The diagnosis of nocardiosis, a severe opportunistic infection, is challenging. We assessed the specificity and sensitivity of a 16S rRNA PCR-based assay performed on clinical samples. In this multicenter study (January 2014 to April 2015), patients who were admitted to three hospitals and had an underlying condition favoring nocardiosis, clinical and radiological signs consistent with nocardiosis, and a PCR assay result for a clinical sample were included. Patients were classified as negative control (NC) (negative culture results and proven alternative diagnosis or improvement at 6 months without anti- treatment), positive control (PC) (positive culture results), or probable nocardiosis (positive PCR results, negative culture results, and no alternative diagnosis). Sixty-eight patients were included; 47 were classified as NC, 8 as PC, and 13 as probable nocardiosis. PCR results were negative for 35/47 NC patients (74%). For the 12 NC patients with positive PCR results, the PCR assay had been performed with respiratory samples. These NC patients had chronic bronchopulmonary disease more frequently than did the NC patients with negative PCR results (8/12 patients [67%] versus 11/35 patients [31%]; = 0.044). PCR results were positive for 7/8 PC patients (88%). There were 13 cases of probable nocardiosis, diagnosed solely using the PCR results; 9 of those patients (69%) had lung involvement (consolidation or nodule). PCR testing had a specificity of 74% and a sensitivity of 88% for the diagnosis of nocardiosis. PCR testing may be helpful for the diagnosis of nocardiosis in immunocompromised patients but interpretation of PCR results from respiratory samples is difficult, because the PCR assay may also detect colonization.

NCT01520597.