New-onset non-infectious pulmonary manifestations among patients with systemic lupus erythematosus in Sweden

Systemic lupus erythematosus (SLE) is a heterogeneous autoimmune disease that has the potential to affect many organ systems. The lungs can be involved in up to two thirds of patients with SLE at some point in their disease course [1]. Prior studies have demonstrated a wide array of pulmonary manifestations, some of which are transient and not life threatening, such as pleurisy, to others that are more permanent and/or severe, such as interstitial lung disease (ILD) and pulmonary hemorrhage [1–3].

Much of what is known about lung disease in SLE comes from relatively small studies in selected clinical cohorts, such as the US multiethnic LUMINA cohort along with North American and European cohorts [1–3]. The true prevalence and incidence of lung disease in SLE may be underestimated in these studies because of patient selection, small sample sizes, and short follow-up periods. It is therefore important to identify pulmonary manifestations in a larger population-based cohort of individuals with SLE. Using such population-based data, less common phenotypes of SLE-related lung disease may be identified along with their onset relative to SLE diagnosis.

Therefore, we sought to investigate the prevalence and incidence of lung disease and whether lung manifestations occur early in SLE, using the Swedish Lupus Linkage (SLINK) cohort. It includes individuals with SLE and comparators from the general population who were identified through linkage of national registers. Using SLINK, we quantified and compared the history and incidence of pulmonary disease in individuals with SLE compared to those without SLE. This information will help us to understand the burden of lung disease among patients with SLE and if we can diagnose and potentially treat SLE lung manifestations early in the disease course to ultimately prevent pulmonary progression and damage. It will further help us to define a standard approach to detecting and treating lung disease in patients with SLE.

Among those with prevalent SLE, approximately 10% had at least one pulmonary manifestation during follow up compared with 2.4% in the general population (Table 3). The highest incidence rate was observed for pleural disorders (4.6 (95% CI 4.0–5.2)) followed by pulmonary embolism and ILD (3.3 (95% CI 2.9–3.8) for both). Having SLE was associated with a 5.4-fold increased risk of any pulmonary disease during follow up (95% CI 4.9–6.0) in adjusted models comparing to the general population. When considering each pulmonary disease group separately, the highest relative risk was found for ILD (HR 14.3 (95% CI 10.8–18.8)) and the lowest for pulmonary embolism (HR 3.8 (95% CI 3.1–4.6); Table 2). Due to limited power (< 5 cases) we were unable to investigate pediatric risks separately.

Bias analyses suggested that the HR from the main analyses was robust to unmeasured confounding due to smoking even under extreme confounding conditions. This was the case for both incident SLE (HR 5.5 (95% simulation interval 4.5–6.6)) and prevalent SLE (HR 5.1 (95% simulation interval 4.6–5.6)). These results did not change appreciatively when assuming a substantially higher relative risk of 5.4 between smoking and these pulmonary manifestations.

In this large population-based investigation, we have presented results on the incidence of various pulmonary disease manifestations in patients with incident and prevalent SLE. Our analyses showed that there is a much higher rate of overall pulmonary disease in both prevalent and incident cases of SLE compared to the general population. After adjusting for age and sex, there was an almost sixfold increased rate of pulmonary disease among those with incident SLE and a fivefold increased rate of pulmonary disease among those with prevalent SLE compared to the general population. The largest relative risk was for ILD, which appears to be partly driven by the observed low incidence of ILD in the general population.

The incidence rates were similar for all the individual pulmonary outcomes for incident and prevalent SLE. It is particularly intriguing that the rates of ILD for incident SLE were essentially the same as for prevalent SLE (just over three ILD cases per 1000 person-years of follow up). This does not imply that survival is poor but rather suggests that patients in the early stages of SLE experience ILD at a similar rate to patients with prevalent SLE. Data from the literature confirm that pulmonary damage is relatively common in SLE. In the multiethnic LUMINA cohort of over 600 patients with SLE, 5-year and 10-year cumulative risks of pulmonary damage were 7.6% and 11.6%, respectively [3].

Our study has some limitations. We cannot exclude that there may be some misclassification of SLE and pulmonary outcomes. We used a robust definition of SLE requiring multiple visits with a SLE code along with a diagnosis from a specialist. We anticipate that misclassification is minimal [4] and, if present, would most likely dilute the associations we observed towards the null. Pulmonary outcomes (i.e. ILD, pulmonary embolism) have previously been defined similarly using discharge diagnoses [11, 12] along with validation studies showing that the majority of cases met objective diagnostic criteria [13, 14]. Some pulmonary manifestations, including subtypes of ILD, are more difficult to validate and we cannot exclude the possibility of some misclassification.

We did not have data on SLE phenotypes (such as neuropsychiatric lupus or lupus nephritis), disease activity, or antiphospholipid syndrome or antibody status. By incorporating outpatient and inpatient admissions in our SLE definition, a range of phenotypes and disease severities were likely captured. In addition, our study was conducted using a relatively homogenous population in Sweden and therefore may not be generalizable to non-Caucasian populations with potentially different patterns of SLE disease manifestations and severity. Despite the relatively large study population, we had too few events to study outcomes in pediatric SLE separately. This would be interesting for a future study given a recent meta-analysis highlighting the fact that pulmonary manifestations of SLE were more common in patients with SLE who presented after the age of 50 compared to their younger peers, particularly ILD and serositis [15]. Last, we also did not have data on the smoking habits of our population. However, we conducted sensitivity analysis to assess the scope of unmeasured confounding by smoking and showed that our estimates from the main analyses were robust even under extreme bias assumptions.

Using Swedish population-based registers was a major strength of this investigation. We were able to obtain a representative sample of patients with SLE who together with their matched comparators from the general population could be followed with minimal losses. Our results are readily generalizable to other adult populations receiving similar standards of care to the Swedish population. Furthermore, the use of general population comparators is a strength, as it neither requires that individuals be sick with another disease, nor selects “healthy” individuals, both of which would distort the comparison. Last, we considered both prevalent and incident SLE to evaluate pulmonary disease complications in patients with established and newly identified disease, respectively.

This was the first population-based study of its kind addressing lung disease in SLE. We demonstrated that the rate of lung disease is much higher in SLE compared to the general population. Clinicians caring for patients with SLE should have heightened suspicion of lung disease, including ILD, even early within the disease course. We cannot say with certainty from our analyses that these findings of lung disease are clinically significant or important. However, additional investigations could define an important natural history of these findings and later help facilitate early intervention and prevention of pulmonary damage if it occurs.