Optimisation of the polymerase chain reaction and dot-blot hybridisation for detecting cytomegalovirus DNA in urine: comparison with detection of early antigen fluorescent foci and culture

Rapid, sensitive and specific assays are required for the diagnosis of CMV infection following transplantation. We describe our experience in developing assays for detecting CMV in urine. Conventional preparation of probes cloned after amplification in E. coli led to contamination with E. coli nucleic acids; these hybridised to E. coli DNA present in urine and produced false positive results. Two CMV probes (Hind III and gL) hybridised to human DNA despite high stringency; these probes were thus unsuitable for detecting viral nucleic acids in clinical samples. A PCR derived probe from the immediate early gene of CMV detected dot-blotted CMV DNA specifically. Optimal preparation of urine for detection of CMV DNA was as follows; four freeze/thaw cycles and ultracentrifugation before in vitro proteinase K/SDS treatment, phenol:chloroform extraction, heat denaturation and direct application onto a nylon membrane. However, dot-blot hybridisation was a poor test for CMV in urine; it had low sensitivity and specificity compared with virus isolation and DEAFF. Single round PCR of a 293bp region of CMV DNA was sensitive and specific to CMV targets. However, undiluted urine contained PCR inhibitors that could only be partly removed by using PEG precipitation. PCR of CMV DNA from urine was specific but was insensitive compared to conventional culture and DEAFF. A significant proportion of urine samples were toxic in conventional culture and DEAFF tests but, PCP of CMV DNA from urine is insensitive and despite its specificity is unlikely to be advantageous in clinical practice even when DEAFF or culture prove unreliable. (C) 1998 Elsevier Science B.V. All rights reserved.