Abstract 83The European APL91 and the North American Intergroup study I0129 reported in the 1990s that ATRA combined with chemotherapy markedly improves the outcome of APL. Large randomized and single arm studies from around the world with both short and relatively long-term follow-up, confirm this impact of ATRA, while optimizing their mode of administration. However, the incidence and outcome of late relapses have not been well established. Furthermore, as strategies to minimize or eliminate chemotherapy by combining ATRA and arsenic trioxide are being adopted, the long-term durability of ATRA and chemotherapy regimens is even more relevant. We now report on late relapses among the APL patients enrolled in the North American Intergroup study I0129 conducted between 4/92 -2/95. Methods:380 patients (excluding 14 who had never received ATRA, and 3 ineligible) from 6 cooperative oncology groups, were randomized to induction with either ATRA or chemotherapy with daunorubicin and ara-C (DA). Patients then received two cycles of anthracycline-containing consolidation chemotherapy and then irrespective of their induction arm, patients were randomized to receive 1 year of maintenance with ATRA or observation (Obs) (Tallman et al NEJM 1997). Late relapse was defined as relapse 3 years or later from CR. Data analysis as of March 25, 2011 with a median follow-up of 11.8 (ranges 0.4–14.5) years. Results:Of 191 pts on DA and 189 pts on ATRA induction arms, 85 vs. 116 (total 201) were last known to be alive, median OS 3.6 years (95% CI: 2.0, 8.7) vs. not yet reached (p=0.0007), respectively. CR was achieved in 272 (72%) pts, and 113 are known to have relapsed. In addition, 70 relapsed pts. had received ATRA at any time, of them 60 pts relapsed early and 10 pts. relapsed late; 43 pts relapsed patients did not receive any ATRA and all relapsed early (P=0.01).Comparison of late (≥3 yrs. From CR1) with early (<3 yrs. From CR1) relapses, among the 272 CR1 pts.LateEarlyP valueNo of relapses (%)10 (4%)103 (38%)Age, median (range), yrs.41 (14–62)36 (2–75)0.75Male/female, No6/468/350.74At diagnosisWBC, Median (range), 1000/μL1.9 (1.1–29.9)4.3 (0–106)0.51WBC >10,000 μL3 (30%)30 (29%)1.0Yrs. in CR1 prior to relapse, median (range)3.9 (3.1–7.5)*0.9 (0.3–2.9)Induction, No (%)DA (n=129)4 (3.1%)63 (48%)0.31ATRA (n=143)6 (4.2%)40 (27%)Maintenance No (%)**Obs.4 (40%)69 (66%)0.08ATRA.5 (50%)32 (32%)Follow up of relapse known to be alive, yrs., median (range)11.2 (4–13)11.5 (0.7–14.2)0.76OS after CR1, yrs. median (95% CI)Not reached (3.3–∞)2.1 (1.7–3.4)*Only 2 pts. relapsed after 4 yrs.**3 pts not randomized to maintenance therapy.Follow up information is known so far on 5 late relapses. Relapse sites: BM -4 Pts, CNS – 1pt. Late relapse salvage: chemotherapy-5 pts., alloBMT-2 pts., Auto BMT -1pt., arsenic trioxide –none. Of the 10 late relapses, 3 died, all from disease progression. Median duration in CR1 prior to late relapse: survivors vs. died 3.9 and 4.1yrs, respectively. Only 35 (34%) of the early relapses are alive. Two second malignancies in the breast were identified, all in no known relapse pts. Conclusions:This is among the largest studies with very long follow-up to report that, in newly diagnosed APL patients receiving consolidation chemotherapy, the significantly higher OS rate after induction with ATRA compared to chemotherapy, is sustained. Most relapses in APL occur early, including in patients who had received ATRA at any time. Late relapses after 3 years: 1) are uncommon (4%) and are very rare after 4 years (1); at that time CR1 patients are most likely cured; 2) survive longer than early relapses, independent of CR1 duration, implicating the effectiveness of salvage therapy; 3) do not differ from early relapses in WBC and age at diagnosis; 4) were slightly more likely to be on ATRA maintenance. As the number of long-term APL survivors increases, future studies should address survivorship, late treatment-related complications, and determine whether some late relapses represent new therapy-related AML. Finally, if the high rate of early death (Park et al Blood, 2011) can be reduced, more patients will benefit from the high cure rate reported by this and other studies. Disclosures:No relevant conflicts of interest to declare.