SAN DIEGO — Clinically significant interstitial lung disease (ILD) is believed to occur in 5%-10% of patients with rheumatoid arthritis (RA), but robust data are lacking on how to best predict which patients face the highest risk for RA-associated ILD. However, the results of several studies presented at the American College of Rheumatology annual meeting indicate that researchers are making strides in this field of rheumatologic care.
Adding Genetic Factors Improves ILD Risk Prediction
In the realm of risk stratification, Austin M. Wheeler, MD, a rheumatology fellow at the University of Nebraska Medical Center, Omaha, discussed the development and validation of a combined clinical and genetic risk score for ILD. "There is clear and well documented phenotypic and genetic overlap of ILD with idiopathic pulmonary fibrosis (IPF)," Dr. Wheeler said. "A number of clinical risk factors have been described for RA-ILD, including older age, male sex, smoking history, higher disease activity, and seropositivity. There are also well-documented genetic risk factors for RA-ILD. The MUC5B genetic variant is the strongest risk factor for IPF, and it's been described in RA-ILD as well."
A recently published study indicated that a genetic risk score without the MUC5B variant improved predictive ability for IPF and interstitial lung abnormalities better than using the MUC5B variant alone, "but no prior attempts have been made at developing a composite genetic risk score in RA-ILD" using both genetic and clinical risk factors, he said.
For the current study, Dr. Wheeler and colleagues drew from 2,386 participants in the Veterans Affairs Rheumatoid Arthritis (VARA) Registry, a multicenter, prospective cohort of US veterans with rheumatologist-diagnosed RA and who fulfilled the 1987 ACR classification criteria. The researchers validated ILD through a systematic review of medical records, including clinical diagnosis of ILD plus either imaging or lung biopsy findings, and collected whole genome data that included 12 single nucleotide polymorphisms (SNPs) previously identified to be associated with risk for RA-ILD. They then used a meta-analytic approach to create pooled associations for each of those respective SNPs using data from the VARA registry participants as well as participants from the past study where the SNPs were first identified. "Those pooled associations were what we used for our effects size within the genetic risk score," which ended up using five of the SNPs, Dr. Wheeler explained. Next, he and his colleagues combined the genetic risk score with clinical risk factors including age, sex, smoking history, disease activity, and rheumatoid factor (RF) positivity to create their combined risk score.
The mean age of the cohort was 70 years, 89% were male, 78% had a smoking history, and 78% were anti–cyclic citrullinated peptide (CCP) antibody positive. Of the 2,386 participants, 224 (9.4%) had RA-ILD. The full composite risk score had the highest area under the receiver operating curve (AUC) of 0.67, compared with an AUC of 0.623 using the clinical factors alone, 0.651 using the clinical factors plus only the MUC5B variant, and 0.654 using the composite score minus only the MUC5B variant. These AUCs show that "the combined risk score performs better than clinical factors even without the inclusion of the MUC5B variant in the score, which is notable because it supports the importance of further investigation into polygenic risk scores in RA-ILD as there is clearly more at play in a patient's overall genetic risk," Dr. Wheeler said.
As an example of the composite score's ability to discriminate between people with and without RA-ILD, a cutpoint of 0.05 gave a sensitivity of 90.2% and would have eliminated about 25% of the cohort from unnecessary high-resolution CT scans and pulmonary function tests, he said.
"This study demonstrates the potential utility of genetic risk scores in RA-ILD identification and supports further investigation into individual risk stratification and screening," he concluded. "This isn't ready for clinical applicability by any means, but I think it serves as a proof of concept of the idea of a genetic risk score in RA-ILD."
Biomarker Score Investigated
In a separate abstract, Brent Luedders, MD, assistant professor of rheumatology and immunology at the University of Nebraska Medical Center, and colleagues set out to determine if a previously derived biomarker score is associated with prevalent and incident ILD in the same VARA Registry cohort. An abstract presented at the ACR 2022 annual meeting found that a panel derived from IPF peripheral biomarkers was significantly associated with RA-ILD, including matrix metalloproteinase (MMP)-2, -7, and -9, eotaxin, macrophage-derived chemokine (MDC), monocyte chemoattractant protein-1 (MCP-1), fms-like tyrosine kinase 3 ligand (Flt3L) and interleukin-8 (IL-8). For the current analysis, Dr. Luedders and colleagues measured the concentrations of seven biomarkers (MMP-7, MMP-9, eotaxin, MDC, MCP-1, Flt3L, IL-8) from serum/plasma samples collected from VARA's participants at enrollment to develop a score based on the concentrations of each biomarker.
Baseline characteristics were similar between the groups, although those with prevalent RA-ILD were slightly older than those without ILD, and those who developed incident ILD during follow-up had slightly higher RA disease activity at the time of enrollment. When the researchers examined the association of the biomarker score with prevalent RA-ILD as a continuous measure, they found an adjusted OR of 1.08 for prevalent RA-ILD for each 1-point increase in the biomarker score. "When this was divided into quartiles, we found that the highest quartile of the biomarker score was associated with an adjusted odds ratio of 2.31 for prevalent RA-ILD," Dr. Luedders said. "We saw a significant P for trend of < .001, suggesting a dose-response relationship, in which higher scores had higher risk." Similar associations were observed for incident RA-ILD, in which participants with the highest quartile had an adjusted hazard ratio of 2.26 for incident RA-ILD.
The AUC of 0.653 that was obtained with clinical factors did not significantly improve with inclusion of the biomarker score, rising to only 0.669. "In receiver operating characteristic analysis, the addition of the biomarker score to clinical variables (age, sex, race, smoking status, anti-CCP positivity, and RA disease activity by DAS28) did not lead to a significant increase in the area under the curve. Therefore, further work is needed to identify combinations of clinical, biomarker, and other factors to accurately predict which people with RA will develop ILD," he said.
Dr. Luedders acknowledged certain limitations of the results, including the fact that MMP-2 was not measured in this cohort and thus not included in the score. "This was an observational study with usual care; therefore, the absence of systemic evaluation for ILD may miss early or mild RA-ILD cases," he added. "Similarly, a male predominance may limit the generalizability, and we have limited information on the RA-ILD pattern." He concluded that the study results "support the shared pathogenesis of IPF and RA-ILD. However, we found that this score has limited discriminative performance, compared to clinical risk factors alone."
Drilling Down on ILD Subtypes
In a poster abstract presentation at the meeting, Gregory Campbell McDermott, MD, MPH, a rheumatologist at Brigham and Women's Hospital, Boston, highlighted results from a study that investigated differences in demographic, serologic, and lifestyle factors for RA-ILD and the major subtypes of RA-ILD: usual interstitial pneumonia (UIP) and nonspecific interstitial pneumonia (NSIP). "Historically, RA-ILD has been studied as a single entity, even though we increasingly recognized that there are lots of different subtypes that fall under the umbrella of RA-ILD," Dr. McDermott said in an interview. "We are also learning that the different subtypes probably have both prognostic and potentially therapeutic implications. For example, the UIP subtype, which is the most fibrotic subtype, has the worst prognosis but also may be a potential target for antifibrotic therapies. We've been trying to see if we can identify factors that are associated with specific subtypes, in particular the UIP subtype which has the worst prognosis."
He and his colleagues examined 208 patients with RA-ILD with a mean age of 51 years and 547 patients with RA but no ILD with a mean age of 49 years from two RA cohorts comprising 3,328 patients: the Mass General Brigham Biobank RA Cohort and the Brigham RA Sequential Study (BRASS). Of the 208 RA-ILD cases, nearly half (48%) were RA-UIP, 18% were RA-NSIP, 8% were organizing pneumonia, 3% were respiratory bronchiolitis-ILD, and 23% were other/indeterminate. After conducting multivariable adjusted analyses, the researchers found that RA-ILD was associated with male sex (OR, 1.58; 95% CI, 1.09-2.23), seropositivity for RF and/or anti-CCP (OR, 2.22; 95% CI, 1.51-3.24) and being an ever smoker (OR, 1.70; 95% CI, 1.13-2.54). Having all three of these risk factors was strongly associated with RA-ILD (OR, 6.04; 95% CI, 2.92-12.47) and with RA-UIP in particular (OR, 7.1). "We found that a lot of the traditional RA-ILD risk factors like male sex, history of smoking, and seropositive status were most strongly associated with a UIP pattern," Dr. McDermott said. "We think this is a first step in trying to understand how these different ILD subtypes may have different risk factors, pathogenesis, and potentially different treatments, prevention, and screening strategies."
While clinicians wait for guidelines on systemic autoimmune rheumatic disease-associated ILD that are expected to be published by the ACR in 2024, he added that "we probably shouldn't screen every single person with RA for ILD, but we need to identify people who have symptoms or findings on clinical exam. This study wasn't designed to look specifically at who is at high risk, but I think we are moving toward that question: Who is high risk, and who's asymptomatic [but] may need more screening?"
He pointed out limitations of the study, including its retrospective design and the fact that imaging was done for clinical purposes, "so it's probably a higher risk group to begin with than the whole RA population," he said. "We also didn't have data on RA disease activity or erosions, some of these other measures that we think are important for understanding the full RA disease phenotype in these patients."
Dr. Wheeler reported having no disclosures. Dr. Luedders reported that his study was supported by the VA, the Rheumatology Research Foundation, and the University of Nebraska Medical Center Mentored Scholars Program. Dr. McDermott reported that his study was supported by the Rheumatology Research Foundation.
This article originally appeared on MDedge.com, part of the Medscape Professional Network.
Comments