ANATOMIC PATHOLOGY Original Article Tumor Angiogenesis in Stage II Carcinoma Association With Survival BARBARA F. BANNER, MD, ROB WHITEHOUSE, STEPHEN P. BAKER, MSPH, AND RICHARD S. SWANSON, MD We studied the frequency of microvessels in T NO MO colorectal carcinomas from patients with widely different survival times. Microvessels (<5 nm diameter) were enhanced by immunostaining with antibody to factor VHI-related antigen and counted in x high-power fields in sections of resected carcinomas from 9 patients who died of disease in months or less (short-term survivors) and who had no evidence of disease at 9 months or longer (longterm survivors). The means of the highest counts for each case were compared between the long- and short-term survivor groups. The mean ± SD microvessel count was 5. ±.5 for the short-term survivors and. ±. for the long-term survivors. Median counts There is ample evidence that angiogenesis is a major factor supporting the growth of tumors., Morphologic studies enumerating microvessels in tissue sections have shown an association between higher frequency of tumor microvessels and poor patient survival, increased risk for recurrence, and metastasis in tumors from many sites, including breast, lung, genitourinary tract, and skin., These studies suggested that microvessel counts performed on tissue sections of tumors may be of prognostic value in the workup of patients with cancer. Although stage at presentation is the most important prognostic factor for colorectal carcinoma, 5-year survival for stage II (T) colorectal carcinomas ranges from % to 8%. 5 It would be advantageous to have additional prognostic information to further refine the definition of stage II and to help predict whether patients will have good or poor clinical outcomes. Findings in previous studies of tumor vessels in his- From the Departments of'pathology, Academic Computing, and Surgery, University of Massachusetts Medical Center, Worcester, Massachusetts. Manuscript received April, 997; revision accepted October,997. Address reprint requests to Dr Banner: Department of Pathology, University of Massachusetts Medical Center, 55 Lake Ave N, Worcester, MA 55. were 7. and 9., respectively. The distribution of microvessel counts was skewed toward higher counts in the long-term survivors. There was no correlation between microvessel counts and tumor site, size, or grade; lymphovascular invasion; or the presence of a mucinous component. Although there was a trend toward a higher frequency of microvessels in patients with longer survival, it is unlikely that microvessel count is an independent prognostic indicator for patients with T NO M colorectal carcinoma because there is only a small difference in microvessel frequency between patients with widely different survival times. (Key words: Microvessel counts; carcinoma) Am J Clin Pathol 998;9:7-77. tologic sections of resected colorectal carcinomas agree that microvessel counts may be prognostically useful, but they do not agree on what a high or low count should indicate. Some studies showed an association between high microvascular count and shortterm survival -8 and between a high microvascular count and tumor recurrence. 8,9 However, Lindmark et al found higher microvessel counts in patients with longer survivals. The aim of this study was to compare microvessel counts in T NO M colorectal carcinomas from patients with long-term (>9 months) and shortterm (< months) survival to determine whether microvessel counts differ in patients at the same tumor stage but with widely different survival times. Cases Studied MATERIALS AND METHODS From January 979 to June 99, patients with stage T NO M colorectal carcinoma were identified in the University of Massachusetts Medical Center (Worcester) Tumor Registry. Of these, patients who died of disease before months (n = 9) and patients who lived longer than 9 months with no evidence of disease (n = ) and for whom slides and blocks from a resection specimen were available were Downloaded from https://academic.oup.com/ajcp/article-abstract/9//7/7579 by guest on January 9 7
7 ANATOMIC PATHOLOGY / Article selected for study, for a total of cases. For each case, the tumor location and size were obtained from the surgical pathology reports. Follow-up data were obtained from the Tumor Registry. The hematoxylineosin-stained slides were reviewed to confirm the diagnosis, stage, and grade. Tumors were graded according to the amount of solid growth pattern: grade, <% solid pattern;, % to 8% solid pattern; and, >8% solid pattern. Lymphovascular invasion was assessed in the factor VHI-stained sections and rated positive if there were tumor cells in vessels away from the tumor or in channels lined by factor VHI-positive cells within the tumor. Mucinous component was identified if mucin distended glands or created extracellular pools in more than % of the tumor. Tissue Processing The hematoxylin-eosin-stained slides were reviewed, and one section containing the deep invasive edge of the tumor was selected from each case for immunostaining. Although the sections had been taken for the routine diagnostic workup, at least one with the deep invasive edge of the tumor was available for all cases. In many cases, the sections were full-thickness through the tumors. All tissue samples had been fixed in % buffered formalin and routinely processed for light microscopy. Sections were cut at xm, heated at C for minutes, and deparaffinized and hydrated through a series of xylenes and alcohols. Optimum pretreatment and dilutions were determined by testing with known positive and negative material. Rabbit polyclonal antibody against factor VIII (DAKO, Carpinteria, Calif) at a dilution of :, with antigen retrieval gave us the best signal/noise ratio in the positive test material and no staining in the negative test material. The slides were microwaved with a proprietary antigen retrieval solution (citrate buffer) (BioTek Solutions, Santa Barbara, Calif) for 5 minutes in an 8-W microwave oven. Following replenishment of this solution, the slides were microwaved again for an additional 5 minutes and allowed to cool for minutes. The slides were stained on an automated immunostainer (TechMate, BioTek) using an avidin-biotin complex staining procedure (BioTek). Following a hydrogen peroxide block of endogenous peroxide and a serum blocking step, the slides were incubated with the primary antibody for 5 minutes, followed by brief buffer washes and incubation in a cocktail of biotinylated antimouse IgG/IgM and antirabbit IgG (BioTek) for minutes. The sections were then washed, incubated in avidin-biotin complex for minutes, washed, and reacted with diaminobenzidine and hydrogen peroxide to visualize the end product. Sections were counterstained with hematoxylin. A duplicate set of slides was also stained in the same manner substituting normal rabbit serum for factor VTH antibody to serve as a negative control. Identification and Counting of Microvessels Each immunostained section was scanned at low power to identify the deeper half of the tumor, including the deep invasive edge. To prevent granulation tissue in ulcerated tumors from interfering with the tumor microvessel count, the luminal surfaces were excluded from the areas to be analyzed by a line drawn arbitrarily across the section. Abscesses when present adjacent to the tumors were also excluded. The slides were then scanned to identify the areas containing the greatest number of distinct factor Vlll-positive vessels, and the microvessels were counted in x fields in these areas of highest neovascularization. To equalize the cases, the highest counts for each case were used in the data analysis, because all of the cases had at least countable areas. Specific criteria were applied when counting vessels, based on those of Weidner. These criteria were as follows: () vessels counted had to be in stroma within the tumor, ie, stroma surrounded by malignant glands; () vessel-containing stroma had to occupy more than 5% of the x field to be counted; () microvessel diameter had to be 5 urn or less (by eyepiece micrometer); and () positive cells in clusters without an obvious lumen were counted if they otherwise were consistent with vessels by their size and distribution among the other vessels. An example of a typical x field with factor Vlll-positive microvessels is shown in Figure. Data Analysis The mean microvessel counts per high-power field (HPF) for the long-term and short-term survivors were compared statistically using a Mann-Whitney U- Wilcoxon Rank Sum W test and a Kolmogorov- Smirnov--Sample Test. RESULTS As shown in Table, the male/female ratio, the age range of the patients, and the tumor size, grade, and location were comparable in the short-term and long-term survivor groups. However, lymphovascular invasion Downloaded from https://academic.oup.com/ajcp/article-abstract/9//7/7579 by guest on January 9 AJCP J une998
BANNER ET AL 75 Micwvessels in Carcinoma and the presence of a mucinous component were more frequent in the short-term survivors. Of the 9 patients in the short-term survivor group who died of disease, carcinoma was the cause of death in 7 cases. Of the patients who survived longer than years, were alive and died with no evidence of carcinoma at the time of the last follow-up. The mean ± SD number of microvessels per HPF was 5. ±.5 for the short-term survivors (range, FIG. A representative x high-power field containing seven factor VM-positive microvessels (arrows, with two arrows on one vessel in lower right). A portion of a tumor gland is present at the upper edge (avidin-biotin complex immunostaining technique, x). Mean age (range) (y) Men Women Tumor dimension, maximum range (mean) (cm) Location Cecum Ascending colon Transverse colon Sigmoid Rectum Unknown Tumor grade Lymphovascular invasion Mucinous component Microvessel counts Mean ± SD Median.-9.8). For the long-term survivors the mean ± SD number of microvesssels per HPF was. ±. (range,.-8.5). The highest counts per HPF for each case were used to compare the short- and longterm survivors with the Mann-Whitney fj-wilcoxon Rank Sum W test. The difference between the two groups was not significant (P =.7). The median counts for the short- and long-term survivors were 7. and 9., respectively. A graph of the distribution of the counts (Fig ) illustrates the overlap, but shows a tendency toward higher counts in the long-term survivors. Microvessel counts were compared with standard prognostic factors of tumor location, size, grade, lymphovascular invasion, and presence of a mucinous component. The distribution of mean microvessel counts per HPF by tumor location was similar for tumors in the cecum, ascending colon, transverse or sigmoid colon, and rectum, although the highest counts were in tumors from the cecum and the ascending colon and the lowest was in a tumor in the rectum. Similarly, there were no differences between the short-term and long-term survivors in the distribution of microvessel counts by tumor size, tumor grade, lymphovascular invasion, or presence of a mucinous component. To further unify the groups, we eliminated the few cases of rectal carcinoma, but the difference in microvessel counts between these smaller groups was also not significant. TABLE. COMPARISON OF SHORT- AND LONG-TERM SURVIVOR GROUPS* Short-Term (n = 9) Long-Term (n = ) (-8) - (.) 5 5. ±.5 7. 8 (55-87) 7-7.5 (.7) ] ' 9. ±. 9. Downloaded from https://academic.oup.com/ajcp/article-abstract/9//7/7579 by guest on January 9 'Data given are numbers. Vol.9 No.
7 ANATOMIC PATHOLOGY Original Article 5 5- - V) a </> 5- CO 5- permit nutrients to reach the tumor cells and permit the tumor cells to enter the circulation, providing a survival advantage for the tumor. It follows that tumors with a higher density of microvessels should behave more aggressively, ie, metastasize earlier, and be associated with a poorer prognosis. Numerous studies assessed the prognostic value of microvessel counts in tumors. - The technique most commonly used was to count microvessels (<5 urn) in the areas of greatest density of vessels ("hot spots") in sections of tumors immunostained using primary antibodies to factor VIII, von Willebrand factor, or CD to enhance the vessels. Higher microvessel counts were associated with increased incidence of metastasis and tumor recurrence and shorter patient survival, suggesting that microvessel density is an independent predictor of metastasis, relapse, and survival, particularly in carcinomas of the breast, lung, prostate, and skin. Similar studies explored the prognostic value of microvessel counts in colorectal carcinoma, but the results conflict (Table ). Saclarides et al 7 counted vessels in microscopic fields in sections from 8 rectal carcinomas immunostained using antibody to factor VIII and found a significant association between higher angiogenesis scores and tumor invasion through bowel wall and death within 5 years. There was a trend toward increased microvessel density and increasing tumor size and incidence of metastasis. This study did not control for tumor stage. Two years later, Frank et al studied 5 cases of colorectal carcinoma, all Astler-Coller B, and found a significant association between higher angiogenesis score and death within 5 years and between higher angiogenesis score and tumor recurrence. TABLE. SUMMARY OF STUDIES OF MICROVESSEL COUNTS (MVCS) AS A PROGNOSTIC FACTOR IN COLORECTAL CARCINOMA Source Saclarides et al 7 Frank et al Takebayashi et al 8 Lindmark et al Engel et al 9 Tanigawa et al Present study Short term Long term -9 - -7 8- -5-9 Mean Microvascular Counts FIG. The distribution of the means of the microvessel counts in the highest areas per case shown as the percentage of cases in each group with the mean count in the incremental count ranges. Although not statistically significant, a tendency was noted for the counts to be higher in the long-term survivor group. DISCUSSION Tumor angiogenesis refers to a proliferation of new vessels, presumably of host origin, around and within a tumor. The beneficial effects of this neovascularization in promoting tumor growth and metastasis have been well established for more than years., Angiogenic factors secreted by tumor and host cells stimulate the formation and proliferation of new vessels into and within the tumor. These vessels are derived from host tissue around the tumor, similar to granulation tissue in a wound. Although derived from host tissue, they retain their own functional characteristics. For example, tumor vessels are leaky. Ultrastructural studies have shown that tumor vessels have fragmented basement membranes. This should N 8 5 7* 5 Site Rectum Colon Stage Astler-Coller A-D Astler-Coller B T NO MO Correlations High MVC correlated with higher stage, poor survival, larger tumor size, and metastases High MVC correlated with survival of <5 y and recurrence High MVC correlated with larger tumor size, relapse, lymph node metastasis, vessel invasion, and poor survival High MVC correlated with longer survival; no correlation between MVC and tumor differentiation or Dukes' stage High MVC correlated with recurrence High MVC correlated with hematogenous metastasis, grade, depth invasion, lymphatic invasion, lymph node metastasis, and peritoneal metastasis Trend toward higher MVC with longer survival Downloaded from https://academic.oup.com/ajcp/article-abstract/9//7/7579 by guest on January 9 Biopsy, ; resection, 5. AJCP June 998
BANNER ET AL 77 Microvessels in Carcinoma Engel et al 9 studied the role of microvessel counts in predicting recurrence and time to recurrence in 5 patients, with a recurrence and with no recurrence. Microvessels were counted in areas of the tumor with highest vessel density. By using statistical models, the authors found that when controlled for stage, a tumor microvessel count of 5 or more per HPF significantly predicted tumor recurrence. Takebayashi et al 8 found a correlation between higher microvessel count and larger tumor size, shorter survival, and tumor relapse in patients with resected colorectal carcinomas, including Dukes' stages A, B, and C. Most recently, Tanigawa et al performed a detailed statistical analysis of prognostic factors in colorectal carcinomas. When the tumors were stratified into hypovascular and hypervascular groups based on median vessel counts greater or less than 5 per HPF, there was a highly significant association between microvessel count and hematogenous metastasis. Association with lymph node and peritoneal metastasis was not as strong. The hypovascular and hypervascular groups also differed in tumor grade, depth of penetration, and lymphatic invasion. All of these studies found a correlation between higher counts and more aggressive behavior. In 99, controversy was introduced by the study results of Lindmark et al who found higher microvessel counts in tumors from patients with longer survival. They counted microvessels in frozen sections of full-thickness biopsy specimens of colorectal carcinomas, immunostained using antibody to von Willebrand factor. There was no correlation between the number of microvessels and tumor differentiation or Dukes' stage. However, higher microvessel counts were found in the group with longer survival. The reasons for the discrepancy between the results of this study and previous studies is unclear. Although previous studies suggested that microvessel counts are an independent prognostic factor, this may not be the case. Microvessel counts may be linked to other factors, such as tumor size, stage, presence of ulceration, and necrosis. In the study by Lindmark et al, % of the tumors were Dukes' A or B, while in the study by Saclarides et al, 7 5% of the tumors were Astler-Coller A or B, raising the question of whether the higher proportion of tumors with a favorable stage in the study by Lindmark et al might have contributed to the difference in the trends detected in the microvessel counts. The aim of our study was to address the whether microvessel counts may be a useful independent predictor of survival in patients with node-negative colorectal carcinoma by comparing microvessel counts in tumors from patients at the same stage at resection but with widely different survival times. Patients with T NO MO (Astler-Coller B) carcinomas were chosen for study because patients with B tumors have the widest range of survival times, from % to 8% at 5 years. In addition to tumor stage, our patient groups were similar in age and sex and location, size, and grade of the tumors. Lymphovascular invasion and the presence of a mucinous component were slightly more frequent in the short-term survivors, but the number of cases with these findings in each group was too small for statistical analysis. Our findings concur with those of Lindmark et al in that microvessel counts were higher in the cases with long-term survival. However, because there was considerable overlap in the counts between the long- and short-term survivors and the difference in counts between the two groups was not statistically significant, we conclude that microvessel count may not be a reliable independent predictor of survival in individual patients with T NO MO tumors. Acknozvledgments: We are grateful to Susan O'Hara, Director, University of Massachusetts Medical Center Tumor Registry, for identifying the T NO MO cases, and to Lou Savas for performing the immunostaining of the tissue sections. REFERENCES. Folkman J. What is the evidence that tumors are angiogenesis dependent? / Natl Cancer Inst. 989;8:-.. Weidner N. Tumor angiogenesis: review of current applications in tumor prognostication. Semin Diagn Pathol. 99;:-.. Weidner N, Folkman J, Pozza F, et al. Tumor angiogenesis: a new significant and independent prognostic factor in earlystage breast carcinoma. / Natl Cancer Inst. 99;8:875-887.. Staibano S, Boscaino A, Salvatore G, et al. The prognostic significance of tumor angiogenesis in nonaggressive and aggressive basal cell carcinoma of the human skin. Hum Pathol. 99;7:95-7. 5. DeVita VT, Hellman S, Rosenberg SA. Cancer: Principles and Practice of Oncology. rd ed. Philadelphia, Pa: Lippincott; 989:9.. Frank R, Saclarides TJ, Leurgans S, et al. Tumor angiogenesis as a predictor of recurrence and survival in patients with nodenegative colon cancer. Ann Surg. 995;:95-99. 7. Saclarides TJ, Speziale NJ, Drab E, et al. Tumor angiogenesis and rectal carcinoma. Dis Colon Rectum. 99;7:9-9. 8. Takebayashi Y, Akiyama S, Yamada K, et al. Angiogenesis as an unfavorable prognostic factor in human colorectal carcinoma. Cancer. 99;78:-. 9. Engel CJ, Bennett ST, Chambers AF, et al. Tumor angiogenesis predicts recurrence in invasive colorectal cancer when controlled for Dukes staging. Am ] Surg Pathol. 99;:-5.. Lindmark G, Gerdin B, Sundberg C, et al. Prognostic significance of the microvascular count in colorectal cancer. / Clin Oncol. 99;:^.. Tanigawa N, Amaya H, Matsumura M, et al. Tumor angiogenesis and mode of metastasis in patients with colorectal cancer. Cancer Res. 997;57:-. Downloaded from https://academic.oup.com/ajcp/article-abstract/9//7/7579 by guest on January 9 Vol.9 No.