bs_bs_banner
Hepatology Research 2015
doi: 10.1111/hepr.12487
Original Article
Microvessel density in hepatocellular carcinoma: Prognostic significance and review of the previous published work Keigo Murakami,1 Atsuko Kasajima,1 Naoki Kawagishi,2 Noriaki Ohuchi2 and Hironobu Sasano1 1
Department of Pathology, and 2Division of Advanced Surgical Science and Technology, Graduate School of Medicine, Tohoku University School of Medicine, Sendai, Japan
Aim: Assessment of the microvessel density (MVD) may yield important information leading to an effective antiangiogenic treatment for hepatocellular carcinoma (HCC). Methods: The intratumoral MVD of 136 HCC patients was retrospectively evaluated using CD34. The correlation between the MVD and clinicopathological findings was assessed. In addition, the prognostic factors influencing the 2-year disease-free survival (DFS) and overall survival (OS) were analyzed.
Results: The MVD of each tumor size group (5 cm) was 196 ± 51, 181 ± 63 and 147 ± 69. The MVD of each histological grade (well-, moderately and poorly differentiated) was 200 ± 56, 184 ± 61 and 114 ± 55. The optimum cut-off values of the MVD for the 2-year DFS and OS were 118.3 and 112.7, respectively. For the 2-year DFS, high tumor marker levels (α-fetoprotein >100 ng/mL and protein induced by vitamin K
INTRODUCTION
H
EPATOCELLULAR CARCINOMA (HCC) is a highly vascularized tumor, and its associated angiogenesis is histologically characterized by the development of sinusoidal capillarization and unpaired arteries.1 Angiogenesis is of crucial importance to tumor growth, because it provides the oxygen and nutrition that are required by the proliferating neoplastic cells.2 The microvessel density (MVD), as assessed by CD31, CD34 or von Willebrand factor (vWF or Factor VIII) staining, has generally been considered as an index of angiogenesis in human malignancies.3 Above all, the anti-CD34 antibody has been suggested to be the most sensitive and specific marker among the other endothelial markers in HCC.4–6 The prognostic
Correspondence: Dr Keigo Murakami, Department of Pathology, Tohoku University Hospital, 1-1 Seiryou-machi, Aoba-ku, Sendai 980-8574, Japan. Email: [email protected] Received 17 August 2014; revision 13 December 2014; accepted 11 January 2015.
© 2015 The Japan Society of Hepatology
absence/antagonist-II >100 mAU/mL), poorly differentiated hepatocellular carcinoma (HCC), a high Ki-67 index (>20%), a large tumor size (>5 cm), vascular invasion, high tumor–node– metastasis (TNM) stage (III/IV) and a low MVD were the significant unfavorable prognostic factors. For the OS, a high Ki-67 index, a large tumor size, vascular invasion, high TNM stage and a low MVD were the significant risk factors for death. By the multivariate analysis, a low MVD was identified as an independent predictor of the 2-year DFS as well as the OS.
Conclusion: A low MVD can be used to predict an unfavorable prognosis in HCC patients. Key words: angiogenesis, CD34 antigen, hepatocellular carcinoma, microvessel density, prognostic factor
significance of the MVD in HCC has remained in dispute. The results of several previously published studies suggested that an increased MVD could serve as an independent prognostic factor for decreased disease-free survival (DFS) or overall survival (OS).7,8 On the other hand, several studies suggested that it had no prognostic value,9,10 and some studies demonstrated that there was a significantly lower MVD in larger or more advanced stage HCC.4 These discrepancies might have been caused by the differences in the endothelial markers employed, the sites evaluated, the counting methods and other reasons. A precise understanding of the MVD in HCC has become increasingly important, because several antiangiogenic agents have started to be used in the treatment of HCC.11 Therefore, in this study, we evaluated the MVD using CD34 as an endothelial marker, following the exclusion of unpaired arteries detected by elastica-Masson (EM) staining of the serial sections from the same case. In order to explore the significance of the MVD in HCC, we correlated the findings with the clinical outcome of patients with HCC.
K. Murakami et al.
METHODS Tissue specimens and clinical data
W
E RETRIEVED THE clinical data and surgical pathology specimens from 136 Japanese patients with HCC who underwent curative surgical resection from 1998 to 2010 at Tohoku University Hospital, Sendai, Japan. The protocol for this study was approved by the ethics committee of Tohoku University School of Medicine, Sendai, Japan. Informed consent was obtained from each patient. All resections were considered to be curative, with macroscopically complete removal of the tumor and histopathologically negative resection margins.12 A representative area for each case was selected from the growing edge of the tumor in all the cases examined. We obtained serial sections from each tissue block, and stained them with hematoxylin–eosin (HE) and performed EM staining. The average age of the patients was 63.3 ± 10.4 years (mean ± standard deviation; range, 28–81). There were 104 men and 32 women. The average tumor size was 5.1 ± 3.8 cm (range, 1.0–21), with 23 tumors smaller than 2 cm, 69 tumors of 2–5 cm and 44 tumors larger than 5 cm. Among the 136 HCC examined in this study, 60 had hepatitis C infection, 39 had hepatitis B infection and three had hepatitis C and B co-infection. The median follow up was 48.5 months. Patients who had undergone a previous hepatectomy or had a history of any other preoperative treatment were excluded from the analysis. In addition, those who did not survive beyond 60 days after surgery were not included.12 Staging was based on the TNM Classification of Malignant Tumours, Seventh Edition by the International Union Against Cancer.13 Each HCC was histologically classified based on the most predominant histological grade (well-, moderately or poorly differentiated) according to the classification of the Liver Cancer Study Group of Japan.14 Early HCC, defined as a welldifferentiated cancer containing Glisson’s triad and showing a replacing growth pattern and stromal invasion,14 was not included in this study because of the scarcity of specimens in our facility. Those who developed recurrence were treated with repeated hepatic resection, transcatheter arterial embolization or radiofrequency ablation.
Immunohistochemistry We performed immunohistochemical staining for CD34, α-smooth muscle actin (α-SMA), cytokeratin (CK)19 and Ki-67. The specimens were fixed in 10% formalin for 48–72 h at room temperature, embedded in paraffin, cut into 3-μm thick sections and placed on glue-coated glass slides. The sections were deparaffinized in xylene and hydrated with graded alcohols and distilled water. The
© 2015 The Japan Society of Hepatology
Hepatology Research 2015
endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 min at room temperature. Antigen retrieval from tissues was performed using 0.1% trypsin (37°C, 30 min) in Tris buffer (0.05 M, pH 7.6) for αSMA, and using an autoclave (120°C, 5 min) in citrate buffer (0.01 M, pH 6.0) for CK19 and Ki-67. We did not perform any antigen retrieval for CD34. Sections were subsequently incubated for 30 min at room temperature in a blocking solution of 10% rabbit serum (Nichirei Bioscience, Tokyo, Japan). The sections were incubated with primary antibodies for 16 h at 4°C. The dilutions of the monoclonal antibodies were as follows: antiCD34 (Nichirei Bioscience) 1:200, anti-α-SMA (Dako, Copenhagen, Denmark) 1:800, CK19 (Dako) 1:200 and Ki-67 (Dako) 1:300. Secondary antibody reactions were performed for 30 min at room temperature using biotinylated rabbit antimouse antibodies (Nichirei Bioscience) diluted 1:100, and peroxidase-conjugated avidin (Nichirei Bioscience) was added according to the manufacturer’s instructions. Reacted sections were visualized using 3,3′-diaminobenzidine-tetracholoride and 30% H2O2 in Tris buffer (0.05 M, pH 7.6) and were counterstained with hematoxylin.
Evaluation of MVD, unpaired arteries, CK19 and the Ki-67 labeling index After a light microscopic evaluation at low magnification, five parenchymal areas with the highest number of microvessels or three parenchymal areas with the highest number of unpaired arteries (hot spots) were independently selected.8,15 The number of vessels was counted at a higher magnification (×200, 0.74 mm2/field). Microvessels were identified by CD34 staining, regardless of the existence of lumen structures. There were two types of microvessels identified: capillary-like microvessels with small, scattered capillaries and having no or a narrow lumen (Figs 1a,b), and sinusoid-like microvessels with continuous branching and an obvious lumen structure (Figs 1c,d).7 For the sinusoid-like microvessels, which were mainly observed in the areas with a large trabecular structure, a modified method introduced by Tanigawa et al. was used.7 Briefly, every 40-μm length of lumen was calculated to be 1 point. The mean number of microvessels in the five hot spots detected by CD34 in tumor areas was regarded as the MVD.16 Unpaired arteries were defined as vessels without portal tracts or other units, such as bile duct. α-SMA positive microvessels without elastic fibers were occasionally observed in neoplastic lesions, especially in the cases with sinusoid-like microvessels (Fig. 1e). Therefore, we regarded vessels covered by circumferential smooth muscle and elastic fibers as unpaired arteries that
Hepatology Research 2015
MVD in HCC
Figure 1 Representative hepatocellular carcinoma (HCC) images of hematoxylin–eosin (HE) staining (original magnifications: [a,c] ×100), immunostaining for CD34 (b,d: ×100), α-smooth muscle actin (α-SMA) (e: ×100) and elastic-Masson (EM) staining (f: ×600). Parts (a) and (b) are images of serial tissue sections from the same block, as are (c–e). (a,b) Capillary-like microvessels with small and scattered capillaries with narrow lumens. (c,d) Sinusoid-like microvessels with continuous branching structures. Some αSMA positive microvessels without elastic fibers were occasionally observed (e). (f) Magnified image of an unpaired artery detected by EM staining. The unpaired artery had circumferential elastic fibers which could be detected by EM staining as a deep green wavy line.
were detected using EM staining in this study (Fig. 1f). We excluded unpaired arteries when counting the microvessels. Expression of CK19 was considered positive if more than 5% of tumor cells were stained. The Ki-67 labeling index (LI) in tumor cells was determined by counting more than 1000 tumor cells at high magnification (×400).
Statistical analysis The Mann–Whitney U-test or Kruskal–Wallis nonparametric tests were performed to compare continuous values among groups. Categorical variables were assessed using a χ 2-test analysis. The classification accuracy was measured using the area under the curve (AUC) of the receiver–operator curves (ROC). The best cut-off value was established according to the point of the ROC farthest from the diagonal line, which was indicated as the maximum of the Youden index (= maximum [sensitivity + specificity 1]). The cumulative OS and DFS rates were
calculated using the Kaplan–Meyer method, and the comparisons between the subgroups were made using the log– rank test. The Cox proportional hazards regression model was used for the multivariate analysis. The results were considered to be significant for values of P < 0.05. All analyses were performed using the IBM SPSS version 19 software program for Windows (PASW Statistics for Windows; SPSS, Chicago, IL, USA).
RESULTS Correlation between the MVD and the clinicopathological factors of the patients
R
EPRESENTATIVE IMAGES OF the tumors following HE staining and immunostaining for CD34 are shown in Figures 1. The mean MVD count of the entire series was 172/field. One hundred and sixteen of all cases (85.3%) exhibited capillary-like microvessels, and 20 cases (14.7%) had sinusoid-like or mixed types of microvessels.
© 2015 The Japan Society of Hepatology
K. Murakami et al.
The analysis of the correlations between the clinicopathological features and the microvessel types revealed that large HCC (>5 cm) and the presence of vascular invasion were significantly more common in the sinusoid-like or mixed type groups (P = 0.036 and 0.001, respectively). Table 1 shows the mean MVD of each clinicopathological factor examined in our present study. A significantly lower MVD was detected in the patients with a lower age, higher α-fetoprotein (AFP) level, higher protein induced by vitamin K absence/antagonist-II (PIVKA-II) level and the presence of vascular invasion. Furthermore, the MVD significantly decreased as the tumor size and histological grade progressed. There were no significant differences detected in the MVD based on the sex, Child–Pugh classification or viral etiology of the patients.
Correlation between the Ki-67 LI and histological grade of HCC The mean Ki-67 LI (%) of poorly differentiated HCC was higher than that of well- and moderately differentiated HCC (well, 11.3 ± 6.2%; moderately, 14.0 ± 9.2%; and poorly, 29.3 ± 13.5%, respectively). A significant increase of the Ki-67 LI in tumor cells was detected based on the histological grade (P < 0.001, data not shown).
Correlation between the number of unpaired arteries and the histological grade of HCC The average numbers of unpaired arteries in well-, moderately and poorly differentiated HCC were 7.1 ± 4.2, 11.8 ± 6.3 and 6.4 ± 4.4/field, respectively (P = 0.012). Unpaired arteries were most frequently detected in the cases with moderately differentiated HCC. The number of unpaired arteries in the poorly differentiated HCC was the lowest among three histological grades.
Prognostic value of the MVD for HCC The ROC curves showed that the optimum diagnostic cutoff values of the MVD for the 2-year DFS and OS were 118.3 (AUC, 0.643; 95% confidence interval [CI], 0.548– 0.737; sensitivity, 87.0%; specificity, 43.3%) and 112.7 (AUC, 0.597; 95% CI, 0.486–0.707; sensitivity, 85.1%; specificity, 42.9%), respectively. Patients were subsequently classified into two different groups according to the MVD; 118.3 or less (n = 38) and more than 118.3 (n = 98) for the 2-year DFS, and MVD of 112.7 or less (n = 32) and more than 112.7 (n = 104) for the OS. In the analysis of the 2-year DFS, there were statistically significant differences, with more favorable survival for the patients with a low AFP level (≤100 ng/mL) (P = 0.016), a low PIVKA-II level (≤100 mAU/ml) (P = 0.008), well- or moderately differentiated histological grade tumors
© 2015 The Japan Society of Hepatology
Hepatology Research 2015 Table 1 Relationship between the MVD and the clinicopathological features of HCC patients MVD (/0.74 mm2) Variable
n
Age ≤60 years 46 >60 years 90 Sex Male 104 Female 32 Child–Pugh A 114 B and C 22 Cirrhosis Absent 92 Present 44 Viral etiology HCV 60 HBV 39 HCV + HBV 3 None 34 AFP (ng/mL) ≤100 90 >100 46 PIVKA-II (mAU/mL) ≤100 65 >100 71 Tumor size (cm) 5 44 Histological grade Well-differentiated 11 Moderately differentiated 100 Poorly differentiated 25 Vascular invasion Absent 81 Present 55 TNM stage I 65 II 53 III 16 IV 2
Mean
SD
158.6 179.3
60.2 67.0
169.8 180.4
68.0 56.0
171.0 179.0
64.4 71.2
169.2 178.9
65.9 64.3
184.2 153.8 144.1 175.0
69.7 54.9 45.8 66.4
187.1 143.4
62.2 62.1
192.6 153.8
55.9 68.1
195.8 180.9 146.5
50.6 62.8 68.8
199.5 183.9 113.8
55.7 60.6 55.4
190.8 145.1
57.7 66.8
196.1 152.5 152.7 80.6
58.2 58.4 81.5 32.0
P 0.046
0.353
0.585
0.404
0.086
100 ng/mL) Tumor marker (PIVKA-II ≤100/>100 mAU/mL) Histological grade (well- or moderately/poorly differentiated) CK19 (negative/positive) Ki-67 (≤20.0/>20.0%) Tumor size (≤5/>5 cm) Vascular invasion (absent/present) TNM stage (I + II/III + IV) MVD (≤118.3/>118.3)
n
P
46/90 104/32 102/34 114/22 91/45 65/71 111/25 120/16 96/40 92/44 81/55 118/18 38/98
0.963 0.112 0.876 0.403 0.016 0.008 0.005 0.563 0.006 20.0%) Tumor size (≤5/>5 cm) Vascular invasion (absent/present) TNM stage (I + II/III + IV) MVD (≤112.7/>112.7)
Multivariate
n
P
P
HR
HR (95% CI)
46/90 104/32 102/34 114/22 90/46 65/71 111/25 120/16 96/40 92/44 81/55 118/18 32/104
0.783 0.896 0.535 0.519 0.522 0.121 0.146 0.618 0.009 0.005 0.003 5 cm, 44
CD34 and CD105 CD34
≤5 cm, 33 >5 cm, 53 ≤5 cm, 17 >5 cm, 23
Growing edge
Yes
Yes computer analysis
No computer analysis
Growing edge
N.A.
No
No
No computer analysis
No
Yes computer analysis No computer analysis
Yes
N.A.
Growing edge
Growing edge
CD34 and vWF CD34
Random
N.A.
CD34
≤5 cm, 75 >5 cm, 25
≤5 cm, 12 >5 cm, 38 ≤5 cm, 46 >5 cm, 54
N.A.
CD34
No
Growing edge
≤5 cm, 40 >5 cm, 38 ≤5 cm, 55 >5 cm, 29 N.A.
Yes
Assessment of sinusoid-like MVD
N.A.
CD34 and vWF vWF
Antibody
Sampling site of tumor
5 cm, 23
Tumor size
Negative
None
Negative (for CD34) Positive trend not significant
None
Negative trend not significant
None
None
None
None
Positive (for CD34) None
Correlation with histological grading
Negative
(5–10 cm) > (>10 cm) > ( (>5 cm) > (≤2 cm) Negative
Negative
Negative
Positive (for vWF) N.A.
Positive (for CD34) (2–5 cm) > (>5 cm) > (
Hepatology Research 2015
doi: 10.1111/hepr.12487
Original Article
Microvessel density in hepatocellular carcinoma: Prognostic significance and review of the previous published work Keigo Murakami,1 Atsuko Kasajima,1 Naoki Kawagishi,2 Noriaki Ohuchi2 and Hironobu Sasano1 1
Department of Pathology, and 2Division of Advanced Surgical Science and Technology, Graduate School of Medicine, Tohoku University School of Medicine, Sendai, Japan
Aim: Assessment of the microvessel density (MVD) may yield important information leading to an effective antiangiogenic treatment for hepatocellular carcinoma (HCC). Methods: The intratumoral MVD of 136 HCC patients was retrospectively evaluated using CD34. The correlation between the MVD and clinicopathological findings was assessed. In addition, the prognostic factors influencing the 2-year disease-free survival (DFS) and overall survival (OS) were analyzed.
Results: The MVD of each tumor size group (5 cm) was 196 ± 51, 181 ± 63 and 147 ± 69. The MVD of each histological grade (well-, moderately and poorly differentiated) was 200 ± 56, 184 ± 61 and 114 ± 55. The optimum cut-off values of the MVD for the 2-year DFS and OS were 118.3 and 112.7, respectively. For the 2-year DFS, high tumor marker levels (α-fetoprotein >100 ng/mL and protein induced by vitamin K
INTRODUCTION
H
EPATOCELLULAR CARCINOMA (HCC) is a highly vascularized tumor, and its associated angiogenesis is histologically characterized by the development of sinusoidal capillarization and unpaired arteries.1 Angiogenesis is of crucial importance to tumor growth, because it provides the oxygen and nutrition that are required by the proliferating neoplastic cells.2 The microvessel density (MVD), as assessed by CD31, CD34 or von Willebrand factor (vWF or Factor VIII) staining, has generally been considered as an index of angiogenesis in human malignancies.3 Above all, the anti-CD34 antibody has been suggested to be the most sensitive and specific marker among the other endothelial markers in HCC.4–6 The prognostic
Correspondence: Dr Keigo Murakami, Department of Pathology, Tohoku University Hospital, 1-1 Seiryou-machi, Aoba-ku, Sendai 980-8574, Japan. Email: [email protected] Received 17 August 2014; revision 13 December 2014; accepted 11 January 2015.
© 2015 The Japan Society of Hepatology
absence/antagonist-II >100 mAU/mL), poorly differentiated hepatocellular carcinoma (HCC), a high Ki-67 index (>20%), a large tumor size (>5 cm), vascular invasion, high tumor–node– metastasis (TNM) stage (III/IV) and a low MVD were the significant unfavorable prognostic factors. For the OS, a high Ki-67 index, a large tumor size, vascular invasion, high TNM stage and a low MVD were the significant risk factors for death. By the multivariate analysis, a low MVD was identified as an independent predictor of the 2-year DFS as well as the OS.
Conclusion: A low MVD can be used to predict an unfavorable prognosis in HCC patients. Key words: angiogenesis, CD34 antigen, hepatocellular carcinoma, microvessel density, prognostic factor
significance of the MVD in HCC has remained in dispute. The results of several previously published studies suggested that an increased MVD could serve as an independent prognostic factor for decreased disease-free survival (DFS) or overall survival (OS).7,8 On the other hand, several studies suggested that it had no prognostic value,9,10 and some studies demonstrated that there was a significantly lower MVD in larger or more advanced stage HCC.4 These discrepancies might have been caused by the differences in the endothelial markers employed, the sites evaluated, the counting methods and other reasons. A precise understanding of the MVD in HCC has become increasingly important, because several antiangiogenic agents have started to be used in the treatment of HCC.11 Therefore, in this study, we evaluated the MVD using CD34 as an endothelial marker, following the exclusion of unpaired arteries detected by elastica-Masson (EM) staining of the serial sections from the same case. In order to explore the significance of the MVD in HCC, we correlated the findings with the clinical outcome of patients with HCC.
K. Murakami et al.
METHODS Tissue specimens and clinical data
W
E RETRIEVED THE clinical data and surgical pathology specimens from 136 Japanese patients with HCC who underwent curative surgical resection from 1998 to 2010 at Tohoku University Hospital, Sendai, Japan. The protocol for this study was approved by the ethics committee of Tohoku University School of Medicine, Sendai, Japan. Informed consent was obtained from each patient. All resections were considered to be curative, with macroscopically complete removal of the tumor and histopathologically negative resection margins.12 A representative area for each case was selected from the growing edge of the tumor in all the cases examined. We obtained serial sections from each tissue block, and stained them with hematoxylin–eosin (HE) and performed EM staining. The average age of the patients was 63.3 ± 10.4 years (mean ± standard deviation; range, 28–81). There were 104 men and 32 women. The average tumor size was 5.1 ± 3.8 cm (range, 1.0–21), with 23 tumors smaller than 2 cm, 69 tumors of 2–5 cm and 44 tumors larger than 5 cm. Among the 136 HCC examined in this study, 60 had hepatitis C infection, 39 had hepatitis B infection and three had hepatitis C and B co-infection. The median follow up was 48.5 months. Patients who had undergone a previous hepatectomy or had a history of any other preoperative treatment were excluded from the analysis. In addition, those who did not survive beyond 60 days after surgery were not included.12 Staging was based on the TNM Classification of Malignant Tumours, Seventh Edition by the International Union Against Cancer.13 Each HCC was histologically classified based on the most predominant histological grade (well-, moderately or poorly differentiated) according to the classification of the Liver Cancer Study Group of Japan.14 Early HCC, defined as a welldifferentiated cancer containing Glisson’s triad and showing a replacing growth pattern and stromal invasion,14 was not included in this study because of the scarcity of specimens in our facility. Those who developed recurrence were treated with repeated hepatic resection, transcatheter arterial embolization or radiofrequency ablation.
Immunohistochemistry We performed immunohistochemical staining for CD34, α-smooth muscle actin (α-SMA), cytokeratin (CK)19 and Ki-67. The specimens were fixed in 10% formalin for 48–72 h at room temperature, embedded in paraffin, cut into 3-μm thick sections and placed on glue-coated glass slides. The sections were deparaffinized in xylene and hydrated with graded alcohols and distilled water. The
© 2015 The Japan Society of Hepatology
Hepatology Research 2015
endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 min at room temperature. Antigen retrieval from tissues was performed using 0.1% trypsin (37°C, 30 min) in Tris buffer (0.05 M, pH 7.6) for αSMA, and using an autoclave (120°C, 5 min) in citrate buffer (0.01 M, pH 6.0) for CK19 and Ki-67. We did not perform any antigen retrieval for CD34. Sections were subsequently incubated for 30 min at room temperature in a blocking solution of 10% rabbit serum (Nichirei Bioscience, Tokyo, Japan). The sections were incubated with primary antibodies for 16 h at 4°C. The dilutions of the monoclonal antibodies were as follows: antiCD34 (Nichirei Bioscience) 1:200, anti-α-SMA (Dako, Copenhagen, Denmark) 1:800, CK19 (Dako) 1:200 and Ki-67 (Dako) 1:300. Secondary antibody reactions were performed for 30 min at room temperature using biotinylated rabbit antimouse antibodies (Nichirei Bioscience) diluted 1:100, and peroxidase-conjugated avidin (Nichirei Bioscience) was added according to the manufacturer’s instructions. Reacted sections were visualized using 3,3′-diaminobenzidine-tetracholoride and 30% H2O2 in Tris buffer (0.05 M, pH 7.6) and were counterstained with hematoxylin.
Evaluation of MVD, unpaired arteries, CK19 and the Ki-67 labeling index After a light microscopic evaluation at low magnification, five parenchymal areas with the highest number of microvessels or three parenchymal areas with the highest number of unpaired arteries (hot spots) were independently selected.8,15 The number of vessels was counted at a higher magnification (×200, 0.74 mm2/field). Microvessels were identified by CD34 staining, regardless of the existence of lumen structures. There were two types of microvessels identified: capillary-like microvessels with small, scattered capillaries and having no or a narrow lumen (Figs 1a,b), and sinusoid-like microvessels with continuous branching and an obvious lumen structure (Figs 1c,d).7 For the sinusoid-like microvessels, which were mainly observed in the areas with a large trabecular structure, a modified method introduced by Tanigawa et al. was used.7 Briefly, every 40-μm length of lumen was calculated to be 1 point. The mean number of microvessels in the five hot spots detected by CD34 in tumor areas was regarded as the MVD.16 Unpaired arteries were defined as vessels without portal tracts or other units, such as bile duct. α-SMA positive microvessels without elastic fibers were occasionally observed in neoplastic lesions, especially in the cases with sinusoid-like microvessels (Fig. 1e). Therefore, we regarded vessels covered by circumferential smooth muscle and elastic fibers as unpaired arteries that
Hepatology Research 2015
MVD in HCC
Figure 1 Representative hepatocellular carcinoma (HCC) images of hematoxylin–eosin (HE) staining (original magnifications: [a,c] ×100), immunostaining for CD34 (b,d: ×100), α-smooth muscle actin (α-SMA) (e: ×100) and elastic-Masson (EM) staining (f: ×600). Parts (a) and (b) are images of serial tissue sections from the same block, as are (c–e). (a,b) Capillary-like microvessels with small and scattered capillaries with narrow lumens. (c,d) Sinusoid-like microvessels with continuous branching structures. Some αSMA positive microvessels without elastic fibers were occasionally observed (e). (f) Magnified image of an unpaired artery detected by EM staining. The unpaired artery had circumferential elastic fibers which could be detected by EM staining as a deep green wavy line.
were detected using EM staining in this study (Fig. 1f). We excluded unpaired arteries when counting the microvessels. Expression of CK19 was considered positive if more than 5% of tumor cells were stained. The Ki-67 labeling index (LI) in tumor cells was determined by counting more than 1000 tumor cells at high magnification (×400).
Statistical analysis The Mann–Whitney U-test or Kruskal–Wallis nonparametric tests were performed to compare continuous values among groups. Categorical variables were assessed using a χ 2-test analysis. The classification accuracy was measured using the area under the curve (AUC) of the receiver–operator curves (ROC). The best cut-off value was established according to the point of the ROC farthest from the diagonal line, which was indicated as the maximum of the Youden index (= maximum [sensitivity + specificity 1]). The cumulative OS and DFS rates were
calculated using the Kaplan–Meyer method, and the comparisons between the subgroups were made using the log– rank test. The Cox proportional hazards regression model was used for the multivariate analysis. The results were considered to be significant for values of P < 0.05. All analyses were performed using the IBM SPSS version 19 software program for Windows (PASW Statistics for Windows; SPSS, Chicago, IL, USA).
RESULTS Correlation between the MVD and the clinicopathological factors of the patients
R
EPRESENTATIVE IMAGES OF the tumors following HE staining and immunostaining for CD34 are shown in Figures 1. The mean MVD count of the entire series was 172/field. One hundred and sixteen of all cases (85.3%) exhibited capillary-like microvessels, and 20 cases (14.7%) had sinusoid-like or mixed types of microvessels.
© 2015 The Japan Society of Hepatology
K. Murakami et al.
The analysis of the correlations between the clinicopathological features and the microvessel types revealed that large HCC (>5 cm) and the presence of vascular invasion were significantly more common in the sinusoid-like or mixed type groups (P = 0.036 and 0.001, respectively). Table 1 shows the mean MVD of each clinicopathological factor examined in our present study. A significantly lower MVD was detected in the patients with a lower age, higher α-fetoprotein (AFP) level, higher protein induced by vitamin K absence/antagonist-II (PIVKA-II) level and the presence of vascular invasion. Furthermore, the MVD significantly decreased as the tumor size and histological grade progressed. There were no significant differences detected in the MVD based on the sex, Child–Pugh classification or viral etiology of the patients.
Correlation between the Ki-67 LI and histological grade of HCC The mean Ki-67 LI (%) of poorly differentiated HCC was higher than that of well- and moderately differentiated HCC (well, 11.3 ± 6.2%; moderately, 14.0 ± 9.2%; and poorly, 29.3 ± 13.5%, respectively). A significant increase of the Ki-67 LI in tumor cells was detected based on the histological grade (P < 0.001, data not shown).
Correlation between the number of unpaired arteries and the histological grade of HCC The average numbers of unpaired arteries in well-, moderately and poorly differentiated HCC were 7.1 ± 4.2, 11.8 ± 6.3 and 6.4 ± 4.4/field, respectively (P = 0.012). Unpaired arteries were most frequently detected in the cases with moderately differentiated HCC. The number of unpaired arteries in the poorly differentiated HCC was the lowest among three histological grades.
Prognostic value of the MVD for HCC The ROC curves showed that the optimum diagnostic cutoff values of the MVD for the 2-year DFS and OS were 118.3 (AUC, 0.643; 95% confidence interval [CI], 0.548– 0.737; sensitivity, 87.0%; specificity, 43.3%) and 112.7 (AUC, 0.597; 95% CI, 0.486–0.707; sensitivity, 85.1%; specificity, 42.9%), respectively. Patients were subsequently classified into two different groups according to the MVD; 118.3 or less (n = 38) and more than 118.3 (n = 98) for the 2-year DFS, and MVD of 112.7 or less (n = 32) and more than 112.7 (n = 104) for the OS. In the analysis of the 2-year DFS, there were statistically significant differences, with more favorable survival for the patients with a low AFP level (≤100 ng/mL) (P = 0.016), a low PIVKA-II level (≤100 mAU/ml) (P = 0.008), well- or moderately differentiated histological grade tumors
© 2015 The Japan Society of Hepatology
Hepatology Research 2015 Table 1 Relationship between the MVD and the clinicopathological features of HCC patients MVD (/0.74 mm2) Variable
n
Age ≤60 years 46 >60 years 90 Sex Male 104 Female 32 Child–Pugh A 114 B and C 22 Cirrhosis Absent 92 Present 44 Viral etiology HCV 60 HBV 39 HCV + HBV 3 None 34 AFP (ng/mL) ≤100 90 >100 46 PIVKA-II (mAU/mL) ≤100 65 >100 71 Tumor size (cm) 5 44 Histological grade Well-differentiated 11 Moderately differentiated 100 Poorly differentiated 25 Vascular invasion Absent 81 Present 55 TNM stage I 65 II 53 III 16 IV 2
Mean
SD
158.6 179.3
60.2 67.0
169.8 180.4
68.0 56.0
171.0 179.0
64.4 71.2
169.2 178.9
65.9 64.3
184.2 153.8 144.1 175.0
69.7 54.9 45.8 66.4
187.1 143.4
62.2 62.1
192.6 153.8
55.9 68.1
195.8 180.9 146.5
50.6 62.8 68.8
199.5 183.9 113.8
55.7 60.6 55.4
190.8 145.1
57.7 66.8
196.1 152.5 152.7 80.6
58.2 58.4 81.5 32.0
P 0.046
0.353
0.585
0.404
0.086
100 ng/mL) Tumor marker (PIVKA-II ≤100/>100 mAU/mL) Histological grade (well- or moderately/poorly differentiated) CK19 (negative/positive) Ki-67 (≤20.0/>20.0%) Tumor size (≤5/>5 cm) Vascular invasion (absent/present) TNM stage (I + II/III + IV) MVD (≤118.3/>118.3)
n
P
46/90 104/32 102/34 114/22 91/45 65/71 111/25 120/16 96/40 92/44 81/55 118/18 38/98
0.963 0.112 0.876 0.403 0.016 0.008 0.005 0.563 0.006 20.0%) Tumor size (≤5/>5 cm) Vascular invasion (absent/present) TNM stage (I + II/III + IV) MVD (≤112.7/>112.7)
Multivariate
n
P
P
HR
HR (95% CI)
46/90 104/32 102/34 114/22 90/46 65/71 111/25 120/16 96/40 92/44 81/55 118/18 32/104
0.783 0.896 0.535 0.519 0.522 0.121 0.146 0.618 0.009 0.005 0.003 5 cm, 44
CD34 and CD105 CD34
≤5 cm, 33 >5 cm, 53 ≤5 cm, 17 >5 cm, 23
Growing edge
Yes
Yes computer analysis
No computer analysis
Growing edge
N.A.
No
No
No computer analysis
No
Yes computer analysis No computer analysis
Yes
N.A.
Growing edge
Growing edge
CD34 and vWF CD34
Random
N.A.
CD34
≤5 cm, 75 >5 cm, 25
≤5 cm, 12 >5 cm, 38 ≤5 cm, 46 >5 cm, 54
N.A.
CD34
No
Growing edge
≤5 cm, 40 >5 cm, 38 ≤5 cm, 55 >5 cm, 29 N.A.
Yes
Assessment of sinusoid-like MVD
N.A.
CD34 and vWF vWF
Antibody
Sampling site of tumor
5 cm, 23
Tumor size
Negative
None
Negative (for CD34) Positive trend not significant
None
Negative trend not significant
None
None
None
None
Positive (for CD34) None
Correlation with histological grading
Negative
(5–10 cm) > (>10 cm) > ( (>5 cm) > (≤2 cm) Negative
Negative
Negative
Positive (for vWF) N.A.
Positive (for CD34) (2–5 cm) > (>5 cm) > (
