Studies in RA and ILD patients treated with csDMARDs are scarce and contradictory, which has led to restriction of their use in this complication.

Methotrexate (MTX). Considered a prototype of drug-induced pulmonary toxicity, MTX has been avoided until a few years ago in the treatment of RA patients with ILDs. We now have evidence from two SRs11,16 and two cohort studies,18,20 which address the relationship of MTX with ILD onset, progression, and mortality. No association was found between MTX exposure and onset of ILD (OR: .85; (.49–1.49); P = .58) in a prospective study conducted by Kiely et al.18 in 2,692 patients with recent-onset RA (1,578 exposed to MTX and 1,114 not exposed). In the patients exposed to MTX, ILD onset was longer than in those not exposed (OR: .41; (.23–.75); P = .004). Conway et al.11 found in an MA of 22 studies with 8,584 RA patients, that MTX produced a small increase in the risk of total respiratory complications (RR: 1.10 (1.02–1.19), due to the increased risk of infections (RR: 1.11 (1.02–1.21). However, they found no increase in the risk of non-infectious respiratory complications (RR 1.02 (.6–1.60), or death from lung disease (RR 1.53 (.46−5.01). MTX produced an increased risk of acute pneumonitis (RR: 7.81 (1.76–34.72), but not of chronic interstitial pneumopathy (IP).

The risk of hospital admission due to ILD in RA patients treated with biologics decreased with MTX exposure (HR: .06 (80.06-.469); P < .001), irrespective of concomitant biologic therapy (anti-TNF vs. non-anti-TNF), in a retrospective study using data from US mutual health insurance companies.28 MTX was also not found to be associated with the onset or exacerbation of ILD in RA in a Greek study comparing the outcome of treatment of RA over 1 year with anti-TNF agents and csDMARDs in patients with ILD (42 and 44) and without ILD (40 and 44, respectively).21 In contrast, the multivariate analysis showed a negative correlation of MTX use with extent of ILD and degree of air entrapment.

The survival of patients with RA-ILD, adjusted for confounding variables, was higher with MTX (median 70 months) than without MTX (log Rank P < .0005), in a retrospective study by Rojas-Serrano et al.20 The 56 cases (25 with associated LEF) and the 26 controls (4 AZA, 9 LEF, 4 SSZ and 9 without DMARDs) had received the same GC regimen.

In an MA, Roubille et al.16 found lower mortality of patients with ILD in RA with MTX (13%) than with TNFi (35.5%).

Leflunomide (LEF). After it was marketed, cases of ILD in RA patients treated with LEF began to be published, especially in Japan. In 2016, Suissa et al.37 published an extensive case-control study nested in an RA cohort, in which they admit that the 2-fold increase in ILD associated with LEF could reveal a treatment channelling bias of patients with pre-existing ILD to LEF. Conway et al.12 also published an SR and an MA of 8 RCTs (only 1 in an Asian population) of LEF in RA. They included 4579 patients, of whom 2274 received LEF and 2305 a comparator (usually MTX) and found 6 cases of pneumonitis and 4 deaths from lung causes, all in the comparator group. LEF was not associated with an increased risk of total respiratory adverse effects (RR: .99 (.56−1.78), infectious (RR: 1.02 (.58−1.82) or non-infectious adverse effects (RR: .64 (.41−.97)), or an increased risk of death from lung disease (RR: 1.53 (.46−5.01). The abovementioned SR of cases of RA-associated ILD16 included 12 studies with 34 cases of ILD in RA treated with LEF. In this study, Roubille et al. estimated a fatal outcome in 18% of patients with LEF, compared to 13% of those with MTX.

The study by Rojas-Serrano et al.,20 which showed increased survival in RA-ILD of patients treated with MTX, observed that combination with LEF (25/58) did not worsen survival.

Glucocorticoids (GC) and Immunosuppressants (IS). Glucocorticoids (GC) and Immunosuppressants (IS). There is little evidence of the effect of IS on RA-associated ILD. They are sometimes used in the treatment of exacerbations of ILD with good response according to some studies, whose authors conclude that intensive therapy with GC and cyclophosphamide (CPM) could improve prognosis in severe cases, with low respiratory function.

The effect of oral GC on mortality in RA-ILD was evaluated in a case-control study,26 which found that patients who received oral prednisone for more than 3 months had a higher relative risk of death than those who did not (RR: 2.06 (1.1–3.8); P = .02), especially from respiratory causes (RR: 2.75 (1.6–4.7); P = .0002). The predominance of the UIP pattern in the GC treatment group may account for some of the excess mortality.

In recent years, mycophenolate mofetil (MMF) has replaced CPM and azathioprine (AZA) in the treatment of ILD associated with systemic autoimmune diseases (SADs). Fisher et al.24 retrospectively analysed the effects of MMF on the progression of ILD in 125 SAD patients, of whom 18 had RA. In general, MMF was well tolerated and allowed the dose of CG to be reduced. In RA-associated ILD, the previous trend of progressive deterioration changed to an improvement in FVC after starting MMF, although it did not reach statistical significance. MMF was also associated with an RR of death from any cause, and from respiratory causes, lower than AZA and prednisone, in a retrospective UK study25 of 290 RA-ILD patients followed up over more than 12 years. These results are consistent with the results of the PANTHER study38on idiopathic pulmonary fibrosis (IPF), which showed an increase in mortality associated with the use of prednisone and AZA, compared with placebo.

Regarding Tacrolimus, with no indication in our setting for treating patients with RA, we only found the anecdotal description of 10 cases of RA39 that developed lung lesions during treatment (with fatal outcome in 2), and therefore we cannot draw any conclusions.

We found 3 SRs13,16,17 that deal with the effect of anti-TNF therapies on patients with RA-associated ILD, but the methodology used calls for caution in interpreting the results.

The SR by Roubille et al.16 were based on published clinical cases. They found that TNFi are associated with a low incidence of ILD (1%), but high mortality (35.5%).

In a more recent SR of cohort, case-control and case report studies, Huang et al.13 analysed the onset or worsening of ILD (ILD event) in RA patients treated with TNFi compared to non-TNFi biologics. TNFis were implicated in 85% of detected ILD events. Events and mortality were associated with older age and concomitant use of AZA. They found that patients with RA-associated ILD treated with TNFi may be at higher risk of adverse events and mortality from ILD than those treated with ABA or RTX, especially at older age and with pre-existing ILD.

An SR conducted in Spain on patients with SAD who developed ILD secondary to the use of biologics included clinical cases, retrospective and post-marketing studies.17 Only 52 cases of ILD had sufficient information, of which 38 had RA. Of these, 13 patients died, who were of more advanced age, had pre-existing ILD and immunosuppressive treatment more frequently than the 25 survivors. The authors conclude that anti-TNF therapy should not be used in patients with pre-existing ILD.

There are several cohort and case-control studies involving anti-TNFs compared with csDMARDs21,22 or non-anti-TNF bDMARDs.27–31 Based on data from the British Society for Rheumatology Biologics Register, Dixon et al.22 conducted an observational study to assess the influence of anti-TNF therapy on overall mortality, and specifically that due to ILD, in patients with RA and pre-existing ILD. A prevalence of ILD of 2.6% in patients with RA was found, being higher in those treated with anti-TNF therapy (299/9294) than with csDMARDs (68/2454) (2.9% vs. 1.8%; P = .02). This prevalence supports the suspicion that only a fraction of patients with this complication are identified in the clinic. No significant differences were found between either group for standardized mortality rate or ILD mortality, although the latter was numerically higher with anti-TNF (15/70) than with csDMARDs (1/14) (21% vs. 7%; NS). Pre-existing ILD was a strong predictor of all-cause mortality in both cohorts. The authors conclude that, despite the results, it should not be assumed that any patient with RA-associated ILD can be treated with anti-TNF therapy. Another study, based on the Spanish registry of biological therapies (BIOBADASER),40 also found no significant difference in the incidence of ILD or mortality from this complication in patients treated with anti-TNF therapy compared to those treated with csDMARDs in another cohort (EMECAR).

A Greek cohort study21 compared the efficacy and safety of anti-TNF therapy with that of csDMARDs (MTX) in 170 RA patients with and without ILD. Anti-TNF agents were not associated with the development or progression of ILD, but instead had a beneficial effect (wall thickening and respiratory function tests) in patients with pre-existing ILD.

Two studies investigated the mortality of patients with ILD treated with anti-TNF therapy.27,31 A multicentre study31 found that mortality due to respiratory causes was higher in patients with ILD than in those without ILD (10% vs. 6%; NS) and higher with anti-TNF therapy than with RTX (15% vs 8%; NS). Although statistical significance was not reached, the study results suggest a trend for patients with ILD to have higher mortality and to receive biologics than patients without ILD, and to have longer survival if they receive RTX rather than anti-TNF.

Druce et al. compared 5-year mortality rates and causes in RA and ILD patients treated first-line with RTX (43) or TNFi (309).27 The all-cause mortality rate was 94.8 with TNFi and 53.0 with RTX per 1000 patients/year. RTX compared with TNFi had an HR adjusted for confounders at 5 years of .49 (.23–1.06). Although not statistically significant, this study also suggests that patients with RA-associated ILD who received first-line RTX had a longer survival than those receiving TNFi.

In another study, conducted by Curtis et al.28 from US mutual insurance data, the incidence and exacerbation of ILD in a cohort of RA patients treated with non-anti-TNF bDMARDs compared to anti-TNF therapy was investigated. No difference in the risk of developing ILD or hospitalization due to pre-existing ILD was found among the cohorts, but patients with pre-existing ILD (419 patients with 499 episodes) were more likely to receive RTX (19.8% vs. 8.5%) and less likely to receive anti-TNF therapy (46.5% vs. 59.8%) than controls (10,800 patients without ILD). The risk of hospitalisation decreased with exposure to MTX (HR: .06; 95% CI:.06-.46; P < .001) and increased in men and previous hospitalisations due to asthma or ILD (P < .05). Therefore, there appear to be initial differences between patients receiving anti-TNF therapy or bDMARDs with another mechanism of action, but not in the incidence of ILD or onset of complications.

Nakashita et al. conducted two retrospective studies in patients with RA and ILD treated with biologics.29,30 In both studies, the extent of lung involvement was measured in 3 stages using HRCT. In the first study30 58 patients, 46 cases (with TNF blockers) and 12 controls (9 TCZ and 3 ABA). After one year of follow-up, the exacerbation rate of pre-existing ILD was high for TNF blockers (30.04%) and zero for non-TNFi bDMARDs (0%) (P = .024).

They then assessed the risk of onset/exacerbation of ILD in 163 patients with RA starting a biologic (102 TNFi and 51 non- TNFi), 58 with pre-existing ILD (30 grade I, 22 grade II and 6 grade III) and 105 without ILD.29 After one year of follow-up, ILD events occurred in 17 patients (10%), more frequently in the patients treated with TNFi.

In a retrospective study of 23 patients with RA and ILD treated with anti-TNF agents32 no statistically significant differences in FVC data were found at 6 months of treatment compared to baseline, suggesting that anti-TNF agents do not worsen lung function in the short term in patients with RA-associated ILD.

Tocilizumab (TZC). TZC is an interleukin-6 (IL-6) receptor blocker useful in the treatment of joint symptoms and some systemic manifestations of RA. While excessive production of IL-6 has been associated with fibrosis in ILD,29 several clinical cases of de novo onset or exacerbation of ILD following initiation of treatment with TZC have been published.

An SR conducted in 2010 to identify non-infectious pulmonary complications with new biological agents used in rheumatological diseases included 3 RCTs of TCZ with 589 patients diagnosed with RA.15 Six (1%) patients developed non-infectious pulmonary adverse events, including 3 exacerbations of pre-existing ILD (with fatal outcome), 1 recent onset ILD and 1 idiopathic pulmonary fibrosis.

Curtis et al.28 found no statistically significant difference in the incidence of ILD or hospital admissions of patients with pre-existing ILD in 59 patients treated with TCZ compared to 232 who received anti-TNF agents, although the number of biologics prior to TCZ was higher (2.1 ± 1.1 vs. 1.4 ± .7; P < .0001). Nakashita and collaborators also found no increase in the development or exacerbation of ILD in 36 patients treated with TCZ,29 but they did with TNFi.

In a post-marketing surveillance study41 that analysed cumulative safety data from 7,901 Japanese patients treated with TCZ, the incidence of ILD in the treated group was 10 cases/1,000 patients/year, clearly higher than the estimated incidence of the disease (4 and 4.5/1,000 patients/year).

A series of 12 patients with RA-associated ILD (11 UIP) treated with TCZ in a Japanese hospital was published recently,42 in which significant differences in biomarkers of joint activity (MMP-3) but not lung activity (KL-6 and LDH) were observed 6 months after administration of TCZ with respect to baseline values.

The evidence is scanty and study results are not uniform, therefore we must be cautious when drawing conclusions. Controlled studies are required to assess the efficacy of TCZ in patients with RA and ILD.

Rituximab (RTX). The observation of follicular B-cell hyperplasia and interstitial plasma cell infiltrates in lung biopsy specimens from RA and ILD patients provides histological justification for anti-CD20 therapy.43 However, further evidence of efficacy and safety in clinical practice is needed to recommend RTX therapy in these patients. The SR by Hadjinicolaou et al.15 included a RCT of RTX in RA in which ILD developed in 1 of the 136 patients in the active arm (.32%) and none of the 149 controls.

Subsequently, a prospective open-label pilot study was conducted with RTX9 in 10 patients with active and progressive RA and ILD (FVC > 50% and DLCO > 30%), who had not received bDMARDs or IS in the previous 8 weeks. RTX1 g was administered at weeks 0, 2, 24 and 26. There were 3 losses (1 infusion reaction and 2 deaths, from pneumonia and hip fracture). Of the 7 evaluable patients (FVC, DLCO or HRCT) at week 48, 5 were stable, 1 had worsened and 1 improved. Although no significant improvement was observed, RTX achieved active and progressive ILD stabilisation in treated patients.

As previously mentioned, two studies on mortality23,27 found better outcomes with RTX than with TNFi, although statistical significance was not reached. One found a lower mortality rate in the 43 patients who received RTX (HR .49; 95%CI: .23–1.06) than in the 309 who received TNFi as their first biologic.27 Previously, in a retrospective multicentre study,31 higher overall and respiratory mortality had been observed in patients who had ILD (22% and 10%) than in those who did not (14% and 6%); and among patients with ILD, in those who received anti-TNF therapy (31% and 15%) than in those who received RTX (8% and 4%).

Curtis et al.28 found no difference, in a retrospective study, in the risk of developing ILD or ILD complications during treatment with RTX, compared to anti-TNF therapy. However, patients were more likely to have received RTX if they had pre-existing ILD than if they did not (19.8% vs. 8.5%; P < .0001), unlike with anti-TNF therapy (46.5% vs. 59.8%; P < .0001).

Yusof et al. conducted a retrospective study44 in a cohort of 700 RA patients treated with RTX, 56 (8%) with pre-existing ILD. In 10 years of follow-up only 1/700 patients developed ILD (.4%). Of those with pre-existing ILD, 9 (16%) died from progression of the ILD. In 44 of the patients it was possible to assess the evolution of lung function (FVC and DLCO): 14 (31.8%) worsened, 23 (52%) remained stable and 7 (16%) improved with RTX. In another retrospective multicentre study, Duarte and collaborators45 evaluated the effectiveness and safety of RTX in patients with SAD (30 RA). Overall, they found lung function (FVC and DLCO) had stabilized at 12 months following treatment in both regimens and on the HRCT in the patients with NSIP. Infections were frequent and there were 4 treatment discontinuations and 2 deaths. According to the authors of these studies, RTX appears to be an acceptable treatment option, even in monotherapy, for patients with RA and ILD. In a recent Italian study46 a statistically significant difference in post-treatment FVC and DLCO values was found between cases (14 RTX) and controls (14 without RTX), despite the small sample size. These results coincide with the abovementioned pilot, prospective, open study conducted on 10 patients,9 and with numerous series presented at conferences. In general terms, the drug achieves stabilisation or improvement of respiratory function parameters in 70% of cases, without differences in response according to histopathological pattern, and is a relatively safe treatment (although usually associated with an increase in respiratory or urinary infections, the majority not severe).

Abatacept (ABA). ABA, a fusion protein that inhibits T lymphocyte co-stimulation, has been favourably positioned by GUIPCAR2 in the treatment of RA patients who develop ILD.

A low incidence of ILD induced or exacerbated by biologics in patients with RA was confirmed in two SRs.14,16 In the SR of clinical cases of ILD due to DMARds by Roubille et al.16 only one case was found to be related to ABA. In the extensive SR by Nelson et al.,14 a lower Incidence Rate (IR) was observed in an MA of 8 RCTs of ABA in RA (.09 (.01–.31) in the short term and .11 (.06−.20) during the combined extension period) than in health insurance studies (1.1 (.1–4.1) in MarketScan and 4.0 (1.6–8.2) MediCare). Furthermore, in this SR, observational studies and case reports showed that there was no increase in the incidence of ILD and that the parameters assessed improved in all but one case of pre-existing ILD. Therefore, despite the limitation of the data, it appears that ABA does not increase the incidence of ILD and may be useful in the management of patients with RA and associated ILD.

Other studies, with a lower level of evidence, have found similar results. Nakashita et al. reported, from a case-control study with 25 ABA controls29 and a series of 16 ABA cases,35 that ABA has less risk of producing new cases of ILD or worsening of pre-existing ILD than TNFi in Japanese RA patients. An extensive case series of RA and ILD patients collected in numerous centres in Spain, has led to different communications. When retrospectively comparing47 patients treated with ABA, 30 treated with RTX and 25 treated with TCZ, they found improvement at 12 months on HRCT in 36.4% of the patients with ABA, 28.6% with RTX and 8.3% with TCZ. In 2019 they reported the results of the largest series of RA and ILD treated with ABA.48 They analysed the clinical, functional respiratory (FVC and DLCO) and radiological (HRCT) progress of 181 patients treated with ABA (81 in monotherapy) during a median follow-up of 12.1 (6.2–24.1) months and observed stabilisation in most of the parameters analysed in a substantial percentage of the patients. Kurata et al. found in another series of 49 RA cases49 treated with different drugs, that the use of ABA can behave as a protective factor.

In contrast, an extensive retrospective US study (11,219 subjects) to assess the incidence of ILD and complications of pre-existing ILD in RA patients receiving their second or subsequent biologic, 28 in which 109 of the 419 patients with pre-existing ILD received ABA, found no difference in the incidence of ILD and complications of pre-existing ILD between the different biologics.

The safety of these drugs at the lung level has not been evaluated in any RCT and there is scanty available information. A post-marketing surveillance drug study50 conducted in 2,882 Japanese patients on tofacitinib identified 14 cases of ILD (.5%), of whom 3 died. In a description of 4 cases of RA with ILD treated with tofacitinib51 the authors state that the patients improved and had no pulmonary exacerbation during follow-up (8–12 months), but they do not provide pulmonary assessment data.

Little information is available on the effect of antifibrotics in patients with RA-associated ILD. The INBULID RCT10 was recently published, comparing nintedanib against placebo in fibrosing connective tissue diseases (CTD) (663 patients, of whom 89 were RA-ILD). In patients with a fibrotic pattern similar to that of IPF, the adjusted rate of decrease in FVC was -82.9 ml per year with nintedanib and -211.1 ml per year with placebo, with a difference of 128.2 ml (95%CI: 70.8–185.6; P < .001). There is a phase II clinical trial with pirfenidone in patients with RA and ILD (NCT02808871), for which we do not yet have results, which is why it has been excluded from the review. According to the considerations of Fisher and Distler,52 pirfenidone and nintedanib could stop the progression of ILD in patients with RA, and with SAD, as they have already demonstrated in patients with IPF.

This review has limitations. In this type of work, publication bias and even potential confusion by indication cannot be ruled out. Many of the included studies did not have a comparator group, which makes it difficult to assign a causal relationship. Finally, many patients received different DMARDs, and therefore the precise contribution of each treatment to the progression of ILD cannot be defined. Despite these limitations, our review provides evidence useful for the routine clinical practice of rheumatologists. On the one hand, MTX does not worsen progression of ILD and it would not be necessary to discontinue it in the treatment of these patients. On the other hand, RTX and ABA have shown promising results in RA-associated ILD and should be considered as a priority option over anti-TNF therapy when these patients need to initiate or change a biologic due to RA activity or ILD progression. However, there is not enough evidence to recommend the discontinuation of an anti-TNF if the RA is in remission and the ILD is not progressing.