1. ¹Translational Cancer Research Group, Lab Pathology University of Antwerp/University Hospital Antwerp, Antwerp, Belgium
2. ²Translational Cancer Research Group, Oncology Center, General Hospital St-Augustinus, Oosterveldlaan 24, B-2610 Wilrijk, Belgium

Correspondence: Dr PB Vermeulen, Department of Pathology, Oncology Center, General Hospital St Augustinus, Oosterveldlaan 24, B-2610 Wilrijk, Belgium. E-mail: peter.vermeulen@gvagroup.be, URL: www.tcrg.be

Received 6 January 2006; Revised 4 April 2006; Accepted 5 April 2006; Published online 2 May 2006.

Abstract

Recently, peritumoural (lympho)vascular invasion, assessed on haematoxylin–eosin (HE)-stained slides, was added to the St Gallen criteria for adjuvant treatment of patients with operable breast cancer (BC). New lymphatic endothelium-specific markers, such as D2-40, make it possible to distinguish between blood (BVI) and lymph vessel invasion (LVI). The aim of this prospective study was to quantify and compare BVI and LVI in a consecutive series of patients with BC. Three consecutive sections of all formalin-fixed paraffin-embedded tissue blocks of 95 BC resection specimens were (immuno)histochemically stained in a fixed order: HE, anti-CD34 (pan-endothelium) and anti-D2-40 (lymphatic endothelium) antibodies. All vessels with vascular invasion were marked and relocated on the corresponding slides. Vascular invasion was assigned LVI (CD34 or /D2-40) or BVI (CD34/D2-40) and intra- (contact with tumour cells or desmoplastic stroma) or peritumoural. The number of vessels with LVI and BVI as well as the number of tumour cells per embolus were counted. Results were correlated with clinico-pathological variables. Sixty-six (69.5%) and 36 (37.9%) patients had, respectively, LVI and BVI. The presence of 'vascular' invasion was missed on HE in 20% (peritumourally) and 65% (intratumourally) of cases. Although LVI and BVI were associated intratumourally (P=0.02), only peritumoural LVI, and not BVI, was associated with the presence of lymph node (LN) metastases (p_peri=0.002). In multivariate analysis, peritumoural LVI was the only independent determinant of LN metastases. Furthermore, the number of vessels with LVI was larger than the number of vessels with BVI (P=0.001) and lymphatic emboli were larger than blood vessel emboli (P=0.004). We demonstrate that it is possible to distinguish between BVI and LVI in BC specimens using specific lymphatic endothelium markers. This is important to study the contribution of both processes to BC metastasis. Furthermore, immunohistochemical detection of lymphovascular invasion might be of value in clinical practice.