Administration of Hypoxia-Activated Prodrug Evofosfamide After

Published OnlineFirst February 23, 2018; DOI: 10.1158/1078-0432.CCR-17-1715

Cancer Therapy: Preclinical Clinical Cancer Research Administration of Hypoxia-Activated Prodrug Evofosfamide after Conventional Adjuvant Therapy Enhances Therapeutic Outcome and Targets Cancer-Initiating Cells in Preclinical Models of Colorectal Cancer Jennifer Haynes1, Trevor D. McKee1,2, Andrew Haller1, Yadong Wang1, Cherry Leung1, Deena M.A. Gendoo1,3, Evelyne Lima-Fernandes4, Antonija Kreso1, Robin Wolman1, Eva Szentgyorgyi5, Douglass C.Vines1,2,6, Benjamin Haibe-Kains1,3,7, Bradly G.Wouters1,3,6, Ur Metser1,8,9, David A. Jaffray1,2,3,6,9, Myles Smith10, and Catherine A. O'Brien1,11,12,13

Abstract

Purpose: Cancer-initiating cells (C-IC) have been described PET imaging was utilized as a noninvasive method to assess in multiple cancer types, including colorectal cancer. C-ICs intratumoral hypoxia. are defined by their capacity to self-renew, thereby driving Results: Hypoxia was sufficient to drive the formation of CC- tumor growth. C-ICs were initially thought to be static entities; ICs and colorectal cancer cells surviving conventional therapy however, recent studies have determined these cells to be were more hypoxic and C-IC-like. Using a novel approach dynamic and influenced by microenvironmental cues such as to combination therapy, we show that sequential treatment with hypoxia. If hypoxia drives the formation of C-ICs, then ther- 5-FU or CRT followed by evofosfamide not only inhibits tumor apeutic targeting of hypoxia could represent a novel means to growth of xenografts compared with 5-FU or CRT alone, but also target C-ICs. significantly decreases the CC-IC fraction. Furthermore, nonin- Experimental Design: Patient-derived colorectal cancer xeno- vasive FAZA-PET hypoxia imaging was predictive of a tumor's grafts were treated with evofosfamide, a hypoxia-activated pro- response to evofosfamide. drug (HAP), in combination with 5-fluorouracil (5-FU) or che- Conclusions: Our data demonstrate a novel means to target the moradiotherapy (5-FU and radiation; CRT). Treatment groups CC-IC fraction by adding a HAP sequentially after conventional included both concurrent and sequential dosing regimens. Effects adjuvant therapy, as well as the use of FAZA-PET as a biomarker on the colorectal cancer-initiating cell (CC-IC) fraction were for hypoxia to identify tumors that will benefit most from this assessed by serial passage in vivo limiting dilution assays. FAZA- approach. Clin Cancer Res; 1–12. 2018 AACR.

1Princess Margaret Cancer Centre, University Health Network, Toronto, Ontario, Introduction Canada. 2STTARR Innovation Centre, University Health Network, Toronto, Ontario, Canada. 3Department of Medical Biophysics, University of Toronto, Hypoxia is a common feature of many solid tumors and is Toronto, Ontario, Canada. 4Structural Genomics Consortium, Toronto, Ontario, often associated with tumor aggressiveness and therapeutic resis- Canada. 5Department of Pathology, University Health Network, Toronto, tance (1–4). More recently, tumor hypoxia was linked to increased Ontario, Canada. 6Department of Radiation Oncology, University of Toronto, self-renewal capacity, the canonical feature deﬁning cancer-initi- Toronto, Ontario, Canada. 7Department of Computer Science, University of 8 ating cells (C-IC; refs. 5, 6). Aside from increasing the self-renewal Toronto, Toronto, Ontario, Canada. Department of Medical Imaging, University capacity of existing C-ICs, hypoxia was shown to promote the of Toronto, Toronto, Ontario, Canada. 9Techna Institute for the Advancement of Technology for Health, University Health Network, Toronto, Ontario, Canada. acquisition of a C-IC like phenotype in a wide range of solid 10Department of Surgery, The Royal Marsden Hospital and Institute of Cancer tumors (7, 8). In the context of colorectal cancer, exposure Research, London, United Kingdom. 11Department of Laboratory Medicine and to hypoxia results in increased nuclear localization and expression Pathobiology, University of Toronto, Toronto, Ontario, Canada. 12Department of of b-catenin, a marker of the colorectal cancer–initiating cell (CC- 13 Physiology, University of Toronto, Toronto, Ontario, Canada. Department of IC) fraction (9). Hypoxia has also been shown to block cellular Surgery, University Health Network, Toronto, Ontario, Canada. differentiation by suppressing expression of CDX1, a differenti- Note: Supplementary data for this article are available at Clinical Cancer ation marker in colorectal cancer, and inducing expression of BMI- Research Online (http://clincancerres.aacrjournals.org/). 1, a key regulator of CC-IC self-renewal (10). Functional assess- Corresponding Author: Catherine A. O'Brien, Princess Margaret Cancer ment of C-ICs in glioblastoma and breast cancers have also shown Research Tower, 101 College St, Room 13-704, Toronto, ON M5G 1L7, Canada. that exposure to hypoxia results in increased C-IC numbers as Phone: 416-581-7537; Fax: 416-340-3808; E-mail: measured by in vivo limiting dilution assays, the gold standard [email protected] assay for self-renewal (11–14). Collectively, these studies indicate doi: 10.1158/1078-0432.CCR-17-1715 that hypoxia is sufﬁcient to drive acquisition of self-renewal 2018 American Association for Cancer Research. capacity in a number of solid tumor-initiating cell subsets.

www.aacrjournals.org OF1

Haynes et al.

Translational Relevance Materials and Methods Colorectal cancer patient-derived xenografts Despite decades worth of research and clinical trials, target- Human colorectal cancer tissue was obtained with informed ing hypoxia has yet to become a standard part of cancer patient consent, as approved by the Research Ethics Board at the treatment. In this study, we show that pretreatment with a University Health Network in Toronto, and processed as 4-day course of 5-fluorouracil (5-FU) resulted in colorectal described previously (31). A summary of patient samples used cancer cells being driven into a hypoxic cancer-initiating in this study is provided in Supplementary Table S1. To establish cell (C-IC) state, which was exquisitely sensitive to the hyp- and maintain PDX models, cells from freshly dissociated colo- oxia-activated prodrug evofosfamide. Using limiting dilution rectal cancer tissue or freshly thawed previously frozen xenograft assays, we demonstrate that sequential treatment with either 5- samples (31) were mixed (1:1) with high concentration Matrigel FU or chemoradiotherapy (CRT) followed by evofosfamide (Corning) and injected subcutaneously (s.c.) into the flanks of specifically targeted the colorectal C-IC fraction. Furthermore, NOD-SCID mice (male or female, 6–8 weeks of age). All animal we identify FAZA-PET as a biomarker for hypoxia that can be experiments were reviewed and approved by the Animal Care used to identify colorectal cancers that will benefit most from Committee at the University Health Network in Toronto. the addition of evofosfamide. Future clinical trials are war- fi ranted to validate our ndings in the context of colorectal Primary cell culture and treatments cancer patients. Patient-derived cell lines established from xenografts (32) or directly from patient tissue were cultured as previously described (31). For hypoxic conditions, cells were maintained at 2% O2 for 7 or 10 days. All cell lines were authenticated using short tandem A canonical feature of C-ICs is decreased response to standard- repeat profiling, and proven to be negative for mycoplasma. For of-care chemotherapy regimens in a range of solid tumors. In the in vitro studies, 5-fluorouracil (5-FU, Sigma) and evofosfamide context of colorectal cancer, we and others have demonstrated (Threshold Pharmaceuticals) were dissolved in DMSO and used fl that standard-of-care chemotherapy agents, such as 5- uorouracil at the concentration indicated in the figure legend for 4, 7, or (5-FU) and oxaliplatin, target more differentiated cancer cells 10 days. Cells were irradiated using an X-RAD 225Cx preclinical – while relatively sparing the CC-IC fraction (2, 3, 15 18). In irradiator (Precision X-Ray). X-rays were collimated using a 4 4 addition to chemoresistance, C-ICs have also been shown to be cm collimator (tube voltage of 225 kV, tube current of 13 mA; relatively radioresistant. Previous work by Bao and colleagues dose rate of 3.215 Gy/min). demonstrated that glioblastoma C-ICs are highly radioresistant compared with the non-C-IC fraction and as a result have Generation of TCF-GFP reporter cells and flow cytometry increased survival postradiotherapy (19, 20). Interestingly, it is Cells were stably transduced with TCF/LEF-GFP reporter well established that like C-ICs, hypoxic cancer cells also dem- or negative control lentivirus particles (Cignal Lenti Reporter, onstrate chemoresistance and radioresistance (2, 4, 21, 22). These Qiagen) as described previously (31). Reporter cells were pro- similarities led us to question whether C-ICs could be specifically cessed as described previously (31) and analyzed using a BD LSR II targeted using the hypoxia-activated prodrug (HAP; refs. 21, 23) flow cytometer. evofosfamide (previously known as TH-302). Evofosfamide is a HAP composed of 2-nitroimidazole conjugated to the cytotoxin qRT-PCR bromo-isophosphoramide mustard that is selectively activated Real-time PCR was performed as described previously (32), under hypoxic conditions (24, 25), and increases the antitumor using SensiFAST SYBR Hi-ROX qPCR kit (FroggaBio) to amplify activity of multiple chemotherapeutic agents in various preclin- cDNA. Housekeeping genes 18S rRNA and TBP were used for ical human tumor xenograft models (26–29). Interestingly, the normalization. Primer sequences are listed in Supplementary concept of utilizing HAPs for cancer is not new and despite Material and Methods. decades worth of research, hypoxia-targeting agents are still not used as standard of care in cancer treatment (4, 21–23). One of the Limiting dilution assays major hurdles in the field of HAP research and its clinical appli- LDAs were performed as described previously (31). For in vitro cation is to understand how to combine HAPs with standard-of- LDAs, live cells were sorted into 96-well plates at 100, 10, or 1 cell care therapies to maximize therapeutic response (23, 26). Another per well, using FACSAria cell sorters and SYTOX Blue (Thermo major hurdle is the selection of patients that will benefit most Fisher Scientific) to exclude dead cells. Plates were incubated in from targeting hypoxia (4, 21, 30). It is evident that there is a wide normoxia or hypoxia for 3–5 weeks, and then wells containing range of hypoxia at baseline in solid tumors. Therefore, identi- spheres were counted. For in vivo LDAs, cells either from cultures fying a clinical biomarker of hypoxia that predicts response to grown in normoxia or hypoxia for 7 days, or from in vivo–treated HAPs could help predict outcome and determine the optimal xenografts were dissociated into single cells, serially diluted, and treatment course. injected subcutaneously into the flanks of NSG mice (male, 6–8 Evidence suggests that hypoxia could be a driver of the C-IC weeks of age) at doses indicated in the figure legend. Sphere- phenotype in colorectal cancer, with CC-ICs preferentially sur- and cancer-initiating cell frequencies and probability estimates viving in the hypoxic niche (6, 9, 10). Therefore, we questioned were calculated using ELDA software (http://bioinf.wehi.edu.au/ whether CC-ICs that survive chemotherapy are also characterized software/elda/). by a relative increase in hypoxia. If this is the case, then chemotherapeutic agents could be utilized to drive cancer cells into a Caspase-3/7 activity and viability assays "hypoxic CC-IC state," which in turn could be exploited to Caspase-3/7 activity and viability assays were performed on augment the response to HAPs such as evofosfamide. cells seeded in 96-well plates at 2,500 or 5,000 cells/well, in

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

Targeting Hypoxia after Therapy Decreases Colorectal Cancer-ICs

quadruplicate. Caspase-Glo 3/7 reagent (Promega) or CellTiter- conjugated a-pimo antibody (1:100, Hypoxyprobe) and DAPI to Glo 3D reagent (Promega) was added, plates were shaken for 5 label nuclei. Images were acquired using a TissueScope confocal minutes, and then incubated at room temperature for 1 hour fluorescence whole slide scanner (Huron Digital Pathology) at (caspase) or 25 minutes (viability). Luminescence was measured 10 magnification (1-mm resolution). Images were analyzed in a CLARIOstar microplate reader (BMG LABTECH). using TissueStudio 3.51 software (Definiens) as described previously (33). In vitro immunofluorescence staining, image acquisition, and quantification of gH2AX foci [18F]-fluoroazomycin arabinoside positron emission Cells were prepared for immunofluorescence as described tomography hypoxia imaging previously (32), using primary antibodies for gH2AX (1:500, FAZA-PET/CT imaging was used to estimate baseline hypoxia in Millipore) and nonphosphorylated (active) b-catenin (1:400, approximately 100 mm3 xenografts one day before the treatment Cell Signaling Technology). Cells were imaged using a 60 commenced. Mice were injected intravenously with 10–15 MBq Plan-Apochromat/1.4 NA oil immersion objective on an of [18F]-FAZA in <150 mL saline containing 10% (v/v) ethanol. inverted laser-scanning confocal microscope (LSM700; Carl After 2 hours, mice were anesthetized and scanned for 20 minutes Zeiss), and images were captured using LSM Zen 2012 Software as described previously (33). FAZA-PET images were fused with (Zeiss). The number of gH2AX foci per nucleus was quantified CT images, and quantified by measuring the maximum % injected using Z-Series projections of confocal images combined from dose/gram (%ID/g) in the tumor region, using a 100 mm3 region 16 optical sections acquired at 0.9-mm intervals, and ImagePro of the paraspinal muscle as a control. FAZA uptake was expressed Plus Software (Media Cybernetics). A minimum of 50 nuclei as tumor maximum (T ) to muscle mean (Mmean) (T/M) ratios per sample were counted. max (%ID/g).

In vivo dosing of chemotherapy and chemoradiotherapy Tumor cell suspensions (2.5 105 for POP92 or 1 106 for Statistical analysis CSC91 and POP74) were mixed (1:1) with high concentration For all experiments except LDAs, statistical analysis was per- Matrigel (Corning) and injected subcutaneously into upper left formed using Prism 6 Software (GraphPad). P values were derived and right flanks of female Ncr/nude mice (Taconic), 8–10 using two-tailed Student t tests, one-way ANOVA followed by weeks of age (2 injections per mouse, 7 mice per treatment Tukey multiple comparisons test, or log-rank test. For LDAs, group). When the average tumor volume reached 100 mm3, frequency determinations and P values were generated using tumor-bearing mice were randomized into control and treat- ELDA software (Walter and Eliza Hall Institute). For all compar- ment groups based on tumor volumes, and dosing commenced isons, a P value of <0.05 was considered statistically significant. on day 0. For in vivo studies, 5-FU and evofosfamide were dissolved in saline and administered by intraperitoneal (i.p.) Results injection. Individual treatment regimens were as follows: saline Hypoxia activates Wnt/b-catenin signaling and enriches þ (control), 5-FU (30 mg/kg 5days),CRT(20mg/kg 2Gy for CC-ICs 5 days), evofosfamide (50 mg/kg 10 days); combination Previous studies have shown that hypoxic stem cells have therapies were given either concurrently (both started on day 0, enhanced Wnt/b-catenin signaling (34) and that high Wnt activity evofosfamide administered 4 hours before 5-FU or CRT) or functionally designates the CC-IC population (35). To confirm sequentially (5-FU or CRT started on day 0, and evofosfamide that our patient-derived colorectal cancer spheroid cultures main- started on day 4). For radiotherapy, mice were anesthetized tain these properties when grown in vitro under conditions that fl with iso urane and placed onto the stage of an X-RAD 225Cx enrich for CC-ICs, we stably transduced three models with a small-animal image-guided irradiator (Precision X-Ray), in lentiviral TCF/LEF promoter-driven GFP reporter (TCF-GFP) and prone position. A cone beam CT scan was taken of the tumor cultured cells under normoxic (21% O2) and hypoxic (2% O2) region, and used to plan a dose of 1 Gy each administered from conditions. Exposure to hypoxia for 10 days significantly the dorsal and ventral sides. A 1-cm circular collimator was increased TCF-GFP reporter activity (2.2- to 4.7-fold) in all three chosen to provide uniform dose across the tumor while min- models (Fig. 1A and B). In addition, hypoxic exposure also imizing normal tissue exposure. No detectable skin or normal upregulated expression of stem genes c-MYC and KLF4 (Supple- tissue toxicity was observed over the course of therapy. Body mentary Fig. S1A), and stabilized hypoxia-inducible factor (HIF) – weights were measured every 1 3 days over the course of proteins (Supplementary Fig. S1B). To evaluate the effects of treatment, and tumor growth was monitored by caliper mea- hypoxia on CC-IC function, we performed in vitro LDAs under – surements every 2 7 days until endpoint was reached. normoxic and hypoxic conditions. Exposure to hypoxia significantly increased sphere formation (1.6- to 7-fold) in all three In vivo pimonidazole immunofluorescence staining, image models tested (Fig. 1C; Supplementary Table S2). To confirm our acquisition, and quantification in vitro results, we performed in vivo LDAs using spheres cultured in Mice were administered pimo (60 mg/kg; Hypoxyprobe) by normoxia or hypoxia for 7 days, injected at limiting dilution into intraperitoneal injection 90 minutes prior to sacrifice. Tumors NSG mice. Both models preexposed to hypoxia displayed a were embedded in O.C.T. compound (Tissue-Tek) and snap significant increase in CC-IC frequency (3.1- and 14-fold) com- frozen, or fixed in 10% neutral buffered formalin, dehydrated, pared with normoxic controls (Fig. 1D; Supplementary Table S3). and embedded in paraffin. Heat-induced epitope retrieval was Collectively, these data show that culturing colorectal cancer cells performed on formalin-fixed paraffin-embedded tissue sections under hypoxic conditions results in an increased number of in citrate buffer. Tissue sections (5-mm) were stained with FITC- phenotypic and functional CC-ICs.

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Haynes et al.

A B POP66 POP92 POP181 *** N 5 100 100 100 * H 4 80 80 80 3 * 60 60 60 N 2 40 40 40 H fluorescence Relative GFP 1 20 20 20 0 Relative to mode (%) 0 0 0 -103 100 103 104 105 -103 100 103 104 105 -103 100 103 104 105 POP66 POP92 GFP GFP GFP POP181

C POP66 POP92 POP181 D POP66 POP92 1 1 1 1 1

* 0.1 0.1 ** 0.1 0.1 0.1 *** *** ** 0.01 0.01 0.01 0.01 0.01 cell frequency cell frequency cell frequency C-IC Frequency C-IC Frequency Sphere-initiating Sphere-initiating Sphere-initiating 0.001 0.001 0.001 0.001 0.001 NH NH NH NH NH

Figure 1. Hypoxia activates Wnt/b-catenin signaling and increases the frequency of CC-ICs. A, GFP intensity in colorectal cancer cells expressing a lentivirally transduced TCF/ LEF transcriptional reporter (TCF-GFP) to monitor Wnt/b-catenin pathway activation. GFP fluorescence was measured by flow cytometry after cells were cultured for 10 days in normoxia (N, 21% O2) or hypoxia (H, 2% O2). B, Quantification of relative GFP median fluorescence intensity for cells shown in A. Values are relative to normoxia control. Data are shown as mean SEM of at least three independent experiments. Student t test was used for statistical significance. C, Sphere- initiating cell frequency of colorectal cancer cells, as measured by LDA in vitro. Colorectal cancer cells were seeded at 100, 10, or 1 cell per well, and cultured in normoxia or hypoxia. D, C-IC frequency of colorectal cancer cells, as measured by LDA in vivo. Colorectal cancer cells were cultured in normoxia or hypoxia for 7 days, dissociated into single cells, and injected subcutaneously into NSG mice at doses of 20,000, 10,000, 1,000, 100, and 10 cells (n ¼ 5 mice, 2 injections per mouse). Data are shown as mean and 95% confidence interval (CI). Frequency and probability estimates were computed using ELDA software (, P < 0.05; , P < 0.01; , P < 0.001).

Chemotherapy enriches for the CC-IC phenotype and hypoxic demonstrate that treatment with 5-FU increases the proportion tumor cells of both hypoxic tumor cells and those with high levels of Wnt/ As both CC-ICs and hypoxia are thought to contribute to b-catenin signaling. chemotherapy resistance (1–4, 18), we tested the effect of the chemotherapy drug 5-FU on the CC-IC phenotype and tumor hypoxia. Colorectal cancer cells treated in vitro with 5-FU (at the Evofosfamide increases the efficacy of chemotherapy or IC50 for each model) for 10 days showed a statistically significant radiation in vitro increase in TCF-GFP reporter activity (1.7- to 6.9-fold) in all three As 5-FU–treated tumors showed an enrichment of the hypoxic models tested (Fig. 2A and B). Consistently, cells cultured in the fraction and hypoxia is known to contribute to chemotherapeutic presence of 5-FU also showed upregulated expression of Wnt resistance (1–4), we asked whether the addition of a HAP would target genes c-MYC and LEF1 (Fig. 2C) and their corresponding increase the efficacy of conventional therapy. 5-FU is a pyrimidine proteins (Supplementary Fig. S2A). In addition, we observed analogue that causes DNA damage by inhibiting thymidylate increased expression of CC-IC surface marker CD133 for both synthase, which disrupts DNA synthesis and repair, whereas spheroid models that express this CC-IC phenotypic marker ionizing radiation directly induces DNA double-strand breaks. (Supplementary Fig. S2B and S2C). Under hypoxic conditions, evofosfamide is converted to the active To determine whether exposure to 5-FU enriches for hypoxic drug bromo-isophosphoramide mustard, which cross-links DNA tumor cells, we injected colorectal cancer spheroid cultures or and renders cells unable to replicate their DNA and divide. To patient-derived xenografts (PDX) subcutaneously into immuno- assess whether evofosfamide enhances 5-FU- or radiation- deficient nude mice. Once tumors were approximately 100 mm3, induced DNA damage in vitro, POP92 spheroid cultures were mice were treated with saline or 5-FU for 5 days, then subse- treated with 5-FU, radiation (X-RAD), or either agent in combi- quently injected with the 2-nitroimidazole hypoxia tracer pimo- nation with evofosfamide. After 4 days, spheres were fixed and nidazole (pimo) and sacrificed 24 hours after the last treatment. labeled with immunofluorescent antibodies for gH2AX to mark We observed an increase in the pimo-positive hypoxic fraction sites of DNA damage (Fig. 3A). In the vehicle control group of cancer cells in tumors exposed to 5-FU (1.6- to 6.4-fold), as (DMSO), there were no gH2AX foci in approximately 50% of compared with control tumors in all four models tested (Fig. 2D cells and one or more foci per cell for the remaining half. As and E). Consistently, cells cultured in the presence of 5-FU also expected, treatment with the single agents alone increased the showed significantly upregulated expression of HIFs (Fig. 2F; proportion of cells with gH2AX foci (Fig. 3B; 87% for 5-FU, 71% Supplementary Fig. S2D) and HIF target genes CXCR4, GLUT1, for X-RAD, or 71% for evofosfamide vs. 51% for DMSO). Treat- and OCT4 (Supplementary Fig. S2E). Taken together, these data ment with 5-FU or X-RAD in combination with evofosfamide

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

Targeting Hypoxia after Therapy Decreases Colorectal Cancer-ICs

A B POP66 POP92 POP181 DMSO 10 100 100 100 5-FU * 8 80 80 80 6 60 60 60 DMSO 4 40 40 40 5-FU * ** fluorescence Relative GFP 2 20 20 20 0 Relative to mode (%) 0 0 0 -103 100 103 104 105 -103 100 103 104 105 -103 100 103 104 105 POP66 POP92 GFP GFP GFP POP181

C c-MYC D POP92 POP66 CSC134 CSC91 DMSO 4 5-FU 3 * 2 * 1 Saline

Relative mRNA levels Relative mRNA 0

POP66 POP92 POP181 LEF1 Pimo / DAPI DMSO 25 * 5-FU 20 5-FU 15 10 * 5 **

Relative mRNA levels Relative mRNA 0

POP66 POP92 E POP181 F HIF-1A HIF-2A 50 Saline DMSO DMSO 2.0 4 5-FU *** 40 5-FU 5-FU * 1.5 3 * * ** ** 30 *** * 1.0 2 * 20 0.5 1 10 Pimo/tumor area (%) Relative mRNA levels Relative mRNA 0 levels Relative mRNA 0 0 POP66 POP92 POP66 POP92 POP181 POP181

POP92 POP66 CSC91 CSC134

Figure 2. Chemotherapy activates Wnt/b-catenin signaling and induces hypoxia. A, GFP intensity of colorectal cancer cells expressing the TCF-GFP reporter. GFP fluorescence was measured by flow cytometry after cells were cultured for 10 days in DMSO (control) or 5-FU (1 mmol/L for POP66/92, 0.5 mmol/L for POP181). B, Quantification of relative GFP median fluorescence intensity for cells shown in A. Values are relative to DMSO control. Data are shown as mean SEM of at least three independent experiments. C, qRT-PCRanalysisofWnttarget(c-MYC, LEF1) gene expression in colorectal cancer cells cultured for 10 days in the presence of 5-FU. Values are relative to DMSO control and normalized to 18S rRNA levels. Data are shown as mean SEM (n ¼ 3 independent experiments). D, Representative images of pimonidazole (pimo) immunofluorescent staining of colorectal cancer PDXs grown subcutaneously in nude mice treated with 5-FU (30 mg/kg 5 days). Mice were injected with pimo (60 mg/kg) 16 hours after the last treatment, and tumors were harvested, fixed, and stained with a-pimo antibody (green) and DAPI to label nuclei (blue). Scale bar, 1 mm. E, Quantification of pimo staining of colorectal cancer PDXs from mice treated with 5-FU. Horizontal lines indicate mean values and error bars represent SEM (n ¼ at least 3 biological replicates each, 1–3 slides per tumor). F, qRT-PCR analysis of hypoxia-inducible factor (HIF-1A, HIF-2A) gene expression in colorectal cancer cells cultured for 7 days in the presence of 5-FU. Values are relative to DMSO control and normalized to 18S rRNA levels. Data are shown as mean SEM (n ¼ at least 4 independent experiments). Student t test was used for statistical significance (, P < 0.05; , P < 0.01; , P < 0.001).

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Haynes et al.

A DMSO5-FUX-RAD Evo 5-FU + Evo X-RAD + Evo H2AX γ DAPI Active β -catenin Merge

B 80 C 4 0 foci * DMSO 1–50 foci *** 5-FU 60 >50 foci 3 Evo 5-FU + Evo 40 H2AX foci (%) 2

20 1 (relative to DMSO) Caspase-3/7 activity

Cells with γ 0 DMSO5-FUX-RAD Evo 5-FU X-RAD 0 POP66POP92 POP181 + Evo + Evo

Figure 3. Evofosfamide increases the efficacy of chemotherapy or radiation. A and B, Immunofluorescent staining showing gH2AX foci as a marker of DNA double-strand breaks. A, Images of POP92 colorectal cancer cells treated with 5-FU (2 mmol/L), X-RAD (2 Gy), evofosfamide (Evo, 1 mmol/L), or the indicated combinations for 4 days. Cells were fixed and immunostained for g-H2AX and non-phosphorylated (active) b-catenin, and nuclei were stained with DAPI. Images were acquired by confocal microscopy using a 60 objective. Arrowheads indicate cells with high-level nuclear b-catenin and >50 gH2AX foci. Scale bar, 10 mm. B, Quantification of gH2AX foci per nucleus for cells in A (n ¼ at least 50 cells per slide, 2–3 replicate slides per group). C, Caspase-3/7 activity of POP66, POP92, and POP181 colorectal cancer cells after 4 days in 5-FU (2 mmol/L), evofosfamide (0.5 mmol/L), or the combination. Values are relative to DMSO control and normalized to viability. Data are shown as mean SEM (n ¼ 3 independent experiments). Two-way ANOVA followed by Tukey multiple comparisons test was used for statistical significance (, P < 0.05; , P < 0.001).

resulted in a further increase in the proportion of cells with gH2AX Sequential dosing with evofosfamide potentiates foci (97% for 5-FU þ evofosfamide or 91% for X-RAD þ evo- standard-of-care agents in vivo fosfamide). This included a 2.0- to 5.2-fold increase in the To validate our in vitro results, we injected POP92 spheroid proportion of cells with >50 foci compared with either agent cultures into nude mice and monitored the growth of tumor given alone (49% for 5-FU þ evofosfamide vs. 24% for 5-FU xenografts in the absence or presence of standard-of-care thera- alone, or 31% for X-RAD þ evofosfamide vs. 6% for X-RAD pies, either alone or in combination with evofosfamide. Once alone). Importantly, in the groups receiving evofosfamide in tumors reached an average volume of approximately 100 mm3, combination with 5-FU or X-RAD, costaining with antibodies mice were randomized into the following treatment groups: for active b-catenin revealed the presence of high-level nuclear saline (control), 5-FU or chemoradiotherapy (CRT), evofosfa- b-catenin in cells with >50 gH2AX foci, suggesting that CC-ICs mide þ 5-FU or CRT þ evofosfamide. In addition, each combi- were not being spared in these groups (Fig. 3A). Consistent with nation group included two different dosing regimens: concurrent, increased DNA damage, colorectal cancer cells treated in vitro with in which 5-FU and evofosfamide were administered on the same the combination of 5-FU þ evofosfamide for 4 days had increased day, and sequential, where evofosfamide was administered 4 days levels of caspase-3/7 activity compared with 5-FU alone (Fig. 3C), after the start of 5-FU (Fig. 4A). We observed minimal effects on indicating enhanced apoptosis in the combination group. Togeth- growth of tumors treated with 5-FU, evofosfamide, or the com- er, these data demonstrate that evofosfamide increases the in vitro bination given concurrently, as indicated by similar growth plots efﬁcacy of conventional chemotherapy or radiation. and time to reach 500 mm3 (Supplementary Fig. S3; Fig. 4B). In

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

Targeting Hypoxia after Therapy Decreases Colorectal Cancer-ICs

A Concurrent Sequential

5-FU or CRT 5-FU or CRT Evo Evo

Day 0821416 0214Day 0821416 0214

B C POP92 + 5-FU POP92 + CRT

100 Saline 100 Saline ) 3 )

5-FU 3 CRT Evo Evo ** 50 5-FU + Evo con 50 CRT+ Evo con * 5-FU + Evo seq CRT+ Evo seq below 500 (mm Percent of tumors below 500 (mm Percent of tumors 0 0 03015 45 60 03015 45 60 Time after treatment (days) Time after treatment (days)

Figure 4. Sequential dosing of chemotherapy or chemoradiotherapy and evofosfamide (Evo) is more effective than concurrent dosing in vivo. A, Schematic representation of chemotherapy (5-FU) or chemoradiotherapy (CRT) regimens given in vivo in combination with evofosfamide either concurrently or sequentially. B and C, Kaplan– Meier survival curves showing tumors less than 500 mm3 after treatment. Nude mice were injected subcutaneously with POP92 colorectal cancer cells and given indicated treatments when the average tumor volume reached 100 mm3 (n ¼ 5 mice, 2 injections per mouse). B, Saline (control), 5-FU (30 mg/kg 5 days), Evo (50 mg/kg 10 d), or the combination given concurrently (con) or sequentially (seq). C, Saline (control), CRT (5-FU, 20 mg/kg þ RT, 2 Gy 5 d), Evo (50 mg/kg 10 d), or the combination given con or seq. The log-rank test was used for statistical significance. , P < 0.05; , P < 0.01, for the 5-FU or CRT þ Evo combination groups given seq versus con. contrast, mice treated sequentially with the combination exhib- sample, there was no statistically significant difference in the ited tumor growth inhibition, as indicated by a significantly tumor growth inhibition between evofosfamide alone versus 5- longer time for tumors to reach 500 mm3. There was an even FU þ evofosfamide treatment groups. The increased sensitivity to more dramatic tumor growth inhibition when CRT and Evo were evofosfamide observed for POP74 likely reflects the finding that dosed sequentially (Fig. 4C; Supplementary Fig. S3). Strikingly, this PDX model possesses the highest baseline hypoxia of all the median time to reach 500 mm3 for the CRT and evofosfamide colorectal cancer xenografts tested in this study (Supplementary sequential group was 47 days compared with 19 days for the Fig. S4). For the rectal tumor model (CSC91), an endpoint of 400 combination given concurrently or 17 days for CRT alone. These mm3 was used due to the markedly slower growth compared with data demonstrate that introducing a HAP in a sequential dosing the previous models. Kaplan–Meier analysis showed that com- regimen after standard-of-care therapy further inhibits tumor bining standard-of-care treatment CRT þ evofosfamide resulted growth of colorectal cancer xenografts, which is consistent with in a longer time to 400 mm3 compared with either agent alone our finding that in vivo treatment with 5-FU resulted in a relative (Supplementary Fig. S5; Fig. 5B, median survival time >125 days increase in the hypoxic fraction after 5 days (Fig. 2D and E). for CRT þ evofosfamide vs. 67 days for evofosfamide alone, 54 days for CRT alone, or 34 days for the saline group). Of note, Evofosfamide combined with conventional therapy increases although we observed that the single agents alone also increased targeting of the CC-IC fraction the median survival time to 400 mm3, only the combination To determine whether the addition of evofosfamide in the treatment yielded a long-term survival benefit for this sample over sequential dosing regimen provides a survival benefit compared the course of the experiment. In total, only 3 of 8 (37.5%) with conventional therapies alone in other colorectal cancer xenografts in the combined therapy group reached 400 mm3 by models, we selected two samples from a panel of established the end of the study (125 days). colorectal cancer xenograft models exhibiting a wide range of To assess the effect of sequential dosing of the combination baseline intratumoral hypoxia for further in vivo testing. Relative treatments on the CC-IC fraction, we performed in vivo LDAs to our other PDX models, POP74 (colon cancer) and CSC91 using colorectal cancer cells isolated from xenografts described (rectal cancer) exhibit high and medium baseline levels of hyp- above. For each sample, one (POP74) or two (CSC91) mice oxia, respectively (Supplementary Fig. S4). For the colon tumor per treatment group were sacrificed 24 hours posttreatment and model (POP74), Kaplan–Meier analysis showed that treatment cells were injected at limiting dilution into NSG mice. POP74 with evofosfamide either alone or in combination with standard- tumors treated with the combination of 5-FU þ evofosfamide had of-care agent 5-FU resulted in a longer time to 500 mm3 compared a significantly impaired ability to repopulate secondary tumors with 5-FU alone or saline groups (Supplementary Fig. S5; Fig. 5A, compared with those treated with 5-FU alone (Fig. 5C; Supple- median survival time of 24 days for evofosfamide alone or 28.5 mentary Table S4; CC-IC frequency 1/30,415 vs. 1/2,644) days for 5-FU þ evofosfamide versus 20 days for 5-FU alone or evofosfamide alone (1/7,275). Similarly, CSC91 tumors trea- or 13 days for the saline group). Interestingly, unlike the previous ted with the combination of CRT þ evofosfamide showed a

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Haynes et al.

A POP74 + 5-FU (Colon cancer) C POP74 Figure 5. 0.1 Evo decreases the frequency of colorectal C-ICs 100 Saline *** in vivo. A and B, Kaplan-Meier survival curves

) showing tumors less than the indicated volume 3 5-FU 0.01 *** after treatment. Nude mice were injected Evo subcutaneously with colorectal cancer cells, 0.001 ** 50 * 5-FU + Evo then imaged and given indicated treatments when the average tumor volume reached 100 0.0001 C-IC Frequency mm3 (n ¼ 7 mice, 2 injections per mouse). A, below 500 (mm Percent of tumors POP74, Saline (control), 5-FU (30 mg/kg 5d), 0 0.00001 04020 60 80 evofosfamide (Evo, 50 mg/kg 10 days), or the Evo combination given sequentially. The log-rank Saline 5-FU Time after treatment (days) test was used for statistical signiﬁcance. 5-FU + Evo , P < 0.05 for the 5-FU þ evofosfamide combination versus 5-FU alone group. B, CSC91, þ B CSC91 + CRT (Rectal cancer) D CSC91 Saline (control), CRT (5-FU, 20 mg/kg RT, 2 Gy 5 days), evofosfamide (50 mg/kg 10 0.1 *** days), or the combination given sequentially. 100 Saline

) C and D, C-IC frequency of colorectal cancer 3 CRT 0.01 *** cells from tumors shown in A and B, as measured Evo *** by LDA in vivo. Tumors were harvested 24 hours 0.001 posttreatment, dissociated into single cells, and 50 CRT + Evo injected subcutaneously into NSG mice at doses 0.0001 C-IC Frequency of 100,000, 10,000, 1,000, and 100 cells (n ¼ 5 below 400 (mm Percent of tumors mice, 2 injections per mouse). C, POP74 þ 5-FU 0 0.00001 (n ¼ 2 tumors per group, pooled). D, CSC91 þ 05025 75100 125 T R Evo CRT (n ¼ 4 tumors per group, pooled). Data are C Time after treatment (days) Saline shown as mean and 95% CI. Frequency and CRT + Evo probability estimates were computed using ELDA software (, P < 0.01; , P < 0.001).

significant reduction in CC-IC frequency compared with combination with evofosfamide, this trend was absent, with CRT alone (Fig. 5D; Supplementary Table S5; CC-IC frequency greater therapeutic benefit in tumors that had high FAZA uptake 1/52,140 vs. 1/9,491) or evofosfamide alone (1/3,417). Collec- at baseline. Indeed, when comparing the four tumors with the tively, these data demonstrate that addition of evofosfamide to 5- lowest FAZA uptake (FAZAlow) to the four with the highest FAZA FU or CRT provides a survival benefit over conventional therapies uptake (FAZAhigh), only the FAZAhigh group showed a statistically alone and increases targeting of the CC-IC fraction. significant decrease in tumor growth rate upon addition of evofosfamide (Fig. 6C). FAZA-PET identifies tumors that benefit most from the addition We observed similar results for the rectal tumor model of evofosfamide to standard-of-care therapies (CSC91), where tumors with higher baseline FAZA uptake gen- Hypoxia is associated with poor prognosis across multiple erally grew faster compared with those with lower baseline FAZA cancers including tumors of the head and neck, breast, cervix, uptake in the CRT alone group but not in the CRT þ evofosfamide and prostate (4, 21). To determine whether hypoxia is prognostic group (Fig. 6D), suggesting the more hypoxic tumors responded for clinical outcome of colon cancer patients, we tested whether less to CRT and greater therapeutic benefit could be seen in the existing hypoxia multi-gene sets correlate with overall survival tumors that had high FAZA uptake at baseline. Consistent with the across primary colon adenocarcinoma samples from The Cancer previous model, only the FAZAhigh group showed a statistically Genome Atlas dataset. Using three published hypoxia gene sets, significant decrease in tumor growth rate upon addition of Evo each encompassing multiple genes that exhibit differential expres- (Fig. 6E). Taken together, these studies indicate that FAZA-PET sion in hypoxia versus normoxia (36–38), we found that colon imaging prior to therapy initiation may serve as an effective cancer patients with high hypoxia gene expression have a signif- clinical biomarker to identify those patients who would benefit icantly worse overall survival compared with those with low most from the addition of evofosfamide to standard-of-care hypoxia gene expression (Supplementary Fig. S6). therapies. We next tested whether we could use noninvasive hypoxia imaging to identify those tumors that would benefit most from addition of evofosfamide to standard-of-care therapies. Discussion [18F]-FAZA-PET/CT imaging was used to determine the baseline Despite a significant body of evidence linking intratumoral intratumoral hypoxia level for a subset of mice from in vivo studies hypoxia to poor prognosis, therapeutic resistance, and enrich- of colorectal cancer xenografts POP74 and CSC91 described ment of C-ICs, targeting hypoxia has yet to become standard of above. One day before treatment commenced, mice from the care in cancer treatment (1, 4, 22). Approaches to targeting 5-FU or CRT and 5-FU or CRT þ evofosfamide groups were hypoxia include the use of bioreductive HAPs and inhibitors of injected with [18F]-FAZA and imaged by PET/CT (Fig. 6A). For hypoxia signaling molecules (4). HAPs have been used in pre- the colon tumor model (POP74), we observed a trend of faster clinical studies and clinical trials both as a single agent and in growth rates for tumors with higher FAZA uptake for the 5-FU combination with chemotherapy and/or radiotherapy (21, 23). alone group (Fig. 6B), suggesting that the more hypoxic tumors The goal of combination studies has been to use chemotherapy or had a decreased response to 5-FU. When 5-FU was given in radiotherapy to target the oxygenated tumor cells and a HAP to

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

Targeting Hypoxia after Therapy Decreases Colorectal Cancer-ICs

Figure 6. A FAZA-PET FAZA-PET is an effective clinical tumor biomarker to identify tumors that will 5-FU or CRT beneﬁt from the addition of Randomized Monitor evofosfamide (Evo). A, Diagram of by tumor tumor experimental protocol for [18F]-FAZA- volume 5-FU or CRT growth PET/CT hypoxia imaging for colorectal + Evo cancer PDX models (n ¼ 5 mice per treatment group). B, Scatterplots 5-FU showing the relationship between B C POP74: 5-FU POP74: 5-FU + Evo 5-FU + Evo FAZA uptake and growth rate for 60 60 60 POP74 tumors treated with 5-FU alone * or 5-FU þ evofosfamide. FAZA uptake was quantiﬁed at day 0 of treatment 40 40 40 /day) /day) /day) 3 3 using the tumor-to-muscle (T/M) 3 ratios. Tumor growth rate was

20 20 (mm 20 (mm calculated as the average growth rate (mm Tumor growth rate Tumor Tumor growth rate Tumor of tumors from day 0 of treatment to growth rate Tumor 0 0 0 time of sacriﬁce. The line of best ﬁtis 0105 15 20 0105 15 20 low high plotted for each treatment group. FAZA uptake (T/M) FAZA uptake (T/M) C, Growth rate of POP74 tumors from FAZA FAZA the indicated treatment group with the lowest FAZA uptake (FAZAlow)orthe highest FAZA uptake (FAZAhigh), D E CRT n ¼ 4 tumors per category for each CSC91: CRT CSC91: CRT + Evo CRT + Evo 30 30 30 treatment group. Data are shown as mean SEM. D, Scatterplots showing * the relationship between FAZA uptake 20 20 20 /day) /day) /day) 3 3 and growth rate for CSC91 tumors 3 treated with CRT alone or CRT þ

10 10 (mm 10 (mm evofosfamide. E, Growth rate of CSC91 (mm low high Tumor growth rate Tumor Tumor growth rate Tumor FAZA or FAZA tumors from the growth rate Tumor indicated treatment group, n ¼ 4 0 0 0 tumors per category for each 063 912 063 912 low high treatment group. Data are shown as FAZA uptake (T/M) FAZA uptake (T/M) FAZA FAZA mean SEM. Student t test was used for statistical significance (, P < 0.05). target the hypoxic compartment (26–28, 39, 40). Here, we show sensitized tumors to evofosfamide. The effect on tumor growth that sequential addition of evofosfamide after 5-FU or CRT is a inhibition was significantly greater when evofosfamide was added novel and highly effective method to target the hypoxic CC-IC after a 4-day course of 5-FU or CRT, compared with concurrent fraction. dosing. One caveat is that we used immunodeficient mice with Numerous reports demonstrate an additive or synergistic effect incomplete tumor microenvironment. However, similar results between evofosfamide and chemotherapy in a wide range of were reported by Benito and colleagues using a syngeneic model preclinical cell line–derived xenograft models of solid tumors of acute myeloid leukemia to show that leukemic bone marrow including: melanoma (26), osteosarcoma (29), colorectal (26), cells surviving chemotherapy remain hypoxic and can be targeted non–small cell lung (26, 27), prostate (28), and pancreatic (41) by the addition of evofosfamide one week after chemotherapy cancer. In these studies, the antitumor activity of cisplatin (26), (40). Our results, as well as those of Benito and colleagues, are docetaxol (26), doxorubicin (26, 28, 29), irinotecan (26), gem- supported by recent evidence that C-ICs reside in hypoxic niches citabine (26, 41), and temozolomide (26) was increased when protected from chemotherapies (5, 6, 42), and result in disease combined with evofosfamide in most cancer models tested. In the recurrence. Together, these findings strongly suggest that targeting context of colorectal cancer, Liu and colleagues showed that in the the hypoxic fraction represents a novel means to target CC-ICs. HT29 xenograft model, administration of Evo 2–8 hours before Intratumoral hypoxia is a known factor contributing to radio- cisplatin yielded superior growth suppression compared with Evo resistance; this is driven in part by the involvement of oxygen in given 2–8 hours after cisplatin or simultaneous administration the initial production of DNA damage and by additional complex (26). The authors hypothesized that administration of cisplatin and multifactorial molecular mechanisms (43). The "oxygen- prior to evofosfamide may have caused reoxygenation of the effect" was established over 50 years ago, and describes the hypoxic compartment; therefore, when evofosfamide was admin- involvement of oxygen in the initial formation of DNA breaks istered after cisplatin, its activity was reduced due to a smaller caused by low LET radiation (44). Hypoxic cells are up to 3-fold hypoxic fraction. It is difficult to make a direct comparison more resistant in terms of the radiation dose needed to cause between our results and those of Liu and colleagues because we equivalent levels of DNA damage and cell death. It is well used different chemotherapies and PDX models are more hetero- established that chronically hypoxic tumors also display geneous than cell line–derived xenografts. We selected 5-FU decreased DNA repair, which results in increased mutation rates because it represents the backbone of chemotherapy regimens and exacerbation of tumor aggressiveness (45). Numerous other for colorectal cancer. We found that pretreatment with 5-FU mechanisms are involved in hypoxia-driven radioresistance, resulted in an enrichment of the hypoxic fraction, which including stabilization of HIF-1a and HIF-2a, oxygen-dependent

www.aacrjournals.org Clin Cancer Res; 2018 OF9

Haynes et al.

epigenetic changes, and effects on cancer cell metabolism (43). evofosfamide as a monotherapy, and identification of these There is significant literature demonstrating that hypoxic cells cancers prior to starting treatment will likely single out patients are also less sensitive to chemotherapy (1–4). This relative che- that will benefit most from a HAP. In our panel of 8 unique moresistance has been explained by the fact that hypoxic cells colorectal cancer PDX models, only 2 samples consistently dem- are non- or slowly proliferating and as such do not respond onstrated a high level of hypoxia at baseline. Importantly, since well to chemotherapy. In addition, hypoxic regions within a we show that treatment with 5-FU enriches for hypoxic cancer tumor are less accessible to chemotherapy resulting in decreased cells, this could potentiate response to evofosfamide even in drug exposure. The relative increase in the hypoxic fraction that we samples with lower baseline hypoxia. observed after 5-FU could be the result of nonhypoxic cells being Finally, we demonstrate the use of FAZA-PET as a clinical targeted, thereby resulting in a relative enrichment in the hypoxic biomarker to predict response to evofosfamide based on intra- fraction. Another possible explanation is that 5-FU can induce tumoral hypoxia. In both PDX models tested, we found that colon cancer cells to undergo metabolic reprogramming toward tumors with higher baseline intratumoral hypoxia, as deter- OXPHOS (46). Genes regulating OXPHOS are upregulated in mined by FAZA-PET/CT imaging, generally exhibited faster chemotherapy-treated tumors, and in response to chemotherapy, growth rates in the presence of 5-FU or CRT. Importantly, the colonospheres can survive by engaging a SIRT1/PGC1a-depen- growth of high hypoxia (FAZAhigh)tumorswassignificantly dent shift from glycolysis to OXPHOS (47). Increased oxygen reduced in the 5-FU or CRT þ evofosfamide combination requirements through OXPHOS could result in increased tumor groups compared with 5-FU or CRT alone, whereas the growth hypoxia. Further work is required to fully elucidate the mechan- of low hypoxia (FAZAlow) tumors was not changed by the isms driving the relative increase in hypoxic cells following addition of evofosfamide. Interestingly, to date clinical trials treatment with 5-FU. of evofosfamide have not stratified patients based on intratu- In support of our findings that hypoxia enriches for CC-ICs, moral hypoxia, and according to our data this likely represents Mao and colleagues have shown that the majority of CC-ICs a confounding factor influencing the trial results. Other studies þ (CD133 population) in colon cancer samples stain positive for have also demonstrated the utility of hypoxia PET tracers the hypoxia marker pimo, whereas non-CC-ICs (CD133 popu- including [18F]-FAZA, [18F]-HX4, and [18F]-FMISO to accurately lation) do not (42). Furthermore, following chemotherapy- measure intratumoral hypoxia in preclinical models (30, 50, þ induced stress, CD133 cells in colon cancer xenografts were 51). Collectively, these findings indicate that FAZA-PET repre- spared, resulting in a relative enrichment of the hypoxic CC-IC sents a noninvasive biomarker of intratumoral hypoxia that fraction. The authors concluded that the hypoxic state of the should be utilized in the clinical setting to identify patients that þ CD133 cells renders them resistant to chemotherapy, lending will benefit from the addition of evofosfamide. further rationale for using a HAP to target CC-ICs. Lohse and colleagues have previously shown that combining radiotherapy Disclosure of Potential Conflicts of Interest with evofosfamide in pancreatic cancer PDX model–targeted C- U. Metser is a consultant/advisory board member for Abbvie. No potential ICs (39). Similarly, our serial passage in vivo LDA results demon- conflicts of interest were disclosed by the other authors. strate that evofosfamide combined with 5-FU or CRT is an effective way to target CC-ICs. Although the sequential combi- Authors' Contributions nation treatments had the most significant effect, 5-FU and CRT Conception and design: J. Haynes, T.D. McKee, A.C. Haller, C. Leung, A. Kreso, alone also decreased CC-ICs, suggesting that these agents did U. Metser, M. Smith, C.A. O'Brien target a subset of CC-ICs. It is known that different subclasses of C- Development of methodology: J. Haynes, T.D. McKee, A.C. Haller, Y. Wang, C. Leung, A. Kreso, D.C. Vines, M. Smith, C.A. O'Brien ICs exist within colorectal cancer, and that these subclasses display Acquisition of data (provided animals, acquired and managed patients, differential responses to treatment with oxaliplatin ranging from provided facilities, etc.): J. Haynes, T.D. McKee, A.C. Haller, Y. Wang, C. Leung, sensitive to resistant (48). The basis of the differential response E. Lima-Fernandes, R. Wolman, D.C. Vines, D.A. Jaffray was hypothesized to be related to the proliferative versus quies- Analysis and interpretation of data (e.g., statistical analysis, biostatistics, cent states of individual CC-ICs. However, another possible computational analysis): J. Haynes, T.D. McKee, A.C. Haller, Y. Wang, C. Leung, explanation is that CC-ICs exist in both the hypoxic and non- D.M.A. Gendoo, E. Lima-Fernandes, E. Szentgyorgyi, B. Haibe-Kains, M. Smith, C.A. O'Brien hypoxic fractions of a tumor; 5-FU and CRT may target CC-ICs in Writing, review, and/or revision of the manuscript: J. Haynes, T.D. McKee, the normoxic zones while sparing CC-ICs in the hypoxic zones A.C. Haller, C. Leung, D.M.A. Gendoo, E. Lima-Fernandes, D.C. Vines, B. Haibe- (39). Alternatively, it is possible that only CC-ICs that can switch Kains, B.G. Wouters, U. Metser, M. Smith, C.A. O'Brien their metabolism from glycolysis to OXPHOS are able to survive Administrative, technical, or material support (i.e., reporting or organizing and give rise to hypoxic CC-ICs (46, 47). Although the molecular data, constructing databases): T.D. McKee, Y. Wang, C. Leung, D.A. Jaffray, mechanisms remain to be elucidated, it is evident from our work C.A. O'Brien Study supervision: B.G. Wouters, C.A. O'Brien that the addition of evofosfamide to 5-FU or CRT increases Other (conducted the assessment of the prognostic value of hypoxia gene sets targeting of CC-ICs. in colon cancer patients (survival analysis): D.M.A. Gendoo Heterogeneity in oxygenation exists within and between patient tumors in every cancer type evaluated (4, 49), and plays Acknowledgments a key role in therapeutic response to HAPs (4, 25). We also found We would like to thank Jennifer Warner, Laurie Ailles, and Charles Hart for considerable intra- and intertumoral heterogeneity in the hypoxic their critical review of the manuscript. We are grateful to all members of the fraction of our colorectal cancer PDX models. For example, O'Brien laboratory for helpful suggestions and constructive discussions, espe- POP74 xenografts expressed the highest level of baseline hypoxia cially Christy Ahn and Jessica Suddaby for technical assistance. We also thank Fannong Meng, Mayleen Sukhram, Viktor Son, and Dianne Chadwick (UHN of all the models tested, and they were the most responsive to Biospecimen Sciences Program) for providing colon cancer samples, as well as treatment with evofosfamide alone. This suggests that at high Farzin Jafari and Maria Monroy (UHN Animal Resources Centre), and Deborah levels of intratumoral hypoxia there is a benefit to treating with Scollard and Teesha Komal (STTARR) for their skillful assistance. The authors

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

Targeting Hypoxia after Therapy Decreases Colorectal Cancer-ICs

would like to acknowledge the STTARR program and its afﬁliated funding embourg and the Marie Curie Actions of the European Commission (FP7- agencies. This work was supported by grants from Colon Cancer Canada (to COFUND), and subsequently of a Banting Postdoctoral Fellowship. C.A. O'Brien and M. Smith), the Canadian Institutes of Health Research (FDN14847 to C.A. O'Brien and 357163 to B. Haibe-Kains), the Terry Fox The costs of publication of this article were defrayed in part by the Research Institute (TFRI PPG 1036 to B.G. Wouters and D.A. Jaffray), and the payment of page charges. This article must therefore be hereby marked Ministry of Economic Development, Employment and Infrastructure and the advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate Ministry of Innovation of the Government of Ontario (ER14-10-121 to B. this fact. Haibe-Kains); and by the Gattuso-Slaight Personalized Cancer Medicine Fund at Princess Margaret Cancer Centre (to B. Haibe-Kains.). E. Lima-Fernandes is a Received June 16, 2017; revised December 21, 2017; accepted February 19, recipient of a postdoctoral fellowship from the National Research Fund Lux- 2018; published ﬁrst February 23, 2018.

References 1. Muz B, de la Puente P, Azab F, Azab AK. The role of hypoxia in cancer 21. Hunter FW, Wouters BG, Wilson WR. Hypoxia-activated prodrugs: paths progression, angiogenesis, metastasis, and resistance to therapy. Hypoxia forward in the era of personalised medicine. Br J Cancer 2016;114:1071–7. 2015;3:83–92. 22. Wilson WR, Pruijn FB. Hypoxia-activated prodrugs as antitumour agents: 2. Crowder SW, Balikov DA, Hwang YS, Sung HJ. Cancer stem cells under strategies for maximizing tumour cell killing. Clin Exp Pharmacol Physiol hypoxia as a chemoresistance factor in breast and brain. Curr Pathobiol Rep 1995;22:881–5. 2014;2:33–40. 23. Baran N, Konopleva M. Molecular pathways: hypoxia-activated prodrugs 3. Uribe D, Torres A, Rocha JD, Niechi I, Oyarzun C, Sobrevia L, et al. in cancer therapy. Clin Cancer Res 2017;23:2382–90. Multidrug resistance in glioblastoma stem-like cells: role of the hypoxic 24. Duan JX, Jiao H, Kaizerman J, Stanton T, Evans JW, Lan L, et al. Potent and microenvironment and adenosine signaling. Mol Aspects Med 2017;55: highly selective hypoxia-activated achiral phosphoramidate mustards as 140–51. anticancer drugs. J Med Chem 2008;51:2412–20. 4. Wilson WR, Hay MP. Targeting hypoxia in cancer therapy. Nat Rev Cancer 25. Sun JD, Liu Q, Wang J, Ahluwalia D, Ferraro D, Wang Y, et al. Selective 2011;11:393–410. tumor hypoxia targeting by hypoxia-activated prodrug TH-302 inhibits 5. Santoyo-Ramos P, Likhatcheva M, Garcia-Zepeda EA, Castaneda-Patlan tumor growth in preclinical models of cancer. Clin Cancer Res 2012; MC, Robles-Flores M. Hypoxia-inducible factors modulate the stemness 18:758–70. and malignancy of colon cancer cells by playing opposite roles in canonical 26. Liu Q, Sun JD, Wang J, Ahluwalia D, Baker AF, Cranmer LD, et al. TH-302, a Wnt signaling. PLoS One 2014;9:e112580. hypoxia-activated prodrug with broad in vivo preclinical combination 6. Dong HJ, Jang GB, Lee HY, Park SR, Kim JY, Nam JS, et al. The Wnt/beta- therapy efficacy: optimization of dosing regimens and schedules. Cancer catenin signaling/Id2 cascade mediates the effects of hypoxia on the Chemother Pharmacol 2012;69:1487–98. hierarchy of colorectal-cancer stem cells. Sci Rep 2016;6:22966. 27. Sun JD, Liu Q, Ahluwalia D, Ferraro DJ, Wang Y, Jung D, et al. Comparison 7. Keith B, Simon MC. Hypoxia-inducible factors, stem cells, and cancer. Cell of hypoxia-activated prodrug evofosfamide (TH-302) and ifosfamide in 2007;129:465–72. preclinical non-small cell lung cancer models. Cancer Biol Ther 8. Vadde R, Vemula S, Jinka R, Merchant N, Bramhachari PV, Nagaraju GP. 2016;17:371–80. Role of hypoxia-inducible factors (HIF) in the maintenance of stemness 28. Saggar JK, Tannock IF. Activity of the hypoxia-activated pro-drug TH-302 in and malignancy of colorectal cancer. Crit Rev Oncol Hematol hypoxic and perivascular regions of solid tumors and its potential to 2017;113:22–7. enhance therapeutic effects of chemotherapy. Int J Cancer 2014;134: 9. Giles RH, Lolkema MP, Snijckers CM, Belderbos M, van der Groep P, Mans 2726–34. DA, et al. Interplay between VHL/HIF1alpha and Wnt/beta-catenin path- 29. Liapis V, Labrinidis A, Zinonos I, Hay S, Ponomarev V, Panagopoulos V, ways during colorectal tumorigenesis. Oncogene 2006;25:3065–70. et al. Hypoxia-activated pro-drug TH-302 exhibits potent tumor suppres- 10. Yeung TM, Gandhi SC, Bodmer WF. Hypoxia and lineage specification of sive activity and cooperates with chemotherapy against osteosarcoma. cell line-derived colorectal cancer stem cells. Proc Natl Acad Sci U S A Cancer Lett 2015;357:160–9. 2011;108:4382–7. 30. Stokes AM, Hart CP, Quarles CC. Hypoxia imaging with PET correlates 11. Li Z, Bao S, Wu Q, Wang H, Eyler C, Sathornsumetee S, et al. Hypoxia- with antitumor activity of the hypoxia-activated prodrug evofosfamide inducible factors regulate tumorigenic capacity of glioma stem cells. Cancer (TH-302) in rodent glioma models. Tomography 2016;2:229–37. Cell 2009;15:501–13. 31. Kreso A, van Galen P, Pedley NM, Lima-Fernandes E, Frelin C, Davis T, et al. 12. Heddleston JM, Li Z, Lathia JD, Bao S, Hjelmeland AB, Rich JN. Hypoxia Self-renewal as a therapeutic target in human colorectal cancer. Nat Med inducible factors in cancer stem cells. Br J Cancer 2010;102:789–95. 2014;20:29–36. 13. Colwell N, Larion M, Giles AJ, Seldomridge AN, Sizdahkhani S, Gilbert MR, 32. O'Brien CA, Kreso A, Ryan P, Hermans KG, Gibson L, Wang Y, et al. ID1 and et al. Hypoxia in the glioblastoma microenvironment: shaping the phe- ID3 regulate the self-renewal capacity of human colon cancer-initiating notype of cancer stem-like cells. Neuro Oncol 2017;19:887–96. cells through p21. Cancer Cell 2012;21:777–92. 14. Lock FE, McDonald PC, Lou Y, Serrano I, Chafe SC, Ostlund C, et al. 33. Zannella VE, Dal Pra A, Muaddi H, McKee TD, Stapleton S, Sykes J, et al. Targeting carbonic anhydrase IX depletes breast cancer stem cells within the Reprogramming metabolism with metformin improves tumor oxygen- hypoxic niche. Oncogene 2013;32:5210–9. ation and radiotherapy response. Clin Cancer Res 2013;19:6741–50. 15. McIntosh K, Balch C, Tiwari AK. Tackling multidrug resistance mediated by 34. Mazumdar J, O'Brien WT, Johnson RS, LaManna JC, Chavez JC, Klein PS, efflux transporters in tumor-initiating cells. Expert Opin Drug Metab et al. O2 regulates stem cells through Wnt/beta-catenin signalling. Nat Cell Toxicol 2016;12:633–44. Biol 2010;12:1007–13. 16. Bharti R, Dey G, Mandal M. Cancer development, chemoresistance, epi- 35. Vermeulen L, De Sousa EMF, van der Heijden M, Cameron K, de Jong JH, thelial to mesenchymal transition and stem cells: a snapshot of IL-6 Borovski T, et al. Wnt activity defines colon cancer stem cells and is mediated involvement. Cancer Lett 2016;375:51–61. regulated by the microenvironment. Nat Cell Biol 2010;12:468–76. 17. Safari M, Khoshnevisan A. Cancer stem cells and chemoresistance in 36. Winter SC, Buffa FM, Silva P, Miller C, Valentine HR, Turley H, et al. glioblastoma multiform: a review article. J Stem Cells 2015;10:271–85. Relation of a hypoxia metagene derived from head and neck cancer to 18. Paldino E, Tesori V, Casalbore P, Gasbarrini A, Puglisi MA. Tumor initiating prognosis of multiple cancers. Cancer Res 2007;67:3441–9. cells and chemoresistance: which is the best strategy to target colon cancer 37. Eustace A, Mani N, Span PN, Irlam JJ, Taylor J, Betts GN, et al. A 26- stem cells? Biomed Res Int 2014;2014:859871. gene hypoxia signature predicts benefit from hypoxia-modifying ther- 19. Bao S, Wu Q, McLendon RE, Hao Y, Shi Q, Hjelmeland AB, et al. Glioma apy in laryngeal cancer but not bladder cancer. Clin Cancer Res stem cells promote radioresistance by preferential activation of the DNA 2013;19:4879–88. damage response. Nature 2006;444:756–60. 38. Dekervel J, Hompes D, van Malenstein H, Popovic D, Sagaert X, De Moor B, 20. Rich JN. Cancer stem cells in radiation resistance. Cancer Res 2007;67: et al. Hypoxia-driven gene expression is an independent prognostic factor 8980–4. in stage II and III colon cancer patients. Clin Cancer Res 2014;20:2159–68.

www.aacrjournals.org Clin Cancer Res; 2018 OF11

Haynes et al.

39. Lohse I, Rasowski J, Cao P, Pintilie M, Do T, Tsao MS, et al. Targeting 46. Denise C, Paoli P, Calvani M, Taddei ML, Giannoni E, Kopetz S, et al. 5- hypoxic microenvironment of pancreatic xenografts with the hypoxia- fluorouracil resistant colon cancer cells are addicted to OXPHOS to survive activated prodrug TH-302. Oncotarget 2016;7:33571–80. and enhance stem-like traits. Oncotarget 2015;6:41706–21. 40. Benito J, Ramirez MS, Millward NZ, Velez J, Harutyunyan KG, Lu H, 47. Vellinga TT, Borovski T, de Boer VC, Fatrai S, van Schelven S, Trumpi K, et al. et al. Hypoxia-activated prodrug TH-302 targets hypoxic bone marrow SIRT1/PGC1alpha-dependent increase in oxidative phosphorylation sup- niches in preclinical leukemia models. Clin Cancer Res 2016;22: ports chemotherapy resistance of colon cancer. Clin Cancer Res 2015; 1687–98. 21:2870–9. 41. Sun JD, Liu Q, Ahluwalia D, Li W, Meng F, Wang Y, et al. Efficacy and safety 48. Kreso A, O'Brien CA, van Galen P, Gan OI, Notta F, Brown AM, et al. of the hypoxia-activated prodrug TH-302 in combination with gemcita- Variable clonal repopulation dynamics influence chemotherapy response bine and nab-paclitaxel in human tumor xenograft models of pancreatic in colorectal cancer. Science 2013;339:543–8. cancer. Cancer Biol Ther 2015;16:438–49. 49. Dhani NC, Serra S, Pintilie M, Schwock J, Xu J, Gallinger S, et al. Analysis of 42. Mao Q, Zhang Y, Fu X, Xue J, Guo W, Meng M, et al. A tumor hypoxic niche the intra- and intertumoral heterogeneity of hypoxia in pancreatic cancer protects human colon cancer stem cells from chemotherapy. J Cancer Res patients receiving the nitroimidazole tracer pimonidazole. Br J Cancer Clin Oncol 2013;139:211–22. 2015;113:864–71. 43. Rey S, Schito L, Koritzinsky M, Wouters BG. Molecular targeting of hypoxia 50. Peeters SG, Zegers CM, Lieuwes NG, van Elmpt W, Eriksson J, van Dongen in radiotherapy. Adv Drug Deliv Rev 2017;109:45–62. GA, et al. A comparative study of the hypoxia PET tracers [18F]HX4, 44. Rockwell S, Dobrucki IT, Kim EY, Marrison ST, Vu VT. Hypoxia and [18F]FAZA, and [18F]FMISO in a preclinical tumor model. Int J Radiat radiation therapy: past history, ongoing research, and future promise. Curr Oncol Biol Phys 2015;91:351–9. Mol Med 2009;9:442–58. 51. Carlin S, Zhang H, Reese M, Ramos NN, Chen Q, Ricketts SA. A comparison 45. Chan N, Koch CJ, Bristow RG. Tumor hypoxia as a modifier of DNA strand of the imaging characteristics and microregional distribution of 4 hypoxia break and cross-link repair. Curr Mol Med 2009;9:401–10. PET tracers. J Nucl Med 2014;55:515–21.

OF12 Clin Cancer Res; 2018 Clinical Cancer Research

Administration of Hypoxia-Activated Prodrug Evofosfamide after Conventional Adjuvant Therapy Enhances Therapeutic Outcome and Targets Cancer-Initiating Cells in Preclinical Models of Colorectal Cancer

Jennifer Haynes, Trevor D. McKee, Andrew Haller, et al.

Clin Cancer Res Published OnlineFirst February 23, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-17-1715

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/04/11/1078-0432.CCR-17-1715.DC2

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2018/04/05/1078-0432.CCR-17-1715. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.