Quantitative Multi-Parametric Magnetic Resonance Imaging of Tumor Response to Photodynamic Therapy

Objective The aim of this study was to characterize response to photodynamic therapy (PDT) in a mouse cancer model using a multi-parametric quantitative MRI protocol and to identify MR parameters as potential biomarkers for early assessment of treatment outcome. Methods CT26.WT colon carcinoma tumors were grown subcutaneously in the hind limb of BALB/c mice. Therapy consisted of intravenous injection of the photosensitizer Bremachlorin, followed by 10 min laser illumination (200 mW/cm2) of the tumor 6 h post injection. MRI at 7 T was performed at baseline, directly after PDT, as well as at 24 h, and 72 h. Tumor relaxation time constants (T1 and T2) and apparent diffusion coefficient (ADC) were quantified at each time point. Additionally, Gd-DOTA dynamic contrast-enhanced (DCE) MRI was performed to estimate transfer constants (Ktrans) and volume fractions of the extravascular extracellular space (ve) using standard Tofts-Kermode tracer kinetic modeling. At the end of the experiment, tumor viability was characterized by histology using NADH-diaphorase staining. Results The therapy induced extensive cell death in the tumor and resulted in significant reduction in tumor growth, as compared to untreated controls. Tumor T1 and T2 relaxation times remained unchanged up to 24 h, but decreased at 72 h after treatment. Tumor ADC values significantly increased at 24 h and 72 h. DCE-MRI derived tracer kinetic parameters displayed an early response to the treatment. Directly after PDT complete vascular shutdown was observed in large parts of the tumors and reduced uptake (decreased Ktrans) in remaining tumor tissue. At 24 h, contrast uptake in most tumors was essentially absent. Out of 5 animals that were monitored for 2 weeks after treatment, 3 had tumor recurrence, in locations that showed strong contrast uptake at 72 h. Conclusion DCE-MRI is an effective tool for visualization of vascular effects directly after PDT. Endogenous contrast parameters T1, T2, and ADC, measured at 24 to 72 h after PDT, are also potential biomarkers for evaluation of therapy outcome.


Methods
CT26.WT colon carcinoma tumors were grown subcutaneously in the hind limb of BALB/c mice. Therapy consisted of intravenous injection of the photosensitizer Bremachlorin, followed by 10 min laser illumination (200 mW/cm 2 ) of the tumor 6 h post injection. MRI at 7 T was performed at baseline, directly after PDT, as well as at 24 h, and 72 h. Tumor relaxation time constants (T 1 and T 2 ) and apparent diffusion coefficient (ADC) were quantified at each time point. Additionally, Gd-DOTA dynamic contrast-enhanced (DCE) MRI was performed to estimate transfer constants (K trans ) and volume fractions of the extravascular extracellular space (v e ) using standard Tofts-Kermode tracer kinetic modeling. At the end of the experiment, tumor viability was characterized by histology using NADH-diaphorase staining.

Results
The therapy induced extensive cell death in the tumor and resulted in significant reduction in tumor growth, as compared to untreated controls. Tumor T 1 and T 2 relaxation times remained unchanged up to 24 h, but decreased at 72 h after treatment. Tumor ADC values significantly increased at 24 h and 72 h. DCE-MRI derived tracer kinetic parameters displayed an early response to the treatment. Directly after PDT complete vascular shutdown was observed in large parts of the tumors and reduced uptake (decreased K trans ) in remaining tumor tissue. At 24 h, contrast uptake in most tumors was essentially absent. Out of 5

Introduction
Photodynamic therapy (PDT) is a photochemistry-based approach for the minimally-invasive local treatment of cancer. It consists of administration of a light-activated chemical, termed a photosensitizer (PS), followed by local irradiation of the tumor with light of the appropriate wavelength, resulting in the generation of cytotoxic species including singlet oxygen [1]. Typically, the light that is used for PDT is in the 600-800 nm wavelength range. PDT is clinically approved and routinely used for treatment of premalignant and malignant non-melanoma skin tumors [2]. Besides, PDT has shown to be promising in the treatment and palliation of head and neck tumors [3], digestive system tumors (e.g. Barrett's esophagus [4]), and prostate cancer [5]. A comprehensive review on PDT principles and applications can be found in Agostinis et al. [6].
Assessment of the treatment outcome is key to therapeutic success, since incomplete treatment may lead to tumor recurrence. The dose-response relation for PDT is notably complex, meaning that therapeutic outcome can vary despite standardized dose. Moreover, light dose and drug dose need to be distinguished. Light dose is usually defined as the fluence at the tumor surface, while drug dose is defined as the amount of injected photosensitizer. However, the actual quantities of photosensitizer and light fluence may vary within tumors and between subjects, due to biological heterogeneity. As a result, treatment response is difficult to predict, and should therefore be evaluated for each individual subject. Preferably, treatment efficacy should be monitored early after therapy, so that poorly responding tumors can be timely identified, in which case repeated or alternative treatment can be started. Furthermore, treatment evaluation should preferably be imaging-based, to enable distinction of spatial heterogeneities in response, which are inherent to tumor non-uniformity of PDT light and drug doses.
PDT causes direct tumor cell damage resulting in apoptosis or necrosis [7]. Additionally, the treatment might lead to vascular occlusion, cutting off the vital supply of oxygen and nutrients to the tumor [8]. Magnetic resonance imaging (MRI) offers techniques to depict both mechanisms of action in a non-invasive and longitudinal fashion. Conventional MRI has been used for PDT response evaluation in a couple of studies, mostly employing T 2 -and diffusionweighted imaging to identify qualitative signal changes or alterations in gross anatomy [9][10][11][12]. Contrast-enhanced (CE) MRI has shown promise for measuring vascular effects of PDT in mammary adenocarcinoma [13] and prostate tumors [11]. The water apparent diffusion coefficient (ADC) is a well-known biomarker to evaluate treatment response in oncology [14]. The ADC is sensitive to necrosis and apoptosis after radiotherapy and chemotherapy [15]. Nevertheless, quantitative MRI has been applied in a few studies only to study PDT response [16][17][18]. Most of these studies focused on a single or a small set of contrast parameters, whereas the spectrum of biological effects associated with PDT calls for a multi-parametric imaging approach. Besides, combined multi-parametric information can often improve accuracy of employed by a commercial company (PhilipsResearch). The funder provided support in the form of salaries for author HG, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the 'author contributions' section treatment evaluation by MRI [19][20][21]. We hypothesize that systematic quantitative multiparametric imaging could help in identifying the most sensitive (combination of) readouts of treatment success.
In view of this hypothesis, the goal of this study was to identify quantitative MRI biomarkers that are suitable for assessment of PDT response. To this end, we evaluated changes in T 1 , T 2 , ADC, and DCE-MRI derived vascular parameters upon PDT in a mouse cancer model up to 72 h after treatment. The CT26.WT colon carcinoma model was used, because this tumor is well-perfused and characterized by little spontaneous necrosis [22], allowing selective detection of treatment-induced vascular effects and tissue necrosis.

Materials & Methods
All animal experiments were performed according to the Directive 2010/63/EU of the European Parliament and approved by the Animal Care and Use Committee of Maastricht University (protocol: 2012-139). Mice were anesthetized using isoflurane in medical air. At the end of experiments, mice were humanely sacrificed by cervical dislocation under anesthesia.

Mouse cancer model
CT26.WT murine colon carcinoma cells (American Type Culture Collection (ATCC), CRL-2638) were cultured as a monolayer at 37°C and 5% CO 2 in RPMI-1640 medium (Invitrogen, Breda, The Netherlands), supplemented with 10% fetal bovine serum (Greiner Bio-One, Alphen a/d Rijn, The Netherlands) and 50 U/ml penicillin/streptomycin (Lonza Bioscience, Basel, Switzerland). Early passages (9-13) of the original ATCC batch were used for inoculation. 10-12 week-old BALB/c mice (Charles River, Maastricht, The Netherlands) were inoculated with 2 Ã 10 6 cells subcutaneously in the right hind limb. Tumors became palpable at 3 to 5 days after inoculation. PDT treatment was performed when tumor diameter reached 10 mm, approximately after 10 days. Average tumor volume was 228 ± 174 mm 3 (mean ± SD) at the day of treatment.
During photosensitizer injection, PDT, and MR imaging, mice were anesthetized using isoflurane (3.5% for induction, 1-2% for maintenance) in medical air flowing at 0.4 L/min. Breathing rate was monitored with a pressure balloon. During PDT and MRI, body temperature was monitored using a rectal probe, and kept at 36-37°C using a warm water circuit.

Study design
In total, n = 22 PDT treated mice and n = 10 untreated control mice were used in this study (S1 Fig). All PDT treated tumor mice were scanned the day before and directly following PDT. After PDT, which was performed outside the MRI scanner, it took approximately 30 min to put the animal in the MRI scanner and start the scanning session. The total scanning time for the complete multi-parametric MRI protocol was approximately 2.5 h. Subsequently, one group of n = 7 mice was killed for histology. Of the remaining 15 mice, n = 10 were scanned at 24 h and killed afterwards, whereas n = 5 were scanned at 72 h and then followed for at most 2 weeks, during which tumor size was measured 3 times a week with a caliper. One control group of n = 5 untreated tumor mice with equal starting tumor size as the PDT group was scanned on 3 consecutive days. After the last scan, these control mice were killed. A second control group of untreated tumor mice (n = 5) with equal starting tumor size as the PDT group was scanned on 2 consecutive days, and at 72 h, and then followed for at most 2 weeks, during which tumor size was measured 3 times a week with a caliper. In all groups, a tumor volume of 1500 mm 3 was used as a humane endpoint.

Photodynamic therapy
The photosensitizer Bremachlorin, also known outside the EU as Radachlorin, was kindly supplied by Harrie Vink and Andrei Reshetnickov from Rada-Pharma International B.V. It is a mixture of chlorins in 0.35% (w/v) aqueous solution for intravenous injection, with chlorin e6 as its main constituent. A dose of 20 mg/kg body weight Bremachlorin was administered via the tail vein 6 h before light exposure. The hind limb was shaved, and a mask of black paper was positioned around the tumor to avoid light entering the surrounding tissue. Subsequently the tumor was exposed to 655 nm light, delivered by a laser diode (WSLB-650-002-H, Wave-Spectrum, Beijing, China) connected to a fiber, which was terminated by a pair of lenses to form a collimated beam with a diameter adjustable to the size of the tumor. The beam was aimed onto the skin covering the tumor, at an irradiance of 200 mW/cm 2 . The laser was turned on for 10 min, resulting in a total fluence of 120 J/cm 2 . Mice received preventive analgesia (s.c. injection of 0.05 mg/kg buprenorphine) 30 min before PDT, and every 12 h during the first 2 days after PDT.

MR image analysis
All image analysis was performed using homemade scripts written in Matlab R2014a. Manual segmentation of tumors was performed based on the T 2 -weighted spin echo scans, in which the tumor signal was hyperintense compared to muscle tissue. 3D FLASH acquisitions were down-sampled in the third dimension to spatially match the 2D multi-slice acquisitions.
For flip-angle correction the signal intensity (S) in each voxel as function of the nominal flip angle (FA n ) was fitted to S = |a Ã FA n + b|, and a flip-angle correction map z was constructed according to z = -(a/b) Ã 180°. For the T 1 -mapping and DCE-MRI scans the actual flip angle in each voxel (FA a ) was obtained by correcting the nominal flip angle using FA a = z Ã FA n .
T 2 -maps were reconstructed by fitting signal intensity (S) as function of TE to S~Exp(-TE/ T 2 ). T 1 -maps were calculated by fitting signal intensity (S) as function of actual flip angle (FA a ) to S~Sin(FA a ) Ã (1-Exp(-TR/T 1 ))/(1-Cos(FA a )Exp(-TR/T 1 )). ADC values per diffusion-gradient orientation were calculated by fitting signal intensity (S) as function of b-value (b) to S~Exp (-ADC Ã b). A mean orientation-invariant ADC value was obtained by averaging the ADC values of the different diffusion-directions.
From the pre-contrast longitudinal relaxation rate R 1,pre = 1/T 1,pre and the dynamic signal changes acquired with a single flip angle during contrast agent influx, the dynamic changes in R 1 (t) = 1/T 1 (t) were calculated. Subsequently, the dynamic changes in contrast agent concentration ([CA]) were calculated using R 1 (t) = R 1,pre + r 1 Ã [CA], where r 1 = 3.53 mM -1 s -1 is the relaxivity of Dotarem at 7 T [26].
For the DCE-MRI, area under the curve (AUC) of Dotarem concentration as function of time was used as a measure of contrast agent uptake by the tumor. Additionally, the standard Tofts-Kermode (TK) model [27] was applied to estimate K trans (transfer constant describing exchange between the blood plasma and the extravascular extracellular space (EES)) and v e (the volume fraction of the EES). The arterial input function (AIF) was constructed from a biexponential function with amplitudes A 1 = 5.36 mM and A 2 = 1.27 mM, and time constant τ 1 = 5.36 s and τ 2 = 915 s. A 1 was estimated from the injected dose divided by the average blood volume for BALB/C mice [28], corrected for a hematocrit level of 0.563 (value provided by animal supplier), whereas time constant τ 1 was taken from literature [29]. A 2 and τ 2 were based on previous experiments [22].

Histological analysis
Immediately after killing the animals, the skin covering the tumor was removed, and tumors were marked with colored lines (MD100-1KT, Sigma Aldrich) as anatomical landmarks. Tumors were excised and snap-frozen in a cold isopentane slush, and then stored at -80°C. For histological analysis, tumors were cut in 8 μm axial sections using a cryomicrotome (Shandon Cryotome, Thermo Fischer Scientific). Sections were stained with a NADH diaphorase tissue viability staining [30] or Hematoxylin and Eosin (H&E).
For the NADH-diaphorase stainging, the cryo-sections were briefly air-dried, and subsequently incubated at 37°C for 1 h in Gomori-Tris-HCl buffer (pH 7.4) containing β-NAD reduced disodium salt hydrate (Sigma-Aldrich, St. Louis, MO, USA, 0.71 mg/ml buffer solution) and nitro blue tetrazolium (Sigma-Aldrich, 0.29 mg/ml buffer solution). Specimens were then washed and mounted for analysis. Brightfield microscopy images of entire sections were performed by mosaic acquisition at 20x magnification.

Statistical analysis
Average endogenous contrast parameter values were calculated in the entire tumor. For DCE-MRI, pixels were classified as non-enhanced if the median contrast agent concentration after injection was smaller than twice the standard deviation of the pre-contrast signal intensity. K trans and v e were averaged over tumor pixels that satisfied the following criteria: 1) pixels were enhanced; 2) R 2 of the TK model fit was larger than 0.80; 3) the maximum concentration was reached before 800 s after bolus injection; 4) K trans and v e values were physiologically realistic (0 K trans and 0 v e 1).
Statistical analyses were performed in SPSS software (version 22). Levene's test was used to test for equality of variances before comparing means of two groups, and an equal-variance or unequal-variance two-tailed t-test was used accordingly. A p value of less than 0.05 was considered statistically significant.

Tumor volume
PDT induced visible and palpable effects in the tumor-bearing hind limb. Within a day, there was swelling and darkening of the skin over the tumor. In some animals, a red ring occurred on the skin around the tumor. Two to three days after PDT, a dry necrotic crust had formed. Based on the MRI-based tumor segmentation, the average tumor volume decreased in PDT treated mice after 72 h, while control tumors continued to grow (Fig 1). For two of the five mice that were followed for 2 weeks after PDT, there was complete remission, and only a small scab remained at the position of the tumor. In the other 3 mice, there was tumor growth arrest or complete necrosis in the bulk of the tumor, but also continued growth of unsuccessfully treated tumor tissue.

Endogenous contrast MRI
The effect of PDT on endogenous MR parameter maps was largely consistent in all mice. Fig 2  shows representative examples of endogenous parameter maps of two animals, up to 24 h and 72 h after PDT, respectively. R 1 and R 2 values in the tumors were rather homogeneous and unaffected by PDT up to 24 h. Typically, widespread increases in tumor R 1 and R 2 were visually observed in the maps only as late as 72 h after treatment, although 5 out of 9 animals showed locally increased R 1 values at 24 h. Tumor ADC values were increased at 24 h and 72 h after treatment. Most mice presented a dry scab of severe necrosis, associated with large R 1 and R 2 , and low ADC in the superficial tumor regions. Except for low ADC values in such regions with advanced necrosis, the ADC in the bulk of the tumor at 72 h was increased compared to the 24 h time point. Decrease in R 1 and R 2 , and increase in ADC were observed directly after treatment in the skin covering the tumor and the muscle underneath, which could be attributed to therapy-induced edema.

Histology
Histology revealed changes in tumor tissue viability early after PDT, despite the absence of changes in R 1 , R 2 , and ADC. Fig 4 shows some representative NADH diaphorase stained sections of tumors excised at different time points. A patchy pattern with viable and non-viable tumor tissue was observed directly after PDT. At 24 h after PDT, all tumors contained mostly non-viable cells. H&E staining confirmed that the regions which did not stain with NADH diaphorase were non viable, showing eosinophilic cytoplasm, hyperchromatic nuclei, and fragmentation of nuclei and cells. Interestingly, a NADH diaphorase viable tumor rim of 50-150 μm width just below the skin was observed in many treated tumors 24h after PDT. Control tumors were mostly completely viable, with the exception of some small necrotic regions.

DCE-MRI
In stark contrast to R 1 , R 2 , and ADC, response in DCE-MRI scans was observed directly after PDT (Fig 5). Area under the curve (AUC) was used as measure for the degree of contrast agent uptake. Whereas prior to treatment tumor AUC was generally similar or higher as compared to surrounding muscle, the AUC was sharply decreased in large areas of the tumor directly after PDT. The region with low AUC extended to the whole tumor and into the surrounding muscle at 24 h and later. Interestingly, while the tumor AUC was decreased, remote muscle tissue distant from the treatment area became more permeable to contrast agent. Fig 6 summarizes the non-enhanced tumor fraction and DCE-MRI kinetic parameters for the treatment and control groups. Directly after PDT, the average non-enhanced tumor fraction for treated mice was significantly higher than baseline. It further increased to 77% at 24 h  and remained at roughly this value up to 72 h post PDT. K trans in tumor tissue with significant AUC was significantly lower directly after PDT, as compared to baseline values and control tumors. The extracellular extravascular volume fraction v e was significantly higher at this time point. K trans and v e at 24 h and 72 h after PDT are not reported, since these parameters could only be calculated in a small number of tumor pixels at these time points, e.g. due to insignificant AUC.
For the five mice that were followed for up to 2 weeks after PDT, Fig 7 shows tumor growth curves based on caliper measurements, and AUC maps obtained right after PDT and 72 h after PDT. At both time points, the two animals with complete response (A and B) had higher nonenhanced fractions than the three animals with tumor recurrence (C-E). Interestingly, the Regions of tumor recurrence were characterized by significant contrast enhancement directly after PDT, and K trans values above the treated tumor average (i.e. K trans > 0.11 min -1 ). However, some successfully treated tumor tissue also showed contrast enhancement and high K trans values. None of the animals with tumor recurrence had lower tumor average K trans right after PDT (0.14, 0.20, and 0.10 min -1 ) than those with complete response (0.10 and 0.07 min -1 ).

Discussion
Multi-parametric MRI was performed to characterize changes in tumor tissue structure and vasculature in response to PDT in a murine tumor model. Our goal was to identify those imaging biomarkers that could be suitable for early evaluation of PDT outcome, preferably within one day after treatment. Quantitative Multi-Parametric MRI of Tumor Response to PDT DCE-MRI showed the fastest response after treatment. Directly after treatment, considerable tumor parts were non-enhanced, which we attribute to vascular occlusion induced by PDT. In the remaining tumor tissue K trans was decreased compared to baseline and controls, indicating that blood perfusion was compromised. In this tissue the process towards vascular stasis may still have been in progress, as at 24 h and later the complete tumor remained nonenhanced after contrast agent injection. The DCE-MRI data of 5 animals that were sacrificed 2 weeks after PDT suggested that remaining contrast-enhancement at 72 h is an indicator for tumor recurrence. Future work could be aimed at optimizing the PDT to prevent tumor recurrence using DCE-MRI as an early indicator of treatment success.
It is known that PDT can induce severe and quick vascular responses [8]. Vascular occlusion is believed to be initiated by endothelial cell damage, as a result of singlet oxygen production by photosensitizer bound to membranes and other vital parts of endothelial cells. This leads to disintegration of the cytoskeletal structure, cell rounding, and opening of the tight junctions between endothelial cells, exposing the basement membrane. Platelets and leukocytes can bind to these sites of damage, triggering vessel constriction and platelet aggregation through the release of signaling substances. Secondly, direct platelet damage caused by activated photosensitizer in the blood is a mechanism that likewise leads to blood stasis.
Similar observations concerning the vascular effects of PDT using DCE-MRI were done by Zilberstein et al. [31], although in their study tumor illumination was started immediately after injection of the photosensitizer. They observed a complete arrest of vascular perfusion 24 h post PDT, but also a marked increase in vessel permeability 1 h post treatment. We did not observe increased K trans directly after PDT, but rather a decrease. This difference might be due to the fact that our first DCE-MRI measurement was performed at approximately 2.5 h instead In contrast to DCE-MRI, the endogenous contrast parameters T 1 , T 2 , and ADC only significantly changed at the later time points. This was somewhat surprising in view of the structural tissue defects that were observed in histological analysis for the treated tumors at this time point. Similarly, Haider et al. found unaltered T 2 -weighted contrast in human prostate 1 week after PDT [11]. On the other hand Fei et al. reported an increase in T 2 at 24 h post PDT [32]. In the latter study, a different tumor model, photosensitizer, and treatment protocol were used. Specifically, tumors were exposed to light 48 h after photosensitizer injection, which might have resulted in more cellular uptake than in our study. Wang et al. studied ADC changes in response to PDT with a phthalocyanine photosensitizer and observed increased ADC values in tumors 24 h post treatment [17], in line with our findings.
The DCE-MRI enhancement maps revealed considerable volumes of non-perfused muscle around the treated tumors. Muscle damage surrounding the tumor was also observed in H&E sections. This could be caused by ischemia as a result of PDT-induced vascular occlusion. Alternatively, the tumor-to-muscle photosensitizer uptake ratio may have been too low to selectively ablate the tumor. In an HT29 human colorectal xenograft model, photosensitizer fluorescence intensity in tumors was generally less than 2 times higher than in muscle [33].
Generally, PDT is based on a combination of direct cellular damage and vascular shutdown. The relative contributions of both mechanisms depend on the pharmacokinetics of the photosensitizer, but can be manipulated by varying the time between injection and light administration. For our study design, we expected that the direct cellular pathway would be Quantitative Multi-Parametric MRI of Tumor Response to PDT most dominant, as Park et al. reported a clearance of Bremachlorin from the blood of C57BL/6 mice: 80% clearance was found 6 hours after i.v. injection [34]. However, our DCE-MRI results suggest that sufficient photosensitizer was present in the blood after 6 h for a strong vascular effect in both tumor and muscle tissue. In order to achieve a more tumor-specific treatment, a longer drug-light interval may be required, to reach sufficient blood clearance and prevent damage to normal tissues through vascular effects.
We have shown in a mouse model that MRI, especially DCE-MRI, is a powerful tool for visualization of tumor response to PDT. Our results also suggested a relation between contrastenhancement at 72 h post PDT and tumor recurrence, which is crucial for the clinical translatability of the methods. Future research will be aimed at investigating if DCE-MRI measurements early after PDT can be used to classify tumor tissue as successfully treated (viable) or non-successfully treated (non-viable), based on a spatial comparison with histological slice stacks of entire tumors. Furthermore, a study in which more animals are followed-up for a longer period after therapy would allow determining the actual predictive value of the MR biomarkers with regard to long-term therapeutic outcome.

Conclusion
We presented a multi-parametric MRI follow-up of PDT in a mouse model of cancer. Shutdown of tumor perfusion was observed by DCE-MRI directly after PDT. A late response to the treatment was observed in the endogenous contrast parameters T 1 , T 2 , and ADC. The methods presented here can be applied for optimization of PDT treatment protocols, mechanistic research, but should also be clinically translatable for treatment evaluation of patients.