Better High Resolution HLA-Matching Is Associated with Lower Transplant Related Mortality after Cord Blood Transplantation

Background: For patients lacking a suitable related or unrelated stem cell donor, cord blood (CB) as the source of stem cells is being increasingly utilized both in pediatric and adult patients. A CB graft is attractive due to the increased level of HLA disparity that can be tolerated, allowing nearly all patients to find a donor. HLA-typing is performed minimally at the antigen level for HLA-A and -B (low resolution LR), and high resolution (HR) HLA-typing is performed for HLA-DRB1 alleles. More recently, many centers are turning to HR HLA-typing as the standard; however the impact on outcomes of HLA disparity using HR typing has not been clearly determined.

Methods: The impact of HR HLA matching was retrospectively analyzed on 202 patients with hematologic malignancies receiving either a single or double cord blood transplant (CBT) between 2007-2014 at the Hutchinson Center. 135 patients (67%) had HR typing for HLA-A, -B, -C, and –DRB1. For double CBT, the HR HLA typing of the predominant cord was used in the outcome analysis. Patients with mixed chimerism (n=2) or graft failure (n=9) were excluded leaving a total of 124 patients (61%) in the final analysis. HR typing at 4 loci revealed that the median HLA match was 5/8 (range 2-7/8). The cause-specific hazards of failure for each endpoint were compared between patients with worse HR typing (≤ 4/8 HR HLA match, n=46) and patients with better HR typing (≥5/8 HLA match, n=78). These models were adjusted for various factors: patient age, weight, risk of disease, presence of minimal residual disease (MRD), patient CMV serostatus, conditioning regimen (myeloablative vs. non-myeloablative), and grades III-IV acute graft-vs.-host disease (aGVHD).

Results: The median patient age and weight were 43 years and 76 kg and 33 years and 69 kg for the ≤ 4/8 and  ≥5/8 groups respectively (p=0.03, p=0.06). The two groups were similar with respect to conditioning intensity, sex, race, CMV serostatus, MRD status, and diagnosis (Fig. 1). The median pre-thaw TNC/kg x10⁷ and pre-thaw CD 34 cells/kgx10⁶ values (Fig. 1) indicated that there was no difference between the two HR HLA groups in regards to both CB unit size or stem cell content (p=0.25). Median time of neutrophil recovery was 20 days for  ≤ 4/8 group and 19 days for ≥5/8 group with a cumulative incidence of engraftment of 93% and 98% respectively (p=0.20). The median time to platelet recovery (>20 x 10⁹/L) was also similar between the two groups: 34 days for the ≤ 4/8 group and 35 days for the ≥5/8 group (p=0.12).  The probability of 4-year DFS was 34% and 62%, respectively (p=0.04) (Fig. 2A). The cumulative incidence of transplant related mortality (TRM) at 4 years was 41% for the ≤ 4/8 group and 15% for the ≥5/8 group (p=0.002) (Fig. 2B). In multivariable analysis the association between worse HLA-matching and higher TRM remained statistically significant [HR=3.22, 95% CI: 1.25-8.20, p=0.01]. The most common causes of death were pulmonary failure (27%), bacterial or fungal infection (24%), and multi-system organ failure (21%). Interestingly, worse HLA matching did not reduce the risk of relapse [HR=0.87, 95% CI: 0.32-2.4, p=0.80] or increase the risk of grades III-IV aGVHD (p=0.45).

Conclusion: Our study demonstrates that TRM was significantly lower in patients receiving a ≥5/8 HR HLA engrafting CB unit leading to a better DFS. The increased incidence of TRM in the less HLA-matched group was not due to an increased risk of aGVHD. Our results support that the use of ≥5/8 allele-matched HLA-units is associated with better outcomes; however how to select CB units based on HLA-allele match in the setting of double CBT still needs to be confirmed in larger studies analyzing how HLA-mismatch at different loci (patient vs. units and unit vs. unit) can impact CB unit selection and dominance.