ELISA Testing of Anti–Desmoglein 1 and 3 Antibodies in the Management of Pemphigus

Objective  To assess the predictive value of anti–desmoglein (Dsg) 1 and anti-Dsg3 antibody (Ab) enzyme-linked immunosorbent assay (ELISA) values for the occurrence of relapses in pemphigus.

Design  Retrospective study.

Setting  Dermatology departments from 13 university hospitals in France.

Patients  The study population comprised 26 patients with typical clinical, histologic, and immunofluorescence findings of pemphigus, who were followed up over a 17-month period.

Main Outcome Measures  Serial anti-Dsg1 and anti-Dsg3 Ab ELISA values were recorded during the patients' follow-up examinations and correlated with the occurrence of skin and/or mucosal relapses.

Results  A significant reduction of anti-Dsg1 (P < .001) and anti-Dsg3 (P < .001) Ab ELISA values was observed in serum samples from patients with pemphigus foliaceus or pemphigus vulgaris after the initial treatment. During the long-term follow-up, anti-Dsg1 Ab ELISA values correlated with the course of skin lesions (P = .03); the 20 U/mL cutoff for the anti-Dsg1 Ab ELISA value provided a 79% positive and an 84% negative predictive value for the occurrence of cutaneous relapses. No correlation was observed between anti-Dsg3 Ab ELISA values and the course of mucosal lesions (P = .13). Anti-Dsg3 Ab ELISA values higher than the 14-U/mL cutoff were observed in 5 of the 5 patients with relapse and in 10 of the 13 patients with ongoing mucosal remission, providing a 100% sensitivity but a poor specificity of 23%. A cutoff value of 130 U/mL for anti-Dsg3 Abs was calculated based on the receiver operating characteristics curve and provided an 84% positive and an 81% negative predictive value.

Conclusions  Anti-Dsg1 Ab ELISA values are more closely correlated than anti-Dsg3 Ab ELISA values with the course of the disease in patients with pemphigus vulgaris or pemphigus foliaceus. This should be taken into account for the management of patients with pemphigus.

Pemphigus includes a group of severe autoimmune diseases characterized by blistering of the skin and mucous membranes. The 2 main classic types of pemphigus have been identified as pemphigus vulgaris (PV) and pemphigus foliaceus (PF), in which pathogenic IgG autoantibodies are directed against desmosomal transmembrane glycoproteins: desmoglein (Dsg) 3 and Dsg1, respectively.¹ The autoimmune response in the mucosal type of PV is directed against Dsg3, whereas both Dsg1 and Dsg3 are targeted by autoantibodies^2,3 in the mucocutaneous type of the disease. Anti-Dsg antibodies (Abs) from patients with pemphigus mainly belong to the IgG4 and IgG1^4-6 subclasses and have been demonstrated to be pathogenic.^7-11

Among the different methods used to detect circulating IgG autoantibodies in serum samples from patients with pemphigus, enzyme-linked immunosorbent assay (ELISA) using recombinant human Dsg1 and Dsg3 have been shown to be more sensitive and specific than indirect immunofluorescence.^12-15 In fact, anti-Dsg–ELISA tests provide a quantitative method for measuring anti-Dsg Ab levels, and are currently used for the diagnosis of pemphigus. The severity of skin and oral mucous membrane lesions in pemphigus is related to anti-Dsg1 and anti-Dsg3 Ab levels, respectively.¹⁶ Moreover, some case reports have suggested a correlation between anti-Dsg–ELISA values and the evolution of pemphigus in individual patients followed up over a prolonged period.^12,14,17 However, the predictive value of anti-Dsg1 and anti-Dsg3 Abs determined by ELISA for the occurrence of cutaneous and/or mucosal relapses has not yet been extensively investigated.

To ascertain the usefulness of serial anti-Dsg1 and anti-Dsg3 Ab ELISA assays in the management of pemphigus, we recorded all the anti-Dsg1 and anti-Dsg3 Ab ELISA values during the follow-up of 26 patients with PV or PF and correlated the evolution of ELISA values with the occurrence of skin and/or mucosal relapse.

This retrospective study was performed in 13 dermatologic centers in France. All patients fulfilled the following inclusion criteria: (1) diagnosis of pemphigus based on the presence of mucosal erosions and/or superficial cutaneous blisters, suggestive of PV and PF, respectively; a histologic picture of intraepidermal acanthosis; and deposition of IgG, complement component 3, or both on the keratinocyte membrane detected by direct immunofluorescence¹⁸; (2) serial anti-Dsg1 and anti-Dsg3 Ab ELISA combined with clinical evaluations at various times over the follow-up period; and (3) a minimal follow-up of 17 months after initial treatment. Twenty of these patients were included in a previous clinical trial.¹⁹

Twenty-one patients (81%) were initially treated with prednisone and immunosuppressants, including azathioprine, methotrexate, thalidomide, mycophenolate mofetil, and rituximab; 4 patients were treated with immunosuppressive therapy alone and 1 patient with prednisone alone. Complete remission with minimal therapy was defined according to the consensus statement²⁰ as the absence of new and/or established lesions while the patient was receiving minimal therapy, ie, 10 mg/d of prednisone or less (or the equivalent) and/or minimal adjuvant therapy for at least 2 months. Relapse was defined as the reappearance of 3 or more new lesions a month that do not heal spontaneously within 1 week or by the extension of established lesions in a patient who has achieved disease control.²⁰ Prednisone doses for the 20 patients included in our previous clinical trial were tapered according to a standardized protocol.¹⁹

We first compared the evolution of anti-Dsg1 and anti-Dsg3 Abs after initial treatment between patients who achieved complete remission with therapy and those who did not. Second, we compared mean anti-Dsg1 and anti-Dsg3 Ab ELISA values between patients with ongoing complete remission with therapy and those who relapsed. Then, to ascertain the evolution of anti-Dsg1 and anti-Dsg3 Abs before the onset of a skin and/or mucosal relapse, we individually assessed in all patients with relapse the evolution of anti-Dsg Abs during the 6-month period before the relapse. In addition, we assessed the levels of anti-Dsg1 and anti-Dsg3 Abs in serum samples from patients with PF or PV that were collected at the last visit before the relapse became clinically evident in patients with no cutaneous or mucosal lesions. Finally, we calculated the sensitivity, specificity, and positive and negative predictive values of anti-Dsg1 and anti-Dsg3 Ab ELISA values for the occurrence of a relapse.

Serum samples were collected at each follow-up visit. A total of 284 samples were analyzed, which represented a mean of 11 samples per patients. To determine anti-Dsg1 and anti-Dsg3 Ab titers, Dsg1- and Dsg3-ELISA tests (MESACUP Desmoglein test; Medical and Biological Laboratories, Nagano, Japan) were performed with 1:100 diluted serum samples according to the manufacturer's instructions as previously described.^12,21 Data are expressed as units per milliliter of serum. For the evaluation of sensitivity and specificity of ELISAs, the cutoff values proposed by the manufacturer, ie, 20 U/mL for anti-Dsg1 Abs and 14 U/mL for anti-Dsg3 Abs, were first considered. Serum samples were collected, and all ELISAs were performed in 1 center.

To assess the usefulness of serial ELISAs during the patients' follow-up examinations, the evolution of anti-Dsg Abs during the 6-month period before the relapse was evaluated. A rise in anti-Dsg Ab levels was defined as an increase of at least 50% of ELISA values between samples collected at the time of the relapse and those collected during the 6-month period before the relapse.

The Wilcoxon signed rank test was used for the comparison of paired data. Two-sided P values less than .05 were considered statistically significant. Continuous variables are expressed as mean (SD). Sensitivity and specificity of Dsg ELISAs for the occurrence of a relapse were calculated as previously described.²² A receiver operating characteristic (ROC) curve was calculated to determine a cutoff value for anti-Dsg1 and anti-Dsg3 Abs with optimal sensitivity and specificity. Each point on the curve represents the relationship between sensitivity and specificity for a single cutoff value. Positive and negative predictive values of ELISAs were calculated based on a 50% rate of relapse.²³

Twenty-six patients (11 men and 15 women; 19 with PV and 7 with PF) were included in the study. Baseline characteristics and the clinical course of patients are given in Table 1. The mean (SD) age was 54.0 (15.8) years. The mean (SD) duration of disease before inclusion was 45 (60) months. All 19 patients with PV had mucosal lesions, and 14 patients (74%) had both skin and mucosal lesions. The mean body surface area involved by skin lesions in the patients with PV or PF was 7% and 20%, respectively.

The median follow-up of patients was 22.0 months (interquartile range, 17.2-24.0 months). Of the 26 patients, 24 (92%) achieved complete remission after initial treatment, after a mean (SD) time of 75 (15) days. Treatment was considered to have failed in 2 patients (1 with PV and 1 with PF) according to the definition of the consensus statement.²⁰ Among the 24 patients who were in complete remission after initial treatment, 11 (8 with PV and 3 with PF) experienced a total of 13 relapses after a mean (SD) time of 12.7 (5.2) months.

Mean anti-Dsg1 and anti-Dsg3 Ab ELISA values at baseline and during the evolution are given in Table 2. Mean (SD) anti-Dsg1 Ab ELISA values of the 6 patients with PF who achieved complete remission of their skin lesions decreased from 123.8 (63.1) U/mL at baseline to 76.8 (58.6) U/mL at day 90 (P = .03). In contrast, anti-Dsg1 Ab ELISA values of the 1 patient with PF in whom treatment failed remained stable from 268 U/mL at baseline to 244 U/mL at day 90. Mean (SD) anti-Dsg1 Ab ELISA values of the 14 patients with PV with mucocutaneous lesions who achieved complete remission of their skin lesions decreased from 113.8 (63.5) U/mL at baseline to 22.3 (23.5) U/mL at day 90 (P < .001). Taken together, anti-Dsg1 Ab ELISA values of the patients with mucocutaneous PV or PF who achieved complete remission with therapy for their skin lesions decreased from 103.0 (65.9) U/mL at baseline to 34.8 (43.9) U/mL at day 90 (P < .001).

Mean (SD) anti-Dsg3 Ab ELISA values of the 18 patients with PV who achieved complete remission of their mucosal lesions after initial treatment decreased from 148.4 (59.3) U/mL at baseline to 99.4 (66.7) U/mL at day 90 (P < .001), whereas Dsg3 Ab ELISA values from the 1 patient with PV in whom treatment failed remained stable from 130.4 U/mL at baseline to 131.6 U/mL at day 90. One should notice that although statistically significant, the decrease of anti-Dsg3 Ab values after initial treatment in patients with PV was less dramatic than that of anti-Dg1 Abs in patients with mucocutaneous PV or PF. In particular, persistent high anti-Dsg3 or anti-Dsg1 Ab ELISA values at day 90 were observed in 8 of the 18 patients with PV who achieved complete remission of their mucosal lesions, compared with 1 of 6 patients with PF and 0 of 14 patients with mucocutaneous PV who achieved clinical remission of their skin lesions, respectively.

To ascertain the long-term evolution of anti-Dsg Abs during the course of the disease, we first evaluated mean anti-Dsg1 and anti-Dsg3 Ab ELISA values from all samples collected during periods of complete remission with therapy and those collected during skin and/or mucosal relapses. The mean (SD) anti-Dsg1 Ab ELISA value of the patients with PF was 55.7 (58.8) U/mL during periods of complete remission with therapy and 182.5 (30.8) U/mL during skin relapses. Corresponding figures in the patients with PV with skin lesions were 12.6 (15.0) U/mL during periods of complete remission with therapy and 48.1 (42.8) U/mL during skin relapses. Mean (SD) anti-Dsg3 Ab ELISA values of the patients with PV in complete remission with therapy or during relapses of mucosal lesions were 76.3 (66.1) U/mL and 149.4 (53.4) U/mL, respectively.

Then, we individually compared the evolution of anti-Dsg1 and anti-Dsg3 Abs in each of the 11 patients with relapse, between all samples collected during periods of complete remission with therapy and those collected during a relapse. Anti-Dsg1 Ab ELISA values were significantly higher during periods of skin relapses than during periods of skin remission in patients with PF or mucocutaneous PV (P = .03). In contrast, no significant difference in anti-Dsg3 Ab ELISA values was found in the patients with PV between periods of complete remission and periods of mucosal relapses (P = .13) (Figure 1).

Third, we evaluated the evolution of anti-Dsg1 and anti-Dsg3 Abs during the 6-month period before the skin and/or mucosal relapse (Figure 2).

Persistent high levels (>100 U/mL) or a rise in anti-Dsg1 Ab values before a skin relapse was observed in 7 of 8 cases, including 4 of 4 patients with PF and 3 of 4 patients with mucocutaneous PV, with higher rates of anti-Dsg1 antibodies among the patients with relapsing PF (182.5 [30.8] U/mL) than among the patients with skin-relapsing PV (48.1 [42.8] U/mL) (Figure 2A). Interestingly, all 4 patients with skin-relapsing PV had high anti-Dsg3 Ab values despite the absence of mucosal relapse, including the 1 patient with PV with no increase in anti-Dsg1 Ab value before the skin relapse (Figure 2B).

Persistent high levels or an increase in anti-Dsg3 Ab values before a mucosal relapse was observed in 5 of 5 patients with PV (Figure 2C) with no associated increase in anti-Dsg1 Ab values (Figure 2D).

In addition, we evaluated the anti-Dsg1 and anti-Dsg3 Ab ELISA values in serum samples from patients with PF or PV that were collected at the last visit before the relapse became clinically evident in patients with no cutaneous nor mucosal lesions. When considering serum samples collected 6 months before the relapse in patients with ongoing complete remission, the mean (SD) anti-Dsg1 Ab ELISA value of the 8 patients with skin relapse was 67.9 (80.8) U/mL. The corresponding figure for anti-Dsg3 Abs in the 5 patients with mucosal relapse was 128.7 (72.9) U/mL. Six patients (3 with PF and 3 with PV) had serum samples collected 1 to 3 months before the relapse. Mean (SD) anti-Dsg1 and anti-Dsg3 Ab ELISA values in these patients were 94.2 (82.9) U/mL and 148.5 (20.5) U/mL, respectively, and were slightly higher than corresponding ELISA values measured in serum samples collected during the 6-month period before the relapse.

Finally, we examined the long-term evolution of anti-Dsg1 and anti-Dsg3 Abs in patients with PF or PV with ongoing cutaneous and mucosal remission. Anti-Dsg1 Ab ELISA values lower than 20 U/mL were observed in 14 of 18 PV and PF patients with complete remission of their skin lesions with therapy (Figure 3A), whereas anti-Dsg3 Ab ELISA values lower than 14 U/mL were observed in only 3 of the 13 patients with PV with long-term remission of their mucosal lesions (Figure 3B).

We calculated the sensitivity, specificity, and positive and negative predictive values of anti-Dsg1 and anti-Dsg3 Ab ELISA values for the course of skin and/or mucosal lesions in patients with PV or PF (ongoing remission vs relapse). When using the cutoff values proposed by the manufacturer (ie, 20 U/mL for anti-Dsg1 Abs and 14 U/mL- for anti-Dsg3 Abs), anti-Dsg1 Ab ELISA values higher than 20 U/mL had an 86% sensitivity and a 78% specificity for the occurrence of a skin relapse, whereas anti-Dsg3 Ab ELISA values higher than 14 U/mL had a 100% sensitivity but a poor specificity of 23% for the occurrence of a mucosal relapse. In fact, 10 of the 13 patients with PV with persistent remission of their mucosal lesions had anti-Dsg3 Ab ELISA values higher than 14 U/mL. An ROC curve was performed to determine cutoff values for the prediction of skin and/or mucosal relapses (Figure 4). The values that produced the highest sum of sensitivity and specificity were defined as the theoretical cutoff values to be used. The cutoff value of anti-Dsg1 Abs calculated from the ROC curve was that proposed by the manufacturer (ie, 20 U/mL). It provided an 86% sensitivity, a 78% specificity, and a 79% positive and an 85% negative predictive value for the occurrence of skin relapses. Owing to the poor specificity of anti-Dsg3 Abs, the cutoff of anti-Dsg3 Abs calculated from the ROC curve was extremely high (130 U/mL), providing an 80% sensitivity, an 84% specificity, and an 84% positive and an 81% negative predictive value for the occurrence of mucosal relapses.

The main result of the present study is the demonstration of a close correlation between anti-Dsg1 Ab ELISA values and the course of skin lesions in patients with PV or PF, whereas anti-Dsg3 Ab ELISA values did not necessarily parallel the course of mucosal lesions in patients with PV. A relationship between ELISA scores and disease activity has been previously suggested in a few case reports,^14,17 but to our knowledge it had not been extensively investigated in a large series of patients with pemphigus until the present study.

Results of our ELISAs at baseline confirmed the previously reported correlation between the clinical phenotype of patients and recognition of Dsg1 and/or Dsg3 by their serum.^12,24 The clinical response to initial treatment of patients with PF or PV was significantly related to the evolution of anti-Dsg1 and anti-Dsg3 Abs, although anti-Dsg1 Abs decreased more dramatically than anti-Dsg3 Abs in patients who achieved complete remission. In fact, 10 of 18 patients with PV and only 1 of 6 patients with PF who achieved complete remission still had high anti-Dsg3 and anti-Dsg1 Ab ELISA values (>100 U/mL) at day 90, respectively.

During the long-term follow-up, a close relationship was found between anti-Dsg1 Ab ELISA values and the evolution of cutaneous lesions in patients with PV or PF, with notably higher anti-Dsg1 Ab ELISA values in patients with relapsing PF (182.5 [30.8] U/mL) than in patients with skin-relapsing PV (48.1 [42.8] U/mL). Indeed, the 20-U/mL cutoff of anti-Dsg1 Ab ELISA value provided a high sensitivity and specificity and a 79% positive and an 84% negative predictive value. Interestingly, high anti-Dsg3 Ab ELISA values were associated with high anti-Dsg1 Ab ELISA values in all patients with skin-relapsing PV, even in those with no associated mucosal relapse, which was in accordance with the compensation theory that states that both anti-Dsg3 and anti-Dsg1 Abs are necessary to trigger acantholysis in the epidermidis.²⁵

High anti-Dsg3 Ab ELISA values were observed in all patients with PV with a mucosal relapse. However, 7 of 13 patients with PV with ongoing remission of their mucosal lesions had high anti-Dsg3 Ab ELISA values (>100 U/mL). This proportion was even higher (10 of 13 patients) when using the 14-U/mL ELISA cutoff value proposed by the manufacturer, leading to a poor specificity of 23%. Because of this poor specificity, a high cutoff value of 130 U/mL was calculated from the ROC curve for anti-Dsg3 Abs, providing an 84% positive and a 81% negative predictive value. A dissociation of anti-Dsg1 and anti-Dsg3 Ab responses with a poor specificity of anti-Dsg3 Abs has initially been reported by Harman et al¹⁶ in occasional cases of quiescent PV with high ELISA values. Similarly, high anti-Dsg3 Ab ELISA values were recently reported by Kwon et al²⁶ in 46% of patients with PV in remission, defined as no eruption for more than 3 months with minimal prednisone therapy. These observations are in accordance with the demonstration of a higher pathogenicity of serum samples from patients with PV in active stage, compared with those in remission, despite similar ELISA values.²⁷ Several hypotheses can be proposed to explain the discrepancy between the persistence of high anti-Dsg3 Ab ELISA values in some patients with PV without mucosal lesions. First, a switch from pathogenic IgG4 in patients with active disease to nonpathogenic IgG1 in patients in remission can be considered, as previously reported by Warren et al²⁸ in patients with PF. However, in an extensive analysis of the immune response of patients with PV treated with rituximab, we still detected anti-Dsg3 IgG4 Abs in most patients with PV in clinical remission.²⁹ Cheng et al¹⁷ showed that a decrease of anti-Dsg3 Ab ELISA values could be evidenced by highly diluting the serum samples that initially failed to show a decline in anti-Dsg3 Ab ELISA values despite ongoing mucosal remission. However, we could only find such a decrease in a few of the highly diluted serum samples that we previously tested.²⁹ Interestingly, the decline of anti-Dsg3 Ab ELISA values was transient and followed by a reincrease of anti-Dsg3 Ab ELISA values during the second year after treatment despite ongoing clinical remission, confirming that these serum samples contained high levels of anti-Dsg3 Abs.³⁰ Finally, we were also able to demonstrate a dramatic decrease of anti-Dsg3 Abs reacting with the pathogenic epitopes of the EC1 and EC2 domains in a few patients with PV with ongoing mucosal remission and high anti-Dsg3 Ab ELISA values after rituximab treatment.¹⁹

Overall, this study demonstrated a dissociation between anti-Dsg1 and anti-Dsg3 Ab ELISA values. This dissociation should be taken into account by physicians in the management of pemphigus because a rise in anti-Dsg1 Ab ELISA values seems to have a good predictive value for the occurrence of a skin relapse in patients with PV or PF, whereas the predictive value of anti-Dsg3 Ab ELISA for the occurrence of a mucosal relapse is far from perfect. In particular, anti-Dsg3 Ab ELISA value should not be taken into account alone in the absence of clinical findings to make a therapeutic decision in patients with PV. However, persistent high levels or a rise in anti-Dsg1 Ab value should be taken into account before decreasing corticosteroid doses in patients with pemphigus.

Correspondence: Pascal Joly, MD, PhD, Department of Dermatology, Clinique Dermatologique, Hôpital Charles Nicolle, 1 rue de Germont, 76031 Rouen CEDEX, France (pascal.joly@chu-rouen.fr).

Accepted for Publication: August 17, 2008.

Author Contributions: Drs Abasq and Joly had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Abasq, Musette, and Joly. Acquisition of data: Abasq, Mouquet, Gilbert, Tron, Musette, and Joly. Analysis and interpretation of data: Abasq, Gilbert, Grassi, Musette, and Joly. Drafting of the manuscript: Abasq, Tron, and Joly. Critical revision of the manuscript for important intellectual content: Mouquet, Gilbert, Grassi, Musette, and Joly. Statistical analysis: Grassi and Joly. Administrative, technical, and material support: Abasq, Mouquet, Tron, Musette, and Joly. Study supervision: Gilbert, Musette, and Joly.

Financial Disclosure: None reported.

Additional Contributions: Richard Medeiros, Rouen University Hospital medical editor, edited the manuscript.