Hemangiomas and Angiosarcomas of the Breast: Diagnostic Utility of Cell Cycle Markers With Emphasis on Ki-67

By Shin, Sandra J; Lesser, Martin; Rosen, Paul Peter

Context.-Vascular tumors comprise a minor subgroup of tumors arising in the breast and represent variants of hemangiomas and angiosarcomas. Diagnostic challenges may arise when differentiating hemangiomas from types I and II angiosarcomas. Ki-67 expression has been used as an adjunct to distinguish between benign and malignant lesions exhibiting histologic overlap at various anatomic sites.

Objective.-To investigate the utility of Ki-67 and other cell cycle regulatory proteins (S-phase kinase-associated protein 2 [Skp2], p27, and cyclin D1) in the differential diagnosis of mammary vascular lesions.

Design.-Thirty-four vascular tumors (21 hemangiomas and 13 angiosarcomas) of the breast were studied. The Ki-67 index and immunoreactivity for Skp2, p27, and cyclin D1 were determined in each case. Appropriate statistical methods were used.

Results.-The mean value of Ki-67 index was statistically different when comparing hemangiomas and angiosarcomas (P < .001). Angiosarcomas were typically positive for Skp2, whereas hemangiomas were negative (P < .001). Sensitivity and specificity cutoffs for Ki- 67 index to distinguish hemangiomas from angiosarcomas showed a candidate cutoff point of 175. The mean values of Ki-67 of lowgrade angiosarcomas were significantly different from all hemangiomas (P < .001) and also different from the subset of atypical hemangiomas (P = .02). Sensitivity and specificity cutoffs for Ki-67 index to distinguish all hemangiomas from low-grade angiosarcomas showed a candidate cutoff point between 150 and 175. Among angiosarcomas, positivity for Ki-67 was inversely related to that of p27 but not to Skp2 or cyclin D1. This was also true among hemangiomas.

Conclusions.-Ki-67 index can be used as a diagnostic tool to distinguish between benign and malignant vascular lesions of the breast. This can be particularly helpful in cases of histologic overlap such as low-grade angiosarcoma and hemangioma.

(Arch Pathol Lab Med. 2007;131:538-544)

Vascular tumors of the breast are uncommon and may be diagnosed at any age, consisting of variants of hemangioma or angiosarcoma. Histologically, mammary angiosarcoma is a heterogenous disease that spans a morphologic spectrum. The histopathologic classification of these tumors has been found to correlate with prognosis. Well- differentiated, type I tumors have the most favorable outcomes, and poorly differentiated, type III tumors have the worst outcomes.1,2 Although most hemangiomas can be easily distinguished from type III angiosarcomas, diagnostic challenges may arise when differentiating hemangiomas from types I and II angiosarcomas. Considerable morphologic overlap can be seen in benign and malignant vascular lesions of the breast, and individual features such as mitotic count are not diagnostically useful except to identify high-grade lesions.

Ki-67 expression has been used as an adjunct to distinguish between benign and malignant lesions exhibiting histologic overlap at various anatomic sites. For example, immunohistochemical stains Bcl-2 and Ki-67 are reportedly helpful in distinguishing basaloid carcinoma from benign basal cell lesions of the prostate gland.3 We investigated the utility of Ki-67 and other cell cycle regulatory proteins (S-phase kinase-associated protein 2 [Skp2], p27, and cyclin D1) in the differential diagnosis of mammary vascular lesions, with the expectation that angiosarcomas would display a significantly higher labeling index than hemangiomas.

MATERIALS AND METHODS

Thirty-four vascular tumors of the breast were retrieved from the pathology files at the New York Presbyterin Hospital-Weill Cornell Medical Center. The benign hemangioma group consisted of 13 typical hemangiomas, 6 atypical hemangiomas, 1 angiolipoma, and 1 angiomatosis.4-7 Four benign lesions were sampled by needle core biopsy; excisional biopsy was performed in the remaining lesions. All malignant vascular tumors were de novo angiosarcomas. Angiosarcomas arising in a setting of prior radiation therapy or postmastectomy lymphedema were excluded from this study. The malignant group consisted of 5 low-grade (type I), 3 intermediate- grade (type II), and 5 high-grade (type III) angiosarcomas. Grading was based on our previously published criteria.1 All malignant lesions were sampled by excisional biopsy.

Immunohistochemistry was performed on 4-m sections from a representative paraffin block containing lesional tissue using monoclonal antibodies against Ki-67 (Zymed Laboratories Inc, South San Francisco, Calif), Skp2 (clone 2C8D9, Zymed), Kip1/p27 (BD Transduction Laboratories, Lexington, Ky), and cyclin D1 (Lab Vision, Newmarket, Suffolk, United Kingdom). Appropriate positive (tonsil and normal breast) and negative controls were used. For Ki- 67, the number of nuclear positive cells was manually counted per 1000 lesional endothelial cells (Ki-67 index) using a BX41 Olympus microscope at 400 magnification. For Skp2, p27, and cyclin D1, a case was deemed positive if 10%, 50%, and 10% of lesional endothelial cells, respectively, were immunoreactive.

Ki-67 index was compared across the 5 histologic types using the Kruskal-Wallis nonparametric test. On finding a significant difference using the Kruskal-Wallis nonparametric test, Bonferroni- adjusted pairwise multiple comparisons were made using the Mann- Whitney test. Other 2-sample comparisons were carried out using the Mann-Whitney test. The Spearman rank correlation was used to estimate the correlation between Ki-67 and each Skp2, p27, and cyclin D1.

RESULTS

The average age of 14 women and 1 man with hemangiomas (13), angiolipoma (1), and angiomatosis (1) was 57 years (median, 54). Six women whose average age was 61 years (median, 58) had atypical hemangiomas. The average age of the women with high-grade and low- grade angiosarcomas was 53 and 39 years, respectively (median, 54 and 42, respectively). The average age of patients with intermediate- grade angiosarcoma (2 women, 1 man) was 49 years (median, 59). There was no predilection for either breast in any of the benign or malignant groups.

The mean, median, and standard deviation values for all hemangiomas and angiosarcomas and relative subgroups were calculated and compared (Table 1; Figure 1). The mean value of Ki-67 index was 46 and 381 in all hemangiomas and all angiosarcomas, respectively. This difference was highly significant (P < .001) (Figure 2; Figure 3, A and B; Figure 4, A and B; Figure 5, A and B). Angiosarcomas were typically positive for Skp2, whereas hemangiomas were negative (P < .001) (Figure 4, C and D; Figure 5, D). There were no statistically significant differences in immunoreactivity using markers p27 or cyclin D1 for either angiosarcomas or hemangiomas (Figure 5, C and E).

Sensitivity and specificity cutoffs for Ki-67 index to distinguish hemangiomas from angiosarcomas showed a candidate cutoff point of 175 at which the sensitivity and specificity were 90% and 95%, respectively (Table 2).

The mean values of Ki-67 were 34, 77, and 294 for all hemangiomas, atypical hemangiomas, and low-grade angiosarcomas (Table 1; Figure 3, A and B; Figure 6). With respect to Ki-67, low- grade angiosarcomas were significantly different from all hemangiomas (P < .001, Figure 7) and also different from the subset of atypical hemangiomas (P = .02, Figure 8).

Sensitivity and specificity cutoffs for Ki-67 index to distinguish all hemangiomas from low-grade angiosarcomas showed a candidate cutoff point between 150 and 175 at which the sensitivity and specificity ranged from 80% to 100% and 91% to 95%, respectively (Table 3). Similar cutoffs were also seen for distinguishing atypical hemangiomas from low-grade angiosarcomas.

Among angiosarcomas, positivity for Ki-67 was inversely related to that of p27 (Spearman correlation coefficient, -0.20) but not to Skp2 or cyclin D1 (Spearman correlation coefficients, 0.32 and 0.39, respectively). These relationships were not statistically significant. This was also true among hemangiomas (Spearman correlation coefficients for p27, Skp2, and cyclin D1 were -0.44, 0.39, and 0.29, respectively, all nonsignificant).

In rare instances, focal Ki-67 expression was relatively higher compared with other areas within the same case as well as to other hemangiomas. Review of the corresponding hematoxylin-eosin sections revealed that these areas were sites of prior biopsy or microthrombus formation (Figure 9, A through E).

COMMENT

Angiosarcoma of the breast accounts for 0.04% of malignant breast neoplasms and 8% of breast sarcomas.8 Nonetheless, angiosarcoma occurs in the female breast more commonly than any other site in the body.9 Until recently, angiosarcoma of the breast was regarded as almost always fatal. Subsequent studies have shown it to be a morphologically heterogeneous group of neoplasms in which grade is prognostically significant.1,2 The initial clinical finding is a palpable, painless mass. In some cases, no external features are evident although angiosarcomas that are particularly large or superficial can have blue or purple discoloration of the overlying skin. Mammary angiosarcomas are grossly typically larger than 2 cm, whereas most hemangiomas in the breast are smaller than 2 cm. However, there is some overlap in size with a few hemangiomas being larger th\an 2 cm. Increasing numbers of nonpalpable vascular lesions of the breast are being sampled by needle core biopsy. In this situation, tumor size is based on the radiologic impression, and the pathologist has only a limited sample to examine.

Benign vascular lesions of the breast are categorized into 6 major groups: perilobular hemangioma, hemangioma, angiomatosis, venous hemangioma, subcutaneous nonparenchymal hemangioma, and aneurysm.4-7 The term atypical hemangiomas has been used to describe benign vascular lesions with atypical cytologic features, including rare mitoses and/or nuclear pleomorphism, in the absence of endothelial growth characteristic for angiosarcoma.4 With adequate surgical samples, it is usually possible to distinguish hemangiomas from high-grade angiosarcomas, especially when lumpectomy specimens are available. Greater difficulty may be encountered with atypical hemangiomas and low-grade angiosarcomas, particularly in needle core biopsy material.

Ki-67 is a monoclonal antibody recognizing nuclear proteins of 345 kd and 395 kd in proliferating cells during the non-G0 phases of the cell cycle.

The utility of the Ki-67 immunohistochemical stain for assessing prognosis8,10,11 or as an ancillary diagnostic tool has been widely studied in other organ sites. Many have found it to be useful in the distinction between benign and malignant lesions possessing histologic overlap.3,12-17 The premise is that malignant lesions at a particular organ site show significantly higher proliferative activity than comparable benign tumors.

Only rare studies investigating Ki-67 in hemangiomas exist,18,19 and none, to our knowledge, has been performed on vascular lesions of the breast. In our study, we found that Ki-67 expression is significantly greater in angiosarcomas than in hemangiomas. This difference was also statistically significant when comparing histologically overlapping subgroups, namely distinguishing low- grade angiosarcomas from hemangiomas or a subset (atypical hemangiomas). We identified a candidate Ki-67 index cutoff of 175 to distinguish hemangiomas from angiosarcomas at which the sensitivity and specificity was 90% and 95%, respectively. A Ki-67 index cutoff for distinguishing low-grade angiosarcomas from hemangiomas or a subset (atypical hemangiomas) is of greater value to practicing pathologists. This value was also calculated and it was found to be similar to the cutoff that distinguishes hemangiomas from angiosarcomas with similar sensitivities and specificities. One drawback is the relatively small sample size, which limits the precision of the estimates of sensitivity and specificity and is largely because of the rarity of these lesions in the breast. Nonetheless, our results support that this difference can be used as a diagnostic aid in distinguishing between these 2 entities, especially in instances in which there is considerable morphologic overlap such as between some hemangiomas and low-grade angiosarcomas. It must be emphasized that increased Ki-67 immunoreactivity can be seen in a hemangioma at the site of a prior core biopsy as well as where there are thrombi. Hence, it is important to examine a concurrent hematoxylin-eosin section and these reactive regions should not be included in determining the Ki- 67 index.

The question arises as to whether Ki-67 can be used reliably in evaluating vascular lesions in needle core biopsy samples. Although our study did not specifically address this issue, we feel that it can be used to help direct further management on a case-by-case basis. For instance, the diagnosis of a hemangioma in the setting of appropriate clinical and/or radiologic findings can be reasonably made if the degree of Ki-67 positivity is appropriately very low. However, if the Ki-67 is increased in an identical setting, this may be reason to be more concerned about the possibility of angiosarcoma. Any vascular lesion encountered on a needle core biopsy that exhibits atypical morphologic features, regardless of the degree of Ki-67 expression, should be completely excised.

In our study, other cell cycle regulatory proteins were investigated using immunohistochemistry. p27 is a key regulator of G1 to S-phase progression. It prevents premature activation of cyclin E-cyclin-dependent kinase 2 in G1 and promotes the assembly and activation of D-type cyclin-dependent kinases.20 Although the p27 gene is rarely mutated in human cancers, the action of p27 is impaired through a variety of mechanisms including proteolysis. 20 Skp2 is required for the ubiquitin-mediated degradation of p27. Recent studies have shown that human cancers with high histopathologic tumor grade and a poor patient outcome demonstrate reduced p27 protein and/or increased Skp2 expression.21-23 In our study, we found Skp2 expression was significantly more common in angiosarcomas than in hemangiomas of the breast, a finding to our knowledge not previously reported. The absence of Skp2 expression can be used as further support of the diagnosis of hemangioma if it is used in conjunction with Ki-67. There were no statistically significant differences between benign and malignant mammary vascular lesions with respect to p27. Positivity for Ki-67 was not found to be inversely related to that of Skp2 in either angiosarcomas or hemangiomas. Furthermore, expression for Skp2 and p27 was weakly inversely correlated in both groups.

Cyclins are important protein molecules involved in the induction and control of the cell cycle. Factors modulating exit from G0 and progression through G1 phase of the cell cycle are critical in determining overall growth rates of cells. Cyclin D1 regulates G1 phase progression via cyclindependent kinase 4/6-dependent phosphorylation and partial inactivation of the repressive activity of the retinoblastoma protein.24 Overexpression of the cyclin D1 protein has been reported in a variety of malignant tumors and is associated with poor prognosis or shorter diseasefree survival.25,26 There were no statistically significant differences in expression between hemangiomas and angiosarcomas with respect to cyclin D1. No relationship between positivity for Ki-67 and cyclin D1 was observed in either angiosarcomas or hemangiomas. Cyclin D1 does not appear to have diagnostic utility in distinguishing benign and malignant vascular lesions of the breast.

The authors wish to thank Meredith Lukin, MS, for assistance with the statistical analysis and the following physicians for contributing material and/or clinical information: V. Agarwal, MD, R. Balaj, MD, J. Barnes, MD, J. Black, MD, M. Broscius, MD, W. Christensen, MD, A. Christiano, MD, E. Dellers, MD,M. Esposito, MD, K. Forstoefel, MD, R. Franklin, MD, J. Hennessy, MD, M. Hitchcock, MD, I. Hitti, MD, V. Hoon, MD, J. Huang, MD, L. Hussey, MD, H. Kim, MD, S. Kremen, MD, A. MacDonald, MD, M. Madigan, MD, T. Mervak, MD, L. Miller, MD, G. Moes, MD, M. Navarro, MD, G. Richard, MD, T. Saminathan, MD, K. Sunshine, MD, S. Thirumala, MD, G. Tillery, MD, J. White, MD, P. Wittenberg, MD, and V. Zemba-Palko, MD.

References

1. Donnell RM, Rosen PP, Lieberman PH, et al. Angiosarcoma and other vascular tumors of the breast: pathologic analysis as a guide to prognosis. Am J Surg Pathol. 1981;5:629-642.

2. Rosen PP, Kimmel M, Ernsberger D. Mammary angiosarcoma: the prognostic significance of tumor differentiation. Cancer. 1988;62:2145-2151.

3. Yang XJ, McEntee M, Epstein JI. Distinction of basaloid carcinoma of the prostate from benign basal cell lesions by using immunohistochemistry for bcl-2 and Ki-67. Hum Pathol. 1998;28:1447- 1450.

4. Hoda SA, Cranor ML, Rosen PP. Hemangiomas of the breast with atypical histological features: further analysis of histological subtypes confirming their benign character. Am J Surg Pathol. 1992;16:553-560.

5. Jozefczyk MA, Rosen PP. Vascular tumors of the breast, II: perilobular hemangiomas and hemangiomas. Am J Surg Pathol. 1985;9:491-503.

6. Rosen PP. Vascular tumors of the breast, III: angiomatosis. Am J Surg Pathol. 1985;9:652-655.

7. Rosen PP. Vascular tumors of the breast, V: non-parenchymal hemangiomas of mammary subcutaneous tissue. Am J Surg Pathol. 1985;9:723-729.

8. Liberman L, Dershaw DD, Kaufman RJ, et al. Angiosarcoma of the breast. Radiology. 1992;183:649-654.

9. Rosen PP. Sarcoma. In: Rosen’s Breast Pathology. 2nd ed. Philadelphia, Pa: Lippincott Williams & Wilkins; 2001:813-862.

10. Engellau J, Persson A, Bendahl P, et al. Expression profiling using tissue microarray in 211 malignant fibrous histiocytomas confirms the prognostic value of Ki-67. Virchows Arch. 2004;445:224- 230.

11. Wolfsberger S, Fischer I, Hftberger R, et al. Ki-67 immunolabeling index is an accurate predictor of outcome in patients with intracranial ependymoma. Am J Surg Pathol. 2004;28:914-920.

12. Lum CA, Binder SW. Proliferative characterization of basal- cell carcinoma and trichoepithelioma in small biopsy specimens. J Cutan Pathol. 2004;31:550-554.

13. Mayerhofer K, Lozanov P, Bodner K, et al. Ki-67 expression in patients with uterine leiomyomas, uterine smooth muscle tumors of uncertain malignant potential (STUMP) and uterine leiomyosarcomas (LMS). Acta Obstet Gynecol Scand. 2004;83:1085-1088.

14. Oda Y, Takahira T, Kawaguchi K, et al. Low-grade fibromyxoid sarcoma versus low-grade myxofibrosarcoma in the extremities and trunk: a comparison of clinicopathological and immunohistochemical features. Histopathology. 2004; 45:29-38.

15. Vang R, Barner R, Wheeler DT, et al. Immunohistochemical staining for Ki-67 and p53 helps distinguish endometrial arias- stella reaction from high-grade carcinoma, including clear cell carcinoma. Int J Gynecol Pathol. 2004;23:223-233.

16. Vargas MP, Vargas HI, Kleiner DE, et al. Adrenocortical neoplasms: role of prognostic markers MIB-1, p53, and Rb. Am J Surg Pathol. 1997;21:556-562.

17. Vollmer RT. Use of Bayes rule and MIB-1 proliferation index to discriminate Spitz nevus from mali\gnant melanoma. Am J Clin Pathol. 2004;122:499-505.

18. Dyduch G, Okon K. Proliferation and apoptosis within the oral mucosa “hemangiomas.” Pol J Pathol. 2000;51(2):93-96.

19. Dyduch G, Okon K, Mierzynski W. Benign vascular proliferations: an immunohistochemical and comparative study. Pol J Pathol. 2004;55(2):59-64.

20. Alkarain A, Jordan R, Slingerland J. p27 deregulation in breast cancer: prognostic significance and implications for therapy. J Mammary Gland Biol Neoplasia. 2004;9:67-80.

21. Hershko D, Bornstein G, Ben-Izhak O, et al. Inverse relation between levels of p27Kip1 and of its ubiquitin ligase subunit skp2 in colorectal carcinomas. Cancer. 2001;91:1745-1751.

22. Langner C, von Wasielewski R, Ratschek M, et al. Biological significance of p27 and Skp2 expression in renal cell carcinoma: a systemic analysis of primary and metastatic tumor tissues using a tissue microarray technique. Virchows Arch. 2004;445:631-636.

23. Sanada T, Yokoi S, Arii S, et al. Skp2 overexpression is a p27-independent predictor of poor prognosis in patients with biliary tract cancers. Cancer Sci. 2004;95:969-976.

24. Swanton C. Cell-cycle targeted therapies. Lancet. 2004;5:27- 36.

25. Kim SH, Lewis JJ, Brennan MF, et al. Overexpression of cyclin D1 is associated with poor prognosis in extremity soft-tissue sarcomas. Clin Can Res. 1998;4:2377-2382.

26. Lopez-Beltran A, Luque RJ, Alverez-Kindelan J, et al. Prognostic factors in survival of patients with stage Ta and T1 bladder urothelial tumors: the role of G1-S modulators (p53, p21Waf1, p27Kip1, cyclin D1, and cyclin D3), proliferation index, and clinicopathologic parameters. Am J Clin Pathol. 2004;122:444- 452.

Sandra J. Shin, MD; Martin Lesser, PhD; Paul Peter Rosen, MD

Accepted for publication September 8, 2006.

From the Department of Pathology and Laboratory Medicine, New York Presbyterian Hospital-Weill Medical College of Cornell University, New York (Drs Shin and Rosen); and the Biostatistics Unit, Feinstein Institute for Medical Research, North Shore-LIJ Health System, Manhasset, NY (Dr Lesser).

The authors have no relevant financial interest in the products or companies described in this article.

The findings in this study were presented at the 94th Annual Meeting of the United States and Canadian Academy of Pathology, San Antonio, Tex, February 28-March 4, 2005.

Reprints: Sandra J. Shin, MD, Department of Pathology and Laboratory Medicine-Starr 1009, New York Presbyterian Hospital- Weill Medical College of Cornell University, 525 E 68th St, New York, NY 10021 (e-mail: sjshin@med.cornell.edu).

Copyright College of American Pathologists Apr 2007

(c) 2007 Archives of Pathology & Laboratory Medicine. Provided by ProQuest Information and Learning. All rights Reserved.