A Practical Technique for Differentiation of Subepidermal Bullous Diseases: Localization of In Vivo–Bound IgG by Laser Scanning Confocal Microscopy

Objective  To develop a practical technique to distinguish autoimmune subepidermal bullous diseases.

Design  A prospective study.

Setting  Academic referral center—the Department of Dermatology, Medical University of Warsaw.

Patients  Forty-two patients fulfilling clinical, immunological, and/or immunoelectron microscopic criteria for bullous pemphigoid (n = 31), mucous membrane pemphigoid (n = 6), or epidermolysis bullosa acquisita (n = 5), diagnosed as having disease and treated from January 1, 1997, to December 31, 2002.

Main Outcome Measures  We applied laser scanning confocal microscopy to determine the localization of in vivo–bound IgG at the basement membrane zone in biopsy specimens taken from patients' skin to compare the localization of basement membrane zone markers: antibody against β₄ integrin, antibody against laminin 5, and antibody against type IV collagen. In vivo–bound IgG was visualized by labeling with fluorescein isothiocyanate–conjugated anti–human IgG antibody, whereas basement membrane zone markers were labeled with anti–mouse Cy5-conjugated antibodies.

Results  In patients with bullous pemphigoid, in vivo–bound IgG was localized on the epidermal side of laminin 5 and co-localized with β₄ integrin. In patients with mucous membrane pemphigoid, IgG was in vivo bound to the dermal-epidermal junction between localization of laminin 5 and type IV collagen. In patients with epidermolysis bullosa acquisita, in vivo–bound IgG was present on the dermal side of type IV collagen.

Conclusions  Laser scanning confocal microscopy allows precise localization of in vivo–bound IgG in patients' skin and, thus, it is a rapid method for the differentiation of mucous membrane pemphigoid from bullous pemphigoid and epidermolysis bullosa acquisita. This tool is suitable for the routine diagnosis of individual patients and for retrospective studies. This method is of special value in those patients in whom circulating autoantibodies are not detectable.

BULLOUS PEMPHIGOID (BP), mucous membrane pemphigoid (MMP), and epidermolysis bullosa acquisita (EBA) belong to the group of autoimmune subepidermal bullous diseases (ASBDs) characterized by the development of tense blisters on apparently healthy skin. These entities clinically can mimic each other, especially at the onset of the disease, but they differ in course, prognosis, and response to treatment.^1-3 The ASBDs are defined by the presence of circulating antibodies directed against different basement membrane zone (BMZ) antigens and the presence of in vivo–bound IgG at the dermal-epidermal junction in patients' skin; thus, differentiation of these dermatoses by routine immunofluorescence (IF) may be difficult or even impossible.^4,5 The characterization of BMZ antigens on a molecular level^6,7 and the introduction of immunoblotting, radioimmunoprecipitation, an enzyme-linked immunosorbent assay, and fusion protein techniques for diagnosing ASBDs allow differentiation of these entities in patients in whom circulating antibodies are detectable. In patients with BP and MMP, BP180 antigen is a target molecule that plays a crucial role in the pathogenesis of these dermatoses.^8-10 BP180 antigen is a transmembrane protein consisting of an amino terminal domain, present in the cytoplasm of basal keratinocytes and in the extracellular portion, that started with the NC16A domain just below the keratinocyte plasma membrane; it extends across the whole lamina lucida into the upper part of the lamina densa, ending by the carboxyterminal domain on the border of the lamina lucida and the lamina densa.¹¹ In patients with BP, the target epitopes for BMZ antibodies are located on the NC16A domain of BP180 antigen, ultrastructurally localized in the upper part of the lamina lucida, whereas the serum samples of most patients with MMP recognize the carboxyterminal domain of BP180 antigen at the lamina lucida–lamina densa border.¹² In some patients with MMP, autoantibodies are directed against the α chain of laminin 5, localized in the upper part of the lamina densa. The production of autoantibodies directed against type VII collagen, localized in the lower lamina densa and the sub–lamina densa, leads to the development of EBA.^13,14

In patients with ASBDs in whom circulating anti–BMZ antibodies are not detectable, a final diagnosis can be established based on direct immunoelectron microscopy, which is time-consuming. In patients with BP, immunoreactants are localized to the lamina lucida¹⁵; in patients with MMP, immunoreactants are localized to the lamina lucida and the lamina densa¹⁶; and in patients with EBA, immunodeposits are bound below the lamina densa.¹⁷ In a previous study, Kazama et al¹⁸ proved that it is possible to distinguish BP from EBA based on the comparison of the localization of target antigens for circulating anti–BMZ antibodies and/or in vivo–bound IgG in patients' skin with the localization of well-defined BMZ markers using laser scanning confocal microscopy (LSCM). In the present study, we applied LSCM to investigate whether it is possible to differentiate MMP from BP and EBA based on the localization of in vivo–bound IgG within the BMZ.

Therefore, this study investigates whether it is possible to differentiate subepidermal bullous diseases based on the localization of in vivo–bound IgG at the BMZ with regard to the localization of lamina lucida and lamina densa markers using LSCM.

Forty-two patients positive for in vivo–bound IgG and complement C3 at the BMZ by direct IF were selected from a group of 102 patients who were diagnosed as having an ASBD and treated at the Department of Dermatology, Medical University of Warsaw, from January 1, 1997, to December 31, 2002. Patients positive for in vivo–bound IgA at the BMZ by direct IF as a predominant or concomitant component were not included in this study.

Thirty-one patients (14 men and 17 women) fulfilled the clinical and immunopathological criteria for BP. Patients were aged between 73 and 90 years. They developed tense blisters on the trunk and extremities, which healed without scars or milia. Patients with BP who had bullous lesions localized on traumatized areas or had mucous membrane involvement were excluded from this study. Of 31 patients' serum samples, 29 were positive for circulating IgG anti–BMZ antibodies and showed reactivity with the epidermal side of sodium chloride salt–split skin by indirect IF. Immunoblot studies revealed the reactivity of 12 patients' serum samples with 180- and /or 230-kDa molecules on epidermal extract. Direct immunoelectron microscopy using a peroxidase technique was performed on the skin originating from 2 patients whose samples were negative for circulating anti–BMZ antibodies, and showed the presence of IgG deposits within the lamina lucida.

Six patients (2 men and 4 women) fulfilled the clinical and immunopathological criteria for MMP.⁸ All patients with MMP had mucous membrane and skin involvement. They had tense bullae on the skin, which healed and left atrophic scars and milia. Mucous membrane involvement in all patients referred to scarring conjunctivitis and chronic painful erosions of oral mucosa. Of 6 patients' serum samples, 2 were positive for circulating IgG anti–BMZ antibodies and reacted with the epidermal and dermal side of sodium chloride salt–split skin by indirect IF. The results of immunoblot studies were negative (in our studies and in the studies performed at another laboratory). Direct immunoelectron microscopy using the peroxidase technique, performed in 4 patients (who were negative for circulating anti–BMZ antibodies), showed the presence of IgG deposits within the lamina lucida and the lamina densa.

Three men (aged 20, 38, and 42 years at the onset of disease) and 2 children (a 3-year-old boy and a 13-year-old girl) were included in this study. Three patients (2 adults and 1 child) fulfilled the criteria of Roenigk et al³ for mechanobullous EBA. They had tense bullae on the skin, localized on traumatized areas that were healing, and scars and milia. Two other patients had an inflammatory type of EBA, clinically resembling BP.² Of 5 patients with EBA, 3 had oral mucous involvement. Of 5 patients' serum samples, 3 were positive for circulating IgG anti–BMZ antibodies and revealed reactivity with the exclusively dermal side of sodium chloride salt–split skin by indirect IF. Immunoblot studies of these serum samples showed reactivity with a 290-kDa molecule on the dermal extract. Serum studies using postembedding immunogold electron microscopy on Lowicryl-embedded healthy human skin demonstrated the reactivity of IgG anti–BMZ antibodies with the lamina densa and anchoring fibrils. Direct immunoelectron microscopy using the peroxidase technique, performed on the skin specimens of 2 patients who were negative for circulating anti–BMZ antibodies, revealed the presence of IgG deposits below the lamina densa. Data are summarized in Table 1.

Punch biopsy specimens taken from perilesional patients' skin were mounted in tissue freezing medium (Leica Instruments, GmbH, Nussloch, Germany) and cut into 10-µm cryosections. These sections were incubated with monoclonal anti–BMZ antibodies directed against β₄ integrin, a marker of the upper part of the lamina lucida (clone 3E1; Chemicon International, Temecula, Calif); laminin 5, also called epiligrin, a marker of the upper part of the lamina densa (clone P3E4; Chemicon International); and type IV collagen, a marker of the lamina densa (clone COL-94; Sigma, Steinheim, Germany), for 30 minutes, followed by 5-minute washings with phosphate-buffered saline, performed 3 times. Then, the cryosections were incubated with a mixture of rabbit anti–mouse Cy5-conjugated antibody (Chemicon International) and fluorescein isothiocyanate (FITC)–conjugated goat anti–human antibodies (Kappel, Aurora, Ohio) for 30 minutes, followed by 5-minute washings with phosphate-buffered saline, performed 3 times; mounted in p-phenylenediamine; and viewed using LSCM (Radiance 2000; BIO-RAD, Oxford, England). Excitation of FITC and Cy5 was simultaneously performed with laser lines of 488 and 637 nm, respectively. To obtain a specific signal for FITC (green color) and Cy5 (red color) and to avoid a cross-talking channel effect, we used a dichroic mirror of 560 nm and a bandpass filter of 500 to 560 nm for epifluorescence of FITC and a longpass filter of 660 nm for Cy5. Green and red images were overlaid by an image processing system integrated with the LSCM, and photographed.

In all patients with BP, the overlay image of β₄ integrin and in vivo–bound IgG showed yellow fluorescence along the BMZ because of an overlap of the red fluorescence of β₄ integrin and the green fluorescence of in vivo–bound IgG (Figure 1, A). The overlay image of laminin 5 and in vivo–bound IgG showed a green reaction of IgG on the epidermal side and red staining of laminin 5 on the dermal side, which indicates that IgG is in vivo bound above laminin 5 (Figure 1, B). The overlay image of in vivo–bound IgG and type IV collagen showed green-red (from epidermis to dermis) staining along the BMZ, suggesting the localization of IgG above the localization of type IV collagen (Figure 1, C).

In all patients with MMP, the overlay images of green fluorescence dependent on in vivo–bound IgG and the red fluorescence of β₄ integrin showed red staining on the epidermal side and green staining on the dermal side of the BMZ, suggesting that IgG is in vivo bound below the localization of β₄ integrin (Figure 1, D). The overlay images of laminin 5 and in vivo–bound IgG showed red staining on the epidermal side and green fluorescence on the dermal side of the BMZ, suggesting that IgG is in vivo bound below the localization of laminin 5 (the epitope is recognized by the monoclonal antibody used in this study) (Figure 1, E). In some areas of the BMZ, a yellow reaction due to the partial co-localization of in vivo–bound IgG and laminin 5 was observed, but there was never green fluorescence of IgG extended above the localization of laminin 5. The overlay images of in vivo–bound IgG and type IV collagen showed green-red (from epidermis to dermis) staining along the BMZ, suggesting the localization of IgG above the localization of type IV collagen (Figure 1, F).

In all patients with EBA, the overlay image of IgG and β₄ integrin, laminin 5, and type IV collagen showed red-green (from epidermis to dermis) staining along the BMZ, indicating that IgG is in vivo bound below the localization of β₄ integrin (Figure 1, G) and below laminin 5 (Figure 1, H) and type IV collagen (Figure 1, I).

Diagnosing MMP and EBA is difficult because in most of the patients, circulating anti–BMZ antibodies are not detectable. An alternative method to time-consuming direct immunoelectron microscopy would be desirable. We developed a useful option using LSCM.

In 1988, one of us (C.K.)¹⁹ was the first, to our knowledge, to introduce the direct IF salt-split skin technique for the differentiation of BP (a reaction on the roof or on the roof and floor of an artificial blister) from EBA (a reaction on the floor of an artificial blister).^20,21 In most of the patients with MMP, in vivo–bound immunoglobulins are present on the roof and floor of the artificial blister mimic, BP, whereas in one fifth of the patients with MMP, immunodeposits react with the floor of the blister, suggesting the diagnosis of EBA. This discrepancy is because of the presence of 2 different targeted antigens recognized in the serum samples of patients with MMP: the C domain of BP180 antigen, present in the lower part of the lamina lucida; and laminin 5 (also known as epiligrin), localized in the upper part of the lamina densa.^11,22,23 Thus, a direct salt-split pattern cannot be conclusive in patients with MMP.

Later, we applied fluorescence overlay antigen mapping, using a regular epifluorescence microscope, to investigate whether it is possible to distinguish the localization of in vivo–bound IgG within the BMZ in patients with BP and MMP from that in patients with EBA. The localization of in vivo–bound IgG in patients' skin was compared with the localization of the carboxyterminal end of type VII collagen (a marker localized 360 nm below the lamina densa).^13,14 We found the co-localization of in vivo–bound IgG in patients with EBA with the carboxyterminal end of type VII collagen and lack of this co-localization in patients with BP and MMP. The resolution of the regular epifluorescence microscope did not allow differentiation of in vivo–bound IgG in the lamina lucida (in patients with BP) from IgG bound in the lamina lucida and the lamina densa (in patients with MMP). In addition, an application of fluorescence overlay antigen mapping for the routine diagnosis of subepidermal bullous diseases might cause many technical problems, known as geometric error and color error.²⁴

In this study, we applied LSCM to compare the localization of in vivo–bound IgG with the localization of different BMZ markers. In contrast to a previous study by Kazama et al,¹⁸ we used a diode laser at 637 nm and a longpass filter at 660 nm for Cy5 dye, instead of a helium-neon laser at 543 nm and a longpass filter at 590 nm for tetramethyl rhodamine isothiocyanate, simultaneously with an argon laser at 488 nm and a bandpass filter at 500 to 560 nm for FITC. The system setting used in this study allows elimination of the cross-talking channel effect. Also, the detection of FITC is increased because of the use of an emission bandpass filter at 500 to 560 nm instead of the 500 to 530 nm used in a previous study.¹⁸

In addition to the antibodies used by Kazama et al,¹⁸ we tested new markers at the lamina lucida–lamina densa border. Finally, we recommend the use of a monoclonal antibody against the α chain of laminin 5 (clone P3E4, called epiligrin by Chemicon International) in the differentiation of in vivo–bound IgG in patients with BP and MMP by LSCM. The precise immunoelectron microscopic localization of the epitope recognized by this antibody is not known; however, by using the cryosectioning immunogold electron microscopic method, different epitopes of laminin 5 were found in the upper part of the lamina densa.¹¹ The antibody against the α chain of laminin 5 (clone P3E4) produced a stronger IF reaction on the dermoepidermal junction compared with the laminin 5 (clone D4B5) used in a previous study¹⁸; thus, it is possible to obtain a high-quality LSCM image with this particular antibody and to increase the resolution.

Our study has shown that EBA could be clearly distinguished from BP and MMP by LSCM based on the presence of in vivo–bound IgG below the localization of type IV collagen. The differential diagnosis between BP and MMP is more complicated because of the close distance between in vivo–bound IgG in both entities.

We found that in vivo–bound immunoglobulins in patients with BP are located above laminin 5. They partially co-localized with, but did not extend below, laminin 5 (at least the epitope recognized by monoclonal antibody used in this study). In patients with MMP, in vivo–bound IgG is localized between laminin 5 and type IV collagen. It partially co-localized with laminin 5 and type IV collagen, but did not extend above laminin 5 and below type IV collagen.

The results of our study indicate that LSCM could be useful for the differentiation of MMP from BP and EBA. This method is of special value in the diagnosis of patients in whom circulating autoantibodies are not detectable.

Corresponding author and reprints: Cezary Kowalewski, MD, Department of Dermatology, Medical University of Warsaw, Koszykowa 82A, 02-008 Warsaw, Poland (e-mail: ckowalew@amwaw.edu.pl).

Accepted for publication March 28, 2003.

This study was supported by grant KBN 3 PO5B 058 22 from the Polish Scientific Research Committee.

This study was presented at the World Congress of Dermatology; July 2, 2002; Paris, France.

We thank Takashi Hashimoto, MD, Department of Dermatology, Kurume University School of Medicine, Kurume City, Japan, for performing some of the immunoblot studies referenced in this article.