Genitourin Med 1992;68:183-188

Diagnosis of cytomegalovirus infection: a review

Deenan Pillay, Paul D Griffiths

IntroductionCytomegalovirus (CMV) infects between 40-100% of adults throughout the world, depend-ing primarily on socio-economic status.However, other relevent factors include thedegree and form of sexual activity such that, inthe UK, homosexual men have a sero-prevalence of 80% compared with 42% ofsexually active male heterosexuals.' Theseindividuals may present to genitourinarymedicine clinics with CMV-related symptoms,most of whom will be infected with HIV.As a member of the herpes group of viruses,

infection with CMV leads to a state of latencywith the capacity to reactivate at any time.Primary infection in the immunocompetent isusually asymptomatic but may be associatedwith a mononucleosis-type illness. Reactiva-tion is of major consequence in those who areimmunocompromised in whom it may beassociated with retinitis, colitis, pneumonitis orencephalitis, all of which carry significantmorbidity.2The availability of effective anti-CMV

therapy has heightened the requirement forrapid, sensitive methods of CMV detection,and these will be discussed in detail. By con-trast, serological diagnosis of CMV infection(ie antibody determination) depends on afunctional immune system and, we believe,plays little role in CMV diagnosis in theimmunocompromised individual.

Despite the severe symptomatologyassociated with CMV, it is important to realisethat infection does not equate with disease.Thus, virological evidence of CMV infectionmust always be interpreted within the clinicalcontext before treatment is considered.

Division ofCommun-icable Diseases, RoyalFree Hospital andSchool of Medicine,Rowland Hill Street,London NW3 2PF, UKD PillayP D GriffithsAddress for correspondence:Dr Deenan Pillay.Accepted for publication19 November 1991

Virus isolationThe detection ofCMV in clinical specimens bygrowth in conventional cell culture (CCC)remains the "gold standard" against which any

new detection system must be compared.Human CMV is highly species specific andthus will only grow in human fibroblasts, a cellline commonly employed in clinical virologylaboratories.' The characteristic CMVcytopathic effect (CPE) consists of small roundor elongated foci of rounded, enlarged refrac-tile cells (fig 1), which usually develop 2-3weeks following inoculation, but may take up to6 weeks. The clinical specimens most com-

monly inoculated are urine, saliva (or throatswab taken into viral transport medium) andthe "buffy coat" component of whole bloodwhich has been taken into preservative-freeheparin. However,CMV can be isolated from a

wide range of tissues, and biopsy samples taken

into viral transport medium and subsequentlyhomogenised prior to inoculation are especiallyvaluable for the diagnosis of invasive disease.Rapid transport of samples from ward tolaboratory is essential since CMV is a labilevirus. If this is not possible the sample shouldbe stored at 4'C and never frozen.

Rapid culture systemsThe major disadvantage of CCC relates to thetime lapse between inoculation of clinicalspecimen and appearance of the characteristicCPE. In addition, this time period allows forpotential contamination of cultures, and theovergrowth of coexisting viruses which may actto hide CMV. Attempts to develop more rapidmethods have been stimulated by the develop-ment ofmonoclonal antibodies directed againstCMV proteins. Following viral infection of apermissive cell, CMV immediate early andearly proteins are produced within the first 8hours, with viral DNA replication occurring upto 24 hours later, followed by late proteinsynthesis.2 The ability to detect early proteinsby fluorescent tagged monoclonal antibodieswas therefore utilised by workers in ourlaboratory to develop a rapid diagnostic assay.This technique, DEAFF (detection of earlyantigen fluorescent foci) is simple to performand can be undertaken on a routine, dailybasis.4 Clinical specimens are inoculated intofibroblast monolayers on eight-well slides.Following 16 hours incubation the cells areacetone fixed, and CMV protein detected by animmediate early protein monoclonal antibody,to which a fluorescence-conjugated anti-species antibody can be bound. It is essentialthat this latter antibody is non-human, in orderto avoid binding to cellular Fc receptors whichare induced by CMV infection. A positiveresult can be detected by bright nuclear stain-ing under the fluorescent microscope (fig 2). Acomparison of this technique with conven-tional cell culture (CCC) has shown the sen-sitivity of DEAFF to be 78% compared with76% for CCC, taking a positive by eithermethod as a true positive. It is likely thatDEAF does not detect very low titres ofvirus inclinical samples, which take a longer thanaverage time to be detected in CCC.' On theother hand it is intriguing to speculate whetherDEAFF, but not CCC, detects virus in thepresence of antiviral drugs, since CMV earlyantigen production is not dependent on priorviral genome replication, a target of thesedrugs. Conversely, drug carry over from aclinical specimen would inhibit viral growth inCCC such that no CPE would be observed.A potential problem associated with any

183

on April 18, 2020 by guest. P

rotected by copyright.http://sti.bm

j.com/

Genitourin M

ed: first published as 10.1136/sti.68.3.183 on 1 June 1992. Dow

nloaded from

Pillay, Griffiths

Fig 1 The typicalcytopathic effect ofcytomegalovirus seen in ahuman embryo lungfibroblast monolayer.

.;,

} 9

*.p.

IPfg

4.., ~~~~~~~~~~~~~~~~~~~~~~~~~~44~~~~~...- A '~~~~~~~~~~~~~~~-IA

.b 0Wst;:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..

.2..... . *.

.......-Ag

Fig 2 Cytomegalovirusinfection diagnosed bydetection of early antigenfluorescent foci(DEAFF).

assay system relying on monoclonal antibodiesis that of antigenic variation, such that therespective antigen in a variety of CMV strainsmay not be detected. Until a single monoclonalantibody can be produced which is shownto react with a group-specific antigen, it isgenerally assumed that the use of a pool ofmonoclonal antibodies in this technique ispreferable.A similar technique to the DEAFF test has

been described in which inoculation of clinicalspecimen and the subsequent fluorescent stain-ing are carried out on a cover slip, held eitherwithin a shell vial6 or on a slide support.7 While

this method may be more labour intensive thanDEAFF, it does allow for centrifugation of theinoculated cell monolayer which can increasethe sensitivity of viral detection.8 "Centrifugalenhancement" is a relatively specific propertyof CMV cultures, and increase in infectivity isgreatest at low (1000 g) rather than high speedcentrifugation. The precise mechanism of cen-trifugal enhancement remains unclear.9

Antigen detectionDirect immunofluorescent techniques can alsobe used on fixed tissue sections, or cytospin

184

on April 18, 2020 by guest. P

rotected by copyright.http://sti.bm

j.com/

Genitourin M

ed: first published as 10.1136/sti.68.3.183 on 1 June 1992. Dow

nloaded from

Diagnosis of cytomegalovirus infection: a review

preparations of cell suspensions in order todetect CMV antigens." The latter methodhas been evaluated for bronchoalveolar lavagematerial, the main advantage over conventionalcell culture and DEAFF being that of speed ofprocessing. However, the sensitivity for CMVdetection from such fluid is low and thismethod alone cannot therefore be recommen-ded for diagnosis of lung infection from lavagefluid.1` On the other hand, immunofluores-cence on frozen sections as well as on formalinfixed and deparaflinised tissue sections, such asliver, brain or lung may be an importantadjunct to a virological or histological diag-nosis.

Immunofluorescent detection ofCMV earlyantigens in blood polymorphonuclear leuco-cytes (PMNC) has been used to assess viraemiain a wide range of immunocompromisedpatients. This requires the preparation ofcytocentrifuged slides to contain around 50,000cells per preparation. The sensitivity of thistechnique appears to be at least equivalent torapid culture techniques with the advantagethat results can be obtained within 5 hours ofsampling.141l A further benefit is that antigenstaining cells can be counted relative to totalfixed PMNC, allowing a semi-quantitativeassessment of CMV viraemia.'6 As discussedabove, the specificity of the monoclonalantibody used in any immuno-assay is impor-tant, and the precise antigen detected in manyof the published studies of this technique hasyet to be fully clarified, as it appears that somerecognise an early protein processed byleucocytes rather than the major immediateearly protein.The above methods relate to detection of

cellular CMV since the virus is very cell-associated, and therefore all involve somepreparative work on the clinical specimen.However, cell-free CMV is excreted in largeamounts in urine and some groups have attem-pted to develop simple enzyme immunoassaysfor urinary detection of virus. This has provedto be problematic, one reason being that suchvirus is coated the host protein B2-microglobulin.`' A consequence of this wouldbe that antigenic determinants are masked,making recognition by antibody difficult. Weare currently investigating methods by whichantibodies directed against both CMV antigensand B2-microglobulin could be incorporatedinto a simple assay system for urinary CMVdetection.

Histology and cytologyHistological diagnosis of CMV infection infixed tissue sections is usually made by detec-tion of intranuclear inclusions, surrounded byan "owl's eye" halo.2 Although such aprocedure lacks sensitivity, it is one ofthe waysof diagnosing invasive CMV disease. If CMVinfection is suspected, immunocytochemicaldetection of viral antigens should also beundertaken on these sections. Inclusion bear-ing cells can also be found in saliva, urine,bronchoalveolar lavage fluid and touchpreparations of CMV infected tissues.18

However, sensitivity of cytological diagnosis issimilar to that for histological diagnosis.

DNA detectionViral genome detection has, until recently,depended on the availability of specific CMVoligonucleotide probes which do not cross reactwith non viral gene sequences. This is ofparticular relevance to the detection of CMVDNA since several regions ofthe CMV genomecontain sequences ofhomology with the humangenome,19 and therefore a prerequisite for theuse of this methodology is the use of probesunique to CMV. These probes are usuallyradioactively labelled to allow detection oftarget sequences by autoradiography. Atpresent few diagnostic laboratories havefacilities for nucleic acid detection as a routineservice; however, this situation is likely tochange in the next few years, especially with thedevelopment of non-isotopic methods for gen-ome detection.20

In situ hybridisation has been shown todetect CMV DNA or RNA in tissue specimensections, even in cells that do not appear to beinfected by standard histological criteria.2122However, in comparison with conventional andrapid cell culture, it was shown to be the leastsensitive for the detection of CMV DNA inbronchoalveolar lavage samples.23 The dot-blotmethod of hybridisation, requiring an initialextraction ofnucleic acid from clinical samples,has been reported to be a more reliable methodfor CMV detection, especially from urine.2425The above techniques have now been super-

seded by the polymerase chain reaction (PCR).This is a method for in vitro amplification ofspecific gene sequences prior to detection. Itrequires the presence of specific primers com-plementary toDNA sequences on either side ofa target CMVDNA segment ofknown size. Byrepeated cycles of heating and cooling, in thepresence of a heat stable DNA polymerase, asingle gene copy may be amplified up to onemillion fold. This may subsequently be detec-ted following electrophoresis of the PCR mix-ture in the presence ofa nucleic acid stain, sincethe gene is of known molecular size; however,confirmation may be made by subsequenthybridisation with a labelled oligonucleotideprobe.20 It follows that the major benefit ofPCR is its high sensitivity for CMV detection.Conversely, PCR is highly susceptible to con-tamination by extraneous viral material, aproblem of particular concern to a diagnosticvirology laboratory in which CMV is com-monly grown to a high titre. The improvedsensitivity ofPCR compared to other methodsof CMV diagnosis has been documented forviral detection in blood and urine samples, inHIV infected as well as other immunocom-promised patients.2129 However, a possiblelimitation of this high degree of sensitivity isbased on the knowledge that normal viralreplication gives rise to a large proportion ofdefective viral particles, unable to take part infurther productive infection. It is likely thatsuch genomes would be detected by a PCRtechnique, despite the fact that their presence

185

on April 18, 2020 by guest. P

rotected by copyright.http://sti.bm

j.com/

Genitourin M

ed: first published as 10.1136/sti.68.3.183 on 1 June 1992. Dow

nloaded from

Pillay, Griffiths

does not necessarily denote active viral infec-tion, nor constitute evidence of disease. Formalassessment of prognostic value of any givenPCR method is therefore mandatory. Alter-natively, a discriminative reporting of PCRresults may be of more use than merely apositive or negative result. Of interest is therecent development of a quantitative PCRmethod for detection of CMV in clinical sam-ples,'0 which has been used to show that a highCMV genome copy number in urine is morelikely to be associated with CMV disease than alow copy number (personal communication;JC Fox, PD Griffiths and VE Emery).

SerologySerological diagnosis of viral infection nor-mally depends on the detection of a seroconver-sion, a four fold or greater increase in titre ofantibody, or the presence of specific circulatingIgM, and sensitive methods are now widelyavailable for the detection of CMV IgG andIgM. The immunocompromised patient,however, is characterised by an inability tomount normal immune responses, especially toa reactivation of latent infection, as is often thecase with CMV.`"3 The disadvantages ofserological methodology over virus detectionare therefore lack of sensitivity, and delay indiagnosis, since the serological changes whichdo occur in response to infection are often latein onset.'2 The measurement of local specificantibody production, such as has been carriedout in bronchoalveolar lavage fluid from trans-plant recipients with CMV pneumonitis, hasalso been shown to be ofno diagnostic benefit.'2In HIV infected individuals the most usefulrole of serology is the determination of CMVIgG status, which allows categorisation ofpatients into those who are liable to CMVreactivation and those who are not.

Diagnosis ofCMV diseaseAs discussed above CMV infection does notequate with CMV disease. A diagnosis ofdisease depends on a combination ofvirologicaland clinical data together with an assessment ofthe individual's risk factors, such as the degreeof this immunocompromise. Close cooperationbetween the clinician and virologist is thereforeessential. Some important aspects in the diag-nosis of CMV related disease in HIV infectedindividuals are discussed below.

Gastrointestinal diseaseAlthough almost all areas ofthe gastrointestinaltract may be infected with CMV, the mostcommon manifestation is diarrhoea due toCMV colitis. Since many other opportunisticand non-opportunistic infections may presentin an identical manner, a biopsy diagnosis isusually required before onset oftreatment. Theascending colon has previously been thought tobe the most common site of CMV infection";however, other workers have demonstratedrectum and sigmoid colon involvement as well,with a relative sparing of the transverse colon."'It is essential that investigations include a

colonscopy with multiple biopsies taken fromascending, transverse and the rectosigmoidregion. Symptoms from other areas of the GItract also require an analysis of relevant biopsysamples for a diagnosis ofCMV infection.

Central nervous system diseaseEncephalitis and myelitis have been attributedto CMV infection.'536 However, although his-tological involvement of the brain suggestive ofCMV infection frequently is noted in postmortem material from AIDS patients,35 andCMV antigens and nucleic acid is detected innervous tissue, it remains difficult to ascribedisease to CMV infection.37 This problem iscompounded by the known neurotropism ofHIV itself. In clinical practice, a brain biopsy isoften undesirable, and in those with otherwiseunexplained neurological symptoms we sug-gest the sampling of CSF, whole blood andother peripheral sites for CMV detectionwhich, if positive, may support a clinical diag-nosis ofCMV disease.

RetinitisThe diagnosis of retinitis is based solely onophthalmological examination following theonset of symptoms.'8 The eyes of AIDSpatients must therefore be examined regularly,in order that treatment can be initiated promp-tly. These individuals often require long termmaintenance therapy. Although retinitis isoften suffered by those with disseminatedCMV disease, the detection ofCMV elsewhereshould have no bearing on the diagnosis ofCMV retinitis.

PneumonitisCMV pneumonitis is a common and severeproblem following organ transplantationrequiring urgent treatment and is diagnosed bydetection of CMV in bronchoalveolar lavage(BAL) fluid.'9 In contrast, although CMV isoften isolated from lavage fluid of AIDSpatients with respiratory signs and symptoms,either alone or in association with other infec-tious agents, it rarely causes disease.' It isthought that CMV pneumonitis is mainlymediated by immuno-pathological mechan-isms, rather than the viral lytic infection whichcauses damage in other tissues.4' As such, thedevelopment of pneumonitis depends to someextent on a functional immune response toCMV. AIDS patients commonly have very lowCD4 counts when CMV is detected, and maynot be able to mount such a response. Wetherefore recommend that BAL samples aresent for CMV detection in these patients butthat a diagnosis of CMV pneumonitis only bemade following the failure of treatment forother suspected opportunistic chest infections,and where the CD4 count is greater than 100 x106/1.42

Prognostic value ofCMV detectionThe severe nature of CMV infection in trans-plant recipients has led to attempts to definerisk factors for disease. One approach has beenfor regular surveillance cultures, from multiple

186 on A

pril 18, 2020 by guest. Protected by copyright.

http://sti.bmj.com

/G

enitourin Med: first published as 10.1136/sti.68.3.183 on 1 June 1992. D

ownloaded from

Diagnosis of cytomegalovirus infection: a review

sites, to be undertaken for CMV detectionduring the high risk post-transplant period.Such studies have shown that CMV isolatedfrom blood, but not urine or saliva, is associatedwith an increased risk of subsequent CMVdisease.4 ' CMV viraemic individuals maythen be given pre-emptive therapy prior toonset of symptoms, or be closely observed suchthat treatment can be initiated early if diseasedevelops. A similar prospective study carriedout in 71 AIDS patients showed that 50% ofthose with CMV viraemia later developedorgan disease compared with 9% of non-viraemic patients, over a mean follow-upperiod of 16 months.45 Further studies are nowrequired in order to determine if anti-CMVpre-emptive therapy can be allocated to AIDSpatients on a similar basis (in those who areCMV viraemic) thereby preventing a majorcause of morbidity and mortality in this groupof individuals.

ConclusionCMV related pathology in AIDS patients isbeing observed to an increasing extent, bothbecause of the growth of this population, andthe fact that they are surviving for a longerperiod. With the availability of effective antiCMV treatments, rapid methods of CMVdiagnosis are required in order that earlyinitiation of such treatment can occur. Conven-tional cell culture remains the gold standard ofdiagnosis, however rapid culture techniques,and antigen detection have been shown to beboth sensitive and specific and are widely usedbecause of their rapidity. Of the array ofmethods available for detecting CMV genomesin clinical specimens, the polymerase chainreaction is most likely to find its way intodiagnostic use. Its major advantage is sen-sitivity; however, it is also quick, and need notrequire the use of radioactivity. A full compar-ison of PCR with other established methods ofCMV diagnosis in a clinical setting is awaited.

In conjunction with sensitive detection sys-tems, another potential development for thefuture is the introduction of prospective diag-nosis, or surveillance for CMV infection inAIDS patients. This would allow for treatmentoptions to be considered prior to the onset ofsymptomatology, and lead to a further enhan-cement in the quality of life of theseindividuals.

1 Berry NJ, MacDonald Burns D, Wannamethee G, et al.Seroepidemiologic studies on the acquisition of antibodiesto cytomegalovirus, herpes simplex virus, and humanimmunodeficiency virus among general hospital patientsand those attending a clinic for sexually transmitteddiseases. J Med Virol 1988;24:385-93.

2 Griffiths PD. Cytomegalovirus. In: Zuckerman AJ, Ban-atvala JE, Pattison JR, eds. Principles and Practice ofClinical Virology, 2nd ed. 1990;69-102, London, Wiley.

3 Stagno S, Britt WJ, Pass RF. Cytomegalovirus. In: SchmidtNJ, Emmons RW. eds. Diagnostic Procedures for Viral,Rickettsial and Chlamydial Infections, 6th edition,1989;321-78, Washington, American Publich HealthAssociation.

4 Griffiths PD, Panjwani DD, Stirk PR, et al. Rapid diagnosisof cytomegalovirus infection in immunocompromisedpatients by detection of early antigenic fluorescent foci.Lancet 1984;ii:1242-5.

5 Stork PR, Griffiths PD. Useofmonoclonal antibodies for thediagnosis of cytomegalovirus infection by the detection ofearly antigenic fluorescent foci (DEAFF) in cell culture.JMed Virol 1987;21:329-37.

6 Paya CV, Smith TF, Ludwig J, Hermans PE. Rapid shellvial culture and tissue histology compared with serologyfor the rapid diagnosis of cytomegalovirus infection inliver transplantation. Mayo Clin Proc 1989;64:670-5.

7 Thiele GM, Bicak MS, Young A, Kinsey J, White RJ,Purtilo DT. Rapid detection ofcytomegalovirus by tissueculture, centrifugation, and immunofluorescence with amonoclonal antibody to an early nuclear antigen. J VirolMethods 1987;16:327-38.

8 Gleaves CA, Smith TF, Shuster EA, Pearson GR. Rapiddetection of cytomegalovirus in MRC-5 cells inoculatedwith urine specimens by using low speed centrifugationand monoclonal antibody to an early antigen. J ClinMicrobiol 1984;19:917-9.

9 Hudson JB. Further studies on the mechanism ofcentrifugalenhancement of cytomegalovirus infectivity. J VirolMethods 1988;19:97-108.

10 Volpi A, Whitley RJ, Ceballos R, Stagno S, Pereira L. Rapiddiagnosis of pneumonia due to cytomegalovirus withspecific monoclonal antibodies. J Infect Dis1983;147:1 119-20.

11 Stirk PR, Griffiths PD. Comparative sensitivity of threemethods for the diagnosis of cytomegalovirus lung infec-tion. J Virol Methods 1988;20:133-42.

12 Bronster 0, Makowka L, Jaffe R. et al. Occurrence ofcytomegalovirus hepatitis in liver transplant patients.J Med Virol 1988;24:423-34.

13 Wiley CA, Schier RD, Denaro FJ, Nelson JA, Lambert PW,Oldstone MBA. Localization of cytomegalovirus proteinsand genome during fulminant central nervous systemicinfection in an AIDS patient. J Neuropathol Exp Neurol1986;45: 127-39.

14 van der Bij W, Torensma R, van Son WJ, et al. Rapidimmunodiagnosis of active cytomegalovirus infection bymonoclonal antibody staining of blood leucocytes. J MedVirol 1988;25:179-88.

15 van der Bij W, Schirm J, Torensma R, van Son WJ, TegzessAM, The TH. Comparison between viraemia andantigenaemia for detection of cytomegalovirus in blood.J Clin Microbiol 1988;26:2531-5.

16 Van den Berg AP, Klompmaker IJ, Haagsma EB, et al.Antigenaemia in the diagnosis and monitoring of activecytomegalovirus infection after liver transplantation.J Infect Dis. 1991;164:265-70.

17 McKeating JA, Grundy JE, Varghese Z, Griffiths PD. Thedetection of cytomegalovirus by ELISA in urine samplesis inhibited by B2-microglobulin. J Med Virol1986;18:341-8.

18 Shulman HM, Hackman RC, Sale GE, Meyers JD. Rapidcytological diagnosis of cytomegalovirus interstitialpneumonia on touch imprints from open-lung biopsy. AmJ Clin Pathol 1982;77:90-94.

19 Emery V, Griffiths PD. Molecular biology ofcytomegalovirus. Int J Exp Pathol 1990;71:905-18.

20 Desselberger U, Collingham K. Molecular techniques in thediagnosis of human infectious diseases. Genitourin Med1990;66:313-23.

21 Myerson D, Hackman RC, Nelson JA, Ward DC,McDougall JK. Widespread presence of histology occultcytomegalovirus. Hum Pathol 1984;15:430-9.

22 Myerson D, Hackman RC, Meyers JD. Diagnosis ofcytomegaloviral pneumonia by in situ hybridisation. J InfDis 1984;150:272-7.

23 Gleaves CA, Myerson D, Bowden RA, Hackman RC,Meyers JD. Direct detection of cytomegalovirus frombronchoalveolar lavage samples by using a rapid in situDNA hybridization assay. J Clin Microbiol 1989;27:2429-32.

24 Chou S, Merigan TC. Rapid detection and quantitation ofhuman CMV in urine through DNA hybridization. NEngi J Med 1983;308:921-5.

25 Musiani M, Zerbini M, Gentilomi G, et al. An amplified dotimmunoassay for the direct quantitation of adapted andwild strains of human cytomegalovirus. J Virol Methods1989;24:327-34.

26 Shibata D, Martin WJ, Appleman MD, Causey DM,Leedom JM, Arnheim N. Detection of cytomegalovirusDNA in peripheral blood of patients infected with humanimmunodeficiency virus. J Infect Dis 1988;158: 1185-92.

27 Einsele H, Steidle M, Vallbracht A, Saal JG, Ehninger G,Muller CA. Early occurrence of human cytomegalovirusinfection after bone marrow transplantation as demon-strated by the polymerase chain reaction technique. Blood1991;77:1 104-10.

28 Cassol SA, Poon M-C, Pal R, etal. Primer mediated enzymicamplification of cytomegalovirus DNA. J Clin Invest.1989;83: 1109-15.

29 Gema G, Zipet D, Parea M, et al. Monitoring of humancytomegalovirus infections and ganciclovir treatment inheart transplant recipients by determination of viraemia,antigenaemia and DNAemia. J Infect Dis 1991;164:488-98.

30 Pass RF, Griffiths PD, August AM. Antibody response tocytomegalovirus after renal transplantation: comparisonof patients with primary and recurrent infections. J InfectDis 1983;147:40-6.

31 Marsano L, Perillo RP, Flye MW, et al. Comparison ofculture and serology for the diagnosis of cytomegalovirusinfection in kidney and liver transplant recipients. J InfectDis 1990;161:454-61.

32 Milburn HJ, Grundy JE, du Bois RM, Prentice HG,

187

on April 18, 2020 by guest. P

rotected by copyright.http://sti.bm

j.com/

Genitourin M

ed: first published as 10.1136/sti.68.3.183 on 1 June 1992. Dow

nloaded from

Pillay, Griffiths

Griffiths PD. Is the measurement of virus-specificantibody in the lungs of transplant recipients withcytomegalovirus pneumonitis of diagnostic of prognosticvalue? J Med Virol 1988;26:197-206.

33 Sutherland DER, Chan FY, Foucar E, Simmons RL,Howard RJ, Najarian JS. The bleeding cecal ulcer intransplant patients. Surgery 1979;86:386-98.

34 Rene E, Marche C, Chevalier T, et al. Cytomegaloviruscolitis in patients with acquired immunodeficiency syn-drome. Dig Dis Sci 1988;33:741-50.

35 Morgello S, Cho ES, Nielson S, Devinsky 0, Petito CK.Cytomegalovirus encephalitis in patients with acquiredimmunodeficiency syndrome: an autopsy study of30 casesand a review ofthe literature. Hum Pathol 1987;18:289-97.

36 Tucker T, Dix RD, Katzen C, Davis RL, Schmidley JW.Cytomegalovirus and herpes simplex virus ascendingmyelitis in a patient with acquired immune deficiencysyndrome. Ann Neurol 1985;18:74-9.

37 Dix RD, Bredesen DE, Erlich KS, Mills J. Recovery ofherpes viruses from cerebrospinal fluid of immuno-deficient homosexual men. Ann Neurol 1985;18:611-4.

38 Henderly DE, Janpol LM. Diagnosis and treatment ofcytomegalovirus retinitis. J Acq Imm Def Synd1991;4(Suppl 1):S6-S10.

39 Rubin RH. Impact of cytomegalovirus infection on organtransplant recipients. Rev Infect Dis 1990;12(Suppl 7):S754-S766.

40 Millar AB, Patou G, Miller RF, et al. Cytomegalovirus in thelungs of patients with AIDS: respiratory pathogen orpassenger? Am Rev Respir Dis 1990;141:1474-7.

41 Grundy JE, Shanley JD, Griffiths PD. Is cytomegaloviruspneumonitis in allogeneic transplant recipients animmunopathological condition? Lancet 1987;ii:996-9.

42 Squire SB, Lipman MCI, Bagdades EK, et al. Severecytomegalovirus pneumonititis pneumonitis in HIVinfected patients with higher than average CD4 counts.Thorax (in press).

43 Webster A, Prentice HG, Pothecary K, Griffiths PD. Thevalue of routine surveillance cultures for detection ofCMV pneumonitis following bone marrow transplanta-tion. Bone Marrow Transplant. (in press).

44 Pillay D, Charman H, Burroughs AK, Smith M, Rolles K,Griffiths PD. Surveillance for CMV infection in orth-otopic liver transplant recipients. Transplantation (inpress).

45 Salmon D, Lacassin F, Harzic M, et al. Predictive value ofcytomegalovirus viraemia for the occurrence of CMVorgan involvement in AIDS. J Med Virol 1990;32:160-3.

188

on April 18, 2020 by guest. P

rotected by copyright.http://sti.bm

j.com/

Genitourin M

ed: first published as 10.1136/sti.68.3.183 on 1 June 1992. Dow

nloaded from