Lymphatic differentiation and microvascular proliferation in benign vascular lesions of skin and soft tissue: Diagnostic features following the International Society for The Study of Vascular Anomalies Classification—A retrospective study

Abstract

Background

Discrepancies have been noted between the clinical and histologic diagnosis of vascular malformations.

Objective

To evaluate the effectiveness of the International Society for Study of Vascular Anomalies (ISSVA) classification in diagnosing benign vascular anomalies based on clinical and (immuno) histologic parameters, focusing on lymphatic differentiation and vascular proliferation.

Method

A retrospective study of 121 consecutive patients with benign skin and soft-tissue vascular anomalies located in the head and neck region (pyogenic granulomas and angioma senilis were excluded) by applying multiplex immunohistochemistry staining for lymph vessels (D2-40), endothelial blood vessels, and proliferating cells (Ki67). Clinical and histologic diagnosis was revised after the ISSVA classification.

Results

Initially, 64 lesions were diagnosed as tumors and 57 as malformations. Revision diagnosis following the ISSVA classification revealed 27 tumors, 90 malformations (22.2% lymphatic), and 4 non-ISSVA. Immunostaining showed lymphatic differentiation in 24 (19.8%) of 121 cases, of which 20 were malformations. Proliferative activity (Ki67+) was found in 41 (33.8%) of 121 cases, of which 8 were arteriovenous malformations.

Limitation

Quality and size of materials (biopsies vs resections) and clinical information.

Conclusion

The diagnostic accuracy of combined histologic and clinical approaches for identifying vascular anomalies following the ISSVA classification can be substantially enhanced by incorporating additional immunostaining techniques to evaluate lymphatic differentiation and proliferative activity, particularly in identifying the occurrence of vascular malformations.

Capsule Summary.

• •A multidisciplinary approach to patients with vascular anomalies of skin and soft tissue is needed for appropriate classification of lesions.
• •Additional immunostaining for lymphatic differentiation and proliferative activity significantly increases the diagnostic yield of a combined clinical and histologic approach, and can be important for planning of targeted therapies.

Introduction

According to the latest International Society for Study of Vascular Anomalies (ISSVA) classification in 2018,¹ vascular anomalies (VA) are grouped into either vascular tumors or (congenital) vascular malformations. This classification is based on a concerted approach of histologic and clinical diagnostic parameters, eventually supplemented with imaging modalities. Erroneous diagnosis of these lesions due to inappropriate diagnostic approach, either clinically, radiologically, or histologically may lead to insufficient or adverse treatment. Therefore, precise diagnostic workup is important to identify the specific types of VA for treatment strategies and predict future prognosis.² In 2017, we reported discrepancies between clinical and histologic diagnosis in a series of vascular malformations.³ An important pitfall was the inability to diagnose vascular malformations or tumors based on conventional histology only or no histology at all.³

Determining the specific vascular nature of each type of lesion, and whether they have a lymphatic origin or substantial lymphatic component, is necessary for targeted therapy and to predict the risk of specific complications.⁴ Infection is a common complication of lymphatic malformations, because malformed lymphatics are less able to clear foreign materials, whereas significant thrombosis is commonly found in venous malformations.5, 6, 7, 8

An accurate diagnosis of VA remains problematic also due to inconsistent nomenclature.9, 10, 11, 12 Vascular malformations can be mistaken for infantile hemangiomas (IH) or other vascular tumors, which may lead to incorrect or delayed therapy. Application of glucose transporter-1, which is a specific and sensitive marker for endothelial cells (ECs) in IH, is recommended to distinguish IH from other lesions.^9,10,12

Vascular tumors and malformations are categorized by their differences in growth behavior. Vascular tumors undergo EC proliferation and malformations are slowly progressive lesions with a low EC turnover.^13,14 Therefore, the identification of vascular proliferative growth in histologic samples of the lesions can be considered as an important discriminator between tumors and malformations. Overall, vascular malformations are characterized by a quiescent type of growth and no tendency to regress, but during the (often lifelong) cause of the disease, disproportionate growth may occur under various circumstances, leading to increased severity of symptoms.¹⁵ Among other factors, such as trauma, thrombosis, and hormonal changes, this disproportionate growth in vascular malformations can also be attributed to areas of microvascular proliferations (MVPs) that may arise in between the large mature vessels of the malformation. A study reported such MVP areas in 30% of patients with symptomatic (surgically treated) arteriovenous malformations (AVM), mostly located in the head and neck area.¹⁶ Recognizing MVP in these types of lesions (using clinical information, histology, and radiology examination) can be useful for planning management strategies of symptomatic vascular malformations.^17,18

Based on previous publications on lymphatic and arterial differentiation of the vessels involved, identification of glucose transporter-1 expression on EC, and the occurrence of MVP, have all emerged as important histopathological features to diagnose VA effectively according to the ISSVA classification, at least when these findings are embedded in a sound clinical diagnosis. In this study, we investigated the occurrence of lymphatic vessel differentiation and MVP area in VA lesions with the use of multiplex immunostaining. Additionally, we examined the effectiveness of the ISSVA classification system in a large consecutive series of symptomatic VA lesions of which the clinical diagnosis was known or could be retrieved retrospectively.

Methods

Selection of study materials

A series of 458 consecutive cases of benign soft-tissue VA, retrieved from the head and neck region (biopsy and excision samples), dating from 2008 to 2017, were collected from the department of Pathology of the Amsterdam University Medical Center (AUMC), location AMC. Because, overall, the lesions retrieved from the head and neck appeared to be clinically well documented for the purpose of this study, we limited our work to this specific topographic location. Three hundred cases of pyogenic granuloma and 8 cases of angioma senilis were excluded from the study, because combined clinical features and histopathologic characteristics of these lesions easily render in accurate diagnosis. From the 150 remaining cases, 2 patients objected to participate in the study, and 27 cases were excluded due to insufficient quality and size of the available tissues stored in paraffin blocks for further pathologic analysis. For inclusion, these tissues had to allow revision on Hematoxylin-eosin and Elastic van Gieson stained sections and further sectioning of the tissue blocks for immunohistochemistry (IHC) staining. Finally, 121 cases were enrolled in the study. Elastic van Gieson stains were used to discriminate between lesional arteries and veins.

Ethical Clearance

Informed consent forms were sent by mail to patients or their families. Review Board of the Amsterdam UMC granted a waiver for the use of leftover materials after surgery (reference number: W20_135 # 20.174). Project was further approved by the Pathology Biobank of AUMC (project number: 2022.0555).

Selection of markers for IHC

A selection of potential antibodies of interest to visualize EC differentiation of the vascular lesions preceded the analysis of the entire study group in order to select the best markers, using IHC single staining. The following antibodies were tested: mouse monoclonal anti-D2-40 (lymph vessels, DAKO) and rabbit monoclonal anti-LYVE-1 (lymph vessels, Abcam). For blood vessels, as follows: rabbit monoclonal anti-ERG (EC, Abcam), mouse monoclonal anti-CD-31 (EC, DAKO), mouse monoclonal anti-CD-34 (EC, DAKO), rabbit monoclonal anti-Fli-1 (EC, Santa Cruz Biotechnology), and rabbit monoclonal anti–von Willebrand factor (vWf) (EC, DAKO). From these series, D2-40 was selected to be the most suitable lymphatic antibody and vWf for EC blood vessel differentiation (arteries and veins). To illustrate this decision, Fig 1 shows combined D2-40 and vWf on a sequential IHC staining as described,¹⁹ using chromogen PermaBlue (Diagnostic Biosystem) for lymph vessels (D2-40) and Vector Novared (Vector Labs) for blood vessels (vWf). Finally, rabbit monoclonal anti-Ki67 (proliferating cells, ThermoFisher Scientific) and mouse monoclonal anti-PCNA (proliferating cells, DAKO) were tested on vascular tumors, of which Ki67 was chosen because of distinct bright nuclear staining (Supplementary Fig 1, available via Mendeley at https://data.mendeley.com/datasets/t5vg2ppbxf/1).

Fig 1.

Lymphatic venous malformations stained with sequential double immunohistochemistry. Positive staining of lymphatic vessels endothelium using D2-40 is shown in blue and blood vessel endothelium using vWf in red. There is no overlapping staining of the 2 markers.

Immunohistochemical multiplex staining

Multiplex IHC staining for D2-40, vWf, and Ki67, respectively, was applied on all cases. First step, sections were dewaxed in xylene and rehydrated in graded-alcohols prior to antigen retrieval with heat-induced epitope retrieval in a pre-treatment Module (ThermoFisher Scientific) using a TRIS-EDTA buffer (ThermoFisher Scientific). Then, endogenous horseradish peroxidase was blocked in methanol + H₂O₂ (30% diluted in methanol into 0.3% solution). Incubation with primary antibody D2-40 for 60 minutes at room temperature, followed by horseradish peroxidase antimouse polymer secondary antibody (Brightvision, Immunologic) for 30 minutes. Vector Novared was used as a chromogen. Sections were counterstained with Hematoxylin, dried on a hotplate and coverslip sealed with Glycerol/Gelatin (Sigma-Aldrich). Next, sections were digitized using a digital pathology slide scanner (Philips IntelliSite Ultra Fast Scanner, Philips Digital Pathology Solutions). After this, a destaining process was performed before continuing to the second staining round by removing the coverslip in warm water and subjecting the sections to a stripping buffer (TRIS-sodium dodecyl sulfate + beta mercaptoethanol) in which the colors and immune complexes were eluted from the tissue. Second, sections were stained and digitized again as described above using vWf, and then the process was repeated with Ki67. After 3 rounds of staining, 2 rounds of destaining, and scanning, the images were ready for analysis.²⁰

Semiquantitative analysis

To evaluate lymph vessel density and morphology (using D2-40 antibody) and proliferative activity in MVP areas (using Ki-67 antibody), a semiquantitative scoring system was applied in both instances using scores of 0 to 3 (Table I). For D2-40 staining, scores 2 and 3 were considered as diagnostic for lymphatic involvement. Lesions with scores 0 and 1 were considered as lesions without lymphatic involvement or inconclusive for diagnosis.

Table I.

Semiquantitative scoring system of assessment for lymphatic differentiation using D2-40 and for proliferating cells using Ki67 antibody in benign vascular anomalies

Scoring for D2-40	Description
(lymphatic vessels)
0	Negative or small scattered vessels in dermis
1	<10% lymph vessels, small conglomerates
2	10%-50% lymph vessels
3	>50% lymph vessels

Scoring for Ki67	Description
(proliferating cells)
0	Negative
1	1 focus (≥1 positive cells in 1 cluster)
2	≥2 positive clusters
3	Diffuse-positive cells

Ki 67 positivity was counted within both blood vessels and lymph vessels. Semiquantitative analysis was performed by 2 observers independently (A.M.U. and M.A.K.). In the case of discrepancy, 2 additional investigators (A.C.W. and L.B.M.J.) reviewed the sections to retrieve definite scores.

Final diagnosis for each case was made together by pathologists and clinicians (M.M.L. and C.M.A.M.H.) following the ISSVA classification, using the histologic, immunohistochemical, and clinical diagnostic features.

Data analysis

Descriptive statistics were analyzed for a set of categorical data (eg, tissue retrieval, sex, topography, D2-40 scores, Ki67 scores, and final diagnosis), and were presented as numbers and percentages. Continuous data following the nonparametric distribution were presented with mean and SD. Statistical analysis was performed using SPSS 26.00 (IBM Corporation).

Results

From a total of 121 patients with VA, 27 biopsy and 94 excision samples were taken. Samples were retrieved from the head and neck region, of which 107 (88.4%) were located in the face. The average age of patients was 44.8 ± 30.4 years, 65 (53.7%) were women and 56 (46.3%) were men (Table II). Initial clinical and histologic data, which were retrieved from clinical and pathology reports, identified 64 vascular tumors and 57 malformations. Further studies on these materials revealed data that are shown in Table II.

Table II.

Characteristic and final International Society for Study of Vascular Anomalies diagnosis of benign vascular anomalies cases included in the study (n = 121)

Age of patients at the time of biopsy/surgical excision, y			Mean (SD) 44.8 (30.4)
Characteristic			Final ISSVA diagnosis
	N	%		N	%
Tissue retrieval			Vascular tumors
Biopsy	27	22.3	Angiokeratoma	1	0.8
Surgical excision	94	77.7	Arteriovenous tumor	7	5.8
Sex			Congenital hemangioma	2	1.7
Females	65	53.7	Epithelioid hemangioma	4	3.2
Males	56	46.3	Hobnail angioma	2	1.7
Topography			Infantile hemangioma	10	8.3
Face	107	88.4	Non-involuting congenital hemangioma	1	0.8
Head and neck	14	11.6	Vascular malformations
D2-40 score			Arteriovenous malformation	14	11.6
0	60	49.6	Capillary arteriovenous malformation	2	1.7
1	37	30.6	Capillary malformation	1	0.8
2 (Lymph+)	18	14.9	Capillary venous malformation	6	5
3 (Lymph+)	6	4.9	Glomovenous malformation	2	1.7
Ki67+			Lymphatic arteriovenous malformation	5	4.1
Blood vessels	41	33.8	Lymphatic malformation	5	4.1
Lymph vessels	8	6.6	Lymphatic venous malformation	10	8.3
			Traumatized malformation	1	0.8
			Venous malformation	32	26.5
			Nonspecific ISSVA diagnosis
			Benign vascular lesion	5	4.1
			Vascular malformation	7	5.8
			Others
			Fibrovascular Polyp	1	0.8
			IgG4 Mediated inflammatory pseudotumor	1	0.8
			Pseudotumor of Masson	1	0.8
			Venous lake	1	0.8

ISSVA, International Society for Study of Vascular Anomalies.

Lymphatic differentiation

Semiquantitative analysis of D2-40 immunostaining showed score 0 in 60 (49.6%), score 1 in 37 (30.6%), score 2 in 18 (14.9%), and score 3 in 6 (4.9%) cases (Supplementary Table I, available via Mendeley at https://data.mendeley.com/datasets/t5vg2ppbxf/1). This resulted in a total of 24 (19.8%) cases diagnostic for lymphatic involvement (scores 2 and 3) (Fig 2). Of note, 2 biopsy and 7 excision samples, which had a score 2 or 3, respectively, were not recognized as lymphatic lesions in the initial clinical and/or histologic diagnosis (Supplementary Fig 2, available via Mendeley at https://data.mendeley.com/datasets/t5vg2ppbxf/1). In these unidentified lymphatic cases, lymph vessel markers had not been applied in the initial examination, and were diagnosed as venous malformations, AVM, or angiomas. After revision, they were diagnosed as lymphatic venous malformation or lymphatic arteriovenous malformations.

Fig 2.

Graph showing semiquantitative scoring of lymphatic differentiation assessed by D2-40 immunostaining in 121 cases of benign vascular anomalies.

Surprisingly, 2 cases initially diagnosed as lymphangioma (based on Hematoxylin-eosin) had a score 0 and 1, indicating no lymph vessel involvement with D2-40 staining. These were revised as benign vascular lesions not otherwise specified, because of a lack of other distinctive diagnostic features.

Microvascular proliferative areas

Ki67 immunostaining was applied to assess the proliferation index of lesional (micro) vessels, which may indicate the MVP areas (Supplementary Fig 3, available via Mendeley at https://data.mendeley.com/datasets/t5vg2ppbxf/1). The occurrence of MVP areas was assessed and semi quantitatively scored in all lesions and was found in lymph and blood vessels. Ki67 positivity in blood vessels (EC and smooth muscle cells) was found in 41 (33.8%) cases. Of these, 6 biopsy and 6 excision cases (29.3%) were scored 1, 1 biopsy and 14 excisions cases (36.6%) were scored 2, and 1 biopsy and 13 excisions cases (34.1%) were scored 3. Presence of vWf+, D240- MVP areas were observed in malformations (AVM type, n = 8) and tumors (mostly IH, n = 7). Ki67 positivity in lymph vessels (D2-40+) was seen in 8 (6.6%) cases, diagnosed as tumors (n = 2, epithelioid hemangioma and hobnail angioma), and in malformations (n = 6, lymphatic venous and lymphatic malformations). Three (1 biopsy and 2 excisions) were scored 2 and 5 (1 biopsy and 4 excisions) were scored 3.

Final diagnosis after revision

After evaluation of lymphatic differentiation, lesional MVP activity, and glucose transporter-1 staining in appropriate cases in combination with the available clinical diagnostic features, a final diagnosis following the ISSVA classification could be made in 117 (96.7%) cases, of which 27 (23.1%) were vascular tumors and 90 (76.9%) were malformations. Without additional IHC staining, only 60 (49.6%) cases could be diagnosed according to the ISSVA classification. The full list of initial and revision diagnosis is shown in Supplementary Table II (available via Mendeley at https://data.mendeley.com/datasets/t5vg2ppbxf/1).

Improper use of nomenclature was found to be a recurring problem in our series. Specifically, 14 cases of malformations were classified as (cavernous) hemangiomas in the initial diagnosis. Lesions that could not be classified according to ISSVA were a fibrovascular polyp, an IgG4-mediated inflammatory pseudotumor, a (pseudo)tumor of Masson, and a venous lake.

Discussion

Our series of VAs were retrieved from a group of lesions that were initially diagnosed as 64 vascular tumors and 57 malformations. Eventually, a different outcome was found with the use of additional IHC and review of clinical signs and symptoms, diagnoses were revised, revealing 27 cases of tumors and 90 cases of vascular malformations, whereas 4 cases could not be diagnosed following ISSVA classification.

Not all VA can be diagnosed easily according to the ISSVA classification. In daily practice, a proportion of lesions are diagnosed and treated on the basis of clinical appearance, eventually combined with imaging modalities. However, in cases of aberrant clinical appearances, disproportionate growth, and suspicion of malignancy, all the criteria on clinical information, radiology, and histology examination must be fulfilled to make a correct diagnosis, and in such cases, a biopsy for histopathologic investigation is highly recommended. One study showed 65.71% agreement between clinical diagnosis and histopathology in vascular malformations.²¹ They concluded that the histopathologic examination helped in subcategorizing various vascular malformations, and thus, combining diagnostic features will reduce diagnostic difficulties and help in proper patient management.²¹

We found that lymphatic differentiation of vessels occurred in 24 cases, whereas 10 (41.7%) cases were not determined previously due to insufficient application of lymphatic marker immunostaining. After testing several EC-specific antibodies, we experienced that combining D2-40 and vWf antibodies resulted in the most suitable method to distinguish lymph vessels from blood vessels. Our present study showed that with the use of a scoring system for lymphatic markers, involvement of lymph vessels can be easily determined. This was supported morphologically by the presence of dilated and malformed lymph vessels in lesions with scores 2 and 3. Lesions with lymphatic components can be difficult to diagnose clinically, and it is essential to predict complication risks, such as infection,⁵ to plan proper treatment. Current management therapies include surgery, sclerotherapy, and laser therapy, or combined, or non-surgical conservative treatments, such as radiotherapy, electrocoagulation, cryotherapy, ligation, and embolization.²² Combination of sclerotherapy and doxycycline drugs may prevent the risk of infection.⁵

Recent insights in the molecular pathways underlying genetic mutations that drive the formation and progression of VA is, at present, increasingly being used to classify VA, ultimately leading to better patient care. Gene-targeted therapies are in development and may provide suitable solutions for patients in which traditional treatments were unsuccessful or not possible. Most mutations in VA, which have been identified thus far, occur within the RAS/MAPK and PI3K/AKT/mTOR pathways.²³ Somatic mutations in the PIK3CA gene causing abnormal activation of PI3K-AKT signaling, may result in increased cellular proliferation, and has been associated with PIK3CA-Related Overgrowth Spectrum.^24,25 However, diagnostic examination of genetic mutations still faces technical difficulty, since some tissue biopsies of patients with PIK3CA-related overgrowth spectrum have low levels of mosaicism (<5%).²⁶ Therefore, conventional histologic examination and IHC are still needed, especially for lesions with symptomatic progressive growth.

Episodic progressive growth in vascular malformations, which can occur during puberty, adolescence, or pregnancy, can lead to severe symptomatic complications.²⁷ In arteriovenous types of malformations, this phenomenon, among other factors, such as trauma, hormonal changes, and inflammation, has also been correlated with substantial areas of intralesional MVP showing high Ki67 labeling indices.^16,28 In case of proliferative activity, a suspected malignancy should always be considered and evaluated histologically.^29,30 However, in the present study, MVP also appeared as a benign proliferative process of several types of benign tumors, but also in symptomatic vascular malformations. These findings indicate that the proliferation of benign microvessels may also occur in lesions that are according to ISSVA usually not considered to be proliferative, such as AVM. This activity is considered to be a reactive feature, and likely results from tissue hypoxia within the AVM.^15,31, 32, 33

Ki67+ proliferative activity in our series was also observed in lymphatic lesions and 1 case of epithelioid hemangioma (angiolymphoid hyperplasia with eosinophilia). It remains to be settled if angiolymphoid hyperplasia with eosinophilia truly embodies a vascular neoplasm, a reactive vascular hyperplasia, or a lymphoproliferative process with subsequent angiogenesis.^34,35

Finally, our study also confirmed a major problem of misuse of the nomenclature as has been recognized previously.^3,21,36 Several cases that were not recognized as malformations were initially diagnosed as angioma, which can lead to improper treatment. Sampling bias may occur in case of small biopsies, and thus, it is not recommended to apply as a single diagnostic tool.

Limitations

This study was limited by its quality and size of a part of the available materials for multiplex IHC staining, incomplete clinical information in some cases, being a single-center study only, the restricted topographic distribution of the lesions, and the retrospective nature of the samples. For some biopsy cases, the materials appeared to be insufficient for a proper diagnosis and had to be excluded from the study.

Conclusion

A multidisciplinary approach to patients with benign VA, as forwarded in the ISSVA classification, is required for accurate diagnosis of the types of lesions involved. This yielded a considerably higher percentage of vascular malformations as opposed to tumors in our consecutive lesions of VA. In this setting, (immuno)histologic features of lymphatic differentiation and occurrence of MVP should be considered systematically. MVP areas are not a feature of vascular tumors only but can also be observed regularly in symptomatic vascular malformations, especially AVM. However, in all these cases, a diagnosis of vascular malignancy should be excluded. This study can be useful for clinicians, radiologists, and pathologists working in the field of VA, since only an accurate diagnosis of specific types of VA allows an adequate treatment of the disease.

Conflicts of interest

None disclosed.