Abstract
Rationale
Rheumatoid arthritis (RA) has been implicated in interstitial lung disease, as the majority of studies have comprised patients with known RA. However, it remains unclear whether an underlying risk for RA in combination with genetic risk for pulmonary fibrosis is associated with radiological markers of early lung injury and fibrosis in broader population samples.
Objective
We sought to determine whether genetic and serological biomarkers of RA risk in combination with the MUC5B (rs35705950) risk allele (T) are associated with interstitial lung abnormalities (ILAs) on computed tomography scans.
Methods
Associations of RA-risk HLA-DRB1 alleles (*04:01, *04:08, *04:05, *04:04, and *10:01) and serum RA autoantibodies with ILA in the Multi-Ethnic Study of Atherosclerosis (MESA; n = 4,018) and COPDGene (n = 5,963) cohorts were modeled using logistic regression and adjusted for age, sex, self-reported race and ethnicity, smoking history, body mass index, and principal components of genetic ancestry.
Results
The prevalence of an RA-risk HLA-DRB1 allele was 16.5% and 21.9% in the MESA and COPDGene cohorts, respectively. ILA was present in 3.9% and 11% of the MESA and COPDGene cohorts, respectively. An RA-risk HLA-DRB1 allele was not significantly associated with ILA in the MESA and COPDGene cohorts. In the MESA cohort, higher serum levels of immunoglobulin (Ig)A rheumatoid factor (RF) and anticyclic citrullinated peptide were associated with odds ratios for ILA of 1.20 (95% confidence interval [CI] = 1.07–1.35) and 1.19 (95% CI = 1.04–1.38), respectively. Among smokers without baseline ILA, per doubling of IgM RF was associated with an odds ratio for ILA 10 years later of 1.25 (95% CI = 1.08–1.44). Associations were not significantly different by MUC5B risk allele status.
Conclusions
RA-related HLA-DRB1 alleles were not associated with ILA, whereas higher serum levels of IgM RF among smokers without baseline ILA were associated with subsequent ILA.
Keywords: rheumatoid arthritis, human leukocyte antigen, interstitial lung disease, MUC5B
Interstitial lung diseases (ILDs) are characterized by varying degrees of lung inflammation and fibrosis (1). Clinical and genetic risk factors have been identified that confer a higher risk of ILD. For example, rheumatoid arthritis (RA)—a systemic autoimmune rheumatic disease characterized by symmetric polyarticular inflammatory arthritis—results in a higher risk of developing ILD (2). It has been postulated that RA contributes to the pathogenesis of ILD in combination with other risk factors for pulmonary fibrosis such as cigarette smoking and the MUC5B common promoter variant (rs35705950) (3). Development of ILD and the detection of computed tomography (CT) patterns related to pulmonary fibrosis (e.g., traction bronchiectasis, reticulation) are more prominent among patients with RA who have the MUC5B risk allele (T) (4, 5). Therefore, it has been posited that a combination of underlying RA risk, genetic predispositions to pulmonary fibrosis, and environmental exposures puts an individual at risk of developing ILD (6).
Most studies that have investigated the relationship between RA and ILD have relied on cohorts that comprise patients who have already been diagnosed with RA (2, 5, 7–9). If RA is a significant contributor to the development of ILD, it is plausible that early ILD-related features on CT scans of the lung would be detected in adults at higher risk of developing RA, either from genetic predisposition or from high levels of circulating autoantibodies. Higher levels of certain RA-related autoantibodies were associated with more interstitial lung abnormalities (ILAs) on CT in older community-dwelling adults (10). However, there were nearly 10 years between blood collection and ILA assessments in that study, and it was unknown whether participants had ILA at baseline. Notably, the study did not account for MUC5B promoter variant status, which is a significant genetic risk factor for ILD development among patients with RA (2). Furthermore, there is a strong genetic component in the development and progression of RA. The most well-studied genetic risk factor for RA is the human leukocyte antigen (HLA) system, a complex of genes that resides on chromosome 6 and encodes proteins critical to the immune system’s regulation (11). Several of these classic shared epitope alleles encode valine at position 11, which confers a nearly fourfold higher risk for RA (12, 13). Whether these RA-risk HLA-DRB1 alleles are associated with a greater burden of radiological interstitial abnormalities in the lung and stronger in those with the MUC5B promoter variant is unknown.
We hypothesized that adults who are predisposed to developing RA, whether because of genetic risk factors (i.e., those with HLA-DRB1 *04:01, *04:08, *04:05, *04:04, and *10:01 alleles) or elevated serum RA-related autoantibody levels, would have a greater burden of ILA, and the associations would be stronger among those with the MUC5B risk allele (T). Recent advances in technology have enabled the extraction of HLA-DRB1 allele frequencies from whole-genome sequencing (WGS), thus providing insights into the complex interactions between immunity and phenotypic traits (12, 14). Using available genomic and serologic biomarkers that are harmonized to ILA imaging data in the National Heart, Lung, and Blood Institute Trans-Omics for Precision Medicine (TOPMed) program, we performed an analysis in the U.S. population–based Multi-Ethnic Study of Atherosclerosis (MESA) and cigarette smoker–based Genetic Epidemiology of Chronic Obstructive Pulmonary Disease (COPDGene) cohorts.
Methods
Study Participants
TOPMed is a National Institutes of Health–sponsored program that performed WGS in several human study cohorts from which HLA-DRB1 alleles were extracted (14, 15). Of the TOPMed-affiliated cohorts, MESA and COPDGene each had HLA-DRB1 allele and ILA data that were available for analysis. MESA originally recruited 6,814 U.S. community-dwelling adults between the ages of 45 and 84 years without known cardiovascular disease at the time of enrollment to investigate subclinical cardiovascular disease (16). Baseline visits occurred at Exam 1 (2000–2002). Participants were invited for follow-up visits, with the most recent visit occurring at Exam 6 (2016–2018). For Phase 1 (2007–2012), COPDGene enrolled 10,198 non-Hispanic White and African-American individuals between the ages of 45 and 80 years with a history of 10 or more cigarette pack-years (17). Follow-up visits occurred at 5 and 10 years after the first visit. Institutional review board IRB approval from participating study sites and written informed consent from each participant were obtained.
HLA-DRB1 Alleles and Genotype Assessments
A summary of the assessments performed in the MESA and COPDGene studies is shown elsewhere (see Figure E1 in the data supplement). HLA typing was performed using WGS datasets from the TOPMed program as previously described (14). A description of the HLA typing is provided in the data supplement. MUC5B genotypes (rs35705950) were obtained from WGS that was performed from whole-blood DNA samples in MESA and COPDGene as part of the TOPMed program (15). For WGS, we used Illumina HiSeq X Ten instruments that targeted a mean depth of at least 30× (paired-end, 150-bp reads) and PCR-free library preparation kits from KAPA Biosystems for all of the sequencing.
RA-related Antibodies
Banked serum samples from MESA Exam 1 were used to quantify isotype-specific immunoglobulin (Ig) M and IgA rheumatoid factor (RF) and IgG anti–cyclic citrullinated peptide 2 (anti-CCP) by enzyme-linked immunosorbent assay using commercially available kits (TheraTest Laboratories, Inc.), with levels expressed in units per milliliter (18). IgA RF, IgM RF, and anti-CCP intraassay coefficients of variation were less than 10% (10). RA-related autoantibodies were not measured in the COPDGene study at the time of this analysis.
ILA Assessments
ILA was defined as the presence of ground-glass opacities, reticular abnormalities, traction bronchiectasis, honeycombing, and/or nonemphysematous cysts affecting at least 5% of nondependent lung (19). In the MESA study, ILA assessments were performed from Exam 1 cardiac CT scans, which capture the lower two-thirds of lung fields (20). In the COPDGene study, ILA was assessed from full-lung CT scans performed at the initial visit. ILA was assessed by one of several trained radiologists in MESA and by up to three trained readers in COPDGene (20, 21). In MESA and COPDGene, scans with focal, unilateral reticular/ground-glass abnormalities or patchy ground-glass abnormalities were labeled indeterminate and excluded from the final analysis. The fibrotic ILA subtype was defined as bilateral fibrosis in multiple lobes associated with honeycombing, traction bronchiectasis, and/or subpleural involvement (22, 23). In MESA, subsets of participants underwent follow-up visits, including at Exam 5 (2010–2012) in which a full-lung CT scan was performed in accordance with the MESA/SPIROMICS protocol and radiologists conducted ILA assessments on these scans (22, 24).
Statistical Analysis
Our primary objective was to examine associations of RA-risk HLA-DRB1 alleles with ILA in the MESA and COPDGene studies. Among the classic shared epitope alleles, we chose the *04:01, *04:08, *04:05, *04:04, and *10:01 alleles, because published studies show that they have the strongest association with RA risk and disease severity (12, 13). We used logistic regression models and adjusted for age, sex, self-reported race and ethnicity, first five principal components (PCs) of genetic ancestry, body mass index (in kilograms per square meter), smoking history, and cigarette pack-years. In MESA, PCs were computed in the entire cohort and pooled across all race/ancestry groups, using SNPs generated from the Affymetrix Human SNP Array 6.0 as part of the MESA SNP Health Association Resource (or, SHARe) project. In COPDGene, PCs were generated from the TOPMed Informatics Research Center using the PC-AiR and PC-relate methods, which provide accurate population structure and kinship estimation (25, 26). (For further details, see the data supplement.) Smoking history was categorized as never, former, or current in MESA and as former or current in COPDGene. To determine whether associations of RA risk HLA-DRB1 alleles with ILA are stronger among those with the MUC5B (rs35705950) risk allele (T), we used likelihood ratio tests with and without the interaction term, “RA risk HLA-DRB1 allele × MUC5B risk allele.” Results are presented as odds ratio (OR) for ILA per copy of the RA-risk HLA-DRB1 allele.
To examine the associations of RA-related autoantibodies with ILA overall and by MUC5B risk allele (T) status in MESA, we used similar logistic regression models as the HLA-DRB1 analysis. RA-related autoantibodies were log-transformed to meet assumptions of normality, and results are presented per doubling of the autoantibody level. Generalized additive models were used to visualize the continuous association of RA-related autoantibodies with ILA and to determine P values for nonlinearity and linearity. In MESA, we examined associations of HLA-DRB1 alleles and RA-related autoantibodies with new onset ILA on Exam 5 full-lung scans among those without ILA on their baseline scans. Similar logistic regression models were used in the primary analysis. Among participants in MESA and COPDGene with RA-risk HLA-DRB1 and ILA assessments, there were no missing covariate data. (To view the consort diagram, see Figure E2.) A two-sided P value < 0.05 was deemed statistically significant for the main analyses. For subgroup analyses that tested for effect modification with multiplicative interaction terms, we used an adjusted P value cutoff of <0.017 to account for the three main subgroup analyses (MUC5B, smoking history, and HLA-DRB1 allele status). Analyses were performed using R, Version 4.2.2 (R Foundation for Statistical Computing).
Results
There were 4,018 MESA and 5,963 COPDGene participants who had baseline HLA-DRB1 allele data and ILA assessments after indeterminate ILA was excluded. Baseline characteristics of the cohorts are shown in Table 1. The prevalence of at least one copy of an RA-risk HLA-DRB1 allele was 16.8% in MESA and 21.9% in COPDGene. There was a lower proportion of men and smokers in MESA compared with that in COPDGene. The prevalence of ILA was 3.9% in MESA at Exam 1 and 11% in COPDGene.
Table 1.
Baseline characteristics
| Characteristic | MESA | COPDGene |
|---|---|---|
| No. of participants | 4,018 | 5,963 |
| RA-risk HLA-DRB1 alleles, n (%)* | ||
| No copies | 3,343 (83.2) | 4,657 (78.1) |
| One copy | 675 (16.8) | 1,284 (21.5) |
| Two copies | 0 (0) | 22 (0.4) |
| IgM RF, U/ml, median (Q1–Q3) | 8 (4–16) | Unavailable |
| IgA RF, U/ml, median (Q1–Q3) | 9 (6–14) | Unavailable |
| Anti-CCP, U/ml, median (Q1–Q3) | 0.2 (0.1–0.5) | Unavailable |
| Female sex, n (%) | 2,069 (51.5) | 2,747 (46.1) |
| Race and ethnicity, n (%) | ||
| Non-Hispanic White | 1,679 (41.8) | 4,083 (68.5) |
| African American | 532 (13.2) | 1,880 (31.5) |
| Hispanic | 937 (23.3) | 0 (0) |
| Asian | 870 (21.7) | 0 (0) |
| Age (yr), mean (SD) | 61 (10) | 59 (8.8) |
| BMI (kg/m2), mean (SD) | 28 (5) | 28.7 (6.3) |
| Smoking history, n (%) | ||
| Never | 1,826 (45.5) | 0 (0) |
| Former | 1,625 (40.4) | 2,813 (47.2) |
| Current | 567 (14.1) | 3,150 (52.8) |
| Cigarette pack-years, mean (SD) | 12 (21) | 43.2 (23.5) |
| Interstitial lung abnormalities, n (%) | 158 (3.9) | 654 (11) |
| Fibrotic subtype, n (%) | 42 (1) | 100 (1.7) |
Definition of abbreviations: BMI = body mass index; CCP = cyclic citrullinated peptide; COPDGene = Genetic Epidemiology of Chronic Obstructive Pulmonary Disease; HLA = human leukocyte antigen; Ig = immunoglobulin; MESA = Multi-Ethnic Study of Atherosclerosis; Q = quartile; RA = rheumatoid arthritis; RF = rheumatoid factor; SD = standard deviation.
HLA-DRB1*04:01, *04:08, *04:05, *04:04, and *10:01.
The HLA-DRB1 allele frequencies by self-reported race and ethnicity subgroups in MESA and COPDGene are shown elsewhere (see Table E1). RA-risk HLA-DRB1 alleles were more common in the non-Hispanic White subgroup compared with other racial and ethnic groups in the MESA and COPDGene cohorts.
HLA-DRB1 Alleles
Associations of RA-risk HLA-DRB1 alleles with ILA are shown in Figure 1A. A copy of the RA-risk HLA-DRB1 alleles was not statistically significantly associated with ILA in the MESA cohort (odds ratio [OR], 1.17; 95% confidence interval [CI] = 0.76–1.82) or in the COPDGene cohort (OR, 0.95; 95% CI = 0.78–1.16). In both cohorts, associations between RA-risk HLA-DRB1 alleles and ILA were not significantly different by MUC5B risk allele (T) status (P values for MUC5B interaction ≥ 0.31). RA-risk HLA-DRB1 alleles were not associated with the fibrotic ILA subtype in the MESA or COPDGene cohort (Figure 1B).
Figure 1.
Associations of rheumatoid arthritis–risk HLA-DRB1 alleles (*04:01, *04:08, *04:05, *04:04, and *10:01) with (A) interstitial lung abnormalities (ILAs) and (B) fibrotic ILA in the Multi-Ethnic Study of Atherosclerosis (MESA) and Genetic Epidemiology of Chronic Obstructive Pulmonary Disease (COPDGene) cohorts. P values for the interaction of RA-risk HLA-DRB1 allele and MUC5B: 0.83 (MESA–ILA), 0.89 (MESA–fibrotic ILA), 0.31 (COPDGene–ILA), and 0.46 (COPDGene–fibrotic ILA). The x-axis indicates log-scale. CI = confidence interval; OR = odds ratio.
Given major differences in cigarette smoking between the two cohorts, sensitivity analyses stratified by smoking status (never vs. ever) and restricted to those with a history of 10 or more cigarette pack-years (n = 1,289) were performed in the MESA cohort (see Tables E2 and E3). The RA-risk HLA-DRB1 alleles were associated with higher odds for fibrotic ILA in this subgroup (OR, 2.91; 95% CI = 1.02–8.31). Among MESA participants without baseline ILA (n = 1,931), 192 (9.9%) had ILA detected on their Exam 5 full-lung CT scan. RA-risk HLA-DRB1 alleles were not associated with ILA on the Exam 5 scan (see Table E4).
RA-related Autoantibodies in MESA
Serum levels of RA-related autoantibodies by RA-risk HLA-DRB1 allele status and presence of ILA are shown elsewhere (see Table E5). Cross-sectional associations of baseline serum RA-related autoantibodies with ILA are shown in Table 2. Higher serum levels of IgA RF and anti-CCP were each associated with a higher OR for ILA on baseline CT. Every doubling of serum IgA RF and anti-CCP was associated with ORs for ILA of 1.20 (95% CI = 1.07–1.35) and 1.19 (95% CI = 1.04–1.38), respectively. IgM RF was not statistically significantly associated with ILA. Continuous associations between the RA-related autoantibodies and ILA are shown in Figures 2A–2C. The associations of IgA RF and anti-CCP with ILA were linear (P values for nonlinearity ≥ 0.62) (Figures 2B and 2C). Associations between RA-related autoantibodies and ILA were not significantly different by MUC5B risk allele status (for MUC5B interaction, P values ≥ 0.31). Each doubling of anti-CCP was associated with an OR for fibrotic ILA of 1.29 (95% CI = 1.07–1.56) (Table 2). Associations of IgM RF, IgA RF, and anti-CCP with ILA were not significantly different by HLA-DRB1 allele status (for HLA-DRB1 interaction, P values ≥ 0.29). Smoking history did not modify associations between serum autoantibody levels and ILA (see Table E6). A sensitivity analysis restricted to participants with a history of 10 or more cigarette pack-years is shown elsewhere (see Table E7).
Table 2.
Associations between rheumatoid arthritis autoantibodies and ILAs from Exam 1 scans in MESA
| Rheumatoid Arthritis Autoantibody | OR for ILA (95% CI) | P Value for MUC5B Interaction | OR for Fibrotic ILA (95% CI) | P Value for MUC5B Interaction |
|---|---|---|---|---|
| IgM RF | ||||
| Overall | 1.04 (0.93–1.16) | 0.78 | 1.10 (0.92–1.31) | 0.92 |
| MUC5B (GG) | 1.04 (0.92–1.19) | 1.05 (0.83–1.32) | ||
| MUC5B (GT/TT) | 1.00 (0.79–1.26) | 1.21 (0.89–1.66) | ||
| IgA RF | ||||
| Overall | 1.20 (1.07–1.35) | 0.57 | 1.11 (0.90–1.36) | 0.39 |
| MUC5B (GG) | 1.19 (1.04–1.36) | 1.01 (0.75–1.34) | ||
| MUC5B (GT/TT) | 1.19 (0.94–1.52) | 1.29 (0.91–1.82) | ||
| Anti-CCP | ||||
| Overall | 1.19 (1.04–1.38) | 0.31 | 1.29 (1.07–1.56) | 0.29 |
| MUC5B (GG) | 1.26 (1.08–1.47) | 1.43 (1.16–1.75) | ||
| MUC5B (GT/TT) | 0.94 (0.62–1.41) | 0.99 (0.53–1.87) |
Definition of abbreviations: CCP = cyclic citrullinated peptide; CI = confidence interval; Ig = immunoglobulin; ILA = interstitial lung abnormality; MESA = Multi-Ethnic Study of Atherosclerosis; OR = odds ratio; RF = rheumatoid factor.
Results are reported as OR for ILAs per doubling of rheumatoid arthritis autoantibody level. Genotypes are GG, GT, and TT. Model was adjusted for age, sex, smoking history, cigarette pack-years, body mass index, self-reported race and ethnicity, and principal components of genetic ancestry.
Figure 2.
Continuous associations of baseline serum (A) immunoglobulin (Ig)M rheumatoid factor, (B) IgA rheumatoid factor, and (C) anti-CCP with ILAs from Exam 1 computed tomography scans in the Multi-Ethnic Study of Atherosclerosis. Models were adjusted for age, sex, smoking history, cigarette pack-years, self-reported race and ethnicity, body mass index, and principal components of genetic ancestry. Solid lines represent the overall adjusted effect estimate; dashed lines represent the 95% confidence interval bands. Each vertical hashmark along the x-axis represents a single participant. The x-axis indicates log-scale. anti-CCP = anti-cyclic citrullinated peptide; ILAs = interstitial lung abnormalities.
Among participants without ILA at baseline, associations of baseline serum IgM RF, IgA RF, and anti-CCP with Exam 5 ILA are shown elsewhere (see Table E8). None of the RA-related autoantibodies were associated with ILA overall or in stratified analyses by MUC5B risk allele or RA-risk HLA-DRB1 allele status. Higher serum levels of IgM RF were more strongly associated with Exam 5 ILA in ever-smokers compared with never-smokers (for smoking interaction, P = 0.015) (Figure 3). Every doubling of serum IgM RF was associated with an OR for ILA of 1.25 (95% CI = 1.08–1.44) among ever-smokers and 0.95 (95% CI = 0.79–1.14) among never-smokers. In a sensitivity analysis restricted to participants with a history of 10 or more cigarette pack-years, higher serum levels of IgM RF and IgA RF were associated with ORs of 1.24 (95% CI = 1.03–1.49) and 1.24 (95% 1.03–1.49), respectively, for Exam 5 ILA (see Table E9).
Figure 3.
Continuous associations of baseline serum IgM rheumatoid factor with interstitial lung abnormalities (ILAs) from Exam 5 full-lung scans (2010–2012) by smoking history among participants without baseline ILA in the Multi-Ethnic Study of Atherosclerosis. Models were adjusted for age, sex, self-reported race and ethnicity, body mass index, and principal components of genetic ancestry. For nonlinearity, P values = 0.38 (for never-smokers) and 0.39 (ever-smokers). For linearity, P values = 0.73 (for never-smokers) and 0.001 (for ever-smokers). Black lines indicate never-smokers, and gray lines indicate ever-smokers. Solid lines represent the overall adjusted effect estimate, solid dots that comprise the lines represent the individual participant and corresponds to the vertical hashmark, and dashed lines represent the 95% confidence interval bands. Each vertical hashmark along the x-axis represents a single participant. The x-axis indicates log-scale. Ig = immunoglobulin.
Discussion
We examined genetic and serological biomarkers that confer a higher risk for RA and their associations with ILAs. We found that RA-risk HLA-DRB1 alleles were not associated with ILA in two independent cohorts of U.S. adults. Higher serum levels of IgA RF and anti-CCP were associated with ILA on CT. Associations of these RA-risk markers with ILA were not significantly different by MUC5B risk allele status. Our findings underscore the complex interaction between autoimmunity, genetics, and environmental factors in the development of ILD.
The multifactorial disease model of intrinsic and extrinsic factors has been proposed for RA-ILD, in which underlying dysregulation of the host immune system related to RA compounded by inhalation of harmful agents such as cigarette smoke leads to protein citrullination and autoantibody production in the lungs (6, 9, 27). This cascade of ongoing lung injury followed by aberrant lung repair may place an individual on a trajectory to develop pulmonary fibrosis. There is a higher prevalence of ILD in patients with RA (28, 29), and adults with both RA and the MUC5B promoter variant have a higher lifetime risk of pulmonary fibrosis, higher odds for a severe radiological subtype of pulmonary fibrosis (i.e., usual interstitial pneumonia), and a greater burden of early lung inflammation and fibrosis detected on CT (2, 4, 5, 7). Thus, it is plausible that a combination of RA, cigarette smoking, and the MUC5B promoter variant contributes to pulmonary fibrosis (6). A common feature of these studies is their reliance on data from patients who were already diagnosed with RA. Likewise, the lack of preclinical models for RA-ILD limits the ability to determine whether RA contributes significantly to ILD development. We leveraged imaging assessments of lung injury and fibrosis on CT scans, which may represent earlier stages of ILD, and common genetic variants and HLA alleles extracted from WGS to partly address this gap (12, 14, 22, 30, 31).
Despite the rationale for our hypothesis, RA-risk HLA-DRB1 alleles were not consistently associated with ILA. It is possible that the underlying pathobiology that contributes to RA has a minimal role in the early development of ILD. Rather, RA may propagate ongoing development of lung fibrosis in its later disease stages. Additionally, it is likely that not all ILA cases progress to fulminant lung fibrosis, which may be why our HLA-DRB1 findings were largely null. We did find the RA-related HLA-DRB1 allele to be associated with fibrotic ILA among participants in the MESA study who had a substantial cigarette pack-year history. This may indicate an enriched subgroup of adults with a combination of RA risk and harmful exposures that may contribute to a more severe subtype of ILA. Conversely, because of the cross-sectional nature of the analysis, this group may comprise patients with RA-ILD, which was not adjudicated for in MESA at enrollment. Nevertheless, we acknowledge that this finding was in a smaller subgroup with wider CIs, and replication is needed.
Studies have identified associations of other HLA-DRB1 alleles with IPF risk and fibrotic ILDs, suggestive of autoimmunity’s contributory role (32, 33). Our study complements these prior publications for two main reasons. First, we took a targeted approach by focusing on HLA alleles that confer the strongest risk for RA, given ongoing interests to elucidate the relationship between RA and ILD risk. Second, we used a cohort study design with a well-studied imaging phenotype that could represent subclinical ILD. The lack of associations between HLA-DRB1 and ILA in our study extends a recent analysis that did not find consistent associations between HLA gene variability and IPF risk (34). ILA may precede non-IPF ILDs, and we cannot rule out that other HLA alleles may contribute to the early development of ILD. Our study was likely underpowered to examine all available markers of HLA variation and ILA, but this is a future research priority.
In the MESA cohort, we found cross-sectional associations of higher baseline serum levels of IgA RF and anti-CCP with ILA. However, these autoantibodies were not associated with ILA on follow-up scans. The presence of ILA may indicate injury to the lung that results in citrullination of peptides that are then leaked into the blood. This would be in concordance with the mucosal-injury model that has been proposed for RA lung involvement (6, 35). Thus, elevated serum levels of IgA RF and anti-CCP may signal recurrent microinjuries to the lung. Also, although RF and anti-CCP are strongly associated with RA, these autoantibodies can be detected in other systemic autoimmune rheumatic diseases (e.g., psoriatic arthritis, myositis), infections, and in individuals without disease. This may explain why our autoantibody findings were discordant with our HLA-DRB1 allele results. Additionally, associations between RA autoantibodies and ILA were not stronger among those with the HLA-DRB1 alleles, which further suggests that elevated autoantibody levels are antecedent to lung injury. Our finding supports a U.K. study that found a unidirectional relationship between IPF and RA, in which observed IPF-related genetic risk variants conferred a higher risk for RA but not vice versa (36). We caution that this is speculative, as repeated measurements of RA autoantibodies were not performed in the MESA study, and we did not have an independent validation cohort.
In smokers without ILA at baseline in the MESA study, higher serum levels of IgM RF were associated with ILA on subsequent scans from Exam 5. A possible explanation for this finding is that harmful environmental exposures to the lungs (e.g., cigarette smoke) combined with systemic autoimmunity may confer a higher risk of developing lung injury that is detected on a CT scan. Notably, before a diagnosis of RA, IgM RF may be detected before IgA RF and anti-CCP (37). Also, higher serum levels of IgM RF and IgA RF were associated with the development of ILA at Exam 5 in a sensitivity analysis restricted to participants with a history of 10 or more pack-years. Further work with independent replication and serial measurements of RA-related autoantibodies will be informative.
Akin to cigarette smoking, it has been proposed that the combination of genetic predispositions to pulmonary fibrosis and RA may lead to the future development of ILD (5, 6). However, a consistent finding in our study was that associations of genetic factors for RA with ILA were weak and did not differ by MUC5B risk allele status. Although the MUC5B risk allele is associated with ILA and is more common in patients with RA-ILD and in patients with RA who have abnormal CT interstitial lung findings (2, 4, 30, 38–41), our study does not extend these findings to ILA. Thus, our study challenges how we conceptualize RA-ILD. Because the MUC5B promoter variant was not a significant effect modifier in our analysis, our findings do not align with the “multi-hit” model that has been invoked to explain RA-ILD (6). Furthermore, the clinical significance of the presence or absence of an RA HLA-DRB1 risk allele among individuals with ILA remains unknown. Conversely, the presence of ILA may indicate ongoing lung injury, resulting in the citrullination of proteins that contributes to the future risk of developing RA and other systemic autoimmune rheumatic diseases.
This study has several limitations. Both cohorts comprised older individuals, so the associations of these RA risk factors with ILA in younger adults are unknown. Although our study included diverse cohorts, we were likely underpowered for our analysis in the Asian, African American, and Hispanic subgroups. This raises an ongoing challenge of undersized cohorts that comprise people from underrepresented groups with comprehensive genomic, radiomic, and phenotype data for ILD studies. We purposely focused on HLA-DRB1 alleles, because they have a much stronger association with RA and account for 30–50% of heritable risk compared with just 15% for SNPs (42, 43). It is possible that there are more rare variants related to RA risk that may associate with ILA; this is a future research area (44). There were some differences in how ILA was assessed in the MESA and COPDGene studies, although overall criteria were similar. The independent assessments of ILA for each cohort, however, reduce ascertainment bias, as prior studies that examined other risk factors have shown replication in both cohorts (39). In the MESA study, ILA was assessed from Exam 1 cardiac CT scans that capture the lower two-thirds of lung fields, whereas ILA assessments were available from full-lung scans at Exam 5. This discordance in lung fields between Exams 1 and 5 may have influenced our results; therefore, future work with serial full-lung scans and RA risk assessments will be informative. Validation of our RA-related autoantibody findings with longitudinal ILA assessments in independent cohorts is needed.
In summary, HLA-DRB1 alleles that confer a higher risk for RA were not associated with ILA. Higher serum levels of IgA RF and anti-CCP were cross-sectionally associated with ILA, whereas IgM RF was associated with subsequent ILA among smokers without baseline ILA. Development and validation of preclinical RA-ILD disease models and prospective assessments of RA and ILD risk in humans are needed to elucidate the pathobiological role of RA in ILD development.
Supplemental Materials
Acknowledgments
Acknowledgment
The authors thank the other investigators, the staff, and the participants who provided biological samples and data for the MESA, COPDGene, and TOPMed studies. A full list of participating MESA investigators and institutions can be found at https://www.mesa-nhlbi.org. For a full list of investigators and institutions for the COPDGene study, see the data supplement.
Footnotes
The Multi-Ethnic Study of Atherosclerosis (MESA) lung study was supported by grants R01-HL077612, R01-HL093081, and RC1-HL100543 from the National Heart, Lung, and Blood Institute (NHLBI). MESA Lung Fibrosis was supported by grant R01-HL-103676 from the NHLBI. The MESA projects are conducted and supported by the NHLBI in collaboration with MESA investigators. Support for the MESA projects are conducted and supported by the NHLBI in collaboration with MESA investigators. Support for MESA is provided by NHBLI contracts 75N92020D00001, HHSN268201500003I, N01-HC-95159, 75N92020D00005, N01-HC-95160, 75N92020D00002, N01-HC-95161, 75N92020D00003, N01-HC-95162, 75N92020D00006, N01-HC-95163, 75N92020D00004, N01-HC-95164, 75N92020D00007, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168, N01-HC-95169, UL1-TR-000040, UL1-TR-001079, UL1-TR-001420, UL1-TR-001881, DK063491, and R01HL105756. J.S.K. was supported by the NHLBI grant K23-HL-150301. K.F.F. was supported by National Institutes of Health Training grant T32-HL-007891. M.R.A. was supported by NHLBI grant K23-HL-150280. A.J.P. was supported by NHLBI grant K23-HL-140199. E.J.B. was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases grant K23-AR-075112, the NHLBI grant R01-HL-164758, and the U.S. Department of Defense grant W81XWH2210163. Y.L. was supported by a Kennedy Trust for Rheumatology Research (KENN202109). J.A.S. was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (grant numbers R01 AR080659, R01 AR077607, P30 AR070253, and P30 AR072577), the R. Bruce and Joan M. Mickey Research Scholar Fund, and the Llura Gund Award for Rheumatoid Arthritis Research and Care. G.C.M. was supported by grant T32 AR007530 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases. G.M.H. was supported by NHLBI awards R01-HL-111024, R01-130974, and R01-135142. Whole-genome sequencing for the Trans-Omics in Precision Medicine (TOPMed) program was supported by the NHLBI. WGS for “NHLBI TOPMed: Multi-Ethnic Study of Atherosclerosis (MESA)” (phs001416.v1.p1) was performed at the Broad Institute of MIT and Harvard (3U54HG003067-13S1). Core support, including centralized genomic read mapping and genotype calling along with variant quality metrics and filtering, was provided by the TOPMed Informatics Research Center (supported by the NHLBI grant 3R01HL-117626-02S1 and contract HHSN268201800002I). Core support, including phenotype harmonization, data management, sample-identity quality control, and general program coordination, was provided by the TOPMed Data Coordinating Center (supported by the NHLBI grants R01HL-120393 and U01HL-120393 and contract HHSN268201800001I). Infrastructure for the CHARGE Consortium is supported in part by NHLBI grant R01-HL105756.
Author Contributions: Study design: J.S.K., K.F.F., V.M., G.M.H., M.H.C., C.K.G., R.G.B., and E.J.B. Data acquisition, analysis, and interpretation: J.S.K., K.F.F., V.M., A.M., S.S., Y.L., C.F.M., M.S., M.R.A., E.A.H., A.J.P., J.H.Y., G.C.M., J.A.S., R.P., M.M., S.S.R., J.I.R., I.N., G.R., J.T.G., R.W., S.R., G.M.H., M.H.C., C.K.G., R.G.B., and E.J.B. First draft of the manuscript: J.S.K., K.F.F., V.M., and E.J.B. Final manuscript approval: all authors.
This article has a data supplement, which is accessible at the Supplements tab.
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1. Wijsenbeek M, Cottin V. Spectrum of fibrotic lung diseases. N Engl J Med . 2020;383:958–968. doi: 10.1056/NEJMra2005230. [DOI] [PubMed] [Google Scholar]
- 2. Juge PA, Lee JS, Ebstein E, Furukawa H, Dobrinskikh E, Gazal S, et al. MUC5B promoter variant and rheumatoid arthritis with interstitial lung disease. N Engl J Med . 2018;379:2209–2219. doi: 10.1056/NEJMoa1801562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. McDermott GC, Doyle TJ, Sparks JA. Interstitial lung disease throughout the rheumatoid arthritis disease course. Curr Opin Rheumatol . 2021;33:284–291. doi: 10.1097/BOR.0000000000000787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Matson SM, Deane KD, Peljto AL, Bang TJ, Sachs PB, Walts AD, et al. Prospective identification of subclinical interstitial lung disease in rheumatoid arthritis cohort is associated with the MUC5B promoter variant. Am J Respir Crit Care Med . 2021;205:473–476. doi: 10.1164/rccm.202109-2087LE. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Palomäki A, Palotie A, Koskela J, Eklund KK, Pirinen M, Ripatti S, et al. FinnGen Lifetime risk of rheumatoid arthritis-associated interstitial lung disease in MUC5B mutation carriers. Ann Rheum Dis . 2021;80:1530–1536. doi: 10.1136/annrheumdis-2021-220698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kadura S, Raghu G. Rheumatoid arthritis-interstitial lung disease: manifestations and current concepts in pathogenesis and management. Eur Respir Rev . 2021;30:210011. doi: 10.1183/16000617.0011-2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. McDermott G, Gill R, Gagne S, Byrne S, Huang W, Cui J, et al. Associations of the MUC5B promoter variant with timing of interstitial lung disease and rheumatoid arthritis onset. Rheumatology (Oxford) . 2022;61:4915–4923. doi: 10.1093/rheumatology/keac152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Oka S, Higuchi T, Furukawa H, Shimada K, Okamoto A, Hashimoto A, et al. Serum rheumatoid factor IgA, anti-citrullinated peptide antibodies with secretory components, and anti-carbamylated protein antibodies associate with interstitial lung disease in rheumatoid arthritis. BMC Musculoskelet Disord . 2022;23:46. doi: 10.1186/s12891-021-04985-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Restrepo JF, del Rincón I, Battafarano DF, Haas RW, Doria M, Escalante A. Clinical and laboratory factors associated with interstitial lung disease in rheumatoid arthritis. Clin Rheumatol . 2015;34:1529–1536. doi: 10.1007/s10067-015-3025-8. [DOI] [PubMed] [Google Scholar]
- 10. Bernstein EJ, Barr RG, Austin JH, Kawut SM, Raghu G, Sell JL, et al. Rheumatoid arthritis-associated autoantibodies and subclinical interstitial lung disease: the Multi-Ethnic Study of Atherosclerosis. Thorax . 2016;71:1082–1090. doi: 10.1136/thoraxjnl-2016-208932. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Rioux JD, Goyette P, Vyse TJ, Hammarström L, Fernando MMA, Green T, et al. International MHC and Autoimmunity Genetics Network Mapping of multiple susceptibility variants within the MHC region for 7 immune-mediated diseases. Proc Natl Acad Sci U S A . 2009;106:18680–18685. doi: 10.1073/pnas.0909307106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Raychaudhuri S, Sandor C, Stahl EA, Freudenberg J, Lee HS, Jia X, et al. Five amino acids in three HLA proteins explain most of the association between MHC and seropositive rheumatoid arthritis. Nat Genet . 2012;44:291–296. doi: 10.1038/ng.1076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Viatte S, Plant D, Han B, Fu B, Yarwood A, Thomson W, et al. Association of HLA-DRB1 haplotypes with rheumatoid arthritis severity, mortality, and treatment response. JAMA . 2015;313:1645–1656. doi: 10.1001/jama.2015.3435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Luo Y, Kanai M, Choi W, Li X, Sakaue S, Yamamoto K, et al. NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium A high-resolution HLA reference panel capturing global population diversity enables multi-ancestry fine-mapping in HIV host response. Nat Genet . 2021;53:1504–1516. doi: 10.1038/s41588-021-00935-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Taliun D, Harris DN, Kessler MD, Carlson J, Szpiech ZA, Torres R, et al. NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium Sequencing of 53,831 diverse genomes from the NHLBI TOPMed program. Nature . 2021;590:290–299. doi: 10.1038/s41586-021-03205-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, et al. Multi-Ethnic Study of Atherosclerosis: objectives and design. Am J Epidemiol . 2002;156:871–881. doi: 10.1093/aje/kwf113. [DOI] [PubMed] [Google Scholar]
- 17. Regan EA, Hokanson JE, Murphy JR, Make B, Lynch DA, Beaty TH, et al. Genetic Epidemiology of COPD (COPDGene) study design. COPD . 2010;7:32–43. doi: 10.3109/15412550903499522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Majka DS, Vu TT, Pope RM, Teodorescu M, Karlson EW, Liu K, et al. Association of rheumatoid factors with subclinical and clinical atherosclerosis in African American women: the Multiethnic Study of Atherosclerosis. Arthritis Care Res (Hoboken) . 2017;69:166–174. doi: 10.1002/acr.22930. [DOI] [PubMed] [Google Scholar]
- 19. Hatabu H, Hunninghake GM, Richeldi L, Brown KK, Wells AU, Remy-Jardin M, et al. Interstitial lung abnormalities detected incidentally on CT: a position paper from the Fleischner Society. Lancet Respir Med . 2020;8:726–737. doi: 10.1016/S2213-2600(20)30168-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. McGroder CF, Hansen S, Hinckley Stukovsky K, Zhang D, Nath PH, Salvatore MM, et al. Incidence of interstitial lung abnormalities: the MESA Lung Study. Eur Respir J . 2023;61:2201950. doi: 10.1183/13993003.01950-2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Washko GR, Lynch DA, Matsuoka S, Ross JC, Umeoka S, Diaz A, et al. Identification of early interstitial lung disease in smokers from the COPDGene Study. Acad Radiol . 2010;17:48–53. doi: 10.1016/j.acra.2009.07.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Podolanczuk AJ, Oelsner EC, Barr RG, Hoffman EA, Armstrong HF, Austin JH, et al. High attenuation areas on chest computed tomography in community-dwelling adults: the MESA study. Eur Respir J . 2016;48:1442–1452. doi: 10.1183/13993003.00129-2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Putman RK, Gudmundsson G, Axelsson GT, Hida T, Honda O, Araki T, et al. Imaging patterns are associated with interstitial lung abnormality progression and mortality. Am J Respir Crit Care Med . 2019;200:175–183. doi: 10.1164/rccm.201809-1652OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Sieren JP, Newell JD, Jr., Barr RG, Bleecker ER, Burnette N, Carretta EE, et al. SPIROMICS Research Group SPIROMICS protocol for multicenter quantitative computed tomography to phenotype the lungs. Am J Respir Crit Care Med . 2016;194:794–806. doi: 10.1164/rccm.201506-1208PP. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Zhao X, Qiao D, Yang C, Kasela S, Kim W, Ma Y, et al. Whole genome sequence analysis of pulmonary function and COPD in 19,996 multi-ethnic participants. Nat Commun . 2020;11:5182. doi: 10.1038/s41467-020-18334-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Gogarten SM, Sofer T, Chen H, Yu C, Brody JA, Thornton TA, et al. Genetic association testing using the GENESIS R/Bioconductor package. Bioinformatics . 2019;35:5346–5348. doi: 10.1093/bioinformatics/btz567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Saag KG, Cerhan JR, Kolluri S, Ohashi K, Hunninghake GW, Schwartz DA. Cigarette smoking and rheumatoid arthritis severity. Ann Rheum Dis . 1997;56:463–469. doi: 10.1136/ard.56.8.463. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Olson AL, Swigris JJ, Sprunger DB, Fischer A, Fernandez-Perez ER, Solomon J, et al. Rheumatoid arthritis–interstitial lung disease–associated mortality. Am J Respir Crit Care Med . 2011;183:372–378. doi: 10.1164/rccm.201004-0622OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Gabbay E, Tarala R, Will R, Carroll G, Adler B, Cameron D, et al. Interstitial lung disease in recent onset rheumatoid arthritis. Am J Respir Crit Care Med . 1997;156:528–535. doi: 10.1164/ajrccm.156.2.9609016. [DOI] [PubMed] [Google Scholar]
- 30. Hunninghake GM, Hatabu H, Okajima Y, Gao W, Dupuis J, Latourelle JC, et al. MUC5B promoter polymorphism and interstitial lung abnormalities. N Engl J Med . 2013;368:2192–2200. doi: 10.1056/NEJMoa1216076. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Kim JS, Manichaikul AW, Hoffman EA, Balte P, Anderson MR, Bernstein EJ, et al. MUC5B, telomere length and longitudinal quantitative interstitial lung changes: the MESA Lung Study. Thorax . 2023;78:566–573. doi: 10.1136/thorax-2021-218139. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Xue J, Gochuico BR, Alawad AS, Feghali-Bostwick CA, Noth I, Nathan SD, et al. The HLA class II Allele DRB1*1501 is over-represented in patients with idiopathic pulmonary fibrosis. PLoS One . 2011;6:e14715. doi: 10.1371/journal.pone.0014715. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Fingerlin TE, Murphy E, Zhang W, Peljto AL, Brown KK, Steele MP, et al. Genome-wide association study identifies multiple susceptibility loci for pulmonary fibrosis. Nat Genet . 2013;45:613–620. doi: 10.1038/ng.2609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Guillen-Guio B, Paynton ML, Allen RJ, Chin DPW, Donoghue LJ, Stockwell A, et al. Association study of human leukocyte antigen variants and idiopathic pulmonary fibrosis. ERJ Open Res . 2024;10 doi: 10.1183/23120541.00553-2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Holers VM, Demoruelle MK, Kuhn KA, Buckner JH, Robinson WH, Okamoto Y, et al. Rheumatoid arthritis and the mucosal origins hypothesis: protection turns to destruction. Nat Rev Rheumatol . 2018;14:542–557. doi: 10.1038/s41584-018-0070-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Leavy OC, Kawano-Dourado L, Stewart ID, Quint JK, Solomon JJ, Borie R, et al. Rheumatoid arthritis and idiopathic pulmonary fibrosis: a bidirectional Mendelian randomisation study. Thorax . 2024;79:538–544. doi: 10.1136/thorax-2023-220856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Deane KD, O’Donnell CI, Hueber W, Majka DS, Lazar AA, Derber LA, et al. The number of elevated cytokines and chemokines in preclinical seropositive rheumatoid arthritis predicts time to diagnosis in an age-dependent manner. Arthritis Rheum . 2010;62:3161–3172. doi: 10.1002/art.27638. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Hobbs BD, Putman RK, Araki T, Nishino M, Gudmundsson G, Gudnason V, et al. Overlap of genetic risk between interstitial lung abnormalities and idiopathic pulmonary fibrosis. Am J Respir Crit Care Med . 2019;200:1402–1413. doi: 10.1164/rccm.201903-0511OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Kim JS, Axelsson GT, Moll M, Anderson MR, Bernstein EJ, Putman RK, et al. Associations of monocyte count and other immune cell types with interstitial lung abnormalities. Am J Respir Crit Care Med . 2022;205:795–805. doi: 10.1164/rccm.202108-1967OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Rice MB, Li W, Schwartz J, Di Q, Kloog I, Koutrakis P, et al. Ambient air pollution exposure and risk and progression of interstitial lung abnormalities: the Framingham Heart Study. Thorax . 2019;74:1063–1069. doi: 10.1136/thoraxjnl-2018-212877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Sack C, Vedal S, Sheppard L, Raghu G, Barr RG, Podolanczuk A, et al. Air pollution and subclinical interstitial lung disease: the Multi-Ethnic Study of Atherosclerosis (MESA) air–lung study. Eur Respir J . 2017;50:1700559. doi: 10.1183/13993003.00559-2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Gregersen PK, Silver J, Winchester RJ. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum . 1987;30:1205–1213. doi: 10.1002/art.1780301102. [DOI] [PubMed] [Google Scholar]
- 43. Kim K, Bang SY, Lee HS, Bae SC. Update on the genetic architecture of rheumatoid arthritis. Nat Rev Rheumatol . 2017;13:13–24. doi: 10.1038/nrrheum.2016.176. [DOI] [PubMed] [Google Scholar]
- 44. Ishigaki K, Sakaue S, Terao C, Luo Y, Sonehara K, Yamaguchi K, et al. BioBank Japan Project Multi-ancestry genome-wide association analyses identify novel genetic mechanisms in rheumatoid arthritis. Nat Genet . 2022;54:1640–1651. doi: 10.1038/s41588-022-01213-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.