Polymerase Chain Reaction (PCR) from Ficoll- Purified Polymorphonuclear Leukocytes for Monitoring Cytomegalovirus Infections in Renal Allograft Recipients: Superior Sensitivity and Similar Specificity Compared with Plasma PCR

To the Editor—Boivin et al. [1] compared the diagnostic value of quantitative polymerase chain reaction (PCR) using polymorphonuclear leukocytes (PMNL) with qualitative plasma PCR to monitor human immunodeficiency virus–infected individuals at risk for cytomegalovirus (CMV) disease. Symptomatic patients had significantly higher CMV DNA copies in both leukocytes and plasma than asymptomatic individuals, but plasma PCR was far less sensitive than leukocyte PCR due to significantly higher CMV DNA load in leukocytes. Although the specificity and positive predictive value were much lower for a positive leukocyte PCR versus plasma PCR result, this problem was overcome by defining a cutoff value for the CMV DNA load in leukocytes. The authors concluded from their data that quantitative leukocyte PCR was superior to plasma PCR. We would like to comment on three of the issues raised in the discussion. First, Boivin et al. [1] see a discrepancy between their own observations and those reported by others, who were unable to define a reliable cutoff value for leukocyte PCR. However, we previously showed for renal allograft recipients [2] that CMV DNA load in total peripheral blood leukocytes (PBL) of ≥10,000 copies of CMV/10 copies of b-globin DNA (the theoretical average DNA yield from ∼1 mL of whole blood) is highly indicative of clinically manifest CMV infection. This is almost the same as the cutoff of 16,000/mL described by Boivin et al. [1]. Furthermore, we observed similar sensitivity, specificity, and positive and negative predictive values (100%, 94%, 83%, and 100%, respectively). These similarities are all the more remarkable because we investigated a different patient clientele and also used a completely different quantification method based on competitive PCR [3]. Second, Boivin et al. [1] hint that reliable performance of quantitative PCR can be hazardous. In our experience with leukocyte PCR, the DNA extraction procedure is the most crucial step. We regard the parallel quantitation of cellular DNA sequences and CMV genomes as indispensable to assess the efficiency of DNA extraction, since we have observed high interassay variability of total DNA yield from leukocyte samples [3]. Finally, due to the technical demands of quantitative PCR procedures, we find it desirable to improve the diagnostic value of qualitative CMV PCR. Since it is generally accepted that leukocyte PCR is more sensitive and usually becomes positive earlier than plasma PCR, the challenge is to improve the inferior specificity of leukocyte PCR by selective detection of CMV DNA originating from active infection. CMV DNA is found predominantly in PMNL during active infection [4], whereas latently CMV-infected cells belong mostly to the peripheral blood mononuclear cell (PBMC) population [5]. PCR analysis of purified PMNL is therefore the most promising strategy. However, the dextran sedimentation method as used by Boivin et al. [1] and many others for isolating “PMNL” in reality yields leukocyte mixtures containing up to 30% PBMC. This might be a reason for the low specificity of leukocyte PCR. In contrast, using ficoll-metrizoate density centrifugation, which yields ∼95%-pure PMNL and PBMC fractions, we recently found [6] that with leukocyte samples of 35 CMV IgG–positive kidney transplant patients with no laboratory evidence for active CMV infection, only 6% of PMNL but 66% of PBMC fractions were PCR-positive. To assess the diagnostic feasibility of our approach, we performed qualitative PCR from total PBL, ficoll-purified PMNL, and PBMC, and plasma samples from 96 renal allograft recipients to diagnose (1) active CMV infections (n 5 ) as defined by pp65 antigenemia [7] or (2) symptomatic CMV 37 infections ( ) as defined by previously described criteria n 5 19 [8] (table 1). Active as well as symptomatic CMV infections were detected in the PMNL fraction with maximal sensitivity and similar specificity compared with plasma. In contrast, the vast majority of unwanted positive results in leukocytes was attributable to the PBMC. These results suggest that qualitative CMV PCR using ficoll-purified PMNL is a true alternative to plasma PCR and to quantitative leukocyte PCR for monitoring patients at risk for CMV infections.