J Clin Pathol 1983;36:683-692

Problems of infection after bone marrowtransplantationJ GRAHAM WATSON

From the Royal Free Hospital, Pond Street, London NW3

SUMMARY Representatives of 17 bone marrow transplant (BMT) teams met to discuss the prob-lems of infection after BMT. With experience of more than 2200 transplanted patients over thepast 10 yr, the changes in the patterns of infection were surprisingly similar.Deaths due purely to bacterial infection have greatly diminished and the major early problems

today result from fungi and cytomegalovirus. The role of non-herpes group viruses has onlyrecently received attention. With increasing numbers of survivors, late bacterial infections areassuming importance. The meeting also provided an opportunity to document the measuresadopted to prevent infection during BMT and following discharge.

Infection remains one of the main problems in bonemarrow transplantation (BMT). Patients are at riskof severe exogenous infection for months after atransplant and may also experience reactivation ofendogenous infections. In pneumonitis, which is amajor hazard especially after total body irradiation(TBI), infection is an important factor in many cases(perhaps in all), while the other main complicationof transplantation, graft versus host disease(GvHD), may be precipitated or complicated byinfection. Full control of infection, therefore, consti-tutes a vital step in the perfection of BMT proce-dures and a step which, theoretically at least, shouldbe attainable.A workshop on this topic sponsored by the

Leukaemia Research Fund was held at the RoyalMarsden Hospital on 25 June 1982. The intention ofthe conference was to bring together experiencefrom many marrow transplant units and to reviewtheir incidence and patterns of infection and themethods of preventing or treating infection so as toprovide an overview not previously available exceptby searching out and correlating individual reportsfrom the published work of each centre. The ques-tions addressed were concerned with the compara-tive incidence and experience of specific infections,with the regimens of decontamination, isolation,recontamination, prophylactic antibiotics, granulo-cyte transfusions and other measures used to pre-vent infection. Diagnostic procedures and treat-

Accepted for publication 6 January 1983

ments were also compared and the general questionof the relation of infection and GvHD in mouse andman was particularly emphasised. A summary of theworkshop with key references has been compiled.

Methods

A questi)nnaire sent to each centre requested theirtransplant methodology and the incidence and sev-erity of certain infections. We wished to know thenumbers, nature and diagnosis of BMT recipients,the pregraft chemoradiotherapy schedules anddetails of the protocol used to prevent GvHD. Wedid not request statistics on survival, leukaemiarelapse or the quality of life nor did we ask fordescriptive details of GvHD. We wished to know themethods used to prevent infection immediately afterBMT and after discharge from hospital; the nature,degree and duration of isolation; the prophylacticantimicrobial agents used and their route, of deliv-ery; the use of prophylactic granulocytes or a pro-tected diet; and how long these measures continuedafter BMT and what new measures were adopted orinstructions given on discharge home.We sought the incidence and severity of infection

with four viruses (varicella-zoster virus (VZV),herpes simplex virus (HSV), cytomegalovirus(CMV) and papovavirus), four groups of bacteria(Staph aureus, Staph epidermidis, Corynebacteriumspp and anaerobes), Candida spp and Pneumocystiscarinii. In addition other infections and infectingorganisms were to be written into the data by teams

683

684

who felt they had particular experience to convey.The Tables (see later) have been constructed fromthese data and the text contains salient points whicharose in discussion.

Results

PREVENTION OF INFECTION:(TableS la and b)These show the measures currently taken in 17BMT centres to limit infection after BMT in bothinpatients and outpatients. Five centres (three in theUK) do not use filtered air. Two centres (Seattle andMemorial) include a comparison of LAF with simpleroom isolation (as part of wider trials) and in timemight be able to comment on whether stringent iso-lation still confers additional benefit today. It isunfortunate that both these centres include antibio-tics normally regarded as important in treatment ascomponents of their prophylactic antimicrobialregimens. No centre (except perhaps Munich) iscomparing minimal isolation with more sophisti-cated isolation techniques while both groups receivenon-absorbable antimicrobial agents.

Watson

It is not possible to compare the efficiency of dif-ferent antibacterial regimens in this report. Onecentre, Baltimore, has used no prophylactic oralantibacterial agents (or protected diet) at the time ofBMT since June 1981 so their reports of Gram-positive infections (Table 5), Gram-negative infec-tions and perhaps indirectly viral infections are ofinterest but include data from the time when oralprophylactic antibacterial agents were used. Futurereports from Baltimore and from Seattle of theirexperiences with isolation plus antifungalprophylaxis only should allow an up-to-date assess-ment of the contribution made by prophylactic anti-bacterial agents in the prevention of infection andGvHD. Together with their more simple isolationstudies the data being generated by these centrescould have wide ranging implications for units start-ing BMT. Nystatin remains the most popularprophylactic antifungal agent although the earlydata and clinical acceptability of mepartricin makethis absorbable methylated relative of amphotericinperhaps worthy of a wider trial than only Genoa.Data from the Royal Free Hospital' suggested that

Table la Measures to prevent infection during BMT

Centre Protected Prophylactic antibacterial agents Prophylactic antifungal Prophylactic Protectedenvironment agents granulocytes diet

Seattle LAF Van Pol Tob po plus Tic, Tob Van iv Nys No SterileLAF Van Pol Tob po Nys No SterileRoom Van Tic Tob iv Ket No NoRoom None Nys No No

Minneapolis Room* Cot Nys No NoMemonal LAF or Gen Cef Nys No Sterile

Room None Nys No CleanUCLA Room Van Col Cot Nys or Ket Not now CleanBaltimore Room* Not now Nys No NoSt Louis LAF "Total" Ket No Ster. GFCreteil Room Varies with antibiogram Amph Yes SterileLeiden-Paed LAF Neo Co Ceph Amph No Sterile

LAF Neo Co Amph No SterileBasel LAF Gen Van Cot Nys Amph No SterileGenoa LAF Neo Co Cot Mep No SterileMunich Room* Neo Co ± Cot Nys Amph ± Mic Rarely Sterile

LAF KetGlasgow LAF Gen Van 5FC + Nys No SterileHammersmith Room* Fram Co Cot Nys Amph Ket No Clean

MicRoyal Free Room Neo Co Cot Nys Amph Ket No SterileWestminster Room Neo Co Amph No CleanRoyal Marsden Room Neo Co or Cot Nys Amph No SterileGt Ormond St Room Neo Co or Cot Nys No Clean

LAF = laminar air flow.* wwith filtered positive pressure air.GF = gluten-free.Amph = amphotericin.Cef = cefamandole.Ceph = cephaloridine.Co = colistin.Cot = cotrimoxazole.Fram = framycetin.Gen = gentamicin.

KetMepMicNeoNysPolTicTobVan

= ketoconazole.= mepartricin.= miconazole.= neomycin.= nystatin.= polymixin-B= ticarcillin.= tobramycin.= vancomycin.

Problems of infection after bone marrow transplantation

Table lb Measures to prevent infection following discharge after BMT

Centre Cotrimoxazole Penicillin Diet Other

Seattle to 120 days Not routinely + VZ transfer factorand duringtreatment ofchronic GvHD

Minneapolis to 1 yr NysMemorial 50-150 days Avoid crowdsUCLA 30-120 days low dose + PneumovaxBaltimore to 180 days + Ket Tetanus and polio (salk)

continued if chronic at 1 yrGvHD

St Louis to 180 days continued if Tetanus, polio immunisationschronic GvHD + at 180 daysPneumovax

Creteil to 120 days From 120 days Neo Co Nal to 90 daysLeiden-Paed 20-100 days 100 days - yr* + Pneumovax* Avoid some foods Avoid crowdsBasel for 5 days after Cooked food Nys Amph loz

discharge Avoid some foods Avoid crowdsWear masks outside

Genoa Yes Cooked food Continue Neo Co CotMunich to 180 days Nys Avoid crowdsGlasgow Yes General advice -y globulin if

measles contactHammersmith to 100 days Amph loz

General adviceRoyal Free to 180 days Ket + Nys or AmphWestminster to 150 days Avoid some foods Avoid close contact with

animals. V IgRoyal Marsden to 150 days Avoid some foods Avoid pets and parties V IgGt Ormond St to 100 days Avoid some foods Nys V Ig

V Ig = Varicella immunoglobulin.VZ = Varicella-Zoster.

*only in patients grafted for leukaemia

See footnote to Table la for abbreviations.

ketoconazole is not adequate alone because of prob-lems with oral absorption in irradiated patients.

Only- Creteil use prophylactic granulocytes in allpatients, giving 1-5 x 1010 granulocytes/m2/day fromday + 1 to about day + 18. Their data on bacterial,fungal and viral infections are shown in the approp-riate sections.Most centres discharge their patients shortly after

the granulocytes exceed 1 x 109/1 and almost nonerelaxes the protected environment while the patientis in hospital. Patients transplanted because ofimmunodeficiency disease remain longer in hospital.All centres except Basel and St Louis giveprophylactic cotrimoxazole to outpatients for 3-6months and some continue for 1 yr. Only a quarterof centres give prophylaxis against pneumococcaldisease at present; this is surprising since half the 65long term (>5 yr) survivors at Seattle have hadthree or more infections after day 100 of which Streppneumoniae and Staph aureus were predominantcauses. Varicella immunoglobulin is given as routineat only three of six British centres reporting.

Several centres restrict outpatient diet in that all

food must be freshly cooked. Most centres suggestavoiding crowded places. One centre insists on nopets or parties, one on no new "body contact" petsand one on wearing masks in the street. Many con-tinue antifungal agents (usually nystatin) and inaddition Creteil and Genoa continue gut decon-tamination for three months after discharge.Tetanus and polio immunisations are given after twomonths at Leiden-Paed, 6 months at St Louis; Salkpolio at 1 yr in Baltimore and the demonstration ofan antibody response is required before discontinu-ing isolation of Westminster s children with congeni-tal immunodeficiency and those at Leiden-Paed.

SPECIFIC INFECTIONS

Fungal infections (Table 2)Fungal infections still cause major problems in diag-nosis and treatment, despite new prophylactic anti-fungal agents. In Minneapolis invasive fungal dis-ease was associated with 30% of all deaths whereasviral and bacterial infections were respectivelyassociated with 18% and 10%. In Seattle 30% of

685

Table 2 Fungal infections after BMT

Centre Due to Candida spp Due to Aspergillus spp Due to others

Incidence Severe forms

Seattle 5% of 'bacteraemia' Various Nocardia30% of necropsies Coccidiomycosis,

CryptococcusTorulopsisPacillomyces

Minneapolis 5/50 patients 2 contributed to death 10/50 Mucor 1Memorial 2/122Los Angeles 45/265 30 severe 18/265 Trichosporon 1Baltimore Incidence reduced Nocardia 1

from 20% to 4% Trichosporon 1St Louis 8/63 Torulopsis 2

Cryptococcus 1Creteil 4/34 0 0Leiden-Paed 1/31 0 fatal 0/31Basel 20 colonisations/50 1 severe 1/50Genoa 4/56 1 fatalMunich 5/48 4 contributed to death 1/48Glasgow 4 oralV7 0Hammersmith 0/43 0 4/5 CMLRoyal Free 14/19' 4/66 Mucor 1Westminster 19/73 1 fatal 4/73 (all fatal) Nocardia 1 (fatal)Marsden 18/66 3 contributed to deathGt Ormond St 1/23 Fatal 1/23

necropsies showed invasive candidiasis. In this nec-ropsy series the incidence of fungal infection afterBMT for aplasia (46%) was twice that after BMTfor acute leukaemia (20%), almost certainly due tothe more prolonged neutropenia of aplasia. Theirpatients with severe neutropenia (<0.1 x 109/1) last-ing up to 20, 40 and 60 days have incidences of21%, 41% and 57% respectively of invasive fungaldisease.

Aspergillus spp occupied considerable discussion.Minneapolis reported 13 episodes in 10 patients(out of 50) of whom nine died. The one survivor hadonly a thigh infection; the other nine had pulmonaryaspergillosis with central nervous system dissemina-tion in five. Only one pulmonary aspergillosis was

diagnosed before death (investigation in all includedlung biopsy) and seven of the ten were receivingempirical amphotericin B. These infections had noseasonal incidence, occurred any time after BMTand five patients had other infections at the time.Eight of ten had been receiving long term broadspectrum antibiotics. Although Westminster washout radiologically abnormal sinuses before BMT, allfour aspergillosis infections (all fatal) in theirexperience originated there.The importance of filtering air to remove Asper-

gillus spp was shown by Peterson who described howplacing HEPA recirculating filters in each roomreduced the incidence of invasive aspergillosis from12/65 (18%) to 3/66 (5%) and additionalnasopharyngeal colonisation from 5/65 to 2/66

patients, without any change in the incidence ofthese problems in the rest of the hospital. Adequatefiltration to exclude the smaller spores ofAspergillusfumigatus was essential in preventing disease.Two studies of prophylactic ketoconazole versus

oral nystatin and amphotericin were reported. BothPrentice' and Gluckman independently describedconsiderable reduction in fungal infections in thepatients who received ketoconazole but controlswere historical in the Gluckman study and the doseof nystatin (1.2 x 106 units/day) considered inade-quate by some in the other. The optimal prophylac-tic regimen is not yet agreed. Prentice describedconsiderable reduction in the absorption ofketoconazole after day 21 associated with abnor-malities of gastrointestinal function.' Mepartricin,an orally absorbed methyl ester of an amphotericin-like heptane is without renal toxicity as describedby Bacigalupo (Genoa)2 and appears of great inter-est although his incidence of positive candida cul-tures when using either prophylactic mepartricin orketoconazole was similar though much less than inthe patients who received nystatin or miconazole.Even adequate treatment may not be sufficient

unless also commenced at an early stage. Because itis unusual to culture Aspergillus sp in the absence ofinvasive disease, it was strongly recommended thatany patient with a positive culture for aspergillishould receive a total of 2 g amphotericin B; and iftreating possible clinical disease the early institutionof treatment is of the greatest importance. A sug-gested schedule is 1 mg amphotericin test dose

686 Watson

Problems of infection after bone marrow transplantation

intravenously followed one hour later by 20 mginfused over 24 h and then 50 mg during each 24 hto a total of 2 g. Some felt 5-fluorocytosine andrifampicin were helpful, and that radiologicalsinusitis with a pulmonary infiltrate should beregarded as due to Aspergillus sp until provedotherwise.

Cytomegalovirus infections and pneumonitis (Table3)Cytomegalovirus (CMV) gave rise to much discus-sion because of the prevalence of fatal CMVpneumonitis, the difficulty of aetiological diagnosisin pneumonitis, the lack of effective treatment andthe suspicion that most CMV infection in seronega-tive BMT patients is iatrogenic.Between 10% and 50% of patients develop CMV

infection after BMT. Fifty-nine percent of USApatients with CMV developed CMV pneumonitiscompared to 23% of European patients, the propor-tion of such pneumonitis proving fatal being 60% onboth sides of the Atlantic. Factors which increasethe incidence of CMV pneumonitis and CMV infec-tion include being a caucasian; GvHD grade II orgreater; age >12 yr; lung radiation greater than 6Gy; preceding recipient CMV seropositivity; reci-pient seronegativity with donor seropositivity (not inSeattle); the giving of granulocytes and degrees ofrecipient/donor HLA matching which are less thanfully compatible.3Pneumonitis was twice as common after BMT for

acute leukaemia as after aplastic anaemia (p <0.005) but even with transplants for aplasia the inci-dence varies with the conditioning regimen. Asses-sed 15 weeks after BMT for aplasia, cyclophos-phamide, PAPACY and TBI preconditioning regi-mens were associated with incidences ofpneumonitis of 8%, 30%, and 70% respectively.3The lesser incidence of interstitial pneumonitis aftersyngeneic BMT is due mainly to an absence of CMVpneumonitis. True idiopathic pneumonitis seems ascommon after syngeneic BMT as after allogeneicBMT. Barrett and Depledge4 have shown a clearrelation between total lung radiation dose and theincidence of lung complications after BMT in manand also how a dose rate of 0 1 Gy/min or greaterseriously and increasingly impairs gas transfer inmice. A recent International Registry report5 con-cerning 176 patients transplanted for acuteleukaemia identified three of 30 prognostic factorssignificantly (p < 0-005) associated a low risk ofpneumonitis-a radiation dose rate less than 0-057Gy/min, cyclosporin A rather than methotrexate asprophylaxis against GvHD and female donor mar-row into a female recipient.There was much interest in the relevance of

Table 3 Cytomegalovirus infections after BMT

Centre Incidence Severe forms

Seattle 50-60% fatal pneumonitis in 15-20% of leukaemias, <5%of aplastics: alsoleukopenia, hepatitis andpossibly graft rejection inaplastics

Minneapolis 13/50 (patients)Memorial 16/122 10 pneumonitisLos Angeles 80/256 55Baltimore 89/200 51 pneumonitis. 31% fatalSt Louis 31/63 5 pneumonitis. 60% fatalCreteil 7/34 1 encephalitis; 1 fatal

pneumonitisLeiden-Paed 4/31 0Basel 4/50 4 fatal pneumonitisGenoa 8/56 2 fatal pneumonitisMunich 2/48 1 fatal pneumonitisGlasgow 0/7Hammersmith 6/43 2 fatal pneumonitisRoyal Free 16/47 6 pneumonitis (4 fatal)Westminister 22/73 1 encephalitis,

1 pneumonitis,2 hepatitis

Royal Marsden 23/86 6 contributed to deathGt Ormond St 1/23 Fatal

donor/recipient CMV serology. Patients who receiveprophylactic granulocytes have an increased risk ofCMV which is dependent upon whether thegranulocyte donors are seropositive or seronegative(p = 0.005). A smaller difference is seen betweensimilar groups of patients who receive therapeuticgranulocytes p=0027). Extending their 1980report6 Young reported that when only CMV-negative recipients were studied, 16/46 (35%) whoreceived prophylactic granulocytes from onlyCMV-negative donors developed CMV comparedto 21/28 (75%) who received prophylactic granulo-cytes from only CMV-positive donors. (p < 0.01).The presumption is that platelets and blood transfu-sions contribute a significant proportion of theremaining infections. Should granulocytes berequired, Seattle use single family member donorsfor individual patients. Creteil (Jean Vernant) werethe only centre using a full prophylactic granulocyteregimen-an average of 1-52 x 10'° granulocytesmr/day from days + 1 to + 18 after BMT. Of 34 suchpatients, seven experienced CMV infection-fivebenign hepatitis, one encephalitis (resolved) andone fatal interstitial pneumonia. Their overall inci-dence of interstitial pneumonia in remission BMTpatients was 11 of 49 (2 fatal), a greater incidence ofinterstitial pneumonia occurring in those who didnot receive granulocytes. More data are required toexplain their low incidence of CMV in the presenceof such quantities of unscreened granulocytes.No useful therapy for CMV has yet emerged in

that bromovinyl deoxyuridine (BVDU), adeninearabinoside, interferon from various sources and

687

Watson

acycloguanosine in various combinations haveproved ineffective and sometimes toxic (eg adeninearabinoside and interferon) and CMV infection hasdeveloped during long-term prophylactic acyclovir(10mg/kg/day). Winston (UCLA) presented data7from 48 patients using 10ml/kg of CMV hyperim-mune plasma (titre 1/256) before and after BMTand showed a substantial reduction only in thosepatients who did not receive granulocytes. Meyers(Seattle)8 also reported a reduction in CMV infec-tions in a 50 patient study of passive protection withexogenous CMV antibody but again the benefitappeared to be confined to those who did notreceive exogenous granulocytes. It would seem thatthese infusions coupled with careful selection ofdonors when possible or applicable are the onlymeasures to prevent CMV disease at the moment.

Peterson reported experience with open lungbiopsy (OLB). In these data 5/9 patients withpneumonitis diagnosed as "idiopathic" by trans-bronchial needle biopsy were shown to be due toCMV on OLB. OLB is associated with no mortalityand negligible morbidity on the West Coast of theUnited States. It satisfies the two major require-ments, adequate specimens and good haemostasisand new aetiological agents might be found-forexample, four recent chlamydial pneumonias inSeattle-where the workers are sufficientlyconfident that if no pathogens are identified on OLBthe diagnosis must be radiation pneumonitis, dexa-methasone 16mg/day is given and not antimicrobialagents. Saral (Baltimore) disagreed, stating that in

Baltimore pneumonitis of unknown cause is treatedwith Septrin and erythromycin and occasionally fourdays of steroids without OLB.

Recent work on the relation between the stemcell, CMV and the maturing donor marrow wasdescribed by Saral. Stem cells undifferentiated whilepropagating in culture, when infected with murineCMV, do not allow CMV replication and no virus,no viral RNA and no viral antigens can be identifieduntil stem cell differentiation occurs. However viralDNA is present. The viral DNA genome is thereforelatent and activation awaits the differentiation of theinfected cell. Saral also described cellular methodsof assessing the prognosis in CMV infection.Cytotoxic T cell responses and NK cells were pre-served in post BMT patients with non-fatal CMVwhereas they were very low in patients whose CMVultimately proved fatal.9 a-interferon could restorethese absent responses in approximately half the invitro tests.

Varicella-zoster and herpes simplex viruses (Table 4)It was striking how uniform the experience ofcentres was in the incidence of VZ infections,approximately 40% of survivors of BMT developingzoster. Between 20% and 40% of these develop dis-seminated disease and there was no evidence thatthe routine prophylactic use in three centres of vari-cella immunoglobulin (Table lb) during the firstthree months after BMT reduced their incidence ofzoster. A mortality rate of about 5% of cases wasquoted, predominantly due to varicella pneumonia.

Table 4 Varicella-zoster and herpes simplex virus infections after BMT

Centire Varicella-zoster Herpes simplex

Incidence Severe forms Incidence Severe forms

Seattle 215/1000+ cumulative 5% fatal 1/3 disseminate 80% of antibody+ 20+ fatal pneumonias40% usually first 1/6 varicella type onset some severe mucositis,5 months oesophagitis

Minneapolis 4/50 (patients) 0 11/50Memorial 26/122 35/122 1Los Angeles 20/265 5 generalised 1 fatal 200/265 15Baltimore 40%, median at 3 months 4% fatal 113/270 72% of 2 encephalitis

antibody+ 6 pneumonitisSt Louis 14/63 27/63Creteil 7/34 0 8/34 0Leiden-Paed 5/31 1 generalised 3/31* 0Basel 6/50 19/50* 2 encephalitisGenoa 3/56 4/56Munich 7/48 0 4/48Glasgow 0/7 1/7 1 fatalHammersmith 12/43 1 generalised 13/43*

1 pneumoniaRoyal Free 5/37 17/371"Westminster 11/73 1 fatal pneumonia 13/73Royal Marsden 18/66 2 generalised 22/66* 1 encephalitis

1 encephalomyelitisGt Ormond St 2/23 0 0/23

*Made specific comment on recurrence.

688

Problems of infection after bone marrow transplantation

It is likely that the early use of acyclovir will reducethis.

Discussion of herpes simplex (HSV) mainlycentred around the use of acycloguanosine (ACG).Intravenous (250mg/m2 tds)'° and to a lesser extentoral (5mg/kg bd)" prophylaxis were effective inpreventing HSV lesions developing and intravenoustherapy led to rapid resolution of active lesions.'2The place of oral therapy is still to be resolved butthe problems of rapid recurrence of lesions aftertreatment is stopped would probably remain. Resis-tance to ACG by HSV was reported but all strainswhich have been tested lacked thymidine kinase (3/3in Seattle, 6/6 in Baltimore and St Louis, Paris) andconsequently had thrived despite ACG.'3 Severalcentres reported recurrent HSV after stoppingtreatment but almost all patients responded well to asecond course. Meyers (Seattle) described a briefsyndrome, occurring in six patients, five of whomhad received high ACG doses, which was character-ised by tremor, disorientation and mood changes butgenerally without focal signs. Although attributedby Meyers to ACG, Saral and Gluckman had seenthis in the absence of ACG and attributed it to a lateeffect of the preconditioning. Not one of the sixencephalitides in Table 4 was proven by biopsy.No-one suggested that strange neurological syn-dromes may be due to cyclosporin A which has beenshown to produce considerably different toxiceffects in younger age groups.'4 Concern wasexpressed by some that ACG might affect bonemarrow recovery despite the difficulty of differen-tiating ACG effects from those of the virus.No data were presented or cited to implicate

ACG in slow marrow recovery. Workers reportedthat 30 mg/kg/dose of ACG if given with one litre offluid per gram of drug caused no renal toxicity.

Other virusesThe Baltimore experience with viral gastroenteritisafter BMT was recently reported.'5 This showed40% ofBMT recipients became infected with one ormore viruses during their initial admission and drewparticular attention to adenovirus, rotavirus andCoxsackie virus as pathogens. Seventeen of 31patients (13 of 22 viral excretors) with entericpathogens in their stools died compared to 6 of 47without enteric pathogens.Other centres reported their commoner isolates.

Papovavirus, usually BK strain, was found in theurine of between 30% and 80% of BMT recipients(Memorial, Baltimore, Westminster), with a poss-ible association with mild hepatotoxicity.16 Nodefinite evidence of neurotoxicity has emerged, evenin those who excreted JC virus. Adenovirus causedat least 70 severe infections with a mortality rate of

about 15% reported from several centres. Serotypes2, 5 and 11 predominated and infection was mani-fest most usually as a respiratory disease whichrapidly led to severe gut and renal damage. Cox-sackie A 1, 2, 5 have predominantly been shedwithout disease, although both Coxsackie virus-infected Baltimore patients died. One patient diedof influenza in Basel. Epstein-Barr virus remains anarea ripe for exploration. Although Westminsterreported 14 seroconversions after BMT, there wasonly a possible association with hepatitis andGvHD.

Infections with Gram-positive cocci (Table 5)Staph aureus infections were not considered a majorproblem. With the exception of Seattle and Balti-more, there were 45 infections reported of whichseven (16%) were fatal. Saral (Baltimore) felt Staphaureus was a major cause of infection in chronicGvHD and Meyers described Staph aureus as a fre-quent cause of late infection (>100 days afterBMT).

In contradistinction, Staph epidermidis infectionsare currently the most common cause of bac-teraemia in Seattle and certain other centres, andalthough not often fatal (8 of 103 on whom full datawere available) protracted illness, extensive soft tis-sue infections, endocarditis, pneumonia and infectedemboli were described. Frequent resistance to allantimicrobials except vancomycin was found insome centres and this antibiotic is included in thecombination of antimicrobials used at the onset ofpyrexia in Seattle despite concern about the toxicityof the combination of cyclosporin A and vancomy-cin. Double lumen "Hickman" catheters allowingseparate channels for giving and withdrawal did notreduce the incidence of Staph epidermidis sep-ticaemia. However, Westminster reported that care-ful attention to sterile technique plus keeping theend of the "Hickman" line in a sterile finger cot orthe catheter/giving set junction in a sterile plasticbag had eliminated their problem writh Staphepidermidis septicaemia, not now seen for 10months.'7 There was general agreement that indwel-ling intravenous lines should be removed in theevent of septicaemia with the possible exception ofStaph epidermidis.Pneumococcal infections remain a problem, par-

ticularly later than 100 days after BMT. Resistanceto Septrin, delayed and poor responses toPneumovax immunisation (particularly the 6Aantigen-the most common serotype in LosAngeles) and impaired splenic function were allcited as reasons, but no-one commented on a changein their local incidence once prophylactic penicillinwas given.

689

Table 5 Gram-positive bacterial infections after BMT

Centre Staph aureus Staph epidermidis Strep pneumoniae Corynebacteria Other Gram-positive

5-10% ofbacteraemias2/50 (patients)4/12212/265 septicaemias

More frequent now

6/630/343/310/507/562/480/72/410/202/735/660/23

33% ofbacteraemias12/5017/12265/265

Major cause ofinfection32/630/343/313/50

14/561/480/75/4360% of bacteraemias19/736/660/23

Frequent late 40-50/1000+infections now rare

Frequent lateinfections4 fatal

31 fatal (late)

5/732 fatal

1/12215/265

declining4/630/34

0/508/560/48

1/43

2/660/23

Anaerobic infectionsA small proportion of bacteraemias (<5%) weredue to anaerobic organisms, mainly Clostridiumdifficile and Bacteroides spp, but these did not con-stitute a problem and many centres had identified noanaerobic infections.

Gram-negative infectionsMeyers (Seattle) reported that from 1969-73Gram-negative organisms caused 71% of infections,from 1974-1976 50%, and from 1976-80 21% ofinfections. The data from all centres mirrored thisdecline and the speciation of isolates in most centreswas similar with E coli, Pseudomonas spp and Kleb-siella spp accounting for almost all Gram-negativesepticaemias. The vast majority were associatedwith neutropenia and although overall about one infive proved fatal, the influences of coexisting GvHD,or in aplasia, graft failure suggest that death due to"pure" Gram-negative septicaemia is a relativelyunusual event.

O[HER INFECTIONSToxoplasmosisSt Louis, reported three cases, all about six monthsafter BMT. Two presented with early CNS involve-ment, rapidly progressed and died. Seattle had seensix cases (all fatal) with, variably, pneumonia,myocarditis and encephalitis.

'TuberculosisBasel reported two cases both treated successfullyand the Royal Free, one fatal case. Fatal dissemi-nated BCG disease in a six-month infant with SCID,who had been immunised at birth with rapid andapparently normal healing of the immunisation site

was reported from Westminster.

The role of infection in precipitating GvHD

The experimental relation between infection andGvHD was explored by two speakers. Heidtdescribed earlier mouse work where the germ-freestate conferred great protection against death fromdelayed GvHD but that continuation of the germ-free state was only necessary for 40 days after trans-plantation. Earlier conventionalisation resulted infatal GvHD.'8 19 No single organism, antigen orendotoxin has been identified which will precipitatethis reaction but Gram-negative bacteria seem toplay an important role. The degree of protectionfrom GvHD by the decontaminated state seems tobe related to the number of immune competent cellspresent in the graft.20Experiments with (C57BLxCBA) Fl hybrids

have shown that after transplantation with CBAbone marrow and spleen cells, the damage to sub-cutaneous implanted fetal Fl gut in a conventionalrecipient by GvHD is twice as great as in Fl fetal gutimplants carried by decontaminated chimeras,despite also not being in contact with alimentaryflora. CBA fetal gut implants developed substantialdamage when present in conventional chimeras, butnot when present in decontaminated chimeras. Thederived hypothesis is that gut microfloral antigensgained access via minor GvHD lesions, cross-reacted with host tissue antigens and induced a sec-ondary response in the transplanted cells which thenreacted with the recipient tissue.2' Germ-free mon-keys survive BMT better than conventional monk-eys because of a reduction in GvHD and even mon-keys who are fully MLA matched are also relatively

Seattle

MinneapolisMemorialLos Angeles

Baltimore

St LouisCreteilLeiden-PaedBaselGenoaMunichGlasgowHammersmithRoyal FreeWestminsterMarsdenGt Ormond St

10/502/122

15/63

3/31

2/563/48 (1 fatal)

690 Watson

Problems of infection after bone marrow transplantation

protected.Similar results were reported by Truitt who used

AKR mice with end-stage leukaemia to show thatdecontamination should commence a minimum of10 days before BMT in order to have maximumreduction in post-BMT mortality. Maximum benefitwas obtained only when both decontamination andisolation in a protective environment were used.Decontamination and isolation per se did not elimi-nate the GvHD reaction in immunosuppressedleukaemic AKR mice; however, much of the mortal-ity associated with GvHD was eliminated. Thedegree of GvHD remaining was dependent on thecompatibility of the donor and host as well as thedose of post-thymic lymphocytes transplanted.Bacteria-free "leukaemia-cured" AKR chimerascould be returned to a conventional microbial envi-ronment without significant increase in mortality butthe timing of this reconventionalisation was crucial.

Vossen22 reported data from Leiden concerninghigh dose total (Neo Co Ceph) and low dose selec-tive (Neo Co) prophylactic oral antimicrobial decon-tamination and the incidence of acute GvHD inchildren transplanted for aplasia (Table 6). Suppres-sed stools were those from which Gram-negativebacteria or fungi could not be isolated. Childrenwith suppressed stools had less GvHD (p < 05 >0-1) but "total" decontamination resulted in a muchlower incidence of GvHD though the proportion ofconsistently suppressed stools was similar. Therewas no difference in the incidence of infections.Early onset acute GvHD is due to the cytotoxicity oftransplanted mature lymphocytes whereas late onset(>3 weeks after BMT) acute GvHD is caused by Tlymphocytes which have matured out of the graftunder the influences prevailing at that time. Severeearly onset GvHD was seen only in children withhereditary forms of aplasia, but exclusion of thesefive children shows late onset GvHD in 3 of 12 sup-pressed.The most striking "effect" on late onset acute

GvHD was seen in the totally suppressed group:no GvHD out of five non-hereditary aplasias. Thesame "effect" was not seen in the selectively sup-pressed group: three GvHD out of seven non-hereditary aplasias. It was suggested that gut bac-

Table 6 Stool flora and the incidence ofGvHD

Nature of Consistent stool Acute GvHD Acute GvHDdecontamination microbiological of late onset

findings

Total Suppressed 6 0 0Not suppressed 4 3 2

Selective Suppressed 8 4 3Not suppressed 3 3 1

teria or their products together may contribute toGvHD. Stool endotoxin concentrations suggestedan association with the development ofGvHD but the inclusion of oral colistin,an endotoxin-binding agent, in the regimens madeinterpretation difficult. The low rate of chronicGvHD (2 of 25 aplastics; 0 of 8 acuteleukaemia patients) may reflect the degree ofmicrobial suppression achieved by Vossen et al.22

Active and passive recontamination were brieflydiscussed. Various centres recounted anecdotes ofacute GvHD developing shortly after either activerecontamination, withdrawal of decontamination ordischarge home yet while continuing oral Septrin,but no centre has any study of the precipitation ofGvHD in relation to the changing stool microbiol-ogy after discharge from hospital.

Conclusion

It was evident that very different degrees of impor-tance are attached in different places to the role ofinfection in causing GvHD. Experimental datawould seem to indicate the need for relatively com-plete and prolonged bacterial decontaminationbefore and after transplant (perhaps as much as 12days before and 60-80 days after) but the practicaldifficulties of doing so (which would among otherthings reduce the number of transplants done) andthe difficulties of assessing the benefits have largelyinhibited attempts at comparative trials. With anever-expanding scope for marrow-transplants-forexample, for thalassaemia and for the use of non-matched donors-this problem will remain withouteasy solution. The workshop was an invaluableopportunity to compare detailed results and theinformation exchanged will undoubtedly lead toimproved practices in many circumstances, but noobvious pathway to clarify the relation of infectionand GvHD in man was proposed.

I am grateful to the Leukaemia Research Fund forsponsoring this workshop and Prof HEM Kay foradvice. Institutions contributing patient data for thisreport were:Academisch Zieckenhuis, Leiden (J Vossen).Fred Hutchinson Cancer Research Center, Seattle

(J Meyers).Hammersmith Hospital, London (J Goldman).Hospital Henri Mondor, Creteil (J Vemaut).Hospital St Louis, Paris (E Gluckman).Johns Hopkins University School of Medicine,

Baltimore (R Saral).Kantonsspital, Basel (B Osterwalder).Ludwig-Maximilians-Universitat, Munich (HJ

Kolb).

691

692

Ospedali Civili di Genova (A Bacigalupo).Royal Free Hospital, London (G Prentice).Royal Hospital for Sick Children, Glasgow (M

Willoughby).Royal Marsden Hospital, London (R Powles).The Hospital for Sick Children, London (R

Levinsky).Sloan Kettering Cancer Center (R O'Reilly).UCLA School of Medicine, Los Angeles (L Young).University of Minnesota (P Peterson).Westminster Children's Hospital, London (T

Rogers).

References

'Hann IM, Prentice HG, Corringham R et al. Ketoconazole versus

amphotericin B plus nystatin for fungal prophylaxis in severelyimmunocompromised patients. Lancet 1982;i:826-9.

2 Van Lint MT, Bacigalupo A, Frassoni F et al Ketoconazole ver-

sus mepartricin for the prevention of candida infections in theimmunocompromised host. Exp Hematol 1982;10 suppl10:6-7.

Meyers JD, Flournoy N, Thomas ED. Non-bacterial pneumoniaafter allogeneic marrow transplantation-review of ten years

experience. Rev Infect Dis 1983; (in press).Barrett A, Depledge MH. Total body irradiation: some factors

affecting outcome. Exp Hematol 1982;10, suppl 10:56-63.'Bortin MM, Kay HEM, Gale RP, Rimm AA. Factors associated

with interstitial pneumonitis following bone marrow transplan-tation for acute leukaemia. Lancet 1982;i:437-9.

6Winston DJ, Winston G, Ho MD, et al. Cytomegalovirus infec-tions associated with leukocyte transfusions. Ann Intern Med1980;93:671-5.

7Winston DJ, Pollard RB, Ho WG, et al. Cytomegalovirusimmune plasma in bone marrow transplant recipients. AnnIntern Med 1982;97:11-18.

8 Meyers JD, Leszczynski J, Zia JA, Levin MJ, Wright GG,Thomas ED. Clin Res 1982;30:374A (abstract).

9 Quinnan GV, Kirmani N, Rook AH, et al. Cytotoxic T cells incytomegalovirus infection. N Engl J Med 1982;307:7-13.

'° Saral R, Burns WH, Laskin OL, Santos GW, Leitman PS.Acyclovir prophylaxis of herpes-simplex-virus infections. NEngl J Med 1981;305:63-7.

Watson

"Hann IM, Prentice HG, Blacklock HA, et al. Acyclovirprophylaxis of herpes virus infections in severely immunocom-promised patients: A randomised double blind trial. Br Med J1983; (in press).

12 Selby PJ, Powles RL, Jameson B et al. Parenteral acyclovirtherapy for herpes virus infections in man. Lancet1979;ii: 1267-70.

13 Burns WH, Saral R, Santos GW, et al. Isolation and characterisa-tion of resistant herpes simplex virus after acyclovir therapy.Lancet 1982;i:421-3.

14 Joss DV, Barrett AJ, Kendra JR, Lucas CF, Desai S. Hyper-tension and convulsions in children receiving cyclosporin A.Lancet 1982;i:906.

,s Yolken RH, Bishop CA, Townsend TR, et al. Infectious gas-troenteritis in bone marrow recipients. N Engl J Med1982;306:1009-12.

16 O'Reilly R, Lee FK, Grossbard E et al. Papovavirus excretionfollowing marrow transplantation: incidence and associationwith hepatic dysfunction. Transplant Proc 1981;13:262-6.

7 Watson AJ. Staph epidermidis infections can be prevented. Nurs-ing Times 1983; (in press).

1 Van Bekkum DW, Roodenburg J, Heidt PJ, van der Waaij D.Mitigation of secondary disease of allogeneic mouse radiationchimeras by modification of the intestinal microflora. J NatlCancer Inst 1974;52:401-4.

'9 Van Bekkum DW. Bone marrow transplantation. TransplantProc 1977;9:147-54.

20 Heidt PJ, Wagemaker G, Knaan-Shanzer S, van Bekkum DW.Two distinct types of late onset graft-versus-host disease afterbone marrow transplantation in lethally irradiated mice.Transplantation 1981;31:263-4.

21 Van Bekkum DW, Knaan S. Role of bacterial microflora indevelopment of intestinal lesions from graft-versus-host reac-tion. J Nat! Cancer Inst 1977;58:787-90.

22 Vossen JM, Heidt PJ, Guiot HFL et al. Prevention of acutegraft-versus-host disease in clinical bone marrow transplanta-tion: complete versus selective intestinal decontamination. In:Sasaki S, ed. Recent advances in germ free research Tokai:Tokai University Press, 1981:573-7.

Requests for reprints to: Dr J Graham Watson, Depart-ment of Paediatrics, Royal Free Hospital, Pond Street,London NW3, England.