Abstract
Objective The HLA-B27 allele is strongly associated with spondyloarthritis (SpA). HLA-B27 is included in SpA classification criteria and referral strategies for axial SpA. Investigations of HLA-B27 testing in usual clinical practice are limited.
Methods We identified all adult patients tested for HLA-B27 from January 1, 2022, to December 31, 2022, in the Mass General Brigham healthcare system. We examined patient demographics; ordering provider specialty; testing indication; concurrent testing with antinuclear antibodies (ANA), rheumatoid factor, and/or anticyclic citrullinated peptide autoantibodies; and rheumatology referral. We compared the rate of rheumatology referral between HLA-B27–positive and HLA-B27–negative patients.
Results HLA-B27 tests were ordered for 1960 patients (62.4% female; average age: 47.4 yrs). The most common specialties testing HLA-B27 were rheumatology (39.7%) and ophthalmology (21.4%). The most common indications for HLA-B27 testing were peripheral arthritis (33%), uveitis (22%), and back pain (16.7%). The majority of HLA-B27 tests (69.3%) were ordered concurrently with other autoantibody tests. A total of 11% of tested patients were HLA-B27 positive. Ophthalmology had the highest positive rate (15.4%), whereas reactive arthritis was the indication with the highest positive test rate (50%). A greater proportion of HLA-B27–positive patients were referred to rheumatology (53% vs 32%; P = 0.002).
Conclusion HLA-B27 testing was frequently performed by rheumatologists and nonrheumatologists for a broad spectrum of indications. Cotesting HLA-B27 with ANA and rheumatoid arthritis autoantibodies was common. Nearly half of HLA-B27–positive patients were not referred to rheumatology. Further efforts are needed to promote judicious use of HLA-B27 testing and optimize referral pathways to rheumatology.
The HLA-B27 allele is an established genetic risk factor for spondyloarthritis (SpA).1 Strong associations between HLA-B27 and SpA were first described for radiographic axial SpA (axSpA) in 1973,2,3 and subsequent studies linked the allele with uveitis,4 reactive arthritis,5 inflammatory bowel disease–associated arthritis, and psoriatic arthritis, particularly with axial involvement.6 Given these associations, HLA-B27 positivity is included in the Assessment of Spondyloarthritis international Society (ASAS) classification criteria for axSpA and peripheral SpA7 and incorporated into referral strategies for patients with low back pain.8-12 Testing for HLA-B27 can be performed using flow cytometry as well as genotyping methods including PCR-based assays.13 Similar to other diagnostic tools, the utility of HLA-B27 testing for a SpA diagnosis depends on the pretest probability of disease.14 The judicious use of HLA-B27 testing has been a focus of medical recommendations, including those of Choosing Wisely Canada, which recommended against ordering HLA-B27 “unless SpA is suspected based on specific signs or symptoms.”15 Despite these guidelines, there has been limited investigation into how HLA-B27 testing is used in routine clinical care.
We investigated HLA-B27 testing by rheumatologists and nonrheumatologists using comprehensive electronic health records (EHRs) from a large multihospital healthcare system. We analyzed the use of HLA-B27 testing across specialties, indications for testing, cotesting for other autoantibodies, and patterns of rheumatology referral among patients with HLA-B27 tests ordered by nonrheumatologists. We hypothesized that HLA-B27 testing is frequently being used by nonrheumatologists for screening purposes and that positive testing would result in more frequent rheumatology referral.
METHODS
Population. We identified all patients aged ≥ 18 years who had HLA-B27 testing performed in the Mass General Brigham healthcare system between January 1, 2022, and December 31, 2022, using EHR data extracted from the Research Patient Data Registry.16 The Mass General Brigham healthcare system is a large, multihospital healthcare system in the greater Boston, Massachusetts area that includes both academic and community hospitals as well as specialty and primary care practices.
HLA-B27 and other autoimmune disease testing. We extracted the HLA-B27 test orders and results for all patients from the EHRs. We also investigated serological testing for other autoimmune diseases sent during the study period. Specifically, we extracted tests for antinuclear antibody (ANA), rheumatoid factor (RF), and anticyclic citrullinated peptide antibodies (anti-CCP). We determined the date for the HLA-B27 testing, the dates for additional serologic autoimmune testing, as well as the specialty of the ordering provider. We considered testing to be concurrent if ANA, RF, and/or anti-CCP testing was ordered by the same provider within 1 month of HLA-B27 testing.
Indications for HLA-B27 testing. We used the International Classification of Diseases, 10th revision (ICD-10) codes associated with the encounter when the HLA-B27 testing was ordered to determine the indication for testing. As detailed in Supplementary Table S1 (available with the online version of this article), we categorized the indications for testing into the following categories: back pain, peripheral arthritis, uveitis, temporomandibular joint pain, other autoimmune condition, psoriasis, inflammatory bowel disease, reactive arthritis, enthesopathy, elevated erythrocyte sedimentation rate or C-reactive protein, or other reasons for testing. If none of the codes detailed in Supplementary Table S1 were associated with the encounter, we performed a manual chart review to determine the indication for testing.
Rheumatology referral and comparison between HLA-B27–positive and HLA-B27–negative patients. To determine the effect of HLA-B27 test results on subsequent rheumatology referral, we examined rheumatology referral rates among HLA-B27–positive vs HLA-B27–negative patients. First, we excluded all patients who had HLA-B27 testing ordered by a rheumatology provider. We analyzed all HLA-B27–positive patients and selected a random sample of 100 HLA-B27–negative patients using a random number generator as the comparator group. We identified any rheumatology visits that occurred after HLA-B27 testing in the EHR by October 30, 2023. For any patient who did not have a rheumatology visit, we manually reviewed the EHRs to determine if there was documentation of rheumatology referral outside of our healthcare system, recommendation for rheumatology referral that was declined by the patient, patient “no show” in a rheumatology clinic, or rheumatology evaluation occurring prior to HLA-B27 testing.
Additional data. We extracted date of birth and sex from the EHRs. We also extracted the patient’s self-reported race, selected from a list of options at enrollment in the EHR.
Analysis. We described the number of HLA-B27 tests ordered by specialty and the percentage of the overall tests ordered by each specialty. We also determined the proportion of HLA-B27 tests that were ordered concurrently with ANA, RF, and/or anti-CCP testing. We described the rate of cotesting overall by ordering provider specialty and by the indication for testing.
We compared demographic features of patients who were positive vs negative for HLA-B27 testing using descriptive statistics. We examined for statistical differences between the HLA-B27–positive and HLA-B27–negative patients using chi-square tests for categorical variables and t tests for continuous variables. We also determined the proportion of tests that were positive for each specialty and for each indication. We calculated 95% CIs for the proportion of positive tests using a binomial distribution and the Wilson method.17 For visualization, we plotted the proportion of positive tests by specialty and the estimated general US population prevalence of HLA-B27 based on US National Health and Nutrition Examination Survey data.18
Finally, we examined the rate of referral to rheumatology among patients tested for HLA-B27 by nonrheumatologists. We described the rate of referral to rheumatology among HLA-B27–positive patients by the specialty of the provider that had ordered the test. We also compared the referral rates of HLA-B27–positive vs HLA-B27–negative patients using the chi-square test.
All statistical analyses were performed using R (version 4.3.1; R Foundation for Statistical Computing). Graphing was done using GraphPad Prism (version 10.4; GraphPad).
RESULTS
We identified a total of 1960 adult patients who were tested for HLA-B27 in our healthcare system during the 2022 calendar year. A total of 1977 tests were performed, and 15 patients were tested for HLA-B27 more than once during the study period. Of those who had multiple tests, 13 patients were tested twice and 2 were tested 3 times. All duplicate tests were performed by the same method as the original test and demonstrated concordant results. None of the duplicate tests were indeterminate. The demographics of the 1960 patients tested for HLA-B27 are detailed in Table 1. The mean age at testing was 47.4 years, and 62.4% of tested patients were female. The majority of patients tested for HLA-B27 (77.1%) self-identified as White race.
Characteristics of patients tested for HLA-B27 in the Mass General Brigham Healthcare system in 2022.
A description of the provider specialties that ordered HLA-B27 testing in the Mass General Brigham healthcare system is provided in Figure 1A. Of the 1977 tests, 784 (39.7%) were ordered by rheumatology. The next most common provider specialties ordering HLA-B27 testing were ophthalmology (423 tests, 21.4% of total tests), oral and maxillofacial surgery (241 tests, 12.2% of total tests), general internal medicine and nonrheumatology internal medicine subspecialties (201 tests, 10.2% of total tests), and family medicine (145 tests, 7.3% of total tests). The indications for testing are detailed in Figure 1B. The most common indication for HLA-B27 testing was peripheral arthritis (652 tests, 33% of total tests), followed by uveitis (434 tests, 22% of total tests), back pain (331 tests, 16.7% of total tests), and temporomandibular joint disease (233 tests, 11.8% of total tests).
HLA-B27 testing by (A) specialty and (B) indication. ANA: antinuclear antibody; anti-CCP: anticyclic citrullinated peptide antibodies; CRP: C-reactive protein; ESR: erythrocyte sedimentation rate; IBD: inflammatory bowel disease; OMFS: oral and maxillofacial surgery; PM&R: physical medicine and rehabilitation; TMJ: temporomandibular joint.
We describe the pattern of HLA-B27 ordering in isolation or in combination with ANA and/or rheumatoid arthritis (RA) autoantibodies (RF and/or anti-CCP) in Table 2. Overall, 30.7% of HLA-B27 tests were ordered without cotesting for ANA or RF/anti-CCP. HLA-B27 was ordered along with ANA in 8.2%, with RF/anti-CCP in 26.2%, and with both ANA and RF/anti-CCP in 35% of cases. Details of cotesting by specialty are described in Table 2 and cotesting by indication in Table 3. Notably, cotesting of HLA-B27 with ANA and/or RF/anti-CCP was observed across specialties including rheumatology. For example, 21.4% of HLA-B27 tests requested by rheumatologists were ordered in combination with both ANA and RF/anti-CCP.
Cotesting of HLA-B27 with ANA and RF and/or anti-CCP by specialty.
Cotesting of HLA-B27 with ANA and/or RF/anti-CCP by indication.
Of the 1960 patients tested for HLA-B27, 215 had a positive test and 1731 had a negative test. Fourteen patients with indeterminate results or a “failed sample” were excluded from a comparison between HLA-B27–positive and HLA-B27–negative patients. All tests with indeterminate results were performed using flow cytometry. A comparison of patients with positive vs negative HLA-B27 testing is detailed in Table 1. There was no significant difference in age at testing between the positive and negative patients (46.4 vs 47.5 yrs; P = 0.35). A lower proportion of the HLA-B27–positive patients were female (51.2% vs 63.9%; P < 0.001). There were no significant differences in self-identified race between the HLA-B27–positive and HLA-B27–negative patients (P = 0.46).
The specialties with the highest proportion of positive tests (Figure 1A) were ophthalmology (15.4% of 423 tests), other specialties (14.3% of 14 tests), and rheumatology (12.1% of 784 tests). We also determined the rate of positive testing based on the indication for testing (Figure 1B). Reactive arthritis (50% of 10 tests), uveitis (15.4% of 434 tests), and back pain (15.1% of 331 tests) were the indications with the highest rate of positive HLA-B27 testing. Details of positive testing rate and 95% CIs are also detailed in Supplementary Table S2 (available with the online version of this article).
We report the rate of rheumatology referral among patients tested for HLA-B27 in Table 4. Among 120 patients who had a positive HLA-B27 test ordered by a nonrheumatologist, 63 (53%) were ultimately referred to rheumatology by December 2023. Ophthalmology referred the most HLA-B27–positive patients to rheumatology, though only 26 of 65 HLA-B27–positive patients (40%) were referred from this specialty. There were 1051 patients who had negative HLA-B27 testing performed by a nonrheumatologist. Among a random sample of 100 of those patients, 32 (32%) were referred to rheumatology. The difference in rheumatology referral rate between HLA-B27–positive and HLA-B27–negative patients was statistically significant (P = 0.002).
Rheumatology referrals among patients tested for HLA-B27 by nonrheumatologists.
DISCUSSION
In our large multihospital healthcare system, we used comprehensive EHR data to examine the use of HLA-B27 testing by rheumatologists and nonrheumatologists. We found that although rheumatology was the specialty that most commonly ordered HLA-B27 testing, the majority of HLA-B27 tests performed in our healthcare system (60.3%) were ordered by nonrheumatologists. There was a wide range of indications for HLA-B27 testing, with back pain being only the third most common. Moreover, we observed frequent cotesting of HLA-B27 with ANA and/or RA autoantibodies, suggesting it may be commonly used as part of broader rheumatologic screening. Ultimately, 53% of patients who were found to be HLA-B27 positive by a nonrheumatologist were referred to rheumatology. These findings suggest that further efforts are needed to promote judicious use of HLA-B27 testing and improve referral pathways to rheumatology for patients who are HLA-B27 positive and have symptoms that are suspicious for HLA-B27–associated diseases.
Our study adds to a limited existing literature about the use of HLA-B27 in usual clinical practice. Ahrari and colleagues reported that family medicine was the most common specialty ordering HLA-B27 for patients who were subsequently referred to rheumatology in the St. Joseph’s healthcare system in Ontario, Canada.19 These findings contrast with our own, where ophthalmology was the specialty that most commonly referred HLA-B27–positive patients to rheumatology. A previous survey of 1690 American physicians found that among nonrheumatology healthcare providers who suspected inflammatory back pain in patients, 90% chose to test for ANA and RF, 75.5% opted for HLA-B27, and 44.7% for anti-CCP to confirm the presence of an autoimmune rheumatic disease, suggesting a lack of knowledge regarding the rationale behind the usage of each biomarker.20 In our study, we also observed frequent cotesting of HLA-B27 with ANA and/or RA autoantibodies, including among patients tested for a back pain indication (23.9% of patients with back pain cotested for HLA-B27 with ANA and RA autoantibodies), suggesting areas for improvement in education and appropriate use of HLA-B27 and autoantibody testing.
To improve the appropriate use of HLA-B27 testing in clinical practice, national “Choosing Wisely” campaigns in both Canada and the US have emphasized that HLA-B27 testing should be limited to patients with signs or symptoms suggestive of SpA.15,21,22 These recommendations reflect the inclusion of HLA-B27 as a criterion in the ASAS classification criteria for axSpA7,23 and peripheral SpA,23 as well as in the Amor criteria for SpA.24 Many referral strategies for axSpA have been developed and evaluated, and most include HLA-B27 as a criterion.8-12 One of the studies that compared 10 different referral strategies found that a positive HLA-B27 test had the second-highest positive likelihood ratio (12.7) among 19 referral criteria, surpassed only by the presence of sacroiliitis on magnetic resonance imaging.10 In our study, although we noted higher rheumatology referral rate among HLA-B27–positive patients, approximately half of HLA-B27–positive patients were not referred, suggesting either a persistent referral gap or unnecessary testing of HLA-B27. Reassuringly, only 3 of 16 HLA-B27–positive patients who were tested for back pain were not referred to rheumatology. In contrast, the role of HLA-B27 testing for patients with peripheral joint symptoms is less well established. Despite the inclusion of HLA-B27 in classification criteria for peripheral SpA,23 we are not aware of any studies of rheumatology referral strategies for peripheral joint symptoms that incorporate HLA-B27 testing. Prior research from ASAS investigators demonstrated a high prevalence of the HLA-B27 allele among patients with peripheral SpA compared to those who were not classified as having SpA.23 Similar findings were observed in the worldwide ASAS–Peripheral involvement in SpA study, where the prevalence of HLA-B27 was 64% among patients with peripheral SpA tested for HLA-B27.25 However, it should be noted that a significant proportion of patients with peripheral SpA also had axial involvement, and the prevalence of HLA-B27 may be lower among patients with isolated peripheral joint involvement. In our study, peripheral arthritis was the most common indication for HLA-B27 testing, emphasizing the importance of further investigation of the use of HLA-B27 testing in patients with peripheral joint symptoms.
Our study also provides insight into HLA-B27 test ordering and subsequent referral practices in our healthcare system. Notably, there was significant variability in both the rate of positive testing and subsequent referral rates by specialty. Ophthalmology clinics had a high rate of HLA-B27 testing (21.4% of all tests were ordered by ophthalmology) and the highest rate of positive tests (15.4% of tests ordered by ophthalmology were positive). Prior literature has demonstrated a high prevalence of undiagnosed axSpA among patients with uveitis, and this prevalence may be higher among HLA-B27–positive patients.26-28 In our study, 60% of HLA-B27–positive ophthalmology patients were not referred to rheumatology, suggesting that the implementation of screening and referral strategies in ophthalmology clinics may be fruitful in identifying undiagnosed patients with axSpA. Several simple screening methodologies to facilitate appropriate referrals for evaluation of axSpA have also been studied in the ophthalmology setting.29-31
Our study has several strengths. We used comprehensive EHRs from a large healthcare system that includes both academic and community hospitals, as well as outpatient clinics, and augmented these records with manual chart review. Using these comprehensive records, we were able to investigate the specialties of providers ordering HLA-B27 testing and the indications for testing. Further, to our knowledge, this is the first study specifically investigating patterns of cotesting HLA-B27 with ANA and RA autoantibodies and also investigating subsequent referral patterns of patients to rheumatology practices.
Our study also has potential limitations to consider. First, we relied on EHRs and ICD-10 diagnosis codes for determining indications for testing. In some instances, these administrative codes may misclassify the true indications for testing, and we were not able to specifically review for the presence or absence of signs or symptoms that may raise suspicion for HLA-B27–associated diseases. Second, we performed our analysis in a single healthcare system with relatively limited racial diversity. This may limit the generalizability of our findings. Third, there is the possibility that we missed rheumatology referrals outside of our healthcare system. Finally, despite the size of our healthcare system, high volume providers who frequently order HLA-B27 testing may bias our results. For example, the frequent use of HLA-B27 for the evaluation of temporomandibular joint disease was an unexpected finding in our cohort, which may reflect local practice patterns and not be representative of testing trends in other healthcare systems and settings.
In conclusion, we found that HLA-B27 testing is frequently performed by nonrheumatologists in our healthcare system. Further, HLA-B27 testing is frequently performed for symptoms and indications that may have a low pretest probability of HLA-B27–associated diseases. The cotesting of HLA-B27 with ANA and RA autoantibodies was common, suggesting the indiscriminate use of HLA-B27 as a part of a screening panel for inflammatory rheumatic diseases. Finally, although rheumatology referral was higher among patients testing positive for HLA-B27, there may be an opportunity to increase rheumatology referrals among patients who have the HLA-B27 allele and additional appropriate symptoms.
Footnotes
G.C. McDermott and M. Choufani contributed equally as co-first authors.
CONTRIBUTIONS
GCM: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, writing – original draft, review and editing. MC: data curation, investigation, visualization, writing – original draft, review and editing. AM: data curation, investigation, writing – review and editing. JE: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, writing – review and editing.
FUNDING
GCM is supported by a Rheumatology Research Foundation Scientist Development Award and a Spondyloarthritis Research and Treatment Network Mentored Fellowship Award. JE is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (R21 AR076040-01).
COMPETING INTERESTS
The authors declare no conflicts of interest relevant to this article.
ETHICS AND PATIENT CONSENT
This study was approved by the Mass General Brigham institutional review board (#2021P002653). Informed consent was waived because of the retrospective nature of the research.
- Accepted for publication May 14, 2025.
- Copyright © 2025 by the Journal of Rheumatology
REFERENCES
SUPPLEMENTARY DATA
Supplementary material accompanies the online version of this article.
