Evaluation of a score system for the severity and outcome of cytomegalovirus interstitial pneumonia in allogeneic bone marrow recipients

Journal of Infection (1997) 35, 117-123

Evaluation of a Score System for the Severity and Outcome of Cytomegalovirus Interstitial Pneumonia in AIIogeneic Bone Marrow Recipients G. Gentile, P. P. Donati, A. Capobianchi, M. Rolli, A. P. Iori and P. Martino* Haematology Section, Department of Human Biopathology, University 'La Sapienza', Via Benevento 16, Rome 00161, Italy We verified whether a clinical score system developed for renal transplant patients predicts the severity and outcome of cytomegalovirus interstitial pneumonia (CMV IP) in allogeneic bone marrow (BMT) recipients. The score system was retrospectively applied to 20 patients at the estimated date of onset of IP and 1 0 - 1 4 days later. Seven patients received ganciclovir (GCV), seven received GCV plus intravenous immunoglobulin (i.v. Ig), and six received only supportive care. Nine out of 14 patients who received GCV with or without i.v. Ig survived the episode of IP (the median score of these patients at diagnosis of CMV IP was 5 (range 3-8)), while the remaining five patients died of respiratory failure during IP and at the diagnosis had a median score of 10 (range 9-11) (P = 0.01). The six patients who received only supportive care survived for a median time of 21 days (range 1 0 - 2 4 days) from the estimated onset of CMV IE and the median score at the diagnosis of IP was 10 (range 8-12). The overall survival correlates strongly with low initial severity of IF as measured by this score system: 11 out of 20 patients who died of respiratory failure during IP had at the estimated onset of IP a score >8, while of the nine patients who survived IP, eight had at the onset a score <7 and the remaining one a score of 8 (/'=0.0007). The sensibility, specificity, predictive positive value and predictive negative value of the score system (with a threshold value of 8) to identify patients who survived IP was: 100%, 88 %, 91% and 100%, respectively. The use of ganciclovir alone or in combination was the most important determinant of outcome. These data support the relevance of this score system with a threshold value of 8; if prospective and controlled studies confirm our observations, it would help physicians to identify BMT recipients during CMV IP with high vs. low risk of poor outcome.

Introduction Infection with cytomegalovirus (CMV) is a major cause of morbidity and mortality in immune compromised patients and, in particular, organ transplant recipients ~-3 and those with HIV infection. 4'5 Pneumonia occurs among the clinical manifestations of CMV infection and is associated with a poor prognosis. Cytomegalovirus interstitial pneumonia (CMV IP) occurs in 15-20% of allogeneic bone marrow (BMT) recipients and was loaded by a case fatality rate of 85%. 6 The survival rate of BMT recipients with CMV IP ranged from 2 2 - 4 5 % when ganciclovir (GCV) was administered as monotherapy, 7'8 whereas it ranged from 3 1 - 5 2 % when ganciclovir was combined with intravenous (i.v.) immunoglobulin, s-~I The reasons for differences in the survival rate of the patients treated with ganciclovir alone or ganciclovir * Address all correspondence to: P. Martino. Preliminary data from this study were presented at the 9th International Symposiumon Infections in the ImmunocompromisedHost, held in Davos, Switzerland in June 1994. Accepted for publication 19 November 1996. (3163-4453/97/050117+07 $12.00/0

and i.v. immunoglobulin, other than the efficacy of the combination therapy, 9-n might be due to several causes: the patient populations and the clinical practices that establish the diagnosis of CMV IP might be different~; there may be differences in levels ofimmunodepression; or the presentation of CMV IP among this patient population might include forms of IP with various degrees of severityJ 2 Further clinical and laboratory markers for the evaluation of prognosis of CMV IP are not yet available. To assess the severity of CMV disease, a score system was developed and applied by SA Plotldn et aI. 13'14 to renal transplant recipients randomized to receive Towne strain CMV vaccine or placebo; vaccinated patients had less severe CMV disease, with a significant reduction in disease score when compared with patients who received the placebo. 14 In order to verify whether this clinical score system may predict the severity and outcome of CMV IP in BMT recipients, we retrospectively applied the score system 14 to 20 BMT patients at the estimated date of onset of IP and 1 0 - 1 4 days after the onset, assuming that: (a) the clinical features of severe CMV disease (i.e. fever, leucopenia, thrombocytopenia, superinfections) © 1997 The British Societyfor the Study of Infection

118

G. Gentile et aL

Table I. Scoring system for severity of cytomegalovirusdisease. Clinical manifestations

Points

Fever (temperature, >38.3°C (101 F) orally or rectally) Mild (2-4 days) Moderate (5-20 days) Severe (>21 days) Leucopenia (<4 x 109/1) Thrombocytopenia (< 100 x 109/I) Hepatitis Elevation of liver enzyme levels > two times normal Jaundice Pneumonia associated with cytomegalovirus Infiltrate on roentgenogram Infiltrate and symptoms Ventilator support Gastrointestinal bleeding or ulceration proved by biopsy specimen to be cytomegalovirus-related Central nervous system change Lethargy Stupor Coma Renal insut~ciency Creatinine level two to four times above the best level after transplantation Creatinine level > four times above the best level after transplantation Nephrectomy, permanent dialysis Arthritis or muscle wasting Superinfection Bacterial, protozoal, fungal Death

1 1 3 1 1 1 3 1 2 3 3 1 2 3 1 2 3 2 3 4

Plotkin SA, Starr SE, Friedman HM et al. Ann Intern Med 1991; 114: 525-531.

were present during the course of CMV IP; (b) severity of pneumonia was not the same w h e n diagnosis was made; (c) interstitial pneumonia did not progress at a similar rate in all patients.

Patients and Methods Patients

Charts of 278 consecutive patients who were submitted to allogeneic BMT at the Department of Haematology of the University 'La Sapienza' of Rome between July 1983 and December 1 9 9 4 were retrospectively evaluated; we found 20 patients with CMV l-P, and they were included in this study. The severity of CMV IP was analysed according to the individual manifestations listed in Table I, and clinical scores were given to each patient who did not receive

antiviral therapy (Fig. 1) and to each patient who received ganciclovir (Fig. 2) or ganciclovir and i.v. immunoglobulin (Fig. 2) at the onset of CMV IP and after 7 - 1 4 days. To receive any points, patients had to have at the estimated date of onset of interstitial pneumonia also fever, leucopenia, and thrombocytopenia. For superinfections to be scored they had to be bloodstream or deep organ infections. The remaining clinical manifestations had to be present during the course of CMV IP (Table I). Interstitial pneumonia was defined as follows: evidence of hypoxemia (arterial Po2<60 m m H g at room air), respiratory alkalosis (arterial PCO2<30 m m H g at room air), and non-lobar pulmonary infiltrate demonstrated radiographically in the absence of congestive heart failure, is Cytomegalovirus interstitial pneumonia was diagnosed by the growth of the virus from lung tissue or bronchoalveolar lavage (BAL) in low-passage-level MRC-5 cells examined regularly for viral cytopathic effect for up to 6 weeks; or by the detection of CMV antigens in alveolar macrophages and epithelial cells using monoclonal antibodies to both immediate-early (MAb CCH2, Dako, Glostrup, Denmark and MAb L 14, Ortho Inc., Raritan, New Jersey, U.S.A.) and late (MAb XP1, Behring, Marburg, Germany) antigens; or by the detection of CMV early (MAb CCH2) antigens in infected MRC-5 monolayers after centrifugation and co-culture with BAL. Pneumocystis carinii IP was diagnosed when P carinii cysts were identified by methenamine-silver staining or by indirect immunofluorescence with monoclonal antibodies in BAL. BAL were cultured also for bacteria, fungi and viruses by standard methods. Cytomegalovirus interstitial pneumonia was diagnosed by BAL in 19 patients, and by pulmonary biopsy in one patient. Cytomegalovirus serology (IgG, IgM) was evaluated for all recipients and before transplant for m a r r o w donors (Enzygnost Cytomegalie IgG, IgM, Behring, Germany). Cut-off of positivity was calculated according to m a n ufacturers' instructions. Surveillance cultures for the virus were taken from urine and buffy coat, weekly until day + 100 and at least once a m o n t h later. Specimens were immediately inoculated onto monolayers of h u m a n fibroblasts, tested with immunofluorescence 4 8 - 7 2 h later (MAb CCH2, Dako, Glostrup, Denmark) and controlled for typical cytopathic effects twice a week for 6 weeks. From January 1991 onwards, CMV antigenaemia was investigated from polymorphonuclear leucocyte fractions taken from peripheral blood using monoclonal antibodies directed against the CMV proteins pp65 (Biotest, C10, C l l , AG Dreiech, Germany). The conditioning regimens administered to BMT recipients are reported in Table II. The graft-versus-host

Score System for Cytomegalovirus Pneumonia

119

20

18-

16-

14-

12-

S 10 8-

6-

4-

Day 0

Day 10-14

Figure 1. Distribution of CMV disease scores among patients who did not receive antiviral therapy. O Patients Q, R, S, T, U, V died during IP.

disease (GVHD) prophylaxis consisted of methotrexate alone or in combination. Total body irradiation was given at 7.5 cGy/min from dual opposing 60 Co sources. The dose-fractionation of 200 rad twice a day for 3 consecutive days was generally given. Viral prophylaxis (oral aciclovir, 200 mg every 8 h) was given to BMT recipients starting in July 1983; treatment with aciclovir was initiated 7 days before BMT infusion and was discontinued after i year. Bone m a r r o w recipients were given trimethoprim (160 mg) and sulphamethoxazole (800 mg) twice a day for 3 consecutive days a week as prophylaxis of R carinii pneumonia; treatment was started when polymorphonuclear count was > 5 0 0 m m s, and was discontinued 12 months after bone m a r r o w infusion. Bone m a r r o w patients received bacterial prophylaxis with oral quinolones (norfloxacin or ciprofioxacin, 4 0 0 m g and 5 0 0 m g , twice daily, respectively); oral amphotericin B (2000 rag/day) as fungal prophylaxis was given from January 1983 to December 1989, and oral amphotericin B (same dosage) or oral fluconazole (150 rag/day) were given from January 1990 onwards.

No prophylactic anti-CMV immunoglobulin regimen was used. The two-tailed Fischer's exact test was employed for assessment of the significance of interval data. Sensitivity, specificity, negative predictive value and positive predictive value were calculated, as described by Griner et al. 16

Results The characteristics of all 20 patients are reported in Table II; they all received grafts from HLA-identical donors. The 20 recipients were seropositive for CMV, as well as 15 donors. All but one patient had previous episodes of viraemia and/or antigenaemia. Seven patients received GCV, seven patients received GCV plus i.v. immunoglobulin, and six patients received only supportive care because at the time of CMV IP ganciclovir therapy was not yet available. There were no survivors among the six patients who got supportive care only, while nine out of 14 patients

120

G. Gentile et aL 22

2o[ 18 16 14

12], g

O9

10 8 6 4 2

-A

0 Before t h e r a p y

After therapy

Figure 2. Distribution of CMV disease scores a m o n g patients receiving ganciclovir (GCV) and GCV +i.v. immunoglobulin. (A) Patients A, B, F and G, alive after 1770, 2370, 1800 and 2001 days, respectively; patient C died 180 days after IP, at autopsy no CMV. (/k) Patients D and E died after 18 and 34 days, respectively, during IP for respiratory failure; at autopsy CMV. (O) Patients L, M and N died after 9, 24 and 17 days, respectively, during IP for respiratory failure; at autopsy CMV. ( 0 ) Patients H, I and O survived to IP and died after 122, 154 and 123 days, respectively; at autopsy no CMV; patient P alive after 4 8 0 days.

Table II. Characteristics of patients. Characteristics

Mean age (range), year Sex (male/female) Disease, n CML ANLL ALL MS Acute GVI-ID grades 0 - 1 grades 2 - 3 Conditioning regimens TBI+CY TBI + etoposide Busulfan + CY Busulfan + etoposide

Treatment group Ganciclovir (n = 7)

Ganciclovir + immunoglobulin (n = 7)

No antiviral therapy (n = 6)

33.5 (17-46) 6/1

32 (17-47) 6/1

29.5 (13-54) 4/2

3 1 2 1

4 0 2 1

4 1 1 0

3 4

3 4

2 4

4 3 0 0

1 0 3 3

4 1 0 1

CIVIL= chronic myeloid leukaemia; ANLL=acute nonlymphoid leukaemia; ALL = acute lymphoid leukaemia; MS = myelodisplastic syndrome; GVHD = graft-versus-host disease; TBI = total body irradiation; CY = cyclophosphamide.

Score System for Cytomegalovirus Pneumonia receiving GCV with or without i.v. immunoglobulin survived the episode of IP (P = 0.011). Overall, 11 out of 20 patients who died of respiratory failure during IP had at the estimated onset of IP a score of >8, while of the nine patients who survived IP, eight had at the onset a score of <7 and the remaining one a score of 8 (P=O.O007). The sensitivity, specificity, predictive positive value and predictive negative value of the score system (with a threshold value at 8) to identify patients who survived IP was 100%, 88%, 91% and 100%, respectively. The score system is composed of clinical parameters, and we evaluated each factor individually at the estimated onset of IP as to their value in determining outcome. Superinfection (P--O.O03), and ventilator support ( P = 0.0004) were associated with survival of the 20 patients with CMV IP; whereas fever (P=0.6), leucopenia ( P = 0.9), thrombocytopenia (P=0.9), hepatitis (P=0.5), gastrointestinal and central nervous system involvement, renal insufficiency, arthritis or muscle wasting were not associated with survival of the patients with IP. The survival of patients who had an infectious agent identified, other than CMV in BAL, was zero of five patients, compared with nine of 15 in those who did not (P=O.1).

Patients who received gancidovir (GCV) For the seven patients who received GCV, the median time from BMT to diagnosis of CMV IP was 53 days (range 4:5-82 days) and the median time from estimated onset of CMV IP to treatment was 2 days (range 1-3 days). Seven patients (A, B, C, D, E, F, G) received GCV (2.5 mg/kg 3 times a day, for a range of 1 0 - 2 4 days). Patients D and N required ventilatory support before diagnostic procedure for CMV IP; patient D had a BAL positive for CMV plus P. carinii and died 18 days after initiation of therapy; at autopsy the lung was positive for CMV and P. carinii. Five patients (A, B, C, F, G) survived to CMV IP for a median of 1770 days (range 180--2370 days) and had scores which reached a maximum value of 7, whereas two patients (D, E) died from IP and had a score at presentation of IP of 8 and 9, respectively; during GCV therapy, both patients reached a score of 14 (Fig. 2).

121

was 2 days (range 1-4 days). Seven patients (H, I, L, M, N, O, P) received GCV (2.5 mg/kg 3 times a day, for a range of 9 - 2 4 days) and CMV hyperimmune globulin (four patients) (400 mg/kg, given every other day, for a range of 5-9 doses) or (three patients) standard i.v. immunoglobulin (500 mg/kg, for five doses). Patients I and N received a maintenance treatment with GCV (5 mg/ kg once a day, for 6 days a week for 3 months). Four patients (H, L, M, N) required ventilatory support before the diagnostic procedure for IP: three of them (L, M, N) died 9, 24 and 17 days, respectively, after treatment started. Bronchoalveolar lavage and/or tissue lung from autopsy (patients I, M and N) proved positive for CMV plus Aspergillus sp., CMV plus respiratory syncitial virus, and CMV, respectively. Four patients (H, I, O, P) survived for 122, 154, 123 and 480 days, respectively, after IP (at autopsy no CMV was found) and had scores which reached the maximum value of 8, whereas patients M, N and L died from IP and had at the presentation of IP a score of 9; their scores reached 16, 17 and 16, respectively, during therapy (Fig. 2). No relapse of CMV IP occurred in the five patients who survived to IP. For the 14 patients who received GCV alone or in combination the sensitivity, specificity, positive predictive value and negative predictive value of the score system was 100%, 88%, 83% and 100%, respectively.

Patients who received supportive care Six patients did not receive antiviral therapy; the median time from BMT to diagnosis of IP was 64 days (range 8 - 8 7 days). None of the patients required ventilatory support before the diagnostic procedure for IE Bronchoalveloar lavage proved positive for CMV plus Aspergillus spp. for patients U and V. The median score at the diagnosis was 10 (range 8-12), and 1 0 - 1 4 days later the median score was 15 (range 14--19) (Fig. 1); the six patients survived for a median time of 21 days (range 1 0 - 2 4 days) from the estimated onset of CMV IP; they died during IP of respiratory failure, and in all six patients at autopsy CMV was found in the lungs.

Discussion Patients who received gancidovir (GCV) and intravenous immunoglobulin For the seven patients who received GCV and i.v. immunoglobulin, the median time from BMT to diagnosis of CMV IP was 75 days (range 4 0 - 9 3 days) and the median time from estimated onset of CMV IP to treatment

The score system developed by SA Plotkin et aL 13.i4 was initially applied to provide objectivity to the determination of CMV disease severity of renal transplant recipients randomized to receive Towne strain vaccine or placebo; the authors considered that a score of 7 or more reflected a CMV disease of moderate severity. The same authors in another study 17 found that CMV disease was significantly

122

G. Gentile et aL

more severe in immune patients receiving a kidney from a seropositive donor (mean point score= 5.8) than the immune patients receiving a kidney from a seronegative donor (mean point score=2.29), P8 died of respiratory failure during IP: the difference is statistically significant (P8. The score system weights different parameters in an arbitrary fashion derived from empirical observation but not manipulated by statistical tests to assign appropriate weights;13.14 we observed that only the occurrence of the superinfection and ventilator support were associated with death in our patients. The score system was devised for renal transplant patients, and renal function was emphasized. 13 Overall, the clinical parameters used in the score system appear applicable also to BMT patients. However, the presence of GVHD might be a potentially confounding variable, and studies are necessary to evaluate the utility of other clinical parameters. Furthermore, the score system with a threshold value at 8 gave a high sensitivity, specificity, positive and negative predictive value to identify patients who survived IP. Although the number of patients studied is small, the score system applied in our BMT patients seems to show that at the time of diagnosis the severity of CMV IP itself is not necessarily the same, and that it does not progress at similar rates. The use of GCV with or without immunoglobulin was the most important determinant of outcome of IP in our patients, and the score system with a threshold value at 8 predicted the outcome of patients. Furthermore, the outcome of IP in patients receiving GCV alone or in combination was not influenced by the interval between the estimated onset of IP and the initiation of antiviral therapy; or by the interval between transplantation and the estimated onset of IP. Based on several studies with poor survival with CMV IP and studies showing benefit of prophylactic ganciclovir19 or high doses of acyclovir2° in preventing CMV IP, the focus in allogeneic bone marrow recipients is now to optimize protocols for early prevention and therapy of CMV prior to development of IP. 21 Nevertheless, despite

these improvements in patients' management, the control of CMV disease remains less than optimal, and CMV IP continues to occur. The concept of studying prognostic factors might be useful in designing randomized and controlled studies on therapy of CMV IP. It is important to mention that the use of GCV is loaded with several problems. The major complications associated with GCV are neutropenia and creatinine rise. 19'22 Neutropenia occurs approximately in 30% of patients 22'23 and it is significantly associated with bacterial infections. 2° Boeckh et al. 24 found an increased incidence of late CMV disease, mostly IP, associated with a high mortality, in patients who received GCV prophylaxis after allogeneic marrow transplantation. Cheng-Rong Li et al. 25 showed that CMV-specific T-cell responses necessary for protective immunity were delayed as a result of GCV prophylaxis, and that may contribute to the occurrence of late CMV disease in BMT recipients. As regards acyclovir, high doses of intravenous acyclovir followed by oral acyclovir improved survival and was of benefit in prophylaxis against the effects of CMV after BMT;2° however, CMV IP developed in six of these patients, even if none of them died of it. The patient population was not homogenous in relation to treatment and prophylaxis, which might have modified the clinical expression of CMV IP and consequently the score system. In conclusion, by retrospectively using a score system it seems feasible to differentiate patients who develop CMV IP according to different degrees of severity and outcome; if prospective and controlled studies will confirm our observations it would be possible to identify, by means of the score system, different severity and outcomes of clinically comparable CMV IP and thus properly evaluate the efficacy of the drugs employed for the treatment of the BMT recipients with CMV IP.

Acknowledgements The study was partially supported by the Progetto Finalizzato ACRO Consiglio Nazionale delle Ricerche, 'Applicazioni cliniche nella ricerca oncologica' (Unit P. Martino) N.92.02219.39. The authors thank Ruggero Raccah, MD, for his editorial assistance.

References 1 Dummer JS, Hardy A, Poorsattar A, Ho M. Early infections in kidney, heart, and liver transplant recipients on cyclosporine. Transplantation 1983; 36: 259-267. 2 Singh N, Dummer JS, Knsne Set aL Infections with cytomegalovirus and other herpes viruses in 121 liver transplant recipients: transmission by donated organ and the effect of OKT3 antibodies. J Infect Dis 1988; 158: 124-131. 3 Kusne S, Dummer JS, Stngh N e t al. Infections after liver transplantation: an analysis of 101 consecutive cases. Medicine 1988; 67: 132-143.

Score System for Cytomegalovirus Pneumonia 4 Blaser 1VII, Cohn DL. Opportunistic infections in patients with AIDS: clues to the epidemiology of AIDS and relative virulence of pathogens. Rev Infect Dis 1986; 8: 21-30. 5 Lerner CW, Tapper ML Opportunistic infections complicating acquired immune-deficiency syndrome. Medicine 1984; 63: 155164. 6 Meyers ]D, Flournoy N, Thomas ED. Nonbacterial pneumonia after allogeneic marrow transplantation: a review of ten years' experience. Rev Infect Dis 1982; 4: 1119-1132. 7 Winston DJ, Ho WG, Bartoni K et al. Ganciclovir therapy for cytomegalovirus infections in recipients of bone marrow transplant and other immunosnppressed patients. Rev Infect Dis 1988; 10 (Suppl 3): $547-$553. 8 Erice A, Jordan C, Chace BA, Fletcher C, Chinnock B], Balfour HH. Ganciclovir treatment of cytomegalovirus disease in transplant recipients and other immunocompromised hosts, lAMA 1987; 257: 3082-3087. 9 Reed EC, Bowden RA, Dandliker PS, Lilleby KE, Meyers ]13. Treatmerit of cytomegalovirus pneumonia with ganciclovir and intravenous cytomegalovirus immunoglobulin in patients with bone marrow transplant. Ann Intern Med 1988; 109: 783-788. 10 Emanuel D, Cunningham I, Jules-Elysee K et aL Cytomegalovirns pneumonia after bone marrow transplantation successfully treated with the combination of ganciclovir and high dose intravenous immune globulin. Ann Intern Med 1988; 109: 777-782. 11 Ljungman P, Engelhard D, Link H et aI. Treatment of interstitial pneumonitis due to cytomegalovirns with ganciclovir and intravenous immune globulin: experience of European bone marrow transplant group. Clin Infect Dis 1992; 14: 831-835. 12 Morris DJ. Ganciclovir and immune globulin in Cytomegalovirus pnemnonia. Ann Intern Med 1989; 110: 575. 13 Plotkin SA, Friedman HM, Fleisher GR et al. Towne-vaccine induced prevention of Cytomegalovirus disease after renal transplant. Lancet ]984; i: 528-530. I4 Plotkin SA, Stair SE, Friedman HM et al. Effect of Towne live virus vaccine on Cytomegalovirus disease after renal transplant. Ann Intern Med 1991; 114: 525-531.

123

15 Gentile G, Micozzi A, Girmenia C et al. Pneumonia in allogeneic and autologous bone marrow recipients. Chest 1993; 104: 371-375. 16 Griner PE, Mayewski RJ, Mushlin AI, Greenland P. Selection and interpretation of diagnostic tests and procedures. Ann Intern Med 1981; 94: 553-600. 17 Smiley ML, Wlodaver CG, Grossman RA et aI. The role of pretransplant immunity in protection from Cytomegalovirus disease following renal transplantation. Transplantation 1985; 40: 157161. 18 Wreghitt T. Cytomegalovirus infections in heart and heart-lung transplant recipients. J Antfmicrob Chemother 1989; 23: 49-60. 19 Goodrich JM, Mori M, Gleaves CA et al. Early treatment with ganciclovir to prevent cytomegalovirus disease after allogeneic bone marrow transplantation. N Engl J Med i991; 325: 1601-1607. 20 Prentice HG, Gluckman E, Pooles RL et al. Impact of long-term acyclovir on cytomegalovirus infection and survival after allogeneic bone marrow transplantation. Lancet 1994; 343: 749-753. 21 Forman SJ, Zaia JA. Treatment and prevention of Cytomegalovirus pneumonia after bone marrow transplantation: where do we stand. Blood 1994; 83: 2392-2398. 22 Goodrich ]M, Bowden RA, Fischer L, Keller C, Schoch G, Meyers JD. Prevention of Cytomegalovirus disease after allogeneic bone marrow transplant by ganciclovir prophylaxis. Ann Intern ivied 1993; 118: 173-178. 23 Schmidt GM, Horak DA, Niland JC et al. A randomized controlled trial of prophylactic ganciclovir for cytomegalovirns pulmonary infection in recipients of allogeneic bone marrow transplants. N Engl J Med 1991; 324: 1005-1011. 24 Boeckh M, Gooley T, Goodrich ], Sullivan K, Bowden RA. Increased incidence of late cytomegalovirus disease in patients who received ganciclovir prophylaxis after allogeneic marrow transplantation. The Eighth International Symposium on Infections in the immunocompromised host. Switzerland 1994, abstr. 060. 25 Cheng-Rong Li, Greenberg PD, Gilbert MJ, Goodrich IM, Riddell SR. Recovery of HLA-restricted cytomegalovirus (CMV) - specific T-cell responses after allogeneic bone marrow transplant: correlation with CMV disease and effect of ganciclovir prophylaxis. Blood 1994; 83: 1971-1979.