PLZF-RARα, NPM1-RARα, and Other Acute Promyelocytic Leukemia Variants: The PETHEMA Registry Experience and Systematic Literature Review

It has been suggested that 1–2% of acute promyelocytic leukemia (APL) patients present variant rearrangements of retinoic acid receptor alpha (RARα) fusion gene, with the promyelocytic leukaemia zinc finger (PLZF)/RARα being the most frequent. Resistance to all-trans-retinoic acid (ATRA) and arsenic trioxide (ATO) has been suggested in PLZF/RARα and other variant APLs. Herein, we analyze the incidence, characteristics, and outcomes of variant APLs reported to the multinational PETHEMA (Programa para el Tratamiento de Hemopatias Malignas) registry, and we perform a systematic review in order to shed light on strategies to improve management of these extremely rare diseases. Of 2895 patients with genetically confirmed APL in the PETHEMA registry, 11 had variant APL (0.4%) (9 PLZF-RARα and 2 NPM1-RARα), 9 were men, with median age of 44.6 years (3 months to 76 years), median leucocytes (WBC) 16.8 × 109/L, and frequent coagulopathy. Eight patients were treated with ATRA plus chemotherapy-based regimens, and 3 with chemotherapy-based. As compared to previous reports, complete remission and survival was slightly better in our cohort, with 73% complete remission (CR) and 73% survival despite a high relapse rate (43%). After analyzing our series and performing a comprehensive and critical review of the literature, strong recommendations on appropriate management of variant APL are not possible due to the low number and heterogeneity of patients reported so far.


Introduction
Acute promyelocytic leukemia (APL) is a relatively rare hematologic malignancy accounting for~10% of acute myeloid leukemia (AML) cases [1]. APL is characterized by M3 morphological subclassification, t(15;17)(q22;q21) chromosomal translocation, and promyelocytic leukemia protein (PML)-retinoic acid receptor alpha (RAR α ) gene fusion, showing high rates of complete remission (CR) and cure using front-line schedules with all-trans-retinoic acid (ATRA) and/or arsenic trioxide (ATO). Apart from the classical PML-RAR α cases, some APL patients are diagnosed with rare APL variants, characterized by a different rearrangement involving RAR α plus another partner gene. According to some reviews, these APL genetic variants account for~1-2% of APL cases, but the real frequency remains unknown [2][3][4][5]. To our knowledge, 14 types of variant APL where RAR α is fused to different genes have been reported . Moreover, the clinical features and optimal treatment of variant APLs remain to be established. In fact, some authors have suggested resistance to ATO and ATRA for the PLZF-RAR α APL, which is probably the most frequent form of rare APL forms (up to 0.8% of all APLs) [3,27].
In this study, we aim to analyze the incidence, characteristics, treatment patterns, and outcomes of variant APLs reported to the PETHEMA epidemiology registry, including patients from Poland, Argentina, Portugal, and Spain from PALG (Polish Adult Leukemia Group), PETHEMA (Programa para el Tratamiento de Hemopatias Malignas), and GATLA (el Grupo Argentino de Tratamiento de la Leucemia Aguda) groups. Given the clinical heterogeneity and scarce number of variant APL patients reported in the literature, we also aim to perform a systematic review in order to shed light on strategies to improve the management of these extremely rare diseases.

Patients and Eligibility
Between November 1996 and January 2020, adult patients from institutions from Spain, Poland, Portugal, Argentina, Colombia, and Uruguay were registered in the PETHEMA database. In all patients, the diagnosis of APL was suspected by cytomorphology and confirmed by conventional cytogenetics and/or reverse transcriptase-polymerase chain reaction (RT-PCR). Patients were reported irrespectively of the treatment administered, including also those dying early before starting ATRA or chemotherapy. Eligibility criteria and trial design for PML-RAR α patients have been reported elsewhere [64][65][66][67]. Patients diagnosed with rare APL variants were not eligible as per PETHEMA trial protocols, but baseline and treatment information were also collected using the same forms than for PML-RAR α APLs. Diagnosis of APL variants was performed after clinical suspicion (M3 morphology and lack of genetic diagnosis of t (15;17) or PML-RAR α ). Specific PCR tests for rare rearrangements involving RAR α were performed locally according to routine clinical practice. Two central laboratories were available in Spain as per physician's demands in case of need. Informed consent was obtained from patients. According to the Declaration of Helsinki, the protocol was approved by the Research Ethics Board of each participating hospital.

Treatment
No specific guidelines or recommendations for front-line or salvage therapy for rare APL variants were available during the study period. If they considered appropriate, physicians could follow front-line protocols for t (15;17) APL, where induction therapy consisted of oral ATRA and intravenous idarubicin (AIDA regimen). All patients in complete remission (CR) received 3 anthracycline-based consolidation courses, which were risk-adapted since protocol LPA99. Consolidation was followed by 2 years of maintenance, as it was previously described [64][65][66][67]. Since 2017, the PETHEMA guidelines recommended ATO+ATRA combination for low-and intermediate-risk t(15;17) APL.

Response Assessment
Remission induction response and relapse were assessed according to the revised criteria by Cheson et al. [68]. There were no guidelines or definitions for molecular remission, molecular persistence, or molecular relapse. Relapse was defined as presence of equal or more than 5% bone marrow blasts or presence of extramedullary disease. Lumbar puncture or other diagnostic tests to assess disease status were performed as per physician's judgement.

Data Collection
Data were collected and registered prospectively; last patient follow-up was updated on 15 February 2020. Following data were collected at diagnosis: age, gender, ECOG score, thrombosis and bleeding, hemoglobin level, platelet count, leukocytes (WBC) count, creatinine, uric acid, urea, lactate dehydrogenase (LDH), alkaline phosphatases, total bilirubin, and albumin, triglycerides, cholesterol, bilirubin, and type of RAR α rearrangement. Coagulopathy was defined as thrombocytopenia plus either: prolonged prothrombin time and/or activated partial thromboplastin; or hypofibrinogenemia and/or increased levels of fibrin degradation products or D-dimers.

Systematic Review: Search Strategy and Selection of Studies
In accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, two independent reviewers (M.S. and P.M.) conducted this systematic review [69]. The following databases were searched without restrictions: PubMed and Excerpta Medica database (EMBASE). In addition, the reference lists of important studies and reviews were hand searched. The last literature search was performed on 23 February, 2020. Similar keywords were used in different databases: acute promyelocytic leukemia and "rare" or "variant" [Mesh], as well as the specific fusion genes. Case reports or studies analyzing a series of patients with rare APL variants were selected. Our systematic search obtained 60 citations from databases and journals (58 original research/case report articles and 2 reviews). The agreement in the study selection between the reviewers was excellent (kappa = 0.97).

Patients and Eligibility
Overall, 2961 patients with newly diagnosed APL from Spain, Portugal, Argentina (GATLA group), Uruguay, Colombia, and Poland (PALG group) were reported to the PETHEMA registry. Of them, 66 were coded as lack of genetic diagnosis, resulting in 2895 patients with genetically confirmed APL. Among those, 11 patients with variant APL were identified (0.4%): 9 PLZF-RAR α (0.3%) and 2 NPM1-RAR α (0.1%). The main characteristics of variant APL patients are presented in Table 1. Briefly, the majority were men (9/11), with median age at diagnosis of 44.6 years (3 months to 76 years) in PLZF-RAR α and 7 and 74 years for NPM1-RAR α patients (two pediatric patients overall).
Median WBC at diagnosis was 16.8 × 10 9 /L, with 50% patients classified as high-risk according to Sanz's relapse-risk score [10,11,13]. Three out of 10 patients (30%) had hyperleukocytic forms with WBC > 50 × 10 9 /L. Two out of 11 patients had therapy-related APL secondary to colon neoplasm's chemotherapy or to methotrexate plus antipaludic drugs for rheumatoid arthritis. The majority of variant APL had coagulopathy at diagnosis (6 out of 7 with available data), and 2 out of 6 patients with CD56 blast expression available were positive.
Post-remission schedule consisted of ATRA and anthracycline-based consolidation cycles in 4 patients (3 of them followed by ATRA plus low-dose chemotherapy maintenance) and intermediate or high dose Ara-C-based (HD-Ara-C) consolidation in 5 patients. Two patients (40 and 50 years old) received an allogeneic hematopoietic stem cell transplant (alloHSCT) in first CR. Only one patient (0.3 years old) received central nervous system (CNS) prophylaxis as part of the front-line regimen.
With a median follow-up of 60 months (8-101 months), 8 patients (73%) were alive at the time of analysis (3 deaths: 2 induction deaths and 1 death due to disease progression after third relapse). Overall, 4 out of 9 patients (43%) achieving first CR subsequently relapsed at 9, 24, 40, and 58 months (3 of them had CNS relapse). Patients relapsing had 53, 100, and 248 × 10 9 /L WBC counts at diagnosis (1 patient no available data). A second CR was achieved in all cases (salvage therapy was ATRA + ATO in 3 patients; and HD-Ara-C in 1; including triple intrathecal chemotherapy for all patients with CNS relapse). An alloHSCT was performed in second CR in 1 patient who relapsed subsequently, and 1 patient was planned to undergo Haploidentical alloHSCT.

Discussion
This study shows that variant APL is probably less frequent (0.4%) than previously reported (1-2%). Our systematic literature search found 60 manuscripts on this topic (58 original reports and 2 reviews), almost equaling the number of patients with variant APL reported to date (n = 79). The characteristics of new 11 variant APL cases registered by the PETHEMA group are in line with prior reports (male predominance, PLZF-RAR α as more frequent rearrangement). The vast majority of PETHEMA patients were treated with ATRA-containing schedules. While survival and CR rates seem better in our series as compared to background literature, prevention of relapse (especially in the CNS) remains as a challenging issue as well. Although prognosis of variant APLs seems better than non-promyelocytic AML, it is worse than PML-RAR α APLs, where ATO + ATRA or AIDA-based regimens results in virtual absence of remission induction resistance, low rate of relapses (<15%), and higher rates of cure (>80%) [1].
As far as we know, this is the larger series analyzing the frequency of variant APL (0.4%); 11 out of 2895 patients with genetically confirmed APL had PLZF-RAR α (0.3%) or NPM1-RAR α (0.1%). This frequency is lower than 1.1% previously reported by Grimwade et al. [3], which showed 5 PLZF-RAR α (0.8%) and 2 NPM1-RAR α (0.3%) out of 611 APL. However, our frequency data should be carefully interpreted due to several reasons: (1) although the PETHEMA registry includes all suspected APL patients regardless of treatment or diagnosis, investigators may include preferably those cases evaluable for PML-RAR α therapeutic protocols; (2) a number of registered APL patients were excluded from this study because of lack of genetic diagnosis. The majority of these cases had no genetic diagnosis because of very early death, but we cannot exclude that some of them had an unidentified variant APL; and (3) although the PETHEMA network offers expert counseling and possibility of genetic testing at central laboratories when required, the diagnosis of variant APL (in particular for very infrequent forms) needs first to be suspected and then guided PCR tests performed. In our opinion, the frequency of variant APL could be higher than herein reported, but probably lower than 1%.
Apart from accurate diagnosis, the main issue for the management of variant APL is the striking lack of evidence to guide therapeutic approaches in this population. Although our study aimed to shed light on this management, making reliable recommendations remains challenging in light of patient's and therapeutic approaches heterogeneity reported so far. Thus, we can only make suggestions for induction therapy, as follows: (1) a chemotherapy-based approach could be administered for STAT3-RAR α and GTF2I-RAR α as 3 reported cases were resistant to ATRA-or ATO-based regimens; (2) an AIDA or AIDA-like induction could be employed for NPM1-RAR α , IRF2P2-RAR α , FIP1L1-RAR α , BCOR-RAR α , NuMa-RAR α , PRKAR1A-RAR α , FNDC3B-RAR α , TFG-RAR α , and NABP1-RAR α (only 1 induction resistance reported). No clinical data regarding ATO sensitivity were available in the literature for these APL forms. However, in our series, one pediatric patient achieved a second CR after ATO plus ATRA plus intrathecal therapy for a concomitant bone marrow and CNS relapse, suggesting that ATO could be active as well in NPM1-RAR α APL; (3) an ATRA plus chemotherapy-based regimen could be employed for TBLR1-RAR α and STAT5B-RAR α , but the CR rate seems much lower than in PML-RAR α APL. Of note, an ATO-based salvage was successfully used in 1 patient with relapsed TBLR1-RAR α APL.
Regarding treatment and outcomes of PLZF-RAR α APL, previously reported cases (n = 35) and data from our cohort of patients (n = 9) allow for more reliable recommendations. Based on previously published data, induction schedules were quite heterogeneous, including some ATO-based regimens, with low CR (34%) and high resistance rates (51%). Of note, better CR rates (64%) were observed among 14 patients receiving ATRA plus chemotherapy-based induction ( Table 2). In contrast, front-line regimens were more homogeneous (AIDA or AIDA-like in 7 patients) in the PETHEMA cohort, no resistances occurred and early death was the only cause of induction failure, similar to PML-RAR α patients [70]. Until we have better evidence, it seems reasonable to recommend a first induction with AIDA or AIDA-like regimens for PLZF-RAR α patients. Regarding ATO therapy, although clinical responses are disappointing, combinations including ATO could play a role at relapse, where some responses have been observed so far (Tables 1 and 2). The mechanism of resistance to ATRA and ATO in PLZF-RAR α APL could be related to distinct nature of leukemogenic process as compared to the PML-RAR α . Rego et al., observed in vitro and in vivo, that therapeutic doses of ATRA and ATO can induce the degradation of both fusion protein, but maintenance of the leukemic phenotype depends on the continuous presence of the PML-RAR α but not of the PLZF-RAR α protein [71]. According to another study, ATRA, but not ATO, can provoke degradation of the PLZF-RARA fusion protein [27]. Moreover, it has been suggested that ATRA has difficulties to completely release the corepressor proteins like N-CoR (nuclear receptor corepressor) or nuclear receptor transcriptional (SMRT) from the PLZ-RAR α fusion protein as there is a second binding site for corepressor complex in the PLZF region [72]. We should also highlight that PML restauration by ATO and ATRA activates p53 by recruiting the protein to PML-nuclear bodies site and promoting its activation. The lack of p53 function could be in relation to the resistance that is observed in APL variants [73]. On the other hand, activation of the hematopoietic growth factor granulocyte colony-stimulating factor (G-CSF) receptor signaling could lead to the release of corepressor proteins from PLZF, supporting the therapeutic combination of ATRA and G-CSF [12]. Other studies provided biological rationale to enhance the efficacy of ATRA through combinations with interferon [74] or histone deacetylase inhibitors [75,76] and efficacy of ATRA and ATO through 8-CPT-CAMP [77,78]. We should mention the possibility of using tamibarotene instead of ATRA for those patients with variants since it is thought that tamibarotene is a better inducer of differentiation and cells death in APL cells [79].
Post-remission outcomes were worse in variant APL as compared to classical forms, mainly due to frequent relapses. In the literature cases, crude relapse rate was 57% among PLZF-RAR α and NPM1-RAR α , even if intensive consolidations were often administered. In our short series, 43% of patients relapsed with a remarkable frequency of CNS involvement, but a second CR was achieved in all patients. It should be noted that variant APLs frequently show risk factors that have been related with an increased risk of relapse in PML-RAR α APL (i.e., hyperleukocytosis and male gender) [80]. In our series, 3 out of 4 relapses occurred in patients with more than 50 × 10 9 /L WBC counts at diagnosis. Given the high rate of relapses observed, we can speculate that consolidation regimens including ATRA and chemotherapy (i.e., anthracycline and Ara-C) may be useful to prevent hematological and extramedullary relapses. Regarding specific CNS prophylaxis, it is difficult to make recommendations given the low number of cases, but it seems judicious to follow similar approaches than for typical APL. Although our short but mature series shows an acceptable proportion of patients still alive (73%), better than in previously published data. We can hypothesize that more homogeneous schedules with ATRA plus chemotherapy-based regimens, successful salvages, and more contemporaneous treatments, could explain our survival results. From our experience, the role of alloHSCT in first CR is debatable, but it should be performed when possible in second CR.
In conclusion, we confirm that variant forms are very rare, accounting for less than 1% of APLs. Main characteristics of 11 patients reported by PETHEMA are in line with previous reports (male predominance, high WBC, median age 45 years, PLZF-RAR α followed by NPM1-RAR α as more frequent rearrangements). Except for STAT3-RAR α and GTF2I-RAR α APLs, ATRA plus chemotherapy-based induction may lead to high CR rates. Further studies are needed to gain insights on optimal post-remission strategies to prevent the relatively high rate of relapse observed in variant APL patients. Funding: This work was partially financed with FEDER funds (CIBERONC (CB16/12/00284)) and with Instituto de Investigación Sanitaria La Fe funds (2016/0158).