The pharmacokinetics of therapeutic arsenic trioxide in acute promyelocytic leukemia patients

Abstract

Arsenic trioxide (ATO) treats Acute Promyelocytic Leukemia (APL). ATO is converted from inorganic arsenic (iAs) to methylated (MAs) and dimethylated (DMAs) metabolites, which are excreted in the urine. Methylation of iAs is important in detoxification, as iAs exposure is deleterious to health. We examined ATO metabolism in 25 APL patients, measuring iAs, MAs, and DMAs. Plasma total iAs increased after ATO administration, followed by a rapid decline, reaching trough levels by 4–6 h. We identified two patterns of iAs metabolism between 6 and 24 h after infusion: in Group 1, iAs increased and were slowly converted to MAs and DMAs, whereas in Group 2, iAs was rapidly metabolized. These patterns were associated with smoking and different treatments: ATO with all-trans retinoic acid (ATRA) alone vs. ATO preceded by ATRA and chemotherapy. Our data suggest that smoking and prior chemotherapy exposure may be associated with ATO metabolism stimulation, thus lowering the effective blood ATO dose.

Introduction

Acute Promyelocytic Leukemia (APL) is an aggressive leukemia, with the pathognomonic translocation t(15;17) (q24;q21) leading to fusion of retinoic acid receptor alpha (RARA) gene on chromosome band 17q21 with nuclear regulatory factor gene on chromosome band 15q24 (PML gene) [1]. This unique PML-RARA fusion gene is targeted by Arsenic Trioxide (ATO) and All-Trans Retinoic Acid (ATRA), which enable promyelocytes to overcome their differentiation blockade [2].

Unlike ATRA, single-agent ATO can induce deep molecular remissions (PCR negative) in APL patients [3] . Frontline treatment for APL historically consisted of ATRA with chemotherapy (Cytarabine and an anthracycline) with ATO in second-line settings. Clinical trials such as CALGB C9710 have shown that upfront ATO treatment regimens were superior to treatments without ATO [4]. Subsequent trials confirmed excellent complete remission rates with fewer toxicities in ATRA plus ATO regimens. Thus, upfront ATRA and ATO combination has become the standard of care in low/intermediate-risk APL patients [5]. Randomized clinical trials, such as AML-17, have also shown feasibility and high cure rates with ATRA and ATO in high-risk patients [6]. Thus, ATO has widely been used in APL treatment and is associated with 90-95% of overall survival [7,8].

In contrast, inorganic arsenic (iAs) chronic environmental exposures are associated with deleterious health effects, from increased risk of cardiometabolic disease to skin, bladder, and lung cancer [9–15]. iAs is metabolized in the liver through methylation by arsenic (+3 oxidation state) methyltransferase (AS3MT), which converts iAs to methyl- and dimethylarsenic metabolites (MAs and DMAs), then excreted in the urine. The efficiency of this conversion has been shown to affect iAs toxicity and risk of diseases [9–15].

ATO is quickly hydrolyzed to active iAs, arsenious acid, which is methylated by AS3MT to MAs and DMAs [16] in a similar manner to environmental iAs. Notably, some of the methylated metabolites, specifically trivalent MAsIII and DMAsIII, are toxic for APL cells [17,18] and may contribute to its therapeutic effects or to toxicities. The ATO therapeutic dose is much higher than environmental iAs exposures, thus may lead to saturation of AS3MT methylation capacity. The pharmacokinetics of ATO and the efficiency of its conversion from iAs to MAs and DMAs in APL patients are not well understood.

The aims of this study were to measure iAs in the blood (plasma) and urine of APL patients who received therapeutic ATO and to assess the efficiency of ATO detoxification through its conversion to MAs and DMAs, as well as residual levels of arsenic at end of treatment.

Materials and methods

Patients

We enrolled 26 adults, newly diagnosed APL patients (18 years and older) treated with ATO at Northwell Health System between 2011 and 2015. The study was conducted according to the standards of good clinical practice, approved by the local Institutional Review Board (trial no. N10-326A), and in accordance with the Declaration of Helsinki. All patients provided written informed consent. One patient’s data were excluded from analysis after revoking consent.

Treatments

APL patients were treated with one of three commonly used regimens (Figure 1(a)): North American Intergroup Study C9710 Regimen [4], LoCoco Regimen [5], and SWOG S0535 Regimen [19].

Figure 1.

The first dose of ATO is considered day 1 (d1) of the (a) treatment regimens and 0 h with respect to (b) the sampling time points. (a) Treatment regimens are aligned such that ATO dosing (0.15 mg/kg/day IV) begins on day 1. ATRA dosing is 45 mg/m2/day PO. Gemtuzumab ozogamicin (GO), Daunorubicin, and Cytarabine (Ara C) were administered at a dose of 9 mg/m2 IV, 50 mg/m2 IV, and 200 mg/m2 IV, respectively. Prior to ATO treatment is not to scale. (b) Time points at which each matrix was sampled are indicated by +.

Sample collection

ATO at 0.15 mg/kg, was given intravenously over 2 h. The sampling times for analyses of iAs, Mas, and DMAs are in Figure 1(b). For peripheral blood, 5 ml were drawn from the contralateral arm to ATO infusion, in the heparinized tube at −1 h, at end of infusion 0 h, at 1, 2, 4, 6, and 24 h, and at 4, 8, and 15 d. Additional samples were collected 30 d after the last infusion. The blood tubes were placed on ice, centrifuged at 4 °C for 10 min for plasma separation. The plasma was aliquoted into a plastic tube, frozen at −70 °C, and shipped overnight on dry ice to the University of North Carolina Chapel Hill (UNC). Urine samples (both 24-h urine and spot urine) were collected at 1, and 4 d and 30 d post-treatment (Table 1b). Exfoliated urothelial cells were isolated from 24-h urine and spot urine samples by centrifugation at 450 × g at 4 °C for 10 min. Samples were frozen immediately after collection and stored at −80 °C. The frozen samples of 2–4 patients were batch-shipped on dry ice to UNC for arsenic speciation analysis.

Table 1.

Characteristics of APL patients treated with ATO.

Characteristic (%)	All patients N = 25	Group 1: slow metabolizers N = 13 (52% )	Group 2: fast metabolizers N = 12 (48%)	p-value
Age (years)
Mean (Range)	51 (27, 81)	52 (28, 76)	51 (27, 81)	.89
Body mass index
Mean (Range)	30.1 (21.8, 45.1)	29.2 (21.8, 40.0)	31.1 (25.2, 45.1)	.5
Sex
Male	14 (56.0%)	9 (69.2%)	5 (41.7%)	.24
Female	11 (44.0%)	4 (30.8%)	7 (58.3%)
APL risk level
High risk	5 (20.0%)	4 (30.8%)	1 (8.3%)	.32
Low/intermediate	20 (80.0%)	9 (69.2%)	11 (91.7%)
Race
Caucasian	12 (48.0%)	7 (53.8%)	5 (41.7%)	.12
Hispanic	3 (12.0%)	1 (7.7%)	2 (16.7%)
African American	6 (24.0%)	5 (38.5%)	1 (8.3%)
Asian	2 (8.0%)	0 (0%)	2 (16.7%)
Other	2 (8.0%)	0 (0%)	2 (16.7%)
Smoking
Current	7 (28.0%)	7 (53.8%)	1 (8.2%)	.03
Former/Never	16 (64.0%)	6 (46.2%)	11 (91.7%)
Alcohol use
Any	5 (20.0%)	4 (30.8%)	1 (8.3%)	.32
Never	20 (80.0%)	9 (69.2%)	11 (91.7%)

t-test was conducted for continuous variables. Fisher’s exact test or Chi-square test was conducted for categorical variables.

Values in bold are statistically significant p values, namely less than 0.05

Arsenic speciation analysis

Concentrations of iAs, Mas, and DMAs in plasma, urine, and exfoliated urothelial cells were measured by Hydride Generation-Atomic Absorption Spectrometry coupled with a CryoTrap (HG-CT-AAS) [20]. All samples were treated with 2% L-cysteine (Sigma-Aldrich, St. Louis, MO, USA) for 1 h at room temperature to reduce pentavalent iAs, MAs, and DMAs to their trivalent counterparts [21]. Cysteine-treated samples were then adjusted to pH 6 using NaOH (Sigma-Aldrich). The treatment of samples with cysteine prior to the HG step reduces all pentavalent arsenic species in the samples to trivalency. Hydrides corresponding to iAs (arsine), methylarsenic (methylarsine), and dimethyl arsenic (dimethylarsine) are then generated from the trivalent species originally present in the samples and the trivalent species formed by the reduction of pentavalent species with cysteine. Thus, all tri- and pentavalent arsenic species are accounted for using this method.

The analysis by HG-CT-AAS was carried out directly without further sample pretreatment or extraction. The limits of detection (LODs) for iAs, MAs, and DMAs were as follows: in plasma, 0.28, 0.14, and 0.14 ng As/mL, respectively; in urine, 0.41, 0.22, and 0.24 ng As/mL, respectively; in exfoliated cells, 27, 13, and 13 pg As/106 cells, respectively. Imputed values of LOD/$\sqrt{2}$ were used for values below LODs. Total As (tAs) concentration was calculated as the sum of the concentrations of iAs + MAs + DMAs.

A standard reference material ‘Species in Frozen Human Urine’ (SRM 2669; NIST) was used with every batch of samples sent to UNC to assure the accuracy of the analysis. Concentrations of iAs, MAs, and DMAs measured by HG-CT-AAS in SRM 2669 exceeded 90% of the certified values for all batches.

Statistical analysis

Patients were grouped by plasma concentrations of iAs between 6 and 24 h. Increasing and decreasing plasma concentrations between hours 6 and 24 after infusion were categorized as Groups 1 and 2, respectively. Patient characteristics were compared by group using the student’s T-test for continuous variables and Fisher’s exact test for categorical variables. The concentration time-course of all species was plotted in plasma, urine (spot and 24-h), and exfoliated urothelial cells (EUC; spot and 24-h). Serum time points after −1 h were linearly interpolated. Five (of 175) such data points were missing, and 1 and 2 values were interpolated at the 6- and 24-h plasma timepoints, respectively. For the 5 subjects missing time points at baseline (−1 h), participants were assigned the average of all values for that time-point. Group-specific urine and EUC were statistically compared using a t-test or non-parametric test (for non-normal data) at all time points, for both spot and 24-h measures. The authors conducted the data analyses and all authors had access to the data. For access to original data, please contact miroslav_styblo@med.unc.edu.

Results

Metabolites of ATO in plasma

From November 2011 to October 2015, 26 APL patients were enrolled, one patient withdrew consent and was thus excluded from data analysis. Demographic data on all patients are in Table 1. On presentation, most patients had low/intermediate-risk APL (defined as white blood cell count ≤ 10,000/μL), had a history of cardiovascular disease (hypertension and hypercholesterolemia), mild liver transaminitis, and were not actively smoking or drinking alcohol. Almost all of the patients (88.0%) had normal renal function. The age distribution was between 27 and 81 years old, with a median age of 51 years. There were more male patients (56.0%) compared to female patients (44.0%) in the study. The most prevalent ethnicity was Caucasian (48.0%), but a significant percentage of patients were African American (24.0%) and Hispanic (12.0%).

Regarding treatment for APL, more patients (68.0%) received ATRA/ATO treatment regimens alone for induction (LoCoco or SWOG 0535 regimens), whereas 32.0% received ATRA with an anthracycline (Daunorubicin) and a pyrimidine analog (Cytarabine) upfront (following C9710 regimen) [4,5,19]. These latter patients received ATO during their consolidation phase, after achieving morphological complete remission. Furthermore, all the patients following C9710 were nonsmokers, whereas active smokers were evenly distributed between the chemotherapy and non-chemotherapy containing regimens. Treatment regimens are in Figure 1(a) [4,5,19].

Plasma concentrations of arsenic species were evaluated during 3 phases: the acute phase −1–24 h), the ‘repetitive phase’ (1–15 d), in which ATO was administered, and the ‘post-treatment phase’ which was 30 d after treatment with ATO ended. We assessed metabolism at all these time points. Individual patients’ treatment regimens were at the discretion of the treating physician. Thus, whereas patients treated following the LoCoco and SWOG 0535 received ATO daily for at least 30 sequential days, patients treated following C9710 were given ATO for only 5 d weekly for 5 consecutive weeks during consolidation (Figure 1(a)). Since this difference in the patients’ exposure to ATO would affect the profiles of iAs and its metabolites, we compared the kinetics of ATO metabolism during the first 4 d, when treatment was consistent in all patients.

As expected, total arsenic in plasma (tAs = iAs + MAs + DMAs) rapidly increased from the baseline level (−1 h) to the time right after therapeutic ATO infusion (0 h), then declined over the next 2 h, reaching a steady-state (mean = 9.0 ng/mL) by 4–6 h after infusion (4–6 h) Figure 2). Between 6 h and 24 h, the plasma tAs level increased from a mean 8.9 ng/ml at 6 h to a mean of 15.26 ng/ml at 24 h. tAs values fluctuated between patients to a greater degree between 6 and 24 h as compared to the first 6 h after infusion (Figure 2).

Figure 2.

Kinetics of ATO metabolism: tAs concentration (ng As/mL) in plasma of all patients during the first 24 h and at day 4 of the treatment. Each line represents a single patient.

During the acute phase (1–24 h), iAs were the main arsenic species in the plasma, accounting for a mean 68.1% of tAs (IQR: 52.0%, 91.6%). Interestingly, when each patient’s concentrations of iAs were graphed over time, it became evident that there were two distinct groups of patients that differed in the kinetics of ATO metabolism between 6 and 24 h: Group 1 (n = 13), the ‘slow’ metabolizers in which the iAs concentrations in plasma increased over time, and Group 2 (n = 12), the ‘fast’ metabolizers which quickly eliminated iAs from plasma (Figure 3(a)). During this time interval, the average change in plasma iAs in Group 1 was +9.6 ng/mL as compared to −4.25 ng/mL in Group 2. At day 4, Group 1 iAs level was 11.4 ng/mL compared to 8.1 ng/mL in Group 2.

Figure 3.

Kinetics of ATO metabolism: (A) iAs, (B) MAs, and (C) DMAs concentration (ng As/mL) in plasma of patients in Group 1 and 2 during the first 24 h and at day 4 of the treatment. Each line represents a single patient. Panel A inset shows the change in serum iAs concentration between 6 and 24 h by the group.

Next, we compared the plasma concentrations of MAs and DMAs, the metabolites of iAs (Figure 3(b,c)). Between 6 and 24 h, the concentrations of MAs and DMAs increased or remained unchanged in most plasma samples; however, values varied significantly among patients and there was no clear difference in MAs and DMAs kinetics between Group 1 and 2.

Notably, the proportions of plasma tAs represented by arsenic species (%iAs, %MAs and %DMAs) changed between 6 and 24 h and the direction of these changes differed between Group 1 and Group 2 (Figure 4). In Group 1, %iAs increased and %MAs and %DMAs decreased while in Group 2, %iAs decreased and %MAs and %DMAs increased, providing additional evidence of a more efficient iAs metabolism (i.e. conversion of iAs to DMAs) by patients in Group 2. The differences between the groups in Δ%iAs (p = 4.7 × 10–7), Δ%MAs (p = 7.5 × 10–5 and Δ%DMAs (p = 1.1 × 10–7) were highly significant. In addition, we calculated the Area Under the Curve (AUC) for concentrations of each of the plasma arsenic species between 6 and 24 h and over the first 4 d of the treatment. We found striking differences in the AUC for iAs, MAs, and DMAs for the 6–24 h interval, and these differences were statistically significant (iAs mean difference was −72.6, p-value .012, MAs mean difference −113.7, p-value .0005, and DMAs mean difference −118.5, p-value .004). AUC also varied for each metabolite over the full 4 d period, though to a lesser significance than for the 6–24 h: iAs mean difference of 385.4, p-value .058, MAs mean difference −183.8, p-value .08, and DMAs mean difference −53.8, p-value .67.

Figure 4.

Change in the proportions (%) of as species in plasma of patients differs by Group 1 and 2 between 6 and 24 h after initial ATO infusion.

When we analyzed the baseline characteristics of these two groups, the major differences between Group 1 and Group 2 were the patients’ treatment regimen (ATO/ATRA alone vs. ATO/ATRA and prior chemotherapy; p = .01) and smoking status (current vs. former/never smokers; p = .03) (Tables 1 and 2). Prior chemotherapy was seen in the faster metabolizers, Group 2 patients (7 patients in Group 2 vs. 1 patient in Group 1) whereas 92.3% of the slow metabolizers, Group 1 patients, had ATO/ATRA alone (Table 2). Smokers were mainly distributed in the slow metabolizers, Group 1 patients (7 patients in Group 1 vs. 1 patient in Group 2) and most former/never smokers were faster metabolizers, Group 2 patients (84.6%) (Table 1). No other variables significantly differed by metabolism group.

Table 2.

Clinical variables by metabolism group.

Characteristic (%)	All patients N = 23	Group 1: slow metabolizers N = 13 (52% )	Group 2: fast metabolizers N = 12 (48%)	p-value
Treatment regimen
ATO/ATRA alone	17 (68.0%)	12 (92.3%)	5 (41.7%)	.01
ATO/ATRA + prior chemotherapy*	8 (32.0%)	1 (7.7%)	7 (58.3%)
Comorbidities **
Cardiovascular	11 (44.0%)	5 (38.5%)	6 (50.0%)	.56
Endocrine	6 (24.0%)	3 (23.1%)	3 (25.0%)	.91
Pulmonary	2 (8.0%)	1 (7.7%)	1 (8.3%)	.95
Gastrointestinal	8 (32%)	3 (23.1%)	5 (41.7%)	.32
Presenting liver transaminitis gradea
None	5 (20.0%)	2 (15.4%)	3 (25.0%)	.81
Grades 1 and 2	16 (64.0%)	9 (69.2%)	7 (58.3%)
Grades 3 and 4	4 (16.0%)	2 (15.4%)	2 (16.7%)
Renal dysfunction
Yes	3 (12.0%)	1 (7.7%)	2 (16.7%)	.49
No	22 (88.0%)	12 (92.3%)	10 (83.3%)
Concurrent QTc prolonging medications
Antidepressants	6 (24.0%)	3 (23.1%)	3 (25.0%)	.91
Antibacterials	15 (60.0%)	10 (76.9%)	5 (41.7%)	.07
Antiemetics	10 (40.0%)	6 (46.2%)	4 (33.3%)	.51
Proton pump inhibitors	7 (28.0%)	4 (30.8%)	3 (25.0%)	.75

^*Chemo refers to Daunorubicin + Cytarabine given during induction before treatment with Arsenic trioxide in the c9710 treatment protocol.

^**Cardiac complications refer to patients with hypertension/myocardial; Endocrine complications refer to patients with type 2 diabetes, hypothyroidism, diabetic nephropathy, and retinopathy; Pulmonary complications refer to patients with asthma, PE infarction/CAD/hypercholesterolemia. Percentages do not add to 100% because the conditions are not mutually exclusive.

^°t-test was conducted for continuous variables. Fisher’s exact test or Chi-square test was conducted for categorical variables.

^aclassified according to the highest recorded grading; grading was conducted according to version CTCAE criteria version 5 (48).

Values in bold are statistically significant p values, namely less than 0.05

Further analysis focused on ATO metabolism during the repetitive phase. However, the differences in the timing of ATO treatment during this phase (daily vs. 5 d per week), made a direct comparison between Group 1 and Group 2 impossible. In general, plasma concentrations of iAs varied substantially in the repetitive phase, but all patients returned to background levels of plasma iAs at 30 d post-treatment. As expected, plasma MAs and DMAs concentrations were elevated in the repetitive phase in all patients, but on average were notably higher in the slow metabolizers (Group 1). Data on iAs metabolism during the repetitive and post-treatment phases are provided in Supplemental Figure S1 but here the patients were grouped by treatment regimens (ATRA/ATO alone vs. ATRA/ATO with prior chemotherapy), which differed in the timing of ATO treatment.

Metabolites of ATO in urine and exfoliated urothelial cells

The spot and 24-h urine samples collected on days 1 and 4 were analyzed for iAs, MAs, and DMAs. In the 24-h urines, tAs, iAs, MAs, and DMAs concentrations on day 1 were higher in Group 2 as compared to Group 1, but this difference failed to reach statistical significance (p = .14, p = .11, p = .18, and p = .14, respectively). On day 4, the elevation in Group 2 became statistically significant in all except the MAs concentrations (Figure 5). This finding is consistent with the observation that the Group 2 patients are ‘fast metabolizers,’ and are thus expected to have more iAs and metabolites in the urine. The same differences in the concentrations of iAs and its metabolites were noted in spot urine samples, with Group 2 concentrations higher than Group 1 on both days 1 and 4, but statistically significant day 4 alone (Supplemental Figure S2). At 30 d after last ATO infusion, urine concentrations of arsenic metabolites were still an order of magnitudes higher than levels typical for unexposed populations [22] with mean values of 371.6 ng/mL, 163.9 ng/mL, 50.5 ng/mL, and 165.1 ng/mL for tAs, iAs, Mas, and DMAs, respectively.

Figure 5.

The concentrations of tAs, iAs, MAs, and DMAs in 24-h urine collected from patients differ significantly between Group 1 and 2 on day 4 of the treatment, but not day 1. P-values indicate a t-test comparison of groups 1 and group 2.

Lastly, concentrations of iAs, MAs, and DMAs were measured in exfoliated urothelial cells from spot urine and 24-h urine samples. Arsenic species in exfoliated urothelial cells provide an estimate of the internal doses of iAs and its metabolites. In addition, bladder epithelium is one of the tissues targeted by iAs exposure, and chronic iAs exposure is associated with bladder cancer [23–25]. The concentrations of iAs, MAs, and DMAs in exfoliated cells from Group 2 patients were higher compared to Group 1 patients, but this difference was not statistically significant (Figures S3 and S4). Thirty days after the last ATO infusion, mean arsenic metabolite concentrations in exfoliated urothelial cells were 208.8 pg/105 cells, 148.5 pg/105 cells, 36.4 pg/105 cells, and 24.0 pg/105 cells for tAs, iAs, Mas, and DMAs, respectively. These values are within the range of the concentrations reported in exfoliated urothelial cells from individuals chronically exposed to iAs in drinking water [23].

Discussion

Frontline ATO treatment is the standard of care in APL patients. However, chronic exposure to iAs (mainly studied in populations exposed via contaminated water) is associated with skin, bladder, lung, and liver cancer as well as with cardiovascular disease, diabetes, and developmental disorders [26,27]. iAs metabolism in chronic environmental exposures has been extensively studied [28–30], but scarce data exist on therapeutic ATO metabolism, and on potential long-term toxicities.

ATO in APL patients yields the same metabolites found in urine and tissues of individuals chronically exposed to environmental iAs. For example, 95% of tAs in the urine of APL patients treated with ATO consists of iAs, MAs, and DMAs in proportions similar to those commonly found in the urine of individuals exposed to iAs in drinking water [31]. The same arsenic species were also found in patients’ saliva [32] plasma [33–37] cerebrospinal fluid [33] bone marrow [37], and hair and nails [38]. With one exception, all these studies provided only static measurements and enrolled a small number of patients. Our study is the first to describe tAs in a relatively large number of patients during the first 4 d of ATO treatment, while also providing data on ATO metabolites 30 d after the end of treatment.

In patients with therapeutically dosed ATO, we identified 2 groups of arsenic metabolizers based on observed plasma iAs concentrations, %iAs, %MAs and %DMAs and urinary tAs between 6 and 24 h after initial infusion: Group 1, slowly metabolized iAs to MAs and DMA (had high iAs concentrations and percentage in plasma and high urinary tAs), and Group 2, which converted iAs to MAs and DMA faster and excreted more tAs in urine. The liver is the main site for iAs methylation, but other tissues have also been shown to retain iAs in laboratory models [39]. The increase in iAs in the plasma levels of Group 1 patients between 6 and 24 h suggests that a portion of iAs was stored in the liver and other tissues after ATO infusion, then was subsequently released unmetabolized (i.e. in the form of iAs) in the plasma.

Multiple factors are known to affect the efficiency of iAs metabolism. Polymorphisms in the gene encoding AS3MT, the key enzyme in iAs methylation, is one factor [40,41]. Several studies published in the past two years have shown that polymorphisms in AS3MT can affect ATO metabolism and be predictive of hyperleukocytosis and of liver dysfunction during ATO treatment [42,43].

In the present study, we could not obtain DNA for AS3MT genotyping. However, we evaluated associations of the ATO kinetic pattern between 6 and 24 h with other factors. When comparing baseline demographic characteristics and clinical variables, we identified two factors statistically significantly associated with differences in metabolism: (a) treatment regimen (ATO with ATRA alone or ATRA plus prior chemotherapy) and (b) smoking (active vs. former/never smoker). The chemotherapy regimen C9710 given during induction consisted of Daunorubicin and Cytarabine [4]. Both chemotherapeutic agents are hepatically metabolized. Daunorubicin is an anthracycline antibiotic converted in the liver to its active metabolite, Daunorubicinol [44]. Cytarabine is a pyrimidine analog, hepatically metabolized by deoxycytidine kinase and other nucleotide kinases to its active metabolite, Aracytidine triphosphate [45]. We observed no significant differences in hepatic function between Group 1 and Group 2 patients. Most patients had grade 1–2 transaminitis (69.2% of Group 1 patients and 53.8% of Group 2 patients, p-value .85) which resolved by the end of hospitalization (Table 2). Prior exposure to chemotherapy (Daunorubicin and Cytarabine) during induction, followed by ATO during the consolidation, was preferentially noted in the faster metabolizers, Group 2. This is an interesting observation, and to our knowledge, there is no published literature comparing the metabolism of ATO in patients receiving different regimens.

Given Daunorubicin, Cytarabine and ATO are all metabolized in the liver, one possible explanation for the differences in ATO kinetics between Group 1 and 2 is that the prior chemotherapy exposure in patients treated following C9710 had an effect on hepatic drug transporter expression in Group 2 patients. For instance, it is known that ATP-binding cassette (ABC) transporters translocate chemotherapeutics across hepatocyte membranes, and the expression of ABC transporters in patients is variable [46]. Some of the same transporters are involved in iAs metabolism in human hepatocytes [47]. Thus, induction of the hepatic ATP-cassette transporters as a result of chemotherapy could explain the faster metabolism and clearance of ATO in Group 2.

In addition, the chemotherapy patient group (following C9710) had arsenic exposure during the consolidation phase of the treatment, after patients had already achieved morphological remission, and were not taking concurrent ATRA. In contrast to that, the non-chemotherapy patients did receive concurrent ATRA and Arsenic, and those patients were newly diagnosed APL patients. Thus, concurrent ATRA and disease status may also play a role in influencing iAs kinetics, and larger studies should be undertaken in order to perform multivariate analyses.

As more patients nowadays receive non-chemotherapy regimens for APL, it is important to monitor long-term arsenic toxicities. The methylation of iAs to DMAs is generally considered a detoxification pathway, and numerous studies have shown patients with more efficient iAs methylation (lower urinary %DMAs or DMAs/MAs ratio) had decreased risk of skin lesions, bladder cancer, and atherosclerotic disease [28–30]. In contrast, inefficient iAs methylation has been associated with a higher risk of these diseases. The same paradigm may be applicable to patients receiving ATO.

The smoking status also emerged as a factor in ATO kinetics between 6 and 24 h after infusion (p = .03), with smokers mainly distributed in the slower metabolizer Group 1. This most likely represents a confounding variable, and given the size of our study, we could not control for smoking as an independent variable. Nevertheless, the role of smoking in ATO metabolism and toxicities should be further investigated.

In summary, the present study is one of the very few clinical studies of ATO metabolism published to date. Results show that ATO metabolism may vary among APL patients, and that treatment regimens and smoking may be at least partly responsible for this variation. The small number of patients involved in this study and relatively few time points, during which the samples were collected after the first day of treatment represent significant limitations. In addition, the implications of the key findings for ATO treatment strategies remain unclear. Prospective, multi-institutional trials are needed to validate our findings and to identify other independent variables in ATO metabolism. These trials should also determine whether the rate of ATO detoxification affects the efficacy of the therapy and short- or long-term deleterious effects associated with this therapy.