Diffusion MRI in early cancer therapeutic response assessment

Abstract

Imaging biomarkers for the predictive assessment of treatment response in patients with cancer earlier than standard tumor volumetric metrics would provide new opportunities to individualize therapy. Diffusion-weighted MRI (DW-MRI), highly sensitive to microenvironmental alterations at the cellular level, has been evaluated extensively as a technique for the generation of quantitative and early imaging biomarkers of therapeutic response and clinical outcome. First demonstrated in a rodent tumor model, subsequent studies have shown that DW-MRI can be applied to many different solid tumors for the detection of changes in cellularity as measured indirectly by an increase in the apparent diffusion coefficient (ADC) of water molecules within the lesion. The introduction of quantitative DW-MRI into the treatment management of patients with cancer may aid physicians to individualize therapy, thereby minimizing unnecessary systemic toxicity associated with ineffective therapies, saving valuable time, reducing patient care costs and ultimately improving clinical outcome. This review covers the theoretical basis behind the application of DW-MRI to monitor therapeutic response in cancer, the analytical techniques used and the results obtained from various clinical studies that have demonstrated the efficacy of DW-MRI for the prediction of cancer treatment response.

INTRODUCTION

Monitoring cancer treatment response

Image-based assessment of cancer treatment response continues to be an active area of research with advances in medical imaging instrumentation providing opportunities to fundamentally change the clinical management of patients with cancer. MRI represents a key modality that has found use in the diagnosis, treatment planning, and assessment of response and recurrence of solid malignancies. By providing high spatial resolution and soft tissue contrast, MRI allows exquisite noninvasive radiographic detection of tumor location, whilst also providing a determination of the tumor number and dimensions.

Computed tomography and, soon after, MRI have been used since the 1960s to measure gross changes in tumor volume following a therapeutic intervention (1). Although there have been advancements in quantitative imaging techniques, such as diffusion-weighted MRI (DW-MRI), dynamic contrast-enhanced MRI (DCE-MRI) and fluorodeoxyglucose-positron emission tomography (FDG-PET), standard practice for patient management and clinical trials continues to employ anatomical images to assess tumor response to treatment (2–4). The World Health Organization (WHO) and the Response Evaluation Criteria in Solid Tumors (RECIST) have proposed guidelines primarily based on a single linear summation of specific lesions, where monitoring of the morphological changes in tumor volume allows for routine measurements for cancer response assessment. Nevertheless, there continues to be growing concerns regarding the adequacy of these criteria as some treatments, such as molecularly targeted agents, may provide therapeutic benefit without significantly reducing the tumor volume (5–7). These concerns underscore the urgency for the development and implementation of reliable response imaging biomarkers or surrogates that can detect response to treatment earlier than current methodologies (8,9).

GENERAL CONCEPTS IN DIFFUSION

The first diffusion MR sequence was demonstrated in 1965 by Stejskal and Tanner (10) and, by the 1980s, DW-MRI of in vivo systems was reported (11–13). Since then, reviews have been generated on the principles and technical aspects of this MR technique, as well as consensus recommendations using diffusion imaging as a response metric for treatment assessment (14–16). Molecular diffusion is the thermally driven random translational motion of molecules in media, which is referred to as ‘Brownian motion’. Key factors that exert their influence on the mobility of a diffusing molecule include medium viscosity, temperature and its molecular mass. Diffusion is not a magnetization-related process such as, for example, T₁ and T₂ magnetization relaxation, which drives conventional MRI contrast. Nevertheless, MRI can be used to noninvasively quantify (image) water diffusion values spatially in vivo. This is accomplished in part through the use of magnetic gradients that allow for the ‘encoding’ of initial locations of constituent water molecules in the tissue. Following a brief interval, the same gradients are used to ‘decode’ the molecular locations. For those water molecules in which displacement has occurred during the time interval, decoding will be incomplete, resulting in the loss of signal through spin dephasing. The dephasing amount increases in proportion to the distance translated between encode/decode diffusion gradient pulses. Highly mobile water molecules will have greater attenuation of the signal relative to water in more restricted/cellular tissue environments. The determination of the degree of signal loss at various diffusion gradient settings provides for the ability to calculate molecular mobility in complex systems, such as tumor tissue. However, because tumor tissue is composed of water located in a highly complex microenvironment, the concept of a single diffusion coefficient is not valid and, as such, it is reported as an ‘apparent diffusion coefficient’ (ADC) (13,14). ADC measurements can be used to assess a myriad of properties that impede molecular motions, including cell membrane integrity, cell density, interactions with macromolecules, and processes that enhance mobility via active transport, convective motion and perfusion.

The ability of water to sample its surrounding environment is the foundation behind its efficacy as a measure of tumor response to cancer. The thermal, i.e. Brownian, motion of free water at body temperature (~35 °C) is approximately 3 × 10^–3 mm²/s. Clinical DW-MRI sequences typically have a bipolar gradient interval around 50 ms, resulting in a displacement of free water molecules of 30 μm. By applying these motion-sensitive gradients, water molecules can be exploited to sample the microenvironment of biological systems well within the resolution of the MRI sequence. Structures within the solid tumor that are sampled by water molecules may include the tumor cell membranes, organelles, myelin layers and macromolecules, as well as additional cellular and subcellular entities, all of which are on the order of micrometers. Transient association of water with large, slow-moving macromolecules and cell membranes that result in water binding, as well as impediment by membranes and other structures, effectively reduce water mobility to an ADC lower than free water diffusion. The greater the bulk density of structures within a tumor tissue that impede water mobility, the lower the ADC value for that tumor. As such, ADC is considered to be a noninvasive imaging biomarker of cellularity or cell density. However, if two tissues have different ADC values, the lower ADC tissue may not necessarily have the greater number of cells per unit volume. Other factors that make up the microenvironment (e.g. cell size, viscosity, vasculature, extracellular matrix and permeability) also affect water mobility and ADC. Within a given tissue or cell type, ADC is useful as an indicator of the relative cellularity, such as in the evolution of a tumor over time following therapy. Cellular alterations caused by disease or intervention, as well as changes in cellular organization or integrity of cellular elements, are available for study by diffusion imaging.

Water diffusion on the order of cellular distances is measurable in spite of the presence of other much larger physiologic motions. A single-shot echo-planar imaging (EPI) approach (17) is the standard imaging sequence for the acquisition of diffusion-weighted imaging. By acquiring the entire set of echoes for an image within one single scanning period, respiratory bulk tissue motion, which would overwhelm the measurement of molecular motion, is essentially eliminated. By decreasing the acquisition times by a factor of 100, EPI also allows DW-MRI to be incorporated as a standard MRI sequence in clinical scanners to be used in routine clinical scanning protocols. However, images generated by EPI are sensitive to artifacts, such as distortion and signal loss owing to magnetic susceptibility. These limitations aside, EPI is the most commonly used clinical sequence, combined with diffusion sensitization gradient pulses, to perform DW-MRI.

ADC AS A MEASURE OF TUMOR CELLULARITY

It is traditionally viewed that, as cellular density increases, the added tortuosity within the microenvironment reduces water mobility. Figure 1 illustrates the effect of an effective therapeutic agent on the water diffusivity in a solid tumor mass (18). Solid tumors typically have a mean ADC value around 1 × 10^–3 mm²/s (Fig. 1). Following the intervention of a therapeutic agent that results in cell killing (i.e. a decrease in tumor cellularity), the extra-cellular space increases as the intracellular space diminishes (Fig. 1). This results in a shift in the tumor water diffusivity to higher values in therapeutically responsive regions of the tumor. Several groups have reported the inverse relationship between ADC and cellular density (19–22). To aid in the interpretation of these results, a biphasic model relating ADC values to cellularity has been proposed in which two pools of water within the tissue exist: a fast diffusion pool and a slow diffusion pool (23). The slow diffusion pool is proposed to consist of a water layer trapped by electrostatic forces of the cellular membranes and associated cytoskeleton. The fast diffusion pool is thought to belong to a combination of intra- and extracellular compartments which are, however, slower than free water. Both the traditional, i.e. monoexponential, and biphasic diffusion models provide for the rationale that water diffusion will decrease during cell swelling or cell proliferation, and increase during treatment-induced loss of cellular viability or density. Regardless of the underlying mechanism, the fact remains that tumor diffusion values increase as tumor tissue initially progresses from a solid, cellular lesion to an acellular, necrotic tumor during successful cytotoxic therapy. This characteristic of tumor water diffusion values provides a key opportunity to use this biophysical and quantifiable ADC parameter as a sensitive biomarker for the detection of the underlying changes in tumor cytoarchitecture associated with treatment (24).

Figure 1.

Schematic diagram of changes in water diffusivity in a tumor following an effective therapeutic agent. Changes in cellularity (left) occur with increasing molecular water mobility, measured as the apparent diffusion coefficient (ADC; right), as a tumor responds to treatment (top to bottom). As a tumor responds to therapy, an increase in extracellular space and membrane permeability occurs, which allows for increased water mobility, and is detected by diffusion-weighted MRI (DW-MRI) as an increase in ADC values. [Courtesy of ref. (18).]

DIFFUSION IMAGING TO ASSESS TREATMENT RESPONSE

Twenty years of research in preclinical studies have supported the notion that water diffusivity is highly dependent on the tumor microenvironment. This suggests that diffusion MR can be used to noninvasively detect cellular changes associated with treatment-induced cell killing in animal models (19,20,22,25–30). The key findings in many studies are that changes in ADC values precede changes in tumor volume regression, as well as being treatment independent and dose/efficacy dependent. All of this supports the claim that this imaging biomarker may indeed be used as an early surrogate for the assessment of treatment outcome.

Diffusion MRI as a method for therapeutic response assessment in the clinic was first demonstrated in patients with glioma (21). Tumors treated with radiation, with or without chemotherapy, demonstrated an increase in ADC values from baseline. The magnitude of change in ADC values correlated with cellularity in the tumor mass, albeit in a pilot study. Through advances in radiofrequency coil design, parallel imaging and rapid pulse sequencing, diffusion MRI has been demonstrated as a biomarker of treatment response in breast cancer (31–38), liver cancer (39–47), prostate cancer (34,48), rectal cancer (49–57), lymphomas (20,58–63), head and neck cancer (64,65) and metastases (29,33,37,66–72). Results from clinical studies have shown a significant difference in the mean ADC values between patients responding to treatment relative to patients who were determined to be nonresponsive to treatment.

An example of the clinical application of DW-MRI for the assessment of early treatment response was reported in patients with stage II/III breast cancer treated with neoadjuvant chemotherapy (NAC) (73). Presented in Fig. 2 are representative slices of ADC tumor maps from two patients with breast cancer who underwent two cycles of NAC. The first patient revealed an increase in tumor diffusion values (Fig. 2A), indicating that cell killing had occurred with no significant reduction in tumor size (Fig. 2B). Following the second cycle of treatment, a significant decrease in tumor volume was noted. In the second patient with breast cancer, ADC values remained stable over the treatment period and the patient was subsequently classified as non-pathological complete response (non-pCR) (Fig. 2C, D). These data reveal the tremendous potential of DW-MRI for the early monitoring of cancer treatment response.

Figure 2.

(A) Apparent diffusion coefficient (ADC) maps superimposed on the post-contrast dynamic contrast-enhanced MR (DCE-MR) images at three time points [pre-treatment, after one cycle and after all cycles of neoadjuvant chemotherapy (NAC)] for a patient achieving pathological complete response (pCR). The numbers for each panel represent the mean ADC values for each time point in the parametric map. (B) The difference image between pre-contrast and post-contrast DCE-MRI at each time point. (C) ADC maps superimposed on the post-contrast DCE-MR images at three time points (pre-treatment, after one cycle and after all cycles of NAC) for a non-pCR patient. The numbers for each panel represent the mean ADC values for each time point in the parametric map. (D) The difference image between pre-contrast and post-contrast DCE-MRI at each time point. [Courtesy of and adapted from ref. (73).]

Although an initial increase in tumor ADC values during treatment is typically associated with cell death, a subsequent decrease in tumor ADC values may occur, indicating tumor regrowth or, possibly, fibrosis. This present understanding is supported by findings in recurrent high-grade gliomas and osteosarcomas, where lower ADC values are observed in viable tumor and higher ADC values in regions of necrosis following treatment (74,75). Thus, ADC values in the context of the determination of the treatment response should probably be limited to early time intervals post-treatment initiation because of the more complex late-stage cellular processes that may complicate interpretation.

Metastatic lesions pose a very distinct problem for the treatment management of patients with cancer with disseminated disease. In many cases, primary tumors that have metastasized will seed osseous regions. Although bone scans using technetium 99m single photon emission computed tomography (Tc99m-SPECT) imaging are standard clinical practice for the diagnosis of metastatic cancer to the bone, RECIST continues to label bony tumors as ‘non-measurable’ because of the complex metabolic state of the bone interacting with the tumor. DW-MRI, with its high soft tissue contrast and resolution, has been shown to be highly sensitive to tumor response to therapy, irrespective of bone turnover. In a preliminary pilot study, Lee et al. (29) first demonstrated the utility of DW-MRI for therapeutic response assessment in two patients with metastatic prostate cancer to the bone, which was later validated in a large dataset by Reischauer et al. (76).

WHOLE-BODY DIFFUSION-WEIGHTED MRI (WBDW-MRI)

Although the aforementioned studies (29,76) focused only on treatment response in individual tumors, advances in wbDW-MRI may allow for multiple lesions to be monitored simultaneously (77,78). This is illustrated in the work by Horger et al. (59), where 20 patients with lymphoma undergoing systemic therapy were monitored using wbDW-MRI. Figure 3 demonstrates the sensitivity of wbDW-MRI for the detection of variations in therapeutic response in a single patient. Multi-focal lesions within the patient were found to have increased ADC values, suggesting that cell killing occurred following treatment, as depicted in these inverted gray-scale images (arrows). In contrast, the large tumor in the pelvic node (arrowhead) revealed a stable ADC value. Through the use of wbDW-MRI, early response assessment can now be obtained over multiple lesions, but at a cost of reduced spatial resolution.

Figure 3.

Whole-body diffusion-weighted MRI (wbDW-MRI) is presented as an early indicator of response to systemic therapy in patients with lymphoma. (A) Image of a 48-year-old man diagnosed with diffuse large B-cell lymphoma obtained at baseline shows the ubiquitous involvement of lymph nodes (e.g. cervical and retroperitoneal, small arrows) and axillary regions (large arrows) with marked restriction of water diffusivity. A larger pelvic node (arrowhead) is also seen left of the midline. (B) At day 7 following the institution of chemotherapy with rituximab (anti-CD20 antibodies) + CHOP (cyclophosphamide, hydroxydaunorubicin, vincristine, prednisolone), wbDW-MRI shows evident reduction in signal intensity in the cervical and retroperitoneal node regions (small arrows) and axillary region (large arrows) (from ADC = 0.90/0.33/0.67/0.61 to ADC = 1.66/0.73/1.36/1.22), with a corresponding increase in ADC (not shown), but a less marked response, in the pelvic node (arrowhead) (from ADC = 0.83/0.51 to ADC = 1.12/0.67) At the interim, the patient achieved complete remission. [Courtesy of ref. (59).]

ANOMALIES IN REPORTED DIFFUSION VALUES FOR TUMOR RESPONSE

Most studies have reported that tumor water ADC values typically increase following successful intervention in solid tumors. Although this trend appears to be the norm, there have been cases in which a decrease, rather than an increase, in ADC measurements has been reported to correlate with a positive response (54,79–81). As the tumor mass will respond dynamically throughout the course of fractionated therapies, the timing of the acquired DW-MRI measurement may have an impact on the findings. For example, two studies that investigated the efficacy of DW-MRI on treated rectal cancer (54,80) showed a brief, transient increase in ADC in the first week post-treatment initiation. Subsequently, a decrease in ADC was observed over the next several weeks. Histology confirmed that chemoradiation of rectal carcinoma resulted in increased interstitial fibrosis, which may have had the effect of reducing the ADC values in the tumor regions (54). The authors also drew attention to the fact that regions of obvious necrosis, as observed by MRI within the tumor mass, were not included in the volume of interest prescribed over the tumor mass. Omission of the necrotic regions would bias the measurement to lower ADC values. Therefore, the reported decreased ADC values that correlated with response appear to be primarily related to the timing of the measurement, as well as fibrotic formation following tumor cell death.

SPATIAL HETEROGENEITY IN TUMOR RESPONSE

Spatial heterogeneity in tumor response is a major confounding factor in assigning a single indicator to a patient. A given lesion often contains wide gradations of viable cellularity and necrosis, and the response of tumor subregions to treatment can be nonuniform and dependent on many factors. Histogram analysis of ADC values throughout the tumor is one approach to address heterogeneity (83,84). Although a variety of scalar quantities are derivable from tumor ADC histogram analysis, the magnitude of regional changes may be underestimated by whole-tumor summary statistics in the presence of heterogeneous response patterns. Figure 4 from ref. (85) illustrates the effect of response heterogeneity on the tumor histogram. Using simulated data, the authors demonstrated that uniform changes in tumor ADC values result in a mean ADC value that can detect alterations in tumor ADC values (Fig. 4B). Although other whole-tumor metrics may provide more sensitivity, such as the standard deviation for the case in which regions of the tumor demonstrate increasing and decreasing ADC values from baseline (Fig. 4C), we would need to know a priori the most appropriate measure. A more comprehensive evaluation has been performed on the efficacy of histogram-based measures for therapeutic response assessment using MRI-derived blood volume maps in patients with glioma (86). Although not performed using DW-MRI acquired parameters, the study observed negligible effectiveness of a variety of whole-tumor quantitative metrics for the detection of tumor response at both 1 and 3 weeks post-treatment initiation.

Figure 4.

Simulated comparison of whole-tumor histogram analysis (top row; blue line, pre-treatment tumor data; red line, post-treatment tumor data) versus the corresponding voxel-based analysis using a joint density histogram (bottom row). Histograms from tumors with no major change (A), significant uniform shift to higher apparent diffusion coefficient (ADC) values with a 34% net mean change (B) and heterogeneous ADC changes (increased and decreased ADC values) resulting in no net detectable histogram shift (C). Parametric response maps from the corresponding histograms are also shown, where, in (D), the confidence interval for the detection of change was set to 95%, and thus no significant change in red voxels (increased values) or blue voxels (decreased values) was detected. (E) An increase in the number of red voxels was detected at 29% of the total tumor voxels. (F) Both an increase and a decrease in tumor voxels of approximately 15% were detected, whereas no major shift was detected using a histogram analysis of the same data (C). [Courtesy of Ref. (85).]

An alternative image processing approach has been developed to quantify and spatially map the intrinsic treatment-associated heterogeneity of diffusion values within a tumor. This technique is referred to as ‘functional diffusion mapping’ (fDM) (87). A key element of fDM is the spatial registration of baseline and follow-up three-dimensional quantitative diffusion maps (i.e. ADC) into a single geometrical space. Further reading on the registration techniques and limitations for therapeutic response assessment is provided in refs. (85,88,89). Once registered, diffusion changes are measured on a voxel-by-voxel basis from spatially aligned pre-treatment and post-treatment initiation ADC maps. Tumor voxels are then classified by their extent of change in ADC. Although fDM was initially evaluated in patients with glioma (87,90–93), this technique has been applied to other tumor types (29,65,76,85). Figure 5 shows fDMs [also referred to as parametric response mapping (PRM_ADC)] with corresponding scatter plots from patients with head and neck squamous cell carcinoma (HNSCC) diagnosed as complete response (CR) (Fig. 5A) and partial response (PR) (Fig. 5B) following therapy. By analysis of the diffusion maps using fDM, heterogeneity in tumor response can be visualized, with red regions denoting response (i.e. increase in ADC from baseline) versus stable and decreased ADC regions depicted as green and blue, respectively. As demonstrated in a variety of tumor types, large regions of increased ADC from baseline (i.e. red voxels) were strongly correlated with treatment response, irrespective of the presence of tumor regions with stable or decreasing ADC values.

Figure 5.

Functional diffusion mapping (fDM) applied to clinical data acquired from patients with head and neck squamous cell carcinoma (HNSCC) diagnosed as pCR (pathological complete response) (A) and PR (partial response) (B). Results from the fDM analysis are presented as color-coded maps superimposed on contrast-enhanced T₁-weighted images and scatter plots with axes pre-treatment ADC (x-axis) and post-treatment ADC (y-axis). Color-coding is as follows: red, increased ADC values; blue, decreased ADC values; green, unchanged ADC values. [Courtesy of ref. (65).]

STANDARDIZATION AND REPEATABILITY OF ADC MEASUREMENTS

As discussed in this review, the biophysical premise and technical feasibility have allowed quantitative DW-MRI to become a clinically viable technique. Nevertheless, for this imaging protocol to become routine in the management of patients and clinical trials, there is a need to standardize DW-MRI acquisition schemes to account for intra and inter-vendor instrument variability (94). In an effort to bring uniformity throughout the various MRI systems, phantoms have been developed to confirm quantitative agreement across platforms. The ideal phantom must be stable throughout the imaging sequences and provide meaningful ADC measurements consistent with biological systems. As a result of the complexity of water diffusion in living tissue, the development of a phantom that is both stable and mimics all tissue properties has its difficulties. Simple fluid-based test objects are the preferred approach to phantom development using fluids that are thermally stable, readily available and safe when properly handled (95,96). In a study by Tofts et al. (97), the diffusion coefficients of 15 organic liquids were evaluated and found to stably provide repeatable ADC measurements within the relevant range of biological systems [(0.36–2.6) × 10^–3 mm²/s]. In 2011, Chenevert et al. (98) reported a temperature-controlled phantom using water cooled to near freezing. This phantom consisted of liquid water jacketed with ice water, such that the inner chamber was cooled to ~0 °C. Although water diffusivity is highly sensitive to temperature (99), jacketing the liquid water with ice allowed a stable environment with temperatures maintained for up to 4 h and a reliable, biologically relevant ADC value (~1 × 10^–3 mm²/s). The availability of stable and reproducible phantoms has allowed multi-center studies to be performed, demonstrating the repeatability of quantitative DW-MRI across platforms (100,101).

In the absence of a standard DW-MRI protocol, investigators of clinical trials are employing strategies to contend with intra-instrument variability. Affectionately referred to as the ‘coffee-break exam’, this approach acquires repeat DW-MRI examinations, minutes to hours apart, to ascertain the variability in the ADC measurement prior to therapeutic intervention. The motivation of this strategy is to characterize the noise associated with the ADC measurement for a given patient and platform in the absence of disease- or treatment-related changes in tumor physiology and anatomy. Various studies, just to name a few, have reported stable quantitative DW-MRI measurements in HNSCCs (64), hepatocellular carcinoma (102), malignant lung lesions (103), rectal cancer (104) and primary breast cancer (105). Until uniformity in DW-MRI protocols between vendors, instruments and sites is obtained, the strategy of repeat examinations prior to therapeutic intervention will help to elevate some of the variability in the ADC measurement within a given instrument.

FUTURE DIRECTIONS

The studies presented here support the use of DW-MRI as an early surrogate biomarker for tumor response assessment. In a growing body of literature, changes in tumor water diffusion values have been reported to correlate with response to therapy, despite the diverse set of tumor types, MRI manufacturers and magnetic field strengths used to collect the data, together with the varying approaches used to analyse the datasets (Figure 6, Table 1). Taken together, this reveals the overall robustness of DW-MRI for oncological treatment assessment. Clinical cancer studies on the efficacy of DW-MRI as a surrogate imaging biomarker of the tumor treatment response have demonstrated that treatment-induced cell death can be detected in responding tumors as an increased ADC value in these regions. As a result of variability in DW-MRI acquisition and analytical post-processing protocols, efforts have solidified in the publication of a consensus paper to provide for standardization across institutions (16). In addition, temperature-controlled phantoms have recently been developed to facilitate multi-center DW-MRI clinical trials (100,101). These standards are needed for data acquisition, post-image processing, timing of evaluation and the method used to generate the quantifiable metric used to report treatment response. Although the momentum for the use of DW-MRI in the context of tumor response assessment is continuing to grow, validation of DW-MRI as a surrogate imaging biomarker of response will require a large, prospective, multi-institutional trial performed in a standardized fashion between sites. Analysis of the data could also be useful for the validation of the image post-processing software and for regulatory approval as a device. Having a Food and Drug Administration (FDA)- or European-approved software package would provide additional momentum for enhancing the probability that DW-MRI will ultimately be incorporated into routine clinical practice for the management of patients with cancer. Future opportunities in employing DW-MRI in the clinical management of patients with cancer may include adaptive therapy protocols based on intra-therapy evaluation of early ADC changes during fractionated dosage schedules, allowing for the modification of interventions and for the quantification of multi-focal disease response using wbDW-MRI (78). Finally, the recent emergence of anticancer immunotherapies raises an urgent need for the establishment of radiological metrics for assessment of the response to such experimental interventions (106–108). Further efforts investigating advanced imaging techniques, such as DW-MRI, are needed to delineate its ability to provide meaningful insights into treatment responsiveness in order for it to have a successful impact on clinical decision making.

Figure 6.

Number of annual publications on the application of diffusion-weighted MRI (DW-MRI) for therapeutic response assessment. Yearly evaluation showed a growing increase in the number of studies demonstrating the efficacy of DW-MRI for cancer response to treatment. The search was performed on Pubmed using the following criteria [((diffusion OR ADC OR “apparent diffusion coefficient”) AND MRI AND response) NOT (stroke OR review)]. Individual references were manually evaluated.

Table 1.

Please provide legend

Site	Reference
Abdominal	(109)
Acoustic neuroma	(110)
Bladder	(111,112)
Bone marrow	(113)
Brain	(26,87,93,114–138)
Breast	(35–38,139–152)
Cervical	(153–160)
Eye	(161,162)
Leiomyoma	(163–165)
Liver	(41,42,44,46,70,166–181)
Lung	(182–185)
Lymphoma	(186–188)
Myeloma	(189,190)
Ovarian	(191–193)
Pancreas	(194)
Prostate	(29,195–198)
Rectal	(54,79,199–207)
Sarcoma	(208–214)
HNSCC	(65,215–220)

¹HNSCC, head and neck squamous cell carcinoma.

Abbreviations used

ADC

apparent diffusion coefficient

CR

complete response

DCE-MRI

dynamic contrast-enhanced MRI

DW-MRI

diffusion-weighted MRI

EPI

echo-planar imaging

FDG-PET

fluorodeoxyglucose-positron emission tomography

FDM

functional diffusion map

HNSCC

head and neck squamous cell carcinoma

NAC

neoadjuvant chemotherapy

PCR

pathological complete response

PR

partial response

PRM

parametric response mapping

RECIST

Response Evaluation Criteria in Solid Tumors

Tc99m-SPECT

technetium 99m single photon emission computed tomography

wbDW-MRI

whole-body diffusion-weighted MRI

WHO

World Health Organization