What does it mean if a patient is positive for anti-Jo-1 in routine hospital practice? A retrospective nested case-control study

Abstract

Background:  It is widely believed that patients bearing auto-antibodies to histidyl tRNA synthetase (anti-Jo-1) very likely have a connective tissue disease including myositis and interstitial lung disease.  The value of positive tests in low disease prevalence settings such as those tested in routine care is unknown.  We sought to determine the value of anti-Jo-1 auto-antibodies in routine practice. Methods: Our study was a nested case control study within a retrospective cohort of all patients tested for anti-ENA our hospital, from any hospital department, between January 2013 and December 2014.  Data was extracted from electronic records of anti-Jo-1 positive patients and randomly selected ENA negative patients (ratio of 1:2), allowing for a minimum follow up of at least 12 months after first testing. Results: 4009 samples (3581 patients) were tested.  Anti-ENA was positive in 616 (17.2%) patients, 40 (1.1%) were anti-Jo-1 positive. Repeat ENA testing was done for 350/3581 (9.8%) patients (428 of 4009 (10.7%) samples) and in 7/40 (17.5%) of anti-Jo-1 positive patients. The median interval between the first and second request was 124 days (inter-quartile range 233 days).  The frequencies of interstitial lung disease (ILD), myositis and Raynaud’s were comparable for anti-Jo-1 positive patients (n=40) and 80 randomly selected ENA negative controls.  Positive tests led to additional diagnostic testing in the absence of clinical disease.  Sensitivity of Jo-1 for ILD was 50% (CI 19-81%), specificity 68% (CI 59-77%), positive predictive value 12.5% (CI 4 to 27%) and negative predictive value 93.8% (CI 86-98%). Of 10 (25%) patients with high anti-Jo1 levels, 3 had ILD, one myositis and two a malignancy (disseminated melanoma and CML).  Conclusion: Anti-Jo-1 is uncommon in a heterogenous hospital population and is only weakly predictive for ILD.  Repeated test requests were common and potentially unnecessary indicating that controls over repeat requests could yield significant cost savings.