Methods: CMV VL in BALF were quantified using the commercially available Cobas® AmpliPrep/Cobas TaqMan CMV Test (CAP/CTM CMV; Roche Molecular Systems, Inc., Branchburg, NJ), modified with sample processing procedures for BALF and calibrated against the WHO standard for CMV DNA. Subjects (n=40) were adult ICP who underwent BALF collection 2000-2014. Viral load results were normalized to the total nucleated cell counts (TNC) in individual BALF specimens. Optimal VL thresholds to distinguish CMVP and control groups (adults who underwent BALF collection for non-infectious causes, negative for CMV in lung biopsy) were determined by performing receiver operating characteristic (ROC) analysis on log-transformed VL values.
Results: Forty BALF specimens were tested: 17 with CMVP (6 were proven by biopsy) and 23 controls (2 cases with CMV viral shedding). Majority of patients (71% of CMVP; 83% of controls) were transplant recipients (75% lung transplants). The limit of detection was 95 IU/mL by probit analysis, while the lower limit of quantification (LLoQ) was 274 IU/mL. A mean VL difference of only 0.17 log10 IU/mL from expected results was determined by testing CMV standards spanning the reportable range of the assay, obviating the need of an assay calibration correction factor. Among the controls, median VL was 0 (range, 0 – 7,640 IU/mL), while among CMVP cases, the median was 656,000 IU/mL (range, 274 – 18,200,000 IU/mL). A VL of 32,400 IU/ml in BALF had an 82% (14/17) sensitivity and 100% (17/17) specificity for possible, probable or proven CMV pneumonia. By limiting the analysis to biopsy proven cases of CMV pneumonia, a VL of 656,000 IU/ml had a sensitivity and specificity of 100%. No change was noted when VL results were normalized to TNC in BALF.
Conclusion: A CMV VL of >32,000 IU/mL in BALF is suggestive of CMVP in ICP. Our findings should be validated prospectively and in a larger cohort to define the optimal VL threshold for diagnosis of CMVP in ICP.
B. Lahr, None
A. Limper, None
M. J. Binnicker, None
J. Yao, None
R. R. Razonable, None