Hypoxia-Activated Prodrug Evofosfamide Treatment in Pancreatic Ductal Adenocarcinoma Xenografts Alters the Tumor Redox Status to Potentiate Radiotherapy

Abstract

Aims: In hypoxic tumor microenvironments, the strongly reducing redox environment reduces evofosfamide (TH-302) to release a cytotoxic bromo-isophosphoramide (Br-IPM) moiety. This drug therefore preferentially attacks hypoxic regions in tumors where other standard anticancer treatments such as chemotherapy and radiation therapy are often ineffective. Various combination therapies with evofosfamide have been proposed and tested in preclinical and clinical settings. However, the treatment effect of evofosfamide monotherapy on tumor hypoxia has not been fully understood, partly due to the lack of quantitative methods to assess tumor pO₂ in vivo. Here, we use quantitative pO₂ imaging by electron paramagnetic resonance (EPR) to evaluate the change in tumor hypoxia in response to evofosfamide treatment using two pancreatic ductal adenocarcinoma xenograft models: MIA Paca-2 tumors responding to evofosfamide and Su.86.86 tumors that do not respond.

Results: EPR imaging showed that oxygenation improved globally after evofosfamide treatment in hypoxic MIA Paca-2 tumors, in agreement with the ex vivo results obtained from hypoxia staining by pimonidazole and in apparent contrast to the decrease in K^trans observed in dynamic contrast-enhanced magnetic resonance imaging (DCE MRI).

Innovations: The observation that evofosfamide not only kills the hypoxic region of the tumor but also improves oxygenation in the residual tumor regions provides a rationale for combination therapies using radiation and antiproliferatives post evofosfamide for improved outcomes.

Conclusion: This study suggests that reoxygenation after evofosfamide treatment is due to decreased oxygen demand rather than improved perfusion. Following the change in pO₂ after treatment may therefore yield a way of monitoring treatment response. Antioxid. Redox Signal.

Introduction

Tumors are able to grow to a size of 2–3 mm³ by relying on passive diffusion of oxygen and nutrients; for further growth and eventual metastasis additional vasculature is required, which is accomplished by recruiting angiogenic pathways to develop de novo vasculature (5). While in normal processes there is a tight control between antiangiogenic and proangiogenic pathways, which results in a well-structured and functional vascular network, in tumors the proangiogenic factors dominate to meet the metabolic demands caused by tumor growth. This results in a tumor vascular network, which is poorly organized both structurally and functionally (5). As a consequence, the tumor microenvironment is characterized by poor perfusion, low oxygen (hypoxia), and high interstitial fluid pressure (12, 14).

In addition, tumor cells often develop other altered chemical characteristics such as elevated aerobic glycolysis and low pH (9). The eventual development of such a harsh microenvironment results in tumors resistant to treatment with chemotherapy and radiotherapy (7). Solid tumors are known to have regions of hypoxia, which render them resistant to radiotherapy (14). Hypoxia-activated prodrugs (HAPs) have been developed recently to target hypoxic regions of the tumor, which are normally resistant to radiation and antiproliferative chemotherapeutic agents (1, 3).

The HAP evofosfamide (aka TH-302) is comprised of a nitroimidazole moiety covalently linked to a bromo-isophosphoramide (Br-IPM) moiety (Fig. 1) (28, 40). Intracellularly, the nitroimidazole moiety can undergo one-electron reduction to the corresponding anion radical through cellular reductases. Once formed, the anion radical undergoes rapid reoxidation at diffusion-limited rates in normoxic conditions, returning the drug back to the original state. Under these conditions, the prodrug formulation is retained and evofosfamide does not exert significant cytotoxicity as the reactive Br-IPM moiety is blocked. The overall equilibrium between the two forms is roughly exponentially dependent on the local pO₂ (41).

FIG. 1.

Mechanism of evofosfamide action. (A) Structure of evofosfamide (also known as TH302). (B) After crossing the cell membrane, the nitro group of evofosfamide is reduced by 1e-reductases to generate the radical form of the drug. (C) Under normoxic conditions, this reaction is rapidly reversed to regenerate the original drug. (D) Under hypoxic conditions, this reaction is not reversed and the unstable radical fragments to generate the alkylating agent Br-IPM. (E) Br-IPM crosslinks DNA, targeting the cell for apoptosis or cell cycle arrest. Br-IPM, bromo-isophosphoramide.

Under hypoxic conditions where the equilibrium between the reduced and oxidized forms is shifted, evofosfamide is susceptible to fragmentation in the reduced form to release the pharmacologically active Br-IPM moiety. Once generated, free Br-IPM is a powerful alkylating agent capable of crosslinking DNA and inducing apoptosis. In vitro, evofosfamide has been tested in various cell lines and found to have selective hypoxic cytotoxicity (28, 40). Based on several preclinical studies, evofosfamide has been tested clinically in a phase I/phase II study in refractory multiple myeloma as a HAP and found to have shown efficacy (21). Evofosfamide has been less successful in phase III trials in solid tumors (8, 43), prompting the search for predictive biomarkers for an effective response (17, 36).

Innovation

Although it is known that evofosfamide is an effective drug for combination use with other anticancer treatments, the potential effect on oxygenation in tumor microenvironment has not fully examined. This study provides strong evidence of decreasing hypoxic fraction exclusively from hypoxic tumor. The result from electron paramagnetic resonance imaging oximetry suggested the importance of evaluating hypoxia in individual tumors. This hypoxia decreasing effect may be utilized in designing regimens for the combination therapy of evofosfamide and radiation therapy. Magnetic resonance imaging methods to provide quantitative assessment of tumor oxygenation and changes in response to treatment will play a key role in combination treatments.

A few preclinical studies have explored the use of evofosfamide in combination therapies (16, 22, 32, 41, 42, 47). When evaluating combination therapies involving oxygen-dependent treatments such as radiation therapy (RT), it is necessary to know the pO₂ levels during the lifecycle of the treatment as RT may cause complications and should be administered at the point of maximum efficacy.

As evofosfamide likely changes these levels, the timing of evofosfamide administration can be manipulated to maximize the treatment efficacy during planning of combination therapy. The evidence for an evofosfamide-induced decrease in the hypoxic fraction of the tumor rests on ex vivo immunohistochemical analyses (35), which are qualitative and require serial tissue biopsies, which may be difficult to implement clinically. To plan the treatment regimen effectively, a noninvasive method to evaluate the change in pO₂ in response to evofosfamide quantitatively would be useful.

EPRI (electron paramagnetic resonance imaging) is one of the most reliable methods of measuring pO₂ in vivo in preclinical models (25, 26). EPRI requires the injection of a nontoxic paramagnetic spin probe, and absolute pO₂ can be calculated from the linewidth of the distributed probe. The paramagnetic spin probe is well tolerated for serial imaging without impacting the physiology or biochemical profile of the tissue being interrogated. With EPRI it is possible to experimentally determine both the fractional volume of the hypoxic region of the tumor with pO₂ <10 mmHg and the pO₂ spatial distribution (27). It is also possible to dynamically monitor changes in tumor oxygenation by EPRI using hyperoxygenation strategies or by chemical hypoxia induction (26, 42, 49).

EPRI has been previously used in preclinical studies to examine the pretreatment tumor oxygenation status quantitatively to examine the efficacy of evofosfamide in combination with pyruvate to induce hypoxia or in combination with radiation (42, 47). pO₂ values from EPRI were successful in predicting evofosfamide sensitivity across cell lines.

These studies showed the importance of determining the pretreatment tumor oxygenation status quantitatively for effective treatment. While evaluating combination therapies that involve oxygen-dependent treatment such as RT, it is also crucial to understand the pO₂ change in response to evofosfamide in tumor microenvironment because the residual fraction of the tumor microenvironment can be modified by the treatment. In such case, we need to consider the timing of evofosfamide administration to maximize the treatment efficacy when combination therapy is planned.

Recent studies suggested that evofosfamide induced a decrease in hypoxic fraction; such effects were monitored ex vivo using histochemical assays method, which are qualitative (20). While planning treatments with radiation or antiproliferatives after evofosfamide treatment, it is important to investigate the difference in pO₂ between pre- and post-treatment with evofosfamide in a unique tumor.

In this study, to evaluate the changes in oxygenation after evofosfamide treatment, tumor blood perfusion was examined using dynamic contrast-enhanced magnetic resonance imaging (DCE MRI). Combined with the pO₂ information obtained from EPRI, we could achieve the comprehensive assessment of treatment efficacy with evofosfamide by noninvasive imaging modalities. The reoxygenation effect after evofosfamide treatment was found to be driven by a decrease in oxygen consumption due to cell death or cell arrest rather than improved perfusion from relief of solid stress on the vascular network.

Results

The differential treatment response to evofosfamide was examined using two human pancreatic ductal adenocarcinoma (PDAC) xenografts that were previously indicated to be sensitive (MIA Paca-2) and resistant (Su.86.86) to evofosfamide therapy (28, 40). We confirmed the sensitivity of each cell line by observing tumor growth after the initiation of the evofosfamide treatment (Fig. 2A, B). A statistically significant reduction in tumor growth is evident in the Kaplan–Meier plots of the MIA Paca-2 tumors treated with 50 mg/kg evofosfamide 5 days a week compared with the groups treated with vehicle (5% dimethyl sulfoxide [DMSO] in phosphate buffered saline) (Fig. 2C: p = 0.02, log-rank test), but no observable inhibition in Su.86.86 was noticed (Fig. 2D: p = 0.8).

FIG. 2.

Assessment of evofosfamide treatment efficacy on MIA Paca-2 and Su.86.86 tumors. (A, B) The tumor growth curve of MIA Paca-2 and Su.86.86 tumors treated with evofosfamide (evofosfamide 50 mg/kg i.p. 5 days/week) or vehicle. The number of animals is indicated in the plot. (C, D) The Kaplan–Meier survival curves of (A, B) experiments, respectively. *Shows the p < 0.05 evaluated by the log-rank test.

These results were also mirrored in in vitro studies under aerobic conditions (Fig. 3). Both cell types were effectively killed at micromolar concentrations of evofosfamide in ambient air-equilibrated condition, suggesting that residual activity still exists even under high oxygen concentrations. Furthermore, MIA Paca-2 cells were still killed at lower concentrations of evofosfamide than Su.86.86 cells, even in the in vitro situation in which differences in oxygen delivery are not expected to be a limiting factor, although consumptive hypoxia cannot be ruled out a priori (34).

FIG. 3.

In vitro cytotoxicity evaluated after 48 h of incubation with evofosfamide at the indicated concentrations (0, 12.5, 25, 50, 100, 200 μM).

The proposed mechanism for evofosfamide cytotoxicity involves a reduction of the inactive prodrug to generate the reactive species, which is reversed under normoxic conditions (28, 40). Since the equilibrium of the fragmentation process generating the pharmacologically active species from the inactive prodrug is dependent on local oxygen levels, the efficacy should be correlated with the hypoxic fraction. Su.86.86 tumors stained higher for host endothelial cell-derived CD31 (Fig. 4), a marker for blood vessels, than the MIA Paca-2 cell line, consistent with prior studies, which indicated that MIA Paca-2 xenografts are more hypoxic in vivo compared with the corresponding Su.86.86 xenografts (27, 47).

FIG. 4.

Difference in vascularity between poorly differentiated MIA Paca-2 tumor and highly differentiated Su.86.86 tumor evaluated by immunohistochemistry assays. (A) CD31, Glut-1, and DAPI were stained in red, green, and blue, respectively. (B) CD31-positive fraction for both tumors. The fraction was calculated from randomly selected five fields per tumor section (n = 4 per group).

Considering the oxygen-dependent activation of evofosfamide, this difference in oxygenation status before treatment may lead to a difference in treatment response. This assessment assumes that the hypoxia is unchanged with treatment. However, the cytotoxic effect of evofosfamide is expected to lower oxygen consumption, while apoptotic or necrotic cell death is expected to relieve the interstitial pressure within the tumor, increasing oxygen supply by alleviating the constriction of the capillaries supplying blood to the tumor (14).

To determine the functional alteration in the tumor microenvironment induced by evofosfamide in MIA Paca-2 tumors, we examined tumor hypoxia using pimonidazole staining on evofosfamide-treated and untreated tissue sections. Substantial differences in pimonidazole-positive area between untreated and evofosfamide-treated tumors were found, in agreement with a previous study evaluating tumor hypoxia after evofosfamide treatment (Fig. 5A) (20). In general, it is accepted that pimonidazole binds tissue at the threshold of pO₂ <10 mmHg (35). Using this threshold, the pimonidazole-stained area decreased by >50% in evofosfamide-treated tumors (Fig. 5B).

FIG. 5.

Histological assessment of tumor hypoxia using pimonidazole staining. (A) Representative sets of evofosfamide-treated and untreated MIA Paca-2 tumors stained with HE and pimonidazole. (B) Comparison of pimonidazole-positive fraction between evofosfamide-treated (n = 4) and untreated MIA Paca-2 tumors (n = 4). HE, hematoxylin and eosin.

Pimonidazole provides a qualitative measure of hypoxia (20, 35). To quantify responses to evofosfamide treatment more precisely, we imaged pO₂ of MIA Paca-2 and Su.86.86 xenografts pre- and post-treatment with evofosfamide by EPR. Figure 6A shows representative images of MIA Paca-2 tumors before treatment (top panels) and after 2-day treatment with evofosfamide (bottom panels), while the corresponding frequency histograms of pO₂ values before and after treatment for the MIA Paca-2 xenografts for this image are shown in Figure 6B. As expected, there was a correlation between tumor size and oxygen levels in both tumor models before treatment (Fig. 6C, D).

FIG. 6.

The effect of evofosfamide treatment on EPR oximetry. (A) Representative EPR pO₂ images from MIA Paca-2 and Su.86.86 tumors taken before and 48 h after treatment showing the improved oxygenation only in MIA Paca-2 tumor. (B) The frequency histogram of pO₂ values of (A). The white arrows indicate the area where improved pO₂ was evident. (C, D) The plot of median PO₂ before and 48 h after treatment in MIA Paca-2 tumors and Su.86.86 tumors, respectively. (E, F) Dot plots of pretreatment HF10 versus ΔHF10 and pretreatment pO₂ versus ΔpO₂ in MIP Paca-2 tumors. (G, H) Dot plots of pretreatment HF10 versus ΔHF10 and pretreatment pO₂ versus ΔpO₂ in Su.86.86 tumors. **p < 0.01. EPR, electron paramagnetic resonance; HF10, hypoxic fraction <10 mmHg.

Change in pO₂ levels after evofosfamide treatment is not evident in any of the tumors for the Su.86.86 evofosfamide-insensitive tumor model (Fig. 6E). The changes after treatment for the evofosfamide-sensitive MIA Paca-2 tumor are more complex. In some MIA Paca-2 tumors, a reduction in the size of the hypoxic fraction and a shift in the distribution toward higher pO₂ levels are evident after treatment (white arrows in Fig. 6A). The change in pO2 and hypoxic fraction <10 mmHg (HF10) in individual tumors is shown in Supplementary Figure S1. This is especially evident in pretreatment tumor regions with pO₂ in the range ≤16 mmHg. In regions with pO₂ >16 mmHg, no significant change or even an increase in hypoxia was detected.

The degree to which evofosfamide improves oxygenation is likely correlated with the extent to which the active drug is generated. It is therefore expected that the degree of change should be correlated with the initial level of hypoxia. We therefore plotted the correlation between the initial pO₂ and HF10 values and the respective changes in these values. MIA Paca-2 tumors showed a statistically significant negative linear correlation between pretreatment pO₂ and ΔpO₂ (ΔpO₂ = post-treatment pO₂ − pretreatment pO₂) (Fig. 6E) and between pretreatment HF10 and ΔHF10 (ΔHF10 = post-treatment HF10 − pretreatment HF10) (Fig. 6F). The Su.86.86 tumor xenografts by contrast did not show major changes in pO₂ and HF10 between untreated and treated tumors (Fig. 6G, H).

Hypoxia can result from either a demand for oxygen that is in excess of a functioning blood supply network or from a defective vascular network delivering an insufficient supply. To determine evofosfamide's effect on oxygen supply, we investigated the treatment effect of evofosfamide on tumor perfusion and permeability by DCE MRI using Gd-DTPA as a contrast agent. K^trans values were evaluated before treatment and after 2 days of daily treatment of evofosfamide.

The representative DCE curves and K^trans images for both MIA Paca-2 and Su.86.86 tumors are shown in Supplementary Figure S2. In the absence of treatment, K^trans increased after 2 days in both MIA Paca-2 and Su.86.86 tumors (Fig 7A, B), and this increase was significantly decreased in both tumor models in the evofosfamide-treated mice, consistent with previous reports of decreased K^trans by evofosfamide treatment (4, 50). We further examined the treatment effect on blood vessels by measuring the CD31-positive fraction in the MIA Paca-2 tumor treated with evofosfamide or vehicle. There was no significant difference in CD31 between evofosfamide and the control group, indicating that the vascular damage caused by evofosfamide treatment was minimal (Fig. 7C).

FIG. 7.

Neither improved perfusion/permeability nor angiogenesis is responsible for higher pO₂ levels after evofosfamide treatment. (A, B) K^trans of Gd-DTPA in MIA Paca-2 and Su.86.86 tumors treated with evofosfamide (50 mg/kg daily) or vehicle for 48 h. (C) CD 31-positive fraction was measured by histological assessment. Each bar represents four tumor samples.

Evofosfamide is known to have a synergy with RT in pancreatic cancers, as evofosfamide targets hypoxic fractions of the tumor and RT is an oxygen-dependent therapy that can benefit from the reoxygenation effect of evofosfamide. Both the hypoxic and normoxic fractions of the tumor can be treated by a combination treatment in this manner, while a monotherapy of either alone primarily targets only one fraction. Previous studies have shown the superiority of evofosfamide+RT combination therapy in PDAC mouse xenografts over RT therapy alone (16), including on MIA Paca-2 and Su.86.86 tumors (27), but it is unknown if RT therapy can be used to decrease the dosing schedule of evofosfamide to reduce adverse reactions.

A combination treatment of evofosfamide and RT was therefore designed to account for the increase in tumor oxygenation 2 days after evofosfamide treatment on MIA Paca-2 tumors (Fig. 8A). With this regimen, RT is supposed to be performed after the hypoxic region in the tumor is depleted by evofosfamide. Figure 8B shows the tumor growth curve of MIA Paca-2 tumors treated with either vehicle, evofosfamide monotreatment, or combination treatment. Although the total evofosfamide dose was lower, the growth curves of the two treatment groups are similar with slightly stronger growth inhibition observed in combination therapy group in comparison with evofosfamide monotherapy.

FIG. 8.

Designing evofosfamide+radiation combination treatment regimen accounting for evofosfamide induced hypoxia. The MIA Paca-2 tumor growth inhibition by evofosfamide+radiation combination treatment regimen accounting for evofosfamide-induced oxygenation. (A) Scheme of combination therapy and evofosfamide monotherapy regimens. (B) The tumor growth curve of no treatment, evofosfamide monotherapy, and evofosfamide+RT combination therapy. RT, radiation therapy.

Discussion

Earlier studies have shown that MIA Paca-2 PDAC tumors are more sensitive to evofosfamide than Su.86.86 tumors. It is believed that the difference is mainly due to the higher hypoxia levels in MIA Paca-2 tumors, which are associated with poor vasculature, since reductive activation of evofosfamide is necessary to release the active fragment of evofosfamide. This cleavage is thought to occur at cytotoxic levels only in hypoxic regions deep within the tumor. A causative role for hypoxia is suggested by the time course of inhibition, as inhibition of tumor growth is not evident in the first few days (Fig. 2A, B) when the small MIA Paca-2 tumors (<600 mm³) have pO₂ levels comparable with their Su.86.86 counterparts (Fig. 6D).

The correlation with pO₂ levels observed is consistent with previous results. However, an underappreciated aspect of evofosfamide treatment is that oxygen levels are not constant during treatment as evofosfamide itself can alter the oxygenation state as apoptosis/necrosis induced by evofosfamide decreases oxygen consumption. Since evofosfamide mechanism is oxygen dependent, this alteration in oxygenation status may potentially decrease the treatment's effectiveness over time.

Despite its potential impact on evofosfamide treatment, only a few studies have evaluated tumor hypoxia after evofosfamide monotherapy, mostly by pimonidazole staining. Most of them showed improved pO₂ levels (32, 40, 42) 1–2 days after the treatment, while one showed no improvement 1 day after treatment (42). The study by Peeters et al. employed both pimonidazole staining and [18F] HX4 positron emission tomography (PET) imaging to evaluate the oxygen modification by evofosfamide treatment (32). Although both methods qualitatively showed improved oxygenation after evofosfamide treatment, there was a disparity in values between the two modalities probably due to the lack of the capability of quantitative assessment with PET imaging (15).

EPRI oximetry is a useful imaging modality to clarify this phenomenon because it allows minimally invasive in vivo quantitative pO₂ assessment. The capability to distinguish quantitatively tumor oxygen status from 5 to 25 mmHg gives this method the capability to assess changes after pharmacologic interventions. By EPRI, we could also observe the effect on pO₂ in each unique tumor and identify tumors likely to be susceptible to evofosfamide treatment. In addition to allowing comparison between tumor types, intratumor pO₂ levels measured by EPRI provide even more detailed information on the effect of evofosfamide treatment on the tumor microenvironment.

Not only can the oxygenation effect of evofosfamide therapy be seen in Figure 6, but it can also be seen that this differential effect is dependent on the initial pO₂ profile before treatment. The ΔpO₂ change pre- and post-treatment linearly correlated with pretreatment pO₂ only in MIA Paca-2 tumors (Fig. 6E), and a similar relationship was also observed between ΔHF10 and pretreatment HF10 (Fig. 6F). This is consistent with the capability of evofosfamide to induce cytotoxicity, indicating a causative relationship between evofosfamide treatment and pO₂ improvement in tumor. Further, a median pO₂ threshold of ∼15 mmHg established from the intercept of the trend line on the horizontal axis indicates the initial pO₂ level at which sensitive tumors can benefit from evofosfamide treatment (Fig. 6E, F). Below this point, the growth of the tumor in the 5 days between the start of treatment and the second imaging step leads to a natural reduction in pO₂ levels and an increase in the hypoxic fraction. The fragmentation efficiency at 15 mmHg is ∼50% (41), suggesting that the majority of locally distributed evofosfamide must be fragmented for a therapeutic effect.

It is noteworthy that evofosfamide-resistant Su.86.86 tumors did not exhibit a pO₂ change after treatment even when the pO₂ level was comparatively low (Fig. 6G, H). This result is consistent with the result of the tumor growth experiment in which evofosfamide was ineffective in Su.86.86 tumors (Fig. 2). EPRI is therefore useful in both predicting evofosfamide sensitivity and monitoring treatment response.

Improvement in oxygenation is often attributed to improved perfusion. The K^trans values of MIA Paca-2 and Su.86.86 xenografts are appreciably different, suggesting higher permeability/perfusion in Su.86.86 xenografts in line with the higher pO₂ levels measured by EPRI and in agreement with previous studies (46). K^trans decreased in both models after treatment, possibly due to reduced VEGF production from selective targeting of the hypoxic fraction (4).

The decrease in K^trans is consistent with a decrease in perfusion/permeability after treatment, which should, other factors being absent, be expected to lead to a decrease in oxygenation. The opposite is in fact observed. This suggests that the dominant cause for reoxygenation is not an opening of the vascular network after a decrease in solid state pressure (19, 37, 38), but rather the direct cytotoxic effect of evofosfamide leading to either cell death or decreased oxygen consumption after cell cycle arrest.

Surprisingly, the relative sensitivity observed in vivo was also mirrored in vitro under aerobic conditions. Using a long drug exposure model (48 h, comparable with continuous i.p. administration regimen in vivo) in air-equilibrated plates, we evaluated hypoxia-independent toxicity in vitro under highly aerobic conditions (Fig. 3). Compared with the control group, MIA Paca-2 cells were still strongly inhibited relative to Su.86.86 cells at all the concentrations examined. Since both cell lines are under ambient air conditions in this experiment, this finding suggests that an additional cellular mechanism of treatment resistance independent of oxygen delivery exists in Su.86.86. Furthermore, Su.86.86 cells have an oxygen consumption rate roughly twice that of Mia Paca-2 cells (6, 47).

On the basis of oxygen consumption alone, Su86.86 would be expected to be more sensitive to evofosfamide than Mia Paca-2, the opposite of what is observed. An oxygen-independent difference must therefore exist. Many possibilities exist, including differences in the activity of one-electron reductases (45), differences in the efficiency of DNA repair, and differences in cell fate as a function of DNA damage. These differences in cellular metabolism are not readily accessible by imaging and must be accounted for by other methods. Understanding these differences is paramount as drugs that inhibit mitochondrial respiration have been found to be effective in enhancing the radiation response (2, 13, 33), which may prove an effective method for combination therapy (42).

Several regimens for evofosfamide combination therapy have been examined in preclinical studies and proven to be effective (16, 27, 41). Such preclinical studies usually used a frequent dosing of evofosfamide. Based on these studies, clinical trials also used weekly dosing regimens. As a result of such frequent dosing, gastrointestinal disorders such as nausea and vomiting were commonly observed in the phase I clinical trials for evofosfamide (21), which may limit compliance. As the decreased hypoxic fraction after evofosfamide treatment sensitizes the tumor to radiation, the synergistic action may reduce the needed dose of evofosfamide.

This hypothesis was examined in the experiment comparing the combination regimen with extended monotherapy regimen in Figure 8A. The result showed that alternating treatment of the combination therapy showed a comparative effect to extended monotherapy with evofosfamide (Fig. 8B). This suggested that when planning the combination therapy of radiation and evofosfamide, decreasing frequency of evofosfamide treatment may decrease the systemic toxicity without compromising the treatment effect. With methodologies to provide quantitative assessment of tumor oxygen with T1-weighted MRI on conventional clinical scanners implemented on human subjects (29, 31) and the possibility of modifying pO₂ levels by hyperbaric gas interventions (23, 32, 48), such strategies will play a key role in tailoring therapies with evofosfamide and in combination with antiproliferative therapies such as conventional chemotherapy or RT (18).

Materials and Methods

Chemicals

Evofosfamide was purchased from Threshold Pharmaceuticals. Gd-DTPA was purchased from Bio-PAL, Inc. The triarylmethyl EPR oxygen tracer OX063 (methyl-tris[8-carboxy-2,2,6,6-tetrakis[2-hydroxyethyl]benzo[1,2-d:4,5-d0]-bis[1,3]dithiol-4-yl]-trisodium salt) was obtained from GE Healthcare.

In vitro cytotoxicity assay

Exponentially growing cells were seeded into a six-well plate (10,000 cells/well) 6 h before the addition of evofosfamide. After the drug addition, the cells were incubated under aerobic, air-equilibrated conditions for 48 h at 37°C in a standard tissue culture incubator. Cellular cytotoxicity was assessed by counting viable cells after treatment and normalized to control conditions.

Animal experiments

All animal experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animal Resources (National Research Council, 1996), and the experimental protocols were approved by the National Cancer Institute Animal Care and Use Committee (RBB-159-2SA). Female athymic nude mice were supplied by the Frederick Cancer Research Center, Animal Production. Human pancreatic adenocarcinoma MIA Paca-2 cells and Su.86.86 (obtained from Threshold Pharmaceuticals) were authenticated in May 2013 by RADIL using a panel of microsatellite markers.

Cells were routinely cultured in RPMI 1640 with 10% fetal calf serum. The tumors were formed by injecting 3 × 10⁶ subcutaneously into the right hind legs of female athymic nude mice. Tumor-bearing mice were treated daily with the intraperitoneal administration of 50 mg/kg evofosfamide 5 days a week when tumor size reached ∼400 mm³. In the imaging experiments, mice were anesthetized by isoflurane inhalation (4% for inducing and 1%–2% for maintaining anesthesia) and positioned prone with their tumor-bearing legs placed inside the resonator. During EPRI and MRI measurements, the breathing rate of the mouse was monitored with a pressure transducer (SA, Instruments, Inc.) and maintained at 60 ± 20 breaths/min. Core body temperature was maintained at 36°C ± 1°C with a flow of warm air.

Immunohistochemistry

Frozen tumor sections were thawed at room temperature, then fixed with ice-cold acetone for 10 min. After blocking, the sections were incubated with Hypoxyprobe^™ rabbit antipimonidazole antibody (1:250; Hypoxyprobe, Inc.) for pimonidazole staining or rat antimouse CD31 antibody (1:250; BD) for CD31 staining overnight at 4°C. Fluorescence microscopy and imaging was performed using a BZ-9000 BIOREVO (KEYENCE). Images were captured with the BZ-9000E viewer at 10 × magnification and were stitched to compose a whole image of the sections using the BZ-II Analyzer. Quantification of the pimonidazole-positive fraction was completed by counting the pixels of the positive area.

Electron paramagnetic resonance imaging

The technical details of the EPR scanner and oxygen image reconstruction were described in earlier reports (10, 11, 24, 39). Homemade resonators tuned to 300 MHz were used for EPRI. After the mouse was placed in the resonator, the EPR oxygen tracer OX063 (1.125 mmol/kg bolus) was injected intravenously under isoflurane anesthesia. The repetition time was 8.0 μs. The free induction decay (FID) signals were collected following the radiofrequency excitation pulses under a nested looping of the x, y, and z gradients, and each time point in the FID underwent phase modulation enabling three-dimensional spatial encoding. Since FIDs last for 1–5 μs, it is possible to generate a sequence of T2* maps, that is, EPR linewidth maps, which linearly correlate with the local concentration of oxygen and enable the pixel-wise estimation of pO₂.

Dynamic contrast enhancement MRI

DCE MRI studies were performed on a 1.0T scanner (ICON; Bruker Bio-Spin MRI GmbH). T1-weighted fast low angle shot (FLASH) images were obtained with echo time = 6 ms, repetition time (TR) = 118 ms, a flip angle of 30°, two slices, 0.4 mm × 0.4 mm resolution, 15 s acquisition time per image, and 45 repetitions. Gd-DTPA solution (0.25 mmol/kg body weight) was injected through a tail vein cannula 1 min after the start of the dynamic FLASH sequence. To determine the local concentrations of Gd-DTPA, T1 maps were calculated from three sets of Rapid Imaging with Refocused Echoes (RARE) images obtained with TR = 500, 1000, and 2000 ms, with the acquisitions being made before running the FLASH sequence. The endothelial transfer coefficient K^trans was calculated by fitting the dynamic MRI signal change to the standardized Tofts model (30, 44).

Statistics

Data were expressed as the means ± standard deviation. The paired Student's t-test was used to compare the values before and after treatment in the same group of mice in EPRI and MRI studies. Other comparisons of means, which required independent sets of mice, were analyzed using the independent Student's t-test. The log-rank test was used to compare the distribution of Kaplan–Meier survival curves. The event for the Kaplan–Meier test was death or the tumor 2000 mm³, mice requiring early or unscheduled euthanasia due to unrelieved pain/distress were treated as censored. p < 0.05 was considered statistically significant.

Abbreviations Used

Br-IPM

bromo-isophosphoramide

DCE MRI

dynamic contrast-enhanced magnetic resonance imaging

EPRI

electron paramagnetic resonance imaging

FID

free induction decay

FLASH

fast low angle shot

HAPs

hypoxia-activated prodrugs

HF10

hypoxic fraction <10 mmHg

PDAC

pancreatic ductal adenocarcinoma

RT

radiation therapy

TR

repetition time

Author Disclosure Statement

There is no conflict of interest.

Funding Information

This work was supported by the intramural research program of National Cancer Institute/National Institutes of Health (grant no. 1ZIABC010476).

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2