Introduction

Pemphigus is an autoimmune blistering disease involving the skin and/or mucosae with autoantibodies targeting desmosomal transmembrane proteins. Over the last few years, rituximab has emerged as an effective first-line therapy for pemphigus. However, frequent relapses (22-100%) remain a challenge.1-3 Identification of biomarkers to predict relapse would be a step towards personalised medicine, as it can aid in stratifying at-risk patients. However, previous studies aimed at identifying relapse biomarkers have led to conflicting results.4-6 We conducted this study to analyse the change trends of immunological biomarkers and predictors of relapse in Indian patients with pemphigus treated with rituximab.

Methods Study design and participants

This was a prospective observational cohort study conducted at All India Institute of Medical Sciences, New Delhi, India from August 2021 to September 2024. The protocol was registered (CTRI/2020/10/028675) and approved by the Institute Ethics Committee. Adult (≥18 years) patients with active histologically confirmed pemphigus vulgaris or foliaceus treated with a rituximab biosimilar (1000 mg infusion on days 0 and 14) along with or without oral prednisolone and/or oral conventional immunosuppressive agents (azathioprine, mycophenolate mofetil, methotrexate, and/or dapsone) were included in the study after informed consent. Oral prednisolone was started at 0.5-1 mg/kg/day according to disease severity and tapered gradually after disease control was achieved, with a plan to stop it in 3-6 months. Pregnant and lactating mothers, patients with inactive pemphigus, those who had received rituximab in the last 12 months, or had other concomitant autoimmune diseases were excluded.

A thorough history and clinical examination, including the Pemphigus Disease Area Index (PDAI) and Pemphigus Oral Lesions Intensity Score (POLIS), were performed at baseline on the day of rituximab infusion. Patients were clinically followed-up at 2 weeks, 1 month, 3 months, and then every 3 months till relapse or till 2 years, whichever was earlier. Venous blood samples were collected at baseline, 3 months, 6 months, 9 months, and 12 months or until relapse, whichever was earlier. Serum anti-DSG1, anti-DSG3, and anti-AchRM3 antibody levels were measured using Enzyme-linked immunosorbent assay (ELISA) kits, and peripheral blood total CD19+ B-cell and CD19+CD27+ memory B-cell populations were analysed by flow cytometry, as described previously.7

Study definitions

Relapse was defined as per the international expert consensus guidelines as the appearance of three or more new lesions in 1 month, which did not subside spontaneously within a week, in a patient who had achieved remission.8 Early relapse was defined as relapse occurring within 12 months of rituximab treatment.4

Incomplete B-cell depletion was defined as B-cells >0.5% of total lymphocytes at the first measurement time point after rituximab infusion (i.e., at 3 months).9 B-cell repopulation was defined as B-cells >1% of total lymphocytes at any time-point between 6-12 months after rituximab treatment.10

Statistical analysis

We compared the baseline clinical (PDAI, POLIS, affected body surface area, pemphigus type, disease duration, starting prednisolone dose, use of oral immunosuppressive adjuvants) and immunological parameters (anti-DSG1, anti-DSG3, anti-AchRM3, total CD19+ B-cells and CD19+CD27+ memory B-cells) at different time points between relapsing and non-relapsing patients. This comparison was done for relapse at 1 and 2 years. Continuous variables were analysed using the Mann–Whitney U test, while categorical variables were compared with the Chi-square test. A p-value of <0.05 was considered statistically significant. A receiver operating characteristic (ROC) curve was generated for parameters with a statistically significant difference to determine cut-off values that predict relapse with optimum sensitivity, specificity, and positive and negative predictive values (PPV and NPV).

Immunological parameters were longitudinally analysed to assess the changing trends at various time points among three patient subsets, viz. relapsers within 12 months (early relapse), relapsers between 12-24 months, and non-relapsers till 24 months, using the Mann–Whitney U test. In the trend analysis of immunological parameters, differences between two time points in the same group were calculated using the Wilcoxon signed-rank test, and intervals showing significant changes were identified. ROC analysis was then used to establish the cut-off percentage change that best predicted relapse.

To predict relapse in a dynamic fashion with regard to B-cell repopulation, data were prepared in the counting process format, and the Kaplan-Meier survival curve and log-rank test were used. Statistical analysis was conducted with Stata 16 software (StataCorp, College Station, TX).

Results

Fifty-two patients were included in the study, with baseline characteristics of 49 of them reported previously.7 Of these 52 patients, three died within the first month (one due to paradoxical disease flare and sepsis, one from sepsis, and one from unknown causes), while five were lost to follow-up within six months, leaving 44 patients for the final analysis.

Patient profile

Out of the 44 included patients, 40 had pemphigus vulgaris and 4 had pemphigus foliaceus. Besides rituximab infusion, 41 patients received other concomitant treatment; 26 (63.4%) patients received oral prednisolone along with conventional immunosuppressive agents, 12 (29.2%) received only oral prednisolone, and 3 (7.3%) received only conventional immunosuppressive agents. Three (7.3%) patients received rituximab monotherapy. 

Most patients (n=40, 91%) had received prior systemic treatment, including oral steroids (n=37, 84%), pulsed steroids (n=7, 16%), conventional oral immunosuppressives (n=35, 80%), and/or rituximab (n=14, 30%). Only 4 (9%) patients were treatment-naïve, while others had either relapsed or had an inadequate response to prior therapies.

Clinical course and relapse rates

Disease control was achieved in all patients in a mean time of 3.9±2.2 (range 1–10) weeks. Complete remission off prednisolone was achieved in 43 patients, in a mean time of 6.4±2.1 (range 3.5–11.3) months, while the remaining one patient maintained complete remission on minimal prednisolone (<10 mg per day) therapy.

Relapse occurred in 24 patients (54.5%) by 2 years, with 18 (75%) of them happening within 12 months (early relapse). The peak time to relapse was between 6-9 months, when 50% (n=12/24) of all the relapses happened [Figure 1]. The mean time to early relapse (n=18) was 8.6±1.7 (range 6.5-12) months, while the mean time to relapse between 12-24 months (n=6) was 16.1+3 (range 12.5-21) months.

Figure 1: Timeline of relapse.

Export to PPT

Longitudinal trends in immunological biomarkers in early relapsers (<12 months), relapsers between >12-24 months, and non-relapsers

Table 1 shows the baseline clinical and immunological parameters of relapsing and non-relapsing patients. The immunological variables at different time points are shown in Supplementary Table S1.

Table 1: Baseline clinical and immunological parameters in relapsing and non-relapsing patients at different time points

Early relapse (n=18) No relapse till 12 months (n= 26) P value Relapse within 24 months (n=24) No relapse till 24 months (n= 20) P value Clinical variables Mean age (years) 40.5 ± 11.42 (29-64) 36.38 ± 11.23 (18-64) 0.48 37.8 ± 10.2 (22-8) 32.2 ± 11.1 (17-1) 0.06 Mean duration of illness (months) 35.33 ± 25.03 (3-84) 37.94 ± 46.7 (2-216) 0.97 31.6 ± 23.1 (3-84) 37.6 ± 39.3 (2-120) 0.71 Pemphigus type (n, %) 1.00 0.73 Vulgaris 17 23 22 18 Foliaceus 1 3 2 2 Disease onset (n, %) Mucosal 14 12 17 9 Cutaneous 4 12 0.06 7 9 0.1 Mucocutaneous 0 2 0 2 Prednisolone used 16 (88.89%) 22 (84.6%) 0.96 21 (87.5%) 16 (80%) 0.79 Oral immunosuppressive adjuvant used (n, %) 11 (61.1%) 18 (69.2%) 0.68 15 (62.5%) 14 (70%) 0.60 Mean PDAI score 31.72 ± 15.36 (12-71) 24.04 ± 15.9 (2-64) 0.07 28.9 ± 15.2 (12-71) 25.05 ± 17.3 (2-64) 0.31 PDAI severity (n, %) <15 2 8 5 5 15-44 13 15 0.33 16 12 0.82 >45 3 3 3 3 Mean mucosal PDAI score 10.89 ± 8.18 (0-28) 7.07 ± 6.8 (0-27) 0.13 10 ± 7.9 (0-28) 7 ± 7.2 (0-27) 0.16 Mean POLIS score 12.39 ± 9.58 (0-33) 9.03 ± 7.7(0-25) 0.28 12.3 ± 9.2(0-33) 8.2 ± 7.6(0-25) 0.16 Immunological variables Mean anti-DSG1(RU/mL) 633.61 ± 714.62 (0-2697.33) 525.8 ± 420.0 (7.8-1495.11) 0.91 584.94 ± 664.14 (0- 2697.33) 549.41 ± 401.82 (7.8-1495.11) 0.76 Mean anti-DSG3(RU/mL) 624.36 ± 469.21 (0-1525.33) 568.58 ± 683.14 (0-3070.67) 0.45 626.81 ± 507.6 (0-1572.44) 549.03 ± 704.18 (0- 3070.67) 0.46 Mean anti-AchRM3 (ng/mL) 47.04 ± 101.1 (0-421.36) 69.21 ± 137.78 (0- 515.04) 0.63 84.22 ± 146.61 (0-515.04) 31.25 ± 91.14 (0- 406.71) 0.23 CD19+ B-cells (% of total lymphocytes) 6.9 ± 4.4 (0.4-15.2) 12.04 ± 7.46 (0.1-27.2) 0.02 7.3 ± 5.99 (0.1-27.2) 12.9 ± 6.5 (1.3-26.4) 0.003* CD19+CD27+ B-cells (% of total B-cells) 42.5 ± 26.3 (5.9-100) 37.31 ± 23.42 (8.5-100) 0.44 46.37 ± 29.36 (5.9-100) 31.6 ± 14.7 (8.5- 59.7) 0.12 Anti-DSG antibodies and anti-AchRM3 antibodies kinetics

All three patient subsets showed a marked decline in anti-DSG3 and anti-DSG1 levels by 3 months after rituximab administration (p<0.05).

Early relapsers showed a rising trend of mean anti-DSG3 levels between 3-6 months (97.8±81.8 to 163.05±105.9 RU/mL, p=0.01), while those who relapsed later showed a delayed rise between 9-12 months (126.63±137.94 to 175.72±159.43 RU/mL, p=0.04). The patients who did not relapse till 24 months maintained a stable anti-DSG3 trend at 3-12 months [Figure 2a].

Figure 2: Line plot with interquartile range showing comparative longitudinal trends in immunological biomarkers in the three patient groups: early relapsing patients, patients relapsing between 12–24 months, and non-relapsing patients (*indicates p value < 0.05): a) Anti-DSG3, b) Anti-DSG1, c) Anti- AchRM3 d) CD19+ B-cells, e) CD 19+ 27+ memory B-cells

Export to PPT

For anti-DSG1, early relapsers similarly showed a rising trend between 3-6 months (79.26±83.69 to 107.69±110.56 RU/ml, p=0.03), but the other two groups did not show a statistically significant rise at any time interval [Figure 2b].

Anti-AchRM3 levels did not show any significant trends in any of the three patient subsets[Figure 2c].

Total and memory B-cell kinetics

Total CD19+ B cells declined sharply in all patient subsets by 3 months (p<0.05).

Early relapsers experienced a statistically significant rise in total CD19+ B cells between 3-6 months (0.8±2% to 3.8±6%, p=0.003), while those who relapsed later showed a statistically non-significant CD19+ B-cell rise between 6-9 months (1.9±3.6% to 5.3±4.7%, p=0.32). In contrast, patients who remained relapse-free did not show significant B-cell increases at any time interval [Figure 2d].

Memory B cells (expressed as a percentage of total B cells) significantly increased (p<0.05) at three months in all patient subsets. Subsequently, as effector B-cells repopulated, memory B-cells declined between 3-6 months (93.8±25% to 63.8±39.8%, p=0.03) in early relapsers only [Figure 2e].

Incomplete B-cell depletion and B-cell repopulation

Incomplete B-cell depletion was seen in 4 (9.1%) patients. Early relapse rates were statistically significantly higher in patients with incomplete B-cell depletion (n=4/4, 100% vs. n=12/34, 35%; p=0.02).

The mean time to B-cell repopulation was 8.1±2.2 months (range 6-12 months). The mean time to relapse following B-cell repopulation was 4.4±3.5 (median 3.5, Interquartile range (IQR) 2-5.8) months. Of the 18 patients with early relapse, 4 (22.2%) relapsed before B-cell repopulation.

Patients with B-cell repopulation at 6 months had a statistically non-significant higher incidence of relapse at 1 year (n=8/14, 57.1% vs. 10/30, 33.3%; p=0.19) and at 2 years (n=10/14, 71.4% vs.14/30, 46.6%; p=0.19) than those without. Kaplan-Meier plot, using the log-rank test, did not show a statistically significant difference in the relapse rates based on B-cell repopulation (p=0.20) [Figure 3a].

Figure 3: B-cell repopulation and pemphigus relapse timeline: a) Kaplan-Meier plot for relapse, with stratification of patients by CD19+ B cell %, b) Patients who relapsed as a proportion of patients with B-cell repopulation at different time points.

Export to PPT

Of the 29 patients who experienced B-cell repopulation by 12 months, 16 (55.2%) relapsed within 2 years. The relapse rates decreased with increasing time to B-cell repopulation (p=0.17) [Figure 3b].

Predictors of relapse

ROC curve analysis determined cut-offs for absolute values and increase in anti-DSG1 and anti-DSG3 levels as predictors of relapse at different time points within 1 and 2 years are shown in Supplementary Tables S2-S3.

Incomplete B-cell depletion at 3 months was a predictor of early relapse with a sensitivity of 22.2% (95% CI 6.4-47.6%) and specificity of 100% (95% CI 86.85-100%).

At month 6, 73.3% (n=11/15) of patients with early relapse had anti-DSG3 ≥120 RU/ml [Figure 4a], as compared to 27.3% (n=6/22) of patients without early relapse (p=0.008). The combination of incomplete B-cell depletion at 3 months and/or anti-DSG3 ≥120 RU/mL at 6 months had a sensitivity of 84.6% (95% CI 54.6–98.1%) and a specificity of 70.0% (95% CI 45.7–88.1%) with a positive predictive value of 64.7% (95% CI 47.5-78.8%) and negative predictive value of 87.5% (95% CI 65.5-96.3%) in identifying patients at risk of early relapse [Figure 4b].

Figure 4: Immunological predictors of early relapse: a) Receiver operating characteristic curve to determine the optimal anti-DSG3 cut-off at 6 months to predict early relapse, b) B-cell repopulation at 3 months and/or anti-DSG3 titer ≥120 RU/ml at 6 months as predictors of early relapse.*Data available for 16 relapsing and 22 non-relapsing patients. **Data available for 15 relapsing and 22 non-relapsing patients. DSG: Demoglein, ROC: Receiver operating characteristic, AUC- Area under the curve, RU- Relative units, CI: Confidence interval, PPV: Positive predictive value, NPV- Negative predictive value

Export to PPT

We also tested the validity of the criteria proposed by Mignard et al.4 in our cohort, but the baseline PDAI≥45 (16.7% vs. 11.5%, p=0.67), anti-DSG1 ≥20 RU/mL at 3 months (55.6% vs. 42.3%, p=0.54), and anti-DSG3 ≥130 RU/mL at 3 months (38.9% vs. 34.6%, p=0.36) were comparable in early relapsing and non-relapsing patients.

Discussion

The 2-year relapse rate in our study with one cycle of rituximab-RA protocol was 54.5%, with most relapses (n=18, 75%) happening between 6-12 months. This is consistent with the RITUX-3 trial, where 73% (n=8/11) of the relapses occurred within the first year.1A maintenance rituximab infusion at 6 months has therefore been suggested to prolong the disease-free period in some patients, and predictive biomarkers may aid in identifying high-risk patients, leading to a more personalised treatment approach.4-6

We found that early relapsing patients show an earlier rising trend of anti-DSG3 (between 3-6 months) and later relapsing patients show a delayed rise (between 9-12 months) of anti-DSG3, while patients who remained relapse-free for 2 years exhibited no significant immunological changes. A previous Indian study also showed a rising anti-DSG3 titre at 3-6 months post-remission to be associated with relapse at 9 months.6 Further, our observations in the later relapsing group illustrate the sequence of immunological events preceding relapse: a >2-fold increase (from 1.9% to 5.3%) in total CD19+ B-cells between 6-9 months was followed by an increase in anti-DSG3 between 9-12 months, while anti-DSG1 did not rise in this group. While the change in total B-cells lacked statistical significance, probably due to the small number of non-relapsing patients in this time period, it is possible that there was delayed anti-DSG1 elevation beyond our data points. Anti-DSG1 rise post-treatment has been shown to lag behind anti-DSG3 previously as well.11 Thus, changes in B-cell population and anti-DSG3 levels appear to be better predictors of relapse than changes in anti-DSG1. Our results, while in contrast to some previous studies, align with those of Albers et al.5 who also found anti-DSG3 to be a more reliable predictor of relapse than anti-DSG1.

We found a higher baseline mean PDAI (31.72 vs. 24.04) and mucosal disease onset (78 vs. 46%) to show a statistically significant trend towards early relapse, consistent with previous literature.4,6 The finding of baseline mean percentage of CD19+ B-cells being significantly higher in non-relapsers was unexpected, but has been reported in a recent multicentre study as well.12 The clinical significance of this finding is not clear, as we could not estimate the absolute number of B-cells in the blood. Nonetheless, incomplete B-cell depletion was seen only in relapsing patients and was shown to predict early relapse with 100% specificity, but poor sensitivity (22%). Though not studied in pemphigus before, incomplete B-cell depletion after rituximab treatment has been shown to be associated with partial response and higher risk of relapse in lupus nephritis and IgG4-related disease.13,14Although incomplete B-cell depletion was a strong predictor of relapse in our cohort, its low sensitivity precluded its use as a standalone marker. We therefore combined it with another biomarker to develop a set of criteria with better predictive values. Anti-DSG3 levels ≥120 RU/mL at 6 months had a sensitivity of 73.3%, and combining it with incomplete B-cell depletion gave a positive predictive value of 64.7% and negative predictive value of 87.5% for early relapse. Mignard et al. reported a similar anti-DSG3 cut-off (≥130 RU/mL) as ours, but at 3 months, as a predictor of early relapse.4 Our finding of 6 months as the time-point for anti-DSG3 cut-off predicting relapse aligns with the results of Tovanabutra et al., who reported that patients with elevated anti-DSG3 titres (≥20 RU/mL) at 6–9 months post-rituximab relapsed earlier than those without (11.3 vs. 34.2 months).15 However, our results are in contrast to those of Mignard et al., who found baseline PDAI score ≥45 and/or persistent anti-DSG1 antibody levels ≥20 IU/mL and/or anti-DSG3 antibody levels ≥130 IU/mL at 3 months to give a positive predictive value of 50% and negative predictive value of 94% for early relapse.4 This may be due to genetic differences between Indian and Caucasian pemphigus patients.16,17 Further, almost all our patients received rituximab as a second-line treatment, while some differences may be due to the use of a rituximab biosimilar in our study. Our findings underscore the need for caution when generalising findings from previous studies to different populations.

Despite a rising trend of CD19+ B-cells preceding relapse, we did not find B-cell repopulation to be a good predictor of relapse. Notably, 22% of early relapses happened before B-cell repopulation. Pemphigus relapse preceding B-cell repopulation has been rarely reported.18,19 These results suggests that B-cell repopulation (>1%) is not essential for relapse and that even smaller changes in B-cell levels may be clinically meaningful. In our study, 38% (n=14/44) patients experienced B-cell repopulation by 6 months. Though in contrast with the findings of Mignard et al., where the frequency of B-cells was still close to 0 in all patients at 6 months,4 our results are similar to other previous studies showing 23-44% patients with B-cell repopulation by this time point.5,10 The delayed B-cell repopulation in the study by Mignard et al. may be explained by their use of rituximab as a first-line treatment, as opposed to this and previous studies.4 The mean time to B-cell repopulation and mean time to relapse following repopulation in our study were 8.1 and 4.4 months, respectively, comparable to previous results (8.2-8.6 and 3.3-7.9 months, respectively).5,10

Limitations

Being a single-center study, our sample size was relatively small. Our study population included a mix of patients with pemphigus vulgaris and foliaceus, as well as treatment-naïve and relapsed patients. Further, patients were treated with oral steroid adjuvants variably. This heterogeneity may act as a potential confounder, but also mirrors real-world settings. The inclusion of patients already on treatment or with relapse may have led to falsely low baseline pre-rituximab biomarker levels.

Though a few previous studies have evaluated immunological biomarkers as predictors of pemphigus relapse, the strengths of our study are its prospective design, a relatively long follow-up period, sequential assessment of multiple biomarkers, correlation of pemphigus relapse timings (<1 year vs. between 1–2 years vs. no relapse till 2 years) with evolution of immunological markers and estimation of predictive cut-offs.

Conclusion

Our study shows that the differing patterns of immunological biomarker evolution determine the timing of pemphigus relapse. Incomplete B-cell depletion and rise in anti-DSG3 levels emerged as better predictors of relapse than other biomarkers. The presence of at least 1 of the 2 factors (incomplete B-cell depletion at 3 months and/or anti-DSG3 ≥120 RU/mL at 6 months) can help in identifying patients at risk of early relapse, who may benefit from closer monitoring or a maintenance rituximab infusion at 6 months.