An Empirical Analysis of the Probability of Finding a Well Matched Cord Blood Unit: Implications for a National Cord Blood Inventory.

Study Objective: To estimate the probability of finding an HLA matched cord blood unit, taking into account ethnicity, and the required size of national inventory of cord blood units that would provide a 6/6 or 5/6 HLA-A, -B, -DRB1 match for 80% of US patients.

Methods: The study was limited to patients and cord blood units in the NYBC inventory through October 1, 2004 that had HLA-A, -B and -DRB1 typing completed at low resolution for class I and high resolution for DRB1. Of 16,222 search requests, 10,007 patients had high resolution DRB1 typing and of 26,675 cord blood units in our bank, 20,773 are DRB1 typed, thus far, at high resolution. One-third of the inventory would provide ≥ 2.0 x 10⁷ TNC for a 70 kg adult. The patients had relatively low frequencies of the common HLA haplotypes, suggesting a reduction because suitable matches were found among marrow donor registries. Patient ethnicity was similar to that of the US population while donors were more likely to include members of ethnic minorities (20% African-American, 21% Hispanic, 8% Asian), as expected (we collect cord blood in some maternity hospitals that serve a high proportion of non-Caucasoid expectant mothers). A simulated search was performed to find matches for all patients in the current inventory and repeated to find potential grafts with mismatches in the GvHD direction only (no rejection mismatch). Grafts with only GvHD mismatches have had outcomes similar to a 6/6 match (

Results: All except 94 patients (0.9%) had one or more cord blood units that provided a 4/6 match or better, 61% had a 5/6 match and 9% a 6/6 match. The probability in finding well-matched units almost doubled when we included both 6/6 matches and units mismatched in the GvHD direction only. The chance of finding a 5/6 or 6/6 match varied with patient ethnicity and was lowest for African-Americans (44% and 1.7%, respectively), despite the relatively high proportion of African-American donors to the inventory. As expected, patients were more likely to find a suitable match within their own ethnic group. When the inventory search was restricted in size (randomly selected units) to 5,000, 10,000, 15,000 and the full inventory of 20,773 units, the chance of finding a 5/6 or 6/6 match increased (36%, 48%, 55% and 61%, respectively) demonstrating a progressively diminishing improvement in match success.

Conclusions: These empirical comparisons indicate that most patients can find a suitably matched cord blood unit, defined as either a 5/6 or 6/6 HLA match, within our relatively small inventory of cord blood units. The probability of finding a match being highest in the patient’s own ethnic group underscores the need to enrich cord blood inventories with cord blood units donated by ethnic minorities. A projection based on these analyses suggest that 55–60,000 units would provide a 5/6 or 6/6 match for 80% of patients assuming that the ethnic distribution of donors and patients is maintained. However, cord blood units must provide a cell dose ≥ 2.0 x 10⁷ TNC/Kg for single unit transplants. Thus, some 155–170,000 cord blood units may be required to also provide 70 Kg adults with ≥ 80% chance of finding a suitable match with an adequate cell dose. Inventory requirements might decrease if, for example, transplants with well matched units or two unit transplants lessen the cell dose limits, as preliminary data now suggests.