Background: Haploidentical family donors represent the ideal solution to offer a potential cure for every high-risk leukemia patient undergoing HSCT. Widespread use of haploidentical HSCT has been historically limited by high rates of late non-relapse mortality (NRM) and relapse, which are associated with the delayed immune reconstitution (IR) due to either ex vivo T-cell depletion or in vivo post-HSCT cyclophosphamide given as graft-vs-host disease (GvHD) prevention. We have previously shown in a phase 2 trial (TK007, Lancet Oncol 2009;10:489) that TK cells (donor T cells genetically modified to express the HSV-TK suicide gene) can safely induce an early IR when given after T-cell depleted haploidentical HSCTMethods: In a confirmatory, ongoing phase 3 trial (TK008, NCT00914628), up to 4 monthly infusions of TK cells are given at 1x107for kg of patient body weight, starting 21 to 49 days after T-cell depleted haploidentical HSCT in the experimental arm. Control arm consists of either T-cell depleted or post-HSCT cyclophosphamide haploidentical HSCT, at physician discretion. High-risk acute leukemia patients lacking an HLA-matched donor are included. So far, 34 patients have been enrolled from eight EU and US sites, with 19 being in complete remission, 18 presenting with AML and 22 having PS of 0 at HSCT. Hypothesis testing for primary efficacy endpoint: 1-year disease-free survival (DFS) of 30% (control arm) vs 52% (TK arm). Secondary endpoints include overall survival (OS), NRM and relapse incidence, proportion and timing to IR, incidence and control of GvHD by suicide-gene induction with ganciclovirResults: Data refer only to 24 patients randomly assigned to experimental arm. Median follow-up was 1.2 years. TK cells were timely given at a median of 28 days after HSCT (95% CI, 25 to 35). IR (defined as a CD3+ cell count > 100/µL) was achieved by 15 of 19 patients (79%) who received TK cells, after a median number of 2 doses (range, 1 to 4) and a median cumulative TK-cell dose of 2.4x107 (range, 1.0 to 3.9). The median time to reach IR computed from HSCT and last TK-cell infusion was 105 days (95% CI, 68 to 125) and 27 days (95% CI, 23 to 33), respectively. No patient received post-HSCT immune-suppressive therapy as GvHD prophylaxis. Acute GvHD developed in 7 patients (6 grade II and 1 grade III) for a 1-year cumulative incidence of 33% (±10) and was rapidly abrogated by suicide-gene induction with ganciclovir administered for a median of 14 days. None of the patients who experienced GvHD subsequently received prophylaxis with long-term immune-suppressive therapy (IST). No progression from acute to chronic GvHD and no GvHD-related death occurred. TK-cell-induced IR was characterized by a wide T-cell repertoire and high frequency of T cells specific for opportunistic pathogens and was associated with high survival rates. By ITT analysis at 1 year, OS was 85% (±8), DFS and IST-free survival were both 74% (±10) and NRM was 10% (±7). For patients achieving IR, the corresponding values of OS, DFS and NRM were 100%, 86% and 0%, respectively. Relapse incidence at 1 year was 16% (±8) for all patients and resulted related to the cumulative TK-cell dose, being 0% for patients who had received higher doses (≥3x107) compared to 20% for those treated with lower doses. By pooling data from TK007 and TK008 trial, these dose-related outcomes were confirmed in a landmark analysis set at 100 days after HSCT to mitigate the guarantee-time bias, with 5-years relapse incidence being inversely related to the TK-cell dose received (0%, 19% and 64% for doses ≥ 3x107, 1.1-2.9x107 and ≤ 1.0x107, respectively; p=0.008)Conclusions: Preliminary results of this ongoing phase 3 trial confirm the potential clinical benefit of T-cell gene transfer technology integrated with T-cell depleted haploidentical HSCT, and highlight the role of early IR as surrogate endpoint for survival outcomes and the dose-related antileukemic effects of TK cells DisclosuresNiederwieser:Novartis, Gentium, Amgen: Honoraria, Membership on an entity’s Board of Directors or advisory committees, Speakers Bureau. Colombi:MolMed: Employment. Antonio:MolMed: Employment. Bordignon:MolMed: Employment.