Assessment of Tumor Grade and Angiogenesis in Colorectal Cancer: Whole-volume Perfusion CT Hongliang Sun, MD, Yanyan Xu, MD, Qiang Yang, MD, Wu Wang, MD, PhD Rationale and Objectives: The preoperative evaluation of tumor grading and angiogenesis has important clinical implications in the treatment and prognosis of patients with colorectal cancers (CRCs). The aim of the present study was to assess tumor perfusion with 256-slice computed tomography (CT) using whole-volume perfusion technology before surgery, and to investigate the differences in the perfusion parameters among tumor grades and the correlation between perfusion parameters and pathologic results in CRC. Materials and Methods: Thirty-seven patients with CRC confirmed by endoscopic pathology underwent whole-volume perfusion CT assessments with a 256-slice CT and surgery. Quantitative values for blood flow, blood volume, and time to peak were determined using commercial software. After surgery, resected specimens were analyzed immunohistochemically with CD105 antibodies for the quantification of microvessel density (MVD). The difference in CT perfusion parameters and MVD among different tumor differentiation grades was evaluated by the Student–Newman–Keuls test. The correlations between CT perfusion parameters and MVD were evaluated using the Pearson correlation analysis. Results: The mean blood flow was significantly different among well, moderately, and poorly differentiated groups (61.17 17.97, 34.80 13.06, and 22.24 9.31 mL/minute/100 g, respectively; P < .05). The blood volume in the well-differentiated group was significantly higher than that in the moderately differentiated group (33.96 24.81 vs. 16.93 5.73 mL/100 g; P = .002) and that in the poorly differentiated group (33.96 24.81 vs. 18.05 6.01 mL/100 g; P = .009). The time to peak in the poorly differentiated group was significantly longer than that in the well-differentiated group (27.81 11.95 vs. 17.60 8.53 seconds; P = .016) and that in the moderately differentiated group (27.81 11.95 vs. 18.94 7.47 seconds; P = .028). There was no significant difference in the MVD among well, moderately, and poorly differentiated groups (33.47 14.69, 28.89 11.82, and 29.89 11.02, respectively; P > .05). There was no significant correlation between CT perfusion parameters and MVD (r = 0.201, 0.295, and 0.178, respectively; P = .233, .076, and .292, respectively). Conclusions: CT whole-volume perfusion technology has the potential to evaluate pathologic differentiation grade of CRC before surgery. However, preoperative perfusion CT parameters do not reflect the MVD of CRC. Key Words: Colorectal cancer; CT; perfusion; angiogenesis. ªAUR, 2014
C
olorectal cancer (CRC) is the third most common cancer and the fourth most frequent cause of cancer deaths worldwide (1). The 5-year survival rate depends on the tumor stage and grade at patient presentation. Tumors with an advanced stage and grade at diagnosis are associated with a poor outcome. Individual treatment strategy based on tumor stage and grade should be applied to improve the prognosis. Thus, the preoperative diagnostic evaluation and grading of CRC are important (2). Preoperative specimens from endoscopic colorectal biopsies are often used but are normally failed to grade tumor because of the
Acad Radiol 2014; 21:750–757 From the Department of Radiology, China-Japan Friendship Hospital, No. 2 Yinghua East Street, Chaoyang District, Beijing 100029, China (H.S., Y.X., W.W.) and Department of Pathology, China-Japan Friendship Hospital, Chaoyang District, Beijing 100029, China (Q.Y.). Received October 9, 2013; accepted February 10, 2014. Disclosure: The authors have nothing to disclose. Address correspondence to: W.W. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.02.011
750
lack of sufficient tissue (3). Angiogenesis, which is important in the growth and metastasis of carcinomas, has been reported to be a promising prognostic marker for the CRC (4). Microvessel density (MVD) count is used to define the degree of angiogenesis in solid tumors for diagnostic purpose and treatment planning, which is calculated by counting the number of angiogenic blood vessels highlighted on a variety of immunohistochemical stains (5,6). However, information pertaining to the MVD can only be gathered in the in vitro setting after the resection of the tumor, and therefore there is no opportunity to evaluate the effect of neoadjuvant radiochemotherapy and antiangiogenic therapy on this parameter. Perfusion computed tomography (CT) can quantify tumor angiogenesis noninvasively by assessing the enhancement of the tissue and vessels over time. Perfusion parameters, including tissue blood flow (BF), blood volume (BV), time to peak (TTP), and permeability–surface area product (PS), are calculated using the mathematical models for contrast agent exchange (7,8). Goh et al. (7) reported
Academic Radiology, Vol 21, No 6, June 2014
the relationship between CT perfusion parameters and MVD counts and their data indicated a positive correlation between tumor PS and BV with MVD in CRC, inconsistent with the report by Li et al. (9), which showed no significant correlation between any perfusion parameters with MVD. Both the studies were based on the deconvolution approach, but the scanning equipments, scanning modes, analytical software were different, which might be the potential explanation for such differences and make the standardization of the technique difficult (10). Romani et al. (11) reported that CD105 (endoglin)-staining intensities in CRCs were correlated with the MVD levels and were better indicators of the state of tumor angiogenesis. Previous CRC CT perfusion studies usually use one slice or a few slices of the tumor to represent the overall tumor angiogenesis state, and to some degree, this sampling rate is not sufficient because of the heterogeneity of tumor angiogenesis. Additionally, the lesion on CT images is not exactly the same as the pathologic specimen in the orientation, shape, and size, which makes it difficult to achieve precise alignment. Therefore, we analyzed primary CRC using whole-volume perfusion CT measurements and assessed whether perfusion CT could be used to evaluate the pathologic grade and the correlation between perfusion parameters and MVD stained with CD105 in CRC. MATERIALS AND METHODS Patients
The institutional review board approved this study, and informed consent was obtained from all patients enrolled in the study. Between August 2010 and March 2012, 42 consecutive patients with CRC who underwent preoperative CT scans were prospectively enrolled in the study. Patients were excluded according to the following criteria: (1) preoperative treatment such as chemotherapy or radiotherapy (n = 2); (2) contraindication to administration of contrast medium (n = 0); (3) severe diseases of the heart, liver, lung, or kidney (n = 0); and (4) descending or ascending colon cancers with severe motion artifacts during the perfusion CT examination (n = 3). A total of 37 patients with CRC who underwent surgery after perfusion CTat our institute were enrolled in this study. The subjects included 20 men and 17 women, with a mean age of 64.5 years (range, 38–85 years). The diagnosis of adenocarcinoma of the colon and rectum was histologically confirmed by colonoscopic biopsy and surgical specimens. Mean tumor length at pathologic evaluation was 9.5 cm (range, 5.2–13.8 cm). The tumors examined were located in the cecum (n = 5), ascending colon (n = 3), descending colon (n = 3), sigmoid colon (n = 4), and rectum (n = 22). Tumors were divided into three histologic subgroups: well differentiated (7 tumors), moderately differentiated (20 tumors), and poorly differentiated (10 tumors). Thirty-seven patients underwent curative surgery without severe complications.
WHOLE-VOLUME PERFUSION CT IN CRC
Imaging Study
Perfusion CT was performed using a 256-slice CT scanner (Brilliant iCT; Philips Healthcare Systems, Netherlands). First, CT scanning was performed without intravenous contrast medium to localize the tumor, and the wholetumor sections were selected at the level of the tumor for cine imaging. A dynamic study of this area was performed with a Jog mode during rapid intravenous bolus injection (5 mL/second) of 50 mL iopromide containing 370 mg of iodine per milliliter (Ultravist 370; Bayer, Berlin, Germany). The following parameters were used: 0.33-seconds gantry rotation time, 100 kV, 80 mA, 7.6-seconds scanning delay from the start of injection, 3.8-seconds scanning interval, 60.8-seconds duration of transverse data acquisition, and 3-mm reconstructed section thickness. Imaging Analysis
The perfusion data were transferred to an image processing workstation (Extended Brilliant workshop 4.02; Philips Healthcare Systems) and then analyzed using software (CT Perfusion; Philips Healthcare Systems) based on the slope method. The parameters generated by the software were BF (in milliliters per minute per 100 g of wet tissue), BV (in milliliters per 100 g of wet tissue), and TTP (in seconds). To derive functional maps of these perfusion parameters, the arterial input curve of the contrast medium concentration was required, and we obtained this arterial input curve from a region of interest (ROI) in the external iliac artery or aorta. The ROI was drawn freehand (using an electronic cursor and mouse) around the peripheral boundary of the visible tumor. Care was taken to exclude pericolonic fat and intraluminal gas, which was facilitated by viewing a cine loop of acquisition to gauge the degree of patient movement and the tumor margins. We used a section-by-section averaging technique to evaluate whole-tumor perfusion. First, the ROI drawing was repeated for each contiguous transverse level of the entire tumor lesion. Then, a global value representing the perfusion of the entire tumor was calculated by taking the mean value of all individual sections involved (Figs 1 and 2). To assess the interobserver agreement, all cases were reevaluated after 3 months and results of the two sets of measurements were compared. Assessment of Tumor Grade, Immunohistochemical Staining, and Quantification of MVD
The surgical specimens were fixed with 10% formaldehyde. The differentiation grades were assessed by an experienced gastrointestinal pathologist and were divided into three subgroups: well, moderately, and poorly differentiated CRCs. To evaluate tumor angiogenesis, three portions of the tumor in the craniocaudal direction were selected as the locations for additional tissue on tumor specimens for further immunohistochemical staining. Tumor necrotic portions 751
SUN ET AL
Academic Radiology, Vol 21, No 6, June 2014
Figure 1. A 59-year-old man with well-differentiated colorectal cancer. Computed tomography series show reformatted contrast-enhanced images from the entire tumor at serial transverse levels in a sigmoid well-differentiated adenocarcinoma.
were avoided to be included in the additional tissue, if possible. Immunohistochemical analysis using an anti-CD105 antibody (titer 1:50, monoclonal; ZSGB-BIO, Beijing, China) was performed using the EnVision antibody complex method (12). Hot spot analysis was used as the criteria for microvessel counting, as established by Weidner (13). At low magnification, the area with high number of microvessels on the grounds had the high potential of tumor-derived angiogenic cell clones, and consequently, had the easiest way to the blood stream and an increased probability of tumor growth and angiogenesis (14). Any single brown-staining endothelial cell or small clusters of brownstaining endothelial cells, which were clearly separate from adjacent microvessels, tumor cells, and other connective tissue elements, with or without a lumen, were regarded as individual vessels. Vessels of a caliber larger than approximately eight red blood cells and vessels with a thick muscular wall were excluded from the final count. Slides were examined at low-power magnification (40) to identify the areas with the highest density of microvessels. In each case, the most vascularized area was selected and the microvessels within a high-power magnification (200) field of this area were 752
counted thrice in the three portions. The mean of the three highest counts per tumor was used in further analyses (Fig. 3). Statistical Analysis
Reproducibility between replicated measurements was evaluated using the Bland–Altman analysis (15). Intraclass correlation coefficients were calculated for each of the three perfusion parameters along with the mean differences, standard deviation of the differences, and 95% limits of agreement. The difference in the three perfusion parameters (BF, BV, and TTP) and MVD among the three tumor differentiation grades of CRC was examined using the Student–Newman–Keuls test. The diagnostic accuracy of perfusion parameters in different grades was calculated with the receiver-operating characteristic (ROC) analysis. The diagnosis of differentiation obtained by pathologic examination was used as the standard. Sensitivity, specificity, and accuracy for each significant parameter in the prediction of tumor grade was calculated using cutoff values chosen on the basis of ROC curves. The correlation between CT perfusion parameters and MVD was analyzed using the Pearson correlation analysis. All analyses
Academic Radiology, Vol 21, No 6, June 2014
WHOLE-VOLUME PERFUSION CT IN CRC
Figure 2. A 59-year-old man with well-differentiated colorectal cancer. Colored parametric maps show blood flow values at multiple transverse levels that encompass the entire tumor in a sigmoid well-differentiated adenocarcinoma. Each pixel location within the tumor region of interest corresponds to a single quantitative perfusion value. (Color version of figure is available online.)
were performed using the Statistical Package for the Social Sciences (SPSS, version 17; SPSS, Chicago, IL). A value of P < .05 was considered statistically significant. RESULTS Thirty-seven patients subsequently underwent surgery after perfusion CT scan. The mean interval between the preoperative CT scan and surgery was 17.1 days (range, 5–31 days). Good agreements were obtained between the replicated whole-volume perfusion measurements in terms of measuring three perfusion parameters in the patients (Table 1). The difference in the perfusion parameters among different tumor differentiation grades is summarized in Table 2. The mean BF was significantly different among different tumor differentiation grades (P < .05) with the highest BF in the well-differentiated group and the lowest BF in the poorly differentiated group. No statistically significant differences in BV values were found between the moderately and poorly differentiated groups (P = .806); however, BV values in these two groups were significantly lower than
that in the well-differentiated group (P = .002, P = .009). The mean TTP in the poorly differentiated group was significantly longer than that in moderately and well-differentiated groups (P = .016, P = .028), but no statistically significant differences were found between the latter two groups (P = .737). Because there was statistically significant difference among different grades in terms of BF values, diagnostic accuracy was assessed using ROC curves. The cutoff values for BF with the best performance in terms of sensitivity, specificity, and accuracy in the characterization of well-moderately or poorly differentiated, well or moderately-poorly differentiated tumors have been chosen. In the characterization of well-moderately or poorly differentiated tumors, BF cutoff value was set at 21.49 mL/minute/100 g with an area under the curve of 0.811. The sensitivity, specificity and accuracy were 92.6%, 60%, and 83.8%, respectively. In the characterization of well or moderately-poorly differentiated tumors, BF cutoff value was set at 61.28 mL/minute/100 g with an area under the curve of 0.919. The sensitivity, specificity, and accuracy were 71.4%, 100%, and 96.6%, respectively. There was no significant difference in MVD among different 753
SUN ET AL
Figure 3. A 59-year-old man with well-differentiated colorectal cancer. Increased CD105 immunostaining was observed. Microvessels were defined as single brown-staining endothelial cells with lumen or small clusters of brown-staining endothelial cells without lumen (arrow) (CD105 immunostaining at 200 magnification). Lumens of diameters greater than that of eight red blood cells (arrow head) were excluded from the analysis. (Color version of figure is available online.)
tumor differentiation grades (P > .05) (Table 2 and Fig. 4). There was no significant correlation between the three perfusion parameters and MVD in CRCs (Fig. 5).
DISCUSSION The importance of angiogenesis in tumor growth is well established (16). In clinical practice, assessment of angiogenesis has relied mainly on assessment of MVD through immunohistochemical staining of postoperative specimens. By using the dynamic data obtained from perfusion CT, physiological parameters including BF, BV, and TTP can be obtained to assess the hemodynamics of tumor vasculature. Moreover, previous studies have shown that dynamic contrast-enhanced CT measures correlate well with MVD in histopathologic assessments of CRC (7). Previously, one of the principle limitations of CT perfusion was often the restriction of the assessment to a single slice or a limited number of slices (7,9,17). The limited slice sequences are likely to underestimate tumor enhancement and lead to measurement deviation, especially in large tumors (18). With the development of multidetector-row CT imaging methods, perfusion imaging has been advanced from a single-slice technique to a volume-based examination. Ng et al. (18) assessed wholetumor perfusion using a CT technique with helical row of 16 detectors and achieved good reproducibility. Their data demonstrated that inter- and intraobserver variabilities were diminished, indicating the potential for reliable assessment with whole-tumor perfusion CT in lung cancer. Previous study showed the first-pass perfusion imaging based on the slope method could provide a feasible method for assessment 754
Academic Radiology, Vol 21, No 6, June 2014
of whole-tumor perfusion (19). Additionally, CD105 can stain a higher number of the proliferating vessels in colon carcinoma and is a more specific and sensitive marker for tumor angiogenesis than the commonly used pan-endothelial markers, such as CD31 or CD34 (20). CD105 staining also showed prognostic significance, positively correlating with angiolymphatic invasion and metastases to lymph nodes and liver (21). In our study, we used whole-tumor CT perfusion technique and CD105 staining of the specimen after surgery to obtain a more accurate assessment of angiogenesis. Our study showed that the mean BF increased with tumor grade in CRCs. The mean BF was significantly higher in well-differentiated CRCs than that in moderately and poorly differentiated CRCs. Among the three perfusion parameters (BF, BV, and TTP), BF is regarded to reflect the flow rate of the tumor vasculature and tumor grade. Generally, the better the tumor differentiation is, the greater the BF value is. Our findings are consistent with the findings of Hayano et al. (22). In addition, Dugdale et al. (23) reported a correlation between higher tumor BF and the grade of lymphoma with histopathologic examination. However, there are some conflicting results between previous reports and our study. Kim et al. (17) found that the BF in CRCs was increased in moderately differentiated CRCs compared with that in well-differentiated CRCs, and was decreased in poorly differentiated CRCs compared with that in moderately and welldifferentiated CRCs. There are some reasons that might have led to different results in different studies. First, the sample used in the previous study was limited (17). Second, they used the value of one level of the tumor BF, and this BF value was used to calculate the overall tumor. Third, software applications from different vendors do not seem to generate comparable quantitative perfusion results (24). We assume that the relatively mature vessels with a lumen but not the small clusters of endothelial cells without a lumen are major contributors to the BF value. The better the tumor differentiation is, the more the mature vessels are. Therefore, the BF value of the well-differentiated tumor is higher than that of the other two groups. Sahani et al. (25) reported that well-differentiated hepatocellular carcinomas had a higher BF value than moderately and poorly differentiated hepatocellular carcinomas, which is consistent with the results described in the present study. The TTP is determined as the time from the arrival of the contrast material in major arterial vessels to the peak tissue enhancement and is considered to be a marker of perfusion pressure (26). In poorly differentiated CRCs, the interstitial pressure is more than that in well and moderately differentiated CRCs. So the mean TTP was significantly shorter in the well-differentiated group than that in the other two groups. In cancer tumor angiogenesis quantification, MVD assessment is the most commonly used technique (5). MVD is currently applied as an indicator for tumor angiogenetic activity (13) and is widely used to quantify tumor angiogenesis. However, changes in MVD do not always correspond with the changes in tissue perfusion, because they are affected
Academic Radiology, Vol 21, No 6, June 2014
WHOLE-VOLUME PERFUSION CT IN CRC
TABLE 1. Reproducibility for BF, BV, and TTP Measurements
Perfusion Parameters BF (mL/minute/100 g) BV (mL/100 g) TTP(s)
Differences between Measurements (Mean SD) 0.07 0.36 0.01 0.66 0.32 0.63
95% CI
95% Limits of Agreement
0.23 to 0.10 0.29 to 0.32 0.03 to 0.61
0.77 to 0.64 1.28 to 1.30 0.91 to 1.55
ICC (95% CI) 0.9997 (0.9993–0.9999) 0.9939 (0.9845–0.9976) 0.9976 (0.9939–0.9990)
BF, blood flow; BV, blood volume; 95% CI, 95% confidence interval; ICC, interclass correlation coefficient; TTP, time to peak.
TABLE 2. BF, BV, and TTP Measurements for the Different Tumor Differentiation Grades
CT Perfusion Parameter BF (mL/minute/100 g) BV (mL/100 g) TTP (second) MVD
Well Differentiated (n = 7)
Moderately Differentiated (n = 20)
Poorly Differentiated (n = 10)
P Value
61.17 17.97 33.96 24.81 17.60 8.53 33.47 14.69
34.80 13.06 16.93 5.73 18.94 7.47 28.89 11.82
22.24 9.31 18.05 6.01 27.81 11.95 29.89 11.02
.05). The data shown are the Pearson correlation coefficient, r = 0.201, 0.295, and 0.178, respectively; P = .233, .076, and .292, respectively. BF, blood flow; BV, blood volume; MVD, microvessel density; TTP, time to peak.
756
The authors acknowledge Yingying Hu from Department of Radiology at our institute, for her support in recruiting and screening the patients; Dr. Aiping Song from Department of Pathology at our institute for her support in pathologic slices preparation. The authors also acknowledge Dr. Kyongtae T. Bae from Department of Radiology, University of Pittsburgh Medical Center, Pittsburgh, PA for reviewing the manuscript.
REFERENCES 1. Weitz J, Koch M, Debus J, et al. Colorectal cancer. Lancet 2005; 365: 153–165. 2. Compton CC, Fielding LP, Burgart LJ, et al. Prognostic factors in colorectal cancer. College of American Pathologists Consensus Statement 1999. Arch Pathol Lab Med 2000; 124:979–994.
Academic Radiology, Vol 21, No 6, June 2014
3. Fleming M, Ravula S, Tatishchev SF, et al. Colorectal carcinoma: pathologic aspects. J Gastrointest Oncol 2012; 3:153–173. 4. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996; 86:353–364. 5. Vermeulen P, Gasparini G, Fox S, et al. Quantification of angiogenesis in solid human tumours: an international consensus on the methodology and criteria of evaluation. Eur J Cancer 1996; 32:2474–2484. 6. Weidner N, Folkman J, Pozza F, et al. Tumor angiogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma. J Natl Cancer Inst 1992; 84:1875–1887. 7. Goh V, Halligan S, Daley F, et al. Colorectal tumor vascularity: quantitative assessment with multidetector CT—do tumor perfusion measurements reflect angiogenesis? Radiology 2008; 249:510–517. 8. Miles K. Measurement of tissue perfusion by dynamic computed tomography. Br J Radiol 1991; 64:409–412. 9. Li ZP, Meng QF, Sun CH, et al. Tumor angiogenesis and dynamic CT in colorectal carcinoma: radiologic-pathologic correlation. World J Gastroenterol 2005; 11:1287–1291. 10. Dighe S, Blake H, Jeyadevan N, et al. Perfusion CT vascular parameters do not correlate with immunohistochemically derived microvessel density count in colorectal tumors. Radiology 2013; 268:400–410. 11. Romani AA, Borghetti AF, Del Rio P, et al. The risk of developing metastatic disease in colorectal cancer is related to CD105-positive vessel count. J Surg Oncol 2006; 93:446–455. €mmerer U, Kapp M, Gassel AM, et al. A new rapid immunohistochem12. Ka ical staining technique using the EnVision antibody complex. J Histochem Cytochem 2001; 49:623–630. 13. Weidner N. Intratumor microvessel density as a prognostic factor in cancer. Am J Pathol 1995; 147:9–19. 14. Vermeulen P, Gasparini G, Fox S, et al. Second international consensus on the methodology and criteria of evaluation of angiogenesis quantification in solid human tumours. Eur J Cancer 2002; 38:1564–1579. 15. Bland MJ, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 327:307–310.
WHOLE-VOLUME PERFUSION CT IN CRC
16. George M, Dzik-Jurasz A, Padhani A, et al. Non-invasive methods of assessing angiogenesis and their value in predicting response to treatment in colorectal cancer. Br J Surg 2001; 88:1628–1636. 17. Kim JW, Jeong YY, Chang NK, et al. Perfusion CT in colorectal cancer: comparison of perfusion parameters with tumor grade and microvessel density. Korean J Radiol 2012; 13(Suppl 1):S89–S97. 18. Ng QS, Goh V, Fichte H, et al. Lung cancer perfusion at multi–detector row CT: reproducibility of whole tumor quantitative measurements. Radiology 2006; 239:547–553. 19. Li Y, Yang ZG, Chen TW, et al. Whole tumour perfusion of peripheral lung carcinoma: evaluation with first-pass CT perfusion imaging at 64-detector row CT. Clin Radiol 2008; 63:629–635. 20. Jung I, Gurzu S, Raica M, et al. The differences between the endothelial area marked with CD31 and CD105 in colorectal carcinomas by computer-assisted morphometrical analysis. Rom J Morphol Embryol 2009; 50:239–243. 21. Saad RS, Liu YL, Nathan G, et al. Endoglin (CD105) and vascular endothelial growth factor as prognostic markers in colorectal cancer. Mod Pathol 2003; 17:197–203. 22. Hayano K, Shuto K, Koda K, et al. Quantitative measurement of blood flow using perfusion CT for assessing clinicopathologic features and prognosis in patients with rectal cancer. Dis Colon Rectum 2009; 52:1624–1629. 23. Dugdale PE, Miles KA, Bunce I, et al. CT measurement of perfusion and permeability within lymphoma masses and its ability to assess grade, activity, and chemotherapeutic response. J Comput Assist Tomogr 1999; 23:540–547. 24. Goh V, Halligan S, Bartram CI. Quantitative tumor perfusion assessment with multidetector CT: are measurements from two commercial software packages interchangeable? Radiology 2007; 242:777–782. 25. Sahani DV, Holalkere NS, Mueller PR, et al. Advanced hepatocellular carcinoma: CT perfusion of liver and tumor tissue—initial experience. Radiology 2007; 243:736–743. 26. D’Onofrio M, Gallotti A, Mantovani W, et al. Perfusion CT can predict tumoral grading of pancreatic adenocarcinoma. Eur J Radiol 2013; 82: 227–233.
757
C
olorectal cancer (CRC) is the third most common cancer and the fourth most frequent cause of cancer deaths worldwide (1). The 5-year survival rate depends on the tumor stage and grade at patient presentation. Tumors with an advanced stage and grade at diagnosis are associated with a poor outcome. Individual treatment strategy based on tumor stage and grade should be applied to improve the prognosis. Thus, the preoperative diagnostic evaluation and grading of CRC are important (2). Preoperative specimens from endoscopic colorectal biopsies are often used but are normally failed to grade tumor because of the
Acad Radiol 2014; 21:750–757 From the Department of Radiology, China-Japan Friendship Hospital, No. 2 Yinghua East Street, Chaoyang District, Beijing 100029, China (H.S., Y.X., W.W.) and Department of Pathology, China-Japan Friendship Hospital, Chaoyang District, Beijing 100029, China (Q.Y.). Received October 9, 2013; accepted February 10, 2014. Disclosure: The authors have nothing to disclose. Address correspondence to: W.W. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.02.011
750
lack of sufficient tissue (3). Angiogenesis, which is important in the growth and metastasis of carcinomas, has been reported to be a promising prognostic marker for the CRC (4). Microvessel density (MVD) count is used to define the degree of angiogenesis in solid tumors for diagnostic purpose and treatment planning, which is calculated by counting the number of angiogenic blood vessels highlighted on a variety of immunohistochemical stains (5,6). However, information pertaining to the MVD can only be gathered in the in vitro setting after the resection of the tumor, and therefore there is no opportunity to evaluate the effect of neoadjuvant radiochemotherapy and antiangiogenic therapy on this parameter. Perfusion computed tomography (CT) can quantify tumor angiogenesis noninvasively by assessing the enhancement of the tissue and vessels over time. Perfusion parameters, including tissue blood flow (BF), blood volume (BV), time to peak (TTP), and permeability–surface area product (PS), are calculated using the mathematical models for contrast agent exchange (7,8). Goh et al. (7) reported
Academic Radiology, Vol 21, No 6, June 2014
the relationship between CT perfusion parameters and MVD counts and their data indicated a positive correlation between tumor PS and BV with MVD in CRC, inconsistent with the report by Li et al. (9), which showed no significant correlation between any perfusion parameters with MVD. Both the studies were based on the deconvolution approach, but the scanning equipments, scanning modes, analytical software were different, which might be the potential explanation for such differences and make the standardization of the technique difficult (10). Romani et al. (11) reported that CD105 (endoglin)-staining intensities in CRCs were correlated with the MVD levels and were better indicators of the state of tumor angiogenesis. Previous CRC CT perfusion studies usually use one slice or a few slices of the tumor to represent the overall tumor angiogenesis state, and to some degree, this sampling rate is not sufficient because of the heterogeneity of tumor angiogenesis. Additionally, the lesion on CT images is not exactly the same as the pathologic specimen in the orientation, shape, and size, which makes it difficult to achieve precise alignment. Therefore, we analyzed primary CRC using whole-volume perfusion CT measurements and assessed whether perfusion CT could be used to evaluate the pathologic grade and the correlation between perfusion parameters and MVD stained with CD105 in CRC. MATERIALS AND METHODS Patients
The institutional review board approved this study, and informed consent was obtained from all patients enrolled in the study. Between August 2010 and March 2012, 42 consecutive patients with CRC who underwent preoperative CT scans were prospectively enrolled in the study. Patients were excluded according to the following criteria: (1) preoperative treatment such as chemotherapy or radiotherapy (n = 2); (2) contraindication to administration of contrast medium (n = 0); (3) severe diseases of the heart, liver, lung, or kidney (n = 0); and (4) descending or ascending colon cancers with severe motion artifacts during the perfusion CT examination (n = 3). A total of 37 patients with CRC who underwent surgery after perfusion CTat our institute were enrolled in this study. The subjects included 20 men and 17 women, with a mean age of 64.5 years (range, 38–85 years). The diagnosis of adenocarcinoma of the colon and rectum was histologically confirmed by colonoscopic biopsy and surgical specimens. Mean tumor length at pathologic evaluation was 9.5 cm (range, 5.2–13.8 cm). The tumors examined were located in the cecum (n = 5), ascending colon (n = 3), descending colon (n = 3), sigmoid colon (n = 4), and rectum (n = 22). Tumors were divided into three histologic subgroups: well differentiated (7 tumors), moderately differentiated (20 tumors), and poorly differentiated (10 tumors). Thirty-seven patients underwent curative surgery without severe complications.
WHOLE-VOLUME PERFUSION CT IN CRC
Imaging Study
Perfusion CT was performed using a 256-slice CT scanner (Brilliant iCT; Philips Healthcare Systems, Netherlands). First, CT scanning was performed without intravenous contrast medium to localize the tumor, and the wholetumor sections were selected at the level of the tumor for cine imaging. A dynamic study of this area was performed with a Jog mode during rapid intravenous bolus injection (5 mL/second) of 50 mL iopromide containing 370 mg of iodine per milliliter (Ultravist 370; Bayer, Berlin, Germany). The following parameters were used: 0.33-seconds gantry rotation time, 100 kV, 80 mA, 7.6-seconds scanning delay from the start of injection, 3.8-seconds scanning interval, 60.8-seconds duration of transverse data acquisition, and 3-mm reconstructed section thickness. Imaging Analysis
The perfusion data were transferred to an image processing workstation (Extended Brilliant workshop 4.02; Philips Healthcare Systems) and then analyzed using software (CT Perfusion; Philips Healthcare Systems) based on the slope method. The parameters generated by the software were BF (in milliliters per minute per 100 g of wet tissue), BV (in milliliters per 100 g of wet tissue), and TTP (in seconds). To derive functional maps of these perfusion parameters, the arterial input curve of the contrast medium concentration was required, and we obtained this arterial input curve from a region of interest (ROI) in the external iliac artery or aorta. The ROI was drawn freehand (using an electronic cursor and mouse) around the peripheral boundary of the visible tumor. Care was taken to exclude pericolonic fat and intraluminal gas, which was facilitated by viewing a cine loop of acquisition to gauge the degree of patient movement and the tumor margins. We used a section-by-section averaging technique to evaluate whole-tumor perfusion. First, the ROI drawing was repeated for each contiguous transverse level of the entire tumor lesion. Then, a global value representing the perfusion of the entire tumor was calculated by taking the mean value of all individual sections involved (Figs 1 and 2). To assess the interobserver agreement, all cases were reevaluated after 3 months and results of the two sets of measurements were compared. Assessment of Tumor Grade, Immunohistochemical Staining, and Quantification of MVD
The surgical specimens were fixed with 10% formaldehyde. The differentiation grades were assessed by an experienced gastrointestinal pathologist and were divided into three subgroups: well, moderately, and poorly differentiated CRCs. To evaluate tumor angiogenesis, three portions of the tumor in the craniocaudal direction were selected as the locations for additional tissue on tumor specimens for further immunohistochemical staining. Tumor necrotic portions 751
SUN ET AL
Academic Radiology, Vol 21, No 6, June 2014
Figure 1. A 59-year-old man with well-differentiated colorectal cancer. Computed tomography series show reformatted contrast-enhanced images from the entire tumor at serial transverse levels in a sigmoid well-differentiated adenocarcinoma.
were avoided to be included in the additional tissue, if possible. Immunohistochemical analysis using an anti-CD105 antibody (titer 1:50, monoclonal; ZSGB-BIO, Beijing, China) was performed using the EnVision antibody complex method (12). Hot spot analysis was used as the criteria for microvessel counting, as established by Weidner (13). At low magnification, the area with high number of microvessels on the grounds had the high potential of tumor-derived angiogenic cell clones, and consequently, had the easiest way to the blood stream and an increased probability of tumor growth and angiogenesis (14). Any single brown-staining endothelial cell or small clusters of brownstaining endothelial cells, which were clearly separate from adjacent microvessels, tumor cells, and other connective tissue elements, with or without a lumen, were regarded as individual vessels. Vessels of a caliber larger than approximately eight red blood cells and vessels with a thick muscular wall were excluded from the final count. Slides were examined at low-power magnification (40) to identify the areas with the highest density of microvessels. In each case, the most vascularized area was selected and the microvessels within a high-power magnification (200) field of this area were 752
counted thrice in the three portions. The mean of the three highest counts per tumor was used in further analyses (Fig. 3). Statistical Analysis
Reproducibility between replicated measurements was evaluated using the Bland–Altman analysis (15). Intraclass correlation coefficients were calculated for each of the three perfusion parameters along with the mean differences, standard deviation of the differences, and 95% limits of agreement. The difference in the three perfusion parameters (BF, BV, and TTP) and MVD among the three tumor differentiation grades of CRC was examined using the Student–Newman–Keuls test. The diagnostic accuracy of perfusion parameters in different grades was calculated with the receiver-operating characteristic (ROC) analysis. The diagnosis of differentiation obtained by pathologic examination was used as the standard. Sensitivity, specificity, and accuracy for each significant parameter in the prediction of tumor grade was calculated using cutoff values chosen on the basis of ROC curves. The correlation between CT perfusion parameters and MVD was analyzed using the Pearson correlation analysis. All analyses
Academic Radiology, Vol 21, No 6, June 2014
WHOLE-VOLUME PERFUSION CT IN CRC
Figure 2. A 59-year-old man with well-differentiated colorectal cancer. Colored parametric maps show blood flow values at multiple transverse levels that encompass the entire tumor in a sigmoid well-differentiated adenocarcinoma. Each pixel location within the tumor region of interest corresponds to a single quantitative perfusion value. (Color version of figure is available online.)
were performed using the Statistical Package for the Social Sciences (SPSS, version 17; SPSS, Chicago, IL). A value of P < .05 was considered statistically significant. RESULTS Thirty-seven patients subsequently underwent surgery after perfusion CT scan. The mean interval between the preoperative CT scan and surgery was 17.1 days (range, 5–31 days). Good agreements were obtained between the replicated whole-volume perfusion measurements in terms of measuring three perfusion parameters in the patients (Table 1). The difference in the perfusion parameters among different tumor differentiation grades is summarized in Table 2. The mean BF was significantly different among different tumor differentiation grades (P < .05) with the highest BF in the well-differentiated group and the lowest BF in the poorly differentiated group. No statistically significant differences in BV values were found between the moderately and poorly differentiated groups (P = .806); however, BV values in these two groups were significantly lower than
that in the well-differentiated group (P = .002, P = .009). The mean TTP in the poorly differentiated group was significantly longer than that in moderately and well-differentiated groups (P = .016, P = .028), but no statistically significant differences were found between the latter two groups (P = .737). Because there was statistically significant difference among different grades in terms of BF values, diagnostic accuracy was assessed using ROC curves. The cutoff values for BF with the best performance in terms of sensitivity, specificity, and accuracy in the characterization of well-moderately or poorly differentiated, well or moderately-poorly differentiated tumors have been chosen. In the characterization of well-moderately or poorly differentiated tumors, BF cutoff value was set at 21.49 mL/minute/100 g with an area under the curve of 0.811. The sensitivity, specificity and accuracy were 92.6%, 60%, and 83.8%, respectively. In the characterization of well or moderately-poorly differentiated tumors, BF cutoff value was set at 61.28 mL/minute/100 g with an area under the curve of 0.919. The sensitivity, specificity, and accuracy were 71.4%, 100%, and 96.6%, respectively. There was no significant difference in MVD among different 753
SUN ET AL
Figure 3. A 59-year-old man with well-differentiated colorectal cancer. Increased CD105 immunostaining was observed. Microvessels were defined as single brown-staining endothelial cells with lumen or small clusters of brown-staining endothelial cells without lumen (arrow) (CD105 immunostaining at 200 magnification). Lumens of diameters greater than that of eight red blood cells (arrow head) were excluded from the analysis. (Color version of figure is available online.)
tumor differentiation grades (P > .05) (Table 2 and Fig. 4). There was no significant correlation between the three perfusion parameters and MVD in CRCs (Fig. 5).
DISCUSSION The importance of angiogenesis in tumor growth is well established (16). In clinical practice, assessment of angiogenesis has relied mainly on assessment of MVD through immunohistochemical staining of postoperative specimens. By using the dynamic data obtained from perfusion CT, physiological parameters including BF, BV, and TTP can be obtained to assess the hemodynamics of tumor vasculature. Moreover, previous studies have shown that dynamic contrast-enhanced CT measures correlate well with MVD in histopathologic assessments of CRC (7). Previously, one of the principle limitations of CT perfusion was often the restriction of the assessment to a single slice or a limited number of slices (7,9,17). The limited slice sequences are likely to underestimate tumor enhancement and lead to measurement deviation, especially in large tumors (18). With the development of multidetector-row CT imaging methods, perfusion imaging has been advanced from a single-slice technique to a volume-based examination. Ng et al. (18) assessed wholetumor perfusion using a CT technique with helical row of 16 detectors and achieved good reproducibility. Their data demonstrated that inter- and intraobserver variabilities were diminished, indicating the potential for reliable assessment with whole-tumor perfusion CT in lung cancer. Previous study showed the first-pass perfusion imaging based on the slope method could provide a feasible method for assessment 754
Academic Radiology, Vol 21, No 6, June 2014
of whole-tumor perfusion (19). Additionally, CD105 can stain a higher number of the proliferating vessels in colon carcinoma and is a more specific and sensitive marker for tumor angiogenesis than the commonly used pan-endothelial markers, such as CD31 or CD34 (20). CD105 staining also showed prognostic significance, positively correlating with angiolymphatic invasion and metastases to lymph nodes and liver (21). In our study, we used whole-tumor CT perfusion technique and CD105 staining of the specimen after surgery to obtain a more accurate assessment of angiogenesis. Our study showed that the mean BF increased with tumor grade in CRCs. The mean BF was significantly higher in well-differentiated CRCs than that in moderately and poorly differentiated CRCs. Among the three perfusion parameters (BF, BV, and TTP), BF is regarded to reflect the flow rate of the tumor vasculature and tumor grade. Generally, the better the tumor differentiation is, the greater the BF value is. Our findings are consistent with the findings of Hayano et al. (22). In addition, Dugdale et al. (23) reported a correlation between higher tumor BF and the grade of lymphoma with histopathologic examination. However, there are some conflicting results between previous reports and our study. Kim et al. (17) found that the BF in CRCs was increased in moderately differentiated CRCs compared with that in well-differentiated CRCs, and was decreased in poorly differentiated CRCs compared with that in moderately and welldifferentiated CRCs. There are some reasons that might have led to different results in different studies. First, the sample used in the previous study was limited (17). Second, they used the value of one level of the tumor BF, and this BF value was used to calculate the overall tumor. Third, software applications from different vendors do not seem to generate comparable quantitative perfusion results (24). We assume that the relatively mature vessels with a lumen but not the small clusters of endothelial cells without a lumen are major contributors to the BF value. The better the tumor differentiation is, the more the mature vessels are. Therefore, the BF value of the well-differentiated tumor is higher than that of the other two groups. Sahani et al. (25) reported that well-differentiated hepatocellular carcinomas had a higher BF value than moderately and poorly differentiated hepatocellular carcinomas, which is consistent with the results described in the present study. The TTP is determined as the time from the arrival of the contrast material in major arterial vessels to the peak tissue enhancement and is considered to be a marker of perfusion pressure (26). In poorly differentiated CRCs, the interstitial pressure is more than that in well and moderately differentiated CRCs. So the mean TTP was significantly shorter in the well-differentiated group than that in the other two groups. In cancer tumor angiogenesis quantification, MVD assessment is the most commonly used technique (5). MVD is currently applied as an indicator for tumor angiogenetic activity (13) and is widely used to quantify tumor angiogenesis. However, changes in MVD do not always correspond with the changes in tissue perfusion, because they are affected
Academic Radiology, Vol 21, No 6, June 2014
WHOLE-VOLUME PERFUSION CT IN CRC
TABLE 1. Reproducibility for BF, BV, and TTP Measurements
Perfusion Parameters BF (mL/minute/100 g) BV (mL/100 g) TTP(s)
Differences between Measurements (Mean SD) 0.07 0.36 0.01 0.66 0.32 0.63
95% CI
95% Limits of Agreement
0.23 to 0.10 0.29 to 0.32 0.03 to 0.61
0.77 to 0.64 1.28 to 1.30 0.91 to 1.55
ICC (95% CI) 0.9997 (0.9993–0.9999) 0.9939 (0.9845–0.9976) 0.9976 (0.9939–0.9990)
BF, blood flow; BV, blood volume; 95% CI, 95% confidence interval; ICC, interclass correlation coefficient; TTP, time to peak.
TABLE 2. BF, BV, and TTP Measurements for the Different Tumor Differentiation Grades
CT Perfusion Parameter BF (mL/minute/100 g) BV (mL/100 g) TTP (second) MVD
Well Differentiated (n = 7)
Moderately Differentiated (n = 20)
Poorly Differentiated (n = 10)
P Value
61.17 17.97 33.96 24.81 17.60 8.53 33.47 14.69
34.80 13.06 16.93 5.73 18.94 7.47 28.89 11.82
22.24 9.31 18.05 6.01 27.81 11.95 29.89 11.02
.05). The data shown are the Pearson correlation coefficient, r = 0.201, 0.295, and 0.178, respectively; P = .233, .076, and .292, respectively. BF, blood flow; BV, blood volume; MVD, microvessel density; TTP, time to peak.
756
The authors acknowledge Yingying Hu from Department of Radiology at our institute, for her support in recruiting and screening the patients; Dr. Aiping Song from Department of Pathology at our institute for her support in pathologic slices preparation. The authors also acknowledge Dr. Kyongtae T. Bae from Department of Radiology, University of Pittsburgh Medical Center, Pittsburgh, PA for reviewing the manuscript.
REFERENCES 1. Weitz J, Koch M, Debus J, et al. Colorectal cancer. Lancet 2005; 365: 153–165. 2. Compton CC, Fielding LP, Burgart LJ, et al. Prognostic factors in colorectal cancer. College of American Pathologists Consensus Statement 1999. Arch Pathol Lab Med 2000; 124:979–994.
Academic Radiology, Vol 21, No 6, June 2014
3. Fleming M, Ravula S, Tatishchev SF, et al. Colorectal carcinoma: pathologic aspects. J Gastrointest Oncol 2012; 3:153–173. 4. Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996; 86:353–364. 5. Vermeulen P, Gasparini G, Fox S, et al. Quantification of angiogenesis in solid human tumours: an international consensus on the methodology and criteria of evaluation. Eur J Cancer 1996; 32:2474–2484. 6. Weidner N, Folkman J, Pozza F, et al. Tumor angiogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma. J Natl Cancer Inst 1992; 84:1875–1887. 7. Goh V, Halligan S, Daley F, et al. Colorectal tumor vascularity: quantitative assessment with multidetector CT—do tumor perfusion measurements reflect angiogenesis? Radiology 2008; 249:510–517. 8. Miles K. Measurement of tissue perfusion by dynamic computed tomography. Br J Radiol 1991; 64:409–412. 9. Li ZP, Meng QF, Sun CH, et al. Tumor angiogenesis and dynamic CT in colorectal carcinoma: radiologic-pathologic correlation. World J Gastroenterol 2005; 11:1287–1291. 10. Dighe S, Blake H, Jeyadevan N, et al. Perfusion CT vascular parameters do not correlate with immunohistochemically derived microvessel density count in colorectal tumors. Radiology 2013; 268:400–410. 11. Romani AA, Borghetti AF, Del Rio P, et al. The risk of developing metastatic disease in colorectal cancer is related to CD105-positive vessel count. J Surg Oncol 2006; 93:446–455. €mmerer U, Kapp M, Gassel AM, et al. A new rapid immunohistochem12. Ka ical staining technique using the EnVision antibody complex. J Histochem Cytochem 2001; 49:623–630. 13. Weidner N. Intratumor microvessel density as a prognostic factor in cancer. Am J Pathol 1995; 147:9–19. 14. Vermeulen P, Gasparini G, Fox S, et al. Second international consensus on the methodology and criteria of evaluation of angiogenesis quantification in solid human tumours. Eur J Cancer 2002; 38:1564–1579. 15. Bland MJ, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 327:307–310.
WHOLE-VOLUME PERFUSION CT IN CRC
16. George M, Dzik-Jurasz A, Padhani A, et al. Non-invasive methods of assessing angiogenesis and their value in predicting response to treatment in colorectal cancer. Br J Surg 2001; 88:1628–1636. 17. Kim JW, Jeong YY, Chang NK, et al. Perfusion CT in colorectal cancer: comparison of perfusion parameters with tumor grade and microvessel density. Korean J Radiol 2012; 13(Suppl 1):S89–S97. 18. Ng QS, Goh V, Fichte H, et al. Lung cancer perfusion at multi–detector row CT: reproducibility of whole tumor quantitative measurements. Radiology 2006; 239:547–553. 19. Li Y, Yang ZG, Chen TW, et al. Whole tumour perfusion of peripheral lung carcinoma: evaluation with first-pass CT perfusion imaging at 64-detector row CT. Clin Radiol 2008; 63:629–635. 20. Jung I, Gurzu S, Raica M, et al. The differences between the endothelial area marked with CD31 and CD105 in colorectal carcinomas by computer-assisted morphometrical analysis. Rom J Morphol Embryol 2009; 50:239–243. 21. Saad RS, Liu YL, Nathan G, et al. Endoglin (CD105) and vascular endothelial growth factor as prognostic markers in colorectal cancer. Mod Pathol 2003; 17:197–203. 22. Hayano K, Shuto K, Koda K, et al. Quantitative measurement of blood flow using perfusion CT for assessing clinicopathologic features and prognosis in patients with rectal cancer. Dis Colon Rectum 2009; 52:1624–1629. 23. Dugdale PE, Miles KA, Bunce I, et al. CT measurement of perfusion and permeability within lymphoma masses and its ability to assess grade, activity, and chemotherapeutic response. J Comput Assist Tomogr 1999; 23:540–547. 24. Goh V, Halligan S, Bartram CI. Quantitative tumor perfusion assessment with multidetector CT: are measurements from two commercial software packages interchangeable? Radiology 2007; 242:777–782. 25. Sahani DV, Holalkere NS, Mueller PR, et al. Advanced hepatocellular carcinoma: CT perfusion of liver and tumor tissue—initial experience. Radiology 2007; 243:736–743. 26. D’Onofrio M, Gallotti A, Mantovani W, et al. Perfusion CT can predict tumoral grading of pancreatic adenocarcinoma. Eur J Radiol 2013; 82: 227–233.
757
