Retrospective, Landmark Analysis of Long-term Adult Morbidity Following Allogeneic HSCT for Inborn Errors of Immunity in Infancy and Childhood

Eligibility criteria for this retrospective study included (i) HSCT before the age of 16 years and (ii) 5 or more years survival following first transplant. We identified 212 IEI patients who were transplanted < 16 years at Great Ormond Street Hospital for Sick Children NHS Foundation Trust (GOSH) or University College London Hospitals NHS Foundation Trust (UCLH). Of these, 81 had died within five years of transplantation. Of the remaining 131 patients contacted, 83 consented to take part in our long-term psychological and medical outcome studies, while 48 declined to do so. All participating patients provided written informed consent as per institutional guidelines.

Data were collected from clinic letters, laboratory results, imaging and inpatient hospital records. Information on genetic/clinical diagnosis, HSCT characteristics (donor, conditioning, age at transplant, era of transplant), chimerism, immune reconstitution, GVHD, infectious complications, non-immunological medical outcomes and fertility was collected.

The median age at last follow-up was 25 years (range: 17–39 years), with a median post-transplant follow-up duration of 20 years (range: 8–39 years; IQR: 7). Median age at transplant was three years (range: 0–18 years), 57 patients were male (69%) and 26 female (31%). Detailed patient demographics and transplant characteristics are shown in Table 1. IEIs were diagnosed using international criteria [26]. Underlying diagnoses were categorized into 3 groups: severe combined immunodeficiency (SCID) (n = 37); combined immunodeficiency (CID) (n = 33) and phagocyte disorders (n = 13), with details in Table 1. Median age at HSCT was significantly younger for SCID patients (5 m) when compared to CID (9 yr 8 m, p = 0.0001) and PD (6 yr, p = 0.0001) patients.

Demographic data of long-term survivors who declined to take part (n = 39, 81% of the 42 not reported) demonstrates the study population can be considered broadly representative of all long-term survivors. There was no significant difference between the distribution of underlying diagnoses (p = 0.834), median length of follow-up post HSCT (p = 0.127), age at HSCT (p = 0.034) or era of HSCT (p = 0.063) between the study patients and the patients not consenting to the study. Further, OS was 98% for those in the study (n = 83) and 96% for non-consented, p = 0.437. Data not shown.

Data on donor status, conditioning regimen intensity, age at transplant and era of transplant are also shown in Table 1. 36 patients received stem cells from 10/10 HLA-matched unrelated donors (MUD), 31 patients had 10/10 HLA-matched related donors (MRD), 10 transplant recipients had haploidentical donors (Haplo) and six patients had either one or two Ag-mismatched unrelated donors (MMUD). Most patients (n = 69; 83%) received conditioning chemotherapy prior to infusion of allogeneic stem cells. 38 patients (46%) received various reduced intensity conditioning (RIC) regimens, in line with the era of transplant including fludarabine and melphalan (Flu/Mel), fludarabine and busulphan (Flu/Bu) or fludarabine and treosulphan (Flu/Treo). Of the patients receiving myeloablative conditioning (MAC), 31 received busulfan and cyclophosphamide (Bu/Cy) and one received fludarabine and total body irradiation (TBI). A total of 30 patients (36%) were transplanted before one year of age, 17 patients between one and four years, and 36 at the age of five years or older (only two patients were transplanted above the age of 15). Due to the time required post-HSCT for patients to transition to adult care and be eligible for recruitment into this study those under the age of 5 yrs at HSCT where HSCT was performed after 2006 would not have reached the age required to be included in this study. To assess the potential impact of change in transplant practice over time, our patients were also grouped by transplant era. 19 patients received their first HSCT before 1996, 26 between 1996 and 2001, 27 between 2001 and 2006, and a further 11 after 2006. Reflecting changes in transplant practice over time, no RIC transplants were performed pre 1996 an no unconditioned transplants were performed after 2006.

Seven patients (8.4% of the study cohort) had received a second transplant at an average of 1.7 years post HSCT, but only 3 of these occurred beyond 5 years post initial HSCT and were included as events.

OS and EFS were calculated from 5 years post-transplant in this landmark analysis. Events occurring prior to 5 years were not considered in these analyses. Probability of EFS was calculated using the Kaplan–Meier method with Stata 16 (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC.) and comparisons of EFS curves were made using a multi-state Cox model, with event-free, event and death as the states and independent transitions between the states. Transplant characteristics, chimerism and immune reconstitution, and the occurrence of medical events were assessed in univariate Cox models, and all factors with p < 0.2 were included in multivariable models. Factors not significant in the multivariable model were then excluded. The proportional hazards assumption of the Cox model was checked by visual inspection of the proportional hazard (log–log) plots (for categorical variables) and by the Schoenfeld residuals test. The only univariate models for which the proportionality test failed were year of transplant, either by five-year interval (p = 0.012) or as a continuous variable (p = 0.011). Likewise, the multivariable model showed no evidence for violation of proportionality (p = 0.34). For time varying factors (such as chimerism) the patients’ time after transplant was further split at each time point when chimerism was tested. Multivariable models were constructed using all terms with > 90% data completeness and with p < 0.2 in univariate models.

Lymphocyte subset counts (Table 4) for adults are presented raw, but those for pediatric patients (while they were < 16 years) were normalized to the median of the age-adjusted reference range. The reference ranges were similarly normalized and averaged over the age range. The reference median is thus 1.0 and the reference range displayed is the mean ratio of the reference limits.

At 20 years post-transplant, the EFS was 51% in the overall cohort (Fig. 1A); 54% in the SCID group, 50% in the CID group and 54% in the phagocyte disorder group (p = 0.35, Table 2). Univariate analysis identified age at transplant and CD4 count at five years post-transplant as significantly impacting 20 yr EFS (Table 2), although only 2 patients had CD4 counts < 0.2 × 109/l at 5 years post-transplant. Patients transplanted at age < 1 yr, 1-4yrs and ≥ 5yrs had subsequent 20 yr EFS of 60%, 51% and 41% respectively (p = 0.04 for age ≥ 5yrs compared to < 1 yr at HSCT, Fig. 1B and Table 2).

However, year of HSCT, donor, conditioning intensity, underlying diagnosis, whole blood chimerism at 1 year and lineage-specific chimerism had no significant impact on EFS by univariate analysis (Table 2).

Multivariable analysis confirmed only age at transplant (p = 0.002, Table 2) and whole blood chimerism, at the fixed time point of five years post HSCT (p = 0.006, Table 2) as independently significant in predicting EFS beyond 5 years post-transplant. However, when analysis was performed in specific IEI subgroups, the impact of WB chimerism at 5 years on subsequent EFS was only significant for the SCID patients (Table 3). Within the SCID group, both age at HSCT and WB chimerism at 5 yrs influenced EFS (p = 0.01, p = 0.03, respectively). However, only age at HSCT was found to be significant for EFS in the PD group (p = 0.03), Table 3.

A total of 59 significant medical events were documented in 45% of the cohort (37 patients) (Supplementary Table 1). The most frequent events were chronic viral infection (n = 14), followed by recurrent/severe acute infection (n = 14), autoimmune disease (n = 10), moderate/severe organ dysfunction (n = 12), malignancy (n = 2), graft related secondary intervention (n = 3), moderate/severe GVHD (n = 2) and death (n = 2). All clinical events are described in Supplementary Table 1.

Two patients died after enrollment in the study at 33 and 10 years post-transplant, resulting in an overall survival in our cohort of 97% at 20 years (Fig. 1A). Cause of death included hepato-pulmonary syndrome secondary to hepatic nodular regenerative hyperplasia and bronchiectasis in the context of failed B cell immune reconstitution in a patient transplanted for SCID, and sudden cardiac death in a patient transplanted for X-CGD (P3 and P24, respectively, Supplementary Table 1).

Overall, 69% (n = 58) of patients were event-free at their last follow-up appointment (at a median of 20 years post-transplant), including those who had never had an event during the study period (n = 46) and those in whom prior events had resolved and no new events had occurred in the last 12 months of follow-up (n = 12). A total of 25 patients (30% of the cohort) were experiencing ongoing significant complications, including two patients who died of these, as discussed above. Only 11 (13%) patients had more than one clinically significant event.

Long-term chronic GVHD requiring systemic therapy was very rare, affecting two patients. While six patients (7% total cohort) suffered from chronic GVHD beyond five years post-transplant, only two had moderate/severe requiring systemic treatment and defined as events, one SCID patient after a CD34 + top-up and the other following RIC MRD allograft for XLP (P15 and P20). Four patients had limited chronic skin GVHD at last assessment, requiring topical corticosteroids only and were not included as significant events.

Nine patients (11% of total cohort) developed 10 de novo autoimmune events at a median of 10.5 years post-HSCT at a range of 3–17 years (P13 developed juvenile idiopathic arthritis, JIA at three years post-HSCT that persisted beyond five years post-HSCT). These diagnoses included autoimmune joint disease (n = 3), type 1 diabetes mellitus (n = 1), autoimmune hemolytic anemia (n = 2), transverse myelitis (n = 1), Guillain–Barre syndrome (n = 1), keratitis (n = 1) and uveitis (n = 1). Autoimmune diseases were distributed across the IEI diagnostic groups. No association between autoimmune disease and peripheral blood chimerism was observed. Two (2%) patients had documented secondary hypothyroidism (one with MAC and the other with RIC) without auto antibodies.

Three patients (4% of the cohort) developed late onset moderate/severe renal or hepatic sequelae (chronic kidney disease (P13, P26) and liver failure (P3)). P13 received an unconditioned MRD infusion for genetically undefined SCID, developed JIA at three years post-HSCT requiring prolonged systemic immunosuppression and subsequently developed CKD-stage 3A at five years post-HSCT, presumed secondary to medications. P26, who received a RIC Fu/Bu/Alemtuzumab MUD allograft for X-CGD, developed CKD-stage 3A with hypertension and proteinuria at 7 years post-HSCT controlled with Ramipril, and related to a pre-HSCT vascular insult. P3 received a conditioned Haplo for homozygous RAG2 Omenn syndrome. She had a CD34 + top-up at 14 years post-HSCT for poor B cell recovery and absent myeloid engraftment associated with recurrent LRTI and bronchiectasis despite adequate IRT. She died of liver failure secondary to nodular regenerative hyperplasia (NRH) and hepato-pulmonary syndrome.

Five patients were diagnosed with chronic lung disease beyond five years post-HSCT: four cases of bronchiectasis and one case of bronchiolitis obliterans. Three patients with bronchiectasis were immunoglobulin dependent at last follow-up in the absence of B cell engraftment: P3 (RAG2 Omenn), P11 (Gamma chain SCID) and P19 (WAS). The fourth patient, P28 (genetically undefined SCID) had recurrent lower respiratory tract infections (LRTI) despite 100% engraftment across all lineages, with initial suboptimal pneumococcal vaccine responses, corrected by re-vaccination.

P27 developed focal bronchiolitis obliterans at five years post RIC MUD allograft for X-CGD. He had a whole blood chimerism level of > 95% at last follow-up and good immune reconstitution with adequate response to tetanus and pneumococcal vaccines. There was no evidence of pulmonary GVHD in our study population.

Three patients received a CD34 + top-up at a median of 17 years post-HSCT (range: 14–19 years) either to correct persistent cytopenias despite full donor chimerism (P37) or to correct persistent mixed chimerism (P3, P15). Two had myeloablative transplants (P3 and P15, respectively) and one received an unconditioned MSD infusion (P37). No impact on B cell engraftment was observed (see Supplementary Table 1).

Two (2%) patients developed late malignancies including squamous cell carcinoma at 25 years (P15) and recurrent basal cell and squamous cell carcinoma at four and six years post-HSCT (P26), respectively. P15 (gamma chain SCID) also had chronic GVHD causing alopecia totalis and recalcitrant warts (HPV serotypes 17, 27, and 18) despite good immune reconstitution. His SCC (scalp) was positive for HPV18. P26 (X-CGD) was previously reported by Unni et al., 2018 and despite achieving 100% full donor chimerism required prolonged voriconazole for prophylaxis of CGD-associated fungal infections. He initially developed pre-malignant actinic keratosis and subsequently developed basal cell carcinoma of the ear and recurrent SCC of the lower leg at four years and then six years post-HSCT. He had chronic HPV-associated viral warts as well as chronic skin GVHD. A further two patients developed benign tumors: tubulo-papillary adenoma of the gallbladder at 17 years post-HSCT (P21) and a schwannoma of the lower jaw (P18) at 18 years post-HSCT.