Tumor Vascular Micro-environment of Colorectal Hepatic Metastasis and Chemotherapy Response

Background: Tumor vascular micro-environment has an important role in tumor progression and metastasis. The objective of this study was to assess the signi ﬁ cance of metastatic hepatic tumor vascular micro-environment in relation to the response to the systemic 5-FU-based chemotherapy (FOLFOX or FOLFIRI). Methods: A total of 48 consecutive colorectal cancer (CRC) patients with hepatic metastasis were retrospectively reviewed, and factors, such as metastatic tumor vascular micro-environment, chemotherapy response, and hepatic resection, were analyzed. Tumor angiogenesis was microscopically evaluated by microvessel density (MVD) in sections stained immunochemically with CD34 in patients with hepatic resection. The angiogenesis in tumor micro-environment in association with ring enhancement (RE) on computed tomography (CT) was also examined. Results: Microscopic examination revealed that peripheral RE on CT of the metastatic tumor is associated with tumor angiogenesis by MVD. The overall response rate after 6 courses of rst-line chemotherapy for the liver metastasis with RE on CT was 64% (23/36), whereas the response rate without RE was 25% (3/12), which was signicantly different, although the survival rate of patients with RE-positive or without RE-negative tumors was not different. Conclusion: The peripheral RE on CT of the metastatic hepatic tumor was associated with the angiogenesis in tumor micro-environment and higher chemotherapy response.


Introduction
Colorectal carcinoma (CRC) is one of the most common cancers. Approximately 50% of patients develop liver metastases at some point during their disease course1-3). Patients who are candidates for surgical resection of their liver metastases can expect a prolonged survival or even cure4,5). However, only 10 to 25% of patients are candidates for liver resection6,7). In patients with unresectable metastases, chemotherapy is the treatment of choice, and although it is often used with palliative intent, it may also be used in an attempt to render the metastases resectable8, 9). Chemotherapy can also be administered as a neoadjuvant treatment for selected cases of resectable metastases10). Thus, most patients receive chemotherapy.
Tumor vascular micro-environment as angiogenesis is an essential process in many physiological and pathological conditions, including embryonic development, organ regeneration, chronic in ammation, and tumor growth and metastasis11,12). Metastatic liver tumors express angiogenic factors, such as vascular endothelial growth factor (VEGF), which is a well-known angiogenic growth factor. The amount of this angiogenic factor affects patient survival with metastatic disease13). In tumor micro-environment, newly formed angiogenic vessels supply the metastatic tumor with nutritional blood and growth factors, which leads to tumor proliferation, growth, and survival. Loading [MathJax]/jax/output/CommonHTML/jax.js Imaging, such as computed tomography (CT), is generally used to monitor chemotherapy according to the new guidelines for response evaluation (RECIST criteria) 14). Complete response is usually de ned as the disappearance of target lesions on imaging and is considered a good indicator to evaluate the e cacy of chemotherapy. The correlation between imaging and pathological status is not well de ned.
Whether radiological features, including complete radiological response, are correlated with pathological features, including complete pathological response, is important for the management of patients.
We examined the signi cance of tumor peripheral ring enhancement (RE) on contrast CT in hepatic metastasis from CRC regarding the response to rst-line systemic chemotherapy. Furthermore, the correlation between the radiological and pathological status, including the angiogenesis in tumor microenvironment, in patients who underwent resection of the site of the initial liver metastasis was investigated. No patients with synchronous liver metastasis simultaneously underwent resection of both the primary and secondary metastatic lesions. All patients rst underwent operation of the primary tumor. Open colectomy was performed in 29 patients and anterior resection of the rectum was performed in 16 patients, including 2 cases of transient colostomy because of cancer ileus. Three patients underwent Mile's operation. After the initial operation, adjuvant systemic chemotherapy following the 5-FU-based regimen, as described in the systemic chemotherapy section, was started within one month in cases of synchronous metastatic disease. In patients without con rmed hepatic metastasis, prophylactic systemic chemotherapy was not performed. Twelve cases of metachronous hepatic metastasis were found during the follow-up period. After con rmation of newly formed hepatic metastasis, adjuvant systemic chemotherapy was started. Adjuvant chemotherapy was performed for all patients with hepatic metastasis as systemic disease to prevent surrounding invasion or distant metastasis and to increase their resectability. Hepatic resection was conducted excluding inoperable multiple bi-lobular deposits of the liver and/or systemic disease such as lung metastasis or peritoneal dissemination.
Second-line chemotherapy was used in cases of disease progression after rst-line chemotherapy. A radiological response to systemic chemotherapy was assessed according to RECIST criteria 14).

Imaging
Triple-phase helical CT with 5-mm reconstruction (TOSHIBA TSX-101A., TOSHIBA, Tokyo, Japan) and abdominal ultrasound (TOSHIBA Nemio SSA550A, Xario SSA660A, TOSHIBA, Tokyo, Japan) was performed for all patients before chemotherapy and after every six cycles of chemotherapy. Scans of the liver were acquired with 16x0.75-mm collimation and pitch of 1.356, and were subsequently reconstructed at 3-mm intervals. Settings were 200 mA and 120 kV. Nonionic contrast medium (Omnipaque; Daiichi Pharmaceutical Co, Ltd, Tokyo, Japan) was injected at a rate of 3 mL/s (2 mL/kg) by an MCT power injector. Arterial-phase abdominal images were obtained 35 seconds after injection and portal-phase images were obtained at 80 seconds. Although colorectal liver metastasis was visualized better on the portal venous phase, tri-phasic CT was performed systematically for the evaluation of colorectal metastasis to improve the CT readings. All CT images were reviewed independently by radiologists. The CT images were compared with previous CT images. Abdominal angiography was also performed via a femoral approach by radiologists for some patients 17).
For quantitative analysis of CT images, regions of interest (ROI) were selected on the basis of the best tumor image on the portal phase of contrast-enhanced CT. Tumor peripheral RE was assessed as follows: the ROI in the peripheral RE area was measured by at least ve independent enhanced spot areas as a CTvalue (Houns eld Unit: HFU). The CT-value of normal liver parenchyma was also measured at the area nearby the tumor without tumor-altered vascularity. The mean CT-value was used, and the difference in the CT-value between the RE area and the normal liver parenchyma was de ned as the RE CT-value (Fig. 3A). The cut-off value between RE positive or negative was de ned as 5HFU.

Tumor Vascular Micro-environment by Pathological and Immunohistochemical Examination
Resected specimens of liver samples were xed in 10% formalin and embedded in para n. Thin sections were depara nized twice with xylene and rehydrated in a series of ethanol solutions. Sections were placed in 0.01 mol/L trisodium citrate dehydrate buffer (pH 6.0) and treated in a microwave oven for 10 min at 500 W.
For CD34 staining, tissue sections were digested with 0.2% trypsin in 0.01 mol/L phosphate-buffer saline for 20 min at 37°C. The tissues were immersed in 3% H 2 O 2 with distilled water for 10 min to inactivate endogenous peroxidases. After blocking nonspeci c binding by normal goat serum, sections were incubated overnight at 4°C with mouse anti-monoclonal CD34 antibody (1:25; QB-END/10, Novo-castra Laboratories, Newcastle, United Kingdom) as the primary antibody. This was followed by reacting with biotinylated anti-immunoglobulin and labeling using streptavidin-biotin reaction kit peroxidase (Dako, Carpinteria, CA). The peroxidase reaction was visualized with 0.01% H 2 O 2 and 3, 3'-diaminobenzidine Loading [MathJax]/jax/output/CommonHTML/jax.js under light microscopy (200X magni cation). Microvessel density (MVD) was used to evaluate the microscopic tumor angiogenesis in colorectal liver metastases18). For MVD after CD34 staining, the average was calculated in the ve most peri-tumor vascular areas in the 14 metastatic liver cancer lesions examined at 200X magni cation (Fig. 2).

Patient Management and Follow-Up
Preoperative systemic chemotherapy was continued postoperatively for six to eight cycles except in cases of grade 3 or 4 toxicity. Patients were followed up every 3-4 months during the rst 2 years and every 6 months thereafter. At each follow-up visit, tumor recurrence was assessed by clinical examination and liver ultrasound. Abdominal and chest CT was performed every 3-6 months. All surviving patients were followed for a minimum of 12 months after surgery.

Statistics
Quantitative data were expressed as the mean and standard deviation. Quantitative and qualitative variables were compared using the Fisher's exact test or the Mann-Whitney U test as appropriate. Overall survivals in both metachronous and synchronous cases were de ned as the period from the starting day of systemic chemotherapy at the development of metastatic hepatic tumors to the day of death from any cause. The survival rate was calculated using the Kaplan-Meier method and the log-rank test was used to assess the survival differences between groups. Signi cance was de ned by a P-value < 0.05.

Patients and Tumor Characteristics
Clinical and tumor characteristics of 48 patients with liver metastases are shown in Table 2. Of these 48 patients, 36 had synchronous metastatic disease and 12 had metachronous disease. The mean age for patients at the time of diagnosis was 66.2 ± 12 years (range 38-82 years). Primary cancers included colon carcinomas in 29 patients and rectal carcinomas in 19 patients. TNM classi cation19) was as follows: T1-T2 in 2 and T3-T4 in 46. There was lymph node involvement in 38 patients. Pathological diagnoses included well-differentiated adenocarcinoma in 27, moderate differentiated adenocarcinoma in 20, and mucinous adenocarcinoma in 1. The mean number of liver metastases was 4.8 (range 1-12). Distant metastases excluding hepatic metastases were observed in 11 patients with lung metastasis on preoperative CT examination. Peritoneal dissemination was con rmed in 7 patients by surgical exploration.
Thirty-six patients had synchronous hepatic metastasis. Thirty-one synchronous cases were initially treated by resection of the primary lesion followed by systemic chemotherapy, and 6 patients eventually underwent liver resection. The other 30 patients were not considered for resection of the liver metastases for the following reasons: inoperable multiple bi-lobular deposits of the liver and/or systemic disease such as lung metastasis or peritoneal dissemination. Two patients initially underwent diversion Loading [MathJax]/jax/output/CommonHTML/jax.js colostomy for colon obstruction. After con rmation of a good chemotherapy response, the primary lesion was removed.
Twelve of 48 metachronous patients had hepatic metastasis. Metachronous hepatic metastasis was de ned by a hepatic tumor found longer than 12 months after the initial treatment. There were 8 resectable metastatic liver tumors and 4 inoperable cases. Hepatic resection was eventually performed in 14 patients, including 42 lesions. The operative procedures included hemihepatectomy (n = 2), segmentectomy or sectionectomy (n = 6), and partial resection (n = 21). Additional ablation therapy by radio frequent ablation (RFA) was performed in 10 patients.
Clinical tumor characteristics of liver metastases with or without RE on CT are summarized in Table 2. There was no signi cant difference between the two groups regarding the clinical characteristics except for pathological diagnosis.

Initial Chemotherapy Responses
Overall chemotherapy responses are shown in Table 3. The total response rate was 54% (26/48), including 2 in CR and 24 in PR. Chemotherapy responses separated by metastatic hepatic tumor characteristics with or without peripheral RE are indicated. The chemotherapy response rate of REpositive tumors was 64% (23/36) and that of RE-negative tumors was 25% (3/12), which was signi cantly different. The disease control rate (CR + PR + SD/CR + PR + SD + PD) of RE-positive tumors was 86% (31/36) and that of RE-negative tumors was 75% (9/12).

Quantitative Analysis of Angiogenesis in Tumor Microenvironment and RE of Metastatic Hepatic Tumors
Tumors with peripheral RE on contrast-enhanced CT, as shown in Fig. 1A, (thick and thin arrows) corresponded to round stained tumors on abdominal angiography (thick and thin arrows, respectively), as shown in Fig. 1B, suggesting the identi cation of RE of the hepatic metastatic lesion to newly formed blood ow.
Histopathological and immunostaining examination of a metastatic hepatic tumor is shown in Fig. 2. A microscopic view (40X magni cation) of the metastatic hepatic tumor revealed moderately differentiated tubular adenocarcinoma (A). The surrounding tissues, including sinusoidal tissue, hepatocytes, broblasts, and endothelial cells, were compressed and invaded by the metastatic hepatic tumor. CD34 staining is shown in B (40X magni cation) and C (200X magni cation). CD34-stained cells were observed in host liver parenchyma, including compressed sinusoidal tissues, and invaded into metastatic cancer tissues. MVD was measured at peri-tumor areas at 200X magni cation using CD34-stained specimens as described in the Methods section.
The relationship between the MVD and CT-value is shown in Fig. 3A and B. The CT-value was also measured as described in the Methods section. MVD was associated with the RE CT-value of the metastatic hepatic tumor, revealing a strong relationship between microscopic tumor angiogenesis and Loading [MathJax]/jax/output/CommonHTML/jax.js the peripheral metastatic tumor RE (correlation coe cient (r) = 0.65, p-value = 0.01). The metastatic tumor peripheral RE on CT may re ect angiogenesis of the tumor.

Overall Survival Rates in Patients with RE-Positive or -Negative Tumors
The overall survival after systemic chemotherapy between patients with RE-positive and RE-negative tumors is shown in Fig. 4. The overall survival (OS) rate of RE-positive or -negative tumors with hepatic resection is shown in Fig. 4A. There were no signi cant differences between the groups. The mean OS rate of patients with RE-positive tumors was 42.1 months and that of those with RE-negative tumors was 48.0 months (p = 0.6). Longer survival was observed in the groups with hepatic resection than without hepatic resection (Fig. 4A,B). The OS rate between patients with RE-positive or -negative tumors without hepatic resection was also not signi cantly different. The mean OS rate of patients with RE-positive tumors was 25.8 months and that of those with RE-negative tumors was 23.7 months (p = 0.8). OS rates in all patients were not signi cantly different between RE-positive tumors and RE-negative tumors (Fig. 4C). The mean OS rate of patients with RE-positive tumors was 28.1 months and that of those with RE-negative tumors was 43.0 months (p = 0.05).

Discussion
We found that the peripheral RE on CT of metastatic hepatic tumors is associated with the angiogenesis in tumor micro-environment and may predict a good chemotherapy response. The combination of liver resection with chemotherapy improved the survival of patients who had multiple hepatic metastases. There are several reports on the concept of peripheral RE on CT of CRC metastasis 20, 21). CT-based morphological criteria, including peripheral rim of enhancement of the hepatic metastatic tumor, was reported to have a strong association with the pathological response and survival. These investigations support our study.
Angiogenesis is associated with tumor aggressiveness and poorer prognosis in patients with hepatic tumors22,23). Tumor angiogenesis facilitates metastatic formation by providing mechanisms to increase the likelihood of tumor cells invading the blood circulation, and provides nutrients for tumor growth and survival at the metastatic site. The interaction of tumor cells with endothelial cells in tumor microenvironment has an essential role in tumor angiogenesis. Blood nutrient supply and tumor-related endothelial cells promote tumor cell proliferation and tumor growth24). The tumor micro-environment is essential for the formation of a newly metastatic lesion. Tumor cells and host-cells, such as endothelial cells or broblast, participate in tumor metastasis. Tumors that are not vascularized at the metastatic site are typically maintained as small dormant nodules, and the tumor volume remains constant because of a balance between cell proliferation and cell death. Thus, tumor growth is dependent on angiogenesis.
We assessed clinical angiogenesis of metastatic hepatic tumors using indirect imaging by enhanced CT. The clinical manifestation of peripheral RE on CT of the metastatic hepatic tumor was con rmed to correspond with tumor angiogenesis on abdominal angiography (Fig. 1). This was supported by the Loading [MathJax]/jax/output/CommonHTML/jax.js investigation of the correlation between angiographically assessed vascularity and blood ow in hepatic metastases from colorectal carcinoma25). Of note, the hemodynamics of contrast medium between abdominal angiography and enhanced CT images were different. Angiography images were taken in the direct celiac arterial phase, whereas enhanced CT images were taken in the indirect portal phase through the intravenous injection of contrast medium. However, there was a possibility that tumor staining on angiography and peripheral RE on CT was the same because both images may re ect newly formed angiogenic vessels.
There are several methods to monitor angiogenesis using conventional imaging such as contrastenhanced US (ultrasound), CT, and MRI (magnetic resonance imaging). Enhanced CT is frequently used in the clinical setting, and can readily access metastatic hepatic tumors and surrounding tissues. Contrast-enhanced CT is useful to evaluate tumor angiogenesis by immunohistochemical quanti cation of the MVD in colorectal adenocarcinoma patients26). Evaluation of angiogenic vessels by the MVD is associated with microscopic tumor angiogenesis18). We found a strong relationship between microscopic tumor angiogenesis and peripheral metastatic tumor RE on CT.
Hepatic tissue including hepatocytes was fed by blood from the portal vein or hepatic artery, and the blood supply drained via the hepatic vein. Metastatic hepatic tumors may be supplied through angiogenesis via the portal vein or arterial blood ow. Metastatic hepatic tumors were reported to have a dual blood supply from both the portal vein and hepatic artery27,28). In our study, clinical angiogenesis was able to be assessed by not only peripheral RE on portal-phase CT, but also by arterial ow on celiac angiography. This suggested that metastatic hepatic carcinoma was fed from dual blood ow from the portal vein and hepatic artery. Therefore, the clinical manifestation of RE on CT of the metastatic hepatic tumor may be closely associated with tumor angiogenesis.
Angiogenic hepatic metastatic tumors responded well to systemic chemotherapy despite their aggressiveness. Angiogenic tumors may readily uptake anti-cancer drugs to the tumor through newly formed angiogenic vessels. Furthermore, immature angiogenic vessels are fenestrated29). Angiogenic factors, such as VEGF, which was rst identi ed as vascular permeability factor30,31), not only stimulate endothelial cells lining nearby microvessels to proliferate and migrate, but also render these vascular endothelial cells hyperpermeable. Hyperpermeable vessels may readily leak plasma proteins and deliver anti-cancer drugs into the extravascular space.
Recent clinical anti-angiogenic therapies, such as angiogenic antibody, have been used in patients with unresectable metastatic hepatic disease32,33). Anti-angiogenic antibody therapy itself is insu cient for anti-cancer effects. However, a single infusion of anti-VEGF antibody reduced tumor perfusion, vascular volume, microvascular density, interstitial uid pressure, and circulating endothelial cells in patients with rectal cancer34). This suggests that anti-angiogenic therapy has direct anti-vascular effects on human tumors. A combination of these drugs with anti-cancer drugs produces anti-cancer effects. Antiangiogenic molecules, such as anti-VEGF antibody, remodel tumor-related endothelial cells into normal endothelial cells with a normal structure35,36). Anti-angiogenic molecules reshape pathologic Loading [MathJax]/jax/output/CommonHTML/jax.js vasculature into normal vasculature, which results in delivery of the anti-cancer drug to the tumor. Although we analyzed only a few patients using anti-angiogenic agents in this study, there were antitumor effects without anti-angiogenic agents. Anti-angiogenic therapy may not be associated with a direct tumor response, but rather maintenance of anti-tumor effects. Normalization of tumor-related vasculature may enable the sustained delivery of anti-cancer drugs.
Hepatic resection remains the only potential curative treatment for metastatic tumors and improves survival 37,38). Similarly, we found that resection of metastatic hepatic tumors improved OS (Fig. 4A, B).
When the groups were divided by hepatic resection, there were no signi cant differences between patients with RE-positive and -negative tumors with hepatic resection (Fig. 4A). In addition, there were no signi cant differences between patients with RE-positive and -negative tumors without hepatic resection (Fig. 4B). A higher response rate to systemic chemotherapy was observed in patients with RE-positive tumors, but the OS rate of patients with RE-positive tumors was not signi cantly different from that of patients with RE-negative tumors (Fig. 4C). This suggests that a higher response to systemic chemotherapy does not always lead to longer survival. After the initial higher response in our patients, additional surgical therapy prolonged survival. To improve survival, additional therapeutic strategies, such as maintenance chemotherapy, use of molecular targeted drugs, or immuno-check point inhibitors, are needed. Clinically, hepatic metastatic tumors can recur or develop other metastatic lesions, such as a lung metastases or peritoneal dissemination, during the follow-up period for RE-positive and -negative tumors, which may affect patient survival.
There were a few limitations in the present study. First, there were patients without clinical manifestations, such as lung metastasis or peritoneal dissemination, during the initial treatment period. Second, the statistical power was weak because the sample size was small. The observational period of 10 years was relatively long, but new molecular drugs, such as anti-VEGF antibody and anti-EGFR antibody, were not frequently used for the initial treatment. Further studies with a larger number of patients and a shorter time period are needed to con rm our results.

Conclusion
We found that tumor peripheral RE on contrast CT in metastatic hepatic tumors from CRC was correlated with the angiogenesis in tumor micro-environment and indicated a good response to recent 5-FU-based systemic chemotherapy.  A, Quantitative analysis of the CT ndings. The regions of interest (ROI) were selected based on the best tumor image on contrast-enhanced CT. The ROI of the peripheral RE area was measured by at least ve independent enhanced spot areas (red cross) as a CT-value (Houns eld Unit: HFU). The mean CT-value was used, and the difference in the CT-value between the RE area and liver parenchyma area (blue cross) was de ned as the RE CT-value. B, The relationship between the MVD and CT-value. MVD was measured using a CD34-stained specimen and CT-values were measured at the peripheral RE area as described in the Methods section.