Unrelated donor search prognostic score to support early HLA consultation and clinical decisions

Bone Marrow Transplant. 2016 Nov;51(11):1476-1481. doi: 10.1038/bmt.2016.162. Epub 2016 Jun 6.


A simple scoring system that can provide a quick search prognosis at the onset of an adult unrelated donor (URD) search could be a useful tool for transplant physicians. We aimed to determine whether patient human leukocyte Ag genotype frequency (GF) could be used as a surrogate measure of whether or not a potential 10/10 and/or 9/10 URD in the Be The Match Registry (BTMR) can be identified for the patient. GF was assigned on a training data set of 2410 patients that searched the BTMR using the reported ethnic group. A proportional odds model was used to correlate GF with defined search productivity categories as follows: 'Good' (>2 10/10), 'Fair' (1-2 10/10 or No 10/10 and >2 9/10) or 'Poor' (No 10/10 and <3 9/10). A second cohort (n=2411) was used to calculate the concordance by the ethnic group in all three categories. In addition, we validated against an independent cohort (n=1344) resolved as having a 10/10 or 9/10 matched URD. Across the ethnic groups, >90% of cases with 'Good' GF prognosis, 20-26% 'Fair' and <10% 'Poor' had a 10/10 URD. Although not a replacement for an actual URD search, GF offers a quick way for transplant physicians to get an indication of the likely search outcome.

MeSH terms

  • Algorithms
  • Decision Making, Computer-Assisted*
  • Genotype
  • HLA Antigens / genetics*
  • Histocompatibility
  • Histocompatibility Testing / methods*
  • Humans
  • Prognosis
  • Referral and Consultation
  • Registries
  • Supervised Machine Learning
  • Time Factors
  • Unrelated Donors


  • HLA Antigens