulin and the insulin:glucose index, as well as the reduced rates of myocardial infarction in patients taking tumor necrosis factor α (TNF-α) inhibitors.7 8
Recommendations have focused on the indications for bDMARDs in rheumatoid arthritis but rarely discuss the management of rheumatoid arthritis in patients with concomitant comorbidities.9 10 In addition, patient input in treatment decisions is seldom discussed, even though shared decision making improves satisfaction with clinical care.11 This review summarizes current data on the use of bDMARDs in people with rheumatoid arthritis and common comorbid diseases. It will cover guidance on the appropriate selection of biologic agents in patients with specific chronic infections, cancer, cardiovascular disease, and other chronic conditions. Additional clinical or laboratory monitoring and modifications to the treatment of comorbid disease for patients with rheumatoid arthritis who are starting bDMARDs will be reviewed.
Sources and selection criteria
Our primary sources of data were PubMed and the Cochrane Database, which we searched from the initial use of bDMARDs for rheumatoid arthritis in 1997 through to August 2014. We looked at TNF-α inhibitors (etanercept, adalimumab, infliximab, certolizumab, and golimumab), abatacept, rituximab, and tocilizumab. The comorbid conditions included cardiovascular disease, cancer, infections, interstitial lung disease, chronic obstructive pulmonary disease (COPD), chronic kidney injury, diabetes mellitus, and pregnancy. The search terms included rheumatoid arthritis, and comorbid diseases. More than 2500 English language articles were identified. Of these, we reviewed all clinical trials, meta-analyses, prospective and retrospective case series, clinical registries, and case reports (n=500). We prioritized studies that reported the safety or complications of bDMARDs in patients with the above comorbid diseases, especially those with long term follow-up. Given the focus of the article, the most common primary sources were meta-analyses and reports from national registries. Case reports were used when information was not available from other studies. These articles were supplemented by a review of published societal and national guidelines for the treatment of rheumatoid arthritis with bDMARDs. The project was discussed with three patients with rheumatoid arthritis who agreed with the proposed comorbid diseases. They recommended that quality of life be a factor in the decision to prescribe bDMARDs.
Infection and vaccinations
The cellular targets of bDMARDs are integral to host defense against microbial pathogens, and themselves have been implicated as risk factors for severe infection. With the advent of TNF-α inhibitors, reactivation of latent tuberculosis and severe fungal infections emerged as serious threats. Patients with chronic viral infections such as HIV, hepatitis B virus (HBV), and hepatitis C virus (HCV) are also at risk of viral reactivation and disease progression.
Hepatitis B virus
TNF-α is raised in patients with chronic HBV and seems to play a role in clearing and controlling viral replication. Reactivation of HBV was initially described in hepatitis B surface antigen (HBsAg) positive patients receiving chemotherapy, TNF-α inhibitors, and rituximab, with subsequent reports in HBsAg negative patients who were positive for hepatitis B core antibody (HBcAb).
Infliximab, etanercept, and adalimumab have all been implicated in the reactivation of HBV.12 13 14 The introduction of routine screening for HBV before the start of biologic therapy and antiviral prophylaxis in high risk patients has reduced the number of reported cases, explaining the paucity of data with the newer TNF-α inhibitors and other biologic agents.
Rituximab, alone or as part of combination chemotherapy, reduces hepatitis B surface antibody (HBsAb) titers, and reactivation rates range from 3.4% to 80% in HBsAg positive patients and 1.5% to 23.8% in HBsAg negative, HBcAb positive patients, with occasional fatal results.15 16 17 18 19 20 In HBsAg positive patients, antiviral prophylaxis appears prudent. A retrospective study of eight HBsAg positive patients with rheumatoid arthritis who were treated with abatacept reported virus reactivation only in those not given HBV prophylaxis.21
HBV reactivation in a patient treated with abatacept who was HBsAg negative and HBcAb positive has also been reported.22 Data on HBV reactivation with tocilizumab are limited, but tocilizumab was used in two patients who received concomitant HBV prophylaxis, and in an HBsAg positive patient who did not receive antiviral prophylaxis, without HBV reactivation.23 24 25
A recent study identified 179 HBsAg negative, HBcAb positive patients exposed to various bDMARDs (including rituximab) without antecedent antiviral prophylaxis, and reported no cases of seroconversion over the two to five years of follow-up, highlighting the lower risk of HBV reactivation in these patients.26
Lamivudine was the most commonly used antiviral agent because of its tolerability and low cost, but concerns about viral resistance nearing 70% after five years of treatment have curtailed its widespread use.27 Other antiviral agents such as entecavir have been successful, with significantly lower rates of reactivation (6.3% v 39.3% for lamivudine; P<0.05).16 28
Recommendations (table 1⇓)
In the 2012 update, the American College of Rheumatology (ACR) advised against using bDMARDs in patients with chronic untreated HBV or treated HBV with liver dysfunction (Child-Pugh class B or higher) but made no recommendations for those with previous HBV infection only.29
Recommendations for the use of biologic disease modifying anti-rheumatic drugs (bDMARDs) in patients with rheumatoid arthritis and infection*
Earlier consensus guidelines on HBV prophylaxis were based mainly on oncology data.38 In 2015, the American Gastroenterological Association (AGA) Institute classified immunosuppressive therapy as high risk (>10% chance of HBV reactivation, notably with rituximab), moderate risk (1-10%, all other bDMARDs), or low risk (<1%, csDMARDs).30 39
Routine screening for HBsAg and HBcAb is recommended before the start of treatment with bDMARDs. In HBsAg positive patients, HBV viral loads should be documented and antiviral therapy started one week before treatment with a moderate or high risk bDMARD begins. Antiviral treatment should then be continued for three to six months after bDMARDs are stopped.31 32 33 Entecavir or tenofovir, which are more potent and have lower rates of induced viral resistance, were recommended over lamivudine by the European Association for the Study of the Liver (EASL).38 In HBsAg negative and HBcAb positive patients, HBV viral loads should be measured, and if detectable, these patients should be treated similarly to HBsAg positive ones.
Ongoing debate continues over the use of antiviral prophylaxis in HBsAg negative, HBcAb positive patients without detectable HBV DNA. Currently, EASL and the AGA Institute both recommend that patients being treated with rituximab, a high risk bDMARD, should be prescribed prophylactic therapy, although the recommendation was weak.30 38 They also recommended this approach, although less strongly and with a moderate level of evidence, for treatment with moderate risk bDMARDs. Monitoring for reactivation was suggested as a reasonable alternative to prophylactic therapy in patients receiving moderate risk bDMARDs, and one that should be discussed with the patient.40 Serial monitoring should include hepatic function, HBsAg seroconversion, and HBV viral loads.33 38
Hepatitis C virus
TNF-α is raised in chronic HCV infection and plays a role in antiviral resistance. In a small RCT of patients with HCV who had not previously received antiviral therapy, etanercept significantly improved the response to interferon alfa 2b and ribavirin—HCV RNA was eradicated in 63% of patients taking etanercept versus 32% of those taking placebo (P=0.04).41 Several case reports document successful and uneventful use of TNF-α inhibitors in HCV positive patients but rituximab based chemotherapy has been associated with an increased risk of HCV reactivation.31 42 43 44
B cell depletion may lead to decreased levels of IgG and viral replication or dissemination of viral particles from destroyed B cells. After the use of rituximab, cell mediated hepatocellular necrosis may occur with immune reconstitution.44 45 46 However, rituximab is successful in treating HCV cryoglobulinemic vasculitis and is considered safe in patients with rheumatic disease and HCV, with only a single report of a hepatitis flare in a patient with rheumatoid arthritis.47 48 49
Data on abatacept and chronic HCV in rheumatoid arthritis are limited but mild fluctuations in HCV viral loads have been reported in two patients.50 Information is also lacking, with few reports on the use of tocilizumab in patients with HCV.51 52 A single report of increased HCV viral load and transaminitis caused tocilizumab to be discontinued.53 However, given the potential protective role of interleukin 6 in hepatocyte regeneration and proliferation and raised levels in chronic HCV, cirrhosis, and hepatocellular cancer, more long term data on tocilizumab are needed.52 54
Recommendations (table 1)
Patients with rheumatoid arthritis should be screened for chronic HCV before starting bDMARDs. On the basis of the evidence presented above, TNF-α inhibitors remain a viable treatment option, with the data favoring etanercept. A hepatologist should be involved because treatment requires monitoring, and a baseline liver biopsy and close surveillance of liver enzymes and HCV viral loads may be needed. The ACR recommends etanercept in patients with chronic HCV.29 Biologic agents should be avoided in patients with acute or chronic HCV who have severe liver injury (Child-Pugh classes B or C).29
Hepatitis E
Hepatitis E virus (HEV) infection usually occurs as a self limiting, acute hepatitis, but rarely chronic HEV infection occurs and may result in fulminant hepatitis and rapidly progressive liver cirrhosis. It has been described almost exclusively in patients who are immunocompromised as a result of organ transplantation, hematologic malignancy, or HIV infection.55 In two reports from France, HEV infection was identified in patients treated with bDMARDs.56 57
Recommendations
No current guidelines exist for screening for HEV before the initiation of bDMARDs, but rheumatologists should be aware of this possibility in patients with unexplained raised transaminases.
Tuberculosis
Patients with rheumatoid arthritis have an increased risk of reactivation of tuberculosis, with one study reporting a relative risk of 2 compared with patients without rheumatoid arthritis.35 The risk seems to be highest with TNF-α inhibitors, with reported rates of 118 per 100 000 person years, four times that seen in patients with rheumatoid arthritis not treated with these agents.35
The most common manifestation is pulmonary tuberculosis, but extrapulmonary tuberculosis has also been reported, and this should be considered when evaluating patients.35 58
TNF-α inhibitors are also associated with increased rates of non-tuberculous mycobacterial infections.59 The risk of reactivating tuberculosis seems to be greatest with infliximab (145/100 000 person years) and lowest with etanercept (80/100 000 person years).35 60 Reports of tuberculosis reactivation with rituximab and abatacept are rare, and rituximab is considered safer.61 62 63 64 A global review of tocilizumab clinical trials found only eight reported cases of reactivation of tuberculosis.65
Recommendations (table 1)
The ACR recommends screening for latent tuberculosis before starting treatment with any bDMARD.29 The European League Against Rheumatism (EULAR) consensus statement notes insufficient data to recommend mandatory screening for tuberculosis before treatment with rituximab but advises against using rituximab in the presence of serious opportunistic infections.66 However, it recommends rituximab as the first line bDMARD if tuberculosis chemoprophylaxis is contraindicated and for patients from endemic regions.9 Patients with rheumatoid arthritis should be screened for latent tuberculosis infection according to regional guidelines before starting a bDMARD. This may be done with either the tuberculin skin test (TST) or interferon γ release assay (IGRA), with IGRA being recommended in patients who have been given BCG vaccination. Indeterminate IGRA tests should be followed by a TST.
Expert opinion is divided on the superiority of either test and both are reported to perform similarly in immunocompromised people.67 Regional guidelines should be followed regarding the use and interpretation of these tests. Patients with positive results (TST or IGRA) should have a chest radiograph to exclude active tuberculosis and, if negative, begin treatment for latent tuberculosis infection one month before the start of bDMARD therapy.29 The value of serial tuberculosis screening thereafter, in the absence of known exposure or high risk, is unclear.67 All TNF-α inhibitors have been associated with reactivation of tuberculosis, and rituximab is considered the safest.
HIV
Immunosuppressive therapy in immunocompromised patients further increases the risk of infection. TNF-α has been implicated in the pathogenesis and progression of HIV infection, and, theoretically, TNF-α inhibitors should have little effect on the progression of HIV disease. Clinically, TNF-α inhibitors have been used safely in patients with HIV who have stable CD4 counts above 200×106/L, with a single report of recurrent polymicrobial infections in those with lower counts.36 68 In patients with HIV associated lymphoma, rituximab does not seem to increase the risk of opportunistic infections.37 Data on the use of other bDMARDs in patients with HIV are limited.
Recommendations (table 1)
We recommend that patients with rheumatoid arthritis should be screened for HIV before they start treatment with bDMARDs. On the basis of the evidence presented above, TNF-α inhibitors seem to be safe in patients with HIV who have a stable CD4 count of 200×106/L or more. The long term safety of TNF-α inhibitors in patients with low CD4 counts, high viral loads, and those on concurrent immunosuppressive therapy is unknown. Data on other bDMARDs are limited and decisions regarding their use should be made together with the patient and infectious disease specialists. Patients should have regular monitoring of CD4 counts and HIV viral loads while taking bDMARDs.
Vaccinations
Inactivated vaccines can be safely administered during bDMARD therapy but may result in an attenuated immune response. TNF-α inhibitors may not affect humoral immune responses to pneumococcal and influenza vaccines, but patients on rituximab had lower rates of seroconversion with both.69 Rituximab did not affect antibody levels resulting from previous influenza vaccinations, indicating that vaccination should occur, ideally four weeks before the next dose, as per EULAR recommendations.70 71 72 Tetanus toxoid vaccination was shown to be effective when given 24 weeks after exposure to rituximab. No data are available on its efficacy within 24 weeks, so passive immunization is advised in patients with a serious risk of contracting tetanus during this period.72 73 Abatacept and tocilizumab do not significantly impair responses to tetanus or pneumonia vaccinations.74 75 76 77
Recommendations
We recommend that patients with rheumatoid arthritis should be vaccinated according to national guidelines and ideally receive all indicated vaccinations before bDMARDs are started. Physicians should be aware that the effectiveness of vaccinations may be decreased in people taking bDMARDs, especially influenza and pneumococcal vaccinations in people taking rituximab. Live vaccines (measles-mumps-rubella, varicella zoster, yellow fever, and oral poliomyelitis) are contraindicated in people taking bDMARDs and should be given at least four weeks before the start of treatment.78
Cardiovascular disease
The inflammatory nature of rheumatoid arthritis increases the risk of, and mortality from, cardiovascular disease (atherosclerosis, cerebrovascular disease, coronary artery disease, heart failure, and myocardial infarction) beyond that attributable to traditional risk factors. However, the use of bDMARDs in patients with rheumatoid arthritis and pre-existing cardiovascular disease, in particular heart failure, remains controversial. The difficulty for rheumatologists is to provide a balance between control of rheumatoid arthritis disease activity and resultant reduced risk of cardiovascular disease against the increased risk of de novo heart failure or its exacerbation.
TNF-α inhibitors and csDMARDs
Clinical registry data from Germany found no significant risk for new onset (adjusted hazard ratio 1.66, 0.67 to 4.10; P=0.28) or exacerbation of heart failure (1.18, 0.30 to 4.73; P=0.81) with TNF-α inhibitors versus csDMARDs. However, these analyses were limited by ill defined outcome measures, high dropout rates, and small numbers of cardiovascular events.79 80
By contrast, a RCT designed to evaluate the efficacy of infliximab in treating heart failure enrolled patients with New York Heart Association (NYHA) classes III-IV heart failure (ATTACH). The 150 patients recruited received placebo, infliximab 5 mg/kg, or infliximab 10 mg/kg at weeks 0, 2, and 6 intravenously. The primary outcome was the predetermined clinical condition at week 14, and secondary outcomes included change in ejection fraction and clinical condition at week 28. Clinical status did not improve in either of the treatment groups at week 14, but at week 28 there was a trend to greater mortality in the 10 mg/kg infliximab group (13% v 5% for placebo).81 Similar results were noted for etanercept in classes II –IV heart failure.82 This led to these agents being contraindicated in patients with rheumatoid arthritis and NYHA classes III-IV heart failure, and a caution being recommended in patients with NYHA class II heart failure.83 2012 ACR guidelines do not recommend TNF-α inhibitors in patients with NYHA classes III-IV heart failure who have an ejection fraction of 50% or less.29 And neither are baseline screening echocardiograms recommended before starting TNF-α inhibitors in those without a history of heart failure.84
Prospective data from the British Society of Rheumatology Biologics Register (BSRBR) reported an incidence rate ratio of 1.44 (0.56 to 3.67) for initial myocardial infarction in patients with rheumatoid arthritis who had been treated with TNF-α inhibitors compared with those treated with csDMARDs. However, the incidence rate ratio for TNF-α inhibitor “responders” (response within six months) was 0.36 (0.19 to 0.69) compared with “non-responders,” indicating that the risk of myocardial infarction is lower in those with an early response to TNF-α inhibitors.7
A retrospective US study reported that the use of TNF-α inhibitors in patients with rheumatoid arthritis was associated with a 55% reduction in the risk of incident coronary artery disease events (defined as coronary revascularization procedure, myocardial infarction, or unstable angina) compared with csDMARD use (excluding methotrexate) (hazard ratio 0.45, 0.21 to 0.96). This was most pronounced in people without diabetes and those who were positive for rheumatoid factor.85 Apart from a single case report of acute coronary syndrome after infusion of infliximab, current data indicate no contraindication to TNF-α inhibitors in patients with rheumatoid arthritis and previous coronary artery disease or myocardial infarction.86
Non-TNF-α inhibitors
Data on non-TNF-α inhibitor bDMARDs in patients with cardiovascular disease are sparse. Tocilizumab is associated with moderate reversible increases in mean total cholesterol, high density lipoprotein, low density lipoprotein, and triglycerides, but currently this atherogenic lipid profile has not been reflected by increased cardiovascular disease events compared with patients with rheumatoid arthritis who are treated with other drugs.65 87
A systematic review published in 2015 of lipid profile changes in patients with chronic inflammatory arthritis treated with bDMARDs found that patients with rheumatoid arthritis who were treated with tocilizumab were more likely to have hypercholesterolemia (odds ratio 4.64, 2.71 to 7.95), increased high density lipoprotein values (2.25, 1.14 to 4.44), and low density lipoprotein values (4.80, 3.27 to 7.05). This effect was not seen in patients treated with TNF-α inhibitors (1.54, 0.90 to 2.66).88
Prescribing information recommends that serum lipids are measured within four to eight weeks of starting tocilizumab and at 24 week intervals, with management of abnormal values as recommended by current lipid lowering guidelines Adult Treatment Panel II (ATP III) of the National Cholesterol Education Program (ATP III NCEP guidelines). During the TOWARD study, 16 patients with raised serum lipid values started statins, which effectively lowered or stabilized cholesterol levels.89
A review of five phase III clinical trials found that increases in mean low density lipoprotein levels induced by tocilizumab were less pronounced for patients also treated with statins at baseline, and initiation of statin therapy after starting tocilizumab reduced low density lipoprotein to below baseline values.90
Few cases (<0.1%) of acute coronary syndrome, including acute myocardial infarction, have been reported in patients with lymphoma who were treated with rituximab, although 8% experienced cardiac arrhythmias.91 92 Cardiac events were also uncommon in patients with rheumatoid arthritis who were treated with rituximab in a pooled observed case analysis of safety data—myocardial infarction rates were similar between patients treated with rituximab or csDMARD.93 Myocardial infarction was the most common serious cardiac event, and patients had at least one traditional cardiovascular disease risk factor. Rituximab is contraindicated in patients who have NYHA class IV heart failure or uncontrolled coronary artery disease.94 Cardiovascular events are uncommon during abatacept therapy and the presence of stable cardiovascular disease does not preclude its use.95
Recommendations (table 2⇓)
Patients with rheumatoid arthritis who have no history of heart failure do not need baseline Doppler echocardiography before starting treatment with a TNF-α inhibitor.83 EULAR recommends that in patients with a higher risk of cardiovascular disease optimum management of all traditional cardiovascular disease risk factors is paramount in conjunction with control of rheumatoid arthritis.108 Patients with previous myocardial infarction but without heart failure can be given TNF-α inhibitors, with clinical surveillance of cardiac function. TNF-α inhibitors are contraindicated in patients with NHYA classes III-IV disease and caution is recommended in milder heart failure.83 The use of bDMARDs is not contraindicated in patients with previous cerebrovascular disease.
Recommendations for the use of biologic disease modifying anti-rheumatic drugs (bDMARDs) in patients with rheumatoid arthritis and specific comorbidities*
Cancer
The association between bDMARDs and cancer is complex, not unlike that between rheumatoid arthritis and cancer. An increase in the relative risk of non-Hodgkin’s lymphoma (2.4, 1.9 to 2.9), Hodgkin’s disease (3.4, 1.8 to 5.6), lung cancer (1.5, 1.3 to 1.7), and non-melanoma skin cancer (1.3, 1.1 to 1.4), but a decrease in the risk of breast cancer (0.8, 0.7 to 0.9) and colorectal cancer (0.8, 0.7 to 0.9), was reported for Danish patients with rheumatoid arthritis.109 Although confounding factors of smoking and non-steroidal anti-inflammatory drugs (NSAIDs) could account for the paradoxical rates of lung and colorectal cancer, respectively, greater rheumatoid arthritis disease activity is associated with the development of lymphoma.110
Biologic agents modulate the immune system and may affect the development or suppression of cancer. High doses of TNF-α induce tumor necrosis, although certain cancers produce TNF-α, which in turn stimulates malignant cells. CD8+ T cells and natural killer cells require TNF-α for their cytotoxic effects against immunogenic tumor cells.111 Similarly, rituximab is used in the management of B cell lymphoma and tocilizumab for Castleman’s disease.112 113
RCT data
RCTs of bDMARDs in rheumatoid arthritis exclude patients with cancer, yet significant differences in the rates of cancer are reported between patients taking these agents and those taking placebo. A meta-analysis of nine RCTs of infliximab and adalimumab in 3493 and 1512 patients, respectively, reported 29 cancers in the TNF-α inhibitor users versus only three in the placebo group (pooled odds ratio 3.3 (1.2 to 9.1).114
The association of cancer with a higher versus lower dosage of TNF-α supports this link and suggests a dose response. Most cancers have latency periods measured in years, thus a cancer diagnosed during the course of a six month clinical trial could have been present in a preclinical state at study entry, with the TNF-α inhibitor accelerating its growth.115 116 However, these results were not replicated in a meta-analysis of 63 RCTs with more than 29 000 patients, which included trials of anakinra, rituximab, tocilizumab, and TNF-α inhibitors.117 It found no increased risk of cancer when these drugs were used either as bDMARD monotherapy or in combination with methotrexate. In fact, anakinra plus methotrexate reduced the risk of cancer versus methotrexate alone (odds ratio 0.11, 0.03 to 0.45).
A more recent meta-analysis of 44 clinical trials in which 11 700 patients with rheumatoid arthritis received a TNF-α inhibitor and 5901 received placebo or a csDMARD also found no increased risk of cancer in those taking a TNF-α inhibitor (odds ratio 1.29, 0.85 to 1.97), whether as monotherapy or combined with a csDMARD.118 In a long term single arm extension study of tocilizumab in 4527 patients with three years of follow-up, 117 patients developed cancer (standardized incidence ratio of 0.79, 0.66 to 0.95).119
Observational data
Observational studies or national registries seem to be a more appropriate source for assessing cancer risk with long term use of bDMARDs. A US observational study of 13 001 patients with rheumatoid arthritis and about 49 000 patient years of follow-up compared cancer rates with data from the US National Cancer Institute Surveillance, Epidemiology, and End-Results (SEER) database.120 bDMARDs were associated with a significantly increased risk of non-melanoma skin cancer (odds ratio 1.5, 1.2 to 1.8) and possibly melanoma (2.3, 0.9 to 5.4), but not solid tumors. The increased risk was independent of age, sex, education level, smoking history, and use of prednisone, and, after controlling for severity of rheumatoid arthritis, was not associated with lymphoma.
The BSRBR also reported a low risk (hazard ratio 0.83, 0.64 to 1.07) of solid tumors.121 Similarly, a Taiwanese study conducted over a 14 year period reported a reduced risk of new cancer diagnoses in 4426 new TNF-α inhibitor users compared with 17 704 csDMARD users (adjusted hazard ratio 0.63, 0.49 to 0.80).122 However, as in RCTs and systematic reviews, these reports excluded patients with previous cancer, which limits any definitive statements on the risk of recurrent cancer with the use of bDMARDs.
Two European clinical registries have looked at the risk of cancer recurrence in users of bDMARDs. The German biologics register (RABBIT) followed 122 patients with 124 previous malignancies (six lymphomas and 118 solid tumors) and compared them with 4998 patients without cancer.98 Over at least one year of follow-up, the incidence rate ratio for patients taking a TNF-α inhibitor versus a csDMARD was 1.4 (0.5 to 5.5; P=0.63).
A study from the BSRBR reported on the risk of incident cancer in 117 patients with rheumatoid arthritis treated with TNF-α inhibitors compared with 117 treated with csDMARDs. The incidence rates of cancer were 25.3 events per 1000 person years versus 38.3 events per 1000 person years for TNF-α inhibitors versus csDMARDs, respectively. The age and sex adjusted incidence ratio of TNF-α inhibitors versus csDMARDs was 0.58 (0.23 to 1.43).99 The authors suggest that the lower rate with TNF-α inhibitors is due to selection bias—clinicians may be more inclined to us TNF-α inhibitors in patients with longer remission periods (“low grade” tumors), whereas those with a greater chance of recurrence were more likely to receive csDMARDs.
Data from the Swedish biologics registry (ARTIS) were examined to assess the risk of breast cancer recurrence in women with rheumatoid arthritis; 120 patients treated with a TNF-α inhibitor were compared with 120 matched patients who had not been treated with a biologic agent. Recurrence rates were similar (15/1000 v 16/1000 patient years), with an adjusted hazard ratio of 1.1 (0.4 to 2.8), suggesting no increased risk.100 However, cases of rapid progression of squamous cell skin cancer with abatacept and TNF-α inhibitors have raised concerns about a trophic effect.123 124
Caveats
Although these data provide useful information, they should be interpreted with caution. Physicians select patients for treatment with bDMARDs on the basis of their cancer type and history. For example, in the RABBIT registry patients with breast cancer were less likely than those with prostate cancer to receive bDMARDs. However, specific data on cancer type is often not reported.
The interval between the cancer and initiation of bDMARD treatment is often prolonged, as in RABBIT (five years) and BSRBR (11.5 years), which limits the applicability of the results to patients with a more recent diagnosis of cancer.
Recommendations (table 2)
Current treatment guidelines for bDMARDs in patients with rheumatoid arthritis and cancer are based on evidenced based data and expert opinion. The Canadian Rheumatology Association (CRA), through consensus opinion, discourages the use of bDMARDs in patients with active cancer that may require surgery, chemotherapy, or radiation.125
The ACR and CRA both recommend using bDMARDs with caution in patients with a remote history of cancer but avoiding these drugs in those with current cancer.29 125 Both societies also recommend considering rituximab in patients with a history of solid tumors, lymphoma, non-melanoma skin cancer, or melanoma within five years of remission.
We suggest that patients with previous cancer should complete cancer specific surveillance with the decision to start treatment made in consultation with an oncologist.
Interstitial lung disease
The prevalence of rheumatoid arthritis associated ILD ranges from 10% to 50% depending on the population studied and criteria for diagnosis. It is associated with older age at symptom onset, male sex, severity of rheumatoid arthritis, and csDMARD use, specifically methotrexate and leflunomide.126 127 128 129
The interactions between bDMARD and innate or secondary rheumatoid arthritis associated ILD are complex, and emerging data underscore the equivocal associations, the need for vigilance, and the lack of specific biomarkers.130 TNF-α inhibitors, tocilizumab, and rituximab have been associated with the development of rheumatoid arthritis associated ILD yet these same bDMARDs are used for its treatment.131 132 133 134 135
A study of more than 8000 patients with rheumatoid arthritis from a US cohort that was followed for 10 years reported no significant difference between TNF-α inhibitors and csDMARDs with regard to the development of rheumatoid arthritis associated ILD (adjusted hazard ratio 1.03, 0.51 to 2.07).136 A recent systematic review concluded that methotrexate, leflunomide, TNF-α inhibitors, rituximab, and tocilizumab may all induce pneumonitis or worsen pre-existing rheumatoid arthritis associated ILD.137
BSRBR data identified 367 patients with pre-existing rheumatoid arthritis associated ILD, 29 of whom were treated with a TNF-α inhibitor and 68 with a csDMARD.138 Smoking was not a determinant, and after controlling for age, sex, disease duration, severity of rheumatoid arthritis, and steroid and methotrexate use, the adjusted mortality rate ratio for TNF-α inhibitors versus csDMARDs was 0.81 (0.38 to 1.73). However, rheumatoid arthritis associated ILD was listed as the cause of death in 21% of TNF-α inhibitor users—three times the rate of that seen in those taking csDMARDs.
Recommendations (table 2)
Because of the lack of data from controlled trials in rheumatoid arthritis associated ILD, no recommendations can be made on the use of bDMARDs or csDMARDs, and a proposed approach is based on assumptions that extrapolate data from idiopathic pulmonary fibrosis and systemic sclerosis associated ILD.101 Screening pulmonary function tests may be performed, and high resolution computed tomography is recommended in patients who have a carbon monoxide pulmonary diffusing capacity (also known as carbon monoxide transfer factor lung diffusion capacity) below 70% and pulmonary symptoms. The decision to start a bDMARD should be made on an individual basis and potential complications limited by completing appropriate immunizations and consultation with a pulmonologist. Patients with rheumatoid arthritis and pre-existing ILD should be asked to report the onset of any respiratory symptoms. Monitoring of pulmonary function tests with carbon monoxide pulmonary diffusing capacity testing every three to six months should be considered in patients without symptoms.101
Chronic obstructive pulmonary disease
Rheumatoid arthritis and COPD are both common, tobacco use is a pathogenic trigger in both conditions, and they can coexist. TNF-α is involved in the pathogenesis of both asthma and COPD, but clinical trials of TNF-α inhibitors in both diseases have had mixed results.139
An observational study examined 15 771 patients with rheumatoid arthritis and COPD from a North American insurance claims database, 1205 of whom were admitted to hospital for COPD and 12 050 of whom served as controls.102 The adjusted relative risk for hospital admission for patients who had used etanercept within the past year was 0.47 (0.26 to 0.85), but no benefit was seen for infliximab (1.14, 0.66 to 1.99). By contrast, in a pivotal rheumatoid arthritis trial of abatacept in which 4% of patients had COPD, 37 received abatacept and 17 placebo. Respiratory complications including cough, rhonchi, dyspnea, and exacerbation of COPD were reported in 43.2% of patients receiving abatacept versus 23.5% on placebo, indicating the need for caution in patients with a history of COPD, and its avoidance in those with uncontrolled COPD.103
Recommendations (table 2)
We suggest that TNF-α inhibitors can be used in patients with rheumatoid arthritis and COPD but physicians should be vigilant for the development of lung infections. Abatacept should be avoided in patients with symptomatic COPD.103
Diabetes
Rheumatoid arthritis is associated with impaired glucose tolerance, which correlates with the degree of inflammation.140 Levels of TNF-α are increased in chronic hyperglycemia, and this decreases insulin sensitivity, impairs wound healing, and may contribute to other complications of diabetes. Type 1 diabetes is characterized by immune mediated destruction of pancreatic β cells.141
Theoretically, the inhibition of TNF-α should reduce insulin resistance by preventing islet cell destruction. Infliximab is reported to improve insulin resistance in patients with rheumatoid arthritis and decrease insulin requirements in those with type 1 diabetes.8 Treatment with rituximab and abatacept delayed the loss of C peptide responses, indicating the promotion of pancreatic β cell survival.142 143 Similar to rheumatoid arthritis, diabetes is associated with an increased risk of cardiovascular disease, and the presence of both diseases could exponentially increase atherosclerosis.
A placebo controlled safety trial of etanercept in patients with rheumatoid arthritis that included 265 patients with diabetes who were followed for a mean of 3.5 years reported no increased risk of infections.104 Tocilizumab was reported to improve glycated hemoglobin in 10 patients with rheumatoid arthritis and type 2 diabetes within one month of treatment.105 Of interest, none of the 29 patients without diabetes reported hypoglycemia, suggesting that the glucose lowering effect was unlikely to be mediated by insulin secretion but was possibly secondary to control of inflammation and decreased insulin resistance. However, at least one case report describes a patient with type 1 diabetes who developed severe hypoglycemia within 12 hours of administration of adalimumab.106 Other bDMARDs seem to be well tolerated but supporting data are sparse.
Recommendations
bDMARDs seem to be well tolerated in patients with diabetes. However, we recommend that patients are made aware of the increased risk of infections with bDMARDs and those with brittle diabetes are monitored for symptoms of hypoglycemia when starting a TNF-α inhibitor.
Chronic kidney injury
Methotrexate is contraindicated in chronic kidney injury, making bDMARDs a potential option if proved to be safe and efficacious. However, many RCTs comparing methotrexate with bDMARDs exclude patients with chronic kidney injury. One open label study of etanercept in five patients on hemodialysis—two with rheumatoid arthritis the others with spondyloarthritis—followed participants for 18 months. Both patients with rheumatoid arthritis achieved disease activity scores less than 3.2 (low disease activity) without complications.144 We found no reports on the use of tocilizumab, abatacept, or rituximab in patients with rheumatoid arthritis and chronic kidney injury.
Recommendations
bDMARDs may be considered for use in patients with rheumatoid arthritis and chronic kidney injury but we recommend discussion with patients and their nephrologist.
Pregnancy
Rheumatoid arthritis primarily affects women of child bearing age, so the use of bDMARDs before conception, during pregnancy, in the postpartum period, and during breast feeding is an important consideration.
All TNF-α inhibitors are Federal Drug Administration Pregnancy Category “B”—animal reproduction studies have failed to demonstrate a risk to the fetus and there are no adequate and well controlled studies in pregnant women. However, abatacept, rituximab, and tocilizumab are FDA Pregnancy Category “C”—animal reproduction studies have shown an adverse effect on the fetus and there are no adequate and well controlled studies in humans, but potential benefits may warrant use of the drug in pregnant women despite potential risks.145
A prospective study reports marked variation of rheumatoid arthritis disease activity during pregnancy, and challenges earlier reports of remission during pregnancy and a risk of exacerbation after birth.146 Given that csDMARDs such as leflunomide and methotrexate are contraindicated during pregnancy, bDMARDs are therapeutic options along with hydroxychloroquine and sulfasalazine at conception and during pregnancy.
When surveyed, practicing rheumatologists expressed uncertainty about the risks from etanercept or infliximab during pregnancy, given the inherent increased rate of adverse pregnancy outcomes (low birth weight, increased rate of spontaneous abortion) in women with rheumatoid arthritis independent of DMARD therapy.147 148
BSRBR data evaluating outcomes from 130 pregnancies in women with rheumatoid arthritis who were receiving TNF-α inhibitors before or at the time of conception were compared with 10 who were on csDMARDs alone. Women exposed to TNF-α inhibitors had a 27% rate of spontaneous abortion (higher for those on concomitant methotrexate or leflunomide) compared with 10% for women on csDMARDs; however, these differences were not significant. Four congenital malformations were reported for women who had been treated with TNF-α inhibitors compared with none for the women who were on csDMARDs alone.107 There were no cases of VACTERL (vertebral anomalies, anal atresia, cardiac defects, tracheoesophageal fistula and/or esophageal atresia, renal and radial anomalies, limb defects) syndrome. A previous review of the FDA database for adverse events related to TNF-α inhibitors reported one patient with VACTERL syndrome after in utero exposure to etanercept; however, 19 other congenital anomalies consistent with the VACTERL spectrum were also identified, with the frequency of congenital malformations in this population being significantly greater than previously reported for the general population (P<0.01).149
Systematic reviews published in 2014 support the absence of risks when TNF-α inhibitors are used during conception or pregnancy.145 150 However, among TNF-α inhibitors, certolizumab was shown to have the least placental transfer when compared with infliximab and adalimumab.151 Median cord blood values at birth were 3.9% that of the mother for certolizumab compared with 160% and 153% for infliximab and adalimumab, respectively.
For rituximab, current guidelines primarily based on data from patients without rheumatoid arthritis recommend effective methods of contraception and avoiding pregnancy for 12 months after exposure. This is because the effects of in utero and neonatal B cell depletion are unknown and congenital malformations, miscarriages, and prematurity have been reported.150
For men receiving a TNF-α inhibitor at the time of conception, no adverse pregnancy related outcomes, either in the mother or infant, have been reported.150 However, in women, it might be reasonable to wait six months or more between the last infusion of rituximab and attempts to initiate pregnancy.94
Currently, the number of publications on other bDMARDs in relation to pregnancy is limited.
Breast feeding
Rituximab has been found in monkey breast milk, and although this has not been confirmed in human breast milk, breast feeding is not advisable.94 Levels of TNF-α inhibitors are lower in breast milk, with values ranging from undetectable to a twentieth of that seen in the maternal circulation, and infant serum drug levels are low or undetectable.150 Case reports indicate no adverse outcomes in breastfed infants of mothers who continue to use TNF-α inhibitors.152
Recommendations (table 2)
Because of the lack of evidence based data, there are no guidelines on the use of bDMARDs in patients with rheumatoid arthritis who are planning to conceive, who are pregnant, or who wish to breast feed, but observational data suggest that TNF-α inhibitors are the least risky option. We recommend an informed discussion between the rheumatologist and women of child bearing age if bDMARDs are considered. If TNF-α inhibitors are continued during gestation, recommendations on safe food consumption (such as avoiding processed meats and unpasteurized milk and cheeses), although applicable to all pregnant women, should be emphasized given the increased risk of intracellular infections such as Listeria monocytogenes. Live vaccinations should be delayed for at least six months after birth in infants who have been exposed to a TNF-α inhibitor in utero.150
What matters to patients
Given the risk of exacerbations of comorbid disease from bDMARDs, it is imperative that the potential complications are fully disclosed to patients. Patients will want data on the impact of bDMARDs on their quality of life in the presence of comorbid illness because this is a major determinant of their willingness to accept the risk of treatment. In our search we found only one article that examined quality of life for patients with rheumatoid arthritis and comorbidities who were taking bDMARDs. Data from the German biologics registry, RABBIT, noted that an increasing number of comorbidities were associated with the failure of single or multiple bDMARDs to be effective.153 Furthermore, the proportion of patients not achieving a minimal detectable improvement in their health related quality of life was related to the number of comorbidities. Zero, one, and two or more comorbidities were associated with 61.9%, 65.8%, 67.5% of patients, respectively, not achieving a minimal detectable improvement. This paucity of data on quality of life makes treatment decisions harder for patients and physicians. Physicians need to assess different components of the patient’s quality of life—physical, mental, and social aspects—and determine whether deficits are secondary to rheumatoid arthritis or comorbid illness. Poor quality of life, unrelated to rheumatoid arthritis, is unlikely to improve with bDMARD therapy, so disease and patient specific factors need to be considered when deciding whether bDMARDs are appropriate.
Conclusion
Although bDMARDs significantly improve disease outcomes in rheumatoid arthritis, their novel mechanism of action presents challenges for patients with comorbid disease. The safety and efficacy of bDMARDs in patients with chronic medical conditions have not yet been confirmed by RCTs, and observational studies often present conflicting results given the differences in design and outcome measures recorded. Chronic disease registries provide the best available evidence for the long term safety of bDMARDs in rheumatoid arthritis. However, it is not enough merely to record the presence or absence of a comorbid disease, and comprehensive data on the onset, severity, and associated complications of these diseases are needed. Analyses of these data will help to ascertain the potential benefit of bDMARDs on comorbid illnesses, such as improved glycemic control in diabetes.
Given the risk of exacerbations of comorbid disease from bDMARDs, it is imperative that the potential complications are fully disclosed to patients with rheumatoid arthritis. Patients will request data about the impact of bDMARD on their quality of life in the presence of comorbid illness because this is a major determinant of their willingness to accept the risk of treatment. In our search we found one article that looked at quality of life for patients with rheumatoid arthritis and comorbid illness who were taking bDMARDs. Both disease and patient factors need to be considered in selecting the appropriate bDMARD. Data on the exacerbation and severity of comorbid conditions in trials and registries of rheumatoid arthritis need to be collected so that physicians and patients can accurately assess risks and benefits in the presence of comorbid disease.
Questions for future research
How sensitive are the tuberculin skin test and interferon gamma release assay in detecting active tuberculosis in patients with rheumatoid arthritis who take biologic disease modifying anti-rheumatic drugs (bDMARDs)?
Do bDMARDs change the incidence of hepatocellular carcinoma in patients with rheumatoid arthritis and chronic viral hepatitis?
Do bDMARDs improve glycemic control in patients with rheumatoid arthritis and diabetes?
Do bDMARDs, specifically tumor necrosis factor α inhibitors, cause or worsen interstitial lung disease, and is so by what mechanism(s)?
Are patients with rheumatoid arthritis who have had cancer within the past five years more likely to experience a recurrence on bDMARDs or conventional synthetic chemical compound DMARDs?
Do patients with rheumatoid arthritis with or without comorbid disease experience a similar improvement in quality of life?
How patients were involved in creation of this article
Patients would like to know more on the effect of biologic disease modifying anti-rheumatic drugs on their quality of life in the setting of comorbid illness. Patients were therefore asked to review the manuscript and comment on the following:
Which comorbidities are most important? The patients agreed with the comorbidities selected
Given the information, would you still agree to recommendations regarding the use of biologic disease modifying anti-rheumatic drugs (bDMARDs)? These recommendations are:
-There is concern about the use of bDMARDs in interstitial lung disease
-Etanercept seems to be a good option for patients with hepatitis C
-Temporary exacerbations of diabetes or chronic obstructive pulmonary disease that are treatable and reversible may be acceptable
-Insufficient data are available on the use of bDMARDs in cardiovascular disease and pregnancy to make informed decisions
Notes
Cite this as: BMJ 2015;351:h3658
Footnotes
Thanks to Jan Edmiston for her assistance with database searches and to the patients at the Washington DC Veterans Affairs Medical Center for their review of the manuscript and comments.
Contributors: All authors helped in the design of the work; the acquisition, analysis, and interpretation of the data; and the revision of the manuscript. All authors approved the version to be published and are accountable for all aspects of the article. JSR is guarantor.
Competing interests: We have read and understood BMJ policy on declaration of interests and declare the following interests: none.
Provenance and peer review: Commissioned; externally peer reviewed.
The views expressed in this article are those of the authors and do not necessarily reflect the positions or policies of the Department of Veterans Affairs or the Federal Government of the United States.