J Cutan Pathol 2010: 37: 723–729doi: 10.1111/j.1600-0560.2010.01531.xJohn Wiley & Sons. Printed in Singapore

Copyright © 2010 John Wiley & Sons A/S

Elastic fiber pattern in regressingmelanoma: a histochemicaland immunohistochemical study

Background: Although histopathologic identification of regression ofmelanoma is usually straightforward, sometimes it can be difficult todistinguish it from scarring fibrosis. Therefore, this study investigatesthe elastic fiber pattern in melanomas associated with either regressionor scars.Methods: We compared 33 invasive melanomas with the fibrosingstage of regression to 10 cases of invasive melanomas with scarringfibrosis. None of the regression cases had a prior surgical procedure.Elastic fiber patterns were evaluated with Verhoeff’s elastic van Giesonstain (EVG) and elastin immunostain.Results: Elastin immunostain was superior to EVG in revealing theelastic fiber patterns. Both regression and scars had decreased to absentelastic fibers in the areas of fibrosis. However, areas of regression had awell-defined compressed layer of thin elastic fibers pushed down fromthe papillary dermis to the base of the fibrosis. In contrast, the base ofscars lacked this compressed elastic layer and had instead an abrupttransition to the thick elastic fibers of the spared reticular dermis.Conclusions: We have identified distinct changes of the elastic tissuenetwork, which more accurately define the presence of regression inmelanoma and distinguish it from scarring fibrosis

Kamino H, Tam S, Roses D, Toussaint S. Elastic fiber pattern inregressing melanoma: a histochemical and immunohistochemical study.J Cutan Pathol 2010; 37: 723–729. © 2010 John Wiley & Sons A/S.

Hideko Kamino1,2, Sam Tam1,Daniel Roses3 and SoniaToussaint4

1Department of Dermatology,2Department of Pathology,3Department of Surgery, New York UniversityLangone Medical Center, New York, NY,USA, and4Departamento de Dermatologia, HospitalManuel Gea Gonzalez, Mexico City, Mexico

Hideko Kamino, MD, Dermatopathology Section,New York University School of Medicine, 530First Avenue, Suite 7J, New York, NY 10016, USATel: +1 212 263 7250Fax: +1 212 684 2991e-mail: HIDEKO.KAMINO@nyumc.org

Accepted for publication January 11, 2010

IntroductionSpontaneous regression of a neoplasm is defined asthe complete or partial disappearance of the tumorwithout any evidence of prior or concurrent externalfactors.1 Regression may be mediated in part by animmunologic host response.2,3

Many benign and malignant tumors of the skinhave been reported to undergo regression.4 How-ever, regression is more common with melanomathan other neoplasms.1,5 Partial regression occurs in10 to 35% of melanomas of any thickness.6 Thefrequency of complete regression in melanomas isreportedly lower, and figures between 2.4 and 8.7%have been forwarded.7,8 It has been speculated that

this phenomenon may account for unknown pri-mary lesions in some cases of metastatic melanoma.Regression is more commonly seen in melanomasof less than 1.5 mm in thickness, with its reportedincidence ranging from 36 to 58%.6,9– 11 Partialregression in thin melanomas occurs more frequentlyin men than in women and often involves lesionslocated on the trunk.6,9,10,12

The prognostic value of regression in melanomaremains controversial. Some studies have reportedno effect on clinical outcome,9,11,13,14 whereas othershave suggested a worse prognosis.10,12,15– 18 It haseven been stated that an unfavorable prognosis isassociated with histopathologic regression of 75% ormore of the melanoma.19

723

Kamino et al.

Histopathological identification of regression isusually straightforward. However, there are someinstances when it can be difficult to distinguishregression from scarring fibrosis caused by a pre-vious procedure, external trauma, or an associatedruptured cyst or hair follicle. The pattern of elasticfibers in the regressed areas of melanoma has notbeen previously reported in the literature.

Recently, we have described distinctive elastic fiberpatterns in different types of melanocytic nevi andmelanomas.20 The aims of the current study are toidentify the changes and patterns of elastic fibers infibrotic areas of late stage regression of melanomaand to compare with the elastic tissue network inmelanomas with scarring fibrosis caused by a previ-ous surgical procedure.

Materials and methodsWe studied 33 primary invasive malignantmelanomas with evidence of late stage regressionobtained from the files of the DermatopathologySection, Department of Dermatology, New YorkUniversity School of Medicine. Late stage regressionwas identified according to the following publishedcriteria: (i) epidermal atrophy with loss of the reteridge pattern, (ii) papillary dermis expanded byfibrosis with thickened collagen bundles orientedparallel to the skin surface, (iii) diminished or absentmelanocytes at the dermoepidermal junction and inthe dermis, (iv) proliferation of blood vessels and(v) sparse to moderate lymphocytic infiltrate withor without melanophages.21 Our cases ranged frompartially (20 to 40% of the lesion) to markedly (40to 80%) regressed lesions. Focally (less than 20%)and totally regressed neoplasms were excluded. Wealso excluded ulcerated lesions. None of the casesof melanoma with regression had a documentedprevious surgical procedure.

In addition, we also analyzed 10 cases of malignantmelanoma with a recent scar secondary to a previousbiopsy, which had been performed from 2 weeks to4 months earlier. These cases were used to comparethe elastic fiber pattern in the area of scarring withthose seen in regression. No regression was seen inthe prior biopsy of melanoma with associated scar.

All melanomas were removed by an excisionalbiopsy. The specimens were fixed in 10% bufferedneutral formalin, step sectioned, processed routinelyin paraffin and stained with hematoxylin-eosin (H&E)and Verhoeff’s elastic van Gieson (EVG) stains.Immunohistochemical staining was also performedfor elastin (Sigma-Aldrich, St Louis, MO, USA;1 : 3000). For those lesions that were inflamed,the extent of the melanocytic proliferation wasdelineated by immunohistochemical staining for

S-100 protein (Dako North America, Carpenteria,CA, USA; 1 : 4000), HMB-45 (Enzo Life Sciences,Farmingdale, NY, USA; 1 : 100) and/or Melan-A(Dako North America; 1 : 200), using the labeledstreptavidin-biotin technique. In heavily pigmentedlesions, the slides were counterstained with azure todecrease interference by melanin and to help visual-ize the antigens stained with diaminobenzidine.22

The distribution of the elastic fibers was evaluatedby light microscopy using H&E, EVG and anti-elastin immunostain. The elastic tissue network fromnonlesional skin adjacent to the melanoma was usedas a baseline internal control. We evaluated theelastic fibers for the following features: (i) baselineelastic fiber network in the skin adjacent to themelanoma, (ii) pattern of elastic fiber network withinthe melanoma, (iii) appearance of elastic fibers in thearea of regression and (iv) appearance of elastic fibersin the area of scarring fibrosis.

ResultsThe melanomas with associated regression (33 cases)and melanomas with scar (10 cases) were all invasivelesions. The Breslow thickness for both groups rangedfrom 0.4 to 3.4 mm. In the regression cohort,23/33 (73%) of the melanomas were less than1 mm in Breslow thickness, 25/33 (76%) had partialregression and 8/33 (24%) had marked regression.

In inflamed areas, immunostains for S-100protein, HMB-45 and/or Melan-A highlighted thearchitectural configuration of the residual melanomain the epidermis and dermis for both groups.Melanocytes were scant to absent in the areas ofregression, but no melanocytes were identified in theareas with scarring fibrosis.

The distribution of the elastic fiber network inthe adjacent uninvolved skin, which was used asthe baseline pattern, was identified using both EVGstain and elastin immunostain. Thin elastic fibers inthe papillary dermis formed a superficial networkperpendicular to the skin surface, resembling a‘candelabra’ or the tines of a ‘fork’ (Fig. 1). Coarser,branched, wavy elastic fibers, located at the base ofthe papillary dermis and in the reticular dermis, wereoriented parallel to the skin surface. Although theelastic fibers found in nonlesional skin and at thebase of the melanomas could be readily identifiedusing the traditional EVG stain, we found the elastinimmunostain to be superior.

Solar elastosis was found in the adjacent skin in17/33 (52%) of melanomas with regression, with15/33 (46%) having moderate solar elastosis and2/33 (6%) containing marked elastotic material.Among the total cases with solar elastosis, six casesalso contained elastotic globules in the papillary

724

Elastic fiber pattern in regressing melanoma

Fig. 1. Normal skin adjacent to a malignant melanoma showing the thin elastic fibers in the papillary dermis oriented perpendicularly to theepidermis in a ‘candelabra’ or ‘fork’ appearance; (A) elastin immunostain, (B) Verhoeff’s elastic van Gieson stain, both ×400.

dermis. EVG stain and elastin immunostain clearlyhighlighted these thick, curly to clumped, elastoticfibers and elastotic globules.

In the areas of residual invasive malignantmelanoma, elastic fibers were markedly decreasedto absent, both in the stroma and in the nests ofmelanocytes (Fig. 2). In all cases, we observed a well-defined, dense, continuous layer of thin elastic fibers

from the papillary dermis that had been pusheddown and compressed (‘pushing border’) to the baseof the invasive melanoma. In this layer of compressedelastic fibers, the ‘candelabra’ or ‘fork’ pattern couldstill be identified, although the elastic fibers wereshorter and thicker than those normally seen in thepapillary dermis. Similarly, in melanoma evolvingon sun-damaged skin, the solar elastosis and elastotic

Fig. 2. Invasive malignant melanoma. There is absence of the thin elastic fibers in the papillary dermis and around the neoplastic melanocytes.Instead, they are found as a compressed layer (arrows, B and D) at the base of the melanoma where they have been pushed down;(A) hematoxylin-eosin (H&E) stain, (B) elastin immunostain, both ×100. Higher power of the layer of compressed thin elastic fibers at the baseof the invasive malignant melanoma; (C) H&E stain, (D) elastin immunostain, both ×200.

725

Kamino et al.

Fig. 3. Malignant melanoma with fibrosing stage of regression.There is absence of melanocytes and marked loss of elastic fibers inthe area of fibrosis, but the thin elastic fibers previously found in thepapillary dermis now form a compressed layer (arrows, B and C) atthe base of the fibrosis. Note faint staining of regenerated thin elasticfibers in fibrosis; (A) hematoxylin-eosin stain, (B) Verhoeff’s elasticvan Gieson stain, (C) elastin immunostain, all ×200.

globules, which are usually present in the papillarydermis, were pushed down into the reticular dermisby the invasive tumor.

There was a marked loss of elastic fibers in thefibrotic areas of regressed melanoma (Figs. 3 and 4).

At the base of the fibrotic area, a distinct, contin-uous, dense layer of compressed thin elastic fiberswas found. This layer was previously located in thepapillary dermis and was displaced down into thereticular dermis by the invasive melanoma and itsstroma prior to regression. The layer of compressedelastic fibers appeared similar and was contiguousto the layer seen at the base of the adjacent resid-ual melanoma. Melanophages were present in theareas of regression above the displaced layer of elas-tic fibers (Fig. 4). In 17 melanomas on sun-damagedskin, including 6 with elastotic globules, the elasticfibers and globules were also pushed down into thereticular dermis (Fig. 4). Regenerated elastic fibers,which appeared short and thin, were seen in thefibrotic areas of regression in 14/33 (42%) of thecases (Fig. 3). In all cases, the extent of regression wasbetter visualized with either the EVG stain or elastinimmunostain when compared with the H&E stain.

In cases of melanoma with a scar as a result of aprevious surgical procedure, the elastic fibers weremarkedly decreased to absent in the area of scarringfibrosis (Fig. 5). In this cohort, 7/10 (70%) had a scarless than 3 months old and these cases had a completeabsence of elastic fibers in the scar. Three melanomashad surgical procedures performed 4 months earlierand the fibrosis contained thin, fragmented and hap-hazardly arrayed elastic fibers, which were consistentwith regenerated fibers. For this cohort, the base of allscars had thick, coarse and wavy elastic fibers, whichwere histopathologically characteristic of the fibers atthe level of the dermis prior to the surgical procedure.In contrast to the ‘pushing border’ pattern seen at thebase of melanoma with regression, scars caused bya prior surgical procedure lacked any preservationof the compressed layer of thin elastic fibers with a‘candelabra’ or ‘fork’ appearance (Fig. 5).

DiscussionIn the current study, we report the use of histochemi-cal and immunohistochemical staining procedures tohighlight differences between the elastic fiber patternat the base of late fibrosing stage of melanoma regres-sion and at the base of scarring fibrosis secondaryto a prior surgical procedure. In all cases of regres-sion, elastic fibers were decreased or absent withinthe fibrosis because the thin elastic fibers from thenormal papillary dermis had been pushed down bythe melanoma and its associated fibrous stroma, thusforming a compressed layer at its base. This layer issimilar to that seen at the base of the residual invasivemelanoma, which we have previously described.20

On the other hand, the thin elastic fibers of the pap-illary dermis were absent in cases of melanoma asso-ciated with a scar. Instead, at the base of the scarring

726

Elastic fiber pattern in regressing melanoma

Fig. 4. Malignant melanoma with fibrosing stage of regression in sun-damaged skin. Note the absence of dermal melanocytes and elastic fibers,epidermal atrophy and fibrosis. The thicker elastotic fibers and globules are seen at the base of the regression. A few melanophages are seen ininfiltrate above pushing border; (A) hematoxylin-eosin stain, (B) elastin immunostain with azure counterstain, both ×400.

fibrosis, we found surgically cut, thick and wavy elas-tic fibers, which would normally have been seen in thereticular dermis. The compressed, dense layer of thinelastic fibers was absent because it had been removedby the previous surgical procedure. In all cases, theelastin immunostain was superior to the EVG stainin visualizing these contrasting elastic fiber patterns.

Regression in malignant melanoma can bedescribed as developing in stages. In the earlystage, a dense band-like infiltrate of lymphocytesobscures the nests of atypical melanocytes locatedat the dermoepidermal junction and in theupper dermis. In the intermediate stage, asthe junctional and dermal melanocytes begin todisappear and the inflammatory response decreases,the residual fibrous stroma of the melanoma, withits increased vascularity because of angiogenesis,becomes more evident. In the late stage of regression,additional changes include further progression ofdense fibrosis, variable number of melanophages(melanosis), diminished to absent inflammatorylymphocytic infiltrate, epidermal atrophy, andmarkedly decreased to absent melanocytes.21

During the process of regression, the inflammatorylymphocytic infiltrate and the fibrosis do not destroypre-existing elastic fibers. Instead, the elastic fibersare found at the base of the area of regression,where they have been pushed down by the invasivemelanoma prior to regression.

When the history of a previous surgical proce-dure or trauma is unknown, scarring fibrosis can beconfused with the late stage of regression. This is espe-cially problematic if only H&E stained histopatho-logic sections are available because scarring fibrosisis the last step of wound healing. In the early stageof wound healing, growth factors cause fibroblasts toactivate, multiply and produce fibrosing granulationtissue resulting in deposits of collagens I and III; thefibroblasts also secrete proteoglycans but initially donot produce elastic fibers. Ultimately, as scars evolve,

they are composed predominantly of abundant colla-gen fibers with fewer fibroblasts, scarcer blood vesselsand thin regenerated elastic fibers.23

Roten et al. studied the chronology of elastic fiberproduction in 116 scars.24 In scars of less than3 months duration, the elastic fibers were totallyabsent. As time passed, both the number and thick-ness of the regenerated elastic fibers progressivelyincreased, reaching their maximum density anddiameter in 12 months. However, newly synthesizedelastic fibers neither developed into mature elas-tic fibers nor reached the density of those seen inadjacent normal skin.24,25 We also noted that newlyregenerated elastic fibers were absent from the scar-ring fibrosis of melanomas that were biopsied lessthan 3 months prior to excision, thus confirming thefindings of Roten et al.24

In this study, we observed marked regression in24% of melanomas and partial regression in 76%of melanomas. We found that regression was morecommon in melanomas of less than 1 mm in Breslowthickness that we noted in 23/33 (73%) of our cases.These data are in accordance with Blessing’s reportthat melanomas measuring less than 1 mm tended toregress more frequently than thicker lesions.6

According to numerous studies, thin regressingmelanomas may have a worse prognosis comparedwith melanomas of similar thickness withoutregression.10,12,15– 18,26 It is possible that the originallesion prior to regression might have been actuallythicker than indicated by the measurement of theresidual melanoma. Therefore, regression may causesome melanomas to be incorrectly staged, withresultant worse prognosis and greater mortality.

The clinical significance of regression as a prog-nostic factor is controversial. Gromet et al.12 statedthat the presence of regression was a poor prognosticsign when associated with thin melanomas that wereotherwise considered to be of low risk by the crite-rion of tumor thickness. McGovern et al.9 reported

727

Kamino et al.

Fig. 5. Malignant melanoma associated with a scar. In contrast toregression, at the base of the scarring fibrosis, the elastic fibersare thick, coarse, wavy, fragmented and parallel to the epidermis(arrows, B and C). There is no layer of thin compressed elastic fibersat the base of the scar; (A) hematoxylin-eosin stain, (B) Verhoeff’selastic van Gieson stain, (C) elastin immunostain, all ×200.

no independent prognostic significance for regres-sion associated with thin melanomas, whereas otherstudies have found better outcomes in patients whosemelanomas showed evidence of regression.11,27

Although the clinical meaning of regression inmalignant melanoma remains unclear, it is importantto properly measure the extent and thickness of the

area of regressed melanoma. This is essential forclarifying the significance of this parameter as aprognostic indicator in future investigations. In thecurrent study, we have shown changes of the elastictissue network that more accurately reveal the extentof regression and distinguish it from scarring fibrosis.The prognostic value of measuring melanoma tumorthickness and regression using these newly definedelastic pattern criteria will be reported in a separatepublication.

AcknowledgementWe gratefully acknowledge William Putnam for histechnical assistance with the photographic material.

References1. Everson TC, Cole WH. Spontaneous regression of malignant

melanoma. Spontaneous regression of cancer: a study andabstract of reports in the world medical literature and ofpersonal communications concerning spontaneous regressionof malignant disease, Chapter 4. Philadelphia: W.B. SaundersCo., 1966; 164.

2. Halliday GM, Patel A, Hunt MJ, Tefany FJ, Barnetson RS.Spontaneous regression of human melanoma/nonmelanomaskin cancer: association with infiltrating CD4+ T cells. WorldJ Surg 1995; 19(3): 352.

3. Tefany FJ, Barnetson RS, Halliday GM, McCarthy SW,McCarthy WH. Immunocytochemical analysis of the cellularinfiltrate in primary regressing and non-regressing malignantmelanoma. J Invest Dermatol 1991; 97(2): 197.

4. Ceballos PI, Barnhill RL. Spontaneous regression of cutaneoustumors. Adv Dermatol 1993; 8: 229.

5. Nathanson L. Spontaneous regression of malignant melanoma:a review of the literature on incidence, clinical features, andpossible mechanisms. Natl Cancer Inst Monogr 1976; 44: 67.

6. Blessing K, McLaren KM. Histological regression in primarycutaneous melanoma: recognition, prevalence and significance.Histopathology 1992; 20(4): 315.

7. Pack GT, Miller TR. Metastatic melanoma with indeterminateprimary site. Report of two instances of long-term survival.JAMA 1961; 176: 55.

8. Smith JL Jr, Stehlin JS Jr. Spontaneous regression of primarymalignant melanomas with regional metastases. Cancer 1965;18(11): 1399.

9. McGovern VJ, Shaw HM, Milton GW. Prognosis in patientswith thin malignant melanoma: influence of regression.Histopathology 1983; 7(5): 673.

10. Sondergaard K, Hou-Jensen K. Partial regression in thinprimary cutaneous malignant melanomas clinical stage I. Astudy of 486 cases. Virchows Arch A Pathol Anat Histopathol1985; 408(2–3): 241.

11. Trau H, Kopf AW, Rigel DS, et al. Regression in malignantmelanoma. J Am Acad Dermatol 1983; 8(3): 363.

12. Gromet MA, Epstein WL, Blois MS. The regressing thinmalignant melanoma: a distinctive lesion with metastaticpotential. Cancer 1978; 42(5): 2282.

13. Cooper PH, Wanebo HJ, Hagar RW. Regression in thinmalignant melanoma. Microscopic diagnosis and prognosticimportance. Arch Dermatol 1985; 121(9): 1127.

728

Elastic fiber pattern in regressing melanoma

14. McCarthy WH, Shaw HM, McCarthy SW, Rivers JK,Thompson JF. Cutaneous melanomas that defy conven-tional prognostic indicators. Semin Oncol 1996; 23(6):709.

15. Clark WH Jr, Elder DE, Guerry DIV, et al. Model predictingsurvival in stage I melanoma based on tumor progression. J NatlCancer Inst 1989; 81(24): 1893.

16. Paladugu RR, Yonemoto RH. Biologic behavior of thinmalignant melanomas with regressive changes. Arch Surg 1983;118(1): 41.

17. Slingluff CL Jr, Vollmer RT, Reintgen DS, Seigler HF. Lethal‘‘thin’’ malignant melanoma. Identifying patients at risk. AnnSurg 1988; 208(2): 150.

18. Guitart J, Lowe L, Piepkorn M, et al. Histological characteris-tics of metastasizing thin melanomas: a case-control study of 43cases. Arch Dermatol 2002; 138(5): 603.

19. Ronan SG, Eng AM, Briele HA, Shioura NN, Das Gupta TK.Thin malignant melanomas with regression and metastases.Arch Dermatol 1987; 123(10): 1326.

20. Kamino H, Tam S, Tapia B, Toussaint S. The use of elastinimmunostain improves the evaluation of melanomas associatedwith nevi. J Cutan Pathol 2009; 36(8): 845.

21. Kang S, Barnhill RL, Mihm MC Jr, Sober AJ. Histologicregression in malignant melanoma: an interobserver concor-dance study. J Cutan Pathol 1993; 20(2): 126.

22. Kamino H, Tam ST. Immunoperoxidase technique modifiedby counterstain with azure B as a diagnostic aid in evaluatingheavily pigmented melanocytic neoplasms. J Cutan Pathol 1991;18(6): 436.

23. Majno G, Joris I. Wound healing. Cells, tissues, and disease:principles of general pathology, Chapter 14. Cambridge, MA:Blackwell Science, 1996; 465.

24. Roten SV, Bhat S, Bhawan J. Elastic fibers in scar tissue.J Cutan Pathol 1996; 23(1): 37.

25. Tsuji T, Sawabe M. Elastic fibers in scar tissue: scanning andtransmission electron microscopic studies. J Cutan Pathol 1987;14(2): 106.

26. Blessing K, McLaren KM, McLean A, Davidson P. Thinmalignant melanomas (less than 1.5 mm) with metastasis: ahistological study and survival analysis. Histopathology 1990;17(5): 389.

27. Kaur C, Thomas RJ, Desai N, et al. The correlation ofregression in primary melanoma with sentinel lymph nodestatus. J Clin Pathol 2008; 61(3): 297.

729