Current issues of ACP Journal Club are published in Annals of Internal Medicine


Diagnosis

Bronchoalveolar cryptococcal antigen was accurate for diagnosing cryptococcal pneumonia in immunocompromised patients

ACP J Club. 1992 Nov-Dec;117:85. doi:10.7326/ACPJC-1992-117-3-085


Source Citation

Baughman RP, Rhodes JC, Dohn MN, Henderson H, Frame PT. Detection of cryptococcal antigen in bronchoalveolar lavage fluid: a prospective study of diagnostic utility. Am Rev Respir Dis. 1992 May;145:1226-9.


Abstract

Objective

To assess the accuracy of measuring cryptococcal antigen in the bronchoalveolar lavage (BAL) fluid in diagnosing Cryptococcus neoformans pneumonia in immunocompromised patients.

Design

Independent evaluation of the BAL fluid for cryptococcal antigen and of specimens from bronchoscopy and lung biopsy for C. neoformans.

Setting

Pulmonary service department in a university medical center in the United States.

Patients

224 immunocompromised patients (188 patients with AIDS, 16 with solid organ transplant, 11 with malignancy on chemotherapy, and 9 without malignancy who were receiving high-dose corticosteroids or cytotoxic agents) having bronchoscopy and BAL for respiratory symptoms and fever. The lavage specimens from 4 patients were not tested because they were insufficient or lost.

Description of test and diagnostic standard

BAL was done in the area of the infiltrate or in the right middle lobe of patients with diffuse infiltrates or normal roentgenograms. An aliquot of the cell-free fluid was tested for cryptococcal antigen using a latex agglutination system (CALAS; Meridian Diagnostics, Norwich, Ohio). The amount of agglutination was read on a 5-point scale (0 to 4+). The highest dilution that caused at least 2+ agglutination was considered positive.

The standard for diagnosis was the detection of a pulmonary infiltrate and identification of C. neoformans by either culture or cytologic examination of a pulmonary specimen obtained by bronchoscopy or open lung biopsy. Patients were followed for at least 3 months after the bronchoscopy for any evidence of C. neoformans infection.

Main outcome measure

Diagnosis of cryptococcal pneumonia.

Main results

Of the 220 lavage specimens, 8 were from patients diagnosed with cryptococcal pneumonia. All 8 were positive for cryptococcal antigen in their BAL fluid, with titers from 1:8 to 1:2560. 7 patients were positive for cryptococcal antigen who had no evidence of C. neoformans infection. 4 of these 7 patients had cryptococcal antigen titers of 1:8; none was higher. Using a cut point of ≥ 1:8 for antigen titer, sensitivity was 100%, specificity, 98%, positive predictive value, 67%, negative predictive value, 100%, {and likelihood ratio of a positive test of 50 and a negative test of 0.0}.*

Conclusion

The measurement of cryptococcal antigen in the bronchoalveolar lavage fluid was accurate for diagnosing cryptococcal pneumonia in immunocompromised patients with pneumonia.

Source of funding: In part, National Institutes of Health.

For article reprint: Dr. R.P. Baughman, 231 Bethesda Avenue, M.L. 564, Cincinnati, OH 45267-0564 , USA.

*Numbers calculated from data in article.


Commentary

The study by Baughman and colleagues comprises 2 parts. The first part was a retrospective analysis of 51 patients, the results of which yielded 100% sensitivity and 100% specificity. The second part was a prospective study of 220 patients. 15 were positive for cryptococcal antigen in their BAL fluid, but only 8 of them had culture-proven cryptococcal pneumonia. On the basis of a 3.6% prevalence of cryptococcal infection in this group of patients, a negative predictive value of 100% was calculated. The authors also calculated a positive predictive value of 53% to 67% depending on the cut-point titer of antigen used.

Because of the 7 false-positive antigen tests obtained from the prospective study, all antigen-positive and 10 antigen-negative BAL specimens were retested after they were stored at -80 °C for at least 3 months. All culture-positive specimens were antigen positive, but none of the 7 initially false-positive specimens remained positive, thus providing a 100% sensitivity and 100% specificity. It is not clear, however, whether these tests were blinded. If the tests were not blinded, the results would be less convincing.

The criteria for the diagnosis of cryptococcal pneumonia appear to be adequate. The data from the prospective study are probably more reliable than those from the retrospective study. The sensitivity of the test, however, was based on only 8 patients with cryptococcal pneumonia. Nevertheless, the results suggest that testing for the presence of cryptococcal antigen in BAL fluid of patients with cryptococcal pneumonia is a useful adjunct for rapid diagnosis of cryptococcal pneumonia. Because the test is 100% sensitive, a negative antigen test excludes C. neoformans as the cause of a patient's respiratory illness. In a patient with a positive antigen test, particularly of low titer, the test should be repeated and the diagnosis should be confirmed by culture.

Tom D.Y. Chin, MD
University of Kansas Medical CenterKansas City, Kansas, USA

Tom D.Y. Chin, MD
University of Kansas Medical Center
Kansas City, Kansas, USA