PURPOSE: Cytomegalovirus (CMV) can result in significant morbidity and mortality in lung transplantation patients. Clinically significant CMV disease may occur with low viral loads (< log 2 = 100 viral copies). Polymerase chain reaction (PCR) amplification can detect low levels of CMV DNA. PCR that identifies low viral loads, however, may not be reported, as these results may be considered clinically insignificant. We assert that PCR at these low viral loads is both sensitive and specific for the diagnosis of clinically significant CMV disease, and reporting of these results identifies early CMV disease and expedites therapy.
METHODS: 86 lung transplant recipients underwent 187 bronchoscopic procedures with transbronchial biopsy and simultaneous testing for CMV. Patients undergoing surveillance CMV PCR testing without concomitant bronchoscopy were excluded from analysis. Immunoperoxidase staining of lung tissue, shell vial cultures of bronchoalveolar lavage (BAL) fluid, and PCR of whole blood were tested for CMV. CMV disease was defined as either positive lung tissue staining or positive BAL viral culture with simultaneous symptoms (defined as a 10% reduction in forced expiratory volume in one second (FEV1), new radiographic infiltrate, fevers or leukopenia). Sensitivity and specificity of PCR for CMV disease was calculated at 0, 1.0, 1.5 and 2 log copies. The PCR cut-off value was defined as the log copy below which positive results were not reported.
RESULTS: 9/86 (9.3%) patients had documented CMV disease. Sensitivity of PCR for CMV disease was 78 and 89% at the PCR cut-off values of log 1.5 and log 1.0, respectively, while the sensitivity of PCR was reduced to 55% at a PCR cut-off value of log 2.0. Specificity of PCR for CMV disease was 93% or greater at all PCR cut-off values.
CONCLUSIONS: PCR is sensitive and specific for CMV disease at low viral loads (< log 2).
CLINICAL IMPLICATIONS: Reporting of PCR at low viral loads identifies patients with early CMV disease and can expedite therapy.
DISCLOSURE: The following authors have nothing to disclose: James Gagermeier, Mark Landmeier, Zhihong Hu, Farhan Ahmed, Robert Love
No Product/Research Disclosure Information