Graft-versus-host Disease After Double-Unit Cord Blood Transplantation Has Unique Features and an Association with Engrafting Unit-Recipient HLA-match

Patient and Graft Characteristics

This analysis was performed in patients transplanted at Memorial Sloan-Kettering Cancer Center (MSKCC) between October 1, 2005 and February 28, 2011. All CBT recipients during this time period received double-unit grafts. Patients eligible for this analysis included all consecutive adult and pediatric DCBT recipients of first allograft transplanted for the treatment of hematologic malignancies excluding acute leukemia patients transplanted with > 20% bone marrow blasts. All patients provided written informed consent for transplantation according to the principles of the Declaration of Helsinki, and transplantation outcome analysis was approved by the MSKCC Institutional Review and Privacy Board.

CB units were selected on the basis of 4–6/6 HLA-A,-B antigen, -DRB1 allele match to the recipient, a cryopreserved total nucleated cell (TNC) dose of at least 1.5 × 10⁷/ kilogram (kg)/ unit, and the bank of origin as previously described⁷. Unit-unit HLA-match was not considered in CB unit selection. High resolution HLA-A,-B,-C,-DRB1 and DQ allele typing of CB units was performed routinely but usually did not influence unit selection. Units were thawed using albumin-dextran dilution⁸ (n = 205) or thawed with wash (n = 25).

Conditioning Regimens and GVHD Prophylaxis and Treatment

All patients were hospitalized in high-efficiency particulate air filtered rooms and received similar supportive care. Pre-transplant conditioning varied according to patient’s age, diagnosis, remission status, extent of prior therapy, and co-morbidities, and consisted of myeloablative (high-dose and reduced intensity), and non-myeloablative regimens (Table 1). Granulocyte-colony-stimulating factor (5 mcg/ kg/ day) was given to all patients after transplant until neutrophil recovery.

All patients received a calcineurin-inhibitor and MMF for GVHD prophylaxis starting day -3, and none received ATG^4,9. One-hundred and four patients received cyclosporine-A targeting a trough level of 200–400 ng/ml, and eleven patients received tacrolimus targeting a trough level of 5–15 ng/ml. In the first 83 patients MMF was dosed at 1 gram (or 15 mg/kg if < 50 kg) every 12 hours; the total daily dose was increased (1 gram/dose if > 50 kg, 15 mg/kg/dose if < 50 kg and > 12 years, and 20 mg/kg/dose to a maximum of 1 gram if < 12 years, all every 8 hours) in the subsequent 32 patients to augment GVHD prophylaxis¹⁰. MMF was initiated intravenously in all patients and this was maintained during hospitalization and switched to oral prior to discharge. In the absence of GVHD, MMF was tapered after day 45, and the calcineurin-inhibitor was tapered after day 100. Patients with grade II–IV aGVHD were initially treated with either single-agent budesonide (9 mg daily) or systemic corticosteroids (1–2 mg/kg oral prednisone or intravenous methylprednisolone) at the treating physician’s discretion according to the clinical GVHD severity at the time of diagnosis.

Study Definitions

Disease risk, time to neutrophil and platelet recovery, and sustained donor engraftment were defined as previously described^4,11. Patient ancestry was classified as European when there was no known non-European ancestry. Non-Europeans were Asian, African, white Hispanic, and Middle Eastern patients as well as those with mixed non-European and European origins¹². Donor chimerism was determined serially in the bone marrow and blood using semi-quantitative analysis of polymerase chain reaction amplified informative polymorphisms^11,13.

aGVHD and cGVHD were diagnosed clinically with histological confirmation as required and clinically appropriate. Grade of aGVHD was based on the International Bone Marrow Transplant Registry classification¹⁴ except grades A–D were labeled grades I–IV and was applied to GVHD with purely acute features even if it occurred after day 100. Grading was reviewed by a transplant clinician panel to reach consensus of maximum aGVHD grade. cGVHD was defined according to published National Institutes of Health consensus criteria¹⁵, and these criteria were used to assess GVHD severity for classical and overlap cGVHD. Malignant relapse was defined as recurrence or progression of disease over pre- transplant baseline, and TRM as death from any cause in continued remission. PFS was defined according to standard criteria. The algorithm of Copelan et al was used to assign the primary cause of death¹⁶.

Responses to aGVHD therapy were determined using published definitions³. Day 28 was chosen as the optimal time-point to assess GVHD response^17,18. Complete response was the complete resolution of aGVHD symptoms in all organs without secondary GVHD therapy. Partial response was the improvement in GVHD stage in all initially affected organs without complete resolution, and without worsening in other GVHD target organs or requiring secondary GVHD therapy. Progression was worsening of GVHD in at least 1 organ with or without amelioration in any organ. No response was patients with the same grade of GVHD not meeting criteria for partial response or progression.

Statistical Analysis

Data on patient characteristics and transplant-related outcomes were obtained from the prospectively maintained MSKCC transplant database verified by primary source documentation. Significant differences in categorical variables were determined by Chi square or Fisher’s exact test (two-tailed). Significant differences between means were determined by Students t test and between distributions by the Mann-Whitney U test. Incidences of neutrophil engraftment, GVHD and relapse were estimated using the cumulative incidence function. Death was the competing event for engraftment and relapse. Relapse and death were the competing events for GVHD. We included relapse as a competing event for GVHD calculations because treatment of relapse could alter GVHD incidence and severity. PFS was estimated using Kaplan- Meier methodology. Univariate and multivariate Cox regression analyses were used to ascertain associations between patient or graft characteristics with severe (grade III–IV) aGVHD. Analyses were performed using SPSS version 19.

Characteristics of Patients and Grafts

Patient demographics are summarized in Table 1. One-hundred and fifteen patients, predominantly adults (median age 37 years, 23 children and 92 adults) were transplanted. More than half had non-European ancestry. The majority had high-risk acute leukemia and received high dose or reduced intensity conditioning. TNC, CD34+ and CD3+ cell doses, and HLA unit-recipient and unit-unit match are shown in Table 2. Unit selection was based on match at HLA-A,-B antigens and -DRB1 alleles with up to 2 mismatches allowed. However, high resolution typing of HLA-A,-B along with -DRB1 alleles demonstrated 87 units (38%) were < 4/6 HLA-allele matched to the recipient. The median match grade of units at 6 HLA-alleles (high resolution match grade) was 4/6 (range 1–6/6), and for 10 alleles (including HLA-C and -DQ) was 6/10 (range 2–9/10). Since unit-unit match was not a selection criterion, there was a broad range of unit-unit match at all levels of resolution of HLA typing (Table 2).

Engraftment and Donor Chimerism

Six patients had graft failure (5 primary, 1 secondary), and one patient was not evaluable due to early death. The cumulative incidence of sustained donor-derived neutrophil engraftment was 94% (95%CI: 90–98) by day 45, and platelet engraftment was 84% (95%CI: 78–91) by day 180. The median times to neutrophil and platelet recovery were 24 days (range 12–43) and 50 days (range 29–162), respectively, after myeloablative conditioning, and 10 days (range 7–36) and 34 days (range 9–59), respectively, after non-myeloablative conditioning.

One unit dominated in all engrafting patients. Eighty-nine (82%) of the 108 patients with sustained donor engraftment had complete dominance of a single unit documented as early as 21–28 days after transplantation (median donor engraftment: 100%). The percentage of patients with hematopoiesis entirely derived from a single unit further increased to 92 (91%) of 101 evaluable patients at day 60, and 84 (94%) of 89 evaluable patients at day 100 (median 100% donor with the engrafting unit at both time points). The HLA-match of the engrafting unit with the recipient was a median of 4/6 HLA-A,-B,-DRB1 alleles (range 1–6/6), and 6/10 HLA-A,-B,-C,-DRB1,-DQ alleles (range 2–9/10).

Acute GVHD Incidence and Manifestations

Sixty-one patients had grade II–IV aGVHD within the first 6 months after DCBT (2 with late onset after day 100). The diagnosis was supported by biopsy in 55/61 (90%) of patients [skin (n = 29), GI tract (n = 44), and liver (n = 5)]. Twenty-six of these patients had grade III–IV disease (with 3 peaking at grade III–IV after day 100). The day 180 cumulative incidences of grade II–IV and III–IV aGVHD were 53% (95% CI: 44–62) and 23% (95% CI: 15–31), respectively (Figure 1). Among patients with grade II–IV aGVHD, the median onset of aGVHD was 40 days (range 14–169); it was earlier for those with grade III–IV disease (median 35 days) than those with grade II aGVHD (median 42 days, p = 0.012).

Of patients with grade II–IV aGVHD, the gastro-intestinal (GI) tract was the most commonly affected organ (n = 49, 80% of those with grade II–IV disease). Fourteen patients had upper GI tract involvement, 9 had lower, and 26 had both. Skin was affected less often (n = 39, 64% of those with grade II–IV disease). Only 17 patients had stage III skin disease, and 2 had stage IV. The liver was involved in 18% of grade II–IV cases. The combination of skin and GI tract disease was the most common manifestation (n = 25, 41% of those affected), followed by GI tract alone (25%), and skin alone (16%). Four patients had involvement of all 3 organs, 5 had gut and liver involvement, and 2 had only liver involvement.

Acute GVHD Treatment and Responses

Among patients with grade II–IV aGVHD, 29 (48%) were treated with systemic corticosteroids, 27 (44%) with budesonide alone, 4 (7%) with topical corticosteroids, and one recovered without additional immunosuppression. Most patients who received systemic corticosteroids had grade III–IV disease and the majority (79%) responded by day 28 of therapy (Table 3). Six patients (4 adults, 2 children) did not respond, and 2 that initially responded had subsequent progression of GI aGVHD.

Budesonide was used as the sole treatment exclusively in adults for overall grade II aGVHD predominantly affecting the upper and/or lower GI tract. Compared with those given systemic corticosteroids, these patients had a later GVHD onset (Table 3). Twenty-three (85%) of the budesonide-treated patients responded by day 28 of therapy. Four patients without an initial response also required salvage with systemic corticosteroids and 3 responded.

GVHD after Day 100: Late Acute and Chronic GVHD

As day 100 has traditionally distinguished acute from chronic GVHD, the manifestations of late GVHD occurring after day 100 were examined (Table 4). Of the 89 patients who engrafted and were in remission at day 100, 41 had no GVHD after day 100. Seventeen of these 41 patients (42%) had had prior aGVHD (12 grade II and 5 grade III) that had resolved. The remaining 48 patients had active GVHD on or beyond day 100. Twenty-nine of these patients had persistent (n = 14) or recurrent (n = 15) aGVHD (3 subsequently evolved into overlap syndrome), and 2 had late onset aGVHD on days 161 and 169. An additional 17 patients developed cGVHD after day 100. Ten had classical cGVHD (4 interrupted onset, 6 de novo), and 7 had overlap syndrome (6 with prior aGVHD and one de novo).

Thus, of the 48 patients with active GVHD after day 100, 38 (79%) had purely acute features or overlap syndrome. Overall, among patients who engrafted and were cancer-free as of day 100, the 2-year cumulative incidence of GVHD of any kind was 54% (95%CI: 44–65) (Figure 2A). However, if only chronic GVHD by NIH criteria was considered, the 2-year incidence was 23% (95%CI: 14–32) (Figure 2B).

Disease manifestations were examined in the 20 patients who developed NIH defined¹⁵ cGVHD (including the 3 patients who evolved from aGVHD to overlap syndrome). The 10 patients with classical cGVHD had a median onset of 233 days (range 141–505) with skin pigment changes (n = 5), dry eyes (n = 3), joint pain (n = 1), anorexia (n = 2), liver function test abnormalities (n = 4), and nephrotic syndrome (n = 1). The 10 patients with overlap syndrome (median onset 156 days, range 100–418) had aGVHD symptoms of erythematous skin rash (n = 5), nausea/ vomiting and/ or diarrhea (n = 10) combined with dry mouth or taste alteration (n = 3), mouth ulcers (n = 1), dry eyes (n = 3), joint stiffness (n = 1), vaginal dryness (n = 1), and/or liver function test abnormalities (n = 1). No patient had severe ocular or sclerotic skin involvement, contractures, or symptomatic pulmonary involvement. The majority had disease of mild (n = 10, 50%) or moderate (n = 5, 25%) severity, and the GI tract and skin were the organs most commonly affected. Five patients (25%) developed severe disease. All but 4 patients who developed cGVHD during the study period had done so by one year after transplant.

Cessation of Immunosuppression

Sixty-eight patients had sustained donor-derived engraftment and were in remission at one year after DCBT. Of those, 24 (35%) had stopped all immunosuppression (14 patients without GVHD and 10 with prior GVHD) whereas 44 (65%) patients remained on immunosuppression (18 for active GVHD and 26 on a taper) at one year after DCBT. Of the 18 with active GVHD at one year, 10 had aGVHD, 5 had overlap syndrome, and 3 had classical cGVHD.

Mortality and GVHD Deaths

A total of 44 patients have died to date; 14 from relapse, 29 from transplant-related complications, and one patient who was transplanted for treatment-related acute myeloid leukemia from relapsed osteosarcoma. With a median follow-up of 33 months (range 8–73), the 2-year overall survival and PFS are 66% (95%CI: 57–75) and 60% (95%CI: 51–70), respectively.

GVHD was the primary cause of death in 9 patients and was the second most common cause of TRM after organ toxicity. These deaths occurred at a median of 184 days (range 70–445) after DCBT. The 9 GVHD deaths were in adults and due to acute disease (1 grade II, 7 grade III–IV) or grade III aGVHD evolving to severe overlap syndrome (n = 1). Five of these patients had initially received systemic corticosteroids and 4 had failed initial treatment with budesonide. The most common secondary cause of death was infection (n = 7). Two had pneumonia (presumed bacterial), 3 had proven bacterial infections (1 fulminant C. Difficile colitis, 1 Klebsiella septicemia, and 1 Klebsiella pneumonia), and 2 had disseminated adenovirus. The remaining patients died of GI tract perforation (n = 1) and idiopathic diffuse alveolar hemorrhage (n = 1).

Risk Factors for Grade III–IV aGVHD

As the predominant GVHD syndrome was acute disease, and grade III–IV disease accounted for nearly all GVHD mortality, we evaluated risk factors for grade III–IV aGVHD through day 180 (to account for patients whose aGVHD peaked in severity after day 100). Univariate analysis of the relationship between patient characteristics and day 180 grade III–IV aGVHD revealed no association with patient age (0–15 vs > 16 years), diagnosis (acute leukemia/ MDS vs lymphoma/ CLL), conditioning intensity (myeloablative vs non-myeloablative), patient ancestry (European vs non-European), recipient CMV serostatus (sero-negative vs sero-positive recipient), or MMF dosing interval (every 12 vs every 8 hours).

In univariate analysis, there was no association with the infused TNC doses/kg or the CD3+ cell doses/kg of either the engrafting unit or the total dose (unit #1 + unit #2) in the graft. CD34+ cell doses also had no relationship with severe aGVHD risk. There was no association between the unit-unit HLA-match (at any degree of resolution) and severe aGVHD. Differences in severe aGVHD incidence according to HLA-match of the engrafting unit to the recipient (as shown in Figure 3) did not reach significance.

A multivariate Cox regression analysis was performed to take potential confounding variables into account (Table 5). In this analysis, better 6 allele HLA-match between the engrafting unit and the recipient was associated with a lower incidence of severe aGVHD (p = 0.031). Patient age, recipient CMV serostatus, engrafting unit infused TNC, and unit-unit HLA-match were not significant. When 10 allele HLA-match of the engrafting unit was substituted for a 6 allele match in the multivariate analysis, the severe aGVHD hazard ratio was 0.325 for engrafting units that were > 8/10 HLA-matched to the recipient but the comparison was no longer significant (p = 0.073).