Determinants of cerebrospinal fluid arsenic concentration in patients with acute promyelocytic leukemia on oral arsenic trioxide therapy

Central nervous system (CNS) relapse of acute promyelocytic leukemia (APL) occurs in 1% to 5% of patients.^1,2  The most important risk for CNS relapse is a high leukocyte count (> 10 × 10⁹/L) at diagnosis.²  CNS APL fares poorly despite intrathecal/systemic chemotherapy and hematopoietic stem cell transplantation.²  The effective treatment/prevention of CNS APL remains unclear.

Arsenic trioxide (As₂O₃) is a standard treatment for relapsed APL.³  We first showed in an APL patient with CNS relapse that arsenic entered the cerebrospinal fluid (CSF) during oral As₂O₃ treatment,⁴  a finding later confirmed independently in another patient receiving intravenous As₂O₃.⁵  However, the factors controlling arsenic penetration into CSF are undefined.

We studied a series of APL patients receiving intrathecal chemotherapy while on oral As₂O₃, which gave us the unique opportunity of documenting arsenic penetration into CSF, its extent, and the factors controlling it.

Fifty consecutive APL patients with marrow relapse were referred. Initial diagnosis was made morphologically and cytogenetically and/or molecularly, and reaffirmed at relapse.

Oral As₂O₃ (10 mg/day)⁶  and idarubicin (6 mg/m² per day × 3) were administered until complete remission (CR), followed by idarubicin consolidation (6 mg/m² per day × 4)⁷  and maintenance (all-trans retinoic acid, 45 mg/m² per day; oral As₂O₃, 5-10 mg/day; 2 weeks every 2 months for 2 years). As₂O₃ treatment was approved by the institutional review board at Queen Mary Hospital and performed in accordance with the Declaration of Helsinki.

Twenty patients developed post-As₂O₃ relapses, 5 of whom (cases 1, 2, 6-8; Table 1) had isolated CNS relapse. Case 7 had been reported previously.⁴ 

Intrathecal chemotherapy comprising methotrexate (12 mg) and cytarabine (50 mg; 2 times/week) and oral As₂O₃ (5-10 mg/day) were administered until CSF was negative morphologically and molecularly for APL cells. Cranial irradiation was subsequently given in 2 patients (cases 6, 7).

Because of increased risks of CNS relapse in patients in complete remission 3 (CR3) and beyond, 3 patients (cases 3, 4, and 9) consented to CNS prophylaxis. One complete remission 2 (CR2) patient (case 5) with a high leukocyte count (32 × 10⁹/L) at relapse also consented to prophylaxis. Treatment comprised methotrexate (12 mg) weekly for 3 doses.

Elemental arsenic in paired CSF and plasma (obtained immediately after lumbar puncture) was measured by inductively coupled plasma mass spectroscopy.^4,8  Because most patients took the oral As₂O₃ at home, the CSF and plasma sampling could not be timed with oral As₂O₃ administration.

Statistical tests used were described with individual experiments. All analyses were performed with SPSS version 14 software (SPSS, Chicago, IL).

At a median follow-up of 14 months (range, 12-28 months), 3 of 5 patients with CNS relapse were in remission (Table 1). Three of 4 patients receiving intrathecal prophylaxis were in remission, after a median follow-up of 3 months (range, 2-14 months). None of the surviving patients had CNS sequelae from the leukemia or its treatment/prophylaxis.

Arsenic levels were determined in 67 paired CSF and plasma samples (median per patient, 6; range, 2-19). CSF arsenic ranged from 3.5 to 318.9 nmol/L (median, 95.8 nmol/L) and plasma arsenic from 36.3 to 1892.8 nmol/L (median, 498.9 nmol/L). Previous pharmacokinetic studies showed that, after 10 mg of oral As₂O₃, the peak plasma arsenic varied from 200 to 600 nmol/L.⁹  The plasma arsenic levels observed herein were within the expected range, with the high plasma levels reflecting peak, and low plasma levels reflecting trough, levels after oral As₂O₃ administration.

As a group, there was a strong linear correlation between plasma and CSF arsenic (Figure 1A, P < .001), with CSF at 17.7% the plasma level (slope of the regression line, Figure 1A). Analyses of sequential samples of individual patients showed similar correlations (Figure 1B,C, and Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article).

The CSF/plasma arsenic ratio was used as a measurement of the efficiency of arsenic penetration into CSF. Individual patients showed significant variations in their mean CSF/plasma ratios (Figure 1D).

To test whether chemotherapy damaged the meninges and increased arsenic penetration, the CSF/arsenic ratios of the first CSF sample of all patients were grouped, which provided the baseline efficiency of arsenic CSF penetration before chemotherapy. The CSF/plasma arsenic ratios of the second and subsequent CSF samples were similarly grouped. A comparison of the mean CSF/plasma ratios of sequential CSF samples showed no significant changes, implying that repeated intrathecal chemotherapy did not affect arsenic CSF penetration (Figure 1E).

To investigate whether meningeal leukemia might increase arsenic CSF penetration, the CSF/plasma ratios of CSF showing leukemic cells (n = 38) were compared with those without leukemic cells (n = 29). The results showed that meningeal leukemia did not affect CSF/plasma arsenic ratios (Figure 1F).

We documented that arsenic penetrated the CSF over a broad range of plasma arsenic levels. Previous studies showed that arsenic levels in peritoneal dialysates were identical to plasma,⁸  indicating that the peritoneum did not possess an arsenic barrier. On the contrary, the difference between plasma and CSF arsenic implied that the blood-brain barrier maintained an arsenic gradient. Expectedly, individual patients varied in the efficiency of arsenic CSF penetration. Regression analysis of the group showed that CSF arsenic was 17.7% of plasma. This fraction was not significantly altered with repeated intrathecal chemotherapy, an important reassurance that CNS arsenic toxicity would not be increased with concomitant intrathecal chemotherapy.

The median CSF arsenic was approximately 100 nmol/L, highest being more than 300 nmol/L. In primary APL cells, significant cytotoxicity starts with arsenic at 100 to 400 nmol/L, with approximately 50% cell death at 500 nmol/L.¹⁰  Because of variable CSF sampling time after oral As₂O₃, some patients had low plasma and hence CSF arsenic, reflecting sampling at trough levels. However, in at least half of the CSF samples in this study, arsenic was still present at therapeutically meaningful concentrations. Given that the peak plasma arsenic after intravenous As₂O₃ may be several-fold that of oral As₂O₃,⁹  even higher CSF arsenic levels may be reached with better therapeutic efficacy after intravenous As₂O₃, although this needs to be validated. However, the relative cardiac safety of oral As₂O₃ remains an important factor in considering the administration route.¹¹  Because CNS APL is unlikely to be treated with As₂O₃ solely without intrathecal chemotherapy, the issue of whether systemic As₂O₃ alone would be adequate for CNS disease will not be resolved.

None of the drugs at the dosages used in APL¹²  enters the CSF at significant levels, explaining why the CNS may be a leukemia sanctuary site. Patients at high risk for CNS relapse may therefore need intrathecal prophylaxis. Findings here suggest that prophylactic As₂O₃ may be another option. Our routine maintenance during CR comprised oral As₂O₃ (5-10 mg/day) 2 weeks every 2 months for 2 years, which was well tolerated without untoward toxicities.¹³  Whether a more intensive arsenic maintenance might be beneficial for patients at high risk for CNS relapse will need to be validated.

Our results showing arsenic penetration into the CSF are of obvious significance in toxicology studies of environmental/accidental arsenic poisoning. Finally, we and others have shown that aquaglyceroporin 9 is involved in transmembrane trafficking of arsenic.^14,15  Its role in controlling arsenic CSF penetration awaits to be evaluated.

Oral arsenic trioxide was provided free to the patients by the S.K. Yee Medical Foundation. The authors thank Stanley Yeung, who performed the statistical analysis.

This work was supported by the Ruby and Minoo N. Master Charity Foundation.

Contribution: W.-Y.A. conceived the study, treated the patients, obtained the specimens, and wrote and approved the manuscript; S.T. and B.M.F. performed the arsenic analysis and approved the manuscript; and Y.-L.K. conceived the study, treated the patients, and wrote and approved the manuscript.

Conflict-of-interest disclosure: The University of Hong Kong holds a provisional patent on the use of oral arsenic trioxide in the treatment of leukemia. W.-Y.A., S.T., and Y.-L.K. are associated with or employed by the University of Hong Kong. B.M.F. declares no competing financial interests.

Correspondence: Yok-Lam Kwong, Department of Medicine, Professorial Block, Queen Mary Hospital, Pokfulam Road, Hong Kong; e-mail: ylkwong@hkucc.hku.hk.