The current American College of Rheumatology criteria for diagnosing giant cell arteritis have low sensitivity and specificity and should be abandoned, according to the authors of a study that analyzed the clinical utility of the guidelines in a retrospective chart review.
The current ACR criteria, published in 1990, "should not be used to determine the presence or absence of GCA," concluded Dr. Ann P. Murchison of the Wills Eye Institute, Philadelphia, and her coauthors. Instead, they recommended that a temporal artery biopsy at least 2 cm long be done for all patients suspected of having GCA, and all patients with a positive results "must be treated with systemic corticosteroids, even in the absence of other" ACR criteria (Am. J. Ophthalmol. 2012 July 20 [doi:10.1016/j.ajo.2012.03.045]).
They also recommended that all physicians who treat GCA should be educated about the importance of the temporal artery biopsy in diagnosing GCA, and that biopsy results – not the ACR criteria – "should be used as the only indicator of the presence or absence of disease in research regarding GCA."
The current ACR criteria, which the authors pointed out are frequently used to diagnose GCA, do not require a positive biopsy for diagnosis: The criteria state that the diagnosis of GCA can be made when patients meet three of five criteria: age older than 50 years; new onset of a localized headache; temporal artery tenderness or a decreased temporal artery pulse; an ESR (erythrocyte sedimentation rate) equal to or greater than 50 mm/hr; or a positive temporal biopsy.
The investigators reviewed the charts of 112 patients who had a temporal artery biopsy between October 2001 and 2006 at the Wills Eye Institute as part of an evaluation for possible GCA, and looked for the presence or absence of the ACR GCA criteria, the results of biopsies, and the progression of vision loss after the cases were diagnosed.
Of the 112 patients, 74 met their criteria for inclusion in the analysis. (Exclusion criteria included having been treated with corticosteroids before presentation and having a biopsy specimen that was shorter than 2 cm in length.) Of those 74 patients, 35 had a positive biopsy, and 39 had a negative biopsy.
Of the 35 patients with a positive biopsy, 9 (25.7%) met only two of the ACR criteria (age older than 50 years and a positive biopsy), and so would not have been diagnosed if the ACR criteria alone had been used to make the diagnosis, despite the positive biopsy. Based on the criteria, these patients would not have been biopsied or treated "and would have been at higher risk for further catastrophic visual loss," the authors wrote.
In the biopsy-positive group, 16 (4.7%) met two criteria before having a biopsy and then required a biopsy for the diagnosis. Another 10 (28.5%) met three or four criteria before having a biopsy.
Of the 39 patients with a negative biopsy, 11 (28.2%) met three or four of the ACR criteria. In these patients, if the criteria alone (without the biopsy) had been used to make the diagnosis, they would have been treated with corticosteroids, which would have exposed them "to potential adverse events from chronic corticosteroid administration," the investigators said. Another 17 (43.6%) met two criteria and had a biopsy to rule out GCA. And 11 (28.2%) met only one of the criteria, so the biopsy would not have changed the diagnosis, based on the ACR criteria, they noted.
Based on their calculations, the sensitivity of the ACR criteria was 74.3% and specificity was 71.8%. Sensitivity dropped to 28.6% when the biopsy results were not included. These calculations, they pointed out, are significantly lower than the sensitivity (93.5%) and specificity (91.2%) cited by the ACR in 1990.
The authors noted that questions regarding the ability of the ACR criteria to reliably diagnose GCA are not new, and cited studies finding that "none of the criteria alone or in combination are as sensitive or specific as a temporal artery biopsy." However, they added, "temporal artery biopsy still is not considered essential" in diagnosing GCA by some physicians.
The authors of the study had no potential conflicts of interest to disclose.