Clinical Effect of High-Dose, Donor-Derived Natural Killer Cells Infused After HLA-Haploidentical Hematopoietic Cell Transplantation

Abstract 1900

Haploidentical HCT is feasible without ex vivo T cell depletion after conditioning with reduced-dose busulfan, fludarabine, and ATG (Lee K-H, Blood 2011). Furthermore, infusion of donor-derived NK cells (DNKI) after haploidentical HCT has been shown to suppress leukemia recurrence without GVHD in a murine model. Clinically, DNKI up to 2.5×10⁷/kg have been given to patients after haploidentical HCT without obvious increase in GVHD (Yoon SR, BMT 2010). Clinical effect of higher dose of DNKI, however, remains to be studied. Between February 2009 and February 2012, 37 patients with hematologic malignancies (AML, 28; ALL, 7; MDS, 1; DLBCL, 1) underwent haploidentical HCT and received NK cells derived from the same HCT donors twice at 2 (DNKI#1) and 3 weeks (DNKI#2) after HCT. The donors (median age 29 years, range 7–62; offspring 22, siblings 8, and parents 7) underwent 3 to 4 daily leukapheresis after receiving G-CSF. Donated cells during the initial 2 to 3 days were transplanted without further manipulation. Cells collected on the last day were CD3 depleted and differentiated into NK cells ex vivo. In the initial study phase, as per the protocol design, three-patient cohorts each received 0.2×10⁸/kg, 0.5×10⁸/kg, and 1.0×10⁸/kg donor NK cells twice. Among these initial 9 patients, only one patient who received 1.0×10⁸ cells/kg experienced grade-3 acute GVHD. Thereafter, 3 additional patients received 1.0×10⁸ cells/kg twice and none experienced °Ãgrade-3 acute side effect or acute GVHD. For patients enrolled subsequently, donor NK cell doses were based on the quantity of cells generated. The mean viability of final donor NK cell products for DNKIs #1 and #2 were 85% and 71%, respectively; mean CD56+CD122+, 87% and 94%, respectively; and mean CD3+CD56-, 0.4% and 0.3%, respectively, excluding one case with 16% and 7%. The final cell products exhibited additional NK cell features, such as granzyme/perforin gene expression, NK cell receptor expression, TNF-α/IFN-γ production, and cell cytotoxicity against K562 cells. All 37 patients in the study received at least one DNKI. Four patients did not receive the DNKI#2 (3 due to rapid clinical deterioration and 1 due to cell expansion failure). The median cell doses were 1×10⁸/kg (range, 0.2–2.0 ×10⁸/kg) and 1×10⁸/kg (range, 0.2–4.0 ×10⁸/kg) for DNKIs #1 and #2, respectively. In all patients, DNKI was tolerated well without acute toxicities such as fever or hypotension. After HCT, neutrophil engraftment (cumulative incidence, 87%), grade 2–4 acute GVHD (19%), chronic GVHD (27%), and TRM (29%) were observed. The median follow-up period of survivors was 20.0 months (range, 6.2–41.5). Of 33 patients with refractory leukemia (AML, 25; ALL, 7) or lymphoma (n=1), 7 of 25 with AML and 6 of 8 with ALL/lymphoma experienced progression/recurrence of their underlying malignancies (cumulative incidences, 29% vs. 75%; P=0.0325). When considered together with 31 historical patients with refractory acute leukemia (AML 22; ALL 9) treated with haploidentical HCT without high-dose DNKI, univariate analysis showed that the diagnosis (AML vs ALL/lymphoma, P=0.0048), disease status at HCT (primary refractory vs relapse then refractory, P=0.0410), and patient cohort (study vs historical, P=0.0420) were the significant variable predicting less diseases progression/recurrence. Upon multivariate analysis, the diagnosis (AML vs ALL/lymphoma, P=0.005; odd ratio, 0.356; 95% CI, 0.174–0.729) and patient cohort (study vs historical, P=0.033; odd ratio, 0.471; 95% CI, 0.236–0.941) were independent variables predicting less disease progression/recurrence. Kaplan-Meier event-free and overall survival rates for study AML patients, study ALL/lymphoma patients, historical AML patients, and historical ALL patients were 39% and 36%, 0% and 0%, 7% and 0%, and 0% and 0%, respectively. Our study showed that DNKI with median total dose 2×10⁸/kg given over 2–3 weeks after haploidentical HCT was tolerated well without obvious increase in GVHD or TRM. Furthermore, DNKI after haploidentical HCT may decrease leukemia progression especially in patients with refractory AML.