Pyrazoloacridine Is Active in Multidrug-resistant Neuroblastoma Cell Lines with Nonfunctional p53¹ Free

Aggressive chemoradiotherapy followed by autologous BMT³ and 13-cis-retinoic acid improved survival of high-risk neuroblastoma patients; however, progressive disease is still fatal for over 50% of such patients (1). Drug resistance acquired during the course of treatment is one of the major reasons for treatment failure (2), with p53 loss of function being identified as one of the major mechanisms responsible for neuroblastoma drug resistance (3). Therefore, identifying drugs that are p53 independent may provide agents active against recurrent, drug-refractory neuroblastomas.

PZA (NSC366140, PD115934), a novel DNA intercalator, is a rationally synthesized acridine derivative that has been shown (a) to have a broad-spectrum activity against tumor cells in vitro and in vivo (4, 5, 6), (b) to be equally potent in normoxic and hypoxic cells (5, 6), (c) to be equally effective in cycling and noncycling cells (5, 6), (d) to be active in cells resistant to agents due to overexpression of Pgp (7) or MRP (8), and (e) to be toxic to cells with selective loss of topoisomerase I or II (9).

Phase I (10, 11, 12, 13) and II (14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26) clinical trials of PZA showed that major toxicities were myelosuppression in pediatric patients (12) and, in adults, neurotoxicity (prominent with 1-h PZA infusion times) as well as myelosuppression (13). However, only modest tumor responses were seen at the administered schedules (15, 17).

We found that PZA effectively induced cytotoxicity in MDR cell lines, and this prompted us to determine whether PZA could be effective in drug-resistant p53-nonfunctional neuroblastoma cell lines. Here we demonstrate that PZA showed a dose-dependent cytotoxicity for a range of neuroblastoma cell lines in a p53-independent manner, even under hypoxic conditions, but only with PZA exposure times that exceeded those tested in previous clinical trials.

We used a panel of 22 neuroblastoma cell lines (Table 1) obtained from patients at various phases of therapy: 6 at diagnosis before any therapy [SMS-SAN, SK-N-BE(1), SMS-KAN, SMS-KCN, CHLA-15, and CHLA-42]; 3 at progressive disease during dual-agent induction therapy (SMS-LHN, SMS-KCNR, and SMS-KANR); 6 at progressive disease during intensive multiagent chemotherapy [SK-N-BE(2), LA-N-6, CHLA-20, CHLA-119, CHLA-140, and CHLA-171]; and 7 derived at relapse after myeloablative therapy and BMT (CHLA-8, CHLA-51, CHLA-79, CHLA-90, CHLA-134, CHLA-136, CHLA-172). Neuroblastoma origin of these cell lines has been confirmed previously (2, 3, 27, 28).

None of the tumors that gave rise to cell lines used in this study had been treated with PZA. As we have shown previously, the cell line panel displays a spectrum of resistance (acquired during the therapy) to L-PAM, CBDCA, ETOP, cisplatin, doxorubicin, SN-38, and topotecan (2, 3, 27). Cell lines with LC₉₀ (i.e., the drug concentration that was lethal for 90% of the cell population) values greater than clinically achievable drug levels for at least two of three agents (ETOP, L-PAM, and CBDCA) were considered MDR. Clinically achievable reference drug concentrations were 10 μg/ml for L-PAM, 3 μg/ml for CBDCA, and 5 μg/ml for ETOP (2). Thus, 12 drug-sensitive and 10 MDR cell lines were used in this study (Table 1). SK-N-BE(1), SMS-SAN, SMS-LHN, LA-N-6, CHLA-171, CHLA-79, CHLA-134, CHLA-136 (3), and SMS-KAN (29) have been shown to carry wild-type TP53, whereas SK-N-BE(2), CHLA-119, CHLA-90, and CHLA-172 (3) carry mutant TP53.

SMS-KAN, SMS-KANR, SMS-KCN, SMS-KCNR, SK-N-BE(1), SK-N-BE(2), SMS-SAN, SMS-LHN, and LA-N-6 were cultured in complete medium made from RPMI 1640 (Irvine Scientific, Santa Ana, CA) supplemented with 10% heat-inactivated FBS (Gemini Bioproducts, Inc., Calabasas, CA). CHLA-15, CHLA-20, CHLA-42, CHLA-119, CHLA-140, CHLA-171, CHLA-51, CHLA-8, CHLA-79, CHLA-90, CHLA-134, CHLA-136, and CHLA-172 were cultured in complete medium made from Iscove’s modified Dulbecco’s medium (BioWhittaker, Walkersville, MD) supplemented with ≈3 mm l-glutamine (Gemini Bioproducts), 5 μg/ml insulin, 5 μg/ml transferrin, and 5 ng/ml selenous acid (ITS Culture Supplement; Collaborative Biomedical Products, Bedford, MA) and 20% heat-inactivated FBS. All neuroblastoma cell lines used in the study were under passage 30. Non-neuroblastoma cell lines were maintained in complete medium made from RPMI 1640 supplemented with 10% heat-inactivated FBS. The cell lines were cultured at 37°C in a humidified incubator containing 95% air + 5% CO₂ atmosphere without antibiotics. All cell lines tested negative for Mycoplasma. Cell lines were not selected for drug resistance in vitro.

SMS-SAN (MYCN gene amplified) and SMS-LHN (MYCN gene not amplified; both are drug-sensitive cell lines) were transduced with HPV-16 E6, which encodes a protein that promotes degradation of p53 and renders cells p53 nonfunctional (30), or with the LXSN retrovirus empty vector as a control; G418-resistant clones were selected from each. HPV-16 E6-transduced clones were designated as SAN/E6 D4, SAN/E6 B3, LHN/E6 2–6 and LHN/E6 5-B5. LXSN-transduced clones were designated as SAN/LXSN and LHN/LXSN. Reduced p53 activity in HPV-16 E6 clones was confirmed by the lack of p53 expression and the lack of induction of p53, p21, and MDM2 in L-PAM-challenged samples using Western blotting and also using a p53 transactivation assay (3). Using this system, we have shown previously that transduction of HPV-16 E6 conferred high-level multidrug resistance for L-PAM, CBDCA, and ETOP to both SMS-SAN and SMS-LHN cell lines (3).

PZA, L-PAM, CBDCA, and ETOP were obtained from the Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, National Cancer Institute (Bethesda, MD). FDA was from Eastman Kodak Co. (Rochester, NY), and eosin Y was from Sigma Chemical Co. (St. Louis, MO).

The cytotoxicity of chemotherapeutic agents for cell lines was determined with the DIMSCAN assay system (2, 3, 27, 31). DIMSCAN uses digital imaging microscopy to quantify viable cells, which selectively accumulate FDA. DIMSCAN is capable of measuring cytotoxicity over a 4–5-log dynamic range by quantifying total fluorescence/well (which is proportional to viable, clonogenic cells) after eliminating background fluorescence with digital thresholding and eosin Y quenching (2). Cell lines were seeded at 10,000–15,000 cells in 150 μl of complete medium per well into 96-well plates. After overnight incubation, various concentrations of PZA in 100 μl of complete medium were added to each well. The final drug concentrations ranged from 0 to 20 μm. Each condition was tested in 12 replicates. After incubation of cell lines with PZA for 3 days, 150 μl of medium were removed from each well, FDA in 50 μl of medium (final concentration of FDA 10 μg/ml) was added, and plates were incubated for an additional 25 min at 37°C before adding 30 μl of 0.5% eosin Y to each well. Total fluorescence was then measured using digital image microscopy, and results were expressed as surviving fractions of treated cells compared with control cells.

Using four cell lines (SMS-KANR, SMS-KCNR, LA-N-6, and CHLA-90), PZA activity was tested in atmospheric (20.7% oxygen, 5% CO₂, and 74% N₂) and hypoxic (5% or 2% oxygen, 5% CO₂, and the balance N₂) conditions. The stated hypoxic conditions were created and maintained by flushing respective gas mixtures into the hermetic chambers containing 96-well plates as described previously (32). The chambers were incubated at 37°C for 3 (SMS-KANR, LA-N-6, and CHLA-90) and 4 (SMS-KCNR) days, and the plates were then assayed for cytotoxicity by DIMSCAN as described above.

CHLA-90, and CHLA-15 cell lines were treated with PZA (1, 2, 3, and 5 μm) in 96-well plates. After incubation for 1, 3, 6, and 24 h with PZA, 150 μl of medium were removed from each well and replaced with an equal amount of fresh complete medium without disturbing the cell monolayer. This procedure was repeated immediately twice. Plates were then incubated for 3 days and subjected to DIMSCAN analysis as described above.

CHLA-90 was assayed with a range of PZA concentrations and times (for 1–72 h) in such a manner that a PZA concentration × time (here designated AUC) of 30, 60, 90, and 120 was achieved. We have adopted the term AUC (AUC was calculated as μm PZA·h) from clinical pharmacokinetics to refer to PZA cytotoxicity as a function of drug exposure. Cell lines seeded in 96-well plates were exposed to concentrations of PZA for 1, 2, 3, 6, 8, 16, 24, and 72 h, such that for each point on a graph, the indicated AUC was obtained. Plates were then washed three times with fresh medium (except for the 72-h condition), incubated for 3 days, and subjected to DIMSCAN analysis as described above.

The LC₉₀ and LC₉₉ (the drug concentration that was lethal for 90% or 99% of the cell population) values were calculated using the software “Dose-Effect Analysis with Microcomputers” (33). A cell line with a LC₉₀ value greater than or equal to the clinically achievable concentration was considered resistant to PZA. The reported PPL for PZA varies with administration schedule (10, 12, 13) from 2.2 to 7 μm. We chose 2 μm as a clinically achievable reference value for PZA. The area under the time-concentration curve that results in cytotoxicity for 90% of treated cells (AUC₉₀) was calculated as (LC₉₀ × exposure time to the drug).

Before analysis, the data were examined to evaluate compatibility with the assumptions of the normal distribution and homoscedasticity that would permit the use of standard parametric methods of analysis, in particular, the ANOVA. For the effect of hypoxia on PZA cytotoxicity, the logarithm transformation was used, and for the comparison analysis of LC₉₀ and LC₉₉ values for drug-sensitive versus drug-resistant cell lines and for p53-functional versus p53-nonfunctional cell lines, the square root transformation was used. These transformations rendered the data compatible with the assumptions required for ANOVA.

To adjust for the dependence of the assay-to-assay variability on dose, the observations were weighted by the inverse of PZA dose + 0.1. Each of the cell lines was analyzed separately using a factorial design ANOVA employing the generalized linear models procedure. When comparing LC₉₀ and LC₉₉ values for drug-sensitive versus drug-resistant cell lines and for p53-functional versus p53-nonfunctional cell lines, a one-way ANOVA was used. The overall P was based on the ANOVA F test. If the F test P was <0.05, the least significant difference method (34) was used for multiple comparisons. The means and the associated 95% CIs were calculated to summarize the data.

To perform statistical analysis, SAS Software, Version 8.2 was used (SAS Institute, Inc., Cary, NC).

The cell line panel used in this study has been characterized previously for the profile of drug sensitivity to commonly used agents in neuroblastoma: L-PAM; CBDCA; and ETOP (2, 3, 27). Cell lines with LC₉₀ values greater than clinically achievable drug levels for at least two of the three agents were considered drug resistant. This panel included 12 drug-sensitive and 10 drug-resistant cell lines (Table 1). Representative dose-response curves for L-PAM, ETOP, CBDCA, and PZA are shown in Fig. 1 for a drug-sensitive neuroblastoma cell line established at diagnosis (CHLA-15) and two MDR cell lines established at relapse after myeloablative therapy and BMT (CHLA-90 and CHLA-134). The LC₉₀ and LC₉₉ values for PZA and the AUC₉₀ calculated from the LC₉₀ values are listed in Table 1. The LC₉₀ and LC₉₉ of treated cells were calculated from dose-response curves for each drug tested using the DIMSCAN assay.

PZA showed dose-dependent cytotoxicity against all cell lines tested. The LC₉₀ values of drug-sensitive cell lines ranged from 0.003 to 1.1 μm, and the LC₉₉ values ranged from 0.07 to 5.9 μm. The LC₉₀ values of MDR cell lines ranged from 0.8 to 2.4 μm, and the LC₉₉ values ranged from 2.2 to 17.4 μm. PZA was more effective against the drug-sensitive cell lines than against MDR cell lines (LC₉₀ values P < 0.001, LC₉₉ values P = 0.028), but PZA was highly cytotoxic at ∼5 μm levels (72-h exposure) for most cell lines with the MDR phenotype.

PZA LC₉₀ values of cell lines with loss of p53 function ranged from 0.9 to 2.1 μm, and the LC₉₉ values ranged from 2.2 to 4.7 μm (Figs. 2 and 3; Table 1). PZA LC₉₀ values for p53-funtional/drug-resistant cell lines ranged from 0.8 to 2.4 μm, and the LC₉₉ values ranged from 2.5 to 5 μm. There was no statistical difference between PZA activity in drug-resistant/p53-functional cell lines versus MDR/p53-nonfunctional cell lines (P > 0.05).

To confirm that PZA is a p53-independent agent, PZA efficacy was tested in neuroblastoma cell lines in which p53 function was selectively abrogated by HPV-16 E6 transduction (3). We compared dose-response curves for PZA (Fig. 2) with other drugs (L-PAM, CBDCA, and ETOP) in the p53-nonfunctional cell lines.

SAN/E6 and LHN/E6 clones have been studied previously for their sensitivities to L-PAM, ETOP, and CBDCA (3). The LC₉₀ values of SAN/E6 and LHN/E6 clones were generally higher than drug concentrations achieved in patients treated with myeloablative doses, whereas those values in LXSN controls were in the range of clinically achievable levels. The LC₉₀ values of SAN/E6 clones were 8.3 and 10.8 μg/ml for L-PAM, 19.6 and 32 μg/ml for CBDCA, 29 and 110 μg/ml for ETOP, whereas the LC₉₀ values of SAN/LXSN controls were 1.8 μg/ml for L-PAM, 1.3 μg/ml for CBDCA, and 0.7 μg/ml for ETOP. Similarly, the LC₉₀ values of LHN/E6 clones were 13 and 32 μg/ml for L-PAM, 53 and 694 μg/ml for CBDCA, and 7.6 and 166 μg/ml for ETOP. The LC₉₀ values of LHN/LXSN controls were 4.7 μg/ml for L-PAM, 6.4 μg/ml for CBDCA, and 3.7 μg/ml for ETOP.

The PZA LC₉₀ values of SAN/E6 clones were 0.15 and 0.76 μm, whereas the LC₉₀ value of the SAN/LXSN control was 0.1 μm. Similarly, LC₉₀ values of LHN/E6 clones were 1.8 and 2.04 μm, whereas the LC₉₀ value of LHN/LXSN cells was 0.5 μm. Although LC₉₀ values of SAN/E6 (clones B3 and D4) and LHN/E6 (clones 5-B5 and 2-6) were 1.5–7.6 times higher than those of SAN/LXSN and LHN/LXSN controls, they were within the range of the clinically achievable levels of PZA, except for the LHN/E6 2-6 clone, whose LC₉₀ value (2.04 μm) reached the upper limit.

We compared the LC₉₀ values obtained when testing p53-functional (n = 16), p53-nonfunctional (n = 6), and p53 nonfunctional via HPV-16 E6 transduction (n = 4) neuroblastoma cell lines against ETOP, CBDCA, L-PAM, and PZA (Fig. 3). Most p53-nonfunctional cell lines were highly resistant to ETOP, CBDCA, and L-PAM (LC₉₀ > clinically achievable concentrations in the myeloablative setting), whereas PZA did not show any relationship between p53 functionality and drug sensitivity.

We have tested PZA in drug-sensitive (SMS-KANR and SMS-KCNR) and MDR [LA-N-6 and CHLA-90 (the latter is p53 nonfunctional)] cell lines at atmospheric and reduced oxygen conditions, and we found that this agent was equally effective in standard culture conditions and hypoxia (Fig. 4). A 3-day exposure to PZA resulted in equivalent cell kills for LA-N-6 (drug-resistant but p53 functional) and CHLA-90 (MDR, p53 nonfunctional) cell lines in atmospheric (20.7% O₂, 5% CO₂, and 74% N₂) and hypoxic (2% O₂, 5% CO₂, and 93% N₂) conditions (P > 0.05). PZA dose-response curves obtained in 5% O₂, 5% CO₂, and 90% N₂ after 3-day exposure in SMS-KANR and after 4-day exposure in SMS-KCNR cell lines were comparable with those obtained in atmospheric (20.7% O₂, 5% CO₂, and 74% N₂) conditions (Ps >0.05; data not shown).

In the clinical setting, PZA has been given as a 3-h infusion (Tables 2 and 3). To model the activity of PZA at various delivery schedules, we tested the dose-response to PZA after 1-, 3-, 6-, 24- and 72-h exposures in the CHLA-15 and CHLA-90 cell lines (Fig. 5). Exposure to PZA of 6 h or greater appeared to increase activity relative to shorter exposures.

We tested the MDR, p53-nonfunctional cell line CHLA-90 in the presence of various PZA concentrations for 1, 2, 3, 6, 8, 16, 24, and 72 h (Fig. 6), such that for each concentration tested, the cells were exposed to a time-concentration for PZA of 30, 60, 90, or 120 μm·h, here designated as AUC. These in vitro assays were designed to model the responses to varying exposure times to PZA, which may occur during continuous infusion clinical trials. The highest PZA cytotoxicity (∼2.5 logs and greater of cell killing) was achieved with brief exposures to >20 μm PZA, but such high concentrations are clinically unachievable. However, exposure to 5 μm PZA for 6–24 h, which may be clinically achievable with stem cell support, was cytotoxic for the CHLA-90 cell line. Increasing the time of exposure (i.e., increasing the AUC without increasing the drug concentration) for 5 μm PZA led to a multilog increase in the degree of cell kill achieved: 0.7 log at 3 h; 1.0 log at 6 h; 1.4 logs at 12 h; 2.1 logs at 18 h; and 3.2 logs at 24 h.

We have shown that neuroblastomas manifest sustained resistance to multiple drugs (non-Pgp- and non-MRP-associated multidrug resistance) in vitro and that the resistance correlates with the intensity of chemotherapy used in patients (2, 35, 36). We have also observed that all drug-sensitive neuroblastoma cell lines established before chemotherapy had functional p53, whereas in MDR cell lines, p53 was frequently nonfunctional due to mutations or nonmutational events [such as MDM2 genomic amplification (3, 37)]. Moreover, abrogation of p53 function in drug-sensitive neuroblastoma cell lines conferred the MDR phenotype (3), even at drug levels obtained during myeloablative therapy. Drugs that are independent of p53 may be non-cross-resistant with those drugs in current use for neuroblastoma and could have activity in patients who develop chemotherapy-refractory disease.

PZA is a novel drug that has shown impressive activity in vitro and in preclinical studies conducted in mice (4, 5). Previous in vitro studies showed that cells resistant to drugs due to Pgp (7) or MRP (8) overexpression or due to altered topoisomerases I or II were invariably affected by PZA (9). Although responses (2 complete responses, 12 partial responses, and 5 minor responses) were seen in clinical trials (13, 15, 16, 17), PZA antitumor activity in Phase II trials was disappointing (Table 3). The present study was undertaken to investigate the effect of PZA on drug-resistant neuroblastoma in general and to determine whether PZA had activity against neuroblastomas manifesting MDR due to loss of p53 function (3).

It has been well documented that p53, a regulatory nuclear protein, is activated in response to a variety of DNA-damaging agents, resulting in cell cycle arrest (38) and/or apoptosis (39). PZA-mediated cytotoxicity has been shown to correlate with DNA damage and inhibition of DNA synthesis (40), yet we found that PZA was cytotoxic for neuroblastoma cell lines that lacked p53 function and were resistant to other drugs that cause DNA damage (ETOP, CBDCA, and L-PAM). In the panel of cell lines we tested, there were four cell lines with a loss of p53 function due to TP53 mutations and two additional cell lines with loss of p53 function due to nonmutational events (in one instance, associated with MDM2 genomic amplification). All of these cell lines were highly resistant to various drugs (2, 3). We also tested four HPV-16 E6-transduced clones derived from two different drug-sensitive/p53-functional cell lines, which, due to abrogation of p53 function, showed high-level resistance to ETOP, CBDCA, and L-PAM (3). In this study we showed that, in general, PZA was more active in cell lines lacking the MDR phenotype than in MDR cell lines (P < 0.05; Table 1). However, this trend was due to reasons other than a lack of p53 function because PZA was equally toxic in resistant cell lines with nonfunctional p53 (LC₉₀ values ranged from 1.0 to 2.1 μm; LC₉₉ values ranged from 2.5 to 4.7 μm) and those with functional p53 (LC₉₀ values ranged from 0.9 to 2.4 μm; LC₉₉ values ranged from 2.5 to 5 μm; P > 0.05; Table 1 and Fig. 3). The LC₉₀ values of HPV-16 E6-transduced clones were 0.2–2.04 μm (i.e., within the range of clinically achievable levels).

Phase I clinical trials of PZA have tested several dose schedules [Table 2 (10, 12, 13)]. With an administration schedule using a 1-h infusion every 21 days, the reported MTD was 600 mg/m² with a PPL of 7.1 μm (13), and neurotoxicity was dose-limiting (13). However, prolongation of infusion duration from 1 to 3 h resulted in a reduction in the incidence of neurotoxicity and permitted further dose escalation to 750 mg/m² (PPL = 2.3 μm), during which neutropenia emerged as the principal dose-limiting toxicity. The total dose of 750 mg/m² administered as a single dose or as 150 mg/m²/day for 5 consecutive days was the recommended dose for adults (13). In another Phase I clinical trial, a 1-h PZA infusion at the recommended 600 mg/m² dose resulted in an AUC of 30 ± 25 μm·h and peak drug levels of 2.2 ± 1.05 μm (10). In a Phase I clinical trial conducted in adults as a weekly 24-h infusion, 281 mg/m² was reported as the MTD, and the maximum plasma level was 1.6 μm with a maximum AUC of 56 μm·h. This schedule was well tolerated, and neurotoxicity was not problematic (11). This Phase I clinical trial of PZA given as a 24-h continuous infusion was designed to minimize neurological toxicity and to achieve steady-state plasma concentrations that were reported to be associated with in vitro cytotoxicity [IC₅₀ levels ranging from 0.03 to 0.88 μm as determined by colony-forming assay in over 40 cell lines of various tumor types (5, 6)]. Further dose escalation was precluded due to myelosuppression and one case of grade 3 mucositis; however, the authors suggested that dose escalation could have been feasible with colony-stimulating factor support.

In a study conducted in children and young adults, 640 mg/m² PZA administered as a 1- or 24-h infusion was reported as the recommended dose (12). The predictive AUC₅₀ (i.e., the AUC that produces 50% of the maximum effect) was reported as 45 μm·h for leukopenia and 34 μm·h for neutropenia. Phase I pharmacokinetic studies have shown that the PZA half-life t_1/2α was 1–85 min, and the t_1/2β ranged from 2.7 to 158 h (10, 12). Actual PZA plasma concentrations, measured in two patients at the end of a 24-h infusion, were 7 and 2.3 μm. In this age group, myelosuppression was reported as the dose-limiting toxicity; in contrast to the adult population, neurotoxicity was not prominent (12).

Multiple Phase II clinical studies used a total dose of PZA ranging from 600 to 750 mg/m² PZA administered as a 3-h infusion every 21 days, which produced only limited clinical activity (Table 3). Our in vitro data demonstrate that PZA cytotoxicity is dose and time dependent. This observation was in agreement with a study of Grem et al. (40), which showed that with an increased PZA exposure time, the concentration required for a given cytotoxic effect in MCF-7 breast cancer cells was decreased. We determined dose-response curves for CHLA-15 (drug-sensitive) and CHLA-90 (MDR due to loss of p53 function) cell lines for 1, 3, 6, 24, and 72 h with PZA. We found that longer exposures resulted in higher cell kills at a given concentration in these cell lines. In the CHLA-90 cell line, 1-, 3-, and 6-h PZA exposures resulted in LC₉₀ values 7, 4, and 1.5 times greater than that of a 72-h exposure, respectively. After a 24-h exposure, the LC₉₀ value was identical to that obtained after a 72-h PZA exposure. In CHLA-15, the LC₉₀ values after 1-, 3-, 6-, and 24-h exposures were 10, 8.8, 6.7, and 2.7 times higher, respectively, relative to a 72-h exposure. We have also determined the (concentration × time) relationship to cytotoxicity for PZA in the CHLA-90 TP53-mutated, MDR neuroblastoma cell line, such that PZA exposures of 30, 60, 90, or 120 μm·h were achieved. Over a 3-log cell kill was achieved with a 1–3-h exposure to 20 μm PZA, but a similar effect was seen with 12–24-h exposures to 5–10 μm PZA (Fig. 6).

The AUC₉₀ for all 10 MDR neuroblastomas and for 3 of the 12 non-MDR cell lines (Table 1) exceeded that reported in clinical trials. These data suggest that neuroblastoma patients with drug-resistant tumors receiving PZA at a schedule used in Phase II studies would be unlikely to respond (according to in vitro data, cell kills <1 log), consistent with the lack of response seen in the seven neuroblastoma patients treated in the Phase II study (19). However, our in vitro data suggest that if a concentration of 5–10 μm PZA was sustained for 8–24 h (equal to an AUC of 80–120 μm·h), PZA could be active against recurrent neuroblastomas. Due to the significant hematopoietic toxicity of PZA, such prolonged exposure would require hematopoietic stem cell support, precluding a study of such a dose schedule against xenografts in immunocompromised mice. Most Phase I clinical trials reported a PZA C_max of 1.6–2.3 μm. Only two pediatric patients received PZA in a 24-h infusion, and the 7 μm PZA level measured in one patient at the end of 24-h infusion (12) suggests that plasma concentrations over a time >8h of 5–10 μm PZA may be achievable in pediatric patients, especially if given with hematopoietic stem cell support. Our data suggest that achieving >5 μm PZA for >8 h may be effective against recurrent neuroblastoma, but such a schedule has yet to be tested against neuroblastoma in a clinical trial.

Preclinical investigation of PZA in solid tumors in mice concluded that PZA was a schedule category III agent, i.e., a schedule-independent agent (4). Due to the variety of tumor types used in that study, distinct drug sensitivity profiles of tumor cells, and drug administration schedules, it is difficult to interpret the results. In general, treatment schedules delivering higher total drug doses showed better activity, but such high levels tended to be lethal for treated mice. Therefore, the data in mice do not adequately address the possibility that a higher systemic exposure of PZA could be more effective and might be achievable using hematopoietic stem cell support.

(a) Our observation that PZA activity increased with a more prolonged drug exposure, (b) pharmacokinetic data showing a relatively short t_1/2α for PZA, (c) peak levels being limited by neurotoxicity, and (d) myelosuppression being the dose-limiting toxicity in pediatric patients, taken together, suggest that dose escalation to achieve a high PZA AUC using prolonged (6–24 h) infusions supported by hematopoietic stem cell infusion should be tested. Increasing the AUC of PZA may be tolerable with stem cell support and may prove effective in recurrent neuroblastoma patients (and potentially other tumor types), despite discouraging results seen in Phase II clinical trials using shorter infusions of PZA. A Phase I trial testing this hypothesis (N2002-01) is ongoing in the New Approaches to Neuroblastoma Therapy consortium.⁴

We thank Dr. June Biedler for providing SK-N-BE(1) and SK-N-BE(2) cell lines and Dr. Susan Groshen for helpful advice in statistical analysis.