Donald W. Milligan, Department of Haematology, Birmingham Heartlands Hospital, Birmingham B9 5SS, or Suparno Chakrabarti, Department of Haematology, City Hospital, Dudley Road, Birmingham, B18 7QH, UK. E-mail: firstname.lastname@example.org or email@example.com
Summary. Respiratory virus infections can cause serious morbidity and mortality after conventional allogeneic stem cell transplantation. However, the incidence and outcome of these infections after reduced intensity conditioning has not been reported. Between 1997 and 2001, 35 episodes of respiratory virus infections were noted in 25 of 83 transplant recipients conditioned with fludarabine, melphalan and Campath-1H, and 80% of them received early antiviral therapy. Parainfluenza virus (PIV) 3 was the commonest isolate (45·7%) followed by respiratory syncytial virus (37%). Patients with myeloma were more susceptible to these infections [odds ratio (OR) 4·1, P = 0·01] which were often recurrent in patients with severe acute or chronic graft-versus-host disease (GVHD) (OR 10·6, P = 0·03). Infection within the first 100 d (OR 5·0, P = 0·05) and PIV 3 (OR 9·2, P = 0·01) isolation were risk factors for developing lower respiratory infection. Although more than half of the episodes progressed to lower respiratory infection, the mortality was only 8%. This could have been due to early initiation of antiviral therapy, but the attenuation of pulmonary damage due to the reduced-intensity conditioning, low incidence of GVHD and, paradoxically, the low CD4+ T-cell subset in this setting might also have been contributory factors.
Respiratory virus infections, mainly respiratory syncytial virus (RSV) and parainfluenza virus (PIV), have been reported to cause significant morbidity and mortality after allogeneic haemopoetic stem cell transplantation (SCT), following conventional conditioning with high-dose chemoradiotherapy. About 25–50% of the RSV and PIV infections can progress to cause pneumonia and, depending on the clinical setting, the mortality from pneumonia due to RSV and PIV varies between 30% and 100% (Wendt et al, 1992; Lewis et al, 1996; Whimbey et al, 1996, 1997; Bowden, 1997; Elizaga et al, 2001; Nichols et al, 2001). The outcome of antiviral treatment for pneumonia due to respiratory viruses has remained dismal, although uncontrolled studies suggest that early treatment with ribavirin might be associated with improved outcome (Adams et al, 1999; Chakrabarti et al, 2000, 2001a,b; Ghosh et al, 2000).
The use of high-dose chemoradiotherapy in the conditioning regimen precludes the application of this potentially curative therapy to elderly patients and those with major organ dysfunction. Pre-clinical studies have demonstrated the feasibility of achieving donor engraftment using minimal conditioning (McSweeney & Storb, 1999). The reduction in the intensity of pretransplant conditioning in clinical studies, using various non-myeloablative conditioning schedules, has been shown to achieve similar engraftment with reduction in regimen-related toxicities (Slavin et al, 1998; Barrett & Childs, 2000; Giralt et al, 2001; McSweeney et al, 2001; Chakrabarti et al, 2002). It is currently unknown whether the incidence or outcome of respiratory virus infections following non-myeloablative transplants will be different from that following conventional transplants.
In a multicentre study from United Kingdom, we have reported a low incidence of GVHD, but a high incidence of cytomegalovirus (CMV) infection with the addition of Campath-1H in vivo to the conditioning regimen of fludarabine and melphalan (Chakrabarti et al, 2002). We report here the incidence and outcome of respiratory virus infections in 83 patients from two centres, conditioned with this schedule, the majority of whom received early treatment with antiviral agents.
Patients and methods
Patients. Eighty-three patients underwent non-myeloablative conditioning in two major collaborating centres between June 1997 and August 2001. The eligibility criteria for receiving non-myeloablative conditioning have been previously described (Chakrabarti et al, 2002), and local ethics committees of both the participating centres approved the study design.
Conditioning regimen. Conditioning treatment consisted of Campath-1H (the IgG1-humanized monoclonal antibody against CD52) 20 mg/d on d −8 to −4; fludarabine 30 mg/m2 from d −7 to −3 and melphalan 140 mg/m2 on d −2. Cyclosporin A was started from d −1 as GVHD prophylaxis. Patients received unmanipulated peripheral blood stem cells from a matched family donor or unmanipulated bone marrow from an unrelated donor (UD).
Donor lymphocyte infusion (DLI) was not a part of the protocol and was at the discretion of the treating physician. The usual indications were persistent mixed chimaerism or worsening donor chimaerism, and persistence, progression or relapse of the disease that warranted the transplant.
CMV surveillance. Patients at risk of CMV infection were monitored weekly from the start of conditioning treatment to 100 d post transplant by a qualitative PCR assay and pre-emptively treated with ganciclovir 5 mg/kg b.i.d. or foscarnet 180 mg/kg in three divided doses or at half the above doses for both drugs if used in combination (Chakrabarti et al, 2002).
Virology. A naso–pharyngeal aspirate (NPA) and/or sputum sample were collected from any patient with acute upper and lower respiratory illness and sent for virological investigation. In addition, patients with severe lower respiratory symptoms and respiratory failure, requiring mechanical ventilation, underwent a bronchoscopy and broncho–alveolar lavage (BAL), and specimens were virologically investigated. Chest radiography was performed on all symptomatic patients.
NPA, sputum and BAL specimens were examined for respiratory viruses by direct immunofluorescence and viral culture. Direct immunofluorescence (DIF) was carried out using fluorescein isothiocyanate-conjugated specific monoclonal antibodies (Dako Diagnostics, Ely, Cambridgeshire, UK) for PIV 1, 2 and 3, influenza A and B, and RSV. Specimens for culture were inoculated into primary rhesus monkey kidney cells, human embryonic lung cells or MRC5 cell lines, and in some cases into primary liver carcinoma cells (Bio Whittaker, Wokingham, UK).
Treatment. Infections with PIV 3 and RSV were treated with ribavirin in an early initiation schedule. Asymptomatic or mildly symptomatic ambulatory patients were treated with oral ribavirin (15 mg/kg to 45 mg/kg in incremental doses) at one centre as a pilot study (Chakrabarti et al, 2001b). Otherwise, symptomatic patients were treated with aerosolized ribavirin (6 g/d in three divided doses for 5–7 d). Those with probable or definite lower respiratory illness (LRI) not responding to oral or aerosolized ribavirin were treated with intravenous (i.v.) ribavirin at 60 mg/kg/d in four divided doses for 5–10 d at one of the centres (Chakrabarti et al, 2001b). Ribavirin (i.v.) was also used as the first-line therapy if nebulized ribavirin could not be used as a result of logistical problems.
Influenza infections were treated with amantadine 100 mg b.i.d. and replaced by zanamivir inhaler 10 mg b.i.d. in cases of non-response or disease progression.
Definitions. Upper respiratory illness (URI) was defined as the acute onset of any rhinorrhoea, sinusitis, pharyngitis or cough without clinical or radiological evidence of lower respiratory tract involvement and/or hypoxia, combined with the detection of the virus in upper respiratory secretions. Lower respiratory illness (LRI) was defined as clinical signs and symptoms of lower respiratory involvement with radiological evidence of new pulmonary infiltrates with or without hypoxia associated with the detection of the virus in NPA or BAL specimens (Ljungman et al, 2001). Bronchoscopy and BAL were only carried out in patients with respiratory failure requiring ventilation. If co-pathogens were detected and a BAL specimen was not available, the LRI was defined as probable. The remainder, fulfilling the above criteria, were defined as definite LRI. Nosocomial infections were defined as infections diagnosed in patients hospitalized for at least 7 d.
Clinical response was defined as disappearance of respiratory symptoms and resolution of pyrexia, and improvement of radiological abnormalities, if any. Virological response was defined as a negative result on both DIF and culture on two consecutive adequate respiratory samples.
An episode was defined as the duration of an infection attributable to a respiratory pathogen, either URI or LRI, and the duration was considered to be from the onset of clinical symptoms to a virological response. Simultaneous infection with two or more different viruses was considered as a single episode. To qualify for a second episode with the same or a different virus, complete absence of clinical signs or symptoms and at least two negative virology tests had to be documented over 14 d.
A co-pathogen was defined as a bacterial, fungal or a non-respiratory virus pathogen isolated from the respiratory tract or blood culture during the episode of respiratory virus infection.
Supportive care. Antimicrobial prophylaxis consisted of aciclovir and fluconazole or itraconazole from the beginning of conditioning treatment. Pneumocystis carinii prophylaxis was given to every patient with co-trimoxazole 480 mg every 12 h three times weekly, after the absolute neutrophil count exceeded 1·0 × 109/l. Febrile neutropenic patients were treated with broad-spectrum antibiotics, according to the institutional policy of each centre. Granulocyte colony stimulating factor (G-CSF) was administered subcutaneously at the discretion of the transplant physician to hasten neutrophil engraftment. Irradiated red cell and platelet transfusions were administered to maintain haemoglobin and platelets above 9 g/dl and 10 × 109/l respectively.
Estimation of T-lymphocyte subsets. Peripheral blood was collected in EDTA for T-lymphocyte subset analysis at least once every 3 months post transplant. T-lymphocyte subsets were analysed by flow cytometry using a FACScan and simulset software (Becton Dickinson, Oxford, UK), as previously described (Chakrabarti et al, 2001a). The absolute lymphocyte count on the day of obtaining a positive respiratory isolate was considered for statistical analysis.
Statistical calculations. Univariate P-values and odds ratios were calculated from 2 × 2 contingency tables, using epi info version 6 (CDC, Atlanta, USA). Continuous variables were analysed using the non-parametric Mann–Whitney U-test. A multiple logistic regression model was fitted to the data using a stepwise approach and the statistical package for the social sciences (SPSS) version 9 for windows (SPSS UK, Worthing, UK) for risk factor analysis. Cumulative probabilities and survival were analysed by the Kaplan–Meier method, censoring for competing risk factors as necessary, and the difference between the groups was compared using the log rank x2 test.
Detailed characteristics of this patient cohort are shown in Table I. Briefly, 83 patients were studied for a median follow-up of 17 months (range, 2–52 months). The median age was 44 years (range, 18–59 years). The majority were transplanted for lymphoproliferative disorders or myeloma. Thirty-five patients had received previous transplants (34 autografts, one allograft). Thirty-three patients received a bone marrow graft from an unrelated donor (UD). Eighteen of these patients received an UD graft as a second transplant.
Table I. Characteristics of non-myeloablative transplant recipients with and without respiratory virus infection.
Twenty-five patients (30·1%) developed a respiratory virus infection with a cumulative probability of 35% at 18 months post transplant [six of these patients have been reported previously from one of the centres (Chakrabarti et al, 2000, 2001a, b) in separate contexts]. Seven patients had multiple episodes of respiratory virus infection, five had two episodes, one had three episodes and another five episodes. The total number of episodes of respiratory virus infection in 25 patients was 35. The median time of onset of these infections was 123 d (range 1–561 d). Eight of the 35 episodes (22·8%) occurred within first 30 d post transplant with seven of them in the pre-engraftment period. Sixteen episodes (45·7%) occurred in the first 100 d. Nine episodes were nosocomially acquired. Co-pathogens were isolated in eight episodes (aspergillus two, pneumococcus two, Haemophilus influezae one, moraxella 1, enterococcus 1, adenovirus 1).
PIV 3 was the commonest isolate accounting for 16/35 episodes (45·7%), followed by RSV with 13/35 episodes (37%). Influenza A and B accounted for five episodes and PIV 1 accounted for one episode. The characteristics of individual virus infections are detailed in Table II. PIV 3 infections occurred earlier, persisted longer and had a higher incidence of LRI (Table II).
Table II. Outcome of respiratory virus infections according to the infectious aetiology.
Parainfluenza (episodes = 17)
RSV (episodes = 13)
Influenza (episodes = 5)
Median onset, days (range)
Lower respiratory illness
Infection within 30 d post transplant
Infection within 100 d post transplant
Related to GVHD
Median treatment duration (range)
Median duration of virus detection, days (range)
Risk factors for respiratory virus infection
Age, sex, donor status, underlying disease, previous transplant, CMV sero-status, CMV reactivation and concurrent GVHD were variables considered in univariate analysis. Myeloma as the underlying disease (Fig 1) was the only significant risk factor for respiratory virus infections [9/16 compared with 16/67 with other diseases, P = 0·01, OR 4·1 95% confidence interval (CI) 1·1–15].
Respiratory virus infections occurred in 5/11 patients with severe (grade 3–4 acute or extensive chronic) GVHD, compared with 20/72 patients with less severe or no GVHD (P = 0·1). However, of the infected patients, multiple episodes of respiratory virus infection were more common amongst patients with severe GVHD (4/6 compared with 3/19 without severe GVHD, P = 0·03, OR 10·6, 95% CI 1–150).
Lower respiratory illness and risk factors
All but one patient had URI symptoms. LRI (both probable and definite) was documented in 19/35 episodes (54·2%). Eleven episodes (31%) fulfilled the definition of definite LRI. Four had respiratory failure requiring ventilation and all had a respiratory virus detected in BAL specimens. The rest of the patients (7/11) met the definition of definite LRI by fulfilling clinical and radiological criteria along with a positive respiratory virus isolate in NPA and/or sputum samples in the absence of co-pathogens. Other patients who met these criteria were categorized as probable LRI because of the presence of co-pathogens and the lack of BAL samples.
The median onset of LRI was 53 d as compared with 195 d for those with URI alone (P = 0·03). There was no age or sex predilection for the patients developing LRI. There was no relationship to GVHD or donor type. More episodes with nosocomial infection were associated with LRI (7/9 vs 11/26 in community-acquired infections, P = 0·1), but this was not statistically significant.
The median lymphocyte count at the time of respiratory virus isolation was 0·6 × 109/l (0–3·2 × 109/l) and there was no difference in the lymphocyte count between episodes of LRI and URI alone. The serial T-cell subset estimation was available in 49 patients and 19 of them had respiratory virus infection. The median CD4+ T-cell count for the entire cohort was 0·06 × 109/l and 0·16 × 109/l at 3 and 6 months respectively. However, there was a trend towards higher CD4+ T-cell counts at 3 and 6 months post transplant amongst patients with LRI (Table III). There was no relationship between LRI and neutrophil engraftment.
Table III. Lower respiratory tract illness due to respiratory virus.
Respiratory virus infection before 100 d post transplant was associated with a greater risk of developing LRI (12/16 episodes vs 7/19 episodes in infections beyond 100 d, P = 0·04). Infection with PIV 3 (Fig 2) also had a higher risk of developing LRI (13/16 episodes vs 6/19 episodes with other viruses, P = 0·01). A multivariate analysis of risk factors for LRI was carried out with nosocomial infection, time of onset (before or after 100 d) and causative virus (PIV 3 or others) as variables. PIV 3 infection (P = 0·01, OR 9·2, 95% CI 1·7–50·9) was the most significant risk factor in the model, followed by onset of infections before 100 d (P = 0·05, OR 5·0, 95% CI 1·0–27).
Twenty-eight (80%) episodes were treated with antiviral drugs. All episodes of LRI and 9/16 (56·2%) episodes of URI alone were treated. Antiviral therapy was initiated early (mean ± SD 3·7 ± 4·0 d) after the onset of symptoms and within 24–36 h of virus detection in all but one episode. Fourteen of the 16 episodes of PIV 3 infection and 12/13 RSV infections were treated with ribavirin.
Treatment was initiated with oral ribavirin in five episodes (Chakrabarti et al, 2001b), i.v. ribavirin in six episodes and aerosolized ribavirin in the rest. Two patients received i.v. ribavirin as a result of failure of aerosolized ribavirin and three patients were switched from i.v. to aerosolized ribavirin when the nebulized formulation became available. Two patients with influenza virus infection received amantadine and one of them received zanamivir inhaler because of progression of symptoms on amantadine. Seven episodes (PIV 3, two; RSV, one; PIV 1, one; influenza, three) were not treated because of mild symptoms and all were self-limiting.
The median duration of antiviral therapy was 5 d (range 5–40 d). The duration of treatment was longer for LRI episodes (median 9·5 d vs 5 d with URI alone, P = 0·03). Treatment was associated with partial or complete clinical response in all but three patients. Two patients with PIV 3 infection died of pneumonia and the third patient had symptomatic isolation of RSV for 6 weeks. However, virological persistence was protracted and virological response coincided with clinical improvement in 10 episodes only. The median time to virological response was 15 d (range 7–95). This was significantly delayed for those with LRI (median 21 d vs 8·5 d for URI alone, P = 0·005) and PIV 3 infections (median 20·5 d vs 12 d for other viruses, P = 0·04).
Outcome and survival
Four patients required mechanical ventilation. There were only two deaths (8%) attributable to LRI due to respiratory viruses. A 50-year-old patient, receiving a sibling graft for lymphoma, acquired PIV 3 infection 10 d post transplant and died 10 d later without engraftment. The other mortality was in a 53-year-old recipient of a sibling graft for myeloma who was infected with PIV 3, 43 d post transplant, and died 20 d later. Both patients received ribavirin but died of progressive respiratory failure. Fungal and bacterial co-pathogens were identified in both patients.
The mortality at 100 d post transplant was 12% for patients with respiratory virus and 13·8% for those without it. The non-relapse mortality (Fig 3A) was 41% (95% CI 5·8–76·2) amongst patients with respiratory virus infection compared with 25·4% (95% CI 13·7–37·1) in those without it (P = 0·8). The overall survival (Fig 3B) was 50·7% (95% CI 18·2–83·2) and 57·5% (95% CI 43–72) amongst those with and without respiratory virus infection respectively (P = 0·8).
In our study, the probability of respiratory virus isolation in patients conditioned with this Campath-1H-based regimen was 35%. While such a high incidence can reflect the intensity of the surveillance, that alone surely cannot explain these findings. The factors determining the propensity to develop severe infection with these viruses in the post-transplant period is not well understood, and the relative contribution of humoral and cellular immunity to this process is unclear (Couch et al, 1997). We have previously reported the relationship between the risk of respiratory virus infection and the use of in vivo Campath in both conventional and non-myeloablative transplant recipients (Chakrabarti et al, 2001a). The cumulative incidence of respiratory virus infection in patients receiving Campath-1H 100 mg in vivo in that study was 43% which is comparable to this cohort of patients who received a similar dose of Campath. Campath-1H persists for several weeks in the circulation when used in vivo (Rebello et al, 2002) and this, in addition to T-cell depletion of the graft, may also impede immune recovery in the early post-transplant period. This was reflected in the delayed recovery of CD4+ T cells in this cohort of patients.
Our study has shown that patients with myeloma were at a greater risk of respiratory virus infections. This was independent of age, donor type or previous transplants. RSV infections did not cause serious morbidity in myeloma patients undergoing autologous transplants (Aslan et al, 1999). Previous reports on respiratory virus infections in allogeneic transplant recipients included very few myeloma patients, if any. Our findings could be explained by the anti-B-cell effect of Campath antibodies, compounding the pre-existent humoral immunodeficiency in myeloma patients over and above the severe T lymphopenia caused by Campath, as described above.
The incidence of LRI was 51% in this study. Infections occurring before 100 d, particularly with PIV 3, were more likely to be associated with LRI which is much higher than the previously reported rates of 24–50%. PIV 3 infections are probably under-reported as surveillance for respiratory viruses is often limited to winter seasons, whereas PIV 3 infection tends to occur round the year in North America (Wendt et al, 1992; Nichols et al, 2001) and in late spring or summer in United Kingdom (Laurichesse et al, 1999). Secondly, half of the cases of PIV 3 occurred in the summer outbreak of 2000 and thus might have been over-represented. Finally, the majority of the PIV 3 infections occurred in the early post-transplant period when T-cell immunity is poorly reconstituted, which resulted in protracted and progressive infections. The other possibility is that PIV 3 is less susceptible to ribavirin, as suggested in earlier studies (Chakrabarti et al, 2001b; Elizaga et al, 2001; Nichols et al, 2001).
The key finding of this study was a very low mortality associated with respiratory virus infections. Overall mortality related to respiratory viruses was only 8%. In the setting of conventional transplants, progression to LRI and associated mortality was 50–60% in the early post-transplant period (Wendt et al, 1992; Lewis et al, 1996; Whimbey et al, 1996). On the other hand, 22% of the patients developed PIV 3 or RSV infections within the first 30 d and 45% by d 100 with a mortality of only 12·5%.
There may be a number of reasons for such a favourable outcome. We initiated antiviral therapy very early in the course of the infections. Although uncontrolled data suggest that earlier initiation of ribavirin might be associated with an improved outcome (Adams et al, 1999; Chakrabarti et al, 2000, 2001a,b; Ghosh et al, 2000), it is difficult to attribute this to the antiviral therapy alone. Ribavirin therapy was associated with partial or complete clinical response in the majority of episodes, but virological response was achieved only in a third of them.
Pneumonia due to paramyxoviruses does not result from viral cytopathology alone but is also due to inflammatory lung damage (Hertz et al, 1989). The high patient fatality rate of respiratory virus in the early post-transplant period following conventional transplants may be caused by a combined effect of chemoradiotherapy-induced lung damage, release of pro-inflammatory cytokines and the inflammatory milieu caused by GVHD. None of these factors were present in our cohort as a result of the reduced intensity of conditioning and a low incidence of GVHD. In animal models of RSV infection, CD4+ T cells were more immunopathogenic than CD8+ T cells in terms of lung injury, despite being most potent in antiviral efficacy (Alwan et al, 1992). In children with congenital T-cell deficiencies (Crooks et al, 2000) or in those with human immunodeficiency virus infection (King, 1997), RSV or PIV infection is persistent without serious pneumonia, which often results following post-transplant engraftment in the former, highlighting the importance of T cell-mediated mechanisms in the genesis of RSV- or PIV-induced lung damage. The low number of CD4+ T cells in our cohort most likely predisposed to a high incidence of respiratory virus infections, but might have paradoxically played a role in attenuation of the pulmonary pathology in the absence of other pro-inflammatory factors, similar to that hypothesized for CMV pneumonia (Grundy et al, 1987).
In conclusion, although a high incidence of respiratory virus infections was noted in this cohort of patients with prolonged impairment of T-cell immunity, the mortality was low. Further studies are necessary to explore the impact of various reduced-intensity conditioning regimens on the outcome of respiratory virus infections.
We thank the staff of the Therapeutic Antibody Centre, University of Oxford for their contributions to the production of Campath-1H antibody. Their work was supported by the United Kingdom Medical Research Council, Leukosite Inc., and the E.P. Abraham's Trust.