Optimization for detection of cytomegalovirus by the polymerase chain reaction (PCR) in clinical samples.

PCR is 100 times more sensitive than traditional tube culture for detecting cytomegalovirus (CMV) but may require up to 12 reactions per specimen (Sandin et al., 1991). In order to make the assay practical for use in a clinical laboratory the procedure used to detect CMV must be simplified. In this study, the effect of reducing the number of reactions per specimen on sensitivity and specificity of the PCR assay was evaluated. 53 residual samples from specimens processed for CMV by shell vial assay/routine tube tissue culture (SVA/TTC) were analyzed by PCR. The residual samples were separated into a supernatant and pellet fractions, then tested for CMV with primers to the immediate early (IEP) and late protein (LP) genes using a nested procedure. To exclude false negatives due to the presence of inhibitors in the sample fractions, all fractions were tested for the presence of the human myosin heavy chain gene also using a nested procedure. SVA/TTC had a sensitivity and specificity of 52/96% in comparison to PCR when data from all 12 PCR reactions was considered. However, high sensitivity and specificity were retained when only the data of the IEP primers with two samples were considered. The results from examining only the 1:10 dilution of pellet and the undiluted supernatant by PCR provided a 60% increase in sensitivity over SVA/TTC, high specificity and a clinically feasible assay.