Oral arsenic plus imatinib versus imatinib solely for newly diagnosed chronic myeloid leukemia: a randomized phase 3 trial with 5-year outcomes

Chronic myeloid leukemia (CML) is characterized by the BCR::ABL1 oncoprotein, which produces a constitutively active tyrosine kinase that leads to pathogenesis (Heisterkamp et al. 1983). Tyrosine kinase inhibitors (TKIs) can inhibit the activity of the BCR::ABL1 fusion protein to trigger cell apoptosis. They have shown high rates of molecular responses and improved prognoses in patients with chronic phase (CP) of chronic myeloid leukemia (CML) (Hochhaus et al. 2020). The major molecular response (MMR) is recognized as the main optimal milestone of CML treatment (Deininger et al. 2020; Hochhaus et al. 2020). Nevertheless, a deep molecular response (DMR, molecular response 4 or better), particularly molecular response 4.5 (MR^4.5) or more, has been recently regarded as the gateway to treatment discontinuation and treatment-free remission (TFR) for patients with CP-CML, resulting in a low risk of progression (Deininger et al. 2020; Cortes et al. 2021). However, the rate of MR^4.5 is only 1–11% with TKIs in 1 year and 8–25% in 2 years (Hehlmann et al. 2014; Cortes et al. 2016, 2018; Hochhaus et al. 2016a, b; Hochhaus et al. 2016a, b). Therefore, it is crucial to identify other methods to increase the proportion of deeper molecular responses.

Intravenous arsenic trioxide and oral tetra-arsenic tetrasulfide (As₄S₄) have proven to be highly effective and safe and are used as a standard treatment for acute promyelocytic leukemia (APL) (Lu et al. 2002; Zhu et al. 2018). Based on in vitro data in the K562 cell line and CML primary cells from patients, arsenic can directly induce cell apoptosis and degrade BCR::ABL1 rather than inhibiting BCR::ABL1 activity as would a TKI (Li et al. 2002; Yin et al. 2004; Mao et al. 2010). Subsequently, additional results were observed in a mouse model of BCR::ABL-positive CML, where arsenic acted on BCR::ABL by directly binding the RING finger domain of c-CBL to inhibit its self-ubiquitination/degradation, while imatinib inhibits the PI3K/AKT/mTOR pathway (Zhang et al. 2009; Mao et al. 2010; Liu et al. 2014). The synergistic effect of the two drugs at the molecular level might be a promising approach to improving response rates in CML patients.

In addition, the realgar–indigo naturalis formula (RIF), an oral arsenic agent (As₄S₄)-containing formula, is convenient for managing medication and not inferior to intravenous arsenic trioxide (Zhu et al. 2018).

Arsenic is a standard treatment agent for APL and has a long history of use in CML. It was the standard treatment for CML before the emergence of chemotherapy in the late nineteenth century (Forkner and Scott 1931, Zhu et al. 2018). Our study was based on several previous studies that had confirmed the synergistic effects of RIF and imatinib in CML cells and mice (Li et al. 2002; Yin et al. 2004; Zhang et al. 2009; Mao et al. 2010). While As₄S₄ triggers BCR::ABL1 degradation, imatinib inhibits its tyrosine kinase activity. The combination of these two agents was found to be able to lower protein and enzymatic activity levels of BCR::ABL (Yin et al. 2004). Arsenic considerably could upregulate c-CBL, which serves as an E3 ligase for several receptor/protein tyrosine kinases, including BCR::ABL, and mediate the ubiquitination and degradation of BCR::ABL (Mao et al. 2010).

In this randomized phase 3 trial, we compared the arsenic-containing combination therapy, RIF plus imatinib, with the current standard first-line therapy, imatinib, as an initial trial for newly diagnosed CP-CML. The recruitment of this study was terminated earlier than the initial design because of the coronavirus COVID-19 pandemic. Of the planned enrollment of 488 participants, only 191 participants were recruited; 157 participants reached the 12-month time point for analysis of the molecular response, and 159 participants were still on follow-up at 5 years.

In this trial, the primary endpoint of MMR at six months of combination therapy was similar to imatinib monotherapy based on the ITT population. The MMR rate in the imatinib arm at 12 months was 35.8%, which is consistent with that in other randomized trials (22–36.9%) (Kantarjian et al. 2010; Saglio et al. 2010; Wang et al. 2015; Cortes et al. 2018). No significant differences were observed in MMR and MR⁴ rates between the two groups at any time within 12 months. However, there was a trend toward higher rates of achieving these endpoints with combination therapy compared to imatinib alone. A significant improvement in MR^4.5 with combination therapy versus imatinib was identified at 12 months (p = 0.043), although the efficacy analyses were limited by early termination of recruitment. Moreover, at 2 years since randomization, 16.9% more participants who received combination therapy achieved MR^4.5 than those who received imatinib alone (55.6% vs. 38.6%). In recent years, several studies have confirmed that the depth of a patient’s molecular response is positively associated with the probability of TFR success and is essential for determining whether TFR is achieved (Hochhaus et al. 2017b, Takahashi et al. 2018a, b; Takahashi et al. 2018a, b). However, the 5-year MMR, MR⁴, and MR^4.5 rates were similar in the treatment groups. This could be owing to the longer duration of the effect by imatinib required against CML-initiating cells or alternative treatment with the second-generation TKIs after discontinuing the study treatment.

Current therapeutic aims are directed at achieving sufficient DMR to reduce the risks of blast crisis transformation and increase the rates of TFR, which have been extensively investigated and are now part of the management of CML patients (Deininger et al. 2020; Stuckey et al. 2020; Cortes et al. 2021; Krishnan et al. 2022). Etienne et al. reported that 233 patients treated with front-line imatinib who achieved MR^4.5 had better event-free survival and failure-free survival than those with a complete cytogenetic response and MMR status (Etienne et al. 2014). More consensus has been reached that sustained MR^4.5 is an ideal objectif and is associated with higher TFR rates than sustained MR⁴ (Etienne et al. 2014; Branford 2020; Deininger et al. 2020; Cortes et al. 2021). In those trials on the second-generation TKIs in patients with newly diagnosed CP-CML, the cumulative rates of MR^4.5 by 12 months with dasatinib (DASISION trial), bosutinib (BFORE trial), nilotinib 300 mg twice daily, and nilotinib 400 mg twice daily (ENESTnd trial) were 5%, 6.4%, 7%, and 11%, respectively, and after 5 years were 33%, 46%, 52%, and 54%, respectively (Cortes et al. 2016, 2018; Hochhaus et al. 2016a, b; Brümmendorf et al. 2022), which were superior to that with imatinib. Although comparisons between trials should be considered with caution, the MR^4.5 rate with RIF plus imatinib (20.8% by 12 months) may be potentially equal to or even higher than observed with the second-generation TKIs. Several TKI discontinuation trials with limited long-term follow-ups have been reported. TFR rates in imatinib discontinuation studies for patients with a sustained MR^4.5 or better were estimated as 47%–65% at 12 months and 33%–64% at 24 months (Rousselot et al. 2014; Lee et al. 2016; Etienne et al. 2017); in dasatinib, discontinuation studies were 48% at 12 months and 46% at 24 months (Shah et al. 2020), and in nilotinib, discontinuation studies were 51.6–58% at 12 months and 49–53% at 24 months (Hochhaus et al. 2017a, b; Mahon et al. 2018; Ross et al. 2018). Therefore, TKIs combined with RIF may be an efficacy treatment strategy for patients with CP-CML who would be willing to attempt discontinuation.

PFS and OS rates remain high and comparable between the trial groups (more than 90%). These results are consistent with the long-term outcomes in patients with newly diagnosed CP-CML who received the second-generation TKIs nilotinib, dasatinib, and bosutinib (Cortes et al. 2016; Hochhaus et al. 2016a, b; Brümmendorf et al. 2022). Patients with CML treated with TKI can expect a near-normal life expectancy (Cortes et al. 2021).

Safety data were consistent with the known AE profiles of RIF in newly diagnosed patients with APL and imatinib in newly diagnosed patients with CP-CML (Preudhomme et al. 2010; Zhu et al. 2018, Chen, Zhu et al. 2021). Most AEs occurred primarily during the first six months of treatment and were generally controllable, which is consistent with other trials (Zhu et al. 2018, Chen et al. 2021). No severe adverse effects were observed at a cumulative dose of 105,840 mg realgar in this study. The predominant AEs of the RIF plus imatinib regimen were diarrhea, vomiting, nausea, edema, upper respiratory infection, cytopenia, and liver function abnormalities; most participants tolerated the regimen well and did not need to adjust the drug dose in our trial. The incidence and duration of hematological AEs in the two groups were similar, implying that RIF did not significantly increase hematological toxicity. The incidence of hepatotoxicity in the combination group was similar to that reported in other studies involving arsenic agents (Zhu et al. 2018, Chen et al. 2021), though all being grade 1 or grade 2 in our trial. It has been reported that a small prolongation of the QTc interval occurred in patients who received arsenic trioxide, however, to a lesser extent in patients who received RIF, which is consistent with previous reports (Lo-Coco et al. 2013; Zhu et al. 2018).

The combination therapy was more toxic than imatinib alone regarding gastrointestinal side effects, such as diarrhea (p = 0.036), vomiting (p = 0.004), nausea (p = 0.006), and liver function abnormalities, including ALT (p = 0.006) and AST (p < 0.001), which led to temporary interruptions and/or reductions in the combination group (most cases stopped both RIF and imatinib) (Table S6). The rates of MMR and MR⁴ in the combination group were lower than the imatinib group in the early stages of our trial; however, a reversal emerged in response between the trial groups at 9 months when AEs decreased. Furthermore, the difference between combination therapy and imatinib increased subsequently over time, and at 12 months, the MR^4.5 rate showed a significant increase in the combination arm (p = 0.043). In addition, the reasons for the more frequent occurrence of muscle cramps (p = 0.003), rash (p = 0.086), pruritus, muscle ache, and bleeding with imatinib alone might be due to dose interruptions and/or reductions and gastrointestinal symptoms affecting drug absorption at the early stage in the combination group. This also supports our speculation that the time of occurrence of hematological AEs in the combination group lagged behind that in the imatinib group. Therefore, an early molecular response might not reflect the efficacy of the two arms well.

This study had several limitations. First, the significant limitation of our study was low enrollment due to recruitment termination during the COVID-19 pandemic in early 2020, which underpowered the result. Second, our trial found a higher rate of achieving MR^4.5 in participants who received RIF plus imatinib than those who received imatinib monotherapy. Thus, in future, we could plan to design a trial to further evaluate RIF plus TKI for this index. Third, the follow-up time is a little short for TFR, which is another limitation of our trial, and we will continue to follow these patients for a long time.

In summary, there was no statistically significant difference in MMR rates between combination and imatinib arms at 6 months. However, we found that the combination regimen may increase the rates of MR^4.5 which are a prerequisite for TFR, than imatinib alone in participants with de novo CP-CML. In addition, the safety profile of combination treatment was similar to that of imatinib. As enrollment in this trial was terminated early, the efficacy of RIF with imatinib in this setting remains to be established.