The case of methotrexate and the lung: Dr Jekyll and Mr Hyde

Involvement of the respiratory system, the commonest extra-articular manifestation of rheumatoid arthritis (RA), occurs to some extent in a large proportion of patients with RA. Any of the pulmonary compartments can be affected, either secondary to RA itself, or in relation to drug toxicity or opportunistic infection [1]. RA-associated interstitial lung disease (RA-ILD) is an increasingly recognised complication of RA, in many ways similar to idiopathic pulmonary fibrosis [2]. Clinically relevant RA-ILD may be present in up to 10% of patients with RA, and is associated with significant morbidity and mortality [1, 3], with no optimal treatment determined.

Drugs used to treat RA may cause various respiratory complications, including obliterative bronchiolitis (D-penicillamine, tiopronine and gold salts) [4–6], pulmonary oedema [1], lupus (sulfasalazine and tumour necrosis factor (TNF)-α inhibitors) [1], and ILD. In fact, virtually all drugs used to treat RA can cause ILD. Some can cause subacute onset, inflammatory hypersensitivity pneumonitis, including but not limited to anakinra, azathioprine, cyclophosphamide, gold salts, leflunomide, methotrexate (MTX), rituximab, sulfasalazine, TNF inhibitors and tocilizumab [1] (www.pneumotox.com). Others have been associated with the development or progression of pulmonary fibrosis, including azathioprine, cyclophosphamide, gold salts, MTX and sulfasalazine. In this field, the treatment armamentarium has grown considerably over the past years, and one must remain vigilant to the possibility of drug-induced ILD caused by recently marketed drugs, especially the so-called biologics. Among them, anti-TNF monoclonal antibodies (etanercept, infliximab, adalimumab, certolizumab, golimumab), chimeric anti-CD20 antibody rituximab, anti-CTLA4-Ig abatacept, and anti-interleukin-6 antibody tocilizumab are frequently considered to trigger the onset of pneumonitis or acute exacerbations of pre-existing ILD.

However, MTX, a drug marketed for the treatment of RA in the 1970s, remains the preferred first-line disease modifying anti-rheumatic drug (DMARD) in RA [7]. Yet, MTX use in this setting continues to feed considerable debate about its possible “toxicity” to the lung, and patients are often referred to pulmonologists to enquire about possible pulmonary contra-indication to the drug. Indeed, many lung “toxicities” have been associated with MTX treatment [8], ranging from mild cough to lethal acute pneumonitis [9].

The best-characterised complication of the drug is MTX-induced subacute hypersensitivity pneumonitis (table 1), which has an estimated incidence of about 1% [10, 11] (0.3% in a large randomised trial [12], yet with a shorter follow-up as compared to clinical practice). The typical presentation consists of dry cough, dyspnoea, and fever, with a subacute disease onset over a period of several weeks, later progressing to more severe disease. Mild pleural effusion and pleuritic-type chest pain may also be present. The median time from MTX initiation to drug-induced hypersensitivity pneumonitis was 9 months in a large literature review [9]. Rechallenge is dangerous [9], as it is thought to be a hypersensitivity reaction, with upregulation of a number of pro-inflammatory cytokines [13]. Bronchoalveolar lavage is generally performed to look for alternative diagnoses, mainly infection, and may show an increased number of lymphocytes, neutrophils and/or eosinophils [14]. Imaging typically shows diffuse lung infiltrates (mostly ground glass attenuation) (figure 1), combined with pre-existing abnormalities, if present. Changes on lung imaging and pulmonary function test are best interpreted when baseline data are available prior to initiating MTX. As with other drugs, the diagnosis of MTX-induced hypersensitivity pneumonitis is one of exclusion and is based on the combination of the appropriate clinical setting, clinical manifestations, radiographic features, bronchoalveolar lavage abnormalities, lung histology (when available), and response to drug discontinuation [15]. Central in the management is drug discontinuation, often assisted by short-term glucocorticoids.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Representative chest computed tomography (CT) images in a female patient with methotrexate-induced hypersensitivity pneumonitis (parenchymal window). a, b) CT obtained 2 days after the onset of symptoms (dry cough, shortness of breath, mild fever); methotrexate was continued. c, d) CT obtained 8 days later while in intensive care unit (ICU); methotrexate was discontinued. e, f) CT obtained 3 years after the acute episode.

Risk factors for the development of MTX-induced hypersensitivity pneumonitis include age greater than 60 years, female sex, pre-existing lung disease, previous use of DMARDs or anti-TNF medications, hypoalbuminaemia and diabetes, whereas the weekly or cumulative dose of MTX are not associated with an increased risk of toxicity [16]. Cardiovascular diseases have also been identified as a risk factor in one study [16], although it could be just a confounding factor [12, 17, 18]. As the development of MTX-induced pneumonitis cannot be predicted, and systematic serial lung function tests are unhelpful for detection of hypersensitivity reactions [14], experts often advocate for obtaining baseline chest imaging and pulmonary function tests [9, 12]. Other important measures include informing patients about the possible risk and associated symptoms, such as new onset of dyspnoea or cough. Rheumatologists should be aware of rales at lung auscultation, and consider discontinuation of MTX (or possibly lowering the weekly dose) [17] in case of suspected pulmonary manifestations prior to risking progression to a more severe condition.

In contrast to the well-established MTX-induced pneumonitis, the question of whether or not the chronic use of MTX may negatively affect the outcome of RA-ILD in the long-term, or even cause or worsen pulmonary fibrosis, has been a matter of debate for decades. A meta-analysis of 22 studies with 8584 participants found a significant increase in the risk of infections (relative risk 1.11, 95% CI 1.02–1.21) and acute pneumonitis (relative risk 7.81, 95% CI 1.76–34.72) in patients with RA treated with MTX compared with other DMARDs and biologics [19]. However, in this and in another meta-analysis [20], MTX-induced subacute pneumonitis was not clearly distinguished from pre-existing or new-onset chronic RA-ILD. In contrast, in a large, multicentre prospective study, MTX treatment was not associated with an increased risk of ILD [21]. Several other observational studies did not report an increased risk of RA patients to develop incident chronic ILD when treated with MTX [22, 23]. In one cohort study, patients with RA receiving MTX as part of their ILD treatment had better prognosis compared to those not receiving this drug [24]. These conflicting findings justify the physicians’ reluctance to use MTX in patients with RA-ILD.

In this issue of the European Respiratory Journal, Juge et al. [25] present a retrospective, case–control study of a population of RA patients, with (n=410) or without ILD (n=673). Unexpectedly, they found a negative association between MTX exposure and the presence of chronic RA-ILD. In other words, patients who were taking MTX for their RA were less likely to have ILD than those who were not! In addition, among patients with RA-ILD, ILD was detected later in those treated with MTX than in those who were not (11.4±10.4 years versus 4.0±7.4 years after the onset of RA, respectively; p<0.001) [25], consistent with findings from a previous study [21].

Strengths of the analysis include the relatively large population, the use of chest computed tomography (CT) to identify ILD, and the separation into distinct derivation and replication patient cohorts. The comparison was controlled for age at RA onset, sex, smoking history, duration of MTX exposure, and periods of MTX use at RA onset. The findings did not differ between important subgroups, including whether the ILD pattern on CT was usual interstitial pneumonia or not.

Despite the commendable effort, there were some limitations and pitfalls. Due to the retrospective design of the study, one cannot exclude that the prescription of MTX may have differed depending on pre-existing lung disease. Pulmonary function tests were not collected, and it is plausible that patients with impaired lung function were less likely to receive MTX, which indirectly creates a statistical association between MTX and absence of ILD. Selection of the population was not exhaustive: patients in whom ILD preceded the onset of RA and those who did not have a chest CT were not included. Age, sex, smoking history, cumulative dose of MTX, and missing data were imbalanced between groups. Therefore, some confounding factors could have affected the relationship between MTX use and RA-ILD.

As in any observational study, two important biases must be considered: immortal time bias and confounding by indication. Immortal time bias relates to the time-dependency of the exposure to the risk (MTX), which by design cannot be excluded in this analysis. Confounding by indication is also possible, as fewer patients were treated with MTX among those with RA and ILD (60%) than those with RA without ILD (83%), suggesting that presence of ILD did influence treatment choices. It is conceivable that the negative association between MTX use and RA-ILD might be related to physicians prescribing less frequently MTX to patients with ILD or respiratory symptoms.

Only an experimental trial design would provide a better and probably definite answer to this question, but randomising patients to using MTX versus placebo would raise serious ethical concerns. In observational studies, one could restrict the analysis to incident cases of ILD among patients with RA, receiving MTX or not. In this regard, one study showed MTX exposure not to be associated with incident RA-ILD; a non-significant trend was also found in favour of delayed ILD onset [21].

The straightforward interpretation of the study by Juge et al. [25] is that MTX might protect patients with RA against the risk of developing ILD due to their RA disease, which would be ideal given the widespread use of this drug to treat RA. If confirmed, MTX would not only reduce RA disease activity, morbidity and mortality [26], but would also partly prevent the development of ILD, a severe complication of RA. However, we should not forget that association does not equal causation, and this study cannot be interpreted as a clear demonstration that MTX protects against ILD in patients with RA. The association found between MTX use and decreased frequency or delayed onset of ILD might well be due to confounders and not to a protective role of MTX.

In conclusion, one must keep in mind that the relationship between MTX and the lung is two-fold. MTX can induce an unpredictable, subacute hypersensitivity pneumonitis, which is an inflammatory, reversible, yet potentially lethal condition (the Mr Hyde's side of MTX), which should lead to permanent discontinuation of the drug. In contrast, MTX use does not seem to be associated with an increased risk of chronic fibrotic ILD in RA, and perhaps even reduces it (the Dr Jekyll's side of MTX and the lung). More chapters of Dr Jekyll and Mr Hyde will need to be written!

Shareable PDF