Long-Term Follow-Up of a Phase II Study of All-Trans Retinoic Acid, Arsenic Trioxide and Gemtuzumab Ozogamicin in Acute Promyelocytic Leukemia

Introduction

Acute promyelocytic leukemia (APL) is a distinct subtype of acute myeloid leukemia (AML) characterized by the t(15;17)(q24.1;q21.2) translocation, which results in the fusion of the promyelocytic leukemia (PML) and retinoic acid receptor-α (RARA) genes, causing differentiation block.¹ The differentiating agents all-trans retinoic acid (ATRA) and arsenic trioxide (ATO) have transformed the treatment landscape of APL,² with long-term overall survival rates exceeding 90%.³ The APL0406 study of ATRA-ATO was done in standard-risk APL and excluded patients with a presenting white blood cell (WBC) >10×10⁹/L,³ which defines high-risk disease.^3–5

The optimal treatment for high-risk patients is an area of active investigation. The UK AML17 study randomized patients to ATRA-ATO or ATRA-idarubicin, with gemtuzumab ozogamicin (GO) permitted for patients randomized to ATRA-ATO.⁶ Twenty-eight (93%) high-risk patients and seven (8%) standard-risk patients in the ATRA-ATO arm received GO, with a 4-year event-free survival (EFS) of 91%. The Southwest Oncology Group (SWOG) 0535 study of ATRA + ATO + GO in high-risk APL demonstrated a complete remission (CR) rate of 86%, with a 3-year EFS of 78%.⁷ The APML4 phase II study of ATRA, ATO and idarubicin included 24 (20%) high-risk patients.⁸ These patients had a lower 2-year failure failure-free survival of 76%, compared to standard-risk patients. Despite the excellent long-term outcomes for APL, early mortality remains an issue, ranging from 4 – 11% in the ATO-ATRA era.^6,7 Differentiation syndrome (DS) complicates induction therapy in 10 – 37% of patients,^6,7,9 even when prophylactic corticosteroids are used.^8,10,11,12 A less-well recognized complication is the insidious development of fluid retention and consequent pulmonary and multiorgan failure requiring intensive care unit support, temporary withholding of ATRA -ATO, intensive diuresis, and steroids.¹³

GO is an antibody-drug conjugate of a CD33-directed antibody and calicheamicin. CD33 expression is ubiquitous in APL,¹⁴ and GO is highly active in APL, both in combination with ATRA and as a single agent.^15,16 In addition to being a feature of high-risk APL, leukocytosis >10×10⁹/L develops in 43% of standard-risk patients treated with ATO-ATRA.¹⁰ Leukocytosis increases the risk of APL-related complications such as differentiation syndrome and bleeding, including life-threatening intracranial haemorrhage.¹⁷ We previously reported the pooled outcomes of patients with APL treated on three different clinical trials at our institution.¹² Herein, we report the long-term outcomes of the most recently conducted phase II trial of ATRA + ATO and GO in APL, where all patients with WBC >10×10⁹/L at any point from diagnosis to end of induction received GO. This represents our current strategy for treating APL.

Materials and Methods

Results

From October 2011 to October 2023, 153 newly diagnosed APL patients were screened. There were seven screen failures for prior treatment, changes in treatment plan and insurance denials; 146 patients were treated on the study (Figure 1). One hundred and six (72.6%) were standard risk, while 40 (27.4%) were high risk. Baseline characteristics are presented in Table 1. The median age for the entire cohort was 53.0 years (range, 19.3 – 83.9), with no significant differences in age, gender, Eastern Cooperative Oncology Group performance status, ethnicity, PML::RARA isoform or additional cytogenetic abnormalities between standard-risk and high-risk patients. The most common additional cytogenetic abnormality was trisomy 8, which was observed in 16 (11.0%) patients, followed by a complex karyotype, defined as ≥3 abnormalities, in 15 (10.3%).

High-risk patients had a median presenting WBC of 24.4×10⁹/L (range, 10.0 – 197.6), compared with 1.4×10⁹/L (range, 0.2 – 9.4) in standard-risk patients. The characteristics of the patients by risk are shown in Table 1. As expected, high-risk patients were more likely to have internal tandem duplications mutations in the Fms-like tyrosine kinase 3 (FLT3) gene (FLT3^ITD), with 24 (60.0%) high-risk patients having FLT3^ITD compared with 29 (27.4%) standard-risk patients (p = 0.0004). Multi-gene panel NGS testing was performed on 122 (83.6%) patients. Other frequently mutated genes included FLT3 tyrosine kinase domain (FLT3^TKD), WT1, NRAS and KRAS (Table 1). No patients had somatic mutations in CEBPA, DDX41, IDH1, IDH2 or NPM1. One patient had a TP53 mutation.

Overall, 141 (96.6%) patients completed induction, 102 (96.2%) in the standard-risk group and 39 (97.5%) in the high-risk group. Reasons for not completing induction were induction mortality in four (2.7%, 95% CI, 1.1 – 6.8) patients and transfer of care due to insurance withdrawal in the fifth patient (Figure 1). Of the patients who completed induction, 100% achieved a CR/CRi. One hundred and thirty-seven patients (93.8%, 95% confidence interval [CI], 92.2 – 98.5) achieved a CR, 96.2% (95% CI, 90.7 – 98.5) in the standard-risk group and 97.5% (95% CI, 87.1 – 99.6) in the high-risk group. CRi was noted in four patients (3 [2.8%, 95% CI, 1.0 – 8.0] standard-risk and 1 [2.5%, 95% CI, 0.4 – 12.9] high-risk). These patients were discharged from hospital after achieving morphologic remission but were unable to return to our institution for follow-up care and documentation of CR because of insurance coverage.

The median time to CR/CRi was 31.4 days (95% CI, 30.4–33.5 days). Standard-risk patients had a slightly longer median time to CR/CRi than high-risk patients (33.5 days [95% CI, 31.5–35.5 days] vs. 28.4 days [95% CI, 27.4–31.5 days]; p = .0144; Figure 2A). MRD negativity was achieved in 133 of 137 patients (97.1%) who achieved CR. Of the four patients without documented MRD negativity, all four were taken off study before completion of treatment; two for insurance issues, one because consent was withdrawn, and one for nonadherence to study procedures. The median time to MRD clearance was 3.4 months (95% CI, 3.2–3.7 months) from the start of treatment, and 95% achieved MRD clearance by 5.0 months. There was no difference in the median time to MRD negativity between standard-risk and high-risk patients (3.4 months [95% CI, 3.2–3.8 months] vs. 3.6 months [95% CI, 3.0–4.1 months]; p = .4346; Figure 2B).

All patients in the high-risk group and 68 (64.2%) patients in the standard-risk group received GO. Eighty-seven (80.6%) of those receiving GO were given the 9mg/m² dose, 19 (17.6%) received 6mg/m² and two (1.9%) received 3mg/m². Only one standard-risk patient required GO during consolidation. This patient received ATRA-ATO without GO during induction, which was uncomplicated, and achieved CR. RT-qPCR for PML::RARA was persistently positive (0.01%) after cycle 2 of consolidation. One dose of GO 6mg/m² was given during consolidation cycle 3, with subsequent clearance of PML::RARA by RT-qPCR. No further doses of GO were required.

The median follow-up was 61.8 months (95% CI, 4.7–128.4 months) in surviving patients. The median EFS for the whole cohort was not reached, with a 5-year EFS of 92.4% (95% CI, 87.9%–97.1%; Figure 3A). There was no significant difference in EFS between the standard-risk and high-risk groups (median EFS was not reached in either group; p = .2657; Figure 3B). The 5-year EFS for standard-risk patients was 93.6% (95% CI, 88.7%–98.8%) and, for high-risk patients, it was 89.4% (95% CI, 80.1%–99.9%). When patients were stratified by age into groups younger than 60 years and 60 years and older, irrespective of risk status, there was no significant difference between EFS (median EFS was not reached for either group; p = .4126; Figure 3C). The 5-year EFS for patients younger than 60 years was 92.7% (95% CI, 87.6%–98.1%) and, for those aged 60 years and older, it was 92.3% (95% CI, 84.2%–100.0%). Events are shown in Table S1.

Stratifying patients by presence or absence of FLT3^ITD (p = 0.6060), mutational complexity (one or more mutations vs. no mutations [p = .8834; Figure S2A] or two or more mutations vs. one or less mutation [p = .5249]), or the presence of additional cytogenetic abnormalities (p = .7609) also had no effect on EFS. There was no significant difference in EFS among standard-risk patients who received GO compared with those who did not (median EFS for patients not receiving GO, 111.8 months [95% CI, 93.3 months to not estimable] vs. not reached for patients receiving GO; p = .1179).

The median DFS for the entire cohort was not reached. The 5-year DFS was 93.6% (95% CI, 89.5%–97.8%), which was identical to the 10-year DFS (Figure 3D). Standard-risk patients had a 5-year DFS of 95.2% (95% CI, 91.1%–99.4%) compared with 89.4% (95% CI, 80.1%–99.9%) in high-risk patients, with a median DFS that was not reached in either group (p = .2558; Figure 3E). There was no difference in DFS when stratified by age (p = .5891; Figure 3F), presence of one or more mutation(s) (p = .9292; Figure S2B), two or more mutations (p = .3898), or additional cytogenetic abnormalities (p = .3386). The median DFS was the same for standard-risk patients regardless of whether GO was administered (not reached for both groups; p = .8492).

The median OS for the entire cohort was not reached (Figure S3A). The 5-year OS was 93.1% (95% CI, 88.9%–97.7%), with a 10-year OS of 74.7% (95% CI, 61.8%–90.2%). All deaths after 4 years were unrelated to APL. There was no difference in OS between standard-risk and high-risk patients (median not reached; p = .1795; Figure S3B), patients younger than 60 years versus those aged 60 years and older (p = .2934; Figure S3C), or when stratified by the presence of one or more mutation(s) (p = .8880; Figure S3D), two or more mutations (p = .2101), or additional cytogenetic abnormalities (p = .5563).

Only two patients in the entire cohort relapsed (Figure S4), a standard-risk patient at 9.5 months and a high-risk patient at 27.6 months after attaining CR. The 5-year CIR was 1.1% (95% CI, 0.0 – 3.1) for standard-risk patients and 3.3% (95% CI, 0.0 – 9.5) for high-risk patients (p = 0.4769, Figure S5). Both were given GO during induction and had achieved MRD negativity at 2.4 and 4.5 months respectively. The standard-risk patient had WT1 and NRAS mutations at diagnosis, which were again present at relapse. The patient presented with hematological relapse and was successfully salvaged with ATO, ATRA and idarubicin, followed by consolidation allogeneic stem cell transplant in second complete remission. The high-risk patient was had PML::RARA detected by RT-qPCR, t(15;17) and a FLT3^ITD mutation at diagnosis. The patient was lost to follow-up after completion of therapy. The patient relapsed at another institution and died two months later; no other information was available.

There were four (2.7%) deaths during induction, three in standard-risk patients and one in a high-risk patient. Two of the standard-risk patients had received GO during induction. The causes of death for the four patients were: (1) diffuse alveolar hemorrhage with secondary Stenotrophomonas bacteremia and multiorgan failure, (2) myocardial infarction, (3) intracranial hemorrhage, and (4) pneumonia with secondary, diffuse alveolar hemorrhage and respiratory failure. AEs in which at least one grade 3 event occurred in a patient are shown in Table S2. The most common AE was infection, which occurred in 20 patients (13.7%). Differentiation syndrome occurred in 15 patients (10.3%) and was grade ≥3 in seven (4.8%). There were no deaths from differentiation syndrome. QTc prolongation was observed in 18 patients (12.3%), whereas elevated transaminases were noted in 60 (41%). Most cases of elevated transaminases were grade 1–2, and 10 (6.8%) were grade ≥3. Biochemical elevations in lipase were seen in eight patients (5.5%), with two cases of grade 2 pancreatitis. There were no cases of veno-occlusive disease in our cohort. There were two cases of grade ≥3 acute kidney injury requiring dialysis; details of these patients have been reported previously.¹⁸ Three patients experienced grade ≥3 headache, of whom two responded to acetazolamide. None of these patients had objective evidence of raised intracranial pressure on imaging or fundoscopy. There was no significant difference in the rate of AEs between patients who received 3, 6, and 9 mg/m² of GO.

Discussion

In this study we update the excellent long-term outcomes of patients with APL treated with ATRA, ATO and GO at our institution. GO was administered to all high-risk and 64.2% of standard-risk patients. At a median follow-up of over 5 years, this regimen is risk-agnostic and curative in >90% of patients, with only two relapses observed to date. Late deaths after four years were unrelated to APL, resulting in a 10-year DFS of 93.5%. The regimen is tolerable, with induction mortality <3%, a low rate of differentiation syndrome (<5% grade ≥3), no cases of veno-occlusive disease and no new safety signals. Our results re-affirm the favorable outcomes of APL treated with ATRA + ATO ± GO.

Our results compare favorably to other studies of ATO, ATRA and GO. The UK AML17 study compared ATRA-ATO (GO added for high-risk patients) with ATRA and idarubicin, showing that ATRA-ATO was associated with a significantly lower requirement for supportive care.⁶ The morphological remission rate was 91%, with a 30-day mortality of 4% in the ATRA-ATO arm. In contrast to the AML17 study, we administer GO to standard-risk patients who develop leukocytosis >10×10⁹/L at any point during induction. In our study, all evaluable patients achieved a morphological remission. The addition of GO did not result in increased early mortality, which was only 2.7%. In addition, the rate of grade ≥3 differentiation syndrome was low compared with historical rates of 20 – 30%.⁶ It is notable that patients age ≥60 had similar outcomes to younger patients. While this was a smaller subset of our cohort, 16 patients ≥ 60 years with standard risk APL received GO and we did not observe any cases of veno-occlusive disease or early mortality in these older individuals. This is likely due to improved experience with ATO-ATRA-GO, and represents an improvement over our previously published experience.¹² A high index of suspicion is required for specific rare complications such as insidious fluid retention,¹³ distinct from differentiation syndrome, renal impairment¹⁸ and pseudotumor cerebri¹⁹.

In the SWOG study of high-risk APL, GO was added to ATO-ATRA in induction.⁷ However, differences from our study include the use of daunorubicin in consolidation and two further doses of GO for all patients, regardless of MRD status. The SWOG study also utilized a maintenance regimen of ATRA, 6-mercaptopurine and methotrexate, the role of which is debatable in the ATO-ATRA era with MRD monitoring.¹⁷ ATRA-ATO has also been consistently shown to improve outcomes compared with ATRA-chemotherapy combinations, which require more supportive care and are associated with more toxicities.^3,6 The induction mortality in the SWOG study was 10% and the CR rate 86%.⁷ The 3-year EFS of 78% included four patients who relapsed and two deaths in remission; the 3-year CIR was 7.1%. In contrast, our 5-year EFS was 92%, with a 5-year CIR of 3% in high-risk patients. Our strategy has been to monitor patients with MRD, reserving intensification with repeated GO dosing only for patients who fail to achieve molecular clearance. This strategy, which was only required in one patient, minimizes toxicity for patients who do not require intensification, while preserving efficacy for slow responders.

FLT3^ITD mutations were present in 27% of standard-risk patients and 60% of high-risk patients. The presence of FLT3^ITD was not associated with a significant difference in EFS. FLT3^ITD mutations are well-described in APL, but their effect on prognosis is conflicting.²⁰ Studies which demonstrate an adverse prognostic effect of FLT3^ITD were largely conducted in the pre-ATO era.^21,22 A study in mouse models demonstrated that FLT3^ITD was associated with a blunted response to ATRA by reducing the degradation of PML-RARA.²³ Combination with ATO re-induced differentiation and RARA degradation, abrogating FLT3^ITD-mediated resistance to ATRA, which explains the clinical observation that ATO-ATRA combinations have an excellent prognosis regardless of FLT3^ITD mutation status.^8,24 In keeping with the high cure rates of our regimen, we also demonstrate that other traditional adverse prognostic markers, such as baseline WBC, and other exploratory prognostic markers such as additional mutations or cytogenetic abnormalities, have no bearing on outcomes. This is unsurprising given the effectiveness of the treatment and the low event rates.²⁵

A consensus statement from the European LeukemiaNet (ELN) on MRD monitoring in AML has suggested that MRD in APL should be monitored from the time of completion of consolidation to marrow negativity in standard-risk patients and for 24 months in high-risk patients.²⁶ This is based on data from the pre-ATO era which demonstrated that positive MRD in remission was highly predictive of relapse, typically within four months,^27,28 and that treatment of molecular relapse can prevent progression to overt hematological relapse²⁸ and lead to longer leukemia-free survival.²⁹ However, time-to-event analysis when comparing molecular and hematological relapses are subject to lead-time bias, especially when it has been demonstrated that molecular relapses invariably precede hematological relapses. The relapse rate in our study was extremely low, with an overall 5-year CIR of 1%, one patient in each risk category. It is possible that with the excellent cure rates observed, MRD monitoring, especially in high-risk APL, may not be required in the future.³⁰

In conclusion, we report on the long-term outcomes of a large cohort of APL patients, including a cohort of standard-risk patients who were administered GO with a high WBC. Because of the non-randomized nature of the study, we are unable to definitively conclude if the administration of GO was associated with improved outcomes in terms of improved EFS, reduced induction mortality and less differentiation syndrome. However, the results remain extremely encouraging, with excellent long-term DFS, and demonstrate the outstanding activity and tolerability of the ATO, ATRA and GO in APL.

Supplementary Material

Data sharing:

The study details are available online on ClinicalTrials.gov, reference number NCT01409161. The study data is not publicly available to respect participant confidentiality. Requests for sharing of deidentified data should be directed to the corresponding author.