Management of relapsed and refractory childhood acute promyelocytic leukaemia: recommendations from an international expert panel

Acute promyelocytic leukaemia (APL) comprises 5–10% of childhood acute myeloid leukaemia (AML) (de Botton et al, 2004). Its diagnostic hallmark is translocation t(15;17)(q22;q21.1), fusing the genes encoding promyelocytic leukemia protein (PML) and retinoic acid receptor alpha (RARA), generating the PML-RARA oncoprotein (Lo-Coco et al, 2008). Treatment of childhood APL with all-trans retinoic acid (ATRA) and anthracyclines yields complete remission (CR) rates of >90%, and 10-year event-free survival (EFS) of 76% (Testi et al, 2005). Relapse occurs in 17-27% of children with APL (de Botton et al, 2004, 2005; Testi et al, 2005; Lo-Coco et al, 2008; Creutzig et al, 2010; Bally et al, 2012) and its treatment remains a challenge due to lack of consensus over management and risk of cardiotoxicity with further anthracycline administration. Except for sporadic paediatric reports, most of the evidence for the treatment of relapsed APL comes from adult literature. For patients with relapsed APL, evidence shows that arsenic trioxide (ATO) salvage can lead to second CR (CR2) in 85% of patients (Soignet et al, 2001), although some patients can also achieve this with chemotherapy (CT) and/or ATRA (Ortega et al, 2005). Based on personal experience and informative literature on relapsed APL, recommendations were established by members with APL expertise from the North American Children's Oncology Group (COG) and the International Berlin-Frankfurt-Münster-Study Group (I-BFM SG). The quality of evidence for these recommendations, focused on relapsed APL with translocation t(15;17)(q22;q21.1) and the corresponding PML-RARA fusion gene, was mainly derived from expert opinion while a final agreement was by consensus. Refractory disease is defined as the persistent morphological detection of malignant promyelocytes at the end of induction, or as molecular/cytogenetic evidence of APL after consolidation (Sanz et al, 2009). Confident assessment of induction failure requires at least 60 days of ATRA plus either ATO and/or CT (including anthracyclines), and/or gemtuzumab ozogamicin (GO). In clinical trials utilizing ATO induction, predominantly of adults, the rate of refractory APL is virtually zero (Ravandi et al, 2009; Mathews et al, 2010; Zhou et al, 2010; Iland et al, 2012; Lo-Coco et al, 2013, 2016). As contemporary trials of paediatric APL utilize ATO-based induction, we expect primary refractory paediatric APL will be rare in the future. The bone marrow (BM) of patients receiving differentiating agents during induction may remain hypercellular with persistence of atypical promyelocytes for up to 50 days from the start of therapy. Care should be taken not to wrongly classify patients as refractory or to change therapy in this setting. In cases of true ATO resistance (Zhu et al, 2014), anthracyclines and high-dose cytarabine (HD Ara-C) should be considered. However, this approach can be myelosuppressive and rigorous supportive care is required to prevent infectious complications; thus, the use of GO (Estey et al, 2002; Lo-Coco et al, 2004; Breccia et al, 2007), a much less toxic option, should be considered. Regardless of the type of salvage therapy, allogeneic haematopoietic stem cell transplant (allo-HSCT) as consolidation offers a higher chance of long-term cure in patients successfully re-induced into haematological remission. However, in the absence of a suitable well-matched human leucocyte antigen (HLA) donor, and in patients with negative polymerase chain reaction (PCR) after a few salvage cycles, our consensus is that an autologous (auto) HSCT must be considered (Meloni et al, 1997; de Botton et al, 2005). Haematological relapse is defined as the reappearance of >5% abnormal promyelocytes in the BM. Molecular relapse is defined as the reappearance of the PML-RARA fusion transcript, after previously negative reverse transcription PCR (RT-PCR), detected in two successive BM samples repeated within 2 weeks of each other (Sanz et al, 2009). For patients with APL, monitoring minimal residual disease (MRD), by sequential measurement of PML-RARA transcripts using nested RT-PCR throughout treatment, can predict relapse (Grimwade et al, 2009). Burnett et al (1999) showed that children and adults treated on the UK Medical Research Council-ATRA trial with detectable PML-RARA transcripts after 3 cycles of treatment had 57% risk of subsequent relapse at 5 years versus 26% in those who became PCR negative at the same time point. Although MRD monitoring can predict relapse, most relapses occur in patients with negative PCR after consolidation (Breccia et al, 2004). This could be due to the limited sensitivity of the nested PCR technique compared to the quantitative RT-PCR (RQ-PCR). A study by the German AML cooperative group using RQ-PCR showed that patients who failed to achieve 3-log reduction of the PML-RARA transcripts within the first 3–4 months of therapy had an increased risk of early relapse (Schnittger et al, 2003). Minimal residual disease monitoring in BM is considered the 'gold standard' in APL. Although peripheral blood (PB) PCR has good concordance with BM PCR at early stages of therapy, Grimwade et al (2009) have shown an advantage for BM PCR after consolidation, with 1·5 log greater sensitivity than PB. It is recommended that sequential RT-PCR measurement of PML-RARA transcripts be performed every 3 months for 3 years from the end of consolidation (Sanz et al, 2009). Given that most relapses in APL occur within 3 years after consolidation (Gallagher et al, 2003), molecular monitoring can be stopped at this time point. However, contemporary ATO-based therapy for newly diagnosed APL has made relapse a rare event, especially in those with an initial white blood cell count <10 × 109/l (Iland et al, 2012; Lo-Coco et al, 2013, 2016). Thus, it is unclear whether this frequency and duration of BM monitoring are warranted in the ATO era. Two studies in the ATRA era suggested a survival benefit in patients with molecular relapse given pre-emptive therapy compared with those treated at the time of overt haematological relapse (Lo-Coco et al, 1999; Esteve et al, 2007). Indeed, patients presenting with haematological relapse are at higher risk of haemorrhagic death and APL differentiation syndrome. Several studies have identified prognostic factors in patients with relapsed APL who underwent salvage therapy with or without HSCT (Roman et al, 1997; Ramadan et al, 2012; Chakrabarty et al, 2014; Lou et al, 2014; Marjerrison et al, 2014; Lengfelder et al, 2015). However, most studies involved adults with very few subjects being <18 years of age (Chakrabarty et al, 2014; Marjerrison et al, 2014; Lengfelder et al, 2015). The most controversial issue is the definition of early and late relapse. A paediatric APL study reported that time from diagnosis to relapse of <18 months was significantly associated with worse EFS (P = 0·001) and overall survival (OS) (P = 0·002). However, the number of children in this study was too small (n = 24) to draw strong/definitive conclusions (Marjerrison et al, 2014). A recent European Leukaemia Net registry of relapsed APL addressed the outcome of 155 patients (14 children) treated with ATO following the first relapse. Univariate and multivariate analyses reported the favourable impact of first CR (CR1) duration ≥18 months (P = 0·03), achievement of molecular CR2 after consolidation (P = 0·01) and allo- or auto-HSCT (P = 0·01) on the OS and leukaemia-free survival for patients treated with ATO (Lengfelder et al, 2015). A study of the role of HSCT in 294 patients with relapsed APL (79 children) demonstrated that CR1 duration <12 months was associated with higher mortality (P = 0·02) (Chakrabarty et al, 2014). However, the significance of CR1 duration in the paediatric patients alone on this trial was not reported. Adult studies showed that prior ATO therapy is a poor prognostic factor (Lou et al, 2014). A Chinese study compared the outcome of patients with haematological relapse (n = 52, age range: 14–67 years) after previous ATO therapy (n = 20) with that of ATO-naïve patients (n = 32). The CR2 rate was 80% vs. 93·8%, respectively. Moreover, relapse rate was 68·8% vs. 33·3%, respectively (P = 0·03), and 4-year OS and relapse-free survival (RFS) rates were 62·4% and 29·8% in those who had prior ATO versus 71·2% and 66·2% in the ATO-naïve group, respectively (Lou et al, 2014). Among the 52 patients, only 6 CR2 patients received auto-HSCT (n = 2) or allo-HSCT (n = 4); thus, the role of HSCT cannot be determined. However, the high relapse rate in relapsed patients with prior ATO exposure suggests that subsequent HSCT might be the better option after ATO-induced CR2, although treatment response must be taken into account. Molecular CR at the time of transplantation has been shown to be an important prognostic factor. Ramadan et al (2012) reported a significant difference in the 4-year OS between patients transplanted in molecular CR and those with positive RT-PCR at the time of allo-HSCT (64% and 27%, respectively). Another study demonstrated that continued RT-PCR positivity after HSCT predicts subsequent relapse (Roman et al, 1997). Although it is essentially unknown whether patients who have achieved molecular CR after consolidation therapy require HSCT, we recommend HSCT in a subset of patients, as illustrated in Figs 1 and 2. Prognostic factors in relapsed APL arise from adult (Roman et al, 1997; Ramadan et al, 2012; Chakrabarty et al, 2014; Lou et al, 2014; Lengfelder et al, 2015) and more limited paediatric literature (Marjerrison et al, 2014) (Table 1): time to relapse <18 months from diagnosis, prior ATO therapy and failure to clear PML-RARA transcripts. These factors will be used to predict the risk of further relapse (Table 2) and consequently to guide which children can be treated with further differentiating agents and CT alone, and who would benefit from either auto- or allo-HSCT. Arsenic trioxide is the most active agent against refractory/relapsed APL. In an adult study, 40 patients with relapsed APL received ATO during re-induction, consolidation and maintenance: the CR rate was 85%, and OS and RFS were 66% and 56% at 18 months, respectively (Soignet et al, 1998). A Japanese study reported a 2-year OS of 56% after single agent ATO in induction and consolidation in 34 patients with relapsed APL (Shigeno et al, 2005). A Chinese report suggested that patients with relapsed APL who receive ATO followed by CT have better outcomes compared to patients who receive ATO alone (Niu et al, 1999). A synergistic effect between ATO and ATRA accelerates the differentiation and apoptosis of abnormal cells (Zhou et al, 2007), and improves the survival of APL patients (Breccia & Lo-Coco, 2012). Current adult guidelines support ATO ± ATRA as salvage therapy for relapsed APL (National Comprehensive Cancer Network 2014). However, front-line therapies now include ATRA/ATO (Iland et al, 2012; Lo-Coco et al, 2013, 2016), and it remains unclear how this approach will affect the outcome of relapsed patients. A recent Chinese study of 25 patients with relapsed APL who received front-line therapy with ATRA/ATO and CT, reported a 40% CR rate in patients salvaged with ATRA/ATO; again, however, molecular CR was low and relapse rates high (Lu et al, 2014). In childhood APL, ATO is increasingly used in front-line therapy as single agent (Mathews et al, 2010; Zhou et al, 2010) or with ATRA (Zhang et al, 2008). In relapsed APL, however, ATO-containing salvage therapies have been only sporadically described. A paediatric phase I trial reported an 85% CR2 rate in 13 children with relapsed/refractory APL using single-agent ATO; 6 received HSCT as consolidation, two of whom were in molecular CR at HSCT (Fox et al, 2008). Long-term molecular remissions using ATO, either alone or with ATRA, have been reported in children with relapsed APL (Ebinger et al, 2011; Au et al, 2012; Rock et al, 2014). Au et al (2012) reported four children with relapsed APL who were re-induced with oral ATO/ATRA; 3 were consolidated with oral ATO/ATRA, 1 with cytarabine/idarubicin. Maintenance therapy included oral ATO/ATRA for 2 years. All patients were in molecular CR at 122 months. Further, prolonged CR2 was reported in relapsed paediatric APL using ATO/ATRA re-induction followed by consolidation with CT (Zhang et al, 2008) or HSCT (Bally et al, 2012). Successful re-induction of multiply relapsed paediatric APL with ATRA and low-dose antimetabolites has been reported (Dvorak et al, 2007). Creutzig et al (2010) reported the outcome of 81 children with APL treated on the BFM 93/98/2004 trials with ATRA plus anthracycline/cytarabine. Nine patients relapsed and two died: one did not receive further therapy and 1 died 6 years later after second relapse despite allo-HSCT. The remaining 7 patients received anthracycline/cytarabine as re-induction and they all achieved CR; 4 received concurrent ATO, and 5 underwent HSCT; all were alive at 1·4–9·6 years from relapse (Creutzig et al, 2010). Gemtuzumab ozogamicin (Mylotarg®, Pfizer Philadelphia, PA, USA) is a humanized monoclonal antibody directed against CD33, linked to the cytotoxic agent calicheamicin. The latter is an anthracycline, a class of agents known to be very effective in APL. Virtually all cases of APL are CD33-positive (Guglielmi et al, 1998), and Jurcic et al (2000) reported that single-agent GO could induce a molecular remission in adults with APL. This was followed by similar reports in both de novo and relapsed APL (Estey et al, 2002; Lo-Coco et al, 2004; Breccia et al, 2007). Lo-Coco et al (2004) reported the benefit of GO as single agent at molecular relapse in a series of 16 adult patients; a single dose of GO at 6 mg/m2 resulted in 2-log reduction of PML/RARA transcripts in responding patients. The same group subsequently reported that the lower dose of GO 3 mg/m2 appeared equally effective in three adults with molecular relapse, all of whom achieved molecular CR after 2 or 3 doses of GO (Breccia et al, 2007). GO at ≥6 mg/m2/dose is associated with excess toxicity (Castaigne et al, 2012). Furthermore, there is no evidence that a dose of ≥6 mg/m2 is more effective than 3 mg/m2 (Castaigne et al, 2012). To achieve higher cumulative doses, fractionated schedules of GO have been used in both adults (Brethon et al, 2008) and children (Ravandi et al, 2009). The combination of GO with ATRA and ATO in adults (Aribi et al, 2007) and children (Ravandi et al, 2009) with APL shows encouraging efficacy without undue toxicity, in both newly diagnosed and recurrent APL. In children, GO has been used in non-APL AML with encouraging results, as recently reviewed by Parigger et al (2016). Further, GO induced a sustained molecular CR in combination with ATO in a child with multiply relapsed APL (Inoue et al, 2012). The major limitation of GO is the risk of promoting hepatic veno-occlusive disease (VOD)/sinusoidal obstructive syndrome (SOS). However, children seem less susceptible to VOD/SOS when doses ≤6 mg/m2 are used (McKoy et al, 2007). Indeed, children have safely undergone HSCT for AML 1-3 months after receiving GO in combination with CT (Gamis et al, 2014). Thus, HSCT can be performed without major concerns of VOD/SOS in children previously given GO ≤6 mg/m2. The optimal consolidation strategy after ATO-induced CR2 remains controversial. Options include repeated courses of ATO ± ATRA ± GO, CT or HSCT. The best therapeutic strategy, including the type of HSCT, depends on patient age/performance, CR1 duration, availability of GO, prior ATO exposure, donor availability and MRD status. Most of the literature is limited by lack of key data such as: pre-HSCT MRD, data on previous therapies and CR1 duration (Sanz et al, 2009). Molecular response after consolidation guides final therapy, but is not available at the time of relapse. Therefore, we recommend preparing for either auto- or allo-HSCT at the time of relapse. During the ATRA plus CT era, eight retrospective studies were published in adults with relapsed/refractory APL (Tables 3 and 4) (Mandelli et al, 1994; Meloni et al, 1997; de Botton et al, 2005; Sanz et al, 2007; Kohno et al, 2008; Fujita et al, 2013; Pemmaraju et al, 2013; Chakrabarty et al, 2014). These reported less favourable outcomes in allo-HSCT patients due to high treatment-related mortality (TRM) rates (30–39%), compared to the auto-HSCT group (TRM = 0–6%) (de Botton et al, 2005; Fujita et al, 2013; Chakrabarty et al, 2014). However, relapse incidence was lower in the allo-HSCT group, probably due to a graft-versus-leukaemia (GVL) effect. Two adult studies suggested that auto-HSCT is effective when performed with negative PML-RARA RT-PCR, while patients with persistence of the fusion transcript were at high risk of relapse (Meloni et al, 1997; de Botton et al, 2005). Similarly, Yanada et al (2013) reported the results of a phase-2 study aimed at evaluating the efficacy and feasibility of a sequential treatment consisting of induction and consolidation with ATO, peripheral blood stem cell (PBSC) collection after HD Ara-C, and auto-HSCT with PML-RARA negative stem cells in PML-RARA negative patients. With a median of the EFS and OS rates were and respectively in the patients 25 who to who underwent auto-HSCT with PML-RARA negative no death was et al, In the allo-HSCT similar data were reported in a study, OS was vs. for patients with negative versus positive pre-HSCT MRD, respectively (P = et al, 2012). In Chakrabarty et al no of positive pre-HSCT PML-RARA in 6 of autologous and in of allogeneic HSCT on the incidence of relapse, survival or the outcome of APL patients age years) in CR2 who received re-induction with ATO-based therapy alone (n = was compared with those who subsequently underwent auto-HSCT (n = with or without 5 OS was and for the ATO only and auto-HSCT respectively et al, 2016). 2-year 2-year 2-year 4-year 4-year 3 patients 34 5 7 3 76% 18 5 6 6 1 4-year 4-year 16 18 6 7 = = 7 4-year in CR2, in 4-year 13 5 Most of the published data on HSCT as a treatment for relapsed childhood APL comes from small retrospective studies (Tables 3 and 4) (de Botton et al, 2004, 2005; et al, 2004; et al, 2008; et al, 2008). from the paediatric plus trial included relapsed patients who received either auto- (n = or allo-HSCT (n = in of 7 auto-HSCT in CR2, while 2 haematological relapse. In the allo-HSCT group, patients in CR2, while 1 died of (Testi et al, 2005). Among children in the European trial 7 relapsed children achieved 1 with relapse received and with 3 patients received auto-HSCT and 3 underwent one of the patients died in CR2 from disease while 6 patients in molecular CR2 after to months (de Botton et al, 2004). et al reported the outcome of 5 relapsed APL children, who achieved molecular CR2 after re-induction therapy with ATO (n = BM or therapy = and underwent All patients in molecular CR after months et al, 2008). et al (2004) reported performed for relapsed/refractory APL. OS was all were from patient relapsed months after HSCT, a cumulative incidence of relapse et al, 2004). A study by et al reported children with relapsed (n = or refractory (n = APL, either auto- (n = or allo-HSCT (n = EFS and OS were and after and and respectively after with no significant difference between the type of HSCT for both EFS (P = and OS (P = The incidence of after auto- and allo-HSCT was and relapse in 3 patients and 2 allo-HSCT (Dvorak et al, 2008). in the management of such as donor and supportive care et al, 2010; et al, 2016), may as to the reported survival of auto-HSCT et al, 2009; et al, 2010). expert that allo-HSCT should be considered only in patients with prior ATO exposure who relapse primary refractory APL, patients in second or greater relapse or not molecular CR after 4 salvage cycles, especially an donor is available and the patient is in good clinical for ATO-naïve patients who are in molecular CR2 after 4 salvage cycles (Soignet et al, In such patients, consolidation with ATO/ATRA plus GO may be considered. cycles should of one of including ATO/ATRA morphological CR in ATO-naïve patients, with the of 3 doses of GO or GO is not in patients with prior ATO and 3 cycles of consolidation, including ATO for 5 weeks and ATRA for 2 weeks for 3 cycles in ATO-naïve patients, while patients with prior ATO exposure should have one of high-dose followed by a of ATO/ATRA and a of high-dose GO can be to 1 of consolidation in all patients. with should be on 1 of each of consolidation in all patients. A of treatment for relapsed paediatric APL is given in Figs 1 and 2. These recommendations to patients with both haematological and molecular relapse of disease that occur early months from or late months from Treatment of very late relapses is sporadic reports of patients with very late relapse of APL have been published et al, et al, 2004; et al, 2007; Breccia et al, 2011; & 2014). Most of these relapses are to be due to the of the as to a second Relapse at years from diagnosis in APL is and 7 years. Therefore, very late relapse is defined as relapse at months from With a median of months, an adult European APL group et al, 2010) reported a incidence of very late relapse. All 18 patients were previously treated with and were re-induced with achieved molecular CR and had 4-year survival et al, 2010). Similarly, the North American reported long-term results of adults and children with APL, who received front-line therapy with et al, patients achieved molecular CR and children) of these had a very late relapse. were salvaged with CT with ATRA in 4) and one with ATRA patients received consolidation with HSCT. a median of patients were in CR2 et al, Two adult and paediatric studies (Lo-Coco et al, 2010) and European et al, the favourable outcome of patients with very late relapse after prolonged In other APL studies including adults and children treated with few very late relapses and the outcome for these patients was significantly better compared to those with early relapse (Lou et al, 2014). Although no of care for the treatment of very late APL relapse resistance is in these patients. cumulative anthracycline doses, in front-line exposure to anthracyclines in salvage therapy. Although very limited data are available on the use of ATO for very late relapsed APL in children (Ebinger et al, the use of this agent should be considered. The literature is mainly to patients who relapsed after treatment with with only a few reports of patients after ATO-based However, the results of these reports of front-line therapy with ATO, salvage with ATO-based remains effective (Lou et al, 2014; Lengfelder et al, 2015). reports that remission duration is when ATO is with either ATRA or CT as to when it is given as single agent (Lou et al, 2014; Lengfelder et al, 2015). therapy with ATO/ATRA plus GO (Aribi et al, 2007), or ATO plus GO (Inoue et al, 2012) or prolonged ATO/ATRA therapy can be effective for very late relapses (Breccia et al, The benefit of HSCT is for patients after very prolonged CR1 in whom ATO/ATRA or 6 might be (Breccia et al, The OS of the 155 relapsed patients in the European Leukaemia Net was a better outcome was in patients with late compared to early relapse, of treatment (Lengfelder et al, 2015). Thus, more consolidation cycles of ATO without HSCT might be considered in very late relapse to Furthermore, in very late relapse, GO may be effective (Breccia et al, 2007). relapsed APL can but most the or et al, et al, 2007) and most relapses are by evidence of BM of clinical relapses have a et al, but the incidence of in APL patients at relapse is paediatric literature reported a very low incidence of relapse & 2009). consensus on the management of relapse in the or other in patients with APL ATO in and the up to of et al, 2008). Therefore, a therapeutic response at these is In patients with relapse, treatment with ATO ± ATRA demonstrated high efficacy and low toxicity et al, These that ATO is as single agent or in to ATRA for the treatment of relapse, that BM and molecular remission. data from patients with APL that with and is effective for disease (de Botton et al, However, the optimal frequency and number of are although the most recent APL trial after et al, 2015). with high such as HD Ara-C, have been used in childhood relapse followed by auto-HSCT and et al, 2008). The role of HSCT in relapse remains although it was recommended by the European in (Sanz et al, 2009). The for is also although some this approach all APL patients with relapse should to with APL relapse should have BM and for relapse, as they may require more treatment than those with relapse & 2010). are limited reports of in paediatric APL although ATO and have been the results are et al, et al, thus, the optimal therapeutic approach is Arsenic children with late relapse months from and those with very late relapse of ATO exposure, can be re-induced with plus GO followed by ATO consolidation without HSCT, they clear PML-RARA transcripts after 4 salvage ATO-naïve children with early relapse and children with prior ATO exposure and early or late relapse who clear transcripts after 4 cycles can be consolidated with with ATO-based therapy is also patients who are and with early relapse who clear PML-RARA transcripts after 4 cycles, or when auto-HSCT is not due to patient or limited transplant with primary refractory APL, with previous ATO exposure and early relapse, those with second relapse and patients with persistence of PML-RARA after 4 cycles of prior ATO exposure or time to should be considered for consolidation with allo-HSCT an donor is of relapse must be These recommendations are an to in therapy for children with relapsed APL. that relapsed APL patients are a and these may require on patient as as the that are available to the a can ATRA resistance and in relapsed/refractory APL patients et al, 2015). Furthermore, oral ATO is effective in relapsed children therapy is et al, 2012). sequential MRD monitoring with RQ-PCR might rates of clinical relapse in patients with APL (Grimwade et al, 2009). to the low risk of relapse after front-line ATO-based studies for relapsed paediatric APL are not between the during the ATO era will treatment recommendations to be The would to all within the and with whom treatment of paediatric APL was over the years. and the study and the and 1 and 2 and Figs 1 and 2. 3 and the relapse and the recommendations, the and All final The have no of to

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.