Abstract
Introduction:
Patients with primary immunodeficiencies (PIDs) are potentially cured by allogeneic hematopoietic cell transplantation (HCT). The spectrum of PIDs has expanded greatly beyond those that present in infancy or are diagnosed on newborn screening and require urgent, preemptive HCT. Many PID diagnoses are now made later in life and the role of HCT is only considered upon severe disease manifestations; in these cases, the kinetics and goals of a donor search may be different than for severe combined immunodeficiency. Across all PIDs, related donor searches have the additional selection factor of the inherited disease, such searches may yield more limited options than searches for patients with hematologic malignancies; thus, unrelated donor options often become more critical in these patients. We retrospectively evaluated the outcomes of donor searches among PID patients referred for HCT at the National Institutes of Health (NIH), where the minimum patient age for evaluation is 3 years old and where donor options included matched sibling or matched related (MSD/MRD), HLA-haploidentical (haplo), or 7–8/8 HLA-matched unrelated (mMUD/MUD) donors.
Methods:
Patient (n=161) and donor demographics, MUD search results, HLA typing, pedigrees, mutation testing, and donor selection data were collected. The National Marrow Donor Program HapLogic 8/8 HLA-match algorithm was used to predict the likelihood of a successful MUD search and categorized as very good, good, fair, poor, very poor, or futile per the Memorial Sloan Kettering Cancer Center (MSKCC) Search Prognosis method.
Results:
There were significant differences by PID mode of inheritance in patient age, disposition (receipt of HCT or not), donor source, and donor relatedness. A related or unrelated donor option could be identified for 94% of patients. Of living 1st degree relatives (median 3 (range 0–12) per patient), a median of 1 donor remained for autosomal dominant and X-linked (XL) diseases after HLA typing, mutation testing, and other exclusions, and a median of 2 donors remained for autosomal recessive (AR) diseases. Among patients with a PID of known mode of inheritance (n=142), the best related donor was haplo for 99 (70%) patients, with 56 (39%) haplos age 40 years or older and 5 (4%) 2nd degree haplos; 13 (9%) had no family donor options. The best related donor was a heterozygote/asymptomatic carrier of the PID mutation in 36 (49%) patients with AR or XL disease (n=73). Among patients with MUD search performed (n=139), 53 (38%) had very poor/futile 8/8 MUD searches, including 6 (32%) of those with unknown PID mutation and therefore no family donor options. The MSKCC Search Prognosis was less favorable for those of non-European ancestry compared to European ancestry, p=0.002. The majority of patients of Hispanic or African ancestry had very poor/futile MUD searches, 71% and 63%, respectively. No HCT recipients with very poor/futile MUD searches (n=38) received 8/8 MUD grafts.
Discussion:
Alternative donor options, including haplo and unrelated donors, are critical to enable HCT for patients with PID. MUD search success remains low for those of non-European ancestry, and this is of particular concern for patients with PID due to an unknown genetic defect. Among PID patients, related donor options are reduced and haplos age 40 years+ and/or mutation carriers are often the best family option.
Introduction
Allogeneic hematopoietic cell transplantation (HCT) is a potentially curative therapy for patients with inherited diseases of the immune and hematopoietic system, globally referred to as primary immunodeficiencies (PIDs). Many of the more common and long-recognized PIDs have an autosomal recessive (AR) or X-linked (XL) mode of inheritance, such as Wiskott-Aldrich syndrome, chronic granulomatous disease (CGD), and severe combined immunodeficiency (SCID), wherein patients are typically diagnosed and treated with HCT or gene therapy at a very early age. For very severe PIDs that are diagnosed in infancy and referred quickly to HCT based solely on the genetic diagnosis, the goal is to find the most readily available (parent or sibling) donor as quickly as possible and move forward. However, over 300 PIDs have now been identified, many with variable disease severity and phenotypes, asymptomatic affected family members, and late diagnoses. Many PIDs are inherited in an autosomal dominant (AD) fashion, with disease manifestations due to gain of function mutations, such as activated PI3K mutations, or haploinsufficiency, such as GATA2 mutations. These PID patients may present much later for HCT and thus the urgency/necessity to use the first available donor, often a carrier, may be a lesser driver of the donor search process, whereas other factors, including the time-consuming need to perform genetic testing on patients and potential related donors, may rise to the forefront.
While the historical “gold standard” donor was a human leukocyte antigen (HLA)-matched sibling donor (MSD), only 25–30% of patients requiring HCT will have this option.1 When the additional selection factor of an inherited disease needs to be considered, the availability of a suitable MSD may further decrease. In other inherited diseases, such as sickle cell disease (SCD), an AR disease, only 21% of patients referred for HCT have an unaffected MSD and, in one study, it was calculated that only 9% of HCT candidates with SCD went on to receive MSD HCT.2 When considering patients with AD diseases for HCT, the chances of finding a suitable MSD may be even lower.
Considering parents, offspring, and siblings, it is generally observed that over 95% of patients referred for HCT will have a readily available 1st degree HLA-haploidentical (haplo) donor option, regardless of ethnicity. However, patients with PID may be less likely to have children (for reasons of young age themselves, chronic illness, or infertility as part of the disease phenotype, treatment with teratogenic drugs, fear of passing on the disease, etc.) and relatives may be affected with the PID as well. Thus, in the PID setting, consideration of haplo options may expand the donor pool for patients with PID but may not afford a donor option for nearly all PID patients in need of HCT as is true for patients with non-inherited hematologic malignancies.
When a MSD is not an option, matched unrelated donors (MUDs) are often considered as a first alternative. The necessity of MUD options may be particularly relevant for patients where there is the additional donor selection factor of an inherited disease, potentially further limiting related donor options. However, it is well established that patients of non-European, particularly non-northern European, ancestry have variable and often poor chances of finding a MUD.3,4 In one study, African-Americans were found to have a MUD option in only 6% of searches using the National Marrow Donor Program.5 In general, 50% of unrelated donor searches will not yield a suitable MUD despite international registry-based searches.6 Furthermore, at present, gene therapy is not an available potential alternative to HCT for the vast majority of PIDs.
While the consideration of both haplos and MUDs as good alternatives to MSDs should improve the ability to offer HCT to anyone who requires it, the impact of an additional selection factor of an inherited PID on the breadth of donor options and the resulting ability (or inability) to move forward to HCT has not been studied. We hypothesized that the extent to which the additional donor selection factor of a germline PID impacts on the related donor pool and the receipt of HCT would depend on whether a specific disease-causing mutation could be identified, thus enabling the screening of family member to find a suitable donor, and the mode of inheritance of the PID. Herein, we evaluated the outcomes of related and unrelated donor searches for patients with PID referred for HCT at the National Institutes of Health (NIH), where only patients ≥3 years old can be evaluated, within a timeframe when donor types utilized were MRDs, 7–8/8 unrelated donors, and haplos.
Patients and Methods
After Institutional Review Board approval, a retrospective review was conducted of HCT candidates referred to the NIH to screen for HCT (, clinicaltrials.gov). Patients referred for HCT at the NIH were all age 3 years or older, as intensive care resources that could be required by a HCT recipient are not available at the NIH Clinical Center for very young children. The timeframe of this retrospective study was limited to times when there was an actively recruiting clinical trial for a given PID which utilized MSD or HLA-matched related donor (MRD), MUD, and haplo donors; thus, the time periods started on March 30, 2012 for HCT candidates with GATA2 haploinsufficiency, May 21, 2014 for HCT candidates for DOCK8 deficiency, January 1, 2015 for HCT candidates with chronic granulomatous disease, and October 6, 2015 for all other PID HCT candidates. This study was performed by the same providers who conducted the original/actual donor searches, so it was possible to gather first-hand data regarding how each search actually proceeded in a retrospective manner. HCT candidates were included if, at the time of the actual donor search, HLA typing was performed, at minimum, on the recipient and if a pedigree of at least 1st degree relatives was constructed. First-degree relatives included a candidate’s biological parents, full siblings, and children. Inclusion of patients for this retrospective study ended on July 27, 2018 for data analysis. Data related to the donor searches, mutation testing, donor and recipient demographics, recipient ancestry, and family history were collected using records in the Clinical Research Information Systems, Crimson, the HLA laboratory, and the transplant coordinator offices of the National Cancer Institute and the National Institute of Allergy and Infectious Diseases of the National Institutes of Health. For recipients who underwent HCT at the NIH, more detailed data regarding the selected donor were collected; patients evaluated at the NIH but transplanted elsewhere were included if donor and HCT demographic data were known due to ongoing patient follow-up at the NIH.
The HapLogicSM matching predictions algorithm (National Marrow Donor Program (NMDP), Traxis application, Minneapolis, MN) was used to determine the chance of a donor within the NMDP/Be The Match network being an 8/8 HLA-allele match.7 The Memorial Sloan Kettering Cancer Center Search Prognosis (MSKCC SP) categorization was then used to categorize the predicted outcome of an 8/8 HLA-allele match unrelated donor search, where the categorization was as follows: Very Good: ≥20 8/8 donors with a ≥85% chance; Good: 5–19 8/8 donors with a ≥85% chance and/or ≥20 8/8 donors with a ≥70% chance; Fair: 1–4 8/8 potential donors with a ≥85% chance and/or 1–19 8/8 potential donors with a ≥70% chance and/or ≥5 8/8 potential donors with a 40%−69% chance; Poor: 1–4 8/8 potential donors with a 40%−69% chance and/or ≥1 8/8 potential donors with a 25%−39% chance; Very Poor: ≥1 8/8 potential donor with a ≤24% chance; and Futile: 0 8/8 potential donors.8 For this study, the MUD search analysis was performed retrospectively, for a second time, with the initial search being the one performed for clinical purposes at the time of HCT referral. Thus, the repeat, retrospective search could potentially contain data on donor availability and predicted likelihood of matching that was more refined than the data available at the time of the initial MUD search.
For the purposes of this study, recipients with unknown PID mutations were considered to have no family donor options since related donors could not be screened for a specific mutation to determine suitability in these cases. Best family donor option for recipients with an identified genetic defect was retrospectively designated according to the following prioritization: 1) unaffected by the same mutation, 2) suitability, eligibility, and willingness to donate, 3) degree of HLA match, 4) age, prioritizing adults < 40 years over adults 40 or older, 5) avoidance of multiparous female donors, female donors for male recipients, donors to which the recipient had donor-specific anti-HLA antibodies (data not available for all haplos considered), 6) donor same weight or larger than recipient, and 7) between parents, preference of father over mother. As many related donor searches were truncated upon finding a MUD, blood type, viral serologic status, and other donor selection factors that might have been considered in practice were not factored in retrospectively. Biologic parents and children of patients referred to HCT were presumed to be guaranteed haplo options, even if HLA typing was not performed on those donors.
Descriptive statistics were used for patient, donor, and HCT characteristics. Groups were compared using the Chi-square test for categorical variables and Kruskall-Wallis test for continuous variables (GraphPad Prism, version 8.0).
Results
Patient (n=161), donor, and HCT (n=109) characteristics are shown in Table 1, with significant differences in patient age, disposition (receipt of HCT or not), and, among those transplanted, donor relatedness and donor type, across PID modes of inheritance.
Table 1.
Patient, donor, and disease characteristics and disposition
| PID patients (n=161) | AD (n=69) | AR (n=40) | XL (n=33) | Unknown (n=19) | P-value | |
|---|---|---|---|---|---|---|
| Male, n (%) | 92 (57%) | 29 (42%) | 22 (55%) | 30 (91%) | 11 (58%) | |
| Age, median years (range)& | 24 (3.8–68) | 25 (4.5–68) | 19 (3.8–44) | 24 (3.9–54) | 27 (9.2–52) | p=0.003 |
| De novo mutation, n (%)¥ | 28 (17%) | 28 (41%) | 0 | 0 | 0 | |
| Disposition, n (%) | p=0.04 | |||||
| Received HCT or gene therapy# | 112 (70%) | 44 (64%) | 27 (68%) | 28 (85%) | 13 (68%) | |
| Pending HCT with donor found | 40 (25%) | 22 (32%) | 12 (30%) | 3 (9%) | 3 (16%) | |
| No HCT donor identified$ | 6 (4%) | 1 (1%) | 1 (2%) | 1 (3%) | 3 (16%) | |
| Died before donor found | 3 (2%) | 2 (3%) | 0 | 1 (3%) | 0 | |
|
Time from HLA typing to HCT, median months (range) |
7 (1–50) | 7 (1–50) | 7 (1–48) | 8 (2–45) | 7 (2–15) | NS |
| MUD search performed, n (%) | 139 (86%) | 59 (86%) | 33 (83%) | 28 (85%) | 19 (100%) | |
| Donor type, n (% of those receiving HCT, n=109) | p=0.01 | |||||
| MUD | 47 (43%) | 17 (39%) | 9 (33%) | 13 (52%) | 8 (62%) | |
| Haplo full-sibling | 9 (8%) | 5 (11%) | 3 (11%) | 1 (4%) | 0 | |
| Haplo parent | 22 (20%) | 7 (16%) | 9 (33%) | 5 (20%) | 1 (8%) | |
| Haplo 2nd degree relative | 3 (3%) | 3 (7%) | 0 | 0 | 0 | |
| Haplo child | 1 (1%) | 1 (2%) | 0 | 0 | 0 | |
| MSD | 22 (20%) | 11 (25%) | 3 (11%) | 6 (24%) | 2 (15%) | |
| MRD | 3 (3%) | 0 | 3 (11%) | 0 | 0 | |
| mMUD | 2 (2%) | 0 | 0 | 0 | 2 (15%) | |
| Donor relatedness, n (% of those receiving HCT) | p=0.04 | |||||
| Related | 60 (55%) | 27 (61%) | 18 (67%) | 12 (48%) | 3 (23%) | |
| Unrelated | 49 (45%) | 17 (39%) | 9 (33%) | 13 (52%) | 10 (77%) | |
| Donor HLA-match and relation, n (% of those receiving HCT) | NS | |||||
| Matched related | 24 (22%) | 11 (25%) | 5 (19%) | 6 (24%) | 2 (15%) | |
| Unrelated | 49 (45%) | 17 (39%) | 9 (33%) | 13 (52%) | 10 (77%) | |
| Haplo related | 36 (33%) | 16 (36%) | 13 (48%) | 6 (24%) | 1 (8%) | |
| Female donor/male recipient, n (% of those receiving HCT) | 17 (16%) | 7 (16%) | 4 (15%) | 5 (20%) | 1 (8%) | NS |
| Diagnosis, n (%) | n/a | |||||
| GATA2 haploinsufficiency | 33 (20%) | 33 (48%) | 0 | 0 | 0 | |
| Chronic granulomatous disease | 24 (15%) | 0 | 2 (5%) | 22 (67%)@ | 0 | |
| PID of unknown genetic defect+ | 19 (12%) | 0 | 0 | 0 | 19 (100%) | |
| PI3K gain of function~ | 17 (11%) | 17 (25%) | 0 | 0 | 0 | |
| DOCK8 deficiency/Hyper-IgE syndrome | 16 (10%) | 0 | 16 (40%) | 0 | 0 | |
| ADA2 deficiency/DADA2 | 8 (5%) | 0 | 8 (20%) | 0 | 0 | |
| CTLA4 haploinsufficiency | 6 (4%) | 6 (9%) | 0 | 0 | 0 | |
| IFNGR1 deficiency | 5 (3%) | 2 (3%) | 3 (8%) | 0 | 0 | |
| MAGT1 deficiency/XMEN disease | 4 (2%) | 0 | 0 | 4 (12%) | 0 | |
| STAT3 deficiency/Hyper-IgE syndrome | 4 (2%) | 4 (6%) | 0 | 0 | 0 | |
| RAG1/RAG2 deficiency | 4 (2%) | 0 | 4 (10%) | 0 | 0 | |
| STAT1 gain of function | 3 (2%) | 3 (5%) | 0 | 0 | 0 | |
| Other diagnoses§ | 18 (11%) | 4 (6%) | 7 (18%) | 7 (21%) | 0 | |
Abbreviations: PID, primary immunodeficiency disease; HCT, hematopoietic cell transplantation; AD, autosomal dominant; AR, autosomal recessive; XL, X-linked; HLA, human leukocyte antigen; MUD, matched unrelated donor; Haplo, HLA-haploidentical; MSD, matched sibling donor; MRD, matched related donor; mMUD, 7/8 matched unrelated donor; DADA2, deficiency of ADA2; XMEN, X-linked immunodeficiency with magnesium defect, EBV infection, and neoplasia
Age at HCT, if transplanted; age at present, if not transplanted; age at death, if died before reaching HCT
De novo status designations were based on family history and mutation testing in parents and siblings, as appropriate
109 received HCT; 3 received gene therapy for X-linked chronic granulomatous disease
No donor identified after both 8/8 MUD search and related donor search
For some patients with unknown mutations who proceeded to HCT, only related donor options were available, despite them being considered, for the purposes of this study, to have no related donor options.
X-linked, male (n=20); X-linked, symptomatic female carrier (n=2)
Phenotypes: Chronic active EBV (n=4), MonoMAC (n=4), Idiopathic CD4 lymphopenia (n=2), common variable immunodeficiency (n=1), other immunodeficiency phenotype (n=8)
PIK3CD (n=16), PIK3R1 (n=1)
AD: PLCG2 gain of function (n=1), PTEN deficiency (n=1), SAMD9L gain of function (n=1), NFKB1 haploinsufficiency (n=1)
AR: PGM3 deficiency (n=2), LRBA deficiency (n=2), Cartilage hair hypoplasia (n=1), IL10R deficiency (n=1), JAK3 deficiency (n=1)
XL: IL2RG deficiency (n=2), XIAP deficiency (n=2), CD40L deficiency (n=1), NEMO deficiency (n=1), Wiskott-Aldrich syndrome (n=1)
Related Donor Search Outcomes and Best Donor Options
For those with a disease of known mode of inheritance (n=142), the results of each PID patient’s related donor search was tabulated (Table 2), regardless of whether or not the patient went on to receive HCT. Of living 1st degree relatives (median 3 (range 0–12) per patient), a median of 1 donor remained for AD and XL diseases after HLA typing, mutation testing, and other exclusions, and a median of 2 donors remained for AR diseases. Only 20% of patients had a MRD as the best related donor option; 9% had no family donor options, even when searches were extended to 2nd degree relatives and beyond. The best related donor option was haplo for 100 (70%) patients, with 56 (39%) being haplos age 40 years or older and 5 being 2nd degree haplos. No patient had solely haplo options against whom donor-specific anti-HLA antibodies (DSAs) were present. The best related donor was a heterozygote/asymptomatic carrier of the PID mutation in 36 (49%) patients with AR or XL diseases (n=73).
Table 2.
Related donor search results
| PID patients (n=142 with known mode of inheritance) | AD (n=69) | AR (n=40) | XL (n=33) | AD, de novo& (n=28) | AD, familial& (n=31) | |
|---|---|---|---|---|---|---|
| Living 1st degree family, median number of individuals (range) | 3 (0−12) | 3 (0−12) | 3 (2−7) | 3 (1−8) | 3 (1−4) | 3 (1−12) |
| Full-match potential (full sibling among living family members), n (%) | 104 (73%) | 51 (74%) | 30(75%) | 23 (70%) | 20 (71%) | 25 (81%) |
| Living full siblings, median number (range) | 1 (0−10) | 1 (0−10) | 1 (0−5) | 1 (0−5) | 1 (0−2) | 1 (0−10) |
| Sibling excluded because of PID, n (%) | 33 (23%) | 13 (19%) | 12 (30%) | 8 (24%) | 0 | 12 (39%) |
| Sibling excluded because of non-PID reason, n (%)# | 50 (35%) | 28 (41%) | 15 (38%) | 7 (21%) | 12 (43%) | 13 (42%) |
| No siblings remaining after typing and medical evaluations, n (%) | 84 (59%) | 43 (62%) | 23 (58%) | 18 (55%) | 16 (57%) | 20 (65%) |
| Haplo potential, n (%) | 141 (99%) | 68 (99%) | 40 (100%) | 33 (100%) | 28 (100%) | 31 (100%) |
| Parent only | 34 (24%) | 15 (22%) | 9 (23%) | 10 (30%) | 8 (29%) | 4 (13%) |
| Offspring | 14 (10%) | 9 (13%) | 3 (8%) | 2 (6%) | 1 (4%) | 7 (23%) |
| Living guaranteed haplos, median number (range) | 2 (0–7) | 2 (0−6) | 2 (1−4) | 2 (0−7) | 2 (1−2) | 2 (0−6) |
| Offspring | 0 (0–5) | 0 (0−4) | 0 (0−2) | 0 (0−5) | 0 (0−1) | 0 (0−4) |
| Parents | 2 (0–2) | 2 (0−2) | 2 (1−2) | 2 (0−2) | 2 (1−2) | 2 (0−2) |
| Parent excluded because of PID, n (%) | 25 (18%) | 21 (30%) | 0 | 4 (12%) | 0 | 18 (58%) |
| Parent excluded because of non-PID reason, n (%)# | 54 (38%) | 26 (38%) | 13 (33%) | 15 (45%) | 16 (57%) | 7 (23%) |
| No parents remaining after typing and medical evaluations, n (%) | 29 (20%) | 20 (29%) | 2 (5%) | 7 (21%) | 7 (25%) | 7 (23%) |
| Offspring excluded because of PID, n (%) | 6 (4%) | 6 (9%) | 0 | 0 | 0 | 6 (19%) |
| Offspring excluded because of non-PID reason, n (%)# | 8 (6%) | 4 (6%) | 2 (5%) | 2 (6%) | 1 (4%) | 2 (6%) |
| Percent of family excluded because of PID, median (range) | 0 (0–100%) | 0% (0− 100%) | 0% (0− 40%) | 0% (0− 100%) | 0% (0− 50%) | 42% (0−100%) |
| Percent of family excluded because of non-PID reason, median (range) | 25% (0− 100%) | 29% (0− 100%) | 25% (0− 83%) | 25% (0− 100%) | 50% (0− 100%) | 8% (0− 75%) |
| Percent of family remaining after typing and medical evaluations, median (range) | 55% (0− 100%) | 50% (0− 100%) | 67% (17− 100%) | 67% (0− 100%) | 50% (0− 100%) | 50% (0− 100%) |
|
1st degree donor options after typing and medical evaluations, median (range) |
2 (0−7) | 1 (0–6) | 2 (1−7) | 1 (0−3) | 1 (0−4) | 1 (0−5) |
| 2nd degree relative search performed, n (%) | 20 (14%) | 13 (19%) | 3 (8%) | 4 (12%) | 6 (21%) | 7 (23%) |
| Number of 1st degree family members remaining after evaluations, median (range) | 1 (0−6) | 1 (0−5) | 2 (0−6) | 1 (0−3) | 1 (0−4) | 1 (0−5) |
| Any family donor after typing and medical evaluations, n (%) | 128 (90%) | 59 (86%) | 40 (100%) | 29 (88%) | 23 (82%) | 29 (94%) |
| Best family option after typing and medical evaluations, n (%)^ | ||||||
| None | 13 (9%) | 9 (13%) | 0 | 4 (12%) | 3 (11%) | 3 (10%) |
| MSD/MRD | 30 (21%) | 14 (20%) | 9 (23%)+ | 7 (21%) | 6 (21%) | 5 (16%) |
| 1st degree haplo, < 40 years old | 38 (27%) | 17 (25%) | 14 (35%) | 7 (21%) | 8 (29%) | 8 (26%) |
| 1st degree haplo, 40 years or older | 56 (39%) | 24 (35%) | 17 (43%) | 15 (45%) | 8 (29%) | 13 (42%) |
| 2nd degree haplo | 5 (4%) | 5 (7%) | 0 | 0 | 3 (11%) | 2 (6%) |
| Best family option a heterozygote/carrier for mutation, n (%) | 36 (25%) | n/a | 29 (73%) | 7 (21%) | n/a | n/a |
Abbreviations: PID, primary immunodeficiency disease; HCT, hematopoietic cell transplantation; AD, autosomal dominant; AR, autosomal recessive; XL, X-linked; MSD, matched sibling donor; MRD, matched related donor; haplo, HLA-haploidentical donor
De novo vs. familial AD disease designations were made based on family history and mutation testing in 1st degree relatives, as appropriate. From the available information, not all patients with AD disease could be designated as de novo vs. familial
Reasons for exclusion could include medical issues, estrangement, lack of willingness/availability to be typed/donate, or HLA disparity on HLA typing. Age, blood type, Cytomegalovirus serostatus, parity, donor-specific anti-HLA antibodies, and asymptomatic mutation carriers were not counted here as reasons for exclusion.
Best family option was determined by prioritizing the following among remaining family options: degree of HLA match (MSD, MRD, haplo with fewest mismatches on the non-shared haplotype), donor age (adults < 40, minors, adults > 40), parity/sex (avoidance of female donor for male recipient; otherwise equivalent male donor preferred to multiparous female), donor size/fitness/availability, family preference (parents of patients often requested that the donor be a parent if choosing among parent and sibling haplo options)
MSD=6, MRD=3
Among those whose best family donor option was a haplo comfirmed by HLA typing (excluding families with known close consanguinity, n=9), 35 of 65 (54%) had a donor who was at least a 6/10 bi-directional HLA-match or better, 17 (26%) had a donor who was at least a 7/10 bi-directional HLA-match or better, and 8 (12%) had a donor who was at least a 8/10 bi-directional HLA-match or better; none of the remaining mismatches were HLA-DQ mismatches for these patients having partial sharing on the unshared haplotype. For the 9 patients whose parents were 1st cousins, none had a suitable MSD, but 3 (33%) had a HLA-identical donor in the form of a parent or cousin.
Of the 19 patients with a PID of unknown genetic defect, all had living, healthy 1st degree relatives who could have served as donors had a mutation been identified to inform donor screening. Ten of the 19 had full siblings who had HLA typing performed, often in the hopes that a mutation would be found during the course of the donor search, and 5 of the 10 (50%) had at least 1 HLA-matched sibling.
MUD Search Results
MUD searches were performed for 139 patients, with MSKCC SP categorization shown in Table 3. The vast majority (88%) of HCT recipients with Very Good/Good MUD search predictions (n=57) received a MUD HCT, while no patient with Very Poor/Futile MUD search (n=53) went on to receive a 8/8 MUD graft, Table 4. Among the patients with Very Poor/Futile MUD searches included 6 of 19 (32%) of those with PID of unknown genetic defect and therefore no mechanism to genetically screen family donors; 3 of these patients went on to receive HCT using a mMUD graft (n=1) or, with trepidation, a family donor after extensive immunologic evaluations (n=2),and 3 have not yet been offered HCT due to lack of donor. Of the 65 patients whose best family donor was haplo and donor HLA typing was available, 30 (46%) had a Very Poor/Futile MUD search.
Table 3.
MSKCC Search Prognosis categorization by ancestry
| All Patients with MUD Searches (n=139) | Very Good/Good (n=57) | Fair/Poor (n=29) | Very Poor/Futile (n=53) | P-value | |
|---|---|---|---|---|---|
| General Ancestry | 0.002 | ||||
| European | 83 | 44 (53%) | 14 (17%) | 25 (30%) | |
| Non-European | 56 | 13 (23%) | 15 (27%) | 28 (50%) | |
| Detailed Ancestry | 0.0003 | ||||
| Northwest European | 73 | 42 (58%) | 9 (12%) | 22 (30%) | |
| Hispanic | 24 | 4 (17%) | 3 (13%) | 17 (71%) | |
| Non-NW European or Slavic | 12 | 3 (25%) | 5 (42%) | 4 (33%) | |
| African | 8 | 1 (13%) | 2 (25%) | 5 (63%) | |
| Asian | 8 | 3 (38%) | 3 (38%) | 2 (25%) | |
| Middle Eastern | 5 | 2 (40%) | 3 (60%) | 0 | |
| European/Non-European Mix | 4 | 2 (50%) | 2 (25%) | 0 | |
| Non-European Mix | 3 | 0 | 0 | 3 (100%) | |
| Native North American | 2 | 1 (50%) | 1 (50%) | 0 | |
Abbreviations: MUD, matched unrelated donor; NW, Northwest
Table 4.
MUD search outcomes by MSKCC Search Prognosis categorization and ancestry
| Number of MUD searches performed | Number of MUD searches formalized | MUD identified, n (% of formalized searches) | Patients receiving HCT, n (% of patients with MUD search) | 8/8 MUD as donor, n (% of HCT recipients) | |
|---|---|---|---|---|---|
| Total | 139 | 65 | 54 (83%) | 105 (76%) | 47 (45%) |
| Very good/Good | |||||
| European | 44 | 33 | 33 (100%) | 33 (75%) | 29 (88%) |
| Non-European | 13 | 10 | 9 (90%) | 10 (77%) | 9 (90%) |
| Fair/Poor | |||||
| European | 14 | 9 | 5 (56%) | 12 (86%) | 5 (42%) |
| Non-European | 15 | 7 | 6 (86%) | 12 (80%) | 4 (33%) |
| Very Poor/Futile | |||||
| European | 25 | 3 | 0 (0%) | 19 (76%) | 0 (0%) |
| Non-European | 28 | 3 | 1 (33%) | 19 (68%) | 0 (0%) |
Abbreviations: MUD, matched unrelated donor; MSKCC, Memorial Sloan Kettering Cancer Center; HCT, hematopoietic cell transplantation
Of the 139 patients with MUD searches performed, 105 (76%) went on to receive HCT, with no significant difference in rates by European compared to non-European ancestry, Table 4. However, utilization of MUDs as a graft type varied by patient ancestry, with 34 (53%) of HCT recipients of European descent (n=64) receiving a MUD graft and 13 (32%) of HCT recipients of non-European descent (n=41) receiving a MUD graft, p=0.03. The predicted likelihood of a successful 8/8 MUD search was lower for those of non-European ancestry, p=0.01. The majority of patients of Hispanic or African ancestry had Very Poor/Futile MUD searches, representing 71% and 63% of prediction designations for these groups, respectively, Figure 1.
Figure 1.
MUD search categorization, by recipient ancestry, using the Memorial Sloan Kettering Search Prognosis algorithm. There are significant differences in the likelihood of a successful MUD search by ancestry, χ2 p=0.0003.
Receipt of HCT and Role of Alternative Donors
For most of these patients, the underlying PID diagnosis was not an absolute indication for HCT, but rather was influenced both by the genetic diagnosis and severe clinical manifestations. Thus, HLA typing was often performed early in the discussion of treatment options for PID, even before HCT was an immediate necessity; thus, the median time from HLA typing to HCT was 7 months (range 1–50 months). Of patients referred to HCT, 70% have gone on to receive HCT with an additional 25% pending HCT with a donor identified. Of those with PID of known mode of inheritance (n=142) with a MRD option, 86% (24 of 28) have gone on to receive HCT and the 3 of the remaining 4 have HCT scheduled in the coming months. While fewer MSDs were utilized for recipients with AR diseases, 44% of patients with AR diseases and parents who were 1st cousins had a donor who was 8–10/10.
Alternative donors were utilized in 77% of all PID HCTs; unrelated donors were utilized in 45% of the HCTs for this cohort, with the second most common donor type being haplo parents, comprising 22% of HCTs. As expected, patients receiving HCT for a PID of unknown genetic defect were more likely to receive an unrelated donor graft, but notably the disposition of these patients was also that they were more likely to have no donor found despite related and unrelated donor searches, Table 1. Of the 19 with PID of unknown genetic defect, 6 (32%) had no MUD option. Use of mMUD was only observed in patients with PID of unknown genetic mutation and use of 2nd degree haplo donors was only observed in those with AD diseases. The most commonly utilized donor for AR diseases was a haplo parent, comprising 37% of the AR HCT grafts. Of those with PID of known mode of inheritance but no family donor options, 71% (10 of 14) have received HCT; of the remaining 4, 2 have MUD HCTs scheduled, 1 is well enough to defer HCT, and 1 died before a donor could be found.
Discussion
In the modern era of successful approaches to haplo HCT such as PTCy-based platforms,9–11 the prevailing notion is that there is a readily available donor option for any patient in need of HCT. However, our study emphasizes the increased complexity of related donor searches for patients with inherited diseases is emphasized here. This complexity is further heightened by focusing on patients with PID who come to HCT later than those who present in and are transplanted in infancy, where there may often be, at minimum, an immediately available, unaffected, relatively young, and guaranteed HLA-haploidentical parent. While family donors are usually considered to be more readily available than unrelated donors, the reverse may often be true for patients with PID, as determining related donor suitability requires genetic testing for the mutation, as well as at times subsequent immune profiling and evaluations of carriers to ensure that the heterozygous state is asymptomatic. This represents an additional time and resource burden as well as expertise to evaluate related donors in the PID setting. Notably, family members in this cohort were at times reluctant to undergo genetic testing, wishing to not know their own mutation status or, in particular, parents wishing to not know the mutation status of their other children, the patient’s siblings. This highlights the importance of proactively outlining the family tree and exploring related donor options as soon as a patient with PID is considered to be potentially in need of HCT. Inquiries into related donor options always have the potential to reveal intra-family psychosocial tensions or unease, and these complicated dynamics seem to be more common in working with the PID population, when genetic testing is also required.
While the most common haplo donor option for a hematologic malignancy patient may be a child or sibling given that the median ages of patients with hematologic malignancies are in 5th–7th decades, the most common haplo donor for a PID patient, who is typically younger, may be a parent, as shown in our cohort. The reliance on haplo parents in this cohort was due to small family sizes, lack of offspring among PID patients, affected siblings, and also often parental preference to serve as donor if the patient’s siblings were not fully matched and the best family option was a choice between haplo parent or haplo sibling. However, age is emerging as an important factor in donor selection and the majority of haplo parents in our series were over age 40 years or greater. In CIBMTR studies of unrelated donor factors that affect HCT survival outcomes, donor age was found to be a major factor, with decreases in overall survival with increasing donor age on multivariable analysis.12,13 On adjusted analysis, survival was superior if MUDs < age 33 years were utilized.13 Smaller studies have also shown that matched related or unrelated donors > 39 years are associated with inferior overall survival,14 and that MUDs < 30 years are associated with improved OS compared to MRDs and older MUDs.15 Age has been shown to be a detrimental factor in hematopoietic stem cell function, as well as associated with an increased risk of clonal hematopoiesis.16,17 How these data on donor age generalize to haplo donors with PTCy-based approaches requires further investigation, although at least one major HCT center that utilizes PTCy, Johns Hopkins, has moved to selecting donors based on younger age over degree of HLA match, including 2nd or 3rd degree family members within the pool of potential donors.18,19
In addition to age considerations when using haplo parents as donors, mutation carrier status is also a factor in AR and XL diseases. In our series, mutation carriers were used for over one-third of AR HCTs. While carriers were evaluated with diligence to confirm that they were asymptomatic, there may be unknown consequences to using a heterozygote as donor in some newly identified PIDs. As an example, it was noted that a donor who was a carrier for ADA2 deficiency (DADA2) had a very low cell count yield on bone marrow harvest.20 To secure a sufficient graft dose, the donor had to urgently undergo mobilization for peripheral blood stem cell collection to augment the marrow graft. Whether this was a donor-specific issue or a feature among heterozygotes of ADA2 (CECR1) mutations is not presently known. Furthermore, there are reports of symptomatic patients heterozygous for ADA2 mutations, with plasma ADA2 activity levels in the carrier range and reports of these patients at times presenting with milder or later-onset disease manifestations.21–23 Another example is the association between discoid lupus erythematosus and CGD, which is primarily reported not in CGD patients but in the female carriers of XL CGD mutations, as well as heterozygotes of AR CGD;24–26 this association is not well understood but is clearly an example of the uncertainty in the true health of carriers of certain mutations. Interestingly, in evaluating the haplo parents of a patient in our cohort with IL10RA mutation (AR inheritance), the father was excluded for a history of tumid lupus, the susceptibility to which may be related, at least partially, to his IL10RA mutation carrier status.27 These cautionary examples serve to illustrate our incomplete understanding of the potential risk associated with using carriers as donors. Finally, another consideration is that mixed chimerism may be insufficient to correct a disease phenotype, as has been demonstrated mathematically in SCD and hemoglobinopathies.28,29 The use of a carrier donor in PID HCT might make it necessary to achieve higher donor chimerism in critical cell lineages for phenotype reversal.
The lack of a MSD does not always equate to no matched related donor options. A high percentage of patients with AR diseases and close consanguinity of their parents had MRDs in this study and more likely could have been found with more extensive family typing. Additionally, haplo donors are frequently more closely matched than 5/10, sharing additional HLA alleles from the non-shared haplotype. In our series, even 8/10 bi-directionally matched haplos were found in families without consanguinity. In sum, all haplo donors are not equal with regard to the degree of HLA disparity, but discerning how this should inform donor selection, if at all, remains unclear at this time and there may be no clinical differences in outcome based on degree of HLA disparity among haplo donors with PTCy-based approaches.30,31 Nonetheless, we note the degree of matching in the graft-versus-host and host-versus-graft direction for all haplo donors during the donor search evaluation. Furthermore, DSAs were not an issue that limited haplo donor options in this study, perhaps reflecting several somewhat unique features of this patient population as compared to patients with hematologic malignancy referred for HCT,32 including lower parity of females with PID, less exposure to transfusions, and perhaps less propensity to form DSAs by nature of their immunodeficiency and/or its treatment.
Demonstrated here is the great importance of unrelated donor options for these patients, as the number of eligible and suitable family donors was often small and weighed heavily on haplo parents who were typically age 40 years or more. Thus, in our practice, we have initiated a MUD search immediately upon referral for HCT, so as to know as quickly as possible what donor options exist, anticipating delays or potential difficulties in the course of related donor searches. This is distinctly different than the current approach to donor searches for hematologic malignancies, where not all patients require and benefit from a MUD search.4 MMUDs were only used for patients with PID of unknown genetic defect, speaking to the willingness to use haplo donors over mMUDs as an alternative donor source when non-HLA selection factors, such as the underlying mutation, were known and therefore could be evaluated in family members. The donor search process for patients with PID of unknown genetic defect can be quite agonizing when the MUD search is unpromising. The process of trying to deem suitable a related donor for a patient with an unknown genetic defect is time consuming, often unsettling for the patient, family, and transplant physician, and ultimately guesswork at best in an effort to weigh the relative risks and benefits of proceeding with HCT versus continued evolution of the underlying PID.Although umbilical cord blood (UCB) grafts were not considered in this study, as PID HCT protocols at the NIH during the study timeframe did not utilize UCB grafts, UCB provides an additional alternative donor source for patients who lack HLA-matched donors. Adequate-cell-dose cord blood units mismatched at 1 or 2 HLA loci are likely available for nearly all pediatric patients and for the large majority of adults, regardless of ethnicity.3 Thus, patients in this study may have had UCB options that could have been considered in the alternative donor search. Gene therapy is also a potential option for a small subset of patients with specific PIDs,33 but does not currently provide an alternative to HCT for the majority of PID patients.
Approaches to HCT that successfully utilize alternative donors are critical for extending this potentially curative therapy to all eligible PID patients. At the NIH, PTCy-based approaches are used for the majority of PID HCTs, even when using matched related or unrelated donors. Yet, while identifying successful platforms to enable alternative donor HCT is one hurdle that has been largely cleared for PID patients, a remaining hurdle is the donor search itself in this unique patient population. Awareness of and attention to the complexities that surround donor searches for PID patients, as highlighted here, are necessary to successfully bring those PID patients in need of HCT to HCT in an optimal and timely fashion.
Highlights.
The best donor options for PID patients are often unrelated or older, HLA-haploidentical donors
Inherited diseases, particularly autosomal dominant, can diminish related donor options
Additional evaluations, such as genetic testing, add complexity to donor searches for PID patients
Acknowledgements
This project has been funded in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.McCullough J, Perkins HA, Hansen J. The National Marrow Donor Program with emphasis on the early years. Transfusion 2006; 46(7): 1248–55. [DOI] [PubMed] [Google Scholar]
- 2.Hsieh MM, Kang EM, Fitzhugh CD, et al. Allogeneic hematopoietic stem-cell transplantation for sickle cell disease. N Engl J Med 2009; 361(24): 2309–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gragert L, Eapen M, Williams E, et al. HLA match likelihoods for hematopoietic stem-cell grafts in the U.S. registry. N Engl J Med 2014; 371(4): 339–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ciurea SO, Bittencourt MCB, Milton DR, et al. Is a matched unrelated donor search needed for all allogeneic transplant candidates? Blood Adv 2018; 2(17): 2254–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dew A, Collins D, Artz A, et al. Paucity of HLA-identical unrelated donors for African-Americans with hematologic malignancies: the need for new donor options. Biol Blood Marrow Transplant 2008; 14(8): 938–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Grewal SS, Barker JN, Davies SM, Wagner JE. Unrelated donor hematopoietic cell transplantation: marrow or umbilical cord blood? Blood 2003; 101(11): 4233–44. [DOI] [PubMed] [Google Scholar]
- 7.Dehn J, Setterholm M, Buck K, et al. HapLogic: A Predictive Human Leukocyte Antigen-Matching Algorithm to Enhance Rapid Identification of the Optimal Unrelated Hematopoietic Stem Cell Sources for Transplantation. Biol Blood Marrow Transplant 2016; 22(11): 2038–46. [DOI] [PubMed] [Google Scholar]
- 8.Davis E, Devlin S, Cooper C, et al. Validation of an Algorithm to Predict the Likelihood of an 8/8 HLA-Matched Unrelated Donor at Search Initiation. Biol Blood Marrow Transplant 2018; 24(5): 1057–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Luznik L, O’Donnell PV, Symons HJ, et al. HLA-haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biol Blood Marrow Transplant 2008; 14(6): 641–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Brodsky RA, Luznik L, Bolanos-Meade J, Leffell MS, Jones RJ, Fuchs EJ. Reduced intensity HLA-haploidentical BMT with post transplantation cyclophosphamide in nonmalignant hematologic diseases. Bone Marrow Transplant 2008; 42(8): 523–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kanakry CG, Fuchs EJ, Luznik L. Modern approaches to HLA-haploidentical blood or marrow transplantation. Nat Rev Clin Oncol 2016; 13(1): 10–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shaw BE, Logan BR, Spellman SR, et al. Development of an Unrelated Donor Selection Score Predictive of Survival after HCT: Donor Age Matters Most. Biol Blood Marrow Transplant 2018; 24(5): 1049–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kollman C, Spellman SR, Zhang MJ, et al. The effect of donor characteristics on survival after unrelated donor transplantation for hematologic malignancy. Blood 2016; 127(2): 260–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ayuk F, Zabelina T, Wortmann F, et al. Donor choice according to age for allo-SCT for AML in complete remission. Bone Marrow Transplant 2013; 48(8): 1028–32. [DOI] [PubMed] [Google Scholar]
- 15.Kroger N, Zabelina T, de Wreede L, et al. Allogeneic stem cell transplantation for older advanced MDS patients: improved survival with young unrelated donor in comparison with HLA-identical siblings. Leukemia 2013; 27(3): 604–9. [DOI] [PubMed] [Google Scholar]
- 16.Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med 2014; 371(26): 2488–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chambers SM, Shaw CA, Gatza C, Fisk CJ, Donehower LA, Goodell MA. Aging hematopoietic stem cells decline in function and exhibit epigenetic dysregulation. PLoS Biol 2007; 5(8): e201. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Elmariah H, Kasamon YL, Zahurak M, et al. Haploidentical Bone Marrow Transplantation with Post-Transplant Cyclophosphamide Using Non-First-Degree Related Donors. Biol Blood Marrow Transplant 2018; 24(5): 1099–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.McCurdy SR, Zhang MJ, St Martin A, et al. Effect of donor characteristics on haploidentical transplantation with posttransplantation cyclophosphamide. Blood Adv 2018; 2(3): 299–307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hsu AP, West RR, Calvo KR, et al. Adenosine deaminase type 2 deficiency masquerading as GATA2 deficiency: Successful hematopoietic stem cell transplantation. J Allergy Clin Immunol 2016; 138(2): 628–30 e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Meyts I, Aksentijevich I. Deficiency of Adenosine Deaminase 2 (DADA2): Updates on the Phenotype, Genetics, Pathogenesis, and Treatment. J Clin Immunol 2018; 38(5): 569–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rama M, Duflos C, Melki I, et al. A decision tree for the genetic diagnosis of deficiency of adenosine deaminase 2 (DADA2): a French reference centres experience. Eur J Hum Genet 2018; 26(7): 960–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Caorsi R, Penco F, Grossi A, et al. ADA2 deficiency (DADA2) as an unrecognised cause of early onset polyarteritis nodosa and stroke: a multicentre national study. Ann Rheum Dis 2017; 76(10): 1648–56. [DOI] [PubMed] [Google Scholar]
- 24.Cale CM, Morton L, Goldblatt D. Cutaneous and other lupus-like symptoms in carriers of X-linked chronic granulomatous disease: incidence and autoimmune serology. Clin Exp Immunol 2007; 148(1): 79–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Garioch JJ, Sampson JR, Seywright M, Thomson J. Dermatoses in five related female carriers of X-linked chronic granulomatous disease. Br J Dermatol 1989; 121(3): 391–6. [DOI] [PubMed] [Google Scholar]
- 26.Rupec RA, Petropoulou T, Belohradsky BH, et al. Lupus erythematosus tumidus and chronic discoid lupus erythematosus in carriers of X-linked chronic granulomatous disease. Eur J Dermatol 2000; 10(3): 184–9. [PubMed] [Google Scholar]
- 27.Qi ZM, Wang J, Sun ZR, et al. Polymorphism of the mouse gene for the interleukin 10 receptor alpha chain (Il10ra) and its association with the autoimmune phenotype. Immunogenetics 2005; 57(9): 697–702. [DOI] [PubMed] [Google Scholar]
- 28.Fitzhugh CD, Cordes S, Taylor T, et al. At least 20% donor myeloid chimerism is necessary to reverse the sickle phenotype after allogeneic HSCT. Blood 2017; 130(17): 1946–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Roberts C, Kean L, Archer D, Balkan C, Hsu LL. Murine and math models for the level of stable mixed chimerism to cure beta-thalassemia by nonmyeloablative bone marrow transplantation. Ann N Y Acad Sci 2005; 1054: 423–8. [DOI] [PubMed] [Google Scholar]
- 30.Kasamon YL, Luznik L, Leffell MS, et al. Nonmyeloablative HLA-haploidentical bone marrow transplantation with high-dose posttransplantation cyclophosphamide: effect of HLA disparity on outcome. Biol Blood Marrow Transplant 2010; 16(4): 482–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.McCurdy SR, Kanakry CG, Tsai HL, et al. Grade II Acute Graft-versus-Host Disease and Higher Nucleated Cell Graft Dose Improve Progression-Free Survival after HLA-Haploidentical Transplant with Post-Transplant Cyclophosphamide. Biol Blood Marrow Transplant 2018; 24(2): 343–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gladstone DE, Zachary AA, Fuchs EJ, et al. Partially mismatched transplantation and human leukocyte antigen donor-specific antibodies. Biol Blood Marrow Transplant 2013; 19(4): 647–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Thrasher AJ, Williams DA. Evolving Gene Therapy in Primary Immunodeficiency. Mol Ther 2017; 25(5): 1132–41. [DOI] [PMC free article] [PubMed] [Google Scholar]