Dynamic contrast-enhanced magnetic resonance imaging: applications in oncology.

Clinical Oncology xxx (2014) 1e12 Contents lists available at ScienceDirect

Clinical Oncology journal homepage: www.clinicaloncologyonline.net

Overview

Dynamic Contrast-enhanced Magnetic Resonance Imaging: Applications in Oncology Q.Q. Teo *, C.H. Thng y, T.S. Koh y, Q.S. Ng z * Duke

NUS Graduate Medical School Singapore, Singapore Department of Oncologic Imaging, National Cancer Centre Singapore, Singapore z Department of Medical Oncology, National Cancer Centre Singapore, Singapore y

Received 6 February 2014; received in revised form 22 April 2014; accepted 28 April 2014

Abstract Dynamic contrast-enhanced magnetic resonance imaging (DCE MRI) allows functional characterisation of tissue perfusion characteristics and acts as a biomarker for tumour angiogenesis. It involves serial acquisition of MRI images before and after injection of contrast, as such, tissue perfusion and permeability can be assessed based on the signal enhancement kinetics. The ability to evaluate whole tumour volumes in a non-invasive manner makes DCE MRI especially attractive for potential oncological applications. Here we provide an overview of the current research involving DCE MRI as a biomarker for the diagnosis and characterisation of malignancies, prediction of the therapeutic response and survival outcomes, as well as radiation therapy planning. Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. Key words: Angiogenesis; dynamic contrast-enhanced imaging; functional imaging; magnetic resonance imaging; perfusion imaging

Statement of Search Strategies Used and Sources of Information Pubmed was used for the literature search.

Introduction Tumour Angiogenesis Inducing angiogenesis is one of the hallmarks of cancer [1]. The ability to stimulate the formation of new blood vessels is a prerequisite for tumour expansion. In tumour cells, angiogenesis is modulated through a variety of signalling pathways targeting the ligand receptors of endothelial cells, for example, vascular endothelial growth factor (VEGF), angiopoietin-1 and placental growth factor, etc. [2]. The resultant tumour neovasculature is structurally and functionally aberrant. Tumours typically present with Author for correspondence: Q.S. Ng, Department of Medical Oncology, National Cancer Centre Singapore, 11 Hospital Drive, S(169610), Singapore. Tel: þ65-64-368-000. E-mail address: [email protected] (Q.S. Ng).

hypervascularity, excessive branching, irregularity and tortuosity [3]. Blood flows in a highly haphazard manner, ranging from near zero to several times that of the surrounding normal tissue. This is accompanied by increased permeability and leakiness, contributed by the lack of muscularis propia, widened inter-endothelial junctions, lack of basement membranes and presence of vesiculovascular organelles [4,5]. Functional characterisation of tumour microvasculature has been of great interest in the field of oncology imaging as there is evidence for tumours to exhibit a wide spectrum of anomalies in perfusion and these have been correlated with tumour characteristics and clinical outcome, and is a potential therapeutic target. Basic Principles of Dynamic Contrast-enhanced Magnetic Resonance Imaging As paramagnetic substances, gadolinium-based contrast agents shorten T1 and T2 relaxation times and enhance signal intensity on T1-weighted magnetic resonance imaging (MRI). It is administered intravenously, and then travels through the heart and arteries to the capillaries, where it extravasates into the interstitial space. The signal

0936-6555/$36.00 Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.clon.2014.05.014

Please cite this article in press as: Teo QQ, et al., Dynamic Contrast-enhanced Magnetic Resonance Imaging: Applications in Oncology, Clinical Oncology (2014), http://dx.doi.org/10.1016/j.clon.2014.05.014

2

Q.Q. Teo et al. / Clinical Oncology xxx (2014) 1e12

enhancement represents the averaged concentration of the contrast agent within the vascular and interstitial spaces. The change in concentration of contrast with time is a function of the blood flow, permeability, density of vasculature and interstitial space. It is well known that most tumours display a characteristic rapid enhancement followed by an earlier contrast wash-out pattern. This may be attributed to the angiogenic process, in which the rapid proliferation of blood vessels results in increased blood flow, vascular space and capillary permeability. Therefore, contrasts allows for non-invasive evaluation of the perfusion and permeability of the tumour microvasculature. Dynamic contrast-enhanced (DCE) MRI involves the acquisition of serial three-dimensional images at high temporal resolution after a rapid injection of contrast, so as to trace the first passage of contrast bolus through the tissue vasculature. Before that, a preliminary anatomical scan is carried out for slice localisation, such that each slice contains both the target lesion and an appropriate large blood vessel, which is used to obtain an arterial input function that allows the estimation of contrast concentration within the tumour blood vessels. In addition, baseline T1 mapping is carried out to calculate contrast concentration, which may not be linearly related to signal intensity due to the underlying native T1 of tissues [6]. Image Analysis Interpretation of the contrast agent concentrationetime curves can be carried out qualitatively, semi-qualitatively or quantitatively. In the qualitative method, a radiologist visually inspects the shape of the curve with respect to its initial rise and delayed phases. The concentrationetime curves may be described as having a ‘rapid’, ‘medium’ or ‘slow’ initial rise and a ‘persistent’, ‘plateau’ or ‘wash-out’ delayed phase. This method has been used for breast cancers as a complement to existing methods of diagnosis [7,8]. However, it is susceptible to both inter-observer and intra-observer variability, especially with respect to differentiating between the persistent and plateau or plateau and wash-out enhancement types [8]. The semi-quantitative analysis uses curve-derived measurements, such as time to onset, time to peak, initial and mean slope of signal enhancement curve, maximum signal intensity and wash-out gradient [9]. However, these methods are subject to variations in the baseline T1 of tissues, data acquisition method, scanner settings and patient factors such as cardiac output. The quantitative analysis is carried out either through a model-free or model-based approach. The former takes an integration of the concentrationetime curve over a period of time to give the initial area under curve (IAUC). Although the IAUC is presumed to be associated with blood flow and permeability, it does not differentiate the effects of each component. The model-based method describes tissue perfusion using tracer kinetic models that incorporate physiologically meaningful parameters such as blood flow, capillary permeabilityesurface area product, intravascular

space, extravascular space and mean transit time [6]. These parameters are computed by fitting the tracer kinetic model to concentrationetime curves derived from the DCE imaging data set. Several models have been developed, including conventional compartmental, generalised kinetic and distributed parameter models.

Applications in Oncology To date, studies have investigated the feasibility of DCE MRI to improve methods of cancer detection, characterisation, response assessment, prognosis and radiation therapy planning (Table 1). Diagnosis and Characterisation Cancer Detection and Localisation The ability to detect cancers at an early stage helps to reduce mortality. In some cancers, standard morphologybased imaging techniques are not able to discriminate between benign and malignant tumours, and functional characterisation of the tumours can aid in classifying these lesions. Multiparametric imaging is gaining popularity in the diagnosis of cancers and DCE imaging is one of the modalities that are commonly used. Breast cancers are difficult to diagnose using mammograms in patients with dense breast tissues, yet MRI has poor specificity. On DCE imaging, malignant breast lesions may be distinguished by its rapid initial enhancement with subsequent wash-out pattern. Jansen et al. [10] showed that contrast uptake and signal enhancement ratios were able to identify malignant lesions with sensitivities and specificities in the ranges of 90% and 20e30%, respectively. Additionally, DCE MRI may aid the diagnosis of sonographically indeterminate adnexal masses, which make up almost 20% of adnexal masses [30]. For such cases, surveillance with repeat imaging is a common approach. However, this may cause a delay in treatment if there is a malignancy. Timely diagnosis is important, as treatment differs from simple resection in benign lesions, to radical surgical exploration in malignant lesions. Thomassin-Naggara et al. [11] showed the feasibility of characterising these masses by showing that malignant adnexal tumours had greater tissue blood flow and blood volume fraction and smaller interstitial volumes as compared with benign tumours. Treatment-induced permeability changes result in nonspecific contrast uptake that may resemble residual tumour or tumour recurrences, making it especially challenging to identify tumours by contrast-enhanced imaging alone. In glioblastoma surgery, where it is crucial to maximise tumour resection without causing additional neurological deficits, intraoperative DCE MRI may provide a quick assessment for residual tumours at the resection border. € Based on a study by Ozduman et al. [12], there were significant differences in the time-intensity curves, Ktrans (permeability constant), Kep (efflux constant), Ve (volume of extravasculareextracellular space) and IAUC, between remnant tumour and surgically induced enhancement. In



3

Table 1 Selected studies investigating the use of dynamic contrast-enhanced magnetic resonance imaging (DCE MRI) for diagnosis and characterisation, response monitoring and prediction, prognosis assessment and radiation therapy planning Applications

DCE biomarker

Diagnosis and characterisation BI-RADs classification, To differentiate benign and malignant breast modified empirical lesions mathematical model parameters (a, b), IAUC30, slopeini, Tpeak, SER, kpeak

End point

Significant results

Reference

Histological classification (malignant: IDC, DCIS, ILC; benign: fibrocystic change, fibroadenoma, papilloma, others)

a(contrast uptake), slopeini, SER and kpeak were significantly greater in malignant as compared with benign lesions (P < 0.03, P < 0.04, P < 0.0007, P < 0.02, respectively) Malignant tumours had significantly greater tissue blood flow, blood volume fraction, AUC (P < 0.0001, 0.0006, 0.04, respectively) and smaller interstitial volume fraction (P ¼ 0.0002). Area under ROC for discriminating malignant tumours by tissue blood flow was 0.86. At the resection border, climbing-type time-intensity curve (rapid initial enhancement followed by plateau) was seen with solid tumour tissue, whereas mildly enhancing low-amplitude-type time-intensity curve (slow increase in enhancement) was seen with non-tumour tissue. Measurements from tumour, normal brain and surgically induced enhancement of Ktrans, Kep, Ve and IAUC were significantly different (P ¼ 0.023, 95% necrosis)

A greater percentage decrease in DKtrans from baseline to 2 weeks after initiation of sorafenib is associated with a higher percentage of histological necrosis (P ¼ 0.012) and was able to discriminate 2 of 3 optimal responders from 5 suboptimal responders

[20]

Anatomical response by RECIST criteria

Baseline Kep over whole region of tumour was significantly higher for responders than nonresponders (P < 0.001) Responders showed significant reductions in Kep over selective regions (areas of highest enhancement in early arterial phase) at 6 weeks (P ¼ 0.003), 12 weeks and 18 weeks after therapy. Non-responders showed significant elevation in Kep over selective regions of interest at 18 weeks (P ¼ 0.044) The earliest discrimination of PD was 6 and 18 weeks for DCE MRI and RECIST, criteria respectively. For DCE MRI acquired before start of external beam radiotherapy there were significant positive correlations between several parameters (slope, maximum slope, contrast enhancement ratio, Ktrans and kep) and tumour regression (P ¼ 0.022e0.046), whereas peak time was negative correlated to tumour regression (P ¼ 0.045)

[21]

To predict and monitor the therapeutic response in patients with inoperable liver metastasis from colorectal carcinoma treated with a combination of FOLFIRI and bevacizumab

To predict the response of cervix cancer to concurrent chemoradiotherapy

Arrival time, peak time, slope, maximum slope, contrast enhancement ratio, wash-out slope, Ktrans, kep, fraction plasma volume and IAUGC

Prediction of survival outcomes Brix model: ABrix, kep, kcl To predict survival outcomes and disease Tofts model: control in patients with Ktrans, ve locally advanced cervical cancer using pretreatment DCE MRI

Percentage tumour regression

Locoregional control and progression-free survival

Patients with greater ABrix, kcl and Ktrans had a significantly longer progression-free survival and locoregional control (P < 0.05), with Abrix showing the strongest association. From multivariate analysis, Kcl and Ktrans were independent predictors of locoregional control. Kep correlated positively with progression-free survival but not locoregional control, whereas ve was not associated with any of the end points.

[22]

[23]

(continued on next page)


6

Q.Q. Teo et al. / Clinical Oncology xxx (2014) 1e12 Table 1 (continued ) Applications

DCE biomarker

End point


Reference

To investigate pre- and post-treatment DCE MRI parameters as prognostic biomarkers for disease-free and overall survival in patients with breast cancer undergoing NAC

Ktrans, ve, kep, IAUGC60, relative blood volume, relative blood flow, mean transit time

Disease-free survival and overall survival

Higher values of post-treatment Ktrans and IAUGC60 were significantly associated with worse disease-free survival (P ¼ 0.048 and 0.035, respectively) and overall survival (P ¼ 0.043 and 0.029, respectively).

[24]

Haemtoxylin and eosin stained prostatectomy specimens

Tumour coverage by MRI-based delineations was 44e76% before and 85e100% after application of 2-voxel margins. Volumes of MRI-based delineations were 64e174% of the actual tumour volumes. Errors in tumour delineation were due to registration errors of MRI images with threedimensional reconstruction and heterogeneity of tissue characteristics Signal enhancement slope on baseline DCE was significantly higher for patients with locoregional recurrence during follow-up as compared to those without (26.2 versus 17.5s-1, P ¼ 0.003)

[25]

The local control group has significantly smaller subvolumes with low blood volume at baseline and at second week of treatment compared with the local failure group (baseline median of 9.9 versus 31.9 ml, P < 0.02; second week median of 3.7 versus 23.8 ml, P < 0.01). The local control group has significantly greater decrease in subvolumes from baseline to second week of treatment, compared with the local failure group (56 9% versus 23 12%, P < 0.05). Differences were less significant between the 2 treatment groups in terms of subvolumes with low blood flow at baseline (P ¼ 0.07) and 2 weeks of treatment (P ¼ 0.05). Local failure can be predicted, with area under ROC curve of 0.925 and 0.947 at baseline and 2nd week, respectively.

[27]

Applications of DCE in radiation therapy Ktrans To evaluate the accuracy of gross tumour volume delineations determined by functional MRI (DCE, DWI and T2-weighted) of prostate tumours

To evaluate potential of functional imaging with FDG-PET, FMISOPET, DWI and DCE MRI for dose painting in radiotherapy for head and neck squamous cell carcinoma To characterise tumour subvolumes using DCE MRI and evaluate its association with treatment outcomes in patients with advanced head and neck squamous cell carcinoma treated with concurrent chemoradiotherapy

Maximum slope of signal enhancement

Locoregional recurrence volume delineated on diagnostic computed tomography scan

Blood volume, blood flow

Local failure Local control

[26]



7

Table 1 (continued ) Applications

DCE biomarker

End point


Reference

To develop an approach to quantify early changes in subvolumes of DCE MRI defined physiological parameters and evaluate its ability to predict treatment response in patients with brain metastasis treated with wholebrain radiotherapy

Regional cerebral blood volume, Ktrans

Percentage change in gross tumour volume at posttreatment imaging follow-up: >25% decrease ¼ response; >25% increase ¼ progression; 25% to 25% change ¼ stable

[28]

To evaluate DCE MRI for assessing liver function in patients with unresectable intrahepatic cancers receiving focal radiotherapy

Hepatic arterial and portal venous perfusion (derived from dual input single compartment model)

Indocyanine green clearance rate

A greater extent of reduction in high regional cerebral blood volume-subvolumes was observed in responsive tumours as compared with progressive tumours (P ¼ 0.0072) or progressive and stable tumours combined (P ¼ 0.0057). Similarly, a greater extent of reduction in high Ktrans subvolumes was observed in responsive tumours compared with progressive tumours (P ¼ 0.0406). Combining both parameters, prediction of tumour response was improved (between response and progression, P ¼ 0.0012; between response and stable þ progression, P ¼ 0.0017) Comparison between DCE MRIderived portal venous perfusion with the indocyanine green clearance rate showed good correlation (r ¼ 0.70, P < 0.0000001). Significant reductions in perfusion were observed from pretreatment to 1 month after completion of treatment (104.0 11.1 versus 73.2 6.8, P < 0.03). Regional perfusion correlated linearly with the accumulated local dose received (r ¼ 0.95, P < 0.0001) A combination of local bio-dose, perfusion at baseline and at mid-course (after delivery of 60% of planned dose) significantly improves prediction of post-treatment perfusion as compared with just baseline perfusion and local bio-dose alone. Areas with lower radiation dose displayed compensatory improvement in venous perfusion from 25e58 to 33 e78 ml/100 g/min.

[29]

IAUC, initial area under curve; slopeini, initial slope; Tpeak, time to peak enhancement; SER, signal enhancement ratio; kpeak, curvature of the peak; Ktrans, permeability constant; kep, efflux constant; ve, volume of extravascular-extracellular space; SIslope, signal intensity slope; ABrix (amplitude); kcl, plasma clearance rate; RCC, renal cell carcinoma; pCR, pathological complete response; FOLFIRI, irinotecan, 5-fluorouracil, leucovorin.

patients on follow-up with high-grade gliomas treated with surgical excision and adjuvant radiation and chemotherapy, Larsen et al. [13] found that cerebral blood volume measured by DCE was associated with a higher metabolic

rate and clinical progression and, thus, was able to differentiate tumour recurrences and radiation necrosis. For patients with prostate cancer after external beam radiotherapy or radical prostatectomy, it was shown that


8


DCE MRI improved the diagnostic performance of local recurrence detection as compared with T2-weighted MRI, at a pooled sensitivity of 85 and 90% and a pooled specificity of 95 and 81%, respectively [31]. Tumour Characterisation and Subtyping Prostate cancers usually present with a broad spectrum of aggressiveness. In patients with localised prostate cancers there is often a dilemma between avoiding overtreatment of low-risk prostate cancers and not missing those with high risk of lethality that require immediate radical treatment [32]. There is a need for a more reliable method of characterising tumour aggressiveness that can help to select low-risk candidates who would probably benefit from active surveillance [33]. In a study of the correlation between DCE MRI parameters and the histological grade of tumours it was shown that highly aggressive tumours had significantly higher wash-in, wash-out and Ktrans when compared with less aggressive tumours [14]. The results support the understanding that more aggressive tumours tend to undergo rapid angiogenesis, leading to increased density and leakiness of the vessels. Therefore, DCE MRI parameters may play a role in identifying patients who are suitable for active surveillance. There is a role for DCE imaging in the distinction between tumour subtypes, which may guide the clinical management of patients. For example, tyrosine kinase inhibitors are known to be effective for patients with clear cell renal cell carcinoma (RCC), but may be less effective for other histologies [15]. However, a pretreatment diagnosis of tumour subtype is challenging for patients who have metastatic disease or are unsuitable for surgery, as percutaneous biopsies are subjected to inaccuracies from sampling errors and risks of procedural complications. Sun et al. [15] found significant differences in the pre- to post-contrast changes in signal intensity between clear cell, papillary, and chromophobe RCC. This correlates with histological evidence that clear cell RCC tends to be more hypervascular than papillary or chromophobe RCC. DCE imaging allows for non-invasive assessment of the entire tumour, making it very promising as an adjunct to biopsies for the determination of tumour subtypes. In addition, visualisation of tumour perfusion by DCE may provide further information about its angiogenic and metastatic potential. Lollert et al. [16] found that a higher fractional volume of interstitial space was associated with a greater tendency for lymph node metastasis, whereas k21, which reflects vessel permeability, correlated positively with the presence of distant metastasis and epidermal growth factor receptor (EGFR) expression [16]. Haldorsen et al. [17] showed that increased microvascular proliferation, measured by histomorphological markers such as microvessel density and Ki-67/factor VIII expression, resulted in significantly lower tumour blood flows and increased permeability, revealed by DCE MRI. This suggests that highly proliferating microvasculature contributed to dysfunctional vessels and compromised blood flow and the microvasculature characteristics from DCE MRI were congruent with histological findings.

Biomarkers for Clinical Outcomes Assessment and Prediction of Therapeutic Response The advent of targeted therapies that inhibit angiogenesis has led to improved survival benefits. The response to these drugs varies widely among individuals. The challenge lies in identifying patients who may or may not respond to anti-angiogenic therapy, which will spare them from the unwanted side-effects and costs incurred from futile therapy. However, patients receiving anti-angiogenic therapy often do not achieve remarkable tumour shrinkage, despite improvement in survival. In addition, functional vascular changes often precede structural changes. Therefore, standard morphology-based imaging methods that rely on anatomic dimensions of tumours may not adequately reflect the clinical benefit from anti-angiogenic treatment. On the other hand, histology-based methods of measuring the microvessel density of tissue biopsies will result in sampling errors in a heterogeneous tumour and serial measurements will not be feasible due to its invasive nature. As such, there is a need for an accurate and non-invasive means, such as DCE imaging, for assessing changes in the microcirculation over the entire tumour region (Figure 1). Abramson et al. [18] investigated the feasibility of DCE MRI providing early prediction of breast cancer pathological response to neoadjuvant chemotherapy. Using a semiquantitative approach, voxel-based delayed-phase enhancement was classified as progressive, plateau or washout. It was found that patients with a pathological complete response (pCR) showed significantly greater reductions in terms of the percentage of voxels with wash-out enhancement patterns as compared with those with no pCR. However, changes in the sizes of the lesions did not discriminate patients with pCR and non-pCR. This suggests that the biological response to treatment precedes gross changes in sizes and thus is a superior biomarker for response prediction. DCE imaging can facilitate clinical trials by acting as a surrogate for the functional effects of angiogenesis inhibitors. In a phase I study conducted on children and adolescents with soft tissue sarcoma and other refractory solid tumours who had DCE MRI scans to assess pazopanibinduced changes, there was a significant reduction from baseline to the 15th day of treatment, in both fractional tumour blood flow and permeability [19]. The results from functional imaging were consistent with those of the serum biomarker analysis, in which the anti-angiogenic effect was based on significant decreases in the plasma soluble VEGFR2 and endoglin levels. In another trial by Meyer et al. [20], the effects of adding sorafenib to a preoperative chemoradiotherapy treatment regimen of patients with high-risk extremity soft-tissue sarcomas undergoing limb salvage surgery were evaluated using DCE MRI. The results were consistent with previous studies, showing significant reductions in Ktrans. As such, DCE imaging has the potential to be used in drug trials as a reliable biomarker for evaluating the activity and efficacy of new therapeutic agents. There have been several studies investigating the feasibility of using pretreatment DCE MRI to predict the response



9

Fig 1. Dynamic contrast-enhanced magnetic resonance imaging of a patient with advanced hepatocellular carcinoma showing a reduction in blood flow (Fp) and vascular volume (vp) after treatment with a combination of AZD6244 and sorafenib.

to treatment, therefore facilitating clinical decisions even before therapy. In the study by Coenegrachts et al. [21] of patients with inoperable colorectal liver metastasis treated with a combination of FOLFIRI (irinotecan, 5-fluorouracil, leucovorin) and bevacizumab, it was found that responders, as defined by the RECIST criteria, had a significantly greater baseline Kep than non-responders. Similar results were observed with cervix cancer treated with concurrent chemoradiotherapy, as several parameters (e.g. Ktrans and kep) from pretreatment DCE MRI were positively correlated to the percentage of tumour regression [22]. This is probably because a higher Kep corresponds to a better delivery of contrast or therapeutic agents into the interstitial space, leading to better response rates. As such, DCE MRI may be a viable method for predicting tumour response even before the initiation of therapy.

Several studies have assessed the potential of DCE imaging as a prognostic biomarker. DCE imaging allows characterisation of tumour microvasculature and reflects angiogenesis, which is a key process in cancer invasion. Andersen et al. [23] assessed pretreatment DCE MRI of patients with locally advanced cervical cancer and noted a positive correlation between clinical outcomes such as progression-free survival and locoregional control, and ABrix (amplitude), kcl (plasma clearance rate) and Ktrans. Li et al. [24] studied the association between DCE MRI parameters and survival outcomes in patients with breast cancer undergoing neoadjuvant chemotherapy and noted that greater post-treatment values of Ktrans and IAUC were associated with worse disease-free survival and overall survival.

Radiation Oncology Prediction of Survival Outcomes Prognostication plays a crucial role in the making of medical and personal decisions [34]. Patients who are predicted to have a high risk of a bad outcome without treatment need to be treated aggressively, whereas more conservative treatment options may be offered to those with a better prognosis. If the prognosis is predicted to be very poor with or without treatment, a palliative approach may be considered.

Delineation of Radiation Targets Modern technology for radiation therapy has made feasible conformal dose distributions that enable intensive radiation doses to be given to the targeted tumour region, while minimising exposure to the neighbouring tissues. For example, intensity-modulated radiotherapy permits concentrated doses to be given in a specific manner through non-uniform beam intensities and steep dose gradients.


10


Target delineation for intensity-modulated radiotherapy is relevant in the context of prostate cancers, for which local control with intensive radiotherapy is an important factor in preventing recurrences and yet dose increases are often limited by radiation-induced complications. Groenendaal et al. [25] investigated the validity of functional imaging modalities for the detection and delineation of the gross tumour volume (GTV) and radiation boost volumes in prostatic cancers. The GTV was determined by a radiation oncologist based on information from DCE-derived Ktrans, apparent diffusion coefficient and T2-weighted maps and was compared against delineations determined by pathological evaluation of prostatectomy specimens. There was reasonable correspondence between the two methods, evidenced by tumour coverage of 85e100% using MRI-based delineations with 5 mm margins [25]. As such, there is potential in the use of multimodal functional imaging for improving geometric accuracy of target delineation in radiation planning. Dose Painting and Response Prediction Tumours are highly heterogeneous in nature, displaying varied biological characteristics and radiation sensitivity. Dose painting addresses this by identifying regions within the tumour that are more aggressive or radiation-resistant and intensifying doses to these parts, through subvolume boosting (escalating dose for subvolumes within the tumour) or dose painting by numbers (voxel-level dose prescription based on quantitative information). DCE imaging is promising due to its ability to detect abnormalities in perfusion and permeability characteristics and these have been found to correlate with increased tumour aggressiveness, reduced efficacy of drug delivery and hypoxia, which are recognised factors for treatment failure. Sites of potential treatment failure may be selected for dose escalation with the aim of improving disease control. Dirix et al. [26] evaluated the potential use of DCE imaging, together with other functional imaging modalities, for dose painting and radiotherapy response prediction in patients with head and neck squamous cell carcinoma treated with concomitant chemoradiotherapy. There was a significant positive correlation between baseline signal enhancement slope and locoregional recurrence, suggesting that increased angiogenic activity was associated with high aggressiveness and greater recurrence rates. Conversely, Wang et al. [27] showed that local failure in head and neck squamous cell carcinoma was associated with greater subvolumes of low blood volume and low blood flow, which was explained by poor perfusion causing hypoxia-related resistance. The conflicting results show the need for further evaluation of the relationship between DCE-derived perfusion characteristics and treatment response. Nonetheless, there is potential utility for DCE MRI as a predictor of the response to radiotherapy and a biomarker for the selection of more aggressive and resistant subvolumes for dose intensification. Adaptive Therapy Adaptive radiation therapy allows for alterations to be made to the treatment protocol based on functional imaging-

based evaluation early in the course of radiotherapy. Initial physiological changes aid prediction of the treatment response by providing information about radiation sensitivity, which can be rather heterogeneous within the tumour volume. Farjam et al. [28] delineated subvolumes of high, intermediate and low regional cerebral blood volume and Ktrans in patients with brain metastasis receiving whole-brain radiotherapy and observed a positive relationship between the extent of reduction in high regional cerebral blood volume/high Ktrans subvolumes and treatment response, measured by the percentage change in GTV. The findings exemplify the diversity of radiation sensitivity and the potential benefit of DCE MRI in helping to identify radiationinsensitive subvolumes as targets for radiation boost. In addition, DCE MRI can monitor radiation-induced functional changes in neighbouring healthy tissues and guide further treatment by modifying the total dose according to one’s radiation tolerance or redistributing dose to spare tissues with relatively good function. Cao et al. [29] investigated the use of DCE MRI as an imaging biomarker for liver function by comparing DCE MRI-derived portal venous perfusion with the indocyanine green clearance rate and showed good correlation. The same study also observed compensatory hyper-perfusion of subvolumes receiving a lower radiation dose. Functional reserves may be improved by selectively lowering doses of liver segments with better function. DCE MRI benefits adaptive strategies by providing critical information about the extent and spatial distributions of radiation-induced damage. As such, treatment can be tailored accordingly to improve the therapeutic index of radiotherapy.

Challenges and Future Work Even though there is a wide range of potential applications, integration of DCE imaging into routine clinical practice is challenging. The correlation of DCE parameters with clinical characteristics and outcomes is not always demonstrable. Some studies showed dose-dependent reductions in DCE parameters post-treatment but failed to establish a relationship with response rates and survival outcomes [35]. This could be due to a failure to account for tumour heterogeneity. Tumours are highly heterogeneous in both their mutational genotypes and phenotypes. Different regions are composed of varying degree of cellularity, angiogenesis, hypoxia and necrosis. Large spatial variations within these tumours are often associated with poorer prognosis, treatment failure and drug resistance [36]. However, most methods of studying DCE images take a single-averaged enhancementetime curve, which could mask critical features of the underlying tumour. Heterogeneity analysis of DCE parametric maps has been explored as a prediction biomarker for the therapy response for soft tissue sarcoma [37] and breast cancer [38]. More investigations need to be carried out with regards to validating the utility of the heterogeneity analysis of functional characteristics.



In addition, errors may arise from incorrect use of pharmacokinetic models. Compartmental tracer models fit poorly for regions with necrotic tissues, which typically display slow and linear enhancement patterns, yielding non-physiological parameter values that will inevitably skew the results of the analysis. The current practice of excluding necrotic voxels from analysis, however, discards important information about the tumour characteristics. Solute transport within these regions is diffusion-driven rather than convection-driven, as such adheres better to a diffusivity model proposed by Koh et al. [39]. The authors investigated human tumour xenografts implanted in mice and found that necrotic regions displayed tracer uptake patterns consistent with Fick’s diffusion equation. Fitting of this model to the concentrationetime curves allowed the estimation of diffusivity, which may serve as a surrogate for drug delivery in these poorly perfused necrotic regions. The onset and duration of therapeutic effects play an important role in determining the optimal timing for scheduling post-treatment scans. The effects of vascular disrupting drugs onset within hours and last a day, whereas those of anti-angiogenic drugs onset after days to weeks and their effects are more prolonged [40]. Currently, there is no consensus with regards to the optimal timing for monitoring and predicting the therapeutic response. In studies of patients with colorectal liver metastasis treated with anti-angiogenic therapy, the timings of post-treatment measurements differed, from 28 daily [41] to 6 weekly [21]. As such, more investigations need to be directed at choosing optimum kinetic models and scan schedules. The main technical constraints of DCE concerns image quality. Involuntary patient motion (e.g. respiratory, cardiac) can cause artefacts and distortions. Furthermore, the trade-off between spatial and temporal resolution is significant. Spatial resolution is important for the accurate detection and localisation of lesions. However, it is limited by the need for high temporal resolution in order to adequately capture the dynamic changes within the tissue. For example, in DCE imaging of pulmonary lesions, higher temporal resolutions are desired, as pulmonary capillary circulation is 0.7 s [42]. Therefore, image acquisition techniques ought to be optimised to ensure meaningful pharmacokinetic analysis. Also, failure to account for the non-linear relationship between signal enhancement and the concentration of contrast may lead to the over- or under-estimation of DCE parameters. Most methods of evaluating tumour vasculatures takes the mean signal enhancement across the tumour region of interest, but that does not equal the mean concentration of contrast within the region due to this nonlinear relationship. Another obstacle to the development of DCE biomarkers lies in measurement reproducibility, which includes inter-/ intra-observer, inter-visit and multicentre reproducibility. Robustness to intra-patient variability is very important for them to serve as reliable biomarkers, as it determines the magnitude of differences in tumour perfusion that are considered clinically relevant. Also, a lack of standardisation of image acquisition and analysis among institutions often

11

impedes the corroboration of findings and collaborations among investigators. Analyses of DCE images require intensive postprocessing efforts. Image registration, which is needed to align images and correct for respiratory motion, together with complex model fittings, involves high computational power. Furthermore, manual drawing of the region of interest is time-consuming. These may limit the practicality of DCE in the clinical setting. Lastly, the DCE MRI technique enables quantitative analysis of tumour microvasculature, but this is just one aspect of its biological characteristics. For a more comprehensive functional evaluation of tumours, a multiparametric approach should be undertaken. Other functional imaging modalities include diffusion-weighted MRI, blood oxygen level dependent MRI, magnetic resonance spectroscopy and positron emission tomography, which measure tissue architecture and cellular density, tissue oxygenation, proliferation and metabolism [43].

Conclusion DCE imaging is a useful tool for evaluation of the functional characteristics of tumours and has shown much promise in cancer diagnosis, characterisation, response monitoring and prediction of clinical outcomes. Nonetheless, more work needs to be done with respect to refining protocols for derivation of clinically relevant parameters, overcoming technical limitations of image acquisition and standardisation of methodologies, for it to be implemented into the clinical workflow as a complement to conventional imaging strategies.

References [1] Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646e674. [2] Bergers G, Benjamin LE. Tumorigenesis and the angiogenic switch. Nat Rev Cancer 2003;3:401e410. [3] Ostergaard L, Tietze A, Nielsen T, et al. The relationship between tumor blood flow, angiogenesis, tumor hypoxia, and aerobic glycolysis. Cancer Res 2013;73:5618e5624. [4] Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;407:249e257. [5] Dvorak M, Kohn S, Morgan ES, Fox P, Nagy J, Dvorak HF. The vesiculo-vacuolar organelle (VVO): a distinct endothelial cell structure that provides a transcellular pathway for macromolecular extravasation. J Leukoc Biol 1996;59:100e115. [6] Koh TS, Bisdas S, Koh DM, Thng CH. Fundamentals of tracer kinetics for dynamic contrast-enhanced MRI. J Magn Reson Imaging 2011;34:1262e1276. [7] Daniel BL, Yen YF, Glover GH, et al. Breast disease: dynamic spiral MR imaging. Radiology 1998;209:499e509. [8] Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999;211:101e110. [9] Paldino MJ, Barboriak DP. Fundamentals of quantitative dynamic contrast-enhanced MR imaging. Magn Reson Imaging Clin N Am 2009;17:277e289.


12


[10] Jansen SA, Fan X, Karczmar GS, Abe H, Schmidt RA, Newstead GM. Differentiation between benign and malignant breast lesions detected by bilateral dynamic contrastenhanced MRI: a sensitivity and specificity study. Magn Reson Med 2011;59:747e754. [11] Thomassin-Naggara I, Balvay D, Aubert E, et al. Quantitative dynamic contrast-enhanced MR imaging analysis of complex adnexal masses: a preliminary study. Eur Radiol 2012;22:738e745. € [12] Ozduman K, Yıldız E, Dinçer A, Sav A, Pamir MN. Using intraoperative dynamic contrast-enhanced T1-weighted MRI to identify residual tumor in glioblastoma surgery. J Neurosurg 2014;120:60e66. [13] Larsen V, Simonsen HJ, Law I, Larsson HBW, Hansen AE. Evaluation of dynamic contrast-enhanced T1-weighted perfusion MRI in the differentiation of tumor recurrence from radiation necrosis. Neuroradiology 2013;55:361e369. [14] Vos EK, Litjens GJS, Kobus T, et al. Assessment of prostate cancer aggressiveness using dynamic contrast-enhanced magnetic resonance imaging at 3 T. Eur Urol 2013;64:448e455. [15] Sun MRM, Long N, Genega EM, et al. Renal cell carcinoma: dynamic contrast-enhanced MR imaging for correlation with pathologic findings. Radiology 2009;250:793e802. [16] Lollert A, Junginger T, Schimanski CC, et al. Rectal cancer: dynamic contrast-enhanced MRI correlates with lymph node status and epidermal growth factor receptor expression. J Magn Reson Imaging 2014;39:1436e1442. € ner R, et al. Increased micro[17] Haldorsen IS, Stefansson I, Gru vascular proliferation is negatively correlated to tumour blood flow and is associated with unfavourable outcome in endometrial carcinomas. Br J Cancer 2014;110:107e114. [18] Abramson RG, Li X, Hoyt TL, et al. Early assessment of breast cancer response to neoadjuvant chemotherapy by semiquantitative analysis of high-temporal resolution DCE-MRI: preliminary results. Magn Reson Imaging 2013;31:1457e1464. [19] Glade Bender JL, Lee A, Reid JM, et al. Phase I pharmacokinetic and pharmacodynamic study of pazopanib in children with soft tissue sarcoma and other refractory solid tumors: a Children’s Oncology Group Phase I consortium report. J Clin Oncol 2013;31:3034e3043. [20] Meyer JM, Perlewitz KS, Hayden JB, et al. Phase I trial of preoperative chemoradiation plus sorafenib for high risk extremity soft tissue sarcomas with dynamic contrast-enhanced MRI correlates. Clin Cancer Res 2013;19:6902e6911. [21] Coenegrachts K, Bols A, Haspeslagh M, Rigauts H. Prediction and monitoring of treatment effect using T1-weighted dynamic contrast-enhanced magnetic resonance imaging in colorectal liver metastases: potential of whole tumour ROI and selective ROI analysis. Eur J Radiol 2012;81:3870e3876. [22] Zahra M, Tan LT, Priest AN, et al. Semiquantitative and quantitative dynamic contrast-enhanced magnetic resonance imaging measurements predict radiation response in cervix cancer. Int J Radiat Oncol Biol Phys 2009;74:766e773. [23] Andersen EKF, Hole KH, Lund KV, et al. Pharmacokinetic parameters derived from dynamic contrast enhanced MRI of cervical cancers predict chemoradiotherapy outcome. Radiother Oncol 2013;107:117e122. [24] Li SP, Makris A, Beresford MJ, et al. Use of dynamic contrastenhanced MR imaging to predict survival in patients with primary breast cancer undergoing neoadjuvant chemotherapy. Radiology 2011;260:68e78. [25] Groenendaal G, Moman MR, Korporaal JG, et al. Validation of functional imaging with pathology for tumor delineation in the prostate. Radiother Oncol 2010;94:145e150.

[26] Dirix P, Vandecaveye V, De Keyzer F, Stroobants S, Hermans R, Nuyts S. Dose painting in radiotherapy for head and neck squamous cell carcinoma: value of repeated functional imaging with (18)F-FDG PET, (18)F-fluoromisonidazole PET, diffusion-weighted MRI, and dynamic contrast-enhanced MRI. J Nucl Med 2009;50:1020e1027. [27] Wang P, Popovtzer A, Eisbruch A, Cao Y. An approach to identify, from DCE MRI, significant subvolumes of tumors related to outcomes in advanced head-and-neck cancer. Med Phys 2012;39:5277e5285. [28] Farjam R, Tsien CI, Feng FY, et al. Physiological imagingdefined, response-driven subvolumes of a tumor. Int J Radiat Oncol Biol Phys 2013;85:1383e1390. [29] Cao Y, Wang H, Johnson TD, et al. Prediction of liver function by using magnetic resonance-based portal venous perfusion imaging. Int J Radiat Oncol Biol Phys 2013;85:258e263. [30] Spencer JA, Ghattamaneni S. MR imaging of the sonographically indeterminate adnexal mass. Radiology 2010;256:677e694. [31] Wu LM, Xu J-R, Gu HY, et al. Role of magnetic resonance imaging in the detection of local prostate cancer recurrence after external beam radiotherapy and radical prostatectomy. Clin Oncol (R Coll Radiol) 2013;25:252e264. [32] Carter HB. Active surveillance for prostate cancer: an underutilized opportunity for reducing harm. J Natl Cancer Inst Monogr 2012;2012:175e183. [33] Turkbey B, Mena E, Aras O, Garvey B, Grant K, Choyke PL. Functional and molecular imaging: applications for diagnosis and staging of localised prostate cancer. Clin Oncol (R Coll Radiol) 2013;25:451e460. [34] Mackillop WJ. The importance of prognosis in cancer medicine. Hoboken: John Wiley & Sons; 2006. [35] O’Connor JPB, Jackson A, Parker GJM, Jayson GC. DCE-MRI biomarkers in the clinical evaluation of antiangiogenic and vascular disrupting agents. Br J Cancer 2007;96:189e195. [36] Davnall F, Yip CSP, Ljungqvist G, et al. Assessment of tumor heterogeneity: an emerging imaging tool for clinical practice? Insights Imaging 2012;3:573e589. [37] Alic L, van Vliet M, van Dijke CF, Eggermont AMM, Veenland JF, Niessen WJ. Heterogeneity in DCE-MRI parametric maps: a biomarker for treatment response? Phys Med Biol 2011;56:1601e1616. [38] Golden DI, Lipson J, Telli ML, Ford JM, Rubin DL. Dynamic contrast-enhanced MRI-based biomarkers of therapeutic response in triple-negative breast cancer. J Am Med Inform Assoc 2013;20:1059e1066. [39] Koh TS, Hartono S, Thng CH, Lim TKH, Martarello L, Ng QS. In vivo measurement of gadolinium diffusivity by dynamic contrast-enhanced MRI: a preclinical study of human xenografts. Magn Reson Med 2013;69:269e276. [40] Lind JSW, Meijerink MR, Dingemans A-MC, et al. Dynamic contrast-enhanced CT in patients treated with sorafenib and erlotinib for non-small cell lung cancer: a new method of monitoring treatment? Eur Radiol 2010;20:2890e2898. [41] Morgan B, Thomas AL, Drevs J, et al. Dynamic contrastenhanced magnetic resonance imaging as a biomarker for the pharmacological response of PTK787/ZK 222584, an inhibitor of the vascular endothelial growth factor receptor tyrosine kinases, in patients with advanced colorectal cancer. J Clin Oncol 2003;21:3955e3964. [42] Ohno Y, Hatabu H, Takenaka D, et al. Dynamic MR imaging: value of differentiating subtypes of peripheral small adenocarcinoma of the lung. Eur J Radiol 2004;52:144e150. [43] Padhani AR, Miles K. Multiparametric imaging of tumor response to therapy. Radiology 2010;256:348e364.


New applications of magnetic resonance imaging for thoracic oncology.

Magnetic resonance imaging in oncology.

Techniques and applications of dynamic contrast enhanced magnetic resonance imaging in cancer.

Introduction: Magnetic resonance imaging comes of age in radiation oncology.

Functional magnetic resonance imaging in oncology: state of the art.

Magnetic Resonance Imaging and Spectroscopy: Application to Experimental Neuro-Oncology.

Clinical applications for diffusion magnetic resonance imaging in radiotherapy.

Dynamic contrast enhanced magnetic resonance imaging in chronic Achilles tendinosis.

Contrast agents in dynamic contrast-enhanced magnetic resonance imaging.

Functional magnetic resonance imaging (FMRI) neurofeedback: implementations and applications.

Magnetic resonance imaging for pulmonary hypertension: methods, applications, and outcomes.

Single-shot magnetic resonance imaging: applications to angiography.

Clinical applications of pediatric positron emission tomography-magnetic resonance imaging.

Unsupervised tissue segmentation from dynamic contrast-enhanced magnetic resonance imaging.

Current clinical applications of cardiovascular magnetic resonance imaging.

Resting-state functional magnetic resonance imaging: review of neurosurgical applications.

Cardiovascular applications of magnetic resonance imaging and phosphorus-31 spectroscopy.

Dynamic magnetic resonance imaging and postoperative motion of diaphragm.

Undersampled dynamic magnetic resonance imaging using kernel principal component analysis.

[Magnetic resonance imaging].

Pelvic magnetic resonance imaging.

Gynecologic magnetic resonance imaging.

Intracranial magnetic resonance imaging.

Shoulder magnetic resonance imaging.