Laboratory Investigation (2017) 97, 118–129 © 2017 USCAP, Inc All rights reserved 0023-6837/17 $32.00

PATHOBIOLOGY IN FOCUS

Lymphatic invasion and angiotropism in primary cutaneous Andrea P Moy, Lyn M Duncan and Stefan Kraft

Access of melanoma cells to the cutaneous vasculature either via lymphatic invasion or angiotropism is a proposed mechanism for . Lymphatic invasion is believed to be a mechanism by which melanoma cells can disseminate to regional lymph nodes and to distant sites and may be predictive of adverse outcomes. Although it can be detected on hematoxylin- and eosin-stained sections, sensitivity is markedly improved by immunohistochemistry for lymphatic endothelial cells. Multiple studies have reported a significant association between the presence of lymphatic invasion and sentinel lymph node metastasis and survival. More recently, extravascular migratory metastasis has been suggested as another means by which melanoma cells can spread. Angiotropism, the histopathologic correlate of extravascular migratory metastasis, has also been associated with melanoma metastasis and disease recurrence. Although lymphatic invasion and angiotropism are not currently part of routine melanoma reporting, the detection of these attributes using ancillary immunohistochemical stains may be useful in therapeutic planning for patients with melanoma. Laboratory Investigation (2017) 97, 118–129; doi:10.1038/labinvest.2016.131; published online 19 December 2016

Primary cutaneous melanoma has a propensity for metastasis, However, despite multiple studies, controversy remains both to lymph nodes and to distant sites, which is also the regarding the clinical relevance of lymphatic invasion in main cause of mortality in patients with melanoma. Sentinel melanoma. As yet, it is not a criterion for pathologic staging, lymph node (SLN) metastasis is one of the most powerful although the presence of lymphatic invasion may prompt SLN predictors of survival for patients with melanoma and in patients with thin melanoma.7 intravascular spread of tumor cells within lymphatics is Extravascular migratory metastasis (EVMM), in which widely accepted as a key mechanism of metastasis.1,2 Multiple tumor cells migrate along the external surfaces of vessels, is prognostic factors, including tumor thickness, mitotic rate, another potential mechanism of spread of melanoma to and ulceration, are incorporated into the current American nearby or distant sites.8 Angiotropism, believed to be the Joint Committee on Cancer melanoma staging system and histopathologic correlate of EVMM, has been defined by used to guide recommendations regarding which patients melanoma cells cuffing the endothelium of vessels in a should undergo SLN biopsy. However, these factors are pericytic location, located at the advancing front of the tumor limited in their ability to reliably predict which patients will mass. Although the presence of angiotropism within a develop metastasis in the SLN basin, disease progression, or primary melanoma has been associated with melanoma – recurrence, distinguishing them from patients cured by metastasis,9 11 the frequency of angiotropism in , excision of the primary cutaneous tumor. Thus, many its relationship to intravascular and lymphatic invasion, as investigators have studied additional histopathologic para- well as its clinical significance remain unclear. meters with the aim of identifying melanoma biomarkers associated with aggressive disease. DETECTION OF LYMPHATIC INVASION The prognostic value of lymphatic invasion, defined as Lymphatic invasion in primary cutaneous melanomas can be tumor cells located within lymphatic vessels, in melanoma has difficult to detect on hematoxylin- and eosin (H&E)-stained been the subject of much investigation recently, as immuno- sections alone. Foci of intralymphatic tumor cells may be histochemical markers of lymphatic endothelium have misinterpreted for multiple reasons. For example, tumor become available. In addition, lymphatic invasion has been emboli completely filling or distending lymphatic vessels may shown to be an unfavorable factor in other cancer types.3–6 compress and obliterate the lymphatic endothelial cells,

Dermatopathology Unit, Service, Massachusetts General Hospital, Boston, MA, USA Correspondence: Dr S Kraft, MD, Dermatopathology Unit, Pathology Service, Massachusetts General Hospital, 55 Fruit Street, Warren 829A, Boston, MA 02114, USA. E-mail: [email protected] Received 3 August 2016; revised 24 October 2016; accepted 8 November 2016

118 Laboratory Investigation | Volume 97 February 2017 | www.laboratoryinvestigation.org PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

leading to a false-negative interpretation. Alternatively, some using IHC for both S100 and D2-40 has been most studied artifacts may lead to a false-positive interpretation. For and rates of detection in melanoma have ranged from 33 to instance, stromal retraction artifact adjacent to tumor cells 47% of cases.26–29 Xu et al28 showed that the fraction of can mimic a lymphatic space. Infolding of the wall of melanomas with lymphatic invasion detected using dual IHC lymphatic vessels with abluminal tumor cells may simulate increased with increasing tumor thickness and mitotic rate; intralymphatic tumor depending on the plane of section interestingly, lymphatic invasion was identified in 32% examined. In addition, tumor cells may be artificially (40/125) of thin vertical growth phase (VGP) melanomas introduced into lymphatic spaces during the surgical measuring o1.0 mm in thickness. The utility of other procedure or tissue processing. Thus, careful histopathologic endothelial cell or melanoma markers has also been evaluation is essential. In a multicenter study, Egger et al12 investigated. Sahni et al30 used dual S100 and LYVE IHC to evaluated melanomas 41.0 mm in thickness from 2183 detect definite lymphatic invasion in 17% (6/36) of patients and reported lymphovascular invasion in 7.8% of melanomas, none of which showed lymphatic invasion on cases, as detected on H&E-stained sections. Nagore et al13 initially stained H&E sections. Feldmeyer et al31 used double identified lymphatic invasion in 5.7% (47/823) of melanomas IHC with MiTF, a nuclear melanocytic marker, and D2-40 to (see Table 1 for an overview of studies). identify lymphatic invasion in 38% (21/56) of cases, as The detection of lymphatic invasion is enhanced with compared with 22% (13/64) with D2-40 IHC alone and immunohistochemistry (IHC) using antibodies that are overall 4% (5/120) with H&E staining. specific to the lymphatic endothelium, such as lymphatic Although there is currently no gold standard for the vessel endothelial hyaluronan receptor 1 (LYVE-1) and detection of lymphatic invasion, in our experience, dual IHC podoplanin. LYVE-1 binds to hyaluronan on the luminal with S100 and D2-40 is most helpful in the identification of surface of lymphatic vessels and is absent on blood vessels.14 lymphatic invasion in primary cutaneous melanomas The monoclonal antibody D2-40 detects an epitope on (Figure 1). We recently studied 105 melanomas with dual podoplanin, a transmembrane sialoglycoprotein, and is highly S100/D2-40 IHC and detected lymphatic invasion in 23%.32 sensitive and specific for the endothelium of lymphatic Unequivocal lymphatic invasion can be recognized when vessels.15–18 With the use of LYVE-1 or D2-40 IHC, Doeden S100-positive cytologically atypical cells, stained with a red et al19 detected lymphatic invasion in 16% (15/94) of chromogen, are seen circumferentially surrounded by a thin melanomas studied. Fohn et al20 evaluated 64 melanomas D2-40-positive lymphatic endothelium, stained with a brown measuring 2.0 mm or less in thickness and reported a chromogen. Either solid tumor emboli or single tumor cells sevenfold increase in the detection of lymphatic invasion may be seen within the lymphatic spaces. In contrast, it can be with the use of D2-40 IHC, from 3.1% detected on H&E- difficult to assess the precise identity of cells within stained sections at the time of biopsy to 21.9% detected in lymphatic vessels on D2-40 staining only, as histiocytes and retrospective review with the use of IHC. Additional studies other inflammatory cells may be mistaken for melanocytic using D2-40 IHC have detected lymphatic invasion in cells. Although not specific for melanoma, S100 is a useful 15–33% of cases (Table 1).21–25 Of note, Petersson et al23 stain to aid in this task, as it is most sensitive melanocytic did not detect lymphatic invasion in any of the four marker. In addition, dual staining with S100 and D240 may desmoplastic melanomas nor any of the 29 melanomas also increase specificity, as some of the artifacts seen on o1.0 mm in thickness that were studied. H&E-stained sections can be detected. For example, infolding In addition to highlighting foci of lymphatic invasion, IHC of abluminal tumor cells may be detected with D24-0 staining can confirm the identity of a vessel as lymphatic, as blood even when no endothelial cell nuclei can be seen. However, as vessel invasion may be mistaken for lymphatic invasion on occasional histiocytic and dendritic cells may express S100, H&E-stained sections. Although few studies investigating the IHC evaluation may lead to a false-positive interpretation. presence and clinical significance of blood vessel invasion Therefore, intralymphatic cells must display nuclear have been done, it has been shown that blood vessel invasion enlargement and atypia to diagnose lymphatic invasion of detected using immunohistochemical markers specific for melanoma. The use of MiTF to mark melanoma cells may blood vessel endothelium, such as CD31, occurs more rarely offer some advantage in this assessment, as it highlights the than lymphatic invasion in primary cutaneous melanomas, at nuclear size; however, it is less sensitive than S100 and may a rate of 2–3%.19,24 In a study evaluating melanomas also stain histiocytic cells. Melan A or Mart1 may be measuring at least 0.75 mm in thickness with D2-40 and considered in lieu of S100 but, similar to MiTF, are not as CD34 IHC, Storr et al25 found that 25.5% (47/184) only had sensitive for melanocytic cells. False labeling of cells lymphatic invasion, whereas 2.7% (5/184) had blood vessel surrounding melanoma cells as lymphatic endothelium may invasion in addition to lymphatic invasion and 2.7% (5/184) also lead to a misinterpretation. In addition, although artifacts only had blood vessel invasion. introduced during tissue manipulation are an important Dual IHC with markers for melanocytic cells and lymphatic cause of false-positive interpretation of lymphatic invasion, endothelial cells has been shown to be even more helpful in the use of IHC may not allow such foci to be definitively the identification of foci of lymphatic invasion. The utility of distinguished from true foci of lymphatic invasion. Thus,

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 119 PMoy melanoma in AP angiotropism and invasion Lymphatic 120 Table 1 Studies evaluating lymphatic invasion in primary cutaneous melanoma

Study Number of cases Stains evaluated LI on H&E LI on IHC LI and SLN Metastasis LI and prognosis al et

Nagore et al13 823 ≥ 1.0 mm H&E 6% NA — Significantly associated with DFS and OSa Egger et al12 2183 ≥ 1.0 mm H&E 8% NA Significantly associateda Significantly associated with OSb Significantly associated with OS in cases with regressionc Significantly associated with DFS in patients ≥ 50 yearsc Petitt et al26 27 H&E, D2-40 4% 37% NS — Sahni et al30 36 H&E, S100/LYVE-1 0% 17% —— Niakosari et al21 4440.75 mm H&E, D2-40 0% 16% —— Fohn et al20 64 ≤ 2.0 mm H&E, D2-40 3% 22% Significantly associateda — Petersson et al23 74 H&E, D2-40 0% 23% Significantly associatedb Significantly associated with DFS and OSa Niakosari et al22 96 ≥ 1.0 mm H&E, D2-40 — 33% Significantly associateda — Aung et al29 101 ≥ 1.0 mm H&E, S100/D2-40 5% 47% —— aoaoyIvsiain|Vlm 7Fbur 07| 2017 February 97 Volume | Investigation Laboratory Moy et al32 105 H&E, S100/D2-40 — 23% Significantly associatedb Significantly associated with adverse outcomeb Xu et al27 106 H&E, S100/D2-40 5% 33% — Significantly associated with DFSb Feldmeyer et al31 120 H&E (n = 120) D2-40 (n = 64) 6% (D2-40) 22% (D2-40) Significantly associatedb NS with OS MiTF/D2-40 (n = 56) 2% (D2-40/MiTF) 38% (D2-40/MiTF) Xu et al28 251 VGP present H&E, S100/D2-40 5% 43% — Significantly associated with DFS (overall and in subset analysis of cases o 1.0 mm)a Yun et al39 321 VGP present H&E, S100/D2-40 — 22% (n = 116 with — Significantly associated with DFS in regression) cases with regressiona Doeden et al19 94 H&E, D2-40, LYVE-1, CD31 6% 16% Significantly associatedb NS with DM, DFS, or OS 25 www.laboratoryinvestigation.org Storr et al 202 ≥ 0.75 mm H&E, D2-40, CD34 8% 28% NS NS with DM, DFS, and OS AHBOOYI FOCUS IN PATHOBIOLOGY Rose et al24 246 H&E, D2-40, CD31, S100/D2-40 3% 18% NS NS with DM (n =3) Significantly associated with DFS and OSb

Abbreviations: —, not studied; DFS, disease-free survival; DM, distant metastases; LI, lymphatic invasion; NS, not significant; OS, overall survival; SLN, sentinel lymph node. Selection criteria including tumor thickness and presence of vertical growth phase (VGP) are indicated under the case numbers. Statistical analysis is indicated as follows. aBoth univariate and multivariate analysis. bUnivariate analysis. cMultivariate analysis. PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

Figure 1 Although lymphatic invasion is not definitively identified on H&E-stained sections (a, × 100; b, × 200), dual immunohistochemistry with S100 and D2-40 highlights foci of unequivocal lymphatic invasion (c, × 100; d, × 200).

similar to H&E analysis, the interpretation of dual IHC LYMPHATIC INVASION AND METASTASIS analysis also requires careful evaluation. Although we favor The presence of lymphatic invasion in primary cutaneous the use of S100 with D240, the combination of any melanoma is associated with other adverse prognostic factors melanocytic marker along with D240 is a more sensitive and portends an increased risk of metastasis. In a study of and specific means of identifying lymphovascular invasion 2183 patients with melanoma 41.0 mm in thickness, Egger than with H&E stain alone or with only and endothelial et al12 found that lymphatic invasion, as detected on marker. Monshizadeh et al33 studied concordance rates for H&E-stained sections alone, was associated with increasing various histopathologic parameters between external pathol- tumor thickness, the presence of ulceration, and histopatho- ogists and pathologists from a specialized melanoma logic subtype other than superficial spreading. Importantly, a reference center in Australia, presumably on H&E sections multivariate analysis showed that lymphatic invasion was alone. They found only moderate concordance rates associated with an increased risk of SLN metastasis. This (κ = 0.41–0.60) for vascular invasion. Although concordance finding suggests that the detection of lymphatic invasion for the absence of vascular invasion was high (94.3%), within a primary melanoma may be a clue to the propensity concordance for the presence of vascular invasion was low for the tumor to metastasize to regional lymph nodes, and has (44.4%). To our knowledge, the interobserver variability of been supported by studies using IHC to identify lymphatic the detection of lymphatic invasion in melanoma on invasion. IHC-stained sections has not been previously studied. It Xu et al27 studied 106 melanomas with VGP and showed a would be interesting to determine of concordance rates are significantly shorter time to regional nodal disease in cases higher when IHC stains are used. with lymphatic invasion identified on dual S100/D2-40 IHC,

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 121 Lymphatic invasion and angiotropism in melanoma PATHOBIOLOGY IN FOCUS AP Moy et al

as compared with cases without lymphatic invasion. In a significantly associated with SLN metastasis, this association larger study of 251 melanomas with VGP, Xu et al28 also did not hold true in a multivariate analysis when compared showed a significant difference in tumor thickness, presence with tumor thickness, ulceration, patient age, and patient of ulceration, and mitotic rate between cases with and without gender. Although some studies showed association of lymphatic invasion on S100/D2-40 IHC; the proportion of lymphatic invasion with SLN positivity, others found no patients with lymphatic invasion increased with increasing association.24–26 One possible explanation for the variation in tumor thickness and mitotic rate. In a multivariate analysis, reported clinical significance of lymphatic invasion may be lymphatic invasion was significantly associated with metas- related to differences in the amount of lymphatic invasion tasis in the first 10 years after treatment, both overall and in a detected in the cases studied. To our knowledge, no study has subset analysis of thin (o1.0 mm) melanomas. Petersson specifically evaluated the relationship between the burden of 23 et al also showed that more patients with lymphatic invasion lymphatic invasion present within a patient’s primary had distant metastasis or regional recurrence (53%) as melanoma, ie, number of lymphatics containing tumor cells compared with those without lymphatic invasion (11%). identified and/or number of tumor cells within lymphatics, 4 In an initial study of 44 melanomas measuring 0.75 mm and disease progression. et al21 in thickness, Niakosari found no significant association Of particular interest are thin melanomas, as this subset between lymphatic invasion on D2-40 IHC and patient age, accounts for most new cases of primary cutaneous tumor site, histopathologic subtype, Clark level, tumor melanoma.34 Thin melanomas are generally associated with thickness, mitotic rate, ulceration, regression, satellite lesions, a good prognosis; however, survival is variable as 20% are or precursor lesions. This study was likely underpowered associated with metastasis and 5% prove to be fatal.35 given the small number of patients studied. However, in a Notably, recent studies have shown that nearly all the thin 4 subsequent study of 96 melanomas measuring 1.0 mm in melanomas that develop metastases are ≥ 0.75 mm in thickness, lymphatic invasion detected on D2-40 IHC was thickness.36 Therefore, despite the classification of melanomas significantly associated with deeper Clark level of invasion and measuring o1.0 mm in thickness as stage pT1, some now greater Breslow tumor thickness.22In addition, in a multi- recommend that only tumors with a thickness of o0.75 mm variate analysis, lymphatic invasion was associated with a be considered as ‘thin melanomas’.37 The role of SLN biopsy positive SLN: lymphatic invasion was identified in 65% for patients with melanomas ≤ 0.75 mm remains unclear. (15/23) of melanomas from patients with a positive SLN, as Although the procedure provides valuable prognostic infor- compared with 23.3% (17/73) of melanomas from patients mation that has a role in therapeutic planning, it is not with a negative SLN. Fohn et al20 study of 64 patients showed without risks and additional predictors of prognosis and SLN that melanomas measuring 2.0 mm or less in thickness with metastasis are still needed. Murali et al38 found that lymphatic lymphatic invasion on D2-40 IHC were more likely to have a 4 2 vs invasion was associated with SLN positivity in melanomas high mitotic rate ( 5 mitoses per mm ; 50% 4%), ≤ increased tumor thickness (mean thickness 1.59 vs 1.09 mm), 1.0 mm and suggested that SLN biopsy be considered for and a positive SLN (86% vs 12%). In fact, multivariate patients with lymphatic invasion within the primary tumor. et al28 analysis showed that lymphatic invasion was the greatest Xu used a multivariate logistic regression analysis of predictor of SLN status when compared with mitotic rate, 251 patients who developed metastasis in 10 years to create a tumor thickness, ulceration, age, and gender. In our recent prognostic tree to identify groups of patients with different analysis of melanomas from patients who underwent SLN risks; among thin melanomas, patients with T1b melanomas biopsy, we found the sensitivity and specificity of predicting and lymphatic invasion were shown to have a higher risk of SLN metastasis with lymphatic invasion on dual S100/D2-40 metastasis than those without lymphatic invasion. IHC were 44% and 82%, respectively.32 In addition, patients Although there are conflicting reports of whether lympha- without lymphatic invasion were more likely to be disease free tic invasion in primary melanomas is associated with worse at last follow-up (79%) than patients with lymphatic invasion prognosis, the detection of lymphatic invasion in ambiguous (54%) detected on S100/D2-40 IHC (average follow- melanocytic tumors has been proposed to be useful in up = 22 months and 24 months, respectively). identifying tumors capable of metastasis. In a study by Feldmeyer et al31 also detected a significant association Abraham et al39 of ‘melanocytic tumors of uncertain between SLN metastasis and lymphatic invasion on D2-40 malignant potential’, lymphatic invasion, not evident on IHC alone or in combination with MiTF staining. These H&E-stained sections, was observed in 25% (8/32) cases using authors found no difference in the utility of single or double dual S100/D2-40 IHC. Interestingly, the cases with lymphatic IHC for predicting SLN metastasis. Doeden et al19 showed invasion included two from patients who died of melanoma- that the using D2-40 and LYVE-1 IHC to detect lymphatic related disease and two with bulky nodal metastases. invasion increased the sensitivity (39% vs 13%), positive Although lymphatic invasion was associated with aggressive predictive value (75% vs 50%), and negative predictive value outcomes, including the development of nodal disease and (69% vs 61%) of predicting SLN metastasis. Although a distant metastasis, no significant associations were detected univariate analysis showed that lymphatic invasion was on multivariate analysis in this relatively small study.

122 Laboratory Investigation | Volume 97 February 2017 | www.laboratoryinvestigation.org PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

LYMPHATIC INVASION AND SURVIVAL Petersson et al23 showed that patients with intratumoral The detection of lymphatic invasion has also been associated lymphatic invasion had a hazard ratio of 9.0 (95% confidence with adverse outcomes in primary cutaneous melanoma. Xu interval 1.8–44.9) and those with peritumoral lymphatic et al27 showed a significant difference in melanoma-specific invasion had a hazard ratio of 6.1 (95% confidence interval death in cases with and without lymphatic invasion on dual 0.8–46.3). Prior investigations have suggested that intratu- S100/D2-40 IHC. Petersson et al23 showed that lymphatic moral lymphatics are not functional and that metastatic invasion, in addition to ulceration, was significantly associated dissemination occurs primarily via peritumoral lymphatics.42 with a shorter overall survival (OS). Additional studies have However, other studies reveal that chemokine receptors on also shown an association between lymphatic invasion and lymphatic endothelial cells can bind melanoma tumor cells both disease-free survival (DFS) and OS.13,24 Rose et al24 also and thus facilitate lymphatic dissemination.43 Potential demonstrated that patients with lymphatic invasion in greater differences between melanoma and non-melanoma cancers than one vascular channel had a worse DFS as compared to in the peritumoral and intratumoral location of prognostically those with only one involved vascular channel. Despite these relevant lymphatic invasion have yet to be explained, but may findings, a number of additional studies showed no associa- be related specific characteristics of melanoma vs many tion between the detection of lymphatic invasion in primary carcinomas, eg, the potential to metastasize at a very small size melanomas and survival (Table 1).19,25,31 or low tumor mass. Despite these findings, the clinical Egger et al12 detected a significant difference in OS only by significance of identifying lymphatic invasion within a tumor univariate analysis but showed that lymphatic invasion was mass vs at the periphery of a primary melanoma requires associated with a worse OS in patients with regression. further investigation. Interestingly, in a study of 116 melanomas with regression, Yun et al40 found that patients with lymphatic invasion in EXTRAVASCULAR MIGRATORY METASTASIS AND areas of complete regression had a shorter time to first ANGIOTROPISM metastasis and melanoma-specific death than those who did Intravascular dissemination of melanoma cells is now a well- not have lymphatic invasion, suggesting that the adverse accepted mechanism of spread beyond the primary tumor prognostic effect of regression may be in part mediated by site; however, other mechanisms of metastasis may exist. In lymphatic invasion. In a subsequent multi-institution study, 1829, the term ‘metastasis’ was first used by French physician Maurichi et al41 demonstrated that lymphatic invasion and Recamier and referred to the spread of cancer cells along the extensive regression were both independent predictors of OS. external surfaces of vascular spaces.44,45 The ability of melanoma cells to spread via the external surfaces of vessels LOCATION OF LYMPHATIC INVASION was subsequently discussed by Handley46 in 1907, in his paper In other cancer types such as carcinoma, lymphatic invasion regarding the surgical margins for melanoma. Although associated with poor prognosis often occurs in the stroma Handley believed that melanoma spread via intralymphatic surrounding invasive carcinoma. The location of lymphatic dissemination and the association of melanoma cells with vessels containing luminal tumor cells within a primary blood vessels is secondary to the proximity of blood vessels melanoma, ie, intratumoral or peritumoral, has been specified and lymphatics, he cited a prior paper by Borst, who noted in a few studies. Niakosari et al21 reported that 71% (5/7) of melanoma ‘spread along the perivascular tissues immediately cases with lymphatic invasion on D2-40 IHC had tumor outside the blood vessels’ as the tumor cells ‘are attracted emboli in lymphatics located within the tumor mass, one of towards the blood vessels by a kind of chemotaxis’. More which also had involved lymphatics at the advancing tumor recently, Lugassy and Barnhill have described the migration of edge; 29% (2/7) of cases had tumor emboli within melanoma cells along the abluminal surface of vessels by a subepidermal lymphatics. Although other studies also mechanism termed ‘extravascular migratory metastasis’ reported intratumoral lymphatic invasion,22 Petitt et al26 (EVMM).47 Angiotropism is considered the histopathologic reported that all cases with lymphatic invasion had positive correlate of EVMM and has been defined by aggregates of staining at the periphery of the melanoma on dual S100/D2- melanoma cells cuffing the external surface of blood vessels or 40 IHC. lymphatics within 1–2 mm of the advancing primary tumor Accordingly, the prognostic implications related to the front.8,9 specific location of lymphatic invasion in relationship to the The spread of tumor cells via EVMM was proposed based tumor remains to be determined. Storr et al25 showed that upon ultrastructural and immunohistochemical stains that intratumoral and peritumoral lymphatic invasion were both demonstrated the close approximation of melanoma cells to significantly associated with increased stage, increased tumor blood vessel endothelial cells, in a pericytic location.48–50 It thickness, presence of ulceration, and nodular histopathologic has been shown that melanoma cells are separated from subtype; however, only intratumoral lymphatic invasion was endothelial cells by an amorphous matrix containing ‘free’ associated with high mitotic rate and the presence of laminin, which is not incorporated into the basement microsatellites. With regard to disease-specific survival, as membrane, forming an ‘angio-tumoral complex’.48,49,51 The compared with patients without lymphatic invasion, presence of the C16 laminin peptide increases the distance of

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 123 Lymphatic invasion and angiotropism in melanoma PATHOBIOLOGY IN FOCUS AP Moy et al

angiotropic migration of melanoma cells in vitro.52 Further- melanoma cells, without associated intravasation; serpin B2 more, ex vivo and in vivo studies with time-lapse video expression was also noted in angiotropic human melanoma microscopy have documented the migration of melanoma brain metastases. Thus, it has been proposed that neutrophilic cells along the abluminal surface of vascular channels without inflammation present in ulcerated melanomas may contribute blood vessel invasion, arguing against angiotropism just to pericyte mimicry, angiotropism, and subsequent representing an early stage in the vascular invasion metastasis.59,68 process.53–57 It has been hypothesized that the interaction of melanoma DETECTION OF ANGIOTROPISM cells with the abluminal vascular surface recapitulates As the description by Handley,46 there have been reports of embryonic migration of neural crest cells through pericyte angiotropism in primary and metastatic melanoma, as mimicry.58–60 Prior studies have shown that neural crest cells detected on histopathologic evaluation.69–73 Barnhill et al74 migrate along the external surface of microvascular channels described angiotropism in 36 primary melanomas on H&E- during embryogenesis.61,62 Interestingly, angiotropism has stained tissue sections, as defined by unequivocal cuffing of been identified in many congenital nevi, suggesting that melanoma cells on the external surface of vessels in a linear EVMM may be an important mechanism for the development array or aggregates in at least two foci at the advancing front of these benign melanocytic lesions.63,64 Lugassy et al65 of the tumor and without evidence of intravascular tumor identified 128 genes via gene expression profiling that were cells. Of the 35 cutaneous melanomas studied, most (31) were differentially expressed in angiotropic and non-angiotropic invasive to anatomic level IV, two to level II, and two to level primary melanomas, of which 7 genes were associated with V; the average tumor thickness was 1.64 mm (range: 0.46– neural crest cell migration, including TCOF1 and AHNAK. 8.25 mm). One case of cervical mucosal melanoma, measur- Immunohistochemical studies showed overexpression and ing 35 mm in thickness, was also included in this evaluation. underexpression of the protein products of these two genes, In a retrospective review of H&E-stained sections, Hung respectively, among angiotropic melanomas as compared et al75 studied 20 primary cutaneous melanomas and their with non-angiotropic cases. In a subsequent study, Lugassy respective brain metastasis and identified angiotropism in 14 et al66 showed that the interaction between melanoma cells (70%) primary lesions and in 7 (35%) metastatic lesions. To and the abluminal surface of endothelial cells stimulated the our knowledge, no other study has been performed regarding differential expression of 28 genes, including genes involved the frequency with which angiotropism defined as such is in cancer cell migration (CCL2, ICAM1, and IL6), cancer present in melanomas. Although a large population-based progression (CCL2, ICAM1, SELE, TRAF1, IL6, SERPINB2, study of invasive melanoma would provide the most accurate and CXCL6), epithelial–mesenchymal transition (CCL2 and assessment of the frequency of angiotropism, such a study IL6), embryonic/stem cell properties (CCL2, PDGFB, EVX1, would require careful attention to the cases included. and CFDP1), neural crest development (CCL2, PDGFB, Specifically, primary excisional should be evaluated EVX1, and CDFP1), and pericytic recruitment (PDGFB). as these specimens best allow for the adequate evaluation of Additional immunohistochemical studies demonstrated angiotropism at the tumor periphery. expression of ICAM1 and PDGFRB by angiotropic melanoma Angiotropism has also been identified in locoregional cells, without staining in non-angiotropic melanoma cells.66 melanoma metastases, including microscopic satellites and in- Using a genetically engineered mouse model, it has been transit metastases,56,76 supporting the idea that EVMM is a shown that neutrophilic inflammation promotes pericytic possible mechanism of tumor spread. In fact, Gerami et al77 mimicry, angiotropism, and metastasis in primary cutaneous reported angiotropism in an epidermotropic melanoma melanoma.59 The authors demonstrated that repetitive metastasis and suggested that this histopathological finding ultraviolet irradiation to melanomas led to an increase in may be a useful marker in the distinction of epidermotropic local and systemic neutrophilic inflammatory response, which metastasis from locally recurrent melanoma with an epider- correlated with increased angiotropism and an increased mal component and multiple primary melanomas. All seven number of pulmonary metastases; depletion of neutrophils or cases of epidermotropic metastases studied by Lugassy et al56 inhibition of neutrophilic activation led to a decrease in showed angiotropism. angiotropism and metastasis. In vitro experiments revealed Although IHC for S100 and CD31 has been used to study that melanoma cells migrate towards endothelial cells when angiotropism in melanoma, these techniques show no incubated with neutrophil-conditioned media or media increase in sensitivity over H&E-stained sections.9,74 In a containing the cytokine tumor necrosis factor. Interestingly, study of melanoma metastases, Lugassy et al56 used dual IHC Lugassy et al66 identified that genes involved in inflammation for Melan A and CD31 to highlight vessels involved by (CCL2, IL6, TRAF1, CXCL6, SELE, ICAM1, SLC7A2, angiotropic melanoma, although did not investigate the utility C2CD4B, PDGFB, and SERPINB2) were expressed in of such staining to increase the sensitivity or specificity of melanomas as a result of the interaction of melanoma cells detection of angiotropism. and endothelial cells. More recently, Bentolila et al67 In our experience, angiotropism may be detected on H&E- described the perivascular migration of serpin B2-expressing stained sections of primary melanomas (Figure 2). However,

124 Laboratory Investigation | Volume 97 February 2017 | www.laboratoryinvestigation.org PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

Figure 2 Angiotropism can be identified at the advancing edge of the tumor on H&E-stained sections (a, × 100; b, × 200) and CD31 immunohistochemistry (c, × 100; d, × 200). *, vascular lumina; arrows, angiotropic melanoma cell aggregates.

the role of IHC for CD31 to facilitate the identification of recognize such foci, it may be important to set more specific angiotropic foci remains unclear. Although IHC for CD31 can guidelines for detection. Of note, one analysis of the be helpful in the recognition of angiotropic tumor foci, dual interobserver variability of detecting angiotropism on H&E- IHC with CD31 and a melanocytic marker, such as S100, may stained sections of 20 primary melanomas revealed an overall offer the further advantage of confirming that perivascular good agreement;78 however, additional studies are needed to cells are of melanocytic origin, since it can be difficult to be confirm that angiotropism can be identified consistently. certain on background hematoxylin staining. It is again important to note that angiotropism cannot be adequately ANGIOTROPISM AND MELANOMA PROGNOSIS assessed within primary melanomas on small biopsies as the Several studies have suggested that angiotropism is a deep margin of the specimen transects the tumor. In fact, prognostic biomarker in melanoma. In a study by Barnhill Van Es et al10 excluded two cases from their study, as the et al9 of 40 melanomas with metastasis to regional lymph advancing front of the tumor was not present on the biopsy nodes and/or visceral sites and 40 melanomas without specimen. It is also interesting to note that, although metastasis, angiotropism was detected more often in metas- angiotropism is located at the advancing front of the tumor tasizing melanoma. Definitive angiotropism (defined as two according to the definition proposed by Barnhill and or more foci) and equivocal angiotropsm (defined as one Lugassy,8 foci of melanoma cells adjacent to the abluminal focus) were identified in 40% (16/40) and 12.5% (5/40) of surface of endothelial cells may often also be seen within the metastasizing melanomas, respectively; only 15% (6/40) of tumor mass and the significance of such foci is unknown. As non-metastasizing melanomas had equivocal angiotropism the utility of angiotropism as a biomarker of melanoma and definitive angiotropism was not detected in any of the behavior rests on the ability of pathologists to appropriately non-metastasizing melanomas. Interestingly, the authors

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 125 Lymphatic invasion and angiotropism in melanoma PATHOBIOLOGY IN FOCUS AP Moy et al

noted that vascular invasion was only detected in one case of within a primary cutaneous melanoma is predictive of locally metastasizing melanoma with angiotropism. In a subset recurrent disease or a worse overall prognosis. analysis of 26 pairs of patients matched for patient age and The association of angiotropism and intravascular invasion gender and tumor anatomic site and thickness, angiotropism is not clear and further investigation is needed. Barnhill et al9 in the primary tumor was significantly associated with identified one metastatic melanoma with both vascular metastasis. Similarly, Van Es et al10 studied 32 patients with invasion and angiotropism. Van Es et al10 noted six primary primary cutaneous melanoma who developed local or in- cutaneous melanomas with lymphatic or blood vessel transit metastasis and 59 patients with primary cutaneous invasion, five of which were also noted to have angiotropism. melanoma without metastases to investigate whether the Understanding the relationship of EVMM and intravascular presence of angiotropism on H&E-stained sections predicts dissemination in melanoma might help to clarify which disease recurrence. In a blinded slide review, angiotropism patients are at risk for disease progression. was detected more commonly in primary melanomas that metastasized (63%) as compared with those that did not ANGIOTROPISM IN OTHER MELANOCYTIC TUMORS (37%; P = 0.02). In a logistic regression multivariate analysis, Foci of angiotropism within benign and atypical melanocytic only patient age and angiotropism remained statistically nevi have also been reported. Perivascular cells are significant predictive factors of local or in-transit metastasis. commonly recognized as a characteristic feature of congenital Factors significantly associated with angiotropism on uni- melanocytic nevi.80 Barnhill et al63 have reported a case of a variate analysis included greater tumor thickness, anatomic congenital nevus in a newborn that showed prominent level, mitotic rate, presence of ulceration, and absence of angiotropism on histopathologic analysis; IHC for CD34 and regression; there was no significant difference in anatomic CD31 highlighted foci of angiotropism, which were not site, histopathologic subtype, patient age or gender, or associated with D2-40-positive lymphatic vessels. In a neutropism among cases with and without angiotropism. systematic review, Kokta et al64 evaluated 53 congenital nevi There was a significant difference in the 5-year DFS for using H&E-stained sections as well as IHC for CD31 and patients with melanomas with and without angiotropism reported that 94% demonstrate angiotropism; intravascular (53% vs 81%, respectively). melanocytic cells were not seen. As mentioned above, the Wilmott et al11 studied patients with melanomas with and authors speculate that the development of congenital nevi without microscopic satellites and found that the presence of may be related to aberrant migration of neural crest cells angiotropism was significantly associated with microscopic along vascular channels. satellites, defined as a dermal/subcutaneous nodule measuring The biological behavior of atypical Spitz tumors, which are 40.05 mm and separated from the main tumor mass by often difficult to distinguish from melanoma on histopatho- ≥ 0.3 mm, as compared with those without (52% vs 27%, logical evaluation, has been a subject of debate. Although respectively). Anatomic level, absence of regression, and there is a high frequency of cases associated with SLN microscopic satellites were significantly associated with metastases, completion lymphadenectomy rarely detects angiotropism on univariate analysis. A multivariate binary additional metastases and the vast majority of patients do logistic regression analysis showed that the presence of not have further disease progression.79,81 As it was previously angiotropism, anatomic level V, absence of regression, and shown that angiotropism may be associated with local and an acral primary site were predictive of microscopic satellites. regional lymph node metastasis, Barnhill et al82 hypothesized However, no significant difference in survival was detected that EVMM is a potential mechanism of metastasis in atypical between cases with and without associated microscopic spitzoid tumors and studied H&E-stained sections of nine satellites. In a study of 20 primary cutaneous melanomas cases. The primary tumors had an average tumor thickness of with brain metastasis, Hung et al79 detected angiotropism in 3.54 mm (range: 0.66 mm to 45.35 mm) and the average 70% (14/20) of the primary tumors. The presence of mitotic rate was 2/mm2. Angiotropism, either in the initial angiotropism was significantly associated with greater tumor biopsy or subsequent excision, was seen in all primary thickness and mitotic rate. Although the authors noted that tumors, typically involving small blood vessels but occasion- the time to disease relapse was shorter in the cases with ally lymphatics; no tumor cells were seen within vascular angiotropism as compared to those without angiotropism (33 spaces. Four of five patients who underwent SLN sampling vs 57 months, respectively), there was no significant difference had a positive SLN. One patient with a positive SLN in survival. Angiotropism was identified in 35% (7/20) of the developed local recurrence ~ 2 years after the initial diagnosis metastatic tumors but no association between angiotropism was rendered; the lesions were notable for angiotropism and in the primary tumor and metastatic tumor was detected. interpreted to represent satellite metastases. However, despite Nevertheless, these studies all had some degree of inherent these findings, the utility of detecting angiotropism within bias, eg, patients selected with brain metastases or known atypical Spitzoid tumors remains unclear, as angiotropism microscopic satellites. Prospective studies without selection may be present in both benign congenital nevi and malignant bias may be helpful in determining whether angiotropism melanoma.83 It is possible that benign, atypical/low-grade,

126 Laboratory Investigation | Volume 97 February 2017 | www.laboratoryinvestigation.org PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

and malignant neoplasms are capable of spread via angio- 11. Wilmott J, Haydu L, Bagot M, et al. Angiotropism is an independent tropism/EVMM. predictor of microscopic satellites in primary cutaneous melanoma. Histopathology 2012;61:1889–1898. 12. Egger ME, Gilbert JE, Burton AL, et al. Lymphovascular invasion as a CONCLUSION prognostic factor in melanoma. Am Surg 2011;77:992–997. 13. Nagore E, Oliver V, Botella-Estrada R, et al. Prognostic factors in Both lymphatic invasion and, more recently, angiotropism localized invasive cutaneous melanoma: high value of mitotic rate, have been described in primary melanomas and are believed vascular invasion, and microscopic satellitosis. Melanoma Res 2005;15: to represent means of melanoma dissemination. Although few 169–177. 14. Banerji S, Ni J, Wang SX, et al. LYVE-1, a new homologue of the CD44 studies regarding the incidence of angiotropism in melanoma glycoprotein, is a lymph-specific receptor for hyaluoronan. J Cell Biol have been done, initial studies suggest that it is a common 1999;144:789–801. event in primary cutaneous melanomas.75 IHC may facilitate 15. Kahn HJ, Marks A. A new monoclonal antibody, D2-40, for detection of lymphatic invasion in primary tumors. Lab Invest 2002;82:1255–1257. the detection of both lymphatic invasion and angiotropism, 16. Evangelos E, Kyzas PA, Trikalinos TA. Comparison of the diagnostic but its utility remains uncertain and requires further accuracy of lymphatic endothelium markers: Bayesian approach. Mod Pathol 2005;18:1490–1497. investigation. Additional studies are also needed to clarify 17. Fukunaga M. Expression of D2-40 in lymphatic endotheilum the prognostic value of lymphatic invasion and angiotropism of normal tissues and in vascular tumors. Histopathology 2005;46: in predicting which patients are at risk for disease recurrence 396–402. 18. Schacht V, Dadras SS, Johnson LA, et al. Upregulation of the lymphatic or progression as the detection of primary tumor lymphatic marker podoplanin, a mucin-type transmembrane glycoprotein, in invasion and/or angiotropism may have an important impact human squamous cell carcinomas and germ cell tumors. Am J Pathol on therapeutic planning. For example, the identification of 2005;66:913–921. 19. Doeden K, Ma Z, Narasimhan B, et al. Lymphatic invasion in cutaneous lymphatic invasion, particularly in thin melanomas, may have melanoma is associated with sentinel lymph node metastasis. J Cutan an impact on decisions regarding whether or not to perform a Pathol 2009;36:772–780. SLN biopsy. The detection of angiotropism within a primary 20. Fohn LE, Rodriguez A, Kelley MC, et al. D2–40 lymphatic marker for detecting lymphatic invasion in thin to intermediate thickness melanoma may have an impact on the planning of surgical melanomas: association with sentinel lymph node status and margins.10 Furthermore, understanding the underlying mole- prognostic value – a retrospective case study. J Am Acad Dermatol cular mechanisms of intravascular dissemination and EVMM, 2011;64:336–345. 21. Niakosari F, Kahn HJ, Marks A, et al. Detection of lymphatic invasion in and the relationship between these processes, may also have primary melanoma with monoclonal antibody D2-40: a new selective therapeutic implications for melanoma patients. immunohistochemical marker of lymphatic endothelium. Arch Dermatol 2005;141:440–444. 22. Niakosari F, Kahn HJ, McCready D, et al. Lymphatic invasion identified DISCLOSURE/CONFLICT OF INTEREST by monoclonal antibody D2-40, younger age, and ulceration: The authors declare no conflict of interest. predictors of sentinel lymph node involvement in primary cutaneous melanoma. Arch Dermatol 2008;144:462. 23. Petersson F, Diwan AH, Ivan D, et al. Immunohistochemical detection 1. Balch CM, Soong SJ, Gershenwald JE, et al. Prognostic factors analysis of lymphovascular invasion with D2-40 in melanoma correlates with of 17,600 melanoma patients: validation of the American Joint sentinel lymph node status, metastasis and survival. J Cutan Pathol Committee on cancer melanoma staging system. J Clin Oncol 2001;19: 2009;36:1157–1163. 3622–3634. 24. Rose AE, Christos PJ, Lackaye D, et al. Clinical relevance of 2. Balch CM, Cascinelli N. Sentinel-node biopsy in melanoma. N Engl J detection of lymphovascular invasion in primary melanoma using Med 2006;355:1370–1371. endothelial markers D2-40 and CD34. Am J Surg Pathol 2011; 3. Schoppmann SF, Bayer G, Aumayr K, et al. Prognostic value of 35:1441. lymphangiogenesis and lymphatic invasion in invasive breast cancer. 25. Storr SJ, Safuan S, Mitra A, et al. Objective assessment of blood and Ann Surg 2004;240:306–312. lymphatic vessel invasion and association with macrophage infiltration 4. Walgenbach-Bruenagel G, Tolba RH, Varnai AD et al. Detection of in cutaneous melanoma. Mod Pathol 2012;25:493–504. lymphatic invasion in early stage primary colorectal cancer with the 26. Petitt M, Allison A, Shimoni T, et al. Lymphatic invasion detected by monoclonal antibody D2-40. Eur Surg Res 2006;38:438–444. D2-40/S-100 dual immunohistochemistry does not predict sentinel 5. Mohammed RA, Martin SG, Gill MS, et al. Improved methods of lymph node status in melanoma. J Am Acad Dermatol 2009; detection of lymphovascular invasion demonstrate that it is the 6:819–828. predominant method of vascular invasion in breast cancer 27. Xu X, Gimotty PA, Guerry DP, et al. Lymphatic invasion revealed by and has important clinical consequences. Am J Surg Pathol 2007;31: multispectal imagine is common in primary melanomas and associates 1825–1833. with prognosis. Hum Pathol 2008;39:901–909. 6. Mori D, Yamasaki F, Shibaki M, et al. Lateral peritumoral lymphatic 28. Xu X, Chen L, Guerry DP, et al. Lymphatic invasion is independently vessel invasion can predict lymph node metastasis in esophageal prognostic of metastasis in primary cutaneous melanoma. Clin Cancer squamous cell carcinoma. Mod Pathol 2007;20:694–700. Res 2012;18:229–237. 7. Coit DG, Thompson JA, Algazi A, et al. Melanoma, Version 2.2016, 29. Aung PP, Leone D, Feller JK, et al. Microvessel density, lymphovascular NCCN Clinical Practice Guidelines in . J Natl Compr Canc density, and lymphovascular invasion in primary cutaneous melanoma Netw 2016;14:450–473. - correlation with histopathologic prognosticators and BRAF status. 8. Lugassy C, Barnhill RL. Angiotropic melanoma and extravascular Hum Pathol 2015;46:304–312. migratory metastasis: a review. Adv Anat Pathol 2007;14:195–200. 30. Sahni D, Robson A, Orchard G, et al. The use of LYVE-1 antibody for 9. Barnhill RL, Dy K, Lugassy C. Angiotropism in cutaneous melanoma: a detecting lymphatic involvement in patients with malignant prognostic factor strongly predicting risk for metastasis. J Invest melanoma of known sentinel node status. J Clin Pathol 2005;58: Dermatol 2002;119:705–706. 715–521. 10. Van Es SL, Colman M, Thompson IF, et al. Angiotropism is an 31. Feldmeyer L, Tetzlaff M, Fox P, et al. Prognostic implication of independent predictor of local recurrence and in-transit lymphovascular invasion detected by double immunostaining for metastasis in primary cutaneous melanoma. Am J Surg Pathol 2008;32: D2-40 and MITF1 in primary cutaneous melanoma. Am J Dermato- 1396–1403. pathol 2015;38:484–491.

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 127 Lymphatic invasion and angiotropism in melanoma PATHOBIOLOGY IN FOCUS AP Moy et al

32. Moy AP, Mochel M, Duncan LM, et al. Lymphatic invasion predicts 55. Lugassy C, Kleinman HK, Engbring JA, et al. Pericyte-like location of sentinel lymph node metastasis and adverse outcome in primary GFP-tagged melanoma cells: ex vivo and in vivo studies of cutaneous melanoma. J Cutan Pathol 2016;43:e77–e78. extravascular migratory metastasis. Am J Pathol 2004;164:1191–1198. 33. Monshizadeh L, Hanikeri M, Beer TW, et al. A critical review of 56. Lugassy C, Vernon SE, Busam K, et al. Angiotropism of human melanoma pathology reports for patients referred to the Western melanoma: studies involving in transit and other cutaneous Australian Melanoma Advisory Service. Pathology 2012;44:441–447. metastases and the chicken chorioallantoic membrane. Am J 34. Reintgen C, Shivers S, Reintgen M, et al. The changing face of Dermatopathol 2006;28:187–193. malignant melanoma. J Surg Oncol 2010;101:443–446. 57. Zadran S, McMickle R, Shackelford D, et al. Monitoring extra-vascular 35. Karakousis GC, Gimotty PA, Botbyl JD, et al. Predictors of regional migratory metastasis (EVMM) of migrating cancer cells using an in vitro nodal disease in patients with thin melanomas. Ann Surg Oncol co-culture system. Protoc Exch 2013;22:2013. 2006;13:533–541. 58. Lugassy C, Peault B, Wadehra M, et al. Could pericyte mimicry 36. Han D, Zager JS, Shyr Y, et al. Clinicopathologic predictors of sentinel represent another type of melanoma cell plasticity with emboryonic lymph node metastasis in thin melanoma. J Clin Oncol 2013;31: properties? Pigment Cell Melanoma Res 2013;26:746–754. 4387–4393. 59. Bald T, Quast T, Landsberg I, et al. Ultraviolet-radiation-induced 37. Mihic-Probst D, Shea C, Duncan L, et al. Update on thin melanoma: inflammation promotes angiotropism and metastasis in melanoma. outcome of an international workshop. Adv Anat Pathol 2016;23: Nature 2014;507:109–113. 24–29. 60. Lugassy C, Zadran S, Bentolila LA, et al. Angiotropism, pericyctic 38. Murali R, Haydu LE, Quinn MJ, et al. Sentinel lymph node biopsy in mimicry and extravascular migratory metastasis in melanoma: an patients with thin primary cutaneous melanoma. Ann Surg 2012;255: alternative to intravascular cancer dissemination. Cancer Microenviron 128–133. 2014;7:139–152. 39. Abraham RM, Karakousis G, Acs G, et al. Lymphatic invasion predicts 61. Nagy N, Mwizerwa O, Yaniv K, et al. Endothelial cells promote aggressive behavior in melanocytic tumors of uncertain malignant migration and proliferation of enteric neural crest cells via beta1 potential (MELTUMP). Am J Surg Pathol 2013;37:669–675. integrin signaling. Dev Biol 2009;330:263–272. 40. Yun SJ, Gimotty PA, Guerry DP, et al. High lympahtic density and 62. Schwarz Q, Maden CH, Vieira JM, et al. Neuropilin 1 signaling guides lymphatic invasion underlie the adverse prognostic effect of radial neural crest cells to coordinate pathway choice with cell specification. growth phase regression in melanoma. Am J Surg Pathol 2011;35: Proc Natl Acad Sci USA 2009;106:6164–6169. 235–242. 63. Barnhill RL, Chastain MA, Jerdan MS, et al. Angiotropic neonatal 41. Maurichi A, Miceli R, Camerini T, et al. Prediction of survival in patients congenital : how extravascular migration of with thin melanoma: results from a multi-institution study. J Clin Oncol melanocytes may explain the development of congenital nevi. Am J 2014;32:2479–2485. Dermatopathol 2010;32:495–499. 42. Padera TP, Kadambi A, di Tomaso E et al. Lymphatic metastasis in the 64. Kokta V, Hung T, Dhaybi RA, et al. High prevalence of angiotropism in absence of functional intratumor lymphatics. Science 2002;296: congenital melanocytic nevi: an analysis of 53 cases. Am J 1883–1886. Dermatopathol 2013;35:180–183. 43. Wiley HE, Gonzalez EB, Maki W, et al. Expression of CC chemokine 65. Lugassy C, Lazar V, Dessen P, et al. Gene expression profiling of human receptor-7 and regional lymph node metastasis of B16 murine angiotropic primary melanoma: selection of 15 differentially expressed melanoma. J Natl Cancer Inst 2001;93:1638–1643. genes potentially involved in extravascular migratory metastasis. Eur J 44. Recamier JCA. Recherches Sur le Traitement du Cancer. Gabon: Paris, Cancer 2011;47:1267–1275. 1829. 66. Lugassy C, Wadehra M, Li X, et al. Pilot study on “pericytic mimiry” and 45. Lugassy C, Escande JP. The hematogenous theory of metastasis; potential embroyic/stem cell properties of angiotropic melanoma cells Recamier did not propose it. Virchows Arch 1997;431:371. interacting with the abluminal vascular surface. Cancer Microenviron 46. Handley WS. The pathology of melanotic growths in relation to their 2013;6:19–29. operative treatment. Lancet 1907;927:996–998. 67. Bentolila LA, Prakash R, Mihic-Probst D, et al. Imaging of angiotropism/ 47. Lugassy C, Christensen L, Le Charpentier M, et al. Ultrastructural and vascular co-option in a murine model of brain melanoma: implications immunohistochemical observations concerning laminin in B16 for melanoma progression along extravascular pathways. Sci Rep melanoma. Is an amorphous form of laminin promoting a non 2016;6:23834. hematogenous migration of tumor cells? J Submicrosc Cytol Pathol 68. Landsberg J, Tuting T, Barnhill RL, et al. The role of neutrophilic 1998;30:137–144. inflammation, angiotropism, and pericytic mimicry in melanoma 48. Lugassy C, Eyden BP, Christensen L, et al. Angio-tumoral complex in progression and metastasis. J Invest Dermatol 2016;136:372–377. human malignant melanoma characterised by free laminin: ultra- 69. Moreno A, Espanol I, Romagosa V. Angiotropism malignant melanoma: structural and immunohistochemical observations. J Submicrosc Cytol report of two cases. J Cutan Pathol 1992;19:325–329. Pathol 1997;29:19–28. 70. Kerr S, Going JJ. Angiocentric invasion by maligna melanoma. J 49. Lugassy C, Dickersin GR, Christensen L, et al. Ultrastructural and Clin Pathol 1994;47:183–184. immunohistochemical studies of the periendothelial matrix in human 71. Shea C, Kline MA, Lugo J, et al. Angiotropic metastatic malignant melanoma: evidence for an amorphous matrix containing laminin. melanoma. Am J Dermatopathol 1995;17:58–62. J Cutan Pathol 1999;26:78–83. 72. Saluja A, Money N, Zivony DI, et al. Angiotropic malignant melanoma: a 50. Barnhill RL. The biology of melanoma micrometastasis. Recent Results rare pattern of local metastasis. J Am Acad Dermatol 2001;44:829–832. Cancer Res 2001;158:3–13. 73. Barnhill RL, Benson PJ, Lugassy C. Conspicuous angiotropism of 51. Lugassy C, Torres-Munoz JE, Kleinman HK, et al. Overexpression of malignant melanoma involving the brain: implications for malignancy-associated laminins and laminin receptors by angiotropic extravascular migratory metastasis. Am J Dermatopathol 2009;31: melanoma cells in chick chorioallantoic membrane model. J Cutan 205–208. Pathol 2009;36:1237–1243. 74. Barnhill RL, Lugassy C. Angiotropic malignant melanoma and 52. Lugassy C, Kleinman HK, Vernon SE, et al. C16 laminin peptide extravascular migratory metastasis: description of 36 cases with increases angiotropic extravascular migration of human melanoma emphasis on a new mechanism of tumour spread. Pathology 2004;339: cells in a shell-less chick choriorallantoic membrane assay. Br J 485–490. Dermatol 2007;157:780–782. 75. Hung T, Morin J, Munday WR, et al. Angiotropism in primary cutaneous 53. Lugassy C, Haroun RI, Brem H, et al. Pericytic-like melanoma with brain mtastasis: a study of 20 cases. Am J angiotropism of glioma and melanoma cells. Am J Dermatopathol Dermatopathol 2013;35:650–654. 2002;24:473–478. 76. Barnhill RL, Lemaitre S, Levy-Gabrielle C, et al. Satellite in transit 54. Lugassy C, Kleinman HK, Fernandez PM, et al. Human melanoma cell metastases in rapidly fatal conjunctival melanoma: implications for migration along capillary-like structures in vitro: a new dynamic model angiotropism and extravascular migratory metastasis (description for studying extravascular migratory metastasis. J Invest Dermatol of a murine model for conjuctival melanoma). Pathology 2016;48: 2002;119:703–704. 166–176.

128 Laboratory Investigation | Volume 97 February 2017 | www.laboratoryinvestigation.org PATHOBIOLOGY IN FOCUS Lymphatic invasion and angiotropism in melanoma AP Moy et al

77. Gerami P, Shea C, Stone MS. Angiotropism in epidermotropic 81. Busam KJ, Muarli R, Pulitzer M, et al. Atypical spitzoid metastatic melanoma: another clue to the diagnosis. Am J melanocytic tumors with positive sentinel lymph nodes in Dermatopathol 2006;28:429–433. children and teenagers, and comparison with histologically 78. Barnhill RL, Busam K, From L, et al. Inter-observer concordance for the unambiguous and lethal melanomas. Am J Surg Pathol 2009;33: recognition of angiotropism in human melanoma. Pigment Cell 1386–1395. Melanoma Res 2011;24:582–583. 82. Barnhill RL, Kutzner H, Schmidt B, et al. Atypical spitzoid melanocytic 79. Hung T, Piris A, Lobo A, et al. Sentinel lymph node metastasis is not neoplasms with angiotropism: a potential mechanism of locoregional predictive of poor outcome in patients with problematic spitzoid involvement. Am J Dermatopathol 2011;33:236–243. melanocytic tumors. Human Pathol 2013;44:87–94. 83. Wilsher MJ, McCarthy SW, Scolyer RA. Prominent angiotropism in a 80. Mark GJ, Mihm MC, Liteplo MG, et al. Congenital melanocytic nevi of small atypical spitzoid dysplastic melanocytic naevus: what is its the small and garmet type. Hum Pathol 1973;4:395–418. significance? Pathology 2013;45:701–703.

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 February 2017 129