CD19 CAR-T cells show unprecedented responses in relapsed/refractory Acute Lymphoblastic Leukaemia, but long-term persistence appears critical for their use as a stand-alone therapy. The origin of long-term persisting CAR-T cells has yet to be defined and will be critical in designing manufacturing protocols to optimise long-term persistence. Previous data are conflicting with groups showing prolonged persistence of CAR-T cells from cell products with a predominantly effector memory (TEM) phenotype, whereas others suggesting that the dominant clones originate instead from infused stem cell memory (TSCM) and central memory (TCM) T cells. To date, it has not been possible to isolate long-term (> 1 year) persisting CAR-T cells in patients. In the CARPALL Phase I study, the use of an improved low-affinity CD19 CAR resulted in enhanced expansion and persistence of CAR-T cells in vivo (Ghorashian et al, Nature Medicine, in press). Combining this unique experimental setting with our well-established clonal tracking platform based on high-resolution integration sites (IS) analysis has enabled us to track the fate of the infused CAR-T cells. We analysed 2 patients with long-term persistent CAR-T cells detectable by flow cytometry in peripheral blood. CAR-T cells comprised 13 and 53% circulating CD3+ cells respectively at day 14 post-infusion, 7.1 and 7.7% circulating at 1 month, 0.7 and 1.3% at 6 months and 0.1% for both at the latest follow-ups (24-28 months). Blood samples taken at early (14d, 30d) and later (6m to 28m) time points were flow-sorted for CAR+ TCM/TEM mixed population and TSCM T cells, while the corresponding infused gene-modified products were separated into three subpopulations: TSCM, TCM and TEM. The integration profile of each sorted cell populations was established using linear amplification mediated-PCR (LAM-PCR) combined with high throughput sequencing. We identified a total of 7,105 and 4,692 IS from 2 patients overtime before infusion and up to 28 months after infusion. The infused CAR-T cell population was highly polyclonal before infusion. Although the total number of CAR-T cell clones decreased substantially upon in vivo selection, we did not observe any sign of aberrant clonal drifts and diversity was preserved long-term. Early after infusion during the response peak, TSCM underwent two waves of transient oligoclonal expansion. In both patients two distinct sets of individual TSCM clones contributed to the 73% and 97% of the whole analysed TSCM population at day 14 and 74% and 99% at day 30. Conversely, the largest memory/effector clones detected at the same timepoints spanned from 4% to 21% of the total TCM/TEM population. These TSCM clones subsequently contracted and were not observed at 6-28 months after infusion suggesting that different clones are responsible for the early response and prolonged immune surveillance. After 6 months post-infusion, when the IS profile of circulating CAR-T cells was compared with selected populations from the infused product, only 1.8%-6.1% of long term clones were derived from the infused TCM population, despite this accounting for the majority of IS in the products (72.7/75.8% of clones). Conversely, in both patients the majority of IS associated with long term persistence (90.7%/55.5%) were derived from the TSCM compartment. Our preliminary results raise two hypotheses on the clonal dynamics of infused CAR T cells: 1) There is an early expansion of a defined group of clones during the first 30 days, which is more pronounced in the precursor TSCM compartment. These early waves do not seem to be originated from clones that have substantially expanded in vitro such that their clonal mark could not be retained in the batch of the infused cell product analysed. Further, these clones rapidly disappear after the early anti-tumour response. 2) The long-term population of CAR-T cells seem to have a higher relation with TSCM clones that have expanded in vitro before infusion, supporting the notion that such cells in the infused batch would be the one primarily responsible for the preservation of circulating CAR-T cells in the treated patients. This study suggests for the first time that anti-leukemic response occurs along rapid waves of clonal succession and that TSCM are primarily responsible for the long-term survival of CAR-T cells. Disclosures Ghorashian: novartis: Honoraria; UCLB: Patents & Royalties: UCLB; Celgene: Honoraria. Pule:Autolus: Membership on an entity's Board of Directors or advisory committees. Thrasher:4BIOCapital: Membership on an entity's Board of Directors or advisory committees; Orchard Therapeutics: Consultancy, Equity Ownership, Membership on an entity's Board of Directors or advisory committees; Generation Bio: Consultancy, Equity Ownership, Membership on an entity's Board of Directors or advisory committees; Rocket Pharmaceuticals: Consultancy, Membership on an entity's Board of Directors or advisory committees. Amrolia:UCLB: Patents & Royalties.