Recognition of lymphatic tumor emboli on microscopic section is dependent on the pathologist. There is a potential for significant inter-observer variation in the diagnosis of LVI amongst pathologists. For instance, Gilchrist et al. [5] noted that when three surgical pathologists were told to assess for LVI in a pT1-2 N0 M0 histological mastectomy case, all three concurred in only 12 of 35 cases. One source of inter-observer variation is retraction artifact during histological processing that mimics true LVI. Though retraction artifact was previously considered an insignificant finding that interfered with the ability to make an appropriate diagnosis [6], more recent studies have suggested it is not a random phenomenon from tissue fixation. Instead, it may be a result of tumor-stromal interactions suggesting a poor prognosis [7,8]. The purposes of this article are to provide an illustrative description comparing true LVI to retraction artifact and provide insight of quantifying the significance of retraction artifact in future studies.

In breast stroma, cancer cells form in clusters near lymphatic vessels and as glands (Figure 1). The term "lymphatic vessel" refers to a lymph vessel or capillary; these vessels are endothelial-lined channels, devoid of red blood cells, and lacking a smooth muscle cell wall [2]. In true LVI, cancer cells break through the basement membrane. The lumenal invasion is a transient event; thus, it is rarely captured on histologic specimen. A tumor embolus forms in the lumen. On slide preparation, ideal factors favoring true LVI include: 1) a definite endothelial lining, with endothelial nuclei that seem to protrude into the lymphatic space; 2) invasion in one lymphatic vessel (LV) lumen with nearby cancer glands that have minimal or no retraction; 3) a tumor embolus in a LV clear lumen with outside nearby tumor bulk; 4) a tumor embolus that is different in shape than its surrounding clear LV space; and 5) a positive stain for fibrin, CD31, or CD34 on tumor embolus periphery [9,10].

Figure 2A is an illustration of the ideal characteristics favoring true LVI. Figure 2B shows pathological specimens of true LVI stained with hematoxylin and eosin (H&E, at ×40 magnification), and with CD31 or CD34 (both at ×20 magnification, with detail emphasizing the lymphatic space involved). The involved lymphatic space is often in immediate proximity of an artery and a nerve trunk, histologic features very helpful in identifying LVI with certainty, but depending on the plane of section not all three elements are always present on the slide. For example Figure 2 illustrates LVI and an adjacent artery, but no nerve trunk is present in the plane of section. Figure 3A is an illustration of the ideal characteristics that favor retraction artifact. Figure 3B shows three cases of pathological specimens of retraction artifact stained with H&E.

A variable degree of retraction artifact may be seen in 16% to 60% of invasive breast carcinomas [8,11]. Acs et al. [11] found that extensive retraction artifact stained with D2-40, especially in pT1 and pT2 breast carcinomas, correlates with tumor size, histologic type, histologic grade, presence of lymphatic spread, and poor outcome. Irie et al. [8] observed that retraction artifact was seen in 84% of cases of infiltrating ductal carcinoma and similarly suggested that retraction might represent true prelymphatic space involvement. Though the findings are interesting, they are limited. For example, in the Acs et al. [11] study, D2-40 may not have been an ideal marker for finding LVI, as D2-40 staining of myoepithelial cells and myofibroblasts at the edge of retraction spaces of ductal carcinoma in situ may be misinterpreted as tumor LVI [12], and the direct destruction of the endothelial cell layer by enzymatic digestion of matrix proteins by the neoplastic cells can result in lack of expression of D2-40 [8]. In the Irie et al. [8] study, retraction of >25% of the specimen was seen in 168 of 199 infiltrating ductal carcinoma specimens, but only 1 of 188 ductal carcinoma in situ specimens.

Thus, it is unclear how retraction artifact should be considered in cancer staging (e.g., T or N stage). Multiple stains (e.g., with type IV collagen, cytokeratin, and smooth muscle actin) have been recommended in LVI diagnosis [13,14]. A combination scoring system will likely be necessary in future studies of both LVI and retraction artifact to more clearly prognosticate the incidence of tumor metastasis. To further the assess the significance of retraction artifact, we recommend: 1) quantifying the instances of retraction artifact in one specimen; 2) noting tumor histology; 3) using more than one cellular marker in preparation; and 4) grading the size of the largest glands undergoing LVI, as these may be associated with a stronger force of tumor-stromal interaction.