Min-Ying Su, Jeon-Hor Chen.动态增强磁共振成像在乳腺癌诊断及疗效监测中的应用进展.磁共振成像, 2012, 3(2): 84-97. DOI:10.3969/j.issn.1674-8034.2012.02.002.
[摘要] 经过20余年的相关研究,MRI现已成为临床公认的用于检出和诊断乳腺疾病的影像检查手段之一。笔者简要阐述了乳腺动态增强(DCE)磁共振成像技术及其临床应用。详细介绍了动态增强中有关药代动力学的测量方法、各种相关参数的定性分析、以及Ktrans和kep等反应药代动力学特征参数的定量分析。笔者不但就当前"是否有必要对每位患者进行动脉输入函数进行测量"等有争议的话题进行了讨论,还在对当前广泛应用的各种动态增强MRI后处理软件进行了较详尽的描述之后,进一步对动态增强MRI在乳腺癌的诊断及疗效监测等方面的临床应用情况进行了详尽的回顾。作者认为:总体来讲,对于乳腺癌的诊断,关键在于关注病灶中最具侵袭性部分的MRI特征。而对于乳腺癌疗效的监测,则应该对整个病灶进行分析。最后,笔者就乳腺MRI对乳腺癌高危人群的筛查以及风险的处理等问题进行了讨论。笔者预测乳腺MRI在女性乳腺疾病尤其是乳腺癌的筛查、诊断、治疗及疗效监测等一系列问题上将继续发挥不可替代的重要作用。 |
[Abstract] With research being conducted over 2 decades, breast MRI has become an established clinical imaging modality for management of breast diseases. This review paper summarizes the analysis of breast MRI acquired using the dynamic-contrast enhanced (DCE) imaging protocol as well as its clinical applications. The measurement of DCE kinetics, the qualitative analysis to measure heuristic parameters, and the quantitative analysis to obtain pharmacokinetic parameters such as Ktrans and kep are described. The current debate about whether it is necessary to measure the arterial input function from each individual patient is discussed. The DCE-MRI analysis tools offered in several widely used commercial software are described. Then the clinical application of DCE-MRI for diagnosis and therapy monitoring of breast cancer are described. In general, for diagnosis of breast cancer, the hot spot approach to characterize tissues with the most aggressive pathology should be taken; but for therapy response monitoring, the whole tumor should be analyzed. Lastly, the application of breast MRI in high-risk screening and vey recently, risk management, is discussed. It is highly anticipated that breast MRI will continue to play a very important role in management and care of breast diseases for women in the entire clinical spectrum from risk management, screening, diagnosis, therapy, and surveillance. |
[关键词] 动态增强磁共振成像;乳腺癌;诊断;化疗;监测;进展 |
[Keywords] Dynamic contrast enhanced MRI;Breast cancer;Diagnosis;Neoadjuvant chemotherapy;Monitoring;Progress |
Tu & Yuen Center for Functional Onco-Imaging, Department of Radiological Sciences, University of California, Irvine, CA 92697, USA
Tu & Yuen Center for Functional Onco-Imaging, Department of Radiological Sciences, University of California, Irvine, CA 92697, USA;Department of Radiology, China Medical University Hospital, Taichung 40402, Taiwan, China;Department of Medicine, School of Medicine, China Medical University, Taichung 40402, Taiwan, China
通讯作者:Min-Ying Su, Ph.D., E-mail: msu@uci.edu
基金项目: 美国国家卫生研究院/美国癌症研究所基金 项目编号:R01 CA127927和R03 CA136071 |
收稿日期:2012-01-28 |
接受日期:2012-03-01 |
中图分类号:R445.2; R737.9 |
文献标识码:A |
DOI: 10.3969/j.issn.1674-8034.2012.02.002 |
Min-Ying Su, Jeon-Hor Chen.动态增强磁共振成像在乳腺癌诊断及疗效监测中的应用进展.磁共振成像, 2012, 3(2): 84-97. DOI:10.3969/j.issn.1674-8034.2012.02.002. |
1 Introduction
Dynamic contrast enhanced MRI (DCE-MRI) is a rapidly evolving imaging technique. It is the current standard for breast MR imaging; also many research studies have been conducted to evaluate its application for various clinical problems in different organs. The development of gadolinium-based contrast agents in early 1980 opened a new era for imaging of tumors and vascular systems. Tumors require a higher supply of nutrients to support the rapid growth; as such it is necessary to induce formation of new blood vessels, termed "angiogenesis". One of the earliest applications of Gd-based contrast agents is for enhancement of viable tumors on post-contrast T1-weighted images. The typical scan protocol includes pre-contrast images and post-contrast images, and the enhanced tumors can be detected by visual inspection of post-contrast images referencing to pre-contrast images. If the contrast injection is done while keeping the patient still inside the scanner, the subtraction images can be generated to reveal the contrast-enhanced lesions more clearly. However, while this technique was proven very helpful for imaging of lesions in the brain and musculoskeletal system, the application in the breast encountered a great difficulty. The tumors (malignant and benign), as well as normal breast tissues, all showed contrast enhancements, and they were difficult to be differentiated. Further research has found that if multiple sets of post-contrast images at different times after injection were acquired, the enhancement time course (or, enhancement kinetics) could be measured, and that provided additional information to aid in distinguishing malignant lesions from benign lesions and normal breast tissues [1, 2]. This technique was termed "dynamic contrast enhanced MRI" or "DCE-MRI" .
This review paper is focused on the analysis of DCE-MRI data. In Section II, the measurement of enhancement kinetics by DCE-MRI, and the qualitative and quantitative analysis to obtain parameters such as the exchange rate constants Ktrans and kep will be described. The current debate about whether it is necessary to measure the arterial input function from each individual patient will be discussed. Then in Section III, the DCE-MRI analysis tools offered in several widely used commercial software will be described. The clinical application of DCE-MRI for diagnosis and therapy monitoring of breast cancer will be described in Section IV and Section V, respectively. In general, for diagnosis of breast cancer, the hot spot approach to characterize tissues with the most aggressive pathology should be taken [3]; but for therapy response monitoring, the whole tumor should be analyzed [4]. Lastly, in Section VI the application of breast MRI in screening and risk management is described.
2 Measurement and Quantitative Analysis of DCE Kinetics
2.1 Measurement of enhancement kinetics
The T1-weighted 3-dimensional gradient echo sequence is the most commonly used sequence for acquiring DCE-MRI. It covers the entire breast without gaps, generates good quality images, and can be acquired within a short imaging time. All major scanner manufactures offered DCE-MRI sequences that can be easily prescribed with preferred spatial and temporal resolution, and the total number of pre-and post-contrast imaging frames. But, there is no standard about how to perform the contrast injection. As there may be some variations from site to site, it is typical to describe the injection protocol in publications. The DCE kinetics can be analyzed using a straightforward approach based on the increased signal intensity, or using a more sophisticated approach to convert the measure signal enhancement to the concentration of the contrast agents to allow for pharmacokinetic modeling analysis to obtain physiological parameters. The most commonly used approach measures the percent enhancement at time t as: [S(t)-S0] / S0 x 100%. The signal enhancement is calculated by subtracting the pre-contrast signal intensity S0 from the post-contrast signal intensity S(t), and then normalized to the pre-contrast signal intensity S0 to calculate the percent enhancement. Normalization to S0 is necessary to handle the problem of varying coil sensitivity, thus allows for comparison of tissue enhancements across the entire imaging field of view. This approach is easy and does not require multiple calibration scans, and is commonly used in clinical examinations.
On the other hand, if the purpose is to obtain physiological parameters from the transport kinetics of MR contrast agents as tracer, the concentration of the contrast agents needs to be measured to allow such precise analysis. All Gadolinium-based contrast agents are extracellular agents, and can be distributed in the vascular and interstitial spaces. Under assumption of the fast exchange regime, the concentration is proportional to the increased T1 relaxivity (R1=1/T1), thus the T1 relaxation time before injection (T10) and at post-injection time points have to be measured. The gradient echo sequences using different flip angles are commonly used. The T10 can be estimated using 3 flip angles (5°, 10°, and 15°). Given the needed temporal resolution during the DCE acquisition, most studies only use one flip angle after injection (e.g. 15°), and the T1(t) is estimated by referencing to the proton-density images acquired before injection using 5° flip angle. The next step is to convert the increased R1 to the concentration of the contrast agents based on the linear relationship (R1 [C]-R10 = constant x [C]). The proportional constant is different in different tissues and cannot be accurately measured, but the common approach is to use the constant measured in water or saline at 1.5 Tesla and 3.0 Tesla to give a reasonable estimate. After the [Gd] concentration time course is obtained, it can be analyzed using the pharmacokinetic model to obtain parameters that are associated with vascular properties (vascular perfusion and permeability). All clinically approved Gd-based contrast agents are low molecular weight agents, and they do not yield precise measurements that are respectively associated with perfusion and permeability; rather the combined effects are seen.
2.2 Analysis of enhancement kinetics
The enhancement kinetics can be evaluated using 3 distinct features, the wash-in phase, the maximum enhancement, and the wash-out phase. Several heuristic parameters can be analyzed from the curve, such as wash-in slope (maximum slope, or the slope within a time period), the % maximum enhancement, time to maximum, and the wash-out slope (within a time period). Since these parameters may be affected by the noise level at different data points, a more robust and commonly used parameter is the IAUC (initial area under the curve), which integrates the area under the kinetic curve, usually during the early time period such as the first 90 seconds. This parameter reflects how fast and how much the contrast material is delivered into the lesion.
A more sophisticated analysis method is to perform pharmacokinetic analysis based on two compartmental models, commonly referred as the unified Tofts model [5, 6], as shown in Figure 1. The two compartments are the vascular space and the interstitial space, with the transfer constant Ktrans to leak from the vascular to the interstitial space, and the rate constant kep from the interstitial space back to the vascular space. Another parameter considered in the model is the distribution volume in the extravascular-extracellular space ve (within the interstitial space). The change of concentration in the interstitial space (Ce) is expressed as dCe/dt = Ktrans (Cb) - kep (Ce). The total concentration in the tissue can be written as the contribution from both vascular and interstitial compartments as Ct= vbCb + ve Ce. Many parameters can be included in the fitting model, but the problem of over-fitting needs to be considered. Although the vascular compartment is present, yet whether the fitted parameter, such as vb, is truly reflecting the vascular space is of a great concern. Since all clinical Gd-based contrast agents are small agents that can quickly diffuse from the vascular space to the interstitial space, the fitted vb is very likely to contain the early leakage space in the interstitial space, thus not an accurate measurement of the true vascular volume. Due to this concern, the most commonly used model (Tofts model) assumes a relatively small vascular space compared to the interstitial space and ignores the term vb. Thus, the tissue concentration is expressed as Ct = ve Ce, where Ce is dependent on the blood concentration Cb. However, when the analysis is performed for highly vascularized lesion, the term vb may not be ignored.
The blood kinetics Cb is required to fit the measured concentration in the tissue. When the absolute concentration of the contrast agents (such as mmole/liter) in both the tissue and the blood are measured and used in the fitting, the unit for Ktrans and kep is 1/time (commonly used as [1/min]). If diffusion is the only process involved for transport of contrast agents between vascular and interstitial compartment, the exchange rate between Cb and Ce should be equal, and ve can be obtained as Ktrans/kep (0<ve<1). If the absolute concentration is not obtained for either the tissue or the blood, the fitted parameters Ktrans will carry an arbitrary unit (depending on which parameter is used in fitting, such as the percent enhancement), but the unit for kep is always [1/min]. When the fitted pharmacokinetic parameters are analyzed for therapy monitoring studies, a percent change in the follow-up study compared to the baseline study is usually calculated. As long as the acquisition and the analysis are performed consistently, it is not necessary to convert the signal enhancement and the blood kinetics to the [Gd] concentration.
2.3 The individually-measured versus the general-population arterial input function (AIF)
Since the blood concentration is required for fitting, one area of debate is whether the arterial input function should be measured from each individual patient. The advantage is to provide the most accurate blood kinetics taking into account the different hemodynamics of individual patients; but the disadvantage is the difficulty in measuring the AIF accurately. If a wrong measurement is used, it may lead to a large error in the fitted parameters. The preferred artery for measuring the AIF from is the one directly feeding the lesion. As such, a high spatial resolution is needed to delineate the vessel and to avoid partial volume effect (the voxel has to be completely contained within the vessel); also a very high temporal resolution is needed to catch the maximum enhancement. Further, the image quality has to be good to obtain smooth enhancement kinetics without too much noise or fluctuation. The motion artifact, including respiratory motion and pulsation of vessels, may lead to degradation of the vessel image. All these problems lead to difficulty in measuring AIF accurately from the small feeding artery. The alternative approach is to measure the AIF from the large artery such as the aorta; but the high velocity of blood flow needs to be properly handled, also the AIF measured from the aorta may not truly reflect the AIF of the direct feeding artery to the lesion.
Due to these difficulties, the blood kinetics measured from the healthy general population provides a reasonable and acceptable approach. As long as the patient does not have cardiovascular problems or kidney diseases, the general population blood kinetics can be used in the pharmacokinetic model fitting to obtain Ktrans and kep. Since the results are obtained using the unified model, the fitted parameters may be compared between different studies. However, while this is acceptable for studies dealing with lesion diagnosis or characterization, it may not be appropriate for therapy monitoring studies, particularly for those designed to measure the vascular changes in clinical trials of anti-angiogenic or anti-vascular therapeutic agents. It is recommended that for such studies the individual AIF needs be measured and used as reference [7]. However, given the requirement of both high spatial and temporal resolution, the reliability of the measured AIF needs to be investigated when designing the imaging protocol[8]. The reproducibility of the AIF measurement needs to be performed first to allow subsequent data analysis and interpretation, at least to estimate the possible errors introduced by the variation in the measured AIF [9].
2.4 Image registration for pre-and post-contrast DCE frames
One major problem leading to poor quality of the DCE kinetics is the motion during the DCE acquisition, including both patient movement and physiological motion such as breathing and heart beat pulsation. A rigid-coregistration process is usually applied to spatially align all DCE frames with respect to the selected reference frame. It should be noted that the co-registration is applied to the organ of interest (that is, the breast), not the entire image. For breast imaging, the co-registration should not be done to align the thoracic body region while compromising the co-registration of the breast region. Regional co-registration within a small field of view is the commonly used approach.
2.5 Lesion ROI-based analysis and pixel-by-pixel analysis
The choice of the Region of interest (ROI)-based analysis or pixel-by-pixel analysis is usually dependent on the clinical application. In general, for diagnostic purpose the hot spot ROI approach should be used, and for therapy monitoring study the pixel-by-pixel analysis should be applied. The advantages for ROI-based analysis include that it is less susceptible to noise and signal fluctuation through averaging over many pixels, and the fitting to obtain Ktrans and kep is unlikely to fail so that the obtained results can be directly used in the analysis. The advantage for the pixel-by-pixel analysis is the rich data obtained from the entire lesion that allows for histogram analysis within the lesion. The disadvantage includes that the kinetics measured from some pixels may be very noisy, and the fitting quality needs to be carefully inspected. Usually the pixels with unsatisfactory fitting quality need to be discarded in the analysis.
3 Computer-Aided Analysis Tools
There are several commercially available computer-aided analysis software dedicated for analysis DCE-MRI, such as Merge CADstream, DynaCAD, and iCAD. All these have specially designed product for the breast, as well as for other organs. The provided features in these different systems are similar. The software detects the DCE sequence and imports all images for analysis. The typical display includes rendering maximum intensity projection (MIP) and the corresponding view from 3 planes (axial, sagittal, coronal)-one acquired and two reformatted. The color-coding to label the suspicious enhancements can be turned on based on the choice of the threshold enhancement to show color-coding. Two post-contrast frames are selected for determining the wash-out pattern. Typically the red color is used to label the voxel showing wash-out, indicating a high suspicion of malignancy. The DCE kinetics from the hot spot or the selected ROI is displayed on the screen, and the graphical interface is provided to the operator to change the display, for example, to change the enhancement threshold or to select different frames for analysis of the wash-out slope. Within the selected tumor ROI, the percentage of voxels that show wash-out, plateau and persistent enhancing pattern can be calculated and displayed. The program also provides the lesion segmentation tool based on connectivity of enhanced tissues. The cursor can be placed over the enhanced lesion, and the 3D view of the segmented lesion can be displayed. The 3 dimensional size, as well as the total volume of the lesion, is given. Finally a report summarizing all key findings will be generated for each case. Based on all presented information, the radiologist can give a final diagnostic impression or the BI-RADS score. While these systems can be purchased as standing-alone off-line systems for any scanners, they can also be integrated to the scanner for on-line analysis, mainly for planning MR-guided biopsy.
The software for performing quantitative pharmacokinetic modeling based on DCE-MRI kinetics to extract fitting parameters (Ktrans, kep, vp, and ve) is also commercially available, for example, the "Tissue 4D". The signal enhancement is converted to the concentration using an estimated proportional constant provided by the software. Three blood kinetic curves (fast, medium, and slow flow) are built-in and can be selected based on the organ of interest. The software also supports the measurement of individual arterial input function from the analyzed case and used in the fitting. Two fitting models are provided, the Tofts model (ignoring the vascular space) and Tofts+vp (considering the vascular space). The obtained fitting parameters (Ktrans and kep, or in addition ve and vp) from the pixel-by-pixel analysis within the selected ROI can be displayed as overlaying color-maps and histograms, also the data can be exported for further analysis. It is anticipated that the availability of such a standard analysis tool will facilitate comparison of quantitative DCE studies, which could not be done when using different in-house programs developed by different research groups.
4 DCE-MRI for Diagnosis of Breast Cancer
Tumors, particularly the more aggressive malignant tumors, need angiogenesis to support the rapid tumor growth. The angiogenic vessels are leakier (that is, with a wider endothelial junction), and that allows contrast agents to quickly leak from vascular space into the interstitial space and back diffuse to the vascular space to be cleared [10,11]. DCE-MRI can be used to measure the transport kinetics of contrast agents in the tissue, allowing for analysis of transfer rates associated with vascular perfusion and permeability. Despite this advantage, the motivation for performing DCE-MRI for diagnosis of breast cancer is mainly for problem solving, in order to detect the lesion without being obscured by the strong background tissue enhancements, as well as for differentiating between malignant and benign lesions.
In the brain and the musculoskeletal systems, the normal tissue enhancements are usually unremarkable, and as such the lesions are much strongly enhanced and can be easily detected on the post-contrast enhanced images taken at several minutes following injection. The leakage of contrast agents from the capillary to the interstitial space is low in the normal brain due to blood-brain-barrier, and also low in the muscle due to the continuous type capillary (with tight endothelial junctions) and the small interstitial space; and that results in relatively low background tissue enhancements. In contrast, the fibroglandular tissues in the breast have a relatively high vascular supply and interstitial space, which may allow leakage of contrast agents to show a strong background tissue enhancement which increases with time after the contrast injection [12]. Therefore, the lesions can be better differentiated at early times before the background tissue shows strong enhancements [13]. The earlier DCE-MRI studies were performed using a relatively high temporal resolution [14], and different post-contrast frames acquired at different times can be used to find the optimal frame that show the best signal contrast between lesions and normal tissues. Usually the maximum enhancement is seen around 2 minutes after injection of contrast agents.
Many studies have investigated the diagnostic capability of DCE-MRI to differentiate between benign and malignant breast lesions [15,16,17,18,19,20,21]. In general, malignant lesions are more aggressive, and require a higher angiogenic activity. For example, they have a higher expression of Vascular Endothelial Growth Factor (VEGF) that stimulates formation of new vessels, and these immature vessels are leakier. When the contrast agents are injected, the higher vascular space and higher vascular permeability allow more contrast agents to be quickly delivered into the interstitial space of the lesion, also the agents can quickly diffuse back to the vascular space to be cleared. As such, the enhancement kinetics shows a rapid wash-in, reaches to the maximum enhancement quickly, and then starts to show wash-out. In contrast, benign lesions may not have a large number of angiogenic vessels, but still have a high interstitial space to uptake contrast agents, and the enhancement kinetics shows a slow but persistent enhancing pattern[22]. Although these two patterns have been proven as reliable diagnostic features, many lesions show the enhancement kinetics in between, reaching a plateau during the imaging period without showing a clear wash-out or a clear persistent enhancement pattern [22]. Given the heterogeneous nature of breast lesions, the enhancement kinetics also varies with tissue location, or the placement of the region of interest (ROI) within the lesion [3,23]. For diagnostic purpose, the rule of the most aggressive pathology is applied; therefore, the hot spot approach should be used [3]. The ROI should be placed on the tissue regions with the strongest enhancement, and when the wash-out pattern is seen in some part of the lesion, this lesion is considered suspicious of malignancy. In contrast, the benign lesions are often more homogeneous, and if the enhancement kinetics measured from the entire lesion shows the persistent enhancing pattern, this lesion is more likely benign [22].
The research results obtained over a decade provide a strong evidence suggesting that a higher spatial resolution to reveal the morphology of the lesion in a greater detail has a better diagnostic value [24,25,26]. The current protocol for diagnosis of breast lesions emphasizes the spatial resolution over the temporal resolution. As for the number of post-contrast frames and the total DCE time period, it is not a concern as long as the kinetic pattern can be evaluated. The general rule is that one set of high quality post-contrast images should be acquired within 3 minutes after injection of contrast agents for lesion detection and morphological characterization, and the kinetics over a period of 5-10 minutes should be measured for determination of the DCE pattern. Both lesion morphology and kinetic pattern need to be considered for giving a final diagnostic impression[27,28,29].
The commercial CAD system for breast MRI described in Section III provides many useful functions, but they are not essential. All scanner manufactures also provide built-in analysis software that can be used for evaluating the morphological and kinetic features. The breast lesions can be separated into mass and non-mass-like enhancement [26, 28]. For mass lesion the enhancement kinetics may provide very helpful diagnostic information; but for non-mass lesions, the pattern of DCE kinetics cannot be used to differentiate between malignant and benign lesions, and the diagnosis will need to focus on analyzing the morphological distributions of the enhanced tissues (Figure 2) [28]. Developing intellectual CAD systems that can consider all morphological and kinetic information and make a diagnostic impression is an active research area[27,28,29]. Such a system will be very helpful to provide a second reader, and will particularly help inexperienced radiologists in interpreting breast MRI.
5 DCE-MRI for Neoadjuvant Chemotherapy Response Monitoring & Prediction
In addition to diagnosis, another major application of DCE-MRI is for monitoring response of breast cancer undergoing neoadjuvant chemotherapy (or, pre-operative chemotherapy) [30,31,32,33,34,35,36]. Neoadjuvant chemotherapy is conventionally used to down-stage inoperable locally advanced breast cancer. For operable breast cancer, it has shown that the outcome of neoadjuvant chemotherapy followed by surgery is comparable to that of surgery followed by adjuvant chemotherapy, and with this evidence NAC has been increasingly used for operable breast cancer. The advantages of NAC also include that it may facilitate breast conserving surgery; when a patient achieves complete pathologic response (pCR) or near pCR the prognosis is favorable; and that the sensitivity of each individual patient’s cancer to different drug regimens can be tested in vivo. Of all breast imaging modalities, MRI is considered as the most accurate for assessing the treatment response [37,38,39,40]. MRI can also help selection of patients suitable for receiving breast conserving surgery after NAC [41, 42] (Figure 3). The size or the disease extent of lesions before starting of therapy should be established as the baseline reference. For diagnosis of the lesion size after therapy, the patterns of the DCE kinetics should no longer be used as diagnostic criteria. Rather, any enhanced tissues within the tumor bed are considered as the residual disease [30,31,32] .
Changes in lesion size on DCE-MRI are usually not detected until several weeks following NAC [39]. If early surrogate response indicators could be established to predict final treatment outcome, it would help to achieve the goal of pathological complete response (pCR). There are other attempts to investigate whether functional information provided by MRI may serve as earlier response indicators than the size change. They included pharmacokinetic parameters (area under the curve, Ktrans, kep, etc.), the apparent diffusion coefficient (ADC) measured by diffusion weighted imaging (DWI); and the total choline peak measured by proton MR spectroscopy.
It is well known that the cancer therapy also causes vascular damage, and the enhancement kinetic pattern will change from the wash-out pattern to a less aggressive pattern of plateau or persistent enhancement [43], as shown in Figure 4. The role of DCE pharmacokinetic parameters as earlier response indicator has been extensively investigated, but the results were controversial [33,44,45,46,47,48]. Some researchers found that the early change in pharmacokinetic parameters before size shrinkage could predict final treatment outcome [46], but others found that pharmacokinetic parameters were not reliable, and still the size change was the best response indicator [33, 44]. Although some encouraging results have demonstrated that the changes in the exchange rates (Ktrans, kep) or other heuristic parameters (such as maximum % enhancement, area under the curve) showed significant differences between good responders and poor responders, yet in ROC analysis the area under the curve was not sufficiently high for them to serve as reliable response predictors. Furthermore, when compared to the early change in the tumor size, these kinetic parameters may not have a significantly higher predicting power.
The more attractive role of DCE-MRI is to evaluate the response of anti-angiogenic or anti-vascular therapy [31,49,50,51,52,53,54]. The anti-angiogenic agent, Trastuzumab (Avastin), is a monoclonal antibody that neutralizes the vascular endothelial growth factor (VEGF) to inhibit angiogenesis, and it has been used for treating breast cancer in neoadjuvant setting. DCE-MRI provides a means for assessing the treatment-induced vascular changes to investigate the early therapeutic response to this targeted drug. It may provide insightful information to evaluate the efficacy of drugs in clinical trial phases, and to guide the design for future studies. For this purpose, the changes in the entire tumor need to be evaluated, in contrast to the hot-spot approach used for diagnosis. Since breast tumor is highly heterogeneous, there is a poor correlation between the parameters analyzed from the hot-spot and the entire tumor ROI (Figure 5). The most useful analysis method is to perform pixel-by-pixel analysis of the enhancement kinetics from the entire tumor, and the obtained histograms can be compared between studies performed before and after therapy to evaluate changes. Since the parameters measured in different studies will be compared, other confounding factors that may affect the changes need to be considered. One most important factor is the difference in the arterial input function, as these therapies may very likely affect the circulation of the patients. An international consensus panel for application of DCE-MRI in drug trials recommended that the arterial input function be measured on individual basis, and used as reference to obtain quantitative parameters such as Ktrans and kep, or at least to provide a reference for normalization of lesion enhancements for evaluating changes [7, 55]. However, the subsequently published studies revealed great difficulty in obtaining the AIF reliably and consistently [8]. If the measured AIF were problematic, using that to serve as reference would lead to an even higher error in fitted parameters compared to using the AIF of the general population.
In recent several years the change of diffusion within the lesion measured by DWI has been shown as a promising early surrogate biomarker for detecting early response to NAC. The initial results[56, 57] were encouraging, showing earlier changes preceding the tumor shrinkage. DWI can be added to standard DCE-MRI to improve the evaluation accuracy [58,59,60]. ADC measured after four cycles of NAC has been shown as a strong independent predictor of pCR [61]. The baseline ADC may also predict response. Tumors with a low pretreatment ADC tended to respond better to chemotherapy than those with a higher ADC, consistent with a better response seen in more aggressive tumors with a higher cell density[62].
The role of 1H-MRS for therapy response prediction has received a great attention [63,64,65,66,67]. The total Choline (tCho) level measured by MRS is a marker of cellular proliferation, and a decreased tCho soon after starting of treatment may serve as an early predictor of good response (Figure 6). A significant difference in change of tCho was found between good responders and poor responders evaluated based on the size changes at a later time [63]. Also, it has been shown that the tCho changes were greater than the tumor size changes in the pCR group, but not in the non-PCR group [64]. However, it is extremely difficult to obtain reliable measurement of quantitative tCho, and to date the role of MR spectroscopy in management of NAC patients is still not established yet. Other than tCho, several studies also reported an association between the water:fat ratio measured by MR spectroscopy with NAC response [68,69,70,71]. As NAC has become a very important treatment option for breast cancer patients, more research is needed to investigate the role of MRI in patients electing to receive NAC.
6 DCE-MRI for Screening and Risk Management
Mammography is the most commonly used screening modality in the US and Europe; however, its performance is limited by the density of the breast tissue, and it does not work well for young women and Asian women who have dense breasts. MRI is not limited by the density thus is the preferred modality for screening of young women who have a high risk of developing breast cancer. Many large studies conducted around the world have proven that MRI detects many occult cancers that are not detected by mammography or ultrasound in high-risk women. In March 2007, the American Cancer Society in the United States issued a guideline recommending annual MRI screening for women who have greater than 20%-25% lifetime risk of developing breast cancer[15].
In addition to early detection, a better diagnosis and treatment, another very important research area is on risk management, e.g. using chemoprevention to decrease cancer risk or to prevent the development of breast cancer. This is particularly important for management of the high-risk population. For women who had diagnosis of hormonal positive breast cancer, tamoxifen, reloxifene, and aromatase inhibitors are given to them as secondary chemoprevention drugs to prevent recurrence or future development of secondary cancer. A great research effort has been spent on developing new chemoprevention drugs, as well as developing interim surrogate markers that can be used to evaluate whether the cancer risk has been modified by the drug. Since it will take many years to confirm decreased cancer events, using this final endpoint is not a practical study design for assessing the efficacy of many potential drugs. Breast density is known to be strongly associated with breast cancer risk[72,73,74,75], also it has been established as a surrogate marker for predicting who will benefit from the chemoprevention treatment [76]. Mammographic density (i.e. the density analyzed on mammography) has been extensively studied, but the drawback is that it is susceptible to variations from different patient position, different compression level and intensity, and x-ray source; thus may not be reliable. MRI, on the other hand, acquires 3-dimenisonal images without compression; hence compared to mammography it may provide a more precise measurement for evaluating changes of breast density over time [77,78,79,80], Figure 7. If the density can be measured reliably, it may provide helpful information for a woman to assess her risk more accurately to choose an optimal screening and management strategy, as well as to evaluate the potential benefit of taking chemoprevention drugs. This is a very active research area.
In summary, with all research that has been performed over the last 2-3 decades and the innovative research that is on-going, it is highly anticipated that breast MRI will continue to play a very important role in management and care of breast diseases for women in the entire clinical spectrum from risk management, screening, diagnosis, therapy, and surveillance.