| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Academic Department of Radiology, Royal Marsden Hospital, Downs Rd., Sutton,
Surrey, United Kingdom, SM2 5PT.
2 Royal Marsden Hospital, Institute of Cancer Research, Sutton, Surrey, United
Kingdom, SM2 5NG.
Received May 3, 2006;
accepted after revision September 12, 2006.
Supported by Cancer Research UK Grant C1060/A808.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Twenty patients with potentially operable hepatic lesions larger than 1 cm in diameter metastatic from colorectal carcinoma were prospectively evaluated with diffusion-weighted imaging at three b values before and after chemotherapy. Quantitative ADC maps were calculated with images with b values of 0, 150, and 500 s/mm2 (ADC0-500) and with images with b values of 150 and 500 s/mm2 (ADC150-500). Regions of interest were drawn around metastatic lesions and randomly over liver. The mean ADC0-500 and mean ADC150-500 of metastatic lesions before and after chemotherapy were compared according to response defined by Response Evaluation Criteria in Solid Tumors criteria.
RESULTS. Twenty-five responding and 15 nonresponding metastatic lesions were evaluated. Nonresponding lesions had a significantly higher pretreatment mean ADC0-500 and mean ADC150-500 than did responding lesions (Mann-Whitney U test, p < 0.002). There was a linear regression relation (r2 = 0.34, p = 0.02) between percentage size reduction of metastatic lesions and pretreatment mean ADC150-500. After chemotherapy, responding lesions had a significant increase in mean ADC0-500 and ADC150-500 (Wilcoxon's signed rank, p = 0.025). No significant change was observed in nonresponding metastatic lesions (Wilcoxon's signed rank, p > 0.5) or in normal liver parenchyma (Wilcoxon's signed rank, p >0.4).
CONCLUSION. High pretreatment mean ADC0-500 and mean ADC150-500 of colorectal hepatic metastatic lesions were predictive of poor response to chemotherapy. A significant increase in mean ADC0-500 and ADC150-500 was observed in metastatic lesions that responded to chemotherapy. These findings may have implications for development of individualized therapy.
Keywords: abdominal imaging • apparent diffusion coefficient • cancer • diffusion-weighted imaging • liver • MRI
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Tumor response is conventionally assessed by measurement of percentage reduction in tumor size after chemotherapy. However, tumor size measurement on CT or MRI is insensitive to early treatment changes and may be inappropriate for monitoring the effects of novel therapeutic agents, which are frequently cytostatic. Measurement of metabolic response on PET and PET/CT has been found more sensitive for early response in patients with colorectal cancer [8, 9]. Diffusion-weighted MRI (DWI) provides unique information related to tumor cellularity and the integrity of cell membranes and thus may be sensitive to changes in the tumor microenvironment that occur after treatment.
High-quality images of the liver can be obtained when DWI is performed with breath-hold single-shot echo-planar sensitivity encoding (SENSE) [10-13]. The entire liver can be evaluated in less than 1 minute with three gradient factors (b values of 0, 150, and 500 s/mm2) applied in the phase-encoding, frequency-encoding, and slice-select directions [14].
Diffusion-weighted images can be evaluated quantitatively for calculation of a flow-sensitive apparent diffusion coefficient (ADC) with use of all three b values (0, 150, and 500 s/mm2 [ADC0-500]) and of a flow-insensitive ADC with use of only the b values 150 and 500 s/mm2 (ADC150-500) [14]. It has been shown that hepatic metastatic lesions of colorectal cancer have higher mean ADC0-500 and ADC150-500 values than normal hepatic parenchyma [14]. To our knowledge, however, the value of these quantitative indexes in prediction of disease response of colorectal hepatic metastasis has not been shown. Furthermore, to our knowledge, the change in ADC of hepatic metastasis in response to chemotherapy has not been reported. The aims of this study were to determine whether pretreatment ADC0-500 and ADC150-500 of colorectal hepatic metastasis can be used to predict response to chemotherapy and to compare the ADC0-500 and ADC150-500 of metastasis before and after chemotherapy.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patient Population
Twenty consecutively enrolled patients (12 men, eight women; mean age, 57.3
years; age range, 45-72 years) were prospectively evaluated. All patients
underwent surgical resection for the management of colorectal cancer, which
was histologically proved. Nineteen patients had moderately differentiated
adenocarcinoma, and one patient had mucinous carcinoma. The tumor was stage T2
in one of the patients, T3 in 16 patients, and T4 in three patients. The mean
number of malignant nodes found was 2.9 (range, 0-11 nodes). The mean onset of
metastatic liver disease was 4.9 months (range, 0-12 months) after the
diagnosis of the primary tumor.
The inclusion criteria were presence of colorectal hepatic metastatic lesions measuring more than 1 cm in maximum diameter on sonography, CT, or MRI; potential for surgical resection of hepatic metastasis, defined as resection of any number of metastatic lesions with sparing of at least two contiguous hepatic segments; and no contraindications to preoperative chemotherapy. The exclusion criteria were history of any other malignant disease and contraindications to MRI.
MRI Technique
MRI was performed on a 1.5-T system (Intera, Philips Medical Systems) with
a SENSE phasedarray body coil. Images of the liver for all patients were
acquired with a breath-hold gradient-echo T1-weighted sequence (TR/TE,
128/4.6; field of view, 450 cm; flip angle, 80°; matrix size, 256 x
256; SENSE factor, 1.8; number of slices, 25; slice thickness, 7 mm; slice
gap, 1 mm; scan duration, 22 seconds) and a free-breathing turbo spin-echo
T2-weighted sequence (1,800/40 and 1,800/350; flip angle, 90°; field of
view, 450 cm; matrix size, 256 x 256; SENSE factor, 1.8; slice
thickness, 7 mm; slice gap, 1 mm; respiratory triggering; scan duration, 2.5
minutes).
Axial DWI of the liver was performed with a breath-hold single-shot echo-planar technique. Twelve sections were acquired during each 20-second breath-hold (1,850/56; flip angle, 90°; gradient strength, 30 mT; section thickness, 7 mm; slice gap, 1 mm; number of acquisitions, 1; field of view, 340 cm; matrix size, 112 x 256; SENSE factor, 2). Examination of the entire liver was typically completed in less than 1 minute during two breath-holds.
Diffusion gradients with three b values (0, 150 and 500 s/mm2) were applied in three directions: phase encoding, frequency encoding, and slice select. The second b value of 150 s/mm2 was chosen because it has been shown to null perfusion effects within hepatic parenchyma [14, 15]. Nine images were obtained at each level of the liver: one image at a b value of 0 s/mm2, one image in each direction of b values of 150 s/mm2 and 500 s/mm2, and one index isotropic image at b values of 150 s/mm2 and 500 s/mm2.
Assessment of Treatment Response
All patients in this study received standard preoperative chemotherapy,
which comprised 12 weeks of treatment with oxaliplatin and capecitabine. DWI
was performed within 3 weeks of completion of chemotherapy. The maximum
diameter of each metastatic lesion was measured on MR images before and after
treatment for categorization and analysis of the results of DWI according to
the Response Evaluation Criteria in Solid Tumors
[16]. A metastatic lesion was
considered responding if it had 30% or greater reduction in maximum transverse
diameter. A metastatic lesion was classified as nonresponding if it had less
than 30% reduction in maximum transverse diameter or increased size after
treatment.
Image Analysis
All MR images were evaluated in consensus by an experienced body MR
radiologist with more than 10 years of experience in liver imaging and a
senior MR technologist with more than 15 years of experience in MRI. The
pretreatment and posttreatment MR images were evaluated independently lesion
by lesion. The reviewers did not know the treatment response.
Diffusion-weighted images were transferred onto a PC (Windows XP, Microsoft)
and analyzed with dedicated IDL-based (Research Systems, Inc.) software
(DiffusionView, Institute of Cancer Research). The software was used for
calculation of the ADC maps, data optimization by noise filtration,
cross-correlation analysis, and image registration.
Evaluation of metastasis—For each metastatic lesion encountered, the following were recorded: lesion size, that is, the maximum diameter measured with a caliper tool to the nearest millimeter on axial images with a b value of 0 s/mm2; lesion location according to Couinaud segmental anatomy; and ADC0-500 and ADC150-500. After identification of a metastatic lesion on an image slice, cross-correlation analysis was used to identify the presence of motion before generation of ADC maps. A cross-correlation coefficient of 0.80 or greater indicated absence of significant motion. Simple body-rigid registration was applied to the images if the coefficient was less than 0.80. Cross-correlation analysis was used to compute the similarity of sample populations formed by the pixel values on the individual images. Pairwise analysis was performed for images obtained at a particular slice location. The image with a b value of 0 s/mm2 was used as a reference. A high value for cross-correlation indicates a high degree of similarity. Conversely, misregistration of diffusion-weighted images results in a low cross-correlation value. A 4 x 4 median filter was applied to all the images to reduce image noise.
Maps of the flow-sensitive coefficient (ADC0-500) were generated on images with b values of 0, 150, and 500 s/mm2. Maps reflecting the flow-insensitive coefficient (ADC150-500) were generated only on images with b values of 150 and 500 s/mm2. For each patient, regions of interest (mean size, 288 mm2; size range, 90-1,100 mm2) were drawn around metastatic lesions on images with a b value of 0 s/mm2. These regions of interest were transferred onto the ADC0-500 and ADC150-500 maps for recording of mean values (Fig. 1A, 1B, 1C). Pretreatment mean ADC0-500 and mean ADC150-500 were compared between responding and nonresponding metastatic lesions. Mean ADC0-500 and mean ADC150-500 before and at the end of chemotherapy were compared for responding and nonresponding metastatic lesions.
|
|
|
Statistical Analysis
All statistical analysis was performed with SPSS software (version 11.5,
SPSS, Inc.). Univariate analysis was performed to determine whether age, sex,
and lesion size were significantly related to pretreatment ADC0-500 and
ADC150-500 values. ADC0-500 and ADC150-500 values also were correlated with
features of the primary tumor, such as the tumor grade, tumor stage, number of
malignant nodes, and time between tumor detection and onset of metastatic
liver disease.
Because of the unequal variance between the two groups, pretreatment mean ADC0-500 and ADC150-500 values were compared between responding and nonresponding metastatic lesions by use of the Mann-Whitney U test. Receiver operating characteristic analysis was performed to determine a threshold ADC for use in differentiating nonresponding from responding metastatic lesions. Linear regression analysis was performed to determine whether there was a significant relation between percentage reduction in the size of a metastatic lesion and pretreatment ADC0-500 or ADC150-500.
Pretreatment and posttreatment mean ADC0-500 and mean ADC150-500 were compared for responding and nonresponding metastases by use of Wilcoxon's signed rank test. Pretreatment and posttreatment mean ADC0-500 and mean ADC150-500 of liver parenchyma also were compared by use of Wilcoxon's signed rank test. Two-sided tests were used, and p < 0.05 was considered significant.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Of the 40 metastatic lesions evaluated quantitatively, 25 responded to chemotherapy and 15 did not. Imaging showed that six of the 25 responding lesions had a complete response to chemotherapy. One patient with two metastatic lesions had a mixed response; one lesion regressed markedly but the other did not. Twenty-eight metastatic lesions were surgically resected in 13 patients. All 25 responding lesions and three nonresponding lesions were confirmed at histopathologic examination after surgery. The six lesions that had a complete response had imaging evidence of posttreatment fibrosis at the site of previous disease.
No significant relation was found between pretreatment mean ADC0-500 or mean ADC150-500 of metastatic lesions and patient age (F-value = 0.85, p = 0.36; F-value = 0.002, p = 0.96), sex (F-value = 0.04, p = 0.85; F-value = 0.72, p = 0.40), or lesion size (F-value = 0.93, p = 0.34; F-value = 0.60, p = 0.44). No significant correlation was found between mean ADC0-500 or mean ADC150-500 of metastatic lesions and features of the primary tumor, such as tumor stage, tumor grade, number of malignant lymph nodes, and time to onset of metastatic liver disease (r = -0.3 to 0.05, p >0.05).
Evaluation of Metastatic Lesions
In terms of distribution of metastatic lesions, one lesion was evaluated in
eight patients, two lesions in five patients, three lesions in six patients,
and four lesions in one patient. The mean cross-correlation value of images
was 0.83 (range, 0.79-0.92). Only one case showed cross-correlation less than
0.80, for which rigid body registration was performed for the images.
The mean diameter of metastatic lesions before treatment was 22.8 mm
(range, 10-68 mm). The right lobe (segments V-VIII) contained 87.5% (35/40) of
the lesions. The other 22.5% (5/40) of the lesions were in the left lobe
(segments I-IV). Nonresponding metastatic lesions were found to have
significantly higher pretreatment mean ADC0-500 and ADC150-500 than responding
lesions (Table 1) (Mann-Whitney
U test, p 
0.001).The box-and-whisker plots of the mean
ADC0-500 and ADC150-500 of responding and nonresponding metastatic lesions are
shown in Figure 2. Results of
receiver operating curve analysis (Fig.
3) showed that a threshold ADC150-500 value of 1.69 x
10-3 mm2/s had 60% sensitivity and 100% specificity for
identifying lack of response to chemotherapy.
|
|
|
There was a significant correlation between percentage reduction in size of metastatic lesions and pretreatment mean ADC0-500 (r =0.48, p = 0.002) and mean ADC150-500 (r =0.58, p < 0.001). However, a statistically significant linear regression relation was found only between percentage reduction in lesion size and pretreatment mean ADC150-500 value (r2 =0.34, p = 0.02). The graph of linear regression with 95% mean prediction interval for percentage reduction in size versus pretreatment mean ADC150-500 is shown in Figure 4.
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The development of DWI of the liver has lagged behind its use in the brain and pelvis because of the need to overcome motion-induced artifacts. Use of parallel acquisition technique (SENSE) for single-shot echo-planar DWI improves measured signal intensity [29], shortens acquisition time, and minimizes motion artifacts, allowing acquisition of high-quality images of the entire liver in two breath-holds. Imaging through the left lobe of the liver, however, may still be degraded by artifacts that arise from cardiac pulsations [14] and obscure lesions.
DWI with more than two b values enables calculation of flow-sensitive (ADC0-500) and flow-insensitive (ADC150-500) characteristics of hepatic metastatic lesions [14]. Because the liver is highly perfused and metastatic lesions have variable vascularity, eliminating the contribution of vascular flow to the calculation of ADC150-500 by omitting the image with a b value of 0 s/mm2 should result in a more accurate estimate of diffusion distances in the extracellular space, which indirectly reflects tissue cellularity and cell membrane integrity. By the same reasoning, a change in ADC150-500 after treatment would reflect changes in tumor cellularity and cell membrane integrity unaffected by concurrent perturbations in vascular capillary perfusion.
Previous studies of DWI of primary rectal cancer have shown quantitative DWI findings predictive of response to chemotherapy and chemoradiation treatment [23, 24]. One study [24] showed a strong negative correlation between mean pretreatment tumor ADC and percentage size reduction of tumors after chemotherapy (r = -0.67, p = 0.01) and chemoradiation (r = -0.83, p = 0.001). Another study [23] showed that although there was no significant difference between the mean ADC of patients who responded well to chemotherapy and those who did not, the patients who did not respond had a higher fraction of high ADC values than the patients who did respond. Both studies showed that in primary rectal cancer, high pretreatment ADC in a tumor was predictive of poorer response to chemotherapy.
Although the behavior of primary rectal cancer in response to chemotherapy can be predicted with DWI, the value of DWI in predicting the treatment response of colorectal hepatic metastasis has not been ascertained, to our knowledge. Our study showed that high pretreatment ADC0-500 and ADC150-500 were predictive of poor response to oxaliplatin- and 5-fluorouracil-based chemotherapy. Our results are in agreement with findings observed for primary rectal cancer. Using our imaging protocol, we determined with receiver operating characteristic curves that a mean pretreatment ADC150-500 of 1.69 x 10-3 mm2/s had 60% sensitivity and 100% specificity for identification of metastatic lesions not responding to chemotherapy. In addition, a significant linear regression relation was found between mean ADC150-500 and percentage reduction in size of metastatic lesions after treatment.
The biologic basis for the observed difference in ADCs in responding and nonresponding metastatic lesions can be postulated. Higher ADC is observed in necrotic tissue and in tissue with loss of cell membrane integrity. Before chemotherapy, the presence of these changes may indicate a more aggressive phenotype. Regions of necrosis within a tumor are usually poorly perfused, resulting in less delivery of chemotherapeutic drugs to these areas. Furthermore, because of necrosis, tumor cells in these areas are exposed to a more hypoxic and acidic environment, which diminishes the effectiveness of chemotherapy [30]. The presence of necrosis in colorectal hepatic metastatic lesions is well recognized and was reported in 49% of pathologically proven lesions in one study [31]. However, necrosis within tumor may not always be associated a high ADC. In theory, coagulative necrosis without tumor cell lysis or liquefaction may not increase the ADC. This explanation may apply to the tumors in our study that did not respond favorably to chemotherapy despite having lower ADC0-500 and ADC150-500 values.
We found a difference based on change in ADC in responding and nonresponding metastatic lesions before and after chemotherapy. Lesions that responded to chemotherapy had a significant increase in ADC0-500 and ADC150-500 at the end of treatment. This finding suggests a change from a more cellular pretreatment to a less cellular or necrotic posttreatment phenotype. No significant change in ADC was observed among the nonresponding lesions or within normal-appearing liver parenchyma.
The changes in ADCs in our study in responding metastatic lesions were measured at the completion of treatment. In the future, it may be possible to use DWI to gain insight into early treatment response. A 2005 study [27] with rats showed it is possible to discriminate nonperfused but viable from nonviable necrotic tumor tissue on the basis of differences in ADCs measured 1 and 6 hours after administration of combrestatin A4 phosphate, a vascular targeting agent. In another study [20], in which the subjects were patients with cerebral glioma undergoing radiation therapy treatment, an increase in ADC within 1 week of treatment correlated with tumor response at the end of radiation therapy. DWI is thus emerging as a potentially powerful tool for monitoring drug effects and early treatment response at both cranial and extracranial sites. Novel therapeutic agents (e.g., bevacizumab) are in use for the management of colorectal metastasis. Such drugs may not significantly reduce tumor size, particularly during early treatment, and new potential surrogate markers of treatment response would therefore be welcomed. Future studies should be conducted to explore the relation between ADC (pretreatment or change with therapy) and long-term clinical outcome, such as disease-free and overall survival. To our knowledge, no study has shown a clear relation between an imaging biomarker and clinical outcome.
There were several limitations to this study. First, the DWI technique may not be adequate for assessment of metastasis arising in the left lobe of the liver. Further work is needed to overcome artifacts related to cardiac motion, which propagates into segments II and III of the left lobe of the liver on echo-planar imaging, thus degrading image quality. Second, although the sequence has been optimized for normal liver, metastatic lesions have longer T1 values, which can cause overestimation of ADC0-500 for the TR used in the measurement. However, the current technique allows DWI of the entire liver in three directions in two breath-holds. Calculation of trace images from measurements in three directions improves signal-to-noise ratio. Third, although we found a difference in mean ADC0-500 and mean ADC150-500 before and after chemotherapy in responding metastatic lesions, there was substantial overlap in the ADCs, which may limit use of this coefficient in monitoring therapeutic response. However, this overlap is likely in part due to the use of mean ADCs derived from whole tumor regions, which may not adequately reflect heterogeneous changes with treatment. More sophisticated methods of data analysis, such as histogram and pixelwise analysis, should to be further explored and developed. Finally, the interobserver variability of ADC measurements was not tested in this study. Assessment of the limits of error in obtaining ADC measurements within and between individuals is important to ascertain the magnitude of ADC change that can be confidently detected with the technique. We have recently completed a study addressing observer variability in ADC measurements, but future studies assessing the reproducibility of DWI should be encouraged.
High pretreatment mean ADC0-500 and ADC150-500 of colorectal hepatic
metastasis were predictive of poor response to chemotherapy. A pretreatment
mean ADC150-500 value 
1.69 x 10-3 mm2/s can
be used to identify metastasis unlikely to respond to chemotherapy. A
significant increase in mean ADC0-500 and mean ADC150-500 was observed in
metastatic lesions that responded to chemotherapy. These findings may have
implications for individualized therapy and monitoring of therapeutic
response.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Kandpal, R. Sharma, K. S. Madhusudhan, and K. S. Kapoor Respiratory-Triggered Versus Breath-Hold Diffusion-Weighted MRI of Liver Lesions: Comparison of Image Quality and Apparent Diffusion Coefficient Values Am. J. Roentgenol., April 1, 2009; 192(4): 915 - 922. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. C. Braithwaite, B. M. Dale, D. T. Boll, and E. M. Merkle Short- and Midterm Reproducibility of Apparent Diffusion Coefficient Measurements at 3.0-T Diffusion-weighted Imaging of the Abdomen Radiology, February 1, 2009; 250(2): 459 - 465. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T NIWA, M UENO, S OHKAWA, T YOSHIDA, T DOIUCHI, K ITO, and T INOUE Advanced pancreatic cancer: the use of the apparent diffusion coefficient to predict response to chemotherapy Br. J. Radiol., January 1, 2009; 82(973): 28 - 34. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. C. Sullivan Anatomic and Functional Imaging Techniques for Response Assessment in Oncology ASCO Educational Book, January 1, 2009; 2009(1): 119 - 122. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Kim, D. E. Morgan, D. J. Buchsbaum, H. Zeng, W. E. Grizzle, J. M. Warram, C. R. Stockard, L. R. McNally, J. W. Long, J. C. Sellers, et al. Early Therapy Evaluation of Combined Anti-Death Receptor 5 Antibody and Gemcitabine in Orthotopic Pancreatic Tumor Xenografts by Diffusion-Weighted Magnetic Resonance Imaging Cancer Res., October 15, 2008; 68(20): 8369 - 8376. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Cui, X.-P. Zhang, Y.-S. Sun, L. Tang, and L. Shen Apparent Diffusion Coefficient: Potential Imaging Biomarker for Prediction and Early Detection of Response to Chemotherapy in Hepatic Metastases Radiology, September 1, 2008; 248(3): 894 - 900. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
