Is real-time polymerase chain reaction ready for real use in detecting candidemia?
J. Bennett
DOI: https://doi.org/10.1086/528692
IF: 20.999
2008-03-15
Clinical Infectious Diseases
Abstract:Received 27 November 2007; accepted 28 November 2007; electronically published 7 February 2008. Reprints or correspondence: Dr. John Bennett, Clinical Mycology Section, NIAID, Clinical Center room 11N234, National Institutes of Health, Bethesda, MD 20892 (jbennett@niaid.nih.gov). Clinical Infectious Diseases 2008; 46:897–8 This article is in the public domain, and no copyright is claimed. 1058-4838/2008/4606-0018$15.00 DOI: 10.1086/528692 In this issue of Clinical Infectious Diseases, McMullan et al. [1] present a carefully crafted and cautiously interpreted study of real-time PCR for detection of Candida species in serum samples obtained from nonneutropenic patients with candidemia. The test result was positive for 8 of 9 patients with Candida albicans in their routine blood culture and for both patients with Candida glabrata infection. The test detected only the C. albicans in a patient with mixed fungemia due to both C. albicans and C. glabrata. The test was not designed to detect Candida famata and missed the single case of fungemia due to that rare species. Remarkably, there were no false-positive results in 491 serum samples from 128 patients judged unlikely to have invasive candidiasis. Of the 13 serum samples from 6 patients who were considered to be at high risk of invasive candidiasis, the only positive results were for samples obtained from a patient who later developed candidemia. This provocative study raises several issues that are currently of high interest. McMullan et al. [1], like many others, believe that more rapid identification of candidemia could improve outcome. As logical as that seems, it remains controversial whether early therapy decreases overall mortality associated with this very morbid infection [2, 3]. Nevertheless, a rapid test for candidemia might decrease hospital stay or use of inappropriate treatments. In most hospitals, candidemia is now detected by an automated broth blood culture system. Culture medium is inoculated at the bedside and placed in the automated incubator soon after the specimen reaches the laboratory. Depending on inoculum, volume of blood cultured, Candida species, medium, and detection system, growth is detected in ∼24–48 h [4]. At that time, the clinician is notified that a yeast is growing. There are some clues as to species, such as the more rapid growth of C. glabrata in anaerobic than aerobic culture in at least one system [5]. C. glabrata and C. albicans can be identified within 3 h using a commercial peptide nucleic acid fluorescent in situ hybridization (PNA FISH) method [6]. For a firm identification of other yeast species, subculture and overnight incubation is needed for isolation of individual colonies. The laboratory can then select from a variety of techniques for identifying the yeast colony on subculture, with C. albicans being identified the same day and other species up to 3 days later. As with most PCR and phenotypic tests, the identification is as good as the database. Designation of the isolate as “non-albicans” is not predictive of fluconazole resistance, because most of these species are as azole-susceptible as C. albicans. Whether real-time PCR is faster than routine culture depends on how often the real-time PCR batch is run. The need for specially trained technicians, a dedicated space for PCR and another for real-time PCR, and the costs saved by running the test in batches would limit this to 1 or 2 weekdays each week in most laboratories. The real-time PCR used by McMullan et al. [1] began with manual extraction of DNA from serum, followed by a PCR, followed by a nested real-time PCR. Technician hands-on time is considerable and expensive. Another concern is the extraordinary precautions required of this procedure. Routine laboratories have avoided using nested PCR techniques because the opportunities for contamination are so great that, even with dedicated work areas and skilled employees, false-positive results are difficult to prevent. It is often unclear whether such false-positive results are truly false or are detecting levels of candidiasis below the limit of blood culture. A similar nested real-time PCR found positive results for Candida in 27 of 113