Non-endocrine late complications of bone marrow transplantation in childhood: part I.

Abstract

Current estimations indicate that greater than 30 000 allogeneic and autologous haematopoietic stem cell transplants (HSCT) are undertaken every year, worldwide, and that this figure is rapidly growing. Approximately one-fifth are performed in paediatric patients and it is estimated that a minimum of 1500–2000 of these annually will become long-term survivors (Boulad et al, 1998). Overall, autologous transplants, including those using peripheral blood stem cells, are commoner than allogeneic HSCT (Boulad et al, 1998; Horowitz, 1999), but in children the donor source is still mainly allogeneic marrow. This is confirmed by unpublished analysis of British paediatric data, taken from the United Kingdom Childrens' Cancer Study Group Registry between 1993 and September 2000. Out of the 2331 transplantation procedures performed, 1700 were allogeneic. The commonest indication for allogeneic transplantation in children is leukaemia, with 62% undergoing bone marrow transplantation (BMT) for this or myelodysplasia, 36% for non-malignant conditions and 2% for other cancers. With increasing numbers of long-term survivors, delayed complications, often presenting years after BMT, are becoming a concern. Late sequelae may arise as a result of the disease for which transplantation was performed or from toxicity associated with the wide variety of conditioning regimens. Most of the latter will include high-dose chemotherapy, alone or accompanied by radiotherapy, in the form of total body irradiation (TBI) or total lymphoid irradiation (TLI) and/or agents to effect T-cell depletion. The total dose of TBI usually varies from 7·5 Gy given as a single fraction up to 15 Gy given in multiple fractions over a period of 3–4 d. Additional damage may be sustained through the toxic effects of some antibiotics and antifungals, or from immunosuppressive agents used for prevention and treatment of graft-versus-host disease (GVHD) or from the pathological process of chronic GVHD itself. Other potential variables influencing the impact of late sequelae are the total dose, dose rate and method of fractionation of the radiotherapy, the age and sex of the child, and genetic influences. Particularly difficult to disentangle from sequelae related to the BMT procedure itself, but integral to the outcome, is the effect of previous treatment, especially in the case of malignant disease. This review would be incomplete without relevant discussion of some of these important aspects. As many of the childhood studies are very small and in themselves inconclusive, larger, mixed studies and some adult data have been included when I have felt it to be important to the evidence. Part I of this review describes cardiac, pulmonary, renal, neurological and neuropsychological late sequelae of bone marrow transplantation. The second part deals with ocular, audiological, dental, salivary and skeletal delayed complications, second malignant neoplasms, overall morbidity, late mortality and quality of life. The endocrine system is particularly vulnerable to damage by radiation and alkylating agents in the growing child (Leiper et al, 1987; Boulad et al, 1998; Cohen et al, 1999), but endocrinopathy will not be included here. Deeg, Sanders, Kolb and Bender-Gotze have reviewed late complications of BMT in adults and children, covering the early period of transplantation up to the end of the eighties (Sanders et al, 1989; Deeg, 1990; Kolb and Bender-Gotze, 1990; Sanders, 1990; Bender-Gotze, 1991). This review focuses mainly on paediatric literature published over the last decade. Survivors of childhood cancer represent one of the largest new groups at risk of premature cardiovascular disease (Lipshultz & Sallan, 1993). Many of the multimodal agents used in attempts to cure cancer are now shown to be cardiotoxic and, in the transplant setting, a constellation of adverse influences may injure the heart. These include electrolyte disturbance and sepsis (Parrillo, 1985; Martino et al, 1990), pre-existing iron overload in thalassaemia (Mariotti et al, 1993), radiation (Donaldson & Kaplan, 1982; Arsenian, 1991; Hancock et al, 1993; Jakacki et al, 1993), although less likely at lower doses of 15–26Gy (Donaldson & Kaplan, 1982; Hancock et al, 1993), and antineoplastic agents used both in the initial treatment of malignancy and in the conditioning regimen. For example, damage caused by anthracyclines in the conventional treatment of childhood leukaemias may manifest as acute or late cardiotoxicity (Lipshultz et al, 1991; Sorensen et al, 1997; Nysom et al, 1998a; Levitt, 1999), and acute toxicity from cyclophosphamide is reported in adults (Santos et al, 1970; Mills & Roberts, 1979; von-Bernuth et al, 1980; Braverman et al, 1991) and children following BMT (Steinherz et al, 1981; Shaw et al, 1986). Other commonly used chemotherapeutic drugs, especially the alkylating agents, are also cardiotoxic (Vaickus & Letendre, 1984; Bearman et al, 1990; Kanj et al, 1991; Pihkala et al, 1994) and the concomitant or sequential use of these, especially cyclophosphamide (Steinherz et al, 1981) or radiation (Billingham et al, 1977), may augment anthracycline damage. Thus, it becomes hard to separate the contribution to cardiotoxic injury by previous chemoradiotherapy from that posed by the BMT procedure itself. Data regarding late cardiotoxicity in children after BMT is sparse, with no large studies (Uderzo et al, 1991; Larsenet al, 1992a; Liesner et al, 1994; Pihkala et al, 1994; Rovelli et al, 1995; Thuret et al, 1995; Eames et al, 1997; Leahey et al, 1999). Generally, late cardiotoxicity may present in a number of ways depending on the type and combination of cardiotoxic agent. These include clinically significant congestive cardiac failure, arrhythmias, fatal cardiomyopathy, pericardial and valvular disease, non-specific changes or reduced total QRS voltage on electrocardiogram (ECG), or asymptomatic reduction of left ventricular fractional shortening on echocardiography (ECHO) (Larsen et al, 1992b; Lipshultz & Sallan, 1993; Steinherz et al, 1995). The incidence is therefore is difficult to define, and most of the general overview studies of late sequelae after BMT in paediatric patients report sporadic cases of clinically significant cardiomyopathy, or small numbers of patients with abnormal left ventricular fractional shortening or reduced contractility on ECHO (Uderzo et al, 1991; Michel et al, 1997; Leahey et al, 1999). However, one study found no cardiac abnormalities in 13 patients with acute leukaemia transplanted in first remission (Thuret et al, 1995), while as many as 28% of a small number of BMT patients (8) with acute myeloid leukaemia (AML) and myelodysplasia, treated at Great Ormond Street Hospital for children, London, UK, had reduced fractional shortening on ECHO (Liesner et al, 1994). These children were included in a larger cohort of 83 patients transplanted for leukaemia since 1980 at our centre with minimum survival of 3 years (median 7·9 years; 3·5–16·9 years). Twenty-four out of 78 evaluated (31%) had abnormal fractional shortening (< 28%) post transplant. This had deteriorated in at least half the patients from pretransplant values and almost all had received TBI. Clinically significant cardiomyopathy with cardiac failure developed in 3·5%. At follow up, 13 patients had a shortening fraction of < 25% (unpublished data, Pitcher et al, 1998). The Great Ormond Street Hospital long-term study of 33 patients with AML and myelodysplasia (8 BMT), and another of 52 (26 BMT) with the same diagnosis, carried out in Philadelphia, compare late sequelae after BMT with those of chemotherapy alone, with follow up between 1 and 15·5 years (Liesner et al, 1994; Leahey et al, 1999). The BMT preparative regimens differed in that the British patients mainly received TBI, while the majority of American children received busulphan (Bu) and cyclophosphamide (Cy). In the two studies, all but one patient underwent BMT in first remission and most patients had received anthracyclines, in both the BMT and chemotherapy groups. Mean anthracycline dosage was generally higher in those who were not transplanted. Both studies reported that cardiac function in those undergoing BMT was no worse than with chemotherapy alone and, indeed, the mean shortening fractions in the comparison groups were within the normal range in both studies [although significantly reduced compared with normal subjects in Liesner et al (1994)]. However, the incidence of cardiac dysfunction was much higher in theLondon study with 28% of the BMT and 35% of the chemotherapy patients suffering an abnormal reduction in the shortening fraction (< 28%). This compared with 4% and 17% in the American study. Late cardiac failure occurred in two patients in each study 3–15 years later, none of whom had undergone marrow transplantation and all of whom had received high doses of anthracyclines. These findings suggest a role for pre-existing damage by anthracyclines, and some of the differences in incidence in the two studies could be partly explained by the added cardiotoxicity of TBI in Liesner's group, or possibly by differing methods of anthracycline administration. The incidence of an abnormal shortening fraction found by Leahey et al (1999) was comparable to that found by a French study of 45 similar patients transplanted in first remission using both types of conditioning chemotherapy (Michel et al, 1997), see also `note added in proof'. Recently, the subclinical nature of some cardiac defects have been further explored in longer term studies in both adults and children, using more complex techniques, such as radionuclide angiography, cardiopulmonary exercise testing or dobutamine stress echocardiography, in addition to standard methods (Larsen et al, 1992a; Mariotti et al, 1993; Carlson et al, 1994; Pihkala et al, 1994; Eames et al, 1997). In a cross-sectional study of 20 subjects studied prior to BMT and 31 others after BMT, cardiac performance was assessed during exercise, via cycle ergometry, and left ventricular size and fractional shortening, using resting echocardiography. Significant pulmonary limitations were excluded using spirometry. It was found that adults and children with oncological diagnoses had serious limitations in cardiac performance on exercise, after BMT (reduced exercise times, maximal oxygen consumption, anaerobic thresholds and cardiac output), although all had normal oxygen consumption and cardiac indices at rest, and few patients had ECHO abnormalities. However, impaired exercise performance was found even in those patients tested before BMT and there was no difference in this respect between patients studied before or after BMT, suggesting that previous conventional cancer therapy had contributed to cardiotoxicity (Larsen et al, 1992a). Pihkala's group from Helsinki, Rovelli's Italian group and Eames' group from Minneapolis have reported the first studies dedicated to assessing cardiac or cardiopulmonary function in long-term childhood survivors of BMT (Pihkala et al, 1994; Rovelli et al, 1995; Eames et al, 1997). The vast majority of patients in these three studies had received cyclophosphamide, anthracyclines (median dose 140 mg/m2, 250 mg/m2 and 370 mg/m2 respectively) and two-thirds or more in each study had received TBI or total lymph irradiation (TLI). All 42 patients in the Italian study were asymptomatic and had normal fractional shortening on echocardiography, before and up to 4 years post BMT (Rovelli et al, 1995). Acute, mainly reversible, ECG changes were seen after conditioning chemotherapy in this and the Finnish study (Pihkala et al, 1994; Rovelli et al, 1995). Pihkala et al (1994) undertook a retrospective study in 30 children undergoing autologous and allogeneic BMT for malignant and non-malignant disorders. Myocardial function was evaluated by a series of post-BMT investigations including ECG, chest radiography, radionuclide cineangiography and detailed echocardiography. All patients in this study were asymptomatic at the time of follow up. Overall, 15–25% of patients had evidence of late subclinical cardiotoxicity at a median follow up of 4·8 years (0·5–10·7 years) and several patients had subnormal function on more than one test. Thirteen per cent of those transplanted had abnormal contractility on ECHO, 26% of those tested had reduced left ventricular ejection fraction (LVEF < 50%) on radionuclide angiography, in addition to 24% with persistence of acute ECG changes (> 15% reduction of QRS voltage sum on ECG), all of whom had abnormal left ventricular systolic function (Pihkala et al, 1994). In an attempt to further determine cardiac function, long-term prevalence, severity and type of cardiac abnormality post BMT, and risk factors involved, Eames et al (1997) evaluated cardiac function cross-sectionally in 63 patients undergoing bone marrow transplantation at < 18 years of age. Evaluation consisted of review of past cardiac studies and assignation to a New York heart association (NYHA) class based on activity and cardiac symptoms, exercise and resting ECG, echocardiography, chest radiograph pulmonary function, and an exercise treadmill test, assessing a number of parameters of cardiopulmonary function. Overall, 41·3% had cardiac abnormalities detected at mean follow up of 3·3 years (1–16·3 years), which had not been present in the majority in the pre-BMT assessment. More than two-thirds of these defects were subclinical. The cardiac abnormalities were reduced LVEF on echocardiography in four patients and abnormalities of the resting ECG in 10 (16·4%), despite all but one having a normal shortening fraction. Abnormalities of LVEF were less common, although the incidence of cardiac dysfunction was generally higher than the 15–25% found by Pihkala et al (1994). As in the other studies, the majority of patients were asymptomatic. However, the most significant finding in this study from Minneapolis was the high prevalence of abnormalities on the treadmill exercise test (74% of those tested) which had unmasked cardiac dysfunction. Only 13% of patients in the study were symptomatic, all with treadmill test impairment, but there were no cases of cardiac failure or life-threatening cardiomyopathy (Eames et al, 1997). This was similar to the exposure of cardiac dysfunction by exercise reported by Larsen et al (1992a). As yet, there is no definite evidence in children that BMT worsens cardiac function. Our own unpublished data and that of some other authors suggest that it does (Pihkala et al, 1994; Eames et al, 1997), although Pihkala et al (1994) only had ECG data prior to BMT, while others do not (Larsen et al, 1992a, Rovelli et al, 1995). The Minneapolis group found no correlation between cardiac dysfunction, previous therapy, type of BMT regimen or length of follow up. Neither was there a correlation with high-dose cyclophosphamide or TBI fractionation (Eames et al, 1997). TBI was not found to worsen outcome in this or other studies (Rovelli et al, 1995; Eames et al, 1997; Michel et al, 1997), and there was no difference between Bu/Cy and Cy/TBI in the French study (Michel et al, 1997), or between single fraction versus fractionated TBI (Pihkala et al, 1994; Eames et al, 1997). However, Pihkala's group felt that TBI played an important role as QRS changes occurred in all those who received it, and that myocardial changes seemed worse after cyclophosphamide than with high dose Ara-C (Pihkala et al, 1994). Clearly, anthracyclines had an adverse influence on cardiac function in patients treated for AML and myelodysplastic syndrome (MDS) with either chemotherapy alone or BMT (Liesner et al, 1994; Leahey et al, 1999), and in previously treated patients with oncological diagnoses undergoing BMT (Larsen et al, 1992a). However, despite some adult studies suggesting an increased risk of cardiotoxicity post BMT with pretreatment with anthracyclines (Cazin et al, 1986; von Herbay et al, 1988) and a minor, but significant, change in shortening fraction after BMT in children who had received > 300 mg/m2 anthracyclines (Rovelli et al, 1995), few firm conclusions can be drawn about exacerbation of dysfunction by BMT. In conclusion, late cardiac dysfunction found after BMT is multifactorial in origin. Factors which may increase the risk for its development include previous anthracycline usage, TBI, cyclophosphamide, as well as other cardiotoxic drugs, but there is a clear need for prospective, longitudinal studies with large numbers for further elucidation. Damage is often subclinical and may not be picked up by routine methods of investigation, but by using exercise testing and more invasive techniques, such as radionuclide angiography, the damage can be exposed. Although acute cardiotoxicity may improve at least in the short term, cardiac decompensation may occur many years after treatment has finished. Life-long surveillance is needed therefore to ascertain the significance of these subclinical defects. The prognosis for survival following BMT is generally good, although dependent on the indication for BMT and disease status at the time. Nonetheless, pulmonary complications remain a significant cause of mortality and morbidity following allogeneic BMT (Krowka et al, 1985; Soubani et al, 1996; Palmas et al, 1998) and account for 10–40% of transplant-related deaths (Krowka et al, 1985; Breuer et al, 1993, Palmas et al, 1998). Fungal and cytomegalovirus (CMV) are the major post-BMT pulmonary infections, whereas idiopathic interstitial pneumonitis, a restrictive disorder, and the obstructive disorder bronchiolitis obliterans are the major non-infectious complications in children (Stokes, 1994). Interstitial pneumonitis, renowned for its early development (< 100 d) in 10–20% of allogeneic transplant recipients, may also have a late onset (> 100 d) in a minority (Breuer et al, 1993; Soubani et al, 1996). All may potentially contribute to long-term sequelae. Despite a profusion of literature reporting late-onset obstructive airways disease and restrictive pulmonary disease predominantly in adults reviewed by Soubani et al (1996) and Breuer et al (1993), there are comparatively few reports of late pulmonary dysfunction in children after allogeneic or autologous BMT (Table I). The incidence of chronic pulmonary symptoms in childhood is difficult to determine but approximately 10–20% of predominantly adult long-term survivors will develop symptoms associated with abnormal pulmonary function tests (Wyatt et al, 1984; Urbanski et al, 1987; Schwarer et al, 1992; Palmas et al, 1998). However, this cannot be directly extrapolated to children because of the growth potential of the lungs. The main characteristics of the paediatric BMT studies are shown in Table I. The small sample size and heterogeneity of study populations and treatments is almost certainly responsible for some of the conflicting data. During a follow-up period of 1–13 years, both long-term obstructive disease (Johnson et al, 1984; Uderzo et al, 1991; Kaplan et al, 1992; Schultz et al, 1994) and restrictive disease has been reported (Serota et al, 1984; Uderzo et al, 1991; Kaplan et al, 1992, 1994; Arvidson et al, 1994; Quigley et al, 1994; Jenney et al, 1995; Nenadov Beck et al, 1995; Rovelli et al, 1995; Nysom et al, 1996; Fanfulla et al,1997; Cerveri et al, 1999; Fulgoni et al, 1999; Neve et al, 1999). Among long-term survivors with pulmonary disease, chronic GVHD has been identified as a major risk factor for the development of progressive airflow obstruction in the large, predominantly adult series, suggesting that it may be a pulmonary manifestation (Clark et al, 1987; Holland et al, 1988; Prince et al, 1989; Tait et al, 1991; Schwarer et al,1992; Curtis et al, 1995; Soubani et al, 1996; Palmas et al, 1998). This is also suspected from its infrequent or absent association with autologous BMT in adults (Holland et al, 1988; Paz et al, 1992; Schwarer et al, 1992) and children (Arvidson et al, 1994; Nenadov Beck et al, 1995; Fanfulla et al, 1997; Cerveri et al, 1999; Neve et al, 1999). However, its rare development in the autologous situation implicates additional aetiological factors (see Risk factors). It is defined as a Fev1/FVC < 70% and Fev1 < 80% of predicted and the mortality may be high (Clark et al, 1989). The weight of evidence indicates that post-transplantation obstructive lung disease is less common in children, with the majority of studies showing little or no obstructive change (Kaplan et al, 1992, 1994; Arvidson et al, 1994; Quigley et al, 1994; Nenadov Beck et al, 1995; Rovelli et al, 1995; Nysom et al, 1996; Fanfulla et al, 1997; Cerveri et al, 1999; Fulgoni et al, 1999; Neve et al, 1999). Its association with chronic GVHD is less well established (Uderzo et al, 1991; Schultz et al, 1994; Sargent et al, 1995; Kleinau et al, 1997). Kaplan et al (1992) did not find an association between chronic GVHD and obstructive airways disease in a small series of transplanted young adults and children. In contrast, a relatively high incidence of obstructive lung disease (19·4%) was found among 89 children who had undergone allogeneic BMT more than 1·5 years before, which was strongly associated with chronic GVHD (Schultz et al, 1994). Recently, late-onset pulmonary disease, in which no infectious agents are identifiable, has been collectively termed the late onset pulmonary syndrome (LOPS), and broadened to include the spectrum of obstructive and restrictive disease, reported in two mixed studies of adults and children (Schwarer et al, 1992; Palmas et al, 1998). Histologically, the pathological processes include bronchiolitis obliterans (BO), leading to progressive obstructive airways disease, bronchiolitis obliterans with organizing pneumonia (BOOP), diffuse alveolar damage (DAD), and interstitial pneumonia which may be lymphocytic or non-classifiable (Schwarer et al, 1992; Palmas et al, 1998). LOPS commonly develops between 6 and 12 months after BMT, but can develop at any time up to 20 months (Schwarer et al, 1992; Breuer et al, 1993; Palmas et al, 1998). Clinically, it is characterized by cough, dyspnoea and sometimes wheeze, but signs may be minimal and plain radiography of the chest may be normal or show diffuse or patchy infiltrates (Schwarer et al, 1992; Palmas et al, 1998). High-resolution computerized tomography (CT) is a useful non-invasive technique in evaluation of obstructive disease in children (Sargent et al, 1995). Immunosuppressive therapy has limited success in severe cases of LOPS with Fev1 < 45% or Forced expiratory ratio (FEV1/VC) < 50% (Curtis et al, 1995; Palmas et al, 1998), although BOOP has a better prognosis (Kleinau et al, 1997; Palmas et al, 1998). BOOP may or may not be associated with chronic GVHD in case reports of children (Mathew et al, 1994; Kleinau et al, 1997). However, apart from these overt cases of clinical disease, the subclinical nature of pulmonary dysfunction in children is becoming apparent and most children appear to be symptom-free despite abnormal lung function, even when quite severely so (Uderzo et al, 1991; Arvidson et al, 1994; Nenadov Beck et al, 1995; Nysom et al, 1996; Fanfulla et al, 1997; Cerveri et al, 1999; Fulgoni et al, 1999). A homogeneous study from Copenhagen, with long follow up, evaluated longitudinal pulmonary function data in a population-based cohort of 25 survivors of allogeneic transplantation for childhood leukaemia or lymphoma. Lung volumes and transfer factor of the lung for carbon monoxide (TLCO) were reduced immediately after BMT, but increased or stabilized over the subsequent years. However, at the last follow up (4–13 years), despite the absence of symptoms, patients still had a significantly reduced transfer factor, total lung capacity (TLC) and forced vital capacity (FVC), and an increased forced expiratory volume in 1 s to FVC ratio (Fev1/FVC), indicative of a persistent diffusion defect and restrictive disease (Nysom et al, 1996). The transient reductions in lung volumes and transfer factor, approximately 3–6 months after BMT, followed by a late increase, have been found in other paediatric studies with shorter follow up (Serota et al, 1984; Uderzo et al, 1991; Arvidson et al, 1994; Kaplan et al, 1994; Quigley et al, 1994; Fanfulla et al, 1997; Neve et al, 1999). Improvement in pulmonary function is generally seen within the first year after BMT. The partial recovery and long-term residual impairment of lung volumes and transfer factor, found by Nysom et al (1996), is typical of these otherstudies. Only one short-term evaluation shows returnto baseline values (Quigley et al, 1994). Two small longitudinal studies, however, detected continuing decline in some indices of pulmonary function up to 4 years, although remaining within normal limits in Rovelli's study (Rovelli et al, 1995; Neve et al, 1999). An isolated reduction in diffusing capacity (reduced TLCO), may be the only abnormality demonstrated by the group from Pavia in three papers (Fanfulla et al, 1997; Cerveri et al, 1999; Fulgoni et al, 1999) and others (Arvidson et al, 1994). This occurred in 15% of Cerveri's cohort of 52 leukaemia patients, while 23% had a full restrictive defect. Only 62% had completely normal lung function. Some paediatric BMT studies have shown a reduction in diffusing capacity, with or without restrictive changes, in some patients prior to BMT (Quigley et al, 1994; Fanfulla et al, 1997; Leneveu et al, 1999; Neve et al, 1999). Only 65% of Fanfulla's cohort had normal lung function at baseline and 44% of Neve's. This leads to the suggestion that pulmonary function defects may be engendered by antineoplastic treatment, prior to BMT. Corroborating data show similar impairment of lung function and exercise capacity in survivors of childhood leukaemia and lymphoma, many of whom had never undergone BMT or received mediastinal radiation (Shaw et al, 1989; Jenney et al, 1995; Nysom et al, 1998b, 1998c). In contrast, a recent cross-sectional study of survivors of childhood acute lyphoblastic leukaemia (ALL), treated with Berlin–Frankfurt–Münster (BFM)-type chemotherapy alone or with chemotherapy followed by BMT, concluded that intensive front-line treatment was not associated with late pulmonary dysfunction in most cases,but that, in agreement with Jenney et al (1995), re-treatment including BMT can frequently injure the lung (Fulgoni et al, 1999). Nonetheless, worse lung function and clinical sequelae occurred in a small number of children after autologous BMT, treated more intensively for neuroblastoma, before BMT (Neve et al, 1999), and in patients transplanted for haematological malignancy beyond second complete remission, in whom 54% had impaired function, compared with 21% in first remission (Cerveri et al, 1999) (Table I). Children transplanted for leukaemia may have a worse outcome than those transplanted for aplastic anaemia, once again suggesting the influence of pretreatment (Serota et al, 1984). This could not be formally confirmed by statistical comparison of these two groups by Kaplan et al (1994). However, abnormal pulmonary function tests (PFT) before BMT are not usually considered to be a contraindication to BMT in children (Fanfulla et al, 1997), although some groups argue that pre-BMT tests are predictive for later lung disease in adults (Soubani et al, 1996). The role of chronic GVHD in the development of airflow obstruction has already been discussed. Other factors implicated in an increased risk of obstructive disease, in the predominantly adult literature, are the prolonged use of methotrexate, decreased immunoglobulins, increasing recipient age, male gender and lack of human leucocyte antigen (HLA) matching, and recurrent pulmonary infection (Clark et al, 1987; Holland et al, 1988; Schwarer et al, 1992; Breuer et al, 1993; Soubani et al, 1996). Schultz et al (1994), in their paediatric study, agreed that acute and chronic GVHD, HLA disparity and increasing age correlated with obstructive lung disease, but methotrexate prophylaxis, cytomegalovirus (CMV) reactivity in either donor or recipient, and TBI did not. The aetiology of restrictive disease is probably multifactorial. From adult studies, it appears to include the toxic effects of chemotherapy and radiation, and GVHD (Tait et al, 1991; Beinert et al, 1996; Soubani et al, 1996). It may be exacerbated by previous pulmonary infections in both adults (Soubani et al, 1996) and children (Jenney et al, 1995; Nenadov Beck et al, 1995; Cerveri et al, 1999). In children, although no association was found between long-term pulmonary dysfunction and cytomegalovirus reactivation post BMT by Cerveri et al (1999), a previous study from the same group and Quigley's group found that CMV seropositivity was associated (Quigley et al, 1994; Fanfulla et al, 1997). Quigley et al (1994) also found an association between restrictive disease and chronic GVHD, but others did not (Kaplan et al, 1992; Fanfulla et al, 1997; Cerveri et al, 1999). The use of TBI, particularly when delivered at a high dose and dose rate, and as a single fraction, has been associated with restrictive disease in predominantly adult series (Keane et al, 1981; Barrett et al, 1983, Tait et al, 1991). Paediatric data is conflicting and inconclusive but three groups reported worse lung function with TBI (Serota et al, 1984; Arvidson et al, 1994; Quigley et al, 1994). Others found no such association with late dysfunction (Schultz et al, 1994; Fanfulla et al, 1997; Cerveri et al, 1999). Cytotoxic agents commonly used in conditioning regimens [cyclophosphamide, busulphan, melphalan and BCNU (carmustine)] may induce interstitial pulmonary fibrosis of late onset and methotrexate may also cause pneumonitis (Twohig & Matthay, 1990). Cerveri et al (1999) found no difference between TBI or busulphan in the conditioning regimen, while Quigley et al (1994) have shown that busulphan-conta