Patupilone Induced Vascular Disruption in Orthotopic Rodent Tumor Models Detected by Magnetic Resonance Imaging and Interstitial Fluid Pressure

Patupilone, also known as epothilone B (EPO906), is a potent microtubule stabilizer. Epothilones, which are secondary metabolite macrolides produced by the myxobacterium Sorangium cellulosum (1), represent a novel class of nontaxane microtubule-stabilizing natural products. Although patupilone has the same molecular target as the taxanes, it binds β-tubulin with a higher affinity (2) and the binding sites are probably not identical (3). Binding promotes polymerization of tubulin heterodimers into microtubule polymers and stabilizes microtubules against depolymerization, resulting in mitotic cell cycle arrest and eventually cell death via apoptosis (4–6). In non–small cell lung cancer cells, apoptosis due to patupilone exposure seems to be cathepsin B dependent (7). Preclinical work has shown that patupilone is the most active variant among the natural epothilones, is more potent than the taxanes, and, most importantly, retains activity in vitro and in vivo against taxane-resistant cancer cells overexpressing P-glycoprotein or bearing β-tubulin mutations (4–6, 8). Additionally, as a tubulin-binding agent, patupilone should show antivascular effects by disrupting rapidly proliferating and immature endothelial cells, which have a particularly strong dependence on tubulin to maintain their shape (9). Support for this hypothesis was recently provided with cultured cells using low doses of patupilone (10) and on clinical tumor explants (11). Inhibition of endothelial cell function could also have a negative effect on the metastatic process. Thus, both this compound class and the target are considered to have great potential for the treatment of cancer (6, 12).

Patupilone shows marked activity against solid human tumor xenografts grown in nude mice, including thyroid (13), lung, breast, colon, and prostate (4), when given at 3 to 4 mg/kg as an i.v. bolus on a weekly schedule (4, 14). The drug displays a rapid t_1/2α, is eliminated from plasma with a terminal half-life of ∼6 hours, and shows a large volume of distribution to all tissues, including the tumor xenografts and brain from which tissues it is eliminated more slowly (15). Dose-dependent activity against s.c. grown rat C6 glioma tumors has also been shown, and in this model and a human thyroid carcinoma xenograft model, there was synergy when combined with the novel anticancer agent imatinib (13, 16). In summary, the pharmacokinetic profile and activity of patupilone in experimental tumor models suggest that it could be an effective treatment for a range of solid tumors in the clinic. Patupilone is currently in clinical development and early phase II data indicate promising activity in several different solid tumor indications (12, 14). However, a convenient and predictive surrogate biomarker of patupilone is currently not available because the pharmacokinetic profile limits the possibility of using drug levels in the plasma to obtain an effective pharmacokinetic/pharmacodynamic relationship.

The initial growth of a solid tumor is amid normal tissue where it makes use of existing vasculature. Further growth requires formation of new blood vessels (i.e., angiogenesis), but generally these blood vessels tend to be leaky, highly irregular, and tortuous, exhibit arteriovenous shunts and blind ends, and have a sluggish flow due to the high viscosity of the blood (17, 18). The tumor is also unable to form its own lymphatic system, so that solutes cannot be efficiently drained from the tumor interstitial space. Furthermore, the blood vessels are compressed by proliferating cells and this increases the microvascular pressure. Together, these factors contribute to an elevated interstitial fluid pressure (IFP) in solid tumors compared with normal tissues (17). In experimental tumor models, the IFP was found to be uniform across the whole tumor, except at the periphery where it dropped precipitously (17). Jain (17) predicted that the high IFP would reduce movement of molecules from blood vessels to the interstitial space, particularly large molecules (>5 kDa), which move more by convection than diffusion and thus may be a major barrier to the delivery of larger chemotherapeutics. However, recent data suggest that relatively small chemotherapeutics, such as 5-fluorouracil, paclitaxel, and patupilone, also show increased uptake into experimental tumors when the IFP is lowered pharmacologically (13, 19, 20). Thus, there is considerable interest in understanding better the causes of a high tumor IFP to find means to reduce it and elevate drug uptake. Preclinically, although many groups have shown that drugs can lower the IFP of solid tumors, including prostaglandin E₁ (19), dexamethasone (21), imatinib (22), and paclitaxel (23), it has not been investigated whether this effect might also be a predictive factor for drug response. Although Lee et al. (24) showed that an anti–vascular endothelial growth factor antibody lowered the IFP of human tumor xenografts and decreased tumor growth, this was not done by measuring tumor growth and IFP in the same animals; furthermore, there are no investigations of this type for cytotoxics. IFP has also been measured in the clinic in several different solid tumors, including breast, cervix, head and neck, liver metastases, kidney, lung, and melanoma, and found to be always greater than zero, which is the value for normal tissue (17). In one study, IFP was used as a prognostic factor and this showed that it was the single best indicator of survival for patients with cancer of the cervix where a high IFP was associated with a poor prognosis (25). More recently, it has been shown that Avastin, the vascular endothelial growth factor A antibody, and paclitaxel could lower the IFP of renal and breast tumors, respectively (26, 27). A demonstration of a correlation between changes in tumor growth and IFP in the same tumor would suggest that IFP could be used as a surrogate biomarker of response.

Identifying potential responders is critical to successful drug treatment of patients, and development of biomarkers should be an aid to the Response Evaluation Criteria in Solid Tumors, which can be relatively inaccurate and subjective (28). In any case, Response Evaluation Criteria in Solid Tumors may not predict progression-free survival or overall survival, which in some cases can be better predicted by a biomarker (29, 30). Biomarkers can include proteins circulating in the serum, such CA-125, molecular analysis of expression from biopsies, as well as variables supplied by various noninvasive functional imaging modalities, such as [¹⁸F]fluoro-2-deoxy-d-glucose positron emission tomography or dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI; ref. 31). The latter has been proven useful in the development of antiangiogenic drugs where changes in the tumor vasculature can be a marker of whether the effective dose has been reached as well as tumor response (32).

The aim of this study was to investigate the effects of patupilone on the vasculature of solid tumors using a variety of different techniques, including methods that were noninvasive, using DCE-MRI or dynamic susceptibility contrast MRI (DSC-MRI), minimally invasive (IFP measurement) and invasive using histology ex vivo. For this, we have used two orthotopic and syngeneic rodent models, so that the target, the endothelial cells in the vascular supply, would be more prevalent and thus more likely to resemble tumors in patients. This approach could help identify an independent surrogate biomarker of response using a method(s) that can be applied in the clinic and would thus aid development of patupilone.

All animal experiments were done in strict adherence to the Swiss law for animal protection. Animals were identified by tail markings and kept in groups of 4 to 10 animals under normal conditions with access to food and water ad libitum. Female black C57/BL6 syngeneic mice weighing 20 to 25 g were obtained from Charles River (Lyon, France). Female Brown-Norway rats weighing 160 to 180 g were obtained from Charles River. The Brown-Norway rats had a mild, apparently noninfectious pneumonia in the lungs (defined pathologically as multifocal granulomatous pneumonia), which did not affect their general health but precluded studies of metastasis. Except where stated, all animals were sacrificed by CO₂ inhalation.

All cell culture materials were obtained from Life Technologies (Paisley, United Kingdom) or BioConcept (Allschwil, Switzerland). Gadolinium 1,4,7,10-tetra-azacylododecane N,N′,N″,N‴,-tetra-acetate (GdDOTA; Dotarem), the extravascular contrast agent, and Endorem and Sinerem (dextran-coated iron oxide nanoparticles) the intravascular contrast agents, were obtained from Guerbet SA (Roissy, France).

Patupilone was obtained from Chemical Development, Novartis (Basel, Switzerland) and the powder was stored at −20°C. It was freshly prepared on each treatment day by dissolving in polyethylene glycol 300 and then diluting with physiologic saline [0.9% (w/v) NaCl] to obtain a mixture of 30% (v/v) polyethylene glycol 300 and 70% (v/v) physiologic saline for i.v. administration in the tail vein. The injection volume was 2 and 5 mL/kg for rats and mice, respectively. Treatment with vehicle or patupilone was normally once only or once weekly in the efficacy studies.

Where possible, tumor growth and body weights were monitored at least twice weekly. Toxicity was determined by the percentage change in body weight versus those recorded in the vehicle-treated controls. No other obvious toxicity was observed in mice (e.g., diarrhea; the dose-limiting toxicity observed in clinical trials), although some diarrhea was observed in Brown-Norway rats at the highest patupilone doses tested at 3 to 5 days after injection.

Mouse B16/BL6 metastatic melanoma. Cells were cultured in MEM supplemented with 2 mmol/L glutamine, 5% FCS, 1% sodium pyruvate, 1% nonessential amino acids, and 2% vitamins and kept at confluence for 4 to 5 days (with a medium change every 48 hours) before resuspension in HBSS (+10% FCS). Using a Hamilton syringe, 5 × 10⁴ cells were injected i.d. in a volume of 1 μL into the dorsal pinna of both ears of the anesthetized mouse as described previously (33). After 1 week, the primary tumor was visible as a black dot in the middle of the ear, and after 10 days, the cervical lymph node metastases were detectable by palpitation and were used for all biomarker studies from 14 days after cell inoculation.

Mice were randomized into three groups of six mice per group and treatment with vehicle or patupilone (2 or 4 mg/kg) was once weekly for 2 weeks beginning 7 days after cell inoculation. For primary tumors, growth was assessed by the fractional change in tumor surface area compared with the area at the initiation of treatment. Efficacy was expressed as a treated versus control (T/C; i.e., mean fractional increase of drug-treated mice divided by mean fractional increase of vehicle-treated mice). For the metastases, the mice were sacrificed at the end of the treatment period, the lymph nodes were removed and weighed, and the T/C was determined from the mean volume of the metastases from the drug-treated mice divided by the mean volume of the metastases from the vehicle-treated mice.

Rat BN472 mammary carcinoma. BN472 tumors were created by transplantation of tumor fragments. Tissue frozen at −180°C from earlier in-house studies was washed sterilely in 0.9% NaCl containing 100 μg/mL gentamicin and then placed in a Petri dish containing Ham's F-12 medium with 100 μg/mL gentamicin and 10% FCS. The tumor tissue was divided into small pieces (3 × 3 × 3 mm) and one piece was inserted into a hole made by a 22-gauge sterile needle in the lowest mammary fat pad of anesthetized rats after wiping the skin with 70% alcohol. The hole was closed with two metal wound clips, which were then removed after 3 to 4 days. Tumors were used for efficacy and biomarker studies at 2 to 3 weeks after transplantation.

Rats were randomized into different treatment groups and treated with vehicle or various doses of patupilone (0.3, 0.8, 1, and 1.5 mg/kg once). Efficacy was assessed after 1 week as a T/C determined from the mean change in tumor volume (TVol) (determined from the formula: l * w * h * π / 6) of drug-treated rats divided by the mean change in volume of vehicle-treated rats.

MRI was done on mouse B16/BL6 cervical lymph node metastases and rat BN472 mammary tumors. For all studies, animals were anaesthetized using 1.5% isoflurane (Abbott, Cham, Switzerland) in a 1:2 (v/v) mixture of O₂/N₂O and placed on an electrically warmed pad for cannulation of one of the tail veins using a 30-gauge needle attached to an infusion line of 30 cm and volume of 80 μL to permit remote administration of the contrast agent. The animals were positioned on a cradle in a supine position inside the 30-cm horizontal bore magnet and were anesthetized with 1.5% isoflurane in a 1:2 (v/v) mixture of O₂/N₂O given with a facemask (flow rate, 0.7 L/min). The mouse body temperature was maintained at 37 ± 2°C using a warm air flow and was monitored with a rectal probe. MRI experiments were done on a Bruker DBX 47/30 spectrometer (Bruker BioSpin, Fällanden, Switzerland) at 4.7 Tesla equipped with a self-shielded 12-cm bore gradient system as described previously (33).

Dynamic contrast-enhanced magnetic resonance imaging. In mice, the contrast agent GdDOTA was injected (0.1 mmol/kg) for determination of tumor vascular permeability (initial slope of GdDOTA uptake) and extravasion [i.e., tumor extracellular leakage space (final value for GdDOTA uptake)], and the first 20 points (102 seconds) from the injection of GdDOTA were used to calculate the initial area under the enhancement curve (IAUC) for GdDOTA. After 15 minutes, the iron oxide particle intravascular contrast agent Endorem was injected (6 mmol/kg iron) for determination of the relative tumor blood volume (rBVol), and the initial slope of uptake of Endorem by the tumor was used as an index of blood flow (BFI). The principles behind measurement of these variables have already been fully described (33). Animals were examined at baseline (day 0) before treatment with vehicle or patupilone, and at days 2 and/or 6, repeat DCE-MRI scans were made. The results were then expressed as the individual and mean change compared with baseline (day 0) measurement.

Dynamic susceptibility contrast magnetic resonance imaging. In rats bearing BN472 tumors, the contrast agent Sinerem, which is also an iron oxide particle intravascular contrast agent, was injected i.v. (0.2 mmol/kg iron) for measurement of tumor vessel size imaging (VSI) and absolute BVol as a percentage of the total tumor size (BVol). Briefly, the method is based on the measurement ΔR₂ and ΔR₂* relaxation rate constants induced by the injection of an intravascular slow clearance superparamagnetic contrast agent, such as Sinerem. Based on relaxation theory, it was shown in vivo that the ratio ΔR₂*/ΔR₂ is related to the diameter of the vessels (34). R₂* is determined directly by a multigradient recalled echo sequence (TR 400 ms, TE 3-24 ms, 8 echoes) before and after contrast agent infusion. ΔR₂, on the other hand, is assessed by the acquisition of T₂-weighted MRIs (repetition time 4,000 ms, echo time 18 ms) before and after contrast agent infusion. It is calculated according to (34)

where S_pre and S_post are the signal intensities before and after Sinerem infusion, respectively.

Mean vessel diameter R is calculated according to the following equation (35):

where D is the water diffusion coefficient in tumor tissue, γ is the gyromagnetic constant of the proton, Δχ is the susceptibility difference between blood vessels and the surrounding tissues induced by the contrast agent, and B₀ is the magnetic field strength of the magnet.

Blood fraction ξ₀ (BVol) can be derived according to the equation:

and is expressed as a percentage of the total TVol.

The water diffusion coefficient (D) was calculated as follows. Diffusion-weighted images were acquired using a spin-echo sequence with the following variables: repetition time 1,000 ms, echo time 22.5 ms, field of view 70 mm (rat) or 40 mm (mice), matrix 128 × 64 (acquisition) and 128² (reconstruction), slice thickness 1.5 mm, numerical aperture 2, spectral width 50 kHz, and total acquisition time 8 minutes 32 seconds. Four diffusion weighting acquisitions were done with b-factors 18.3, 43.9, 131.4, and 450.5 seconds/mm² (rats) or 31.5, 62.9, 161.8, and 503.9 seconds/mm² (mice) in read direction. Diffusion time (Δ) was 25 ms and diffusion gradient duration (δ) was 5 ms.

The apparent diffusion coefficient (ADC) maps were calculated by pixel-wise nonlinear fitting procedures of the four diffusion-weighted images by the imager software (ParaVision, Bruker BioSpin) according to the model function:

where S_b and S₀ are the signal intensities with and without diffusion gradients, respectively, and b is the diffusion weighting defined by the diffusion gradient strengths and timings. Diffusion weighting factor b is given by the formula:

where g is the diffusion gradient strength, δ is the diffusion gradient duration, and Δ is the diffusion time.

ADC values are given as the mean of the regions of interest encompassing the entire tumor.

The IFP of B16/BL6 and BN472 tumors was measured using the wick-in-needle method as described previously (13). Briefly, animals were divided into different treatment groups and anesthetized using 2.5% isoflurane delivered at 2 L/min. A standard 23-gauge needle connected to a pressure transducer was inserted into the central part of the tumors, and the pressure was monitored for a period of 10 minutes. Animals were then treated with vehicle or patupilone, and the IFP was monitored again at 2 and 6 days later. Results are expressed as the individual and mean change compared with baseline (day 0) measurement.

In the rat BN472 tumors, repeated measurements were readily made, and consistent with earlier reports in other experimental tumors (17), the IFP was found to be invariable across the tumor, except at the periphery. Furthermore, the IFP was independent of BN472 TVol over the range studied of 0.1 to 4 cm³ (data not shown).

H33342 staining of functional blood vessels. The nuclear staining dye, H33342 (Sigma, Buchs, Switzerland) was injected i.v. at 20 mg/kg (2 mg/mL in normal saline) into anesthetized animals, and after 45 seconds, the animals were sacrificed under anesthesia by cervical dislocation and the tumor was removed. The tumor was sliced through the central plain and analyzed as described previously (33) using a magnification of ×100 to determine the blood vessel width and density number. To display the frequency distribution of blood vessel width, vessels were assigned to groups of 10-μm size range (external diameter).

Apoptosis (caspase-3 staining) and necrosis. Following treatment and noninvasive biomarker measurements (day 6), the rats were sacrificed. Tumor slices were harvested from the largest circumference of the tumor, fixed in 4% phosphate-buffered formaldehyde for 24 hours at 4°C, and processed into paraffin. The skin overlying the tumor was swabbed with 70% ethanol, and using surgical scissors, spindle-shaped cuts were made to isolate the tumor ensuring that the superficial skin remained attached. The tumor mass was removed en bloc to a dissection tray and the tumor was cut into equal pieces using a scalpel (no. 10 curved blade). Tumor pieces were not allowed to exceed 3 to 4 mm in thickness, and care was taken to leave the overlying skin attached to the tumor where possible. Tumor pieces were transferred into prelabeled histocasettes and then immersion fixed in 4% phosphate-buffered formaldehyde (pH 7.4; J.T. Baker, Medite, Nunnigen, Switzerland) at 4°C for ≤24 hours. At least 10 times more fixative volume than tissue volume was used.

After fixation, tissue samples were rinsed in 70% ethanol and immediately processed into paraffin using the Wochenprogramm (run time, 10.5 hours) on the TPCduo tissue processor (Medite). After paraffinization, tissue samples were embedded in paraffin. From each tissue block, 3-μm-thick serial sections were cut on a Mikrom sliding microtome (Medite), spread on a 37°C water bath, mounted on microscope slides (Polysine, Menzel, Medite), and air dried at 37°C overnight. One tissue section was stained with H&E on the COT20 linear stainer (Medite). The remaining serial sections were either immediately used for immunohistochemistry or stored at −80°C until used (see below).

Tissue sections were deparaffinized in two changes of fresh xylene for 20 minutes at room temperature followed by hydration of tissue sections in absolute ethanol for 10 minutes, in 96% ethanol for 10 minutes, in 70% ethanol for several rinses, and in PBS (pH 7.4) without Ca²⁺ and Mg²⁺ for 2 × 5 minutes. Heat-induced antigen retrieval was done in 20 mmol/L boric acid buffer (pH 7.0) for 10 minutes at 98°C using a temperature-controlled microwave (Micromed T/T Mega, Medite). Sections were allowed to cool for 20 minutes and were washed in PBS for 2 × 5 minutes. The endogenous peroxidase was blocked with freshly prepared 3.0% H₂O₂ in methanol for 10 minutes followed by rinsing in PBS for 2 × 5 minutes. Tissue sections were blocked by incubation with 10% normal goat serum in PBS for 30 minutes.

For immunohistochemical detection of active caspase-3, a rabbit anti-cleaved caspase-3 (Asp¹⁷⁵) antibody (9661, Cell Signaling, BioConcept) was used. The primary antibody was applied in a dilution of 1:50 in 10% normal goat serum and incubated overnight at 4°C. Tissue sections were rinsed in PBS for 2 × 5 minutes and incubated with DAKO EnVision+ rabbit (K4003, DAKO, Geneva, Switzerland) for 30 minutes at room temperature. Tissue sections were rinsed in PBS for 2 × 5 minutes and incubated with freshly prepared AEC substrate solution (AEC tablets, Sigma) for 11 minutes. The reaction was stopped by rinsing tissue sections in tap water, and counterstaining was done in hematoxylin (Mayer's hematoxylin, Medite) for 30 seconds. Tissue sections were blued in tap water, coverslipped with water-soluble mounting medium (Aquatex, Merck, Glattbrug, Switzerland), and air dried.

Active caspase-3 immunohistochemistry was done on tissue sections from all collected samples. Stained tumor sections were coded such that the evaluator was unaware from which treatment group the samples were obtained. Tumor sections were evaluated for active caspase-3-positive tumor cells using a E800M Nikon microscope. A grid integrated into the ×10 ocular high-power fields composed of 10 × 10 squares (100 files total) encompassing an area of 250 × 250 μm at × 400 magnification was used for quantification of caspase-3 activity. The content of 10 grids was calculated. Active caspase-3-positive particles were calculated in grids containing viable tumor cells. Necrotic areas or border of necrotic areas were excluded from quantification. Care was taken to randomly select tissue areas and not to include adjacent tissue areas to areas that were already counted. The amount of necrotic tumor present was estimated and expressed as percentage of the total area.

Mice were thoracectomized after having been deeply anesthetized with 3% isoflurane (20 L/min) and perfused through the left ventricle of the heart. The left ventricle was punctured with a 19- or 23-gauge needle (for rats and mice, respectively) from a winged infusion set (SV-19BLK; Termudo, Elkton, MD), which was connected to an airtight pressurized syringe containing the rinsing solution (NaCl 0.9% with 250,000 units/L heparin at 35°C). The right atrium was punctured to provide outflow, and the perfusate was infused under a precise controlled pressure of 120 mm Hg. The perfusion was continued for 2 or 5 minutes (for mice and rats, respectively) at a constant rate (20 mL) followed by a prefixation solution (2% performaldehyde in PBS at 35°C). Finally, up to 10 to 30 mL polyurethane resin (PUII4; Vasqtec, Zurich, Switzerland) was infused at the same rate.

The resin-filled tumor was excised from the animal and kept at room temperature for 48 hours to complete resin curing. The tumor cast was then weighed and the soft tissue was removed by maceration in 7.5% KOH during 24 hours at 50°C. The casts were then thoroughly cleaned with and stored in distilled water before drying by lyophilization. Casts were semiquantified by determination of the vascular index (percentage of vascularized tissue), which was calculated from the ratio of the weight before and after KOH extraction.

All results show the mean ± SE. A significant change in TVol compared with vehicle treatment was assessed using a two-tailed t test or one-way ANOVA applying the Holms-Sidak test post hoc for multiple comparisons on the ΔTVol. For biomarker studies, the individual values for each animal are shown both before and one or two time points after treatment, so that the intertumor variability of the response is clearer. A statistically significant change (P < 0.05) in the biomarkers measured in situ (IFP, BVol, ADC, and VSI) compared with baseline was assessed using a paired t test for two time points or a one-way ANOVA with repeated measures applying the Holms-Sidak test post hoc for three time points (using the software from SigmaStat version 3.1). A two-tailed t test was used to compare mean percentage changes between different treatment groups. Correlations between different variables were made by determining Pearson correlation coefficient, with P < 0.05 considered as significant. Where multiple correlations were made between the different variables measured in vivo and ex vivo in the data set of BN472 tumors treated with vehicle or a dose of 0.8 mg/kg patupilone, a Bonferroni correction was made (P multiplied by the number of correlations). In this particular data set, one rat was excluded from all analyses because it was an outlier by Grubb's test (also called the extreme studentized deviate method; see http://www.graphpad.com) for the magnetic resonance–determined values of BVol, ADC, and VSI. For determination of the significance of the association Δ%IFP with tumor response, Fisher's exact test was applied to a 2 × 2 contingency table. The cutoffs for this analysis were based on the decrease in tumor IFP observed in vehicle-treated rats (9 of 10 in the range of −1% to −35%) and response was defined as a ΔTVol less than or equal to 1 SD from the mean of the vehicle-treated rats (1.8 cm³ minus 1.3 cm³).

Effect of 2-week patupilone treatment in the B16/BL6 mouse melanoma model. In two independent experiments, patupilone dosed weekly induced dose-dependent inhibition of growth of both primary tumor and cervical lymph node metastases of B16/BL6 mouse melanoma cells orthotopically implanted into C57BL6 mice. Figure 1A and B shows one experiment, in which the 2 mg/kg dose significantly reduced primary tumor growth (T/C, 26%) but not the cervical metastases (T/C, 123%), whereas the higher dose (4 mg/kg) significantly reduced growth of both tumors compared with vehicle-treated mice (P < 0.001). In cervical metastases, vessel density was significantly reduced (P = 0.014) compared with vehicle-treated controls metastases by 40%, and the frequency distribution of blood vessel widths revealed a higher frequency of larger vessels after treatment with patupilone (Fig. 1C). The occurrence of lung metastases (animals scoring positive for the presence of lung metastases) in the vehicle-treated groups was ∼40% and 70% in the two experiments, whereas it was 0% in the groups receiving 4 mg/kg. Patupilone was well tolerated, with no obvious effects on the well-being or behavior of the mice at either treatment dose in comparison with vehicle-treated controls.

Effect of acute patupilone treatment on vascularity and interstitial fluid pressure of B16/BL6 metastases. The effect of a single dose of 4 mg/kg patupilone on the vascularity of cervical metastases was investigated noninvasively using DCE-MRI to assess vessel permeability (initial slope of GdDOTA uptake), extracellular leakage space (final value for GdDOTA uptake), BFI, and rBVol. Well-established metastases (>14 days old) were examined both before and 2 days after treatment with patupilone or vehicle to allow determination of changes in individual animals. No significant effects of patupilone could be determined on the DCE-MRI measured variables of initial slope of GdDOTA uptake, final value for GdDOTA uptake (IAUC) or BFI compared with vehicle: the fractional changes for patupilone treatment relative to vehicle treatment were 0.94, 1.09, 1.03, and 1.0 for the respective variables. However, most patupilone-treated mice (18 of 22) showed a decrease in rBVol, and compared with baseline, this was highly significant (P = 0.0005; −21 ± 5%), whereas in the vehicle-treated group there was no significant change (−7 ± 6%; see Fig. 2A). The difference between the mean percentage change between the two groups did not reach significance (P = 0.092, two-tailed t test). Representative rBVol maps are shown for one lymph node tumor before and after patupilone treatment (Fig. 2B and C).

In a separate cohort of mice, the IFP of established cervical metastases (>14 days old) was also examined before and 2 days after treatment with patupilone (4 mg/kg) or vehicle. Vehicle-treated mice showed no overall change in IFP (17 ± 15%), whereas patupilone-treated mice showed a significant decrease (P = 0.02) compared with baseline (−32 ± 16%; see Fig. 2D), and this was significantly different to the percentage change in the vehicle group (P = 0.042, two-tailed t test). Vascular casts from a separate cohort of mice showed a trend for patupilone to reduce the vascular index after 2 days: mean 8.4 ± 1.6% for five vehicle-treated mice and mean 5.5 ± 1.4% for six patupilone-treated mice (casts not shown).

Effect of 1.5 mg/kg patupilone treatment in the BN472 rat mammary carcinoma model. A single dose of patupilone was strongly efficacious against the mammary BN472 tumor transplanted orthotopically in rats. One week after treatment, there was little or no tumor growth (T/C, 10.4%), mild diarrhea in some rats, and full survival (8 of 8), but there was significant body weight loss (−22 ± 2%). The overall exposure in these rats (mg/m²) was similar to the high dose given to B16/BL6 mice (i.e., 9 and 14 mg/m² for rats and mice, respectively).

The effect of this single dose of 1.5 mg/kg patupilone on the vascularity of BN472 mammary tumors was investigated noninvasively using DSC-MRI to determine absolute BVol and IFP in the same animals both before and 2 days after treatment with patupilone in comparison with vehicle treatment. Representative BVol maps are shown in Fig. 3. BVol was not significantly affected by vehicle treatment (27 ± 11%), but patupilone induced a significant decrease compared with baseline (P = 0.034, two-tailed paired t test) in 4 of 4 tumors (−42 ± 9%), and this was significantly different to the percentage change in the vehicle group (P = 0.026, two-tailed t test; Fig. 3E). The basal IFP ranged between 9 and 19 mm Hg and this was also not affected by vehicle treatment (21 ± 17%) but was significantly decreased (P = 0.011, two-tailed paired t test) by patupilone in 4 of 4 tumors compared with baseline (−50 ± 4%), and this was significantly different to the percentage change in the vehicle group (P = 0.029, two-tailed t test; Fig. 3F). These changes in BVol and IFP were very similar; indeed, the changes (Δ values) induced by patupilone in these two variables were significantly positively correlated (R = 1.0; P = 0.018). Thus, a lower IFP was associated with a lower BVol. Consistent with a decrease in BVol, vascular casts of BN472 tumors made 2 days after treatment showed a large decrease in the vascular volume compared with vehicle-treated rats of ∼40% (casts not shown).

BN472 tumors: dose dependency of interstitial fluid pressure and volume changes and their correlation. Patupilone induced no changes in TVol after 2 days (data not shown), but after 6 days significant inhibition was apparent at doses of ≥0.8 mg/kg as indicated by the %T/C, and there was a dose-dependent effect on body weight (Table 1). For IFP, significant decreases in comparison with baseline were only recorded for the ≥0.8 mg/kg dose; in general, the decrease tended to be greater after 6 days than after 2 days. The 0.8 mg/kg dose seemed to provide the widest therapeutic index, inducing tumor regression, moderate body weight loss, and significant decreases in IFP of >50%. Vehicle-treated rats showed no significant change in IFP, although some individual tumors showed quite large increases or decreases (several cluster at −30% to −35%) as illustrated by Fig. 4 from the data shown in Table 1.

Figure 4A shows that the largest percentage decreases in IFP measured on day 2 (Δ%IFP_day2-0) were associated with the greatest efficacy as determined by the change in TVol on day 6, and although there was considerable scatter, the two variables were significantly correlated (Pearson R = 0.42; P = 0.005). A similar correlation was observed for the Δ%IFP_day6-0 (Pearson R = 0.42; P = 0.008). Thus, changes in IFP after just 2 days predicted changes in TVol (i.e., tumor response). Furthermore, there was a highly significant association (P = 0.0002, Fisher's exact test) between large decreases in IFP after 2 days and tumor response after 6 days (Fig. 4B). For any dose or all doses pooled, there was no evidence that the baseline IFP was correlated to the tumor response (data not shown).

BN472 tumors: changes in interstitial fluid pressure and volume and correlation with magnetic resonance measured vascularity and histology after 0.8 mg/kg patupilone. As reported above (Table 1), a 0.8 mg/kg dose caused a >50% decrease in IFP at days 2 and 6 (Fig. 5A) but a significant decrease in BVol (−50 ± 4%) was not detectable until day 6 (P < 0.001), Fig. 5B. Patupilone also caused by day 6 a significant decrease (P = 0.01) in the VSI (−40 ± 11%) from 16.0 ± 1 to 9.6 ± 2.1 μm (see Fig. 5C). However, tumor ADC, considered as a marker of apoptosis and/or necrosis, was unchanged by treatment with either vehicle or patupilone (Fig. 5D). Histologic analysis of these tumors on day 6 showed no significant change in necrosis was induced by patupilone, although there was a 2.4-fold increase in apoptosis, which was highly significant (P = 0.000006; see Fig. 5E-H). Thus, in this model, ADC did not reflect changes in apoptosis.

For this subset of data (0 and 0.8 mg/kg), the percentage changes compared with baseline (day 0) observed in IFP (Δ%IFP) at days 2 and 6 were positively and significantly correlated with the change in TVol from days 0 to 6, Δ%IFP_day2-0 (Pearson R = 0.49; P = 0.026; n = 21), and Δ%IFP_day6-0 (Pearson R = 0.54; P = 0.012; n = 21). A similar correlation was observed for the Δ%VSI_day6-0 (Pearson R = 0.49; P = 0.047; n = 17). The Δ%BVol_day6-0 seemed not to be correlated with response, but this was due to one extreme outlier (Grubb's outlier test) in the vehicle group that had a low BVol on day 0 (1.1%) and showed a 400% increase by day 6. Removal of this single tumor revealed a strong positive correlation between BVol and tumor response (Pearson R = 0.7; P = 0.0014; n = 18).

Multiple correlations between the different variables confirmed that there was no significant correlation between ADC and apoptosis and nor with necrosis (P > 0.1). However, TVol, IFP, and BVol were strongly negatively correlated with apoptosis in individual tumors (P < 0.001; i.e., high apoptosis was associated with a small TVol and a low IFP and BVol; Fig. 6A and C). Strong positive correlations were observed between BVol and VSI (P = 0.0018), and these two variables were also strongly correlated with TVol (P = 0.0007 and 0.0023, respectively). Thus, larger tumors had a larger BVol with wider vessels. Finally, the BVol was also strongly positively correlated to IFP (P = 0.0005; see Fig. 6D), confirming observations described above at the 1.5 mg/kg dose. Four other pairs of variables also reached significance (P = 0.023-0.045) but not following a Bonferroni correction (P multiplied by number of correlations done; i.e., n = 21; data not shown).

We have shown that patupilone was strongly efficacious (T/C, ≤20%) at tolerable doses in two different syngeneic orthotopic solid tumor rodent models, including the metastases. Other data, not reported herein but elsewhere (36), also showed significant activity (T/C, 0-39%) at tolerable doses (≤1.75 mg/kg i.v.) in another primary orthotopic rat mammary tumor, MTLn3, including the metastases, and also against established human HT1080 fibrosarcoma cells growing in the lungs of nude mice.

In the mouse B16/BL6 melanoma cervical metastases and the rat BN472 mammary tumor, several different variables relating to tumor vascularity were studied at 2 and/or 6 days after drug treatment and were compared with vehicle-treated animals. In both models, patupilone caused a rapid destruction of blood vessels before a significant change in TVol was detectable. This effect was measurable by ex vivo methods (e.g., histologic staining) and by plastic casts of blood vessels as well as in situ using MRI and by measuring tumor IFP. In BN472 tumors, the data showed that changes in IFP preceded and correlated with tumor response, suggesting this might be a useful clinical surrogate biomarker of a chemotherapeutic response to this drug.

Two weeks after treatment, histology showed that inhibition of growth of B16/BL6 metastases was associated with a significant decrease in blood vessel density (P = 0.014). The histology dye used stains functional vasculature and is therefore an ex vivo equivalent of the noninvasive DCE-MRI measurements of (relative) BVol. DCE-MRI measurements showed that after just 2 days patupilone significantly decreased the rBVol (P = 0.0005). Other vascular variables, such as vessel permeability and overall perfusion, were unaffected, a distinct difference in vascular response in comparison with therapies that target vascular endothelial growth factor (24, 37) or the vascular endothelial growth factor receptor (33) or the endothelial cells directly (38, 39) and not necessarily expected given that other antivascular/disruptive therapies can decrease permeability and/or perfusion (9, 40). Vascular casts of tumors made 2 days after treatment showed a trend for patupilone to induce a decrease in blood vessel mass, and at this time point, the tumor IFP was also decreased (P = 0.02). Thus, the large changes in blood vessel architecture measurable 2 weeks after patupilone treatment were already detectable after just 2 days by the noninvasive method and this was paralleled by a decrease in tumor IFP. Tumor IFP is raised for several reasons, but a major contributor is likely to be the chaotic blood supply (17, 18); thus, a destruction of blood vessels may lead to a reorganization that relieves some of the tumor pressure (41).

The relationship between tumor vasculature and IFP was further investigated in the rat BN472 tumors. Patupilone was efficacious against these tumors with significant changes in tumor size apparent 1 week after treatment. Two days after treatment with a dose of 1.5 mg/kg, large decreases of ∼50% in the BVol and IFP were detectable (P < 0.05) and the changes in these two variables correlated with each other (P = 0.018), suggesting that a low IFP was associated with a low BVol. Vascular casts, an ex vivo method, made at this time point and dose confirmed a trend for patupilone to decrease blood vessel mass.

More extensive noninvasive studies of not only tumor IFP and BVol but also VSI and ADC were made with lower, better tolerated dose of 0.8 mg/kg, which permitted a longer study of 6 days after treatment. At this lower dose, there was still a large decrease in IFP (>50%) after 2 days (P < 0.001), but no changes were revealed in the BVol, VSI, or ADC. However, 6 days after treatment, both VSI (P = 0.01) and BVol (P < 0.001) showed a significant decrease, and the decrease in IFP persisted (P < 0.001), whereas ADC changes were not affected. Histology done on these tumors showed no overall change in necrosis induced by patupilone in comparison with vehicle but showed a significant 2.4-fold increase in apoptosis (P < 0.001). Thus, although patupilone had the expected effect to increase apoptosis in tumors (as measured by caspase-3 staining), this could not be detected by ADC, which has been described as a method to detect changes in cell density and was expected therefore to relate to apoptosis and/or necrosis (42). The explanation for the absence of an effect on ADC is not clear; it may be that the relatively high necrosis, which ranged from 20% to 50% in vehicle-treated rats and was not significantly altered by patupilone treatment, masked further increases in water diffusion caused by the relatively small decrease in cell density expected from a 2-fold increase in apoptosis. Thus, the cell density may already be fairly low in this tumor and further decreases may be difficult to detect using the surrogate marker of water diffusion (i.e., ADC). Nevertheless, both BVol and IFP were strongly negatively correlated with apoptosis, as was the TVol (P < 0.001), suggesting that these two variables measured in situ could be useful biomarkers of tumor cell kill. In confirmation of this hypothesis, the Δ%IFP induced by a dose of 0.8 mg/kg patupilone after 2 days was significantly positively correlated to the ΔTVol (P = 0.026) well before a change in TVol (ΔTVol) was apparent. After 6 days, the Δ%IFP was also significantly correlated to the ΔTVol (P = 0.012), as were the other two noninvasively measured biomarkers, Δ%VSI_day6-0 and Δ%BVol_day6-0 (P < 0.05). Furthermore, BVol and IFP were strongly positively correlated (P = 0.0005) with each other. Thus, a drug-induced reduction in BVol or vessel width whether measured directly by DSC-MRI or indirectly as IFP were effective markers of the tumor response to drug treatment. These effects in the BN472 model are consistent with a normalization of the tumor vasculature (41), but the data in the B16/BL6 model suggest that it did not lead to overall improved perfusion.

Finally, also in the BN472 model, patupilone dose-dependently inhibited tumor growth and dose-dependently decreased the IFP. The Δ%IFP induced by the different doses of patupilone correlated significantly with the subsequent ΔTVol (P = 0.005) again well before a change in TVol was apparent. Hence, the dose-response data confirmed that Δ%IFP could be used to predict tumor response. Although others have shown that different drugs can decrease the IFP of solid tumors, including prostaglandin E₁ (19), dexamethasone (21), imatinib (22), and paclitaxel (23), as far as we are aware, this is the first report showing that the decrease in IFP is directly related to tumor drug response. As hypothesized by Milosevic et al. (43), this drug-induced decrease in IFP may also be a factor in inhibiting subsequent metastases, as we found in three animal models.

In conclusion, the novel microtubule stabilizer, patupilone, at efficacious and tolerable doses, caused a rapid destruction of tumor blood vessels well before a significant change of TVol was apparent. In the rat mammary BN472 model, four potential biomarkers of response were measured in situ, and IFP seemed to be the best predictor of tumor response and also correlated positively with tumor vascularity and negatively with apoptosis. Hence, our observations confirmed that the raised IFP typical of solid tumors is related to blood vessel architecture and tumor cell density, and decreases in IFP may act therefore as a surrogate measure of destruction of tumor cells and blood vessels. In experimental tumors, IFP is relatively easy to measure and, provided the tumors are not very large, is independent of both tumor size and where the IFP is measured in the tumor. In the clinic, however, IFP is not routinely measured and would be expected to be limited to superficial tumors (43). This makes the MRI measurement of BVol attractive, but except for the brain (40, 44) it requires the use of exogenous contrast agents that have yet to be approved for oncology. Similarly, although many Annexin V probes are under development for noninvasive assessment of apoptosis, none have yet been approved for clinical use (45). Thus, our data suggest that measurement of tumor IFP in the clinic is desirable because drug-induced changes in IFP could be a useful surrogate biomarker of tumor response to treatment with patupilone and might even be used for patient stratification. Furthermore, decreases in IFP induced by patupilone or similar compounds may aid uptake of subsequent chemotherapy, including the epothilone itself (46), a strategy that should aid combination chemotherapy or alternative scheduling of monotherapy.

We thank T. Krucker and E. Meyer (Department of Neuropharmacology, Scripps Research Institute, La Jolla, CA) for showing the technology for making casts.