A Canadian Blood and Marrow Transplant Group (CBMTG) Randomised Trial Comparing G-CSF Mobilized Peripheral Blood Versus G-CSF Stimulated Bone Marrow In Recipients Of Sibling Allografts For Hematologic Malignancies

We conducted a Phase III randomised multicenter trial of matched sibling G-PB vs G-BM allografts in adults with hematologic malignancies. The primary endpoint was time to treatment failure defined as the first occurrence of extensive cGvHD, relapse or death from any cause. Thirteen centers in Canada, Saudi Arabia, U.S., Australia and New Zealand enrolled patients. Adaptive stratification (minimization) balanced treatment groups by center, disease, disease risk and conditioning. Donors received G-CSF 5 mg/kg/d sc for 4 or 5 days and underwent apheresis on Day 5 and, if necessary, on Day 6 or marrow harvest on Day 5. Eligible recipients between 16 and 65 years received myeloablative conditioning and cyclosporine/methotrexate as GvHD prophylaxis. The study began before widespread adoption of the NIH GvHD consensus guidelines so cGvHD was characterized as limited or extensive. Immunophenotyping and functional assays performed on donor cells were correlated with incidence of cGvHD.

Between 2007 and 2012, 230 sibling donor-recipients were randomized to receive either a G-PB or G-BM allograft; 3 randomized to G-PB and 4 to G-BM were inevaluable (4 relapsed before transplant, 1 donor withdrew before collection, 1 recipient withdrew consent and 1 was ineligible (non-myeloablative conditioning)). Indication for allograft was AML (109; 49%), ALL (57; 25%); MDS/MPD(31; 14%), CML(13; 6%), lymphoproliferative disorder(NHL/MCL/CLL)(12; 5%) and biphenotypic leukemia(1). 77(70%)(G-PB) and 71(63%)(G-BM) recipients had low-risk disease (AML or ALL in CR1, RA, RARS, CML in 1^st CP). Donor and recipient ages, sex and CMV immune status were balanced between groups.

Donors tolerated mobilization well although one donor (G-PB) died of a cardiac event while receiving G-CSF. Median nucleated cells per kg were 7.59x10⁸ (2.73-18.64) and 5.92x10⁸ (0.94-13.23) in the G-PB and G-BM arms respectively (P<0.0001). Median CD34 cells per kg were 5.50x10⁶ (2.03-23.94) and 3.22x10⁶ (0.29-8.71) respectively (P<0.0001). 94% of G-PB donors underwent 1 apheresis. Median volume of harvested G-BM was 1127mL (350-1731mL).

With a median follow-up of 36 months (range 9.6-48), using Cox proportional hazards multivariable modelling adjusted for the 4 minimization factors, time to treatment failure did not differ significantly between arms (HR 0.91; 95% CI 0.68-1.22; P= 0.52). There was a significantly delayed time to onset of overall cGvHD in the G-BM arm (HR=0.66; 95% CI 0.46-0.95; P=0.03) but time to onset of extensive cGvHD did not differ significantly between arms (HR=0.76; 95% CI 0.52-1.10; P=0.15). Other results are shown below:

Evaluation of donor immune populations showed that IFNg⁺ CD4 T cells and CD56^bright NK cells were protective against cGvHD in G-PB and G-BM recipients. In contrast, specific B cell populations consistent with memory or exhausted B cells were associated with cGvHD in G-PB recipients (CD19⁺CD27+, p=0.05, CD19⁺ CD21^lo, p=0.009) but not in G-BM.

In this large prospective randomized trial, the primary endpoint of time to treatment failure did not differ between G-PB and G-BM. OS and RFS were also not significantly different between arms. Neutrophil and platelet recovery were slower by a median of 3 days with G-BM. However, time to overall cGvHD was significantly delayed in G-BM recipients with non-significant trends of less acute and less extensive cGvHD with G-BM. cGvHD correlated with different cell populations in G-PB and G-BM suggesting that donor source can influence the mechanism of cGvHD. Further analyses are underway to determine the clinical impact of these differences.

No relevant conflicts of interest to declare.