Although transplants from alternative donors are effective in some patients with leukemia, treatment failure is higher than after HLA-identical sibling transplants. Outcome depends on leukemia state, donor-recipient relationship, and degree of HLA matching. In early leukemia, alternative donor transplants have a more than twofold increased risk of treatment failure compared with HLA-identical sibling transplants. This difference is less in advanced leukemia.

Summary. Recent reports have shown that low nucleated cell dose signi®cantly decreases survival after cord blood transplantation. Prior to starting clinical cord blood banking we investigated the impact of obstetric factors on cell dose and volume of cord blood donations. Cord blood was obtained from 114 normal full-term deliveries. Mean volume collected was 93´5 ml, mean total nucleated cell count (TNC) was 13´1´10 8 . Statistical analysis was by backwards stepwise regression. Signi®cant factors affecting nucleated cell yield were volume of blood collected (P < 0´001), length of gestation (P < 0´0001), time from delivery of the infant to cord clamping (P 0´018) and total length of labour (P 0´002). In clinical cord blood banking we have successfully used these ®ndings for pre-collection assessment of placentae. Out of 476 cord blood donations subsequently collected for banking, only 29 (6´1%) have been discarded due to low volume. The mean TNC of the 409 banked units following volume reduction was 10´1´10 8 . Despite careful optimization of collection, processing and storage techniques, cell dose still limits cord blood transplantation to smaller recipients.

This study was designed to determine the incidence of relapse and factors predictive for relapse in 719 patients with severe aplastic anaemia (SAA) after immunosuppressive treatment (IS). Patients developing myelodysplasia or acute leukaemia after IS, and patients receiving a transplant, were excluded from this analysis. Response was defined as reaching complete independence from transfusions, relapse was defined as becoming again transfusion dependent. This criteria was validated by similar figures when using other 'relapse criteria' such as drop in neutrophil or platelet counts. Of 358 patients responding to IS. 74 patients relapsed after a mean time of 778 d after treatment. The actuarial incidence of relapse is 35.2% at 14 years after IS. The risk for relapse was higher in patients responding within 120 d from IS (48%) compared to patients responding between 120 and 360 d (40%) and only 20% for slow responders (> 360 d from IS) (P < 0.00001). In multivariate analysis this factor still proved significant (P < 0.0001). The mean time between diagnosis and treatment was significantly longer in patients relapsing compared to patients who did not relapse (260 v 134 d, P = 0.037). Relapse was not predicted by the severity of the disease, age, and sex. In 39 of the 74 relapsing patients a second response could be achieved. Responses after relapse were associated in univariate analysis with early response to previous IS and early occurrence of relapse. The actuarial survival of patients not relapsing is significantly better than survival of patients relapsing (79.8% v 67.1%, P = 0.0024). However, the actuarial survival of 39 relapsing patients who responded again to IS was similar to patients not relapsing (86%) and significantly better than in 35 patients not reaching a second response after relapse (49.3%, P = 0.0015). This study indicates that relapse is a relevant problem in the treatment of aplastic anaemia, and does have an impact on overall survival. Prospective studies of immunosuppressive regimens, looking at responses, should also address this problem in the future.

Interest in the isolation and characterization of primitive hemopoietic cells in both the clinical and research fields has rapidly increased. In parallel, different purification systems have been developed to isolate these cells. We have compared five different methods for separation of CD34+ cells from human umbilical cord blood, normal bone marrow and apheresis harvests and analyzed purity, recovery, yield and enrichment of colony forming cells (CFC) for each individual system. Our results indicate that the most reliable methods of purification for all samples were fluorescence activated cell sorting (FACS) and magnetic activated cell sorting (MACS) which consistently yielded high purities (> 70%) and enrichment of CFC. In this respect the enrichment of CFC from the MACS was superior to all the other systems including FACS. Similar results (> 70%) for purity were obtained using avidin affinity columns and a biotinylated antibody but neither yield nor CFC enrichment approached the values achieved with MACS. On average CFC enrichment using these affinity columns was greater than that observed for FACS while the purity was comparable. Both CELLector flasks and immunomagnetic beads coated with CD34 antibodies were less effective in our hands in separating purified populations of progenitor cells. Both purity and CFC enrichment of CD34+ cells using these methods were at least 50% lower than obtained with either FACS, MACS or affinity columns.

An in vitro model system is described that allows Correspondence: Dr 1. C. W. Marsh, Toronto General Hospital, Mulock Larkin Wing 2-036. 200 Elizabeth St.. Toronto, Ontario. Canada M5G 2C4.