Adaptive immunity cooperates with liposomal all-trans-retinoic acid (ATRA) to facilitate long-term molecular remissions in mice with acute promyelocytic leukemia

Abstract

We previously developed a murine model of acute promyelocytic leukemia (APL) by using human cathepsin G gene regulatory elements to direct the expression of promyelocytic leukemia (PML)/retinoic acid receptor α (RARα) and RARα/PML fusion cDNAs to the early myeloid compartment of transgenic mice. To study the efficacy of noncytotoxic therapy in this animal model, cohorts of naive immunocompetent mice were inoculated with primary murine APL cells from a frozen tumor bank. Arsenic trioxide and liposomally encapsulated all-trans-retinoic acid (Lipo ATRA), alone or in combination, were administered for 21 days by i.p. injection using doses that yielded plasma levels similar to those observed in human APL patients treated with these agents. Lipo ATRA was highly effective in inducing durable molecular remissions in immunocompetent mice [C57BL/6 × C3H F₁ (B6C3HF1)]; arsenic therapy was much less effective, and did not clearly synergize with Lipo ATRA to increase the remission rate in immunocompetent mice. The survival of Lipo ATRA-treated severe combined immunodeficient (SCID) animals (lacking functional T and B cells) was inferior to that of immunocompetent B6C3HF1 recipients (40% vs. 88% survival at 1 y, P < 0.001). These data suggest that adaptive immunity cooperates with pharmacologic therapy to induce or maintain remissions in murine APL. It also implies that immunosuppressive anti-leukemia therapies could paradoxically blunt effective anti-leukemia immune responses that are important for clearing small numbers of residual tumor cells after chemotherapy-mediated cytoreduction.

Acute promyelocytic leukemia (APL) is usually associated with a t(15;17)(q22;q11.2-12) balanced translocation that creates in-frame fusions between the gene encoding the promyelocytic leukemia (PML) gene on chromosome 15 and retinoic acid receptor α (RARα) gene on chromosome 17 (1). We and others have demonstrated that the expression of PML/RARα in the early myeloid compartment of transgenic mice is necessary but not sufficient for the development of APL (2–4). We have also shown that coexpression of a reciprocal RARα/PML cDNA together with PML/RARα substantially increases the likelihood of APL development (5, 6).

The incorporation of all-trans-retinoic acid (ATRA) into the standard treatment regimens for APL has rendered it the most curable of the AML subtypes. Despite remission rates of up to 90% with oral ATRA alone, this therapy usually does not generate molecular remissions, and must be combined with cytotoxic chemotherapy to effect cures (7). However, recent studies have shown that liposomally encapsulated ATRA (Lipo ATRA), when given to patients as a single agent (“monotherapy”), can directly induce molecular remissions in APL patients (8–10). The role of chemotherapy to consolidate remissions induced by Lipo ATRA is not yet clear.

In this report, we evaluate the effectiveness of ATRA and arsenic trioxide in a well-characterized mouse model of APL. We define treatment protocols that yield drug levels of ATRA and arsenic that are very similar to those of human patients treated with these compounds, and show a striking difference between the efficacy of Lipo ATRA and arsenic. We developed an assay for minimal residual disease in the treated mice, and showed that the molecular remissions obtained with a single course of Lipo ATRA therapy were durable—without additional treatment. Finally, we demonstrated that Lipo ATRA does not require adaptive immunity to induce durable molecular remissions, but that the likelihood of this outcome more than doubles in immunocompetent mice. These results have potential implications for the design of therapies for APL patients.

Materials and Methods

Tumor Cryopreservation, Thawing, and Inoculation into Recipient Animals.

The incidence and phenotype of the murine APL tumors used in this model system have been described previously (5, 6). Briefly, these tumors arise between 6–12 mo of age in transgenic mice that coexpress the PML/RARα and RARα/PML cDNAs under the control of human cathepsin G regulatory elements, in a mixed B6C3H genetic background. Splenic tumors were harvested and cryopreserved as previously described (5). The tumors used in this study included nos. 9638 and 10552. APL cells were injected i.p. (in a total volume of 500 μl) into healthy, nonirradiated, 8- to 10-wk-old B6C3HF1 or C3H severe combined immunodeficient (SCID) recipients. Animals were monitored for evidence of leukemia development by their overall appearance (listlessness, hunched posture, and failure to thrive), and by serial complete blood count (CBC) determinations on peripheral blood obtained from the retroorbital plexus.

Treatment with ATRA and/or Arsenic Trioxide.

Twenty-one days after tumor inoculation, daily i.p. injections of arsenic trioxide (Sigma; ACS reagent grade no. A5081, 0.2 mg = 10 mg/kg), Lipo ATRA (Atragen; Aronex Pharmaceuticals, The Woodlands, TX; 0.2 mg = 10 mg/kg), or both drugs were administered in 500 μl of PBS for 21 consecutive days. Control animals received PBS alone. Animals treated with ATRA pellets (10 mg, 21-day sustained release pellets; Innovative Research of America) had the pellets implanted s.c. on day 21 after tumor inoculation, using methods recommended by the manufacturer. From day 21 to day 27 after tumor inoculation, dexamethasone (2.0 mg/kg per day) was administered to all groups of mice to prevent early treatment-related toxicity (i.e., retinoid syndrome). Animals were inspected twice a week, and blood was obtained from the retroorbital plexus at regular intervals to monitor for the development of leukemia.

ATRA and Arsenic Pharmacokinetics.

Healthy B6C3H F1, C3H SCID, or C57BL/6 recipients were treated for 7–14 consecutive days with Lipo ATRA (10 mg/kg per day = 0.2 mg total dose per day) administered by daily i.p. or i.v. injections. After 7 or 14 days, serial retroorbital plexus bleeds were performed, and 100-μl plasma samples were obtained and frozen at −20°C before analysis. Plasma ATRA levels were measured by using UV HPLC, as previously described (11, 12). For arsenic pharmacokinetics, four healthy C57BL/6 mice were treated with arsenic trioxide (10 mg/kg per day = 0.2 mg total dose per day) daily for 7 days, and serial retroorbital plexus blood samples were obtained and frozen at −20°C before analysis. Arsenic blood level determinations were performed by the California Veterinary Laboratory System (Davis, CA) by using a hydride inductively coupled plasma (ICP) assay.

The disposition of ATRA was evaluated in two groups of mice, where one group received 10 mg/kg i.p. Lipo ATRA alone and the second group received a combination of 10 mg/kg Lipo ATRA i.p. and 10 mg/kg arsenic trioxide i.p. daily for 7–14 days. Whole blood (100 μl) was obtained before drug administration on treatment day 7 or 14, and at 0.5, 1, 2.5, and 5 h after injection from each treated animal (four mice per single time point). Plasma was isolated for determining ATRA levels, and whole blood was used to quantitate arsenic concentrations. A one-compartment linear model with delayed absorption was used to characterize i.p. administration of Lipo ATRA, using maximum-likelihood estimation as implemented on adapt ii software (13). The area under the concentration-time curve (AUC_0→5h) for ATRA was calculated by integration of the concentration-time data from model estimates. Arsenic trioxide pharmacokinetics were determined by using noncompartmental methods to estimate systemic clearance, apparent volume of distribution, half-life, and AUC_0→4h.

PCR-Based Minimal Residual Disease Detection.

For detection of minimal residual disease, 50-μl peripheral blood samples were obtained by retroorbital plexus bleeding and incubated for 5 min on ice in erythrocyte lysis buffer (154 mM NH₄Cl/10 mM KHCO₃/0.1 mM EDTA, pH 7.4). Leukocytes were then pelleted, washed in PBS, and incubated in 50 μl of PBND buffer (50 mM KCl/10 mM Tris⋅HCl, pH 8.3/0.01% gelatin/2.5 mM MgCl₂/0.45% Triton X-100/0.45% Tween 20) for 2–6 h at 52°C. Cell lysates were subsequently stored at −20°C and thawed immediately before use. PCR standards were prepared by adding a known number of viable cells from a cryopreserved APL spleen specimen to samples of peripheral blood obtained from wild-type B6C3HF1 animals, which were then processed as described above. Immediately before use as PCR templates, DNA samples were subjected to a 5-min, 96°C preincubation step. PCR assays were performed by using 2 μl of cell lysates per reaction in PBND buffer with 0.2 mM dNTPs, 0.2 μM oligonucleotides, and 1 unit of TaqDNA polymerase (Sigma). For detection of transgene DNA, forward and reverse primers corresponding to human CG genomic sequence 429–448 (GGCCTGACCTCATCCCATAG) and 950–931 (GCCCTTTTCCCCATCCTAGG), respectively, were used to generate a 522-bp product (2). As a control for DNA quality and content, forward and reverse primers corresponding to the endogenous murine granulocyte-colony stimulating factor (G-CSF) receptor genomic sequence (CTGCTCTTGACATAAGCCTG and TAGCAGCACTCTTCAATGGC) were used to generate a 350-bp product. PCR products were subjected to 1.5% agarose gel electrophoresis, transferred to nylon membranes, and hybridized with an hCG fragment (bp 429–931) radiolabeled by random hexamer-priming. PCR products were quantitated by phosphorimaging. PCR reaction conditions for both primer sets were as follows: 94°C for 2 min (once only), followed by 35 cycles consisting of 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s.

Results

Natural History of Animals Receiving Adoptively Transferred APL Cells.

Immunophenotypic characterization of the two tumors used in this study is shown in Table 1. The tumors both had typical features of murine APL, including a large population of abnormal Gr-1/CD34 double positive cells in the spleens (5, 6); both tumors displayed both MHC Class I and Class II on the CD34⁺ cells (Table 1). Because the tumors arose in B6C3H mice (2, 5), we used B6C3HF1 mice as recipients to eliminate the possibility that tumors would be rejected on the basis of allogeneic differences. We determined that an i.p. dose of 300,000 APL spleen cells from either tumor caused fatal APL (characterized by leukocytosis, anemia, thrombocytopenia, and splenomegaly) in 100% of nonirradiated recipient mice. The leukemias that developed in the recipient animals were morphologically indistinguishable from those that arose in the transgenic animals from which the tumors were derived (data not shown).

Table 1.

Immunophenotypes of APL tumors

Tumor no.	Gr.1+	CD34+	Gr.1+/CD34+	CD34+/Class I Kb+	CD34+/Class I Kk+	CD34+/Class II IAb+	CD34+/Class II IAk+
9,638	98	69	67	17.6	36.1	4.8	1.3
10,552	93	38	37	65.4	0	4.0	14.0

All numbers indicate the percentage of total spleen cells that stained for the indicated marker. 

The natural history of the transferred disease is shown in Figs. 1 and 2. Serial white blood counts from the animals receiving 300,000 APL spleen cells from tumor 10552 are shown in Fig. 1A. The mice had normal peripheral white blood cell (WBC) counts until day 65; after that, the white counts rapidly rose in all animals, and death from leukemia routinely occurred before day 80.

Figure 1.

Natural history of APL in untreated and Lipo ATRA-treated B6C3HF1 mice. Peripheral blood was obtained at 2- to 3-wk intervals over a 100-day period after tumor inoculation in B6C3HF1 mice treated with saline or 10 mg/kg per day of Lipo ATRA for 21 days (the treatment period is shown in the box in B and D). Blood samples were subjected to Coulter analysis for total WBC/μl (A and C) and PCR analysis of hCG-PML-RARa transgene DNA for detection of adoptively transferred, transgenic leukemia cells (B and D). Absolute leukemia cell numbers were estimated by comparing phosphorimaged band intensities with a standard curve generated with known standards (see Fig. 2).

Figure 2.

PCR detection of transgenic APL cell DNA in peripheral blood. PCR detection of human cathepsin G DNA (hCG) derived from the hCG-PML-RARα transgene was used for quantification of APL tumor cells in peripheral blood. (A) PCR products from tumor cell DNA (lane 1), no DNA (lane 2), and DNA from wild-type peripheral blood samples to which defined numbers of APL spleen cells (0–50,000 cells) were added are shown after hybridization with an hCG-specific probe. (B–E) hCG PCR products from blood samples obtained 10–88 days after tumor inoculation from representative untreated (B and C) and Lipo ATRA-treated (D and E) mice are shown. Ethidium bromide-stained, PCR-amplified murine granulocyte-colony stimulating factor (G-CSF) receptor DNA served as an internal control for DNA quality and content for each sample.

We developed a PCR-based assay to prove that the mice were developing disease from engraftment and expansion of the original transgenic tumors (Fig. 2A). PCR primers that specifically amplified human cathepsin G DNA sequences derived from the transgene were used to detect transgenic APL cells in the peripheral blood of recipient mice. A control assay, in which increasing numbers of transgenic APL cells were added to wild-type mouse peripheral blood, is shown in Fig. 2A. The PCR assay detects 5 APL cells in 50 μl of peripheral blood, which contains about 500,000 white cells (i.e., the sensitivity of the assay is 1 leukemic cell in 100,000 total white cells). As shown in Fig. 2 B and C, transgenic APL cells can be detected in the peripheral blood of two representative recipient mice on day 62 after tumor injection, and the signal rapidly increases before the death of both mice on day 75. These results are graphically depicted in Fig. 1B, where the PCR data (quantitated by phosphorimaging) is extrapolated to determine the total number of leukemic cells per μl of peripheral blood in five individual B6C3HF1 mice. The increase in the total white blood counts of the recipient animals just before death was due to rapid expansion and peripheralization of the transferred leukemic cells.

ATRA and Arsenic Pharmacokinetics.

To determine the effectiveness of ATRA for treating APL in this model system, we first investigated a number of drug delivery strategies. Because ATRA is insoluble in aqueous solution, we initially attempted to administer the drug i.p. in dilute ethanol or vegetable oil solutions, but we observed rapid precipitation of ATRA and fatal chemical peritonitis. We also investigated the use of s.c. implanted ATRA 21-day sustained release pellets, as described in previous reports (14, 15). ATRA pellets (10 mg) were implanted in B6C3HF1 animals, and random plasma ATRA levels were measured on days 1, 2, 3, 6, 8, 13, and 28 after implantation. The maximum ATRA level observed in any animal was 168 ng/ml. Average plasma ATRA levels ranged between 30–67 ng/ml between days 6 and 13, and were undetectable on day 28 (data not shown). Pellets were recovered from two animals on day 13 after implantation, and were found to contain ≈3 mg of residual ATRA, suggesting that the delivery of the drug indeed occurred over 21 days, as suggested by the manufacturer.

Because plasma ATRA levels of 350 ng/ml are routinely observed in human APL patients treated with standard doses of oral ATRA (16) (i.e., ≈5–10 times higher than what the 10 mg ATRA pellet achieved), we investigated the use of a liposomally encapsulated form of ATRA (Atragen; Aronex Pharmaceuticals) to achieve higher drug levels in mice. Mean peak ATRA levels observed in human APL patients treated with Atragen (90 mg/m² i.v. every other day) have been reported to be 3,300 ng/ml (17). Pharmacokinetic studies were performed after 7 consecutive days of Lipo ATRA treatment (10 mg/kg per day i.p. with or without concurrently administered arsenic trioxide at 10 mg/kg per day i.p.; Fig. 3A). The ATRA concentration-time curve was well described by a one-compartment model. The volume of central distribution (Vc; volume of central compartment) was 2.5 ± 0.14 liters/kg with either ATRA alone or combined with arsenic. The mean peak ATRA concentration was 3,240 ng/ml in the ATRA alone group and 3,073 ng/ml in the combination group. Systemic clearance of ATRA was 1.02–1.16 liters/h per kg, and the half-life was 2.2–2.5 h. The AUC for ATRA was 8,590 ng/ml⋅h [95% confidence interval (C.I.) 6,300–9,300 ng/ml⋅h] in ATRA alone group and 9,770 ng/ml⋅h (95% C.I. 7,700–10,000 ng/ml⋅hr) for the combination group. Therefore, there appeared to be no direct interaction between ATRA and arsenic trioxide in mice. This result is not surprising, because ATRA is known to be oxidized in the liver via the cytochrome P450 system, whereas arsenic trioxide undergoes methylation and is eliminated by renal excretion (18). No significant pharmacokinetic differences in drug levels were observed after 7 vs. 14 days of ATRA administration, with the i.v. vs. i.p. routes of administration, or with treatment of C3H SCID animals treated with i.p. Lipo ATRA (data not shown).

Figure 3.

Pharmacokinetics of ATRA and arsenic trioxide clearance. (A) Plasma ATRA concentrations measured over a 5-h period after i.p. administration of 0.2 mg (10 mg/kg) of liposomal ATRA to a cohort of four C57BL/6 mice. Virtually identical results were obtained with concurrent arsenic trioxide (10 mg/kg) administration (data not shown). (B) Plasma arsenic levels were measured over a 4-h period after i.p. administration of 0.2 mg (10 mg/kg) of arsenic trioxide to a cohort of four B6C3HF1 animals.

C3H SCID and C57BL/6 animals were treated with arsenic trioxide doses of 2–20 mg/kg per day i.p.; at the highest dose, 100% of the animals died within 24 h. In contrast, all animals treated with 10 mg/kg per day of arsenic trioxide for 21 consecutive days survived 6 mo without weight loss or clinical illness. To study arsenic pharmacokinetics in mice, whole blood samples were obtained after arsenic administration (10 mg/kg per day) to C57BL/6 mice for 7 consecutive days. Arsenic levels obtained on day 7 displayed a mono-exponential decay (Fig. 3B). Peak arsenic concentrations ranged from 1.6 to 4 μg/ml. The apparent volume of distribution was 3.3 liters/kg, and systemic clearance was 0.89 liters/h per kg. Arsenic half-life was 3.7 h and the AUC was 7.84 μg/ml⋅h. Whole blood arsenic disposition was similar to that observed in humans, with apparent linear elimination (19, 20). However, the peak concentrations and AUC in mice receiving arsenic trioxide 10 mg/kg (≈2.1 μg/ml and 7.84 μg/ml⋅h, respectively) were higher than that observed in previous human studies from our center (<0.05–0.13 μg/ml and 0.91–1.37 μg/ml⋅h, respectively) and China (0.42–0.56 μg/ml and 2.1–3.35 μg/ml⋅h, respectively) (19, 20). Mice therefore appear to have a shorter half-life for arsenic (3.4 h) than humans (≈6.8 h) (19).

Treatment of Immunocompetent APL Recipients with ATRA and/or Arsenic Trioxide.

Murine APL cells were adoptively transferred to B6C3HF1 recipients by i.p. inoculation of 300,000 cryopreserved APL spleen cells from tumor nos. 9638 or 10552. After a 3-wk period of observation (to allow for engraftment of leukemia-initiating cells), animals were treated with daily i.p. injections of Lipo ATRA, arsenic trioxide, or both agents simultaneously for 3 wk. In addition, a cohort of animals underwent s.c. implantation of a 21-day sustained release ATRA pellet (10 mg total dose). Control animals received daily saline injections. Leukemia-free survival probabilities are plotted in Fig. 4. Lipo ATRA treatment resulted in 88% leukemia-free survival at 1 yr (Fig. 4A). Although the sustained release ATRA pellet delivery method resulted in a significant prolongation of survival compared with saline-treated controls, all animals died of APL within 230 days (Fig. 4A). Similarly, arsenic trioxide treatment resulted in significant survival prolongation compared with control saline-treated animals, but only 14% of arsenic-treated animals survived at 1 yr (Fig. 4B). The outcomes of animals treated with Lipo ATRA plus arsenic were not significantly different from animals treated with Lipo ATRA only (Fig. 4B).

Figure 4.

Survival of B6C3HF1 recipients after treatment. The survival probabilities of cohorts of B6C3HF1 mice inoculated i.p. on day 1 with 300,000 cells from murine APL tumors, and then treated on day 22 with ATRA pellets or liposomal ATRA (Lipo AT, A), arsenic trioxide (AS, B), arsenic trioxide plus liposomal ATRA (AS-Lipo AT, B), or saline (“Control” in both panels), in three independent experiments with two different tumors, are plotted against time. All groups received dexamethasone (2.0 mg/kg per day) during the first week of treatment. n, Number of mice treated in each cohort.

Serial WBC counts and PCR assays for minimal residual disease were performed in mice treated with Lipo ATRA to determine when the transferred cells entered the peripheral blood after engraftment, and to determine whether the mice achieved molecular remissions after therapy. The WBC counts of nine B6C3HF1 mice treated with 10 mg/kg Lipo ATRA i.p. daily from day 22–42 after tumor injection (denoted by the box labeled “ATRA”) are shown in Fig. 2C. The total WBC counts remained normal in all mice. Examples of serial PCR studies to detect transgenic APL cells in the peripheral blood are shown in Fig. 1 D and E; a graphic depiction of these studies is shown in Fig. 2D. APL cells could not be detected in the peripheral blood of any mouse in this treated cohort. Because the APL cells that were injected into the untreated and treated mice always came from the same frozen aliquots, and were administered at the same time, it is very unlikely that these results could be explained by a random failure of the tumors to engraft only in the treated cohort.

Reduced Efficacy of Lipo ATRA in SCID Recipients.

To determine whether adaptive immunity is relevant for responses to ATRA and/or arsenic, we transferred identical doses of APL cells into B6C3HF1 and C3H SCID recipients, which were subsequently treated with Lipo ATRA, arsenic trioxide, or both agents. As shown in Fig. 5A, survival of Lipo ATRA-treated SCID recipients was significantly inferior to that of B6C3HF1 recipients (40% vs. 88% leukemia-free survival at 1 yr, P < 0.001; the data from wild-type mice are replotted from Fig. 4 for ease of comparison). Peripheral blood samples obtained 9 mo after Lipo ATRA treatment of either B6C3HF1 or C3H SCID recipients demonstrated no evidence of leukemia by PCR analysis (examples are shown in Fig. 5E).

Figure 5.

Survival of B6C3HF1 vs. C3H SCID recipients after treatment. The cumulative probability of survival for cohorts of immunocompetent B6C3HF1 (WT) and immunodeficient (SCID) animals injected on day 1 with 300,000 APL cells, and treated from day 22–42 with Lipo ATRA (Lipo AT, A), arsenic trioxide (AS, B), arsenic trioxide plus liposomal ATRA (C), and 10 mg sustained release ATRA pellets (D) are plotted against time. Those data for the wild-type cohort are replotted from Fig. 4 for ease of comparison. Data from two different tumors used in three independent experimental cohorts are plotted. For comparison, the survival of the same group of saline-treated control mice is also plotted in each panel. All groups received dexamethasone (2.0 mg/kg per day) during the first week of treatment. In E, the absence of PCR-detectable tumor cells in peripheral blood 9 mo after liposomal ATRA treatment is demonstrated. The PCR assay was performed as described in Materials and Methods. The control samples were concurrently performed to calibrate the sensitivity of the assay.

The survival of arsenic-treated (Fig. 5B) or pellet ATRA-treated mice (Fig. 5D) did not differ significantly between C3H SCID and B6C3HF1 mice. The survival of SCID mice treated with both arsenic and Lipo ATRA was less than that of B6C3HF1 recipients (56% vs. 80% at 1 yr, Fig. 5C); this difference was not statistically significant (P = 0.0765). The survival of SCID recipients treated with Lipo ATRA alone (40% at 1 yr) was less than that of animals treated with both liposomal ATRA and arsenic (56% at 1 yr), but this difference was not statistically significant (P = 0.45).

Discussion

We have used an adoptive transfer model of murine APL to investigate the efficacy of ATRA and arsenic trioxide in immunocompetent and immunodeficient mice. We observed that ATRA administration using a sustained release pellet to provide continuous low levels of drug was marginally effective at prolonging the survival of recipient mice. In contrast, daily i.p. injection of Lipo ATRA for 21 days resulted in much higher levels of ATRA in mice, and produced long term molecular remissions in 88% of immunocompetent animals. Only 40% of identically treated SCID animals achieved long-term molecular remissions. Arsenic trioxide produced short-term remissions that prolonged survival in most immunocompetent animals, but only ≈15% achieved long-term remissions. Arsenic did not increase the effectiveness of liposomal ATRA in either immunocompetent mice or SCID mice.

Our experience with ATRA therapy of murine APL is similar to that of previous reports. Lallemand-Breitenbach et al. (14), using the identical 21-day ATRA pellets, found that survival was prolonged with ATRA therapy in their APL mice, but that no long-term remissions occurred. Similarly, Kogan et al. (15), using 5-mg ATRA pellets, and Rego et al. (21), using 1.5 mg/kg of i.p. ATRA daily (in a dissolved form), reported short term remissions but no cures in their murine APL model systems. Therefore, the durable molecular remissions reported here with Lipo ATRA are distinct from the nondurable remissions previously reported. Why is treatment with Lipo ATRA superior to that of sustained released ATRA pellets in our mouse model of APL? The analysis of ATRA levels provided by these two different delivery systems probably provides a first approximation of the answer. Even though the total dose of Lipo ATRA delivered over 21 days (4.2 mg) is less than that of the sustained release pellet (10 mg), both the peak levels and the area under the concentration time curve (AUC) observed for Lipo ATRA were considerably higher than that observed with pellet ATRA. In addition, the formulation of Lipo ATRA is different from that of conventional ATRA, and the delivery of Lipo ATRA to critical cellular compartments could potentially be a factor that contributes to its efficacy.

Our results in mice treated with Lipo ATRA are similar to those observed in human APL patients. Oral ATRA therapy yields very high remission rates in APL patients, but these remissions are rarely confirmed by sensitive molecular assays, and they are seldom durable (7). In contrast, monotherapy with Atragen (Lipo ATRA) in newly diagnosed APL patients has yielded complete remission rates of 79–87% (9, 10). Ninety-two percent of patients achieving complete remission achieved molecular complete remission after 3 mo of therapy, and many of these remissions have been durable without additional treatment. The reason for this difference in the efficacy of the two formulations is not yet known, but it is reasonably clear that patients treated with Lipo ATRA do not experience the decline in ATRA levels encountered with oral ATRA over time (17), and they therefore have higher drug levels for longer periods of time.

Arsenic monotherapy for APL in our mouse models extended the survival of mice significantly, but it did not frequently yield cures. The reason for the low cure rate is not low levels of the drug in mice; the levels achieved with our dosing regimen were actually higher than those achieved in APL patients treated with standard doses of arsenic trioxide. Our experience with arsenic monotherapy is very similar to that previously reported by Rego et al. (21), who administered 5 mg/kg of arsenic trioxide per day for 21 days in their model system, and that of Lallemand-Breitenbach et al. (14), who delivered a dose of 5 mg/kg per day for 28 days; pharmacokinetic studies were not performed in either of these reports. In both studies, arsenic monotherapy prolonged survival, but resulted in few long-term remissions. The durability of remissions obtained in relapsed APL patients who have received arsenic as salvage therapy is not yet completely known (19, 20, 22, 23). In our experience, remissions can be obtained in most relapsed patients, but they are not durable (20); the remission rates of newly diagnosed APL patients treated with arsenic monotherapy have not yet been reported. The results obtained with the mouse model, however, predict that initial arsenic monotherapy will be less effective than Lipo ATRA monotherapy in newly diagnosed APL patients.

Lipo ATRA and arsenic did not synergize to produce higher remission rates in immunocompetent mice, but the combination slightly improved the long-term survival of SCID mice. In two previous reports exploring the synergy of arsenic and ATRA (14, 21), arsenic was combined with either ATRA pellet therapy (the dose and manufacturer were identical to that used in this report), or retinoic acid was delivered as a once daily i.p. dose of 1.5 mg/kg for 21 days; pharmacokinetic studies were reported for neither. None of the ATRA only-treated animals achieved long-term remissions in the other studies, but arsenic clearly synergized with ATRA to produce longer lasting remissions. In the study of Rego et al. (21), all of the animals treated with arsenic and ATRA eventually died whereas, in the study of Lallemand-Breitenbach et al. (14), the doubly treated animals survived for at least 9 mo, when the study was stopped. In our study, where Lipo ATRA produced long term molecular remissions in 88% of immunocompetent mice, the addition of arsenic clearly did not further improve outcome; clearly, it would be more difficult to detect synergy in this model system, because Lipo ATRA monotherapy is so effective. In SCID mice, where the long-term survival rate was only 40% with Lipo ATRA, the combination led to a modest (but statistically insignificant) improvement in survival. These results suggest that arsenic may not improve Lipo ATRA induction therapy in patients with de novo APL, despite the fact that these two drugs act via different mechanisms to eliminate APL cells (14, 21).

Although SCID mice could achieve durable molecular remissions with Lipo ATRA monotherapy, this outcome was observed only half as often as in immunocompetent mice. SCID animals have markedly impaired adaptive immunity, and lack the capacity to generate functional B and T cell responses, despite retaining natural killer cell function (24). These results suggest that Lipo ATRA can reduce the APL tumor burden in SCID mice to levels that do not require T and B cells for curative tumor clearance. However, our data also suggest that the B and T cells of naive immunocompetent mice must be capable of mounting effective tumor-specific responses against APL cells, and that these responses facilitate durable remissions in a large proportion of mice. This result is supported conceptionally by additional studies in our laboratory (J.L.P., P.W., M.J.W., K.M.O., K. Schrimpf, and T.J.L., unpublished results); immunocompetent B6C3HF1 mice reject APL cells approximately 100 times more efficiently than C3H SCID mice. This difference is clearly due to the SCID mutation and not the C3H background, because wild-type C57BL/6 and C3H animals are equally capable of rejecting lethal doses of the same APL samples, based on allogeneic recognition of the tumor cells (J.L.P., M.J.W., and T.J.L., unpublished results).

Because the APL tumors used in these studies arose in a mixed B6C3H background, these tumors must be recognized in B6C3HF1 mice by virtue of tumor-specific antigens. The APL tumors generated in our mouse model uniformly express MHC Class I (and frequently express MHC Class II), and lethal doses of tumor cells are efficiently cleared when adoptively transferred into allogeneically incompatible mice (see above). These observations contradict the prediction that MHC Class I might be down-regulated by PML-RARα (25). Murine APL cells do not seem to be immunologically “cloaked,” nor are they resistant to clearance in vivo. Similarly, limited studies of human APL cells have revealed that nearly all express Class I, and that a significant proportion express Class II (26–28). It is not yet known whether human APL cells can elicit effective immune responses in vivo, but it is reasonably clear that APL patients do not contain cytotoxic T lymphocytes (CTL) that recognize unique high affinity peptides from the PML-RARα junction region (29–31). The role of adaptive immunity for the clearance of APL cells in humans is therefore unknown, but this issue deserves careful study.

Both preclinical and clinical studies suggest that immune-mediated anti-tumor effects may be most effective in the minimal residual disease setting, consistent with our observations that immunocompetent mice cannot clear large tumor challenges like the ones used in this study. However, tumor-specific immunity—probably initiated by B and/or T cells—appeared to augment the anti-tumor effect of Lipo ATRA. In human APL patients (as well as in patients with other hematologic and nonhematologic malignancies), the standard approach for achieving minimal tumor burden involves the administration of cytotoxic chemotherapy, which is often highly immunosuppressive, and which may therefore blunt the development of effective anti-tumor immunity (7). However, recent reports of durable molecular remissions in patients treated with Lipo ATRA monotherapy—and the results reported here—provide support for APL therapies that tend to preserve the capacity of the immune system to mount effective anti-tumor responses.

Abbreviations

APL

acute promyelocytic leukemia

PML

promyelocytic leukemia

RARα

retinoic acid receptor α

ATRA

all-trans-retinoic acid

Lipo ATRA

liposomally encapsulated all-trans-retinoic acid

SCID

severe combined immunodeficient

AUC

area under the concentration-time curve

WBC

white blood cell