Germany studied by Poetsch et al (2000), compared with 61% (49of 80 chromosomes) and 5% (four of 80 chromosomes) in theAustralian samples in our study. The similarity in polymorphismfrequencies indicates that there is no signi®cant difference ingenetic background between the two populations that may accountfor the major difference in reported mutation frequency. It is notknown whether differences in environment could account for thediscrepancy; however, it is dif®cult to envisage how the majorenvironmental risk factor for melanoma development (exposure toultraviolet radiation; Whiteman and Green, 1999) could contributeto the difference. Unless one invokes an environmental carcinogenother than ultraviolet radiation, it seems more likely that therewould be less somatic mutations detected in north-easternGermany than in Australia.

Overall, in stark contrast to the study of Poetsch et al (2000),our results do not indicate a signi®cant role for TTC4mutations in melanoma development. Although TTC4 initiallyappeared a good candidate, the search for tumor suppressorgenes on 1p continues.

This work was supported by the National Health and Medical Research Council.

Nicole Irwin, Graeme Walker, Nicholas HaywardQueensland Institute of Medical Research,

P.O. Royal Brisbane Hospital, QLD 4029, Australia

REFERENCES

Castellano M, Pollock P, Walters M, et al: CDKN2A/p16 is inactivated in mostmelanoma cell lines. Cancer Res 57:4868±4875, 1997

Davis P, Dowdy S: p73. Int J Biochem Cell Biol 33:935±939, 2001Dracopoli NC, Harnett P, Bale SJ, Stanger BZ, Tucker MA, Housman DE, Kefford

RF: Loss of alleles from the distal short arm of chromosome 1 occurs late inmelanoma tumor progression. Proc Natl Acad Sci USA 86:4614±4618, 1989

Poetsch M, Dittberner T, Cowell JK, Woenckhaus C: TTC4, a novel candidatetumor suppressor gene at 1p31 is often mutated in malignant melanoma of theskin. Oncogene 19:5817±5820, 2000

Smedley D, Sidhar S, Birdsall S, Bennett D, Herlyn M, Cooper C, Shipley J:Characterization of chromosome 1 abnormalities in malignant melanomas.Genes Chromosom Cancer 28:121±125, 2000

Su G, Roberts T, Cowell JK: TTC4, a novel human gene containing thetetratricopeptide repeat and mapping to the region of chromosome 1p31 that isfrequently deleted in sporadic breast cancer. Genomics 55:157±163, 1999

Su G, Casey G, Cowell JK: Genomic structure of the human tetratricopeptide repeat-containing gene, TTC4, from chromosome region 1p31, and mutation analysisin breast cancers. Int J Mol Med 5:197±200, 2000

Thompson FH, Emerson J, Olson S, et al: Cytogenetics of 158 patients with regionalor disseminated melanoma. Subset analysis of near-diploid and simplekaryotypes. Cancer Genet Cytogenet 83:93±104, 1995

Tsao H, Zhang X, Majewski P, Haluska F: Mutational and expression analysis of thep73 gene in melanoma cell lines. Cancer Res 59:172±174, 1999

Walker GJ, Palmer JM, Walters MK, Hayward NK: A genetic model of melanomatumorigenesis based on allelic losses. Genes Chromosom Cancer 12:134±141,1995

Whiteman DC, Green AC: Melanoma and sun exposure: where are we now? Int JDermatol 38:481±489, 1999

Zhang J, Glatfelter AA, Taetle R, Trent JM: Frequent alterations of evolutionarilyconserved regions of chromosome 1 in human malignant melanoma. CancerGenet Cytogenet 111:119±123, 1999

Expression of the 90K Tumor-Associated Protein in Benignand Malignant Melanocytic Lesions

To the Editor:

Histopathologic differentiation of melanoma from benign melano-cytic lesions can be extremely dif®cult (Barnhill et al, 1999; Reed,1999), as a subset of melanomas mimic conventional (Suster et al,1987; Wong et al, 1995) or Spitz's nevi (Muhlbauer et al, 1983;Smith et al, 1989; Wong et al, 1995), but behave aggressively (Reedet al, 1975; Reed, 1978; Muhlbauer et al, 1983; Phillips et al, 1986).Therefore, the identi®cation of molecular markers that allow anaccurate diagnosis of melanoma is of utmost importance. Studyingsuch markers could also improve the understanding of the biologicevents involved in the melanocyte transformation to a highlyinvasive and metastatic tumor. 90K is a large oligomeric proteincomposed of a » 90 kDa subunit, originally identi®ed as a tumor-secreted antigen in the culture medium of human breast cancer cells(Iacobelli et al, 1986). After cDNA cloning (Ullrich et al, 1994), thesequence of 90K showed that it contained a cystein-rich domain,which is homologous to that found in the SRCR group A family ofproteins and is strictly related to the immunoglobulin superfamily(Resnick et al, 1994). 90K was identi®ed independently as a ligandof the lactose-speci®c S-type lectin, galectin-3 (formerly known asMac-2), and also named Mac-2 binding protein (Koths et al, 1993).90K is observed in many tissues and in biologic ¯uids (Koths et al,

1993; D'Ostilio et al, 1996) and is overexpressed by the majority ofhuman tumors (Iacobelli et al, 1986), with serum levels frequentlyelevated, as in patients with melanoma (Natoli et al, 1994). In A375

Table I. 90K staining in benign and malignant melanocyticlesions

CasesTotal cases(no.)

Positivecases (no.)

Distributionpattern Score

Lentigo simplex 5 0Junctional nevus 5 2 Diffuse +/±Intradermal nevus 5 1 Focal +/±Compound nevus 10 5 Focal (2)

Patchy (3)+/±

Spitz's nevus 10 10 Diffuse +/++MetastasizingSpitz's nevus

2 2 Focal (1)Diffuse (1)

+/++

Melanoma in nevusa 9 9 Diffuse ++/+++In situ melanoma 6 6 Diffuse +++Melanoma < 0.76 mm 16 16 Diffuse (12)

Patchy (4)++/+++

Melanoma > 0.76 to< 3.00 mmMelanoma > 3.00 mm

148

148

Diffuse (10)Patchy (4)Diffuse (4)Patchy (4)

++/+++++/+++

Cutaneous metastasis 6 6 Diffuse +/++Nodal metastasis 3 3 Diffuse +/++Bone metastasis 1 1 Focal ++

aResidual nevi were negative.

Manuscript received July 3, 2001; revised December 17, 2001; acceptedfor publication December 20, 2001.

Reprint requests to: Dr. Gian P. Trentini, Dipartimento di ScienzeMorfologiche e Medico Legali, Sezione di Anatomia Patologica UniversitaÁdi Modena, Policlinico, via del Pozzo 71, 41100 Modena, Italy.Email: trengi31@unimo.it

VOL. 119, NO. 1 JULY 2002 LETTERS TO THE EDITOR 187

human melanoma cells, 90K was found to promote homotypiccell±cell adhesion by cross-linking of surface-bound galectin-3residues (Inohara et al, 1996). Results from binding studies in vitrodemonstrated b1 integrin-mediated cell adhesion for 90K, as well asinteractions with some collagens and ®bronectin (Sasaki et al, 1998).This suggests that 90K may play a part in embolization ofdisseminating tumor cells in circulation and favors the progressionand metastatization. In this study, we veri®ed whether 90K couldbe a reliable immunohistochemical marker of melanoma. We alsoevaluated the ability of melanoma cells to adhere to 90K and otherextracellular matrix proteins.

Pathologic specimens, achieved from the section of PathologicAnatomy, comprehended the series of benign and malignantmelanocytic lesions listed in Table I. An immunohistochemicalreaction was performed using the anti-90K monoclonal antibody1A4.22 (Tinari et al, 1997), and results were evaluated as positive ornegative cytoplasmic staining, scored in three grades (+, ++, +++)and in three patterns of distribution (diffuse, focal, patchy); weakstaining was recorded as +/±. Cell adhesion assay was performed oncells of the human melanoma cell line MEL-8863, from metastatichuman melanoma (Mecchia et al, 2000), using recombinant 90Kprotein (Ullrich et al, 1994), laminin, and ®bronectin (Calbiochem,San Diego, CA). The immunohistochemical results (see Table Iand Fig 1) demonstrated that all 65 malignant lesions, including

metastasizing Spitz's nevi and metastases, were positively stained(++/+++); however, only 18 of 44 benign lesions (also includingthe nine residual nevi in melanomas) gave a positive reaction,which was weak (+/±) in eight cases, and mild or moderate (+/++)in the remaining 10, represented by Spitz's nevi. Altogether theseresults demonstrated that the speci®city and sensitivity of 90K indifferentiating melanocytic lesions is 59.09% and 100%, respectively(Table II). Cell adhesion experiments showed adhesion of MEL-8863 to 90K, laminin and ®bronectin higher than adhesion to thenegative control bovine serum albumin (p < 0.0001) (Fig 2A),®bronectin being the best substrate. Competition experiments werealso performed and the results are shown in Fig 2(B).

Studies on markers of malignancy in melanoma have focusedmainly on oncogene alterations (Newton Bishop, 1997), but little isknown about other molecules, such as glycoconjugate ligands, thatinteract with extracellular matrix and may be related to malignanttransformation and tumor progression. In this study, 90K appearedto represent a potential marker of melanocytic cell transformationand a factor possibly involved in tumor invasion and metastasis.Although the expression of 90K is not absolute in distinguishingbetween benign and malignant lesions, this marker was consistentlyexpressed in all types of melanoma, but not in normal melanocytesat the dermoepidermal junction nor in conventional nevi. Only aminority of these nevi showed a weak and usually focal staining.

Figure 1. Immunohistochemical staining pro-®le for 90K. Sections from paraformaldehyde-®xed and paraf®n-embedded tissues were immersedin ethylenediamine tetraacetic acid buffer at pH 6.5and pretreated in a microwave oven at 360 Wfor 30 min; endogenous peroxidase activity wasquenched with 3% hydrogen peroxide for 10 minat room temperature; sections were incubated withanti-90K monoclonal antibody 1A4.22 (Tinari et al,1997) for 30 min at room temperature, then weretreated with the LSAB Plus System (streptavidin±biotin, LabVision, Fremont, CA); diamino-benzidine was used as chromogene; sections werecounterstained with hematoxylin for 2 min. Apositive control (breast carcinoma) and a negativecontrol for each case studied were included.Negative normal melanocytes at the dermo-epidermal junction (A) and negative cells incompound nevus (B); positive melanoma cellscompared with negative nevus cell in a case ofmelanoma arisen in nevus (C); strong positivity in``in situ'' melanoma (D) and in invasive melanoma(E); the same invasive melanoma as negativecontrol after incubation without the primaryantibody (F); moderate positivity in Spitz's nevus(G) and in metastasizing Spitz's nevus (H); positivecutaneous (I) and lymph node (L) metastases. Scalebar: 100 mm.

188 LETTERS TO THE EDITOR THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Also, in melanomas arising in nevus, the benign part of the lesionwas negative, whereas the melanoma cells were strongly positive.Metastases were immunohistochemically positive for 90K (con-®rmed by immunoblotting, not shown), thus providing evidencethat 90K is expressed also by secondary lesions. On the other hand,90K does not appear to be of value in differentiating Spitz's nevusfrom melanoma. Indeed, Spitz's nevi showed a diffuse, althoughless intense, positivity when compared with melanoma, and asimilar positivity was observed in the two cases of metastasizingSpitz's nevus. Intriguingly, these data also suggest a role for 90K as

an early marker of neoplastic transformation of melanocytic lesions,as well as raise some questions about the real nature of Spitz's nevus.This is further substantiated by the ®nding that 90K expression wasstronger and more diffuse in ``in situ'' melanoma than in in®ltratinglesions and metastases.

Our ®ndings show that the expression of 90K can help todistinguish between benign and malignant melanocytic lesions,with the exception of Spitz's nevus, which presents roughly thesame degree of expression of malignant melanoma. There was nodifference in the expression between melanomas of differentthickness, with the exception of the stronger positivity characteriz-ing the ``in situ'' lesions. Therefore, 90K does not seem to representa prognostic factor that may predict the clinical outcome forpatients.

Our in vitro experiments show that this protein, absorbed on aplastic substrate, causes the adhesion of melanoma cells through ab1-integrin-mediated process. It has recently been documentedthat 90K is a cell-adhesive, secreted glycoprotein that accumulatesin the extracellular matrix (Sasaki et al, 1998). As b1-integrins areimportant components of the adhesion receptor repertoire onmelanoma cells (Schadendorf et al, 1993; Hieken et al, 1995), theinteraction between 90K present in the extracellular matrix and b1-integrins on the cell surface may play a part in the interactionbetween tumor cells and surrounding microenvironment; however,our immunohistochemical data do not determine whether thisfavors a more invasive behavior.

In conclusion, this study demonstrated that 90K is highlyexpressed in melanoma as compared with normal melanocytes andbenign nevi. Our data suggest that the immunohistochemicalevaluation of 90K could potentially provide useful diagnosticinformation in selected patients with melanoma.

This work was supported by a grant from COFIN MIUR and agrant from the UniversitaÁ di Modena e Reggio Emilia.

Anna Maria Cesinaro, Clara Natoli,* Antonino Grassadonia,*Nicola Tinari,* Stefano Iacobelli,* Gian Paolo Trentini

Department of Morphological Sciences and Legal Medicine,Section of Pathologic Anatomy, University of Modena and

Reggio Emilia, Italy*Department of Oncology and Neurosciences,

University ``G. D'Annunzio'' of Chieti, Italy

REFERENCES

Barnhill RL, Argenyi ZB, From L, et al: Atypical Spitz nevi/tumors: lack ofconsensus for diagnosis, discrimination from melanoma, and prediction ofoutcome. Hum Pathol 30:513±520, 1999

D'Ostilio N, Sabatino G, Natoli C, Ullrich A, Iacobelli S: 90K (Mac±2BP) in humanmilk. Clin Exp Immunol 104:543±546, 1996

Figure 2. Cell adhesion to 90K protein. (A) Cells from the humanmelanoma cell line MEL 8863 were grown in RPMI 1640 mediumsupplemented with 10% fetal bovine serum, 1% (vol/vol) penicillin(100 U per ml), 1% (vol/vol) streptomycin (100 U per ml), and 1%(vol/vol) L-glutamine (all from Gibco-BRL, Grand Island, NY),maintained at 37°C in 5% CO2 and passaged every week. Cells (50,000cells per well) were seeded in 96-well microtiter plates and coated with100 ml of 90K, laminin, and ®bronectin (10 mg per ml in phosphate-buffered saline) overnight at 4°C. One hour before seeding, plates weresaturated with 1% bovine serum albumin in RPMI at 37°C. Cells weretrypsinized, washed, and resuspended at 500,000 per ml in RPMI serumfree with 0.1% bovine serum albumin. One hundred microliters (50,000cells) were seeded in the wells and, after 1 h of incubation at 37°C,nonadherent cells were removed by gentle washing in phosphate-buffered saline and adherent cells ®xed with 95° ethanol for 10 min andstained with 100 ml of 4% crystal violet for 30 min. The absorbance at550 nm was measured on a plate reader after solubilization with 50 ml of0.25% Triton-X100. Data represent the mean 6 SD of three differentexperiments done in quadruplicate. (B) Adhesion assay was performedseeding MEL 8863 cells in 96-well microtiter plates precoated with100 ml of 90K (10 mg per ml) in presence of the function blocking anti-b1 antibody 4B4 (Coulter, Hialeah, FL), anti-90K SP-2 and 1A4.22antibodies (Tinari et al, 1997) at a concentration of 10 mg per ml, anti-galectin 3M3/38 antibody, or 50 mM lactose. Anti-galectin-3 and lactosedid not affect the binding of MEL 8863 cells to 90K, as well as 1A4.22antibody, whereas adhesion was competed by the anti-90K antibody SP-2; moreover, b1 function-blocking antibody 4B4 markedly reduced thebinding of cells to 90K. Data represent the mean 6 SD of three differentexperiments done in quadruplicate.

Table II. 90K immunostaining in detecting malignancy inmelanocytic lesions

Positive Negative Total

Benign lesionsa 18 26 44Malignant lesionsb 65 0 65Total 83 26 109

(95% con®denceinterval Binomialexact test)

Sensitivity (65/65) 100.00% 94.48% 100.00%Speci®city (26/44) 59.09% 43.25% 73.66%Positive predictivevalue (65/83)

78.31% 67.91% 86.61%

Negative predictivevalue (26/26)

100.00% 86.77% 100.00%

aIncluding nine residual nevi in melanoma.bIncluding two metastasizing Spitz's nevi.

VOL. 119, NO. 1 JULY 2002 LETTERS TO THE EDITOR 189

Hieken TJ, Ronan SG, Farolan M, Shilkaitis AL, Kim DK, Das Gupta TK: Beta 1integrin expression in malignant melanoma predicts occult lymph nodemetastases. Surgery 118:669±673, 1995

Iacobelli S, Arno E, D'Orazio A, Coletti G: Detection of antigens recognized by anovel monoclonal antibody in tissue and serum from patients with breastcancer. Cancer Res 46:3005±3010, 1986

Inohara H, Akahani S, Koths K, Raz A: Interactions between galectin-3 and Mac-2-binding protein mediate cell-cell adhesion. Cancer Res 56:4530±4534, 1996

Koths K, Taylor E, Halenbeck R, Casipit C, Wang A: Cloning and characterizationof a human Mac-2-binding protein, a new member of the superfamily de®nedby the macrophage scavenger receptor cysteine-rich domain. J Biol Chem268:14245±14249, 1993

Mecchia M, Matarrese P, Malorni W, et al: Type I consensus interferon (CIFN) genetransfer into human melanoma cells up-regulates p53 and enhances cisplatin-induced apoptosis: implications for new therapeutic strategies with IFN-alpha.Gene Ther 7:167±179, 2000

Muhlbauer JE, Margolis RJ, Mihm MC Jr, Reed RJ: Minimal deviation melanoma: ahistological variant of cutaneous malignant melanoma in its vertical growthphase. J Invest Dermatol 80:635±655, 1983

Natoli C, Ortona L, Tamburrini E, et al: Elevated serum levels of a 90,000 daltonstumor-associated antigen in cancer and in infection by humanimmunode®ciency virus (HIV). Anticancer Res 14:1457±1460, 1994

Newton Bishop JA: Molecular pathology of melanoma. Cancer Metastasis Rev 16:141±154, 1997

Phillips ME, Margolis RJ, Meerot Y, Sober AJ, Reed RJ, Muhlbauer JE, Mihm MC:The spectrum of minimal deviation melanoma. a clinicopathologic study of 21cases. Hum Pathol 17:796±806, 1986

Reed RJ: Minimum deviation melanoma. Am J Surg Pathol 2:215±220, 1978

Reed RJ: Dimensionalities: borderline and intermediate melanocytic neoplasia. HumPathol 30:521±524, 1999

Reed RJ, Ichinose H, Clarke WH, Mihm MC: Common and uncommonmelanocytic naevi and borderline melanoma. Semin Oncol 2:119±147, 1975

Resnick D, Pearson A, Krieger M: The SRCR superfamily: a family reminiscent ofthe Ig superfamily. Trends Biochem Sci 19:5±8, 1994

Sasaki T, Brakebusch C, Engel J, Timpl R: Mac-2 binding protein is a cell adhesiveprotein of the extracellular matrix which self-assembles into ring-like structuresand binds beta1 integrins, collagens and ®bronectin. EMBO J 17:1606±1613,1998

Schadendorf D, Gawlik C, Haney U, Ostmeier H, Suter L, Czarnetzki BM: Tumourprogression and metastatic behaviour in vivo correlates with integrin expressionon melanocytic tumours. J Pathol 170:429±434, 1993

Smith KJ, Barrett TL, Skelton HG, Lupton GP, Graham JH: Spindle cell andepithelioid cell nevi with atypia and metastasis (malignant Spitz nevus). Am JSurg Pathol 13:931±939, 1989

Suster S, Romero M, Bubis JJ: Verrucous pseudonaevoid melanoma. J Surg Oncol36:134±137, 1987

Tinari N, D'Egidio M, Iacobelli S, et al: Identi®cation of the tumor antigen 90Kdomains recognized by monoclonal antibodies SP2 and L3 and preparation andcharacterization of novel anti-90K monoclonal antibodies. Biochem Biophys ResCommun 23:367±372, 1997

Ullrich A, Sures I, D'Egidio M, et al: The secreted tumor-associated antigen 90K is apotent immune stimulator. J Biol Chem 269:18401±18407, 1994

Wong T-Y, Suster S, Duncan LM, Mihm MC: Nevoid melanoma mimickingspindle and epithelioid cell nevus and verrucous dermal nevus. Hum Pathol26:171±179, 1995

Dendritic Epidermal T Cells (DETC) are Diminished inIntegrin aE(CD103)-De®cient Mice

To the Editor:

A unique cell population of Thy-1+ dendritic epidermal cellsresides within murine epidermis (Bergstresser et al, 1983;Tschachler et al, 1983). When it was recognized that thispopulation belongs to the lineage of gdTCR+ T lymphocytes,they were named dendritic epidermal T cells (DETC) (Koning etal, 1987; Kuziel et al, 1987; Stingl et al, 1987a; 1987b). DETCexhibit a very limited diversity of TCR g and d gene usage withvery little junctional diversity, and most, if not all, bear the TCRVg3-Jg1-Cg1/Vd1-Dd2-Jd2-Cd (Asarnow et al, 1988, 1989;Havran and Allison, 1988; Havran et al, 1989). Thymic origin ofDETC was strongly suggested by the observation that TCR-Vg3+

thymocytes [nomenclature according to Garman et al equivalent tothe Vg5 designation by Tonegawa and Heilig (Hayday and Pao,1998)] appear at days 14±15 of gestational life, and disappear fromthe thymus before their ®rst appearance within the epidermis atdays 17±19 of gestation (Havran and Allison, 1988). In addition,implantation of day-14 thymic lobes from normal mice intoathymic nude mice (which lack Vg3+ DETC) resulted in theappearance of an epidermal CD3+/Vg3+ population of DETC(Havran and Allison, 1990). Whereas some experimental evidencesuggested that DETC participate in tolerance induction, elicitationof contact hypersensitivity, tumor rejection, and other immuno-logic functions (Tigelaar and Lewis, 1995; Girardi et al, 2001),many aspects of their biologic role remain unknown. Morerecently, it was found that integrin aE(CD103)b7 is expressedaround day 16 of gestation by TCR-Vg3+ thymocytes, and it washypothesized that this adhesion receptor is involved in epidermallocalization of DETC (Lefrancois et al, 1994). This suggestion was

consistent with the known functions of aE(CD103)b7 on intestinalintraepithelial T lymphocytes, where it mediates adhesion to theepithelial cell surface molecule E-cadherin (Cepek et al, 1994;Karecla et al, 1995), a counter-receptor that is also constitutivelyexpressed by epidermal keratinocytes (Takeichi, 1990). Recently,we demonstrated that aE(CD103)b7 mediates, at least in part,epidermal localization of some CD8+ T lymphocytes in humanin¯ammatory skin disorders. This was based primarily uponantibody-mediated blocking of adhesive functions of cultured Tlymphocytes expressing aE(CD103)b7 (Pauls et al, 2001). There isno experimental evidence thus far, however, supporting thehypothesis that integrin aE(CD103)b7 is involved in epidermallocalization of murine DETC.

In order to directly investigate the effect of integrin aE(CD103)on DETC in vivo, we have analyzed wildtype mice and micelacking expression of the aE(CD103) integrin subunit, whosegeneration and characterization was described recently (SchoÈn et al,1999, 2000). Given that considerable strain-dependent differencesin the numbers of DETC have been reported, and as BALB/c mice(the strain in which the aE-de®cient mice were ®rst generated)have signi®cantly fewer DETC than other inbred strains(Bergstresser et al, 1983), the aE(CD103)-de®cient animals werebackcrossed for ®ve to seven generations into the C57BL/6 strainand then analyzed for DETC number.

In our initial experiments, the presence of DETC within theepidermis was assessed by immunohistochemical analysis (ABCimmunoperoxidase method according to the manufacturer'sinstructions; Vector, Burlingame, CA) using monoclonal antibodiesspeci®c for CD3e (clone 500A), integrin aE(CD103) (clone 2E7),TCR Vg3 (clone 536), and an isotype-matched control mono-clonal antibody (clone Ha4/8, hamster IgGk monoclonal anti-bodies purchased from PharMingen, San Diego, CA). As expected,aE(CD103) was not expressed in the skin of aE(CD103)-de®cientmice, whereas it was readily detected on numerous cells ofdendritic morphology within the epidermis of wildtype (aE

+/+)mice (Fig 1). Interestingly, when CD3+/Vg3+ DETC were

Reprint requests to: Dr. Michael P. SchoÈn, Department of Dermatologyand Venereology, Otto-von-Guericke-University, Leipziger Str. 44, 39120Magdeburg, Germany. Email: Michael.Schoen@Medizin.Uni-Magdeburg.de

190 LETTERS TO THE EDITOR THE JOURNAL OF INVESTIGATIVE DERMATOLOGY