Cohort 1 was a prospective analysis of adults (N = 740) considered to have a high suspicion for CD, recruited from a single CD subspecialty clinic in the United Kingdom. Patients with a previous diagnosis of CD, those adhering to a gluten-free diet, and those with IgA deficiency were excluded. Study patients had tTG-IgA titers drawn and, within 6 weeks, underwent endoscopy with ≥ 1 biopsy from the duodenal bulb and/or the second part of the duodenum. The PPV of tTG-IgA titers ≥ 10 times the ULN in patients with biopsy-proven CD was 98.7% (95% CI, 97%-99.4%).
Cohort 2 was a retrospective analysis of adult patients (N = 532) considered to have low suspicion for CD. These patients were referred for endoscopy for generalized GI complaints in the same hospital as Cohort 1, but not the subspecialty clinic. Exclusion criteria and timing of IgA titers and endoscopy were identical to those of Cohort 1. The PPV of tTG-IgA titers ≥ 10 times the ULN in patients with biopsy-proven CD was 100%.
Cohort 3 (which included patients in 8 countries) was a retrospective analysis of the performance of multiple assays to enhance the validity of this approach in a wide range of settings. Adult patients (N = 145) with tTG-IgA serology positive for celiac who then underwent endoscopy with 4 to 6 duodenal biopsy samples were included in this analysis. Eleven distinct laboratories performed the tTG-IgA assay. The PPV of tTG-IgA titers ≥ 10 times the ULN in patients with biopsy-proven CD was 95.2% (95% CI, 84.6%-98.6%).
In total, this study included 1417 adult patients; 431 (30%) had tTG-IgA titers ≥ 10 times the ULN. Of those patients, 424 (98%) had histopathologic findings on duodenal biopsy consistent with CD.
Of note, there was no standardization as to the assays used for the tTG-IgA titers: Cohort 1 used 2 different manufacturers’ assays, Cohort 2 used 1 assay, and Cohort 3 used 5 assays. Regardless, the “≥ 10 times the ULN” calculation was based on each manufacturer’s published assay ranges. The lack of assay standardization did create variance in false-positive rates, however: Across all 3 cohorts, the false-positive rate for trusting the “≥ 10 times the ULN” threshold as the sole marker for CD in adults increased from 1% (Cohorts 1 and 2) to 5% (all 3 cohorts).
Continue to: WHAT'S NEW
