Extended treatment with efgartigimod and concomitant low-dose prednisone was gradually introduced in an open-label, adaptive phase 2 trial in participants with mild to moderate PV or PF (ClinicalTrials.gov: NCT03334058). Seven participants (7 PV) were treated for 15 weeks in cohort 3 and 15 participants (8 PV, 7 PF) were treated up to 34 weeks in cohort 4 as reported earlier (28). PV or PF diagnosis had to be confirmed by positive direct immunofluorescence and positive indirect immunofluorescence and/or Dsg-1/3 ELISA. Participants were either newly diagnosed or relapsing, with mild to moderate disease severity (pemphigus disease area index [PDAI] <45 at baseline) (29). The treatment period was preceded by a screening period of up to 3 weeks and followed by a treatment-free follow-up period of 10 weeks. In cohort 3, intravenous (i.v.) efgartigimod was administered weekly at 10 mg/kg for 4 weeks as induction followed by administrations every other week for 12 weeks as maintenance period. In cohort 3, efgartigimod could be initiated as monotherapy or in combination with 20 mg/day of prednisone at the discretion of the investigator, and prednisone could be tapered starting in the maintenance period. In cohort 4, i.v. efgartigimod was dosed weekly at 25 mg/kg body weight until EoC, defined as the time at which no new lesions have developed for a minimum of 2 weeks and the majority, i.e., approximately 80% of established lesions have healed. Thereafter, participants were dosed every other week. In cohort 4, efgartigimod was initiated with concomitant prednisone (20 mg/day) in all newly diagnosed participants and relapsing participants off therapy, or at the tapered dose at which relapse occurred. The oral prednisone dose could be tapered as of EoC. No other systemic treatments for pemphigus were permitted during the study, but topical corticosteroids, analgesics, and supportive care for corticosteroid therapy (e.g., vitamin D, proton-pump inhibitors, and specific diets) were allowed (28).

The study was conducted in accordance with the Good Clinical Practice guidelines, in conformity with the ethical principles of the Declaration of Helsinki and was compliant with all relevant country-specific laws. The study protocol and all other appropriate study-related information were reviewed and approved by the ethics committees or institutional review boards of every participating center.

Photography was conducted as routine practice for documenting dermatological conditions in medical practice. Additional participant consent was obtained for sharing anonymized participants’ pictures with the sponsor.

Pharmacodynamic analysis included determination of serum levels of total IgG, IgG subclasses, and anti-Dsg-1 and -Dsg-3 autoantibodies by ELISA (Euroimmun, Germany). Serum levels of protective vaccine antibodies against tetanus toxoid (TT; Indirect EIA, Virotech), varicella zoster virus (VZV; CLIA, Diasorin), and pneumococcal capsular polysaccharide (PCP; EIA, The Binding Site Group) were measured for all participants. A customized Addressable Laser Bead ImmunoAssay (ALBIA) test was performed to determine the anti-Dsg-3 IgG subclasses at different time points. The ALBIA method used was previously described (30). Briefly, 20 μg of anti-HIS antibody (clone HIS.H8) was coupled to 1.25 × 10⁶ fluorescent Bio-Plex R COOH-microspheres (Bio-Rad, USA) with the Bio-PlexR amine coupling kit (Bio-Rad) according to the manufacturer’s protocol. After coupling, coated beads were either used immediately or stored at −20°C in the dark. Before use, 100 ng of recombinant Dsg-3 expressing a 6×His Tag were incubated with 1,000 beads for 15 min at room temperature (RT) and then washed. Immediately prior to their use, coated beads were vigorously agitated for 30 s. Then, 10 μl of beads (containing 1,000 beads) was added to 100 μl of participants’ or control serum diluted 1/100 in PBS with Ca²⁺ and Mg²⁺ supplemented with 1% FCS, in Bio-Plex Pro Flat bottom plates (Bio-Rad). Plates were incubated for 1 h 30 min at RT in the dark on a plate shaker at 850 rpm. Blank (no serum), negative controls (healthy donor serum), and positive controls (human anti-Dsg-3 positive serum) were included in every assay. Beads were collected with a magnetic washer (Bio-Rad) and washed three times with 150 μl of PBS 1× containing 0.1% Tween-20. Biotinylated mouse anti-human IgG-subclass specific secondary antibody (Southern Biotech, USA) was added (at 1/125 dilution for anti-IgG1 and for anti-IgG2, at 1/200 dilution for anti-IgG3 and for anti-IgG4) for 45 min at RT under shaking. After washing, beads were incubated with 50 μl of streptavidin-R-phycoerythrin at 1/400 dilution for 15 min. Finally, beads were resuspended in 100 μl of PBS and mean fluorescence intensity (MFI) was determined on a Bio-PlexR apparatus using the Bio-PlexR Manager Software 4.0 (Bio-Rad). The anti-Dsg-3 IgG subclass serum levels were determined with the following formula: (MFIserum/MFICalibrator) × 100, in which the calibrator was an anti-DSG3 positive control used on every analysis and set to 100 arbitrary units (AU). For determination of Dsg-3-reactive antibodies, the positivity threshold was set at the mean value obtained from 36 healthy donors plus 2 standard deviations. EUROIMMUN anti-Dsg-1 ELISA (Euroimmun, Germany) with modifications to the detection system was employed to identify anti-Dsg-1 IgG subclasses. Rabbit anti-human IgG HRP detection antibody was replaced by anti-human IgG1 or anti-human IgG4 biotinylated (Southern Biotech, USA), followed by streptavidin HRP used at 1/10,000 dilution to detect Dsg-1 specific IgG subclasses and used at 1/10,000 dilution. CIC-C1q EIA kit (A001, Quidel) was employed to detect levels of C1q-associated IgG aggregates (circulating immune complexes [CIC]) in sera of selected participants at different time points according to the manufacturer’s protocol. Calibrators provided by the kit were used to determine expression level.

A total of 60–80 ml of whole blood was collected into 10 BD Vacutainer CPT tubes, remixed immediately and centrifuged within 2 h of blood collection at room temperature to separate peripheral blood mononuclear cells (PBMCs) and red blood cells. Following centrifugation, PBMCs were resuspended into the plasma and transported to the analytical laboratory of the Department of Dermatology at the Philipps University Marburg, Germany, within 24 h. Upon sample receipt, tubes were subjected to PBMC purification, flow cytometry analysis, and cryopreservation of the remaining cells.

PBMCs were washed twice with PBS supplemented with 1% FCS and 1 × 10⁶ cells were stained for T and B lymphocyte subset analysis including detection of Dsg-3-specific B cells as previously described (13). The following antibodies were used for T lymphocyte panel analysis: mouse anti-human CD4 (RPA-T4, BioLegend), mouse anti-human CD45RA (HI100, BioLegend), mouse anti-human CXCR5 (J252D4, BioLegend), mouse anti-human CD25 (M-A251, BioLegend), mouse anti-human CD127 (A019D5, BioLegend), mouse anti-human CXCR3 (G025H7, BioLegend), and mouse anti-human CCR6 (G034E3, BioLegend). The gating strategy is shown in Supplementary Figure 1A for one representative participant. The following antibodies were used for B-cell panel analysis: mouse anti-human CD45 (2D1, BioLegend), mouse anti-human CD19 (HIB19, BioLegend), mouse anti-human CD27 (M-T271, BioLegend), mouse anti-human CD38 (HB-7, BioLegend), mouse anti-human CD24 (ML5, BioLegend), mouse anti-human IgM (MHM-88, BioLegend), mouse anti-human IgD (IA6-2, BioLegend), and mouse anti-human CD138 (MI15, BioLegend). The gating strategy is shown in Supplementary Figure 1B for one representative participant. AlexaFluor 647-labeled recombinant human Dsg-3 (extracellular domain, aa 1-566), produced in the baculovirus expression system (31), was included in a separate B-cell staining panel. AlexaFluor 647-labeled recombinant collagen 7, produced in the same way as Dsg-3, was used as a control for a non-specific binding. The gating strategy is shown in Supplementary Figure 1C for one representative participant.

The frequencies of circulating total IgG and Dsg-specific IgG-antibody secreting cells (ASCs) were determined by human IgG ELISPOT Basic (Mabtech, Sweden) as described (32). In brief, PBMCs from participants in the trial were pre-stimulated with R848 (1 μg/ml) and rhIL2 (10 ng/ml) in complete medium (RPMI-1640 supplemented with 10% FCS, 2 mM L-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin [Thermo Fisher Scientific]) in 24-well plates for 72 h at 37°C. Plates of ELISPOT MAIPS-4510 (Merck Millipore, Germany) were coated overnight at 4°C with anti-IgG human antibody. Plates were washed and blocked with complete medium before use. Pre-stimulated PBMCs were washed, resuspended in complete medium, and transferred to the plate, and then incubated for 24 h with 1 × 10⁵ to 4 × 10⁵ PBMCs per well to detect anti-Dsg-1 and -Dsg-3 IgG-secreting ASC, and with 2.5 × 10³ to 1 × 10⁴ PBMCs per well to detect total IgG-ASC. IgG-ASCs were detected by addition of biotinylated mouse IgG anti-human IgG. Frequency of anti-Dsg-1 or anti-Dsg-3 IgG ASC was calculated after incubation for 2 h with histidine-tagged recombinant Dsg-1 or Dsg-3 proteins (1 μg/ml) in PBS with Ca²⁺ and Mg²⁺. Biotinylated anti-histidine antibodies (0.5 μg/ml) (Abcam, United Kingdom) were then added. Streptavidin-conjugated peroxidase and tetramethylbenzidine as substrate were used to detect spots. The number of spots were determined using ELISPOT Plate Readers and ImmunoSpot software (CTL Europe GmbH, Germany). Results were expressed as frequencies of Dsg-specific IgG-ASC among total IgG-ASC. The sensitivity and specificity of the ELISPOT assay were estimated at 1 Dsg-3-specific ASC/10⁵ total ASC, and 100%, respectively.

Descriptive statistical methods were used to analyze data. Summaries (mean, standard error, median, and range) were plotted graphically by study days/time points.

The characteristics of participants with PV and PF in cohorts 3 and 4, who received extended treatment with efgartigimod (15 and up to 34 weeks, respectively), are shown in Table 1 . Figure 1A shows the clinical outcomes of the patients. Nine participants in cohort 4 (participants 1, 3, 4, 5, 7, 8, 9, 11, and 12) reached CR at any point during the trial while receiving efgartigimod (until day 239) and low-dose prednisone, of which six (participants 1, 4, 5, 7, 8, and 9) were in CR at the end of the study (end of 10-week efgartigimod-free period). Four participants reached complete clinical remission on minimal therapy (CRmin), defined as the absence of new or established lesions while the patient is receiving minimal therapy where minimal therapy is defined as less than or equal to 10 mg/day of prednisone (or the equivalent) for at least 2 months (33), at any point during the study. Participant 3 exhibited low disease activity with a PDAI activity score of 3 following CR and two participants (11 and 12) relapsed to a PDAI activity score >10. Three participants (2, 6, and 10) improved clinically and maintained clinical improvement status until end of study: participants 2 and 10 reached and maintained EoC, and participant 6 reached DC and had a PDAI activity score of 1 at end of the study. In cohort 3, five of seven participants achieved CR at any point during the study (participants 16, 19, 20, 21, and 22), and four of these participants were in CR at the end of treatment-free follow-up (participants 16, 20, 21, and 22).

In order to characterize the observed sustained clinical responses, representative participants with PV (participant 8) and PF (participant 4) from cohort 4 are shown in Figure 1B . Participant 8 with PV who had a PDAI activity score of 10.3 at baseline showed a decline of disease activity starting after the first treatment and mucosal blisters had healed by day 15. Skin lesions of participant 4 with PF who had baseline PDAI activity of 34.5 declined gradually over time and resolved completely by day 281. Importantly, both participants showed no disease activity during the 10-week observation period after last efgartigimod treatment.

In the participants with sustained clinical response (whose PDAI activity and serological responses are summarized in Supplementary Table 1 ), total serum IgG and pathogenic autoantibodies declined early. Total IgG levels remained suppressed to about 70% during the treatment period, and recovered to baseline levels during treatment-free follow-up, in line with what has been previously reported for the overall study population (28). Anti-Dsg antibodies declined early and prolonged reductions were observed in all responding participants ( Figure 1C ). Unlike protective antibody titers (anti-VZV, anti-TT, and anti-PCP) that returned to baseline levels at the end of the study, stable suppression of anti-Dsg-1 antibodies and anti-Dsg-3 antibodies to lower levels than before treatment were observed during the 10 weeks of efgartigimod-free follow-up ( Figure 1D ).

Since PV and PF primarily involve IgG4 (34), and other IgG subtypes to a lesser extent, ALBIA Dsg-3 assay was performed in six participants with PV, and modified anti-Dsg-1 Euroimmun ELISA in six participants with PF who received extended efgartigimod treatment in cohorts 3 and 4 and achieved a sustained clinical response (28). Participants evaluated in these analyses are specified in Table 1 . At baseline, we detected heterogenous anti-Dsg-1/3 IgG1-4 subclasses in different participants. In participants with baseline levels above the positivity threshold, we observed reduction of anti-Dsg-3 IgG1 autoantibodies to below the positivity threshold in 3 of 3 participants and anti-Dsg-3 IgG4 in 2 of 3 participants ( Supplementary Figure 2A ). Anti-Dsg-3 IgG2 autoantibodies, present in 3 participants, were markedly reduced although only 1 subject decreased below the positivity threshold at end of the study; IgG3 autoantibodies present in one participant at baseline became suppressed below the positivity threshold. The six participants with PF subjected to subclass analysis demonstrated a dominant IgG4 signature of anti-Dsg-1 autoantibodies ( Supplementary Figure 2B ), while one participant presented with IgG1 autoantibodies. In this patient population, analysis of Dsg-1 IgG titers over time showed reduction of Dsg-1-specific IgG1 and IgG4 antibodies following clinical response, which remained suppressed at the end of the study. Thus, decreases in anti-Dsg titer regardless of the IgG subtype analyzed were observed following efgartigimod treatment in this subpopulation.

Immunocomplexes are found in healthy individuals (35), but their formation can be expected to be elevated in autoimmune diseases and partly drive and/or exaggerate their pathologies (36). As the half-life of immune complexes, largely containing IgG, may also be affected by the biology of FcRn (35), we investigated levels of CIC in the same six participants with PV and six participants with PF. These participants were subjected to IgG CIC analysis by C1q ELISA, which detects complement-fixed IgG antibodies (without providing information about antigen specificity of CICs). IgG CIC levels are considered clinically significant if ≥4.0 μg Eq/ml. Four participants presented with elevated CIC levels, but notable reduction of CICs during treatment was observed ( Supplementary Figures 3A, B ), consistent with the observed improvement in their clinical condition. Rise of CICs following efgartigimod discontinuation did not lead to rise in disease activity, suggesting that for these participants, CICs were not directly linked with the disease activity.

The observed differences of anti-Dsg-1/3 titers compared to overall serum IgG prompted us to probe the PBMCs of participants with PV to determine the fate of Dsg-3-specific B cells that could potentially explain the prolonged suppression of anti-Dsg autoantibodies. PBMCs were collected in cohort 4 participants only (n = 15), and were successfully isolated in 12 participants (5 PV, 7 PF), of whom nine achieved a sustained clinical response (3 PV, 6 PF).

Frequency of Dsg-3+ B cells in the periphery was identified by staining for CD45+, CD19+, and CD27+ memory B cells (MBCs) and fluorescently labeled Dsg-3 antigen ( Figure 2A ) as previously described (13). Additional staining with IgM and IgD was included prior to gating on Dsg-3+ cells to identify class-switched memory B cells. Dsg-3+ B cells were identified within CD27+ IgM- IgD- cells, known to harbor antigen-specific B cells (37, 38) ( Figure 2B ). In line with previous reports, Dsg-3-specific B cells were rarely seen in peripheral blood but were detected principally at baseline and were detected at higher frequencies in participants with higher anti-Dsg-3 serum IgG levels in the samples analyzed. A reduction of antigen-specific MBCs was visible when participants 5, 6, and 8 reached CR and at the end of treatment. Two PV participants that relapsed following CR also exhibited reduction of Ag-specific B cells at CR ( Supplementary Figure 4 ) and no increase in frequency of Dsg-3+ B cells was further observed, possibly implicating persistent tissue-resident antigen-specific responses in relapse (39).

In addition to assessing the frequency of Dsg-3+ B cells in participants with PV, Dsg-1 specific ASCs were measured by ELISPOT. Although ELISPOT assays were performed in all samples, low cell viability after thawing and withdrawal of participant’s consent for post-hoc analysis restricted data analysis to three participants who are presented here (participants 3, 4, and 7). At baseline, anti-Dsg-1 IgG ASCs were detected in all 3 PF participants assessed and were detected at higher frequency in participants with higher serum anti-Dsg1 levels ( Figures 2C, D ). Following treatment with efgartigimod and concomitant low-dose prednisone, anti-Dsg-1 IgG ASCs were no longer detectable at the end of the study. For two participants, disappearance of anti-Dsg-1 IgG ASC was detected at the CR visit (on days 44 and 275, respectively). One participant demonstrated loss of anti-Dsg-1 IgG ASC at the end of treatment, while CR was achieved 47 days later during treatment-free follow-up.

Beyond autoantibody-producing B cells, classical type-2 T helper (Th2) cells are typically considered central players in pemphigus pathology; however, recent advances in clinical immunology pointed to additional T-cell subsets, namely, T helper-17/T follicular helper-17 (Th17/Tfh17) cells efficiently promoting autoantibody production in participants with pemphigus (7, 8, 40). Thus, we then investigated whether prolonged treatment had an impact on major T-cell and B-cell subsets in addition to the antigen-specific context. Importantly, treatment did not impact global levels of total leukocytes, total lymphocytes, monocytes, or neutrophils ( Figure 3A ). Longitudinal analysis of PBMCs at 4 different time points in the 9 participants from cohort 4 with viable PBMC did not reveal changes in median overall T-cell subpopulations. CD4+ T cells and subsets were stable with no significant changes in Th1, Th2, Th17 or Tfh1, Tfh2, Tfh17 subsets observed ( Figure 3B ).

Baseline frequency of total B cells as a proportion of total lymphocytes was heterogenous and ranged from 6.2% to 30.9%, indicating high B-cell frequency in some individuals possibly due to B-cell activation and proliferation. Analysis of B cells revealed declining numbers of median total CD19+ B cells in the periphery of all nine participants but without affecting the composition of B-cell subsets for the markers tested including CD27+ memory cells ( Figures 3C, D ). After treatment, median levels of CD19+ B cells remained within normal ranges.

The highest B-cell frequencies in the trial were observed in participants 11 and 12 who relapsed to PDAI activity >10 after having achieved CR. At the last evaluable time point, no increases from baseline in B-cell frequency were noted. Participant 13 with PF showed no clinical response to efgartigimod and did not demonstrate changes in CD19+ B cells ( Supplementary Figure 5 ).

In order to investigate the potential impact of prednisone on the observed pharmacodynamic and immunological effects described, concomitant prednisone exposure was analyzed. As seen in the overall study population, early clinical efficacy of efgartigimod allowed for early tapering of steroids. This effect was further pronounced in participants who achieved sustained clinical response after prolonged treatment with efgartigimod in cohorts 3 and 4. In participants who achieved CR at any point in the study, the average daily prednisone dose until CR was 0.255 mg/kg/day for cohort 4 participants (n = 9) and 0.18 mg/kg/day for cohort 3 participants (n = 5) ( Table 2 ), and reduced frequencies of B cells were not associated with higher prednisone dosages ( Figure 4 ).