An Empiric Model for Defining A Suitable Cord Blood Inventory Size in Brazil Using an HLA Matching Computer Program.

Umbilical cord blood represents an alternative for patients who need Hematopoietic Cell Transplants and cannot rely on identical HLA donors. The Brazilian public cord blood consortium has been operating since 2002 with six centers to date, including the Cord Blood Bank at Albert Einstein Hospital, Sao Paulo, Brazil. However, currently there are no studies concerning the optimal size of a cord blood inventory in our country. It is difficult to define the inventory size due to the different possible levels of HLA compatibility and the minimum cellularity required for transplantation.

To estimate the minimum number of cord blood units that should be stored in Brazil in order to adequately provide 4/6, 5/6 and 6/6 HLA-A, -B, -DRB1 matches for a significant percentage of patients who need stem cell transplantation.

A computer program was developed to match HLA compatibilities between a group of 904 patients searching for unrelated stem cell transplantation and a database of over 10,500 Volunteer Unrelated Donors (VUD) and 1,018 cord blood units stored in the Albert Einstein Hospital Cord Bank. In sequence, the number of units necessary for different levels of matches was estimated. This database consisted of HLA-A, -B and -DRB1 typed with low resolution for class I and high resolution for class II. The sequence of individual units in the DATABASE was randomized in order to avoid any bias in the analyses caused by data clustering. Both groups of donors and patients came from varied ethnical backgrounds and live in the state of Sao Paulo, Brazil. The program compared each patient phenotype with the phenotypes of database subgroups of increasing size (500 to 11,500 at 1,000 intervals). The results were then analyzed according to: a) Likelihood of finding 4/6, 5/6, 6/6 matches using at least one cord blood unit for each patient; b) Likelihood of finding at least four compatible cord units (ie, multiple) per match for each patient.

For a minimum 4/6 match using at least one cord blood unit, over 82% of patients were able to find suitable units within an inventory size of 1,500 cords. That percentage raised to over 95% by increasing the inventory size to 5,500 cords. The curve grew asymptotically from that point onwards. When at least four cord units were required at the same compatibility level (4/6), over 90% of patients were able to find suitable units within an inventory size of 11,500 cords With more restrictive (5/6 or 6/6) matching requirements using at least one cord blood unit, 61,3% of patients were able to find suitable unit within an inventory size of 11,500 cords. When at least four cord units with a minimum of 5/6 match were required, only 22,8% of patients will find these amount of units in a bank with 11.500 stored units. An estimate of the need of stored cord blood units based on these analyses suggests that, considering the same genetic distribution of donors, and taking into account that only 21% of the inventory of our bank would provide > 3,0×10E7 TCN for an adult or multiple cord units for an adequate cell dosage, more than 58.000 units would be necessary to find a suitable unit with 5/6 match for 61,3% of patients.

Despite the high degree of variability of the HLA phenotype in the studied population, it was possible to find a suitable matched cord blood unit within a relatively small inventory. That finding is true only for a low HLA compatibility level. Indeed, with only 2,600 stored units, Albert Einstein's Blood Bank has already sent 70 units for transplantation, mostly for children. However, when a higher degree of compatibility and/or a higher cellularity are necessary, the number of units in the database increases drastically. This study suggests that a minimum inventory of 58,000 cord blood units would be necessary to find a suitable match for almost two thirds of the Brazilian population. Finally, it's important to mention that this study is based on a relatively small sample and should be tested in larger studies including rare phenotypes and ethnic minorities.

No relevant conflicts of interest to declare.