Dynamic contrast-enhanced MRI of primary rectal cancer: Quantitative correlation with positron emission tomography/computed tomography




To assess the correlations between parameters measured on dynamic contrast-enhanced magnetic resonance imaging and 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) in rectal cancer.

Materials and Methods

To assess the correlations between parameters measured on dynamic contrast-enhanced MRI and FDG-PET in rectal cancer.


Significant correlations were only demonstrated between kep and SUVmax (r = 0.587, P = 0.001), and kep and SUVmean (r = 0.562, P = 0.002). No significant differences were found in imaging parameters between well, moderately and poorly differentiated adenocarcinoma groups. However, there was a trend that higher imaging values were found in poorly differentiated adenocarcinomas.


Positive correlations were found between kep and SUV values in primary rectal adenocarcinomas suggesting an association between angiogenesis and metabolic activity and further reflecting that angiogenic activity in washout phase is better associated with tumor metabolism than the uptake phase. J. Magn. Reson. Imaging 2011;33:340–347. © 2011 Wiley-Liss, Inc.

ANGIOGENESIS, THE PHYSIOLOGICAL process of new blood vessel formation, is essential in human development, reproduction, and repair (1). However, pathological angiogenesis, triggered by activation of certain cellular signal pathways, is considered a key factor for solid tumor to develop, grow, and metastasize (2). Colorectal carcinoma has been found highly associated with angiogenesis and malignant rectal tumor tissue generally has greater angiogenic activity than normal rectal tissue (3–6).

Activity of tumor angiogenesis may be assessed by several methods. In human tumors, angiogenesis can be assessed by the expression levels of microvessel density and vascular endothelial growth factor (VEGF) using histopathological methods (4, 7–11). However, these techniques require tumor tissues and need to be performed on either preoperative biopsy samples or postoperative surgical specimens. Dynamic contrast-enhanced MRI (DCE-MRI) is a noninvasive imaging technique assessing angiogenesis activity by measuring signal intensity change in tumor tissue over time, which is related with the concentration change of diluted low molecular weight paramagnetic contrast agent in the extravascular-extracellular space (EES) in vivo (12–14). Quantitative data analysis was based on commonly used two-compartment pharmachokinetic modeling (blood plasma and EES) by which a set of standardized quantity terms have been generated, such as Ktrans (transfer constant between the blood plasma and the EES), kep (rate constant between the EES and the blood plasma) and the EES fractional volume (υe, interstitial space; 15). Some studies have demonstrated the relationship between VEGF or microvessel density and Ktrans or kep (16–18), suggesting that DCE-MRI may be a potential tool to characterize the angiogenic activity of a tumor other than invasive histomorphological approaches.

On the other hand, increased glucose metabolism is another feature commonly seen in many tumors (19). Malignant cells often show increased glucose uptake in vitro and in vivo, which is believed to be facilitated by glucose transporters (20, 21). It has been reported that the levels of glucose transporter expression are associated with tumor aggressiveness and patient survival (20). In addition, glucose levels in the primary rectal adenocarcinoma are significantly higher than normal tissues (22). Positron emission tomography (PET) is capable of imaging tumors based on their increased glucose metabolism. 18F-fluorodeoxyglucose (18F-FDG) uptake on PET, quantified by the standardized uptake value (SUV), is a useful marker for the level of tumor glucose activity. Several studies demonstrated the relationships between higher FDG uptake and a more aggressive course of malignancy (23–26).

Even though several studies have shown that DCE-MRI and PET/computed tomography (PET/CT) are both useful in various aspects of tumor evaluations, to our knowledge, the relationships between quantitative parameters measured by DCE-MRI and PET/CT have not been studied. Because both parameters from DCE-MRI and PET/CT have been associated with biological aggressiveness and treatment response in certain tumors including rectal cancer (27, 28), we hypothesize that a correlation may exist between these imaging parameters in rectal cancer. Therefore, the aim of this study is to explore the possible correlations between kinetic parameters from DCE-MRI and tumor glucose metabolism indices from PET/CT in primary rectal cancer.

Materials and Methods


Between November 2008 and September 2009, 27 consecutive patients (15 men and 12 women; mean age, 65 years; range, 45–88 years) with newly diagnosed rectal adenocarcinoma and having undergone both MRI and PET/CT examinations were included in this study. All patients underwent both MRI and PET/CT scans within 1-week period (mean time interval: 2 ± 1 days). This study was approved by Institutional Review Board and informed consent was obtained from all patients. This study is also compliant with patient confidentiality regulations.

Acquisition of MR Images

All MR examinations were performed on a 3T scanner (Achieva; Philips Healthcare, Best, the Netherlands) with the patients in supine position. SENSE-Torso coil with parallel imaging capability was placed over the pelvis to reduce susceptibility artifact and improve imaging quality. Routine T2-weighted, T1-weighted, and contrast-enhanced sequences were obtained. Turbo spin echo T2-weighted images were obtained in axial plane using the following parameters: repetition time/echo time (TR/TE) = 1862/99 ms, field of view (FOV) = 19 × 23 cm, matrix size = 272 × 318, slice thickness = 6 mm, gap = 0, number of acquisition = 1, sense factor = 1.5. Axial TSE T1-weighted images were obtained of the entire pelvis using the following parameters: FOV = 23 × 38 cm, matrix size = 236 × 314, slice thickness = 7 mm, gap = 1 mm, number of acquisition = 1, sense factor = 1. Subsequently precontrast three-dimensional (3D) T1-weighted fast-field echo (FFE) images of the tumor were obtained in the axial plane: TR/TE = 4.9/1.52 ms, FOV = 27 × 40 cm, matrix size = 236 × 344, slice number = 10, slice thickness = 3 mm, gap = 0, flip angle = 5°, number of acquisition = 5, sense factor = 1.2. This was followed by DCE series using the same sequence except for flip angle = 15°, number of acquisition = 1, temporal resolution of 6.3 s for 10 slices, with total 35 dynamic scans. After one dynamic scan, Gadolinium-tetraazacyclododecanetetraacetic acid (Gd-DOTA, Dotarem, Guerbet, France) was injected at 0.2 mL/kg body weight intravenously followed by 25 mL of saline at 3.5 mL/s immediately using a power injector system (Spectris Solaris, MedRad, Indianola, PA).

Acquisition of PET/CT Images

All PET/CT examinations were performed on an integrated PET/CT scanner (Discovery VCT, GE Healthcare Bio-Sciences Corp., USA). All patients fasted with hydration 6 h before receiving intravenous injection of 18F-FDG at 4.8 MBq/kg body weight. PET/CT scans were performed 60 min after the injection of the 18F-FDG. A whole body emission PET scan with a 70-cm axial FOV, a 128 × 128 matrix and 3.75-mm thickness was obtained with five bed positions within 20 minutes. CT images were performed using the following scan parameters: FOV = 50 cm, matrix = 512 × 512, collimation = 0.625 mm × 64, pitch = 0.984, gantry rotation speed = 0.5 s, tube voltage = 120 kVp, and tube current = 200–400 mA. LASIX (4 mg) was administered intravenously to fill bladder to reduce artifacts from high 18F-FDG activity in urine (24). Attenuation correction was performed on PET images with CT data using an ordered-subset expectation maximization iterative reconstruction algorithm (14 subsets and 2 iterations). The CT images were then reconstructed at 2.5-mm intervals to fuse with the PET images (Advanced Workstation 4.3; GE Healthcare).

Image Analysis

Based on the precontrast T1-weighted and the DCE sequences, Ktrans, kep, and vp maps were automatically generated by DCE_Tool software (NIH, USA) using generalized kinetic model with following Eq. [1]:

equation image(1)

where Ct is tracer concentration in tissue, Cp is tracer concentration in arterial blood plasma, ktrans is volume transfer constant, kep is rate constant, νe is volume of extravascular extracellular space per unit volume of tissue (15). Arterial input function (AIF) was measured in the internal iliac artery on precontrast T1-weighted images by placing regions of interest (ROIs) inside the vessel. Then tumor were manually drawn by placing ROIs along contours of each tumor on precontrast T1-weighted images using the T2-weighted images as a guide by an investigator (J.G., with 3 years of experience in reading and performing volumetric tumor measurements on body MRI) who was blinded to PET/CT images and all clinical information other than that the patient was diagnosed with rectal cancer. Then ROIs were copied to the Ktrans, kep and vp (blood plasma volume per unit volume of tissue) maps (Fig. 2). Tumor volume was measured on T2-weighted images by placing ROIs along contours of each tumor on each slice.

Mean Ktrans value (Ktransi) and the cross-sectional area (Areai) of the tumor ROI on each slice (i representing the slice number) were calculated by Image J software (NIH, Bethesda, MD). Subsequently, Ktransmean of the measured part of tumor was calculated as the weighted average for all Ktransi values in each tumor by Eq. [2]:

equation image(2)

We calculated weighted averages because this would be mathematically identical to calculating averages of Ktrans values directly from all voxels within the measured tumor volume.

Mean kep and mean vp acquisition were calculated as following equations [3, 4]:

equation image(3)
equation image(4)

In addition, mean υe was calculated as:

equation image(5)

For SUV measurement, the PET, CT, and fused PET/CT images were displayed on a workstation (Advanced Workstation, 4.3, GE Healthcare), and a 3D ROI was placed over the entire tumor by an investigator (J.Z., with 2 years of experience in interpreting PET/CT) who was blinded to MRI images and all clinical information other than that the patient was diagnosed with rectal cancer. Maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), volume of tumor and total lesion glycolysis (TLG) were automatically calculated by the workstation (Fig. 3). SUV was defined by the following equation [6]:

equation image(6)

SUVmax was defined as the highest value of SUV among all voxels within the 3D ROI placed over the rectal tumor on PET. Subsequently, a fixed threshold value of 50% of the maximum uptake was used to determine tumor margins automatically, and the tumor volume was calculated by the workstation accordingly (26). The SUVmean was then measured as the average of SUV values in all voxels within the threshold-defined tumor volume. Ratio of SUV (rSUV) was calculated as SUVmax/SUVmean. TLG was then calculated using Eq. [7]:

equation image(7)

Clinical Correlation

All patients had biopsy under colonoscopy that confirmed histopathological diagnosis of rectal cancer before the imaging examinations were performed. Circulating plasma level of carcinoembryonic antigen (CEA) was obtained within two weeks of MRI scan (mean time interval, 9 ± 3 days). Based on the references from our institution, CEA >5.0 mg/mL was regarded as abnormal.

Statistical Analysis

All results were expressed as mean ± standard deviation (SD). Pearson's correlation test was used to detect the relationships between quantitative indices of DCE-MRI and PET. Bland-Altman plot were performed to assess agreement between volumes measured on T2-weighted images and PET. One-way analysis of variance (ANOVA) was used to analyze differences of parameters from DCE-MRI and PET in between well, moderately or poorly differentiated rectal adenocarcinoma groups. A P value of <0.05 was considered to indicate statistical significance. All statistical analyses were performed using SPSS software package (SPSS, Version 16.0.1, Chicago, IL, USA).


Characteristics of Patients and Lesions

All 27 patients were confirmed to have rectal adenocarcinomas by pathological diagnosis. Among these patients, 6 (22.2%) had well differentiated adenocarcinomas, 14 (51.9%) had moderately differentiated adenocarcinomas, and 7 (25.9%) had poorly differentiated adenocarcinomas. CEA levels in the patients were 35.4 ± 70.2 mg/mL (range, 0.2–265.0 mg/mL). Eleven of 27 (40.7%) patients had abnormal CEA levels (>5 mg/mL), whereas the other 16 patients (59.3%) had normal CEA levels. Characteristics of study subjects are shown in Table 1.

Table 1. Characteristics of Patients and Lesions
  1. CEA = carcinoembryonic antigen.

Age65 ± 11 years
GenderF = 12, M = 15
CEA35.4±70.2 mg/mL
Patients with normal CEA level (≤ 5.0 mg/mL)16 (59.3%)
Patients with abnormal CEA level (>5.0 mg/mL)11 (40.7%)
Pathological diagnosisNumber of lesions (%)
Well differentiated rectal adenocarcinoma6 (22.2%)
Moderately differentiated rectal adenocarcinoma14 (51.9%)
Poorly differentiated rectal adenocarcinoma7 (25.9%)
Clinical tumor stagingNumber of patients (%)
Stage I0
Stage II5 (18.5%)
Stage III20 (74.1%)
Stage IV2 (7.4%)

Figure 1 shows the concentration-time curves derived from DCE-MRI in a typical case. Fitted curve with goodness of fit R2 for each tumor was automatically calculated and shown on the same graph. Figure 2 demonstrates the measurement of Ktrans, kep, and vp values in the same patient. The kep values of tumors were 0.74 ± 0.36 min−1 (range, 0.16–1.66 min−1). The Ktrans values were 0.23 ± 0.10 min−1 (range, 0.05 – 0.50 min−1). The ve values were 0.33 ± 0.12 (range, 0.17–0.60). The vp values were 0.09 ± 0.05 (range, 0.02–0.27). Tumor volumes measured by T2-weighted images was 21.3 ± 13.9 cm3 (range, 3.1–59.6 cm3).

Figure 1.

Concentration-time curves derived from DCE-MRI in an 88-year-old male with moderately differentiated rectal cancer. The red curve for arterial input function (AIF) shows contrast uptake in the internal iliac artery. The blue curve for uptake in rectal cancer shows the steep upslope followed by a very short phase of plateau and a long phase of washout. The yellow solid curve is the fitted curve for rectal cancer with goodness of fit R2 of 0.934 in this case. Then maps of kep, Ktrans, and vp were generated by DCE_Tool software (NIH, Bethesda, MD) using generalized kinetic model.

Figure 2.

An 88-year-old man with moderately differentiated rectal adenocarcinoma (arrows). a: On precontrast T1-weighted image, an ROI was drawn (red line) along the contour of the tumor. b: Fused T1-weighted image and map of Ktrans generated by DCE-Tool software using a two-compartment general kinetic model. c–e: ROIs (red line) were copied from a to kep, Ktrans, and vp maps of the tumor, respectively, which were generated by DCE-Tool software and had anatomic location identical to that of the precontrast T1-weighted image.

Figure 3.

A 57-year-old woman with moderately differentiated rectal adenocarcinoma. a–c: Fused PET/CT images in the coronal plane (a), sagittal plane (b), and axial plane (c) show a hypermetabolic lesion in the rectum. A 3D ROI (green box) was placed to cover the entire lesion on the Advanced AW Workstation. d: From the corresponding PET images, SUVmax of tumor was measured. Subsequently SUVmean, tumor volume and TLG were calculated automatically using a threshold of 50% SUVmax.

Figure 3 demonstrates the measurement of SUV, TLG, and tumor volume on PET/CT in a typical case. The values of SUVmax, SUVmean, and rSUV of all tumors were 9.9 ± 3.9 (range, 4.5–20.7), 6.5 ± 2.7 (range, 3.1–14.5), and 1.5 ± 0.1 (range, 1.4–1.9), respectively. TLG values were 105.4 ± 90.7g (range, 8.5–330.2 g). Tumor volumes measured by PET/CT were 17.4 ± 9.6 cm3 (range, 3.0–57.5 cm3), using 50% of SUVmax as threshold.

Correlations between DCE-MRI Indices and PET Indices, CEA Levels

Correlations between quantitative DCE-MRI and PET indices are shown in Table 2. Significant positive correlations were demonstrated between kep and SUVmax (r = 0.587, P = 0.001), kep and SUVmean (r = 0.562, P = 0.002), but not between kep and TLG (r = 0.337, P = 0.086), or kep and rSUV (r = 0. 063, P = 0.755; Table 2; Fig. 4). No correlations were found between Ktrans, ve, or vp values and SUVmax, SUVmean, rSUV, or TLG (Table 2).

Figure 4.

Scatter plots showing the significant positive correlations between kep and SUVmax (a), kep and SUVmean (b). Circles = well differentiated rectal cancer, squares = moderately differentiated rectal cancer, triangles = poorly differentiated rectal cancer.

Table 2. Pearson's Correlation Analyses Between DCE-MRI Indices and PET/CT Indices, CEA Levelsa
  • a

    P values less than 0.05 are shown in boldface font. kep = rate constant; Ktrans = volume transfer constant; vp = blood plasma volume per unit volume of tissue; υe = volume of extravascular extracellular space per unit volume of tissue; SUVmax = maximum of standard uptake value; SUVmean = mean of standard uptake value; TLG = total lesion glycolysis; rSUV = SUVmax/SUVmean; CEA = carcinoembryonic antigen.


No correlations were found between quantitative DCE-MRI indices kep, Ktrans, ve, vp values and CEA levels (Table 2).

Quantitative Imaging Parameters and Tumor Differentiation

Weak differences were found between well, moderately, and poorly differentiated adenocarcinoma groups in terms of kep (p = 0.071) and SUVmax (P = 0.099). Although no significant differences were found between the three groups in imaging parameters, there was a trend that higher values of kep, Ktrans, vp, SUVmax, and SUVmean were present in poorly differentiated adenocarcinomas (Table 3).

Table 3. Imaging Parameters and Tumor Differentiation
 Well differentiated n = 10Moderately differentiated n = 14Poorly differentiated n = 9P value
  1. kep = rate constant; Ktrans = volume transfer constant; vp = blood plasma volume per unit volume of tissue; υe = volume of extravascular extracellular space per unit volume of tissue; SUVmax = maximum of standard uptake value SUVmean = mean of standard uptake value; TLG = total lesion glycolysis; rSUV = SUVmax/SUVmean.

kep (min−1)0.58 ± 0.260.68 ± 0.371.00 ± 0.310.069
Ktrans (min−1)0.23 ± 0.150.20 ± 0.090.28 ± 0.070.248
vp0.08 ± 0.040.09 ± 0.040.11 ± 0.080.726
νe0.39 ± 0.130.32 ± 0.130.30 ± 0.090.371
SUVmax7.28 ± 1.369.93 ± 4.3511.91 ± 3.420.099
SUVmean4.77 ± 0.906.49 ± 3.067.84 ± 2.270.118
rSUV1.53 ± 0.01.55 ± 0.131.52 ± 0.060.877
TLG (g)86.66 ± 116.00115.71 ± 97.44100.95 ± 57.210.636

Agreement Between Tumor Volumes Measured on MR and PET

A Bland-Altman plot analysis showed good agreement between tumor volumes measured on T2-weighted image and PET/CT, although by using the standard 50% threshold, PET/CT underestimated the rectal cancer volume compared to T2-weighted imaging.


Cancers are characterized by variable biological characteristics, such as sustained angiogenesis and increased glucose metabolism which were well reflected by functional imaging modalities, such as DCE-MRI and PET/CT, respectively. In terms of rectal cancer, PET/CT has been shown to play a role for staging, pretreatment planning, as well as posttreatment follow-up (23, 27, 29, 30). SUV is a measurement of metabolic activity per body weight. TLG evaluates metabolic activity in the volume of the entire tumor (product of SUVmean and tumor volume). These quantitative indices of PET/CT have been shown to correlate with biological characteristics and treatment response of many solid tumors including rectal cancer (31–33). On the other hand, DCE-MRI was considered a useful functional imaging tool for measurement of angiogenesis in colorectal cancer (34). Quantitative parameters on DCE-MRI, such as Ktrans, kep, and υe, can be generated from a two-compartment model described by Tofts and Kermode (25). It has been reported that these perfusion parameters may be regarded as surrogate biomarkers to reflect biological characteristics of tumor. For example, highest kep colocalized with MVD hot spots in cervical cancer (16). VEGF was found to correlate with kep in breast cancer (17) and with Ktrans in colorectal cancer (18). kep has also been found to correlate with tissue oxygenation (35). However, Atkin et al found a paradoxical negative correlation between Ktrans and MVD, and no correlations between Ktrans, vp, or υe and tumor angiogenesis markers including VEGF and CD31 (a marker of microvessel density) in colorectal cancer (36). In their study, kep was not included into analysis. The variable results in the literature demonstrate the need for further study in this area.

In this study, we found significant positive correlations between kep and SUV values in rectal cancer. To our knowledge, this is the first study to evaluate the correlations between quantitative indices of DCE-MRI and PET/CT. Therefore, this is a preliminary study, hopefully leading to better understanding of the underlying mechanism of these observed correlations. We do know that there are three kinetic phases during the DCE-MRI: (i) Uptake phase: signal intensity increases sharply which reflects a net leakage of contrast from blood vessels into the EES in the tumor; (ii) Plateau phase: maximum enhancement with an equilibrium in the movement of contrast between the blood vessels and EES; (iii) Washout phase: contrast leaves EES and returns into blood vessels (37). Among these processes, kep happens mainly during the washout phase. Therefore, our findings may indicate that washout phase is better associated with tumor metabolism than the uptake phase. The other possible reason of this finding is that Ktrans is influenced by multiple factors including contrast concentration in vessels, vascular density, blood flow, and permeability of vascular surface area (38). For example, at the peripheral rims of tumor where permeability surface-area product is high compared with blood flow, Ktrans is dominated by blood flow; whereas, at the center of the tumor where permeability is low compared with blood flow, Ktrans mostly indicates the permeability surface-area product (34). Therefore, there may be a greater individual variability in Ktrans. This greater variability in combination of a small sample size may hinder our efforts in unveiling an underlying correlation. A larger study would be helpful in providing a higher statistical power. In contrast, kep is only influenced by the concentration of contrast in tumor EES and fractional volume of EES, and may have less variability, and thus easier to demonstrate any underlying biological characteristics of tumor. However, these speculations will need further experimental studies for confirmation. It would also be interesting to explore whether these correlation can be found in other cancer types.

In terms of correlation between the imaging parameters and tumor pathologic characteristics, we did see a trend of higher values of kep, Ktrans, vp, SUVmax, and SUVmean in the more poorly differentiated rectal tumors. However, these differences did not reach statistical significance. It is possible that the relative small patient population in our study may have led to suboptimal statistical powers. Larger studies in the future would be helpful to further evaluate these potential correlations.

There are certain limitations to our study. First, the number of subjects in our study is relatively small. However, on the other hand, we could find some significant correlations between DCE-MRI and PET parameters, despite the relatively small sample size. Second, although DCE-MRI is a noninvasive and sensitive technique for quantifying the physiology of tumor tissues, the acquisition protocols are far from uniform. Especially, semiquantitative data which analyze signal intensity changes were not readily transferable between different centers due to different machine set-up characteristics (37, 39). Therefore, comparison of studies from different groups is complicated and may result in an underestimation of the ability of DCE-MRI. Third, imaging parameters may have different diagnostic performances for different types of tumors. Therefore, the correlations we found in rectal adenocarcinomas may not be transferrable to other types of tumors, and each tumor type should be tested individually.

In conclusion, we found a positive correlation between kep from DCE-MRI and quantitative indices from PET, including SUVmax and SUVmean, suggesting an association between angiogenesis and metabolic activity in rectal adenocarcinomas. Angiogenic activity in washout phase reflected by kep may be better associated with tumor metabolism than the uptake phase in rectal adenocarcinomas.