(Goldstone et al, 1993). Rates of morbidity and mortality have improved with better supportive measures and administration of HDC at an earlier phase in the course of the disease, but procedure-related mortality remains in the region of 10-20% in most published series using autologous bone marrow transplantation (ABMT) Jones et al, 1990;Reece et al, 1991;Bierman et al, 1993). Approximately 50% of these fatalities will be as a consequence of idiopathic pneumonia syndrome (IPS). This presents with dyspnoea, pulmonary infiltrates and fever, and the differential diagnosis is between infection (often with an uncommon or unusual organism), pulmonary involvement with HD, intrapulmonary haemorrhage or druginduced pneumonitis (Clark et al, 1993). The latter is most frequently associated with high-dose BCNU (carmustine), which induces interstitial fibrosis in a dose-dependent manner. (Litam et al, 1981;Pecego et al, 1986;Weaver et al, 1993). The mechanism is poorly understood but thought to involve toxic metabolites that react with sulphydryl compounds (including glutathione) resulting in neutralization. Depletion of such compounds and inhibition of glutathione reductase in alveolar Received 19April 1995 Revised 11 October 1996 Accepted 14 October 1996 Correspondence to: D Cunningham macrophages is postulated to increase the risk of oxidant injury (Arrick et al, 1984;Clark et al, 1993).Pulmonary toxicity may occur 9 days to 12 years after starting BCNU (Durant et al, 1979) and may present with the sudden onset of dyspnoea and progress rapidly to death, or be more insidious with a fatal outcome within 2 years (Phillips et al, 1983). In a dosefinding study published in 1990 using cyclophosphamide, etoposide and BCNU (Wheeler et al, 1990) 450 mg m-2 of BCNU was the maximum tolerated dose, with a 5% mortality rate compared with 22% observed following a dose of 600 mg ni-2. Another study using the same regimen and 600 mg m-2 BCNU, published the following year, reported a 16% incidence of IPS and a 12% mortality rate (Reece et al, 1991). A more recent series (Weaver et al, 1993) reported no IPS in patients who received BCNU 300 mg m-2 a rate of 23% in patients who received 600 mg m-2.The aim of our study was to quantify the risk and time of onset of IPS at our own institution and determine the risk factors for its development and prognostic factors for outcome.
PATIENTS AND METHODS
Using