A Comparative Effectiveness Research of Azathioprine and Cyclophosphamide on the Clinical and Serological Response in Pemphigus Vulgaris.

CONTEXT: A prospective study was carried out to examine the efficacy of cyclophosphamide and azathioprine in pemphigus vulgaris. AIMS: To compare the clinical and serological effect of azathioprine and cyclophosphamide in pemphigus patients.
MATERIALS AND METHODS: Prospective, institutional based study was conducted twenty-one patients of pemphigus vulgaris were initiated on either azathioprine (n = 9) or cyclophosphamide (n = 7) in addition to prednisolone and were evaluated clinically (mucosal and cutaneous severity) and serologically enzyme-linked immunosorbent assay (ELISA) at 0, 3 and 6 months.
RESULTS: Azathioprine had a slower onset of action with a statistically significant improvement seen by 6 months (P = 0.016). Cyclophosphamide had a faster onset of action (3 months) though there was no statistical difference in the efficacy between the two at the end of 6 months. The (RonT) was 33.3-44.4% for azathioprine and 28.8-42.9% for cyclophosphamide at 6 months. Though ELISA had a high sensitivity and specificity for diagnosis, as a tool for assessing therapeutic response a significant decrease was seen only till 3 months. This was restricted to Dsg1 for the azathioprine group and both Dsg3 and Dsg1 levels for the cyclophosphamide group. There were two deaths, both in the cyclophosphamide group.
CONCLUSIONS: Azathiorpine and cyclophosphamide are equally effective for mucosal and cutaneous disease in pemphigus after 6 months of therapy. Dsg ELISA is useful for diagnosis of pemphigus but is not a useful tool for monitoring response to therapy.

What was known?

Steroids are an essential part of pemphigus therapy

Immunosuppressants are required in most cases to reduce side effects of steroids and achieve remissions

Very few studies have compared the onset of action and efficacy in terms of remission on minimal therapy of azathioprine and cyclophosphamide in pemphigus

The use of enzyme-linked immunosorbent assay to monitor therapy has not been consistently found to be useful.

Introduction

Pemphigus vulgaris remains an eminently treatable condition, though the choice of the adjuvant immunosuppressant agent (ISA) is variable.[123] Few studies have focused on the efficacy of drugs on the mucosal manifestations of the disease.[4] Though there is evidence to show that azathioprine may be an effective and safe drug, other drugs have been used, such as mycophenolate mofetil and cyclophosphamide.[56789101112] Very few comparative trials have been done between these drugs[56789] and thus the clinicians choice is based more on personal choices than concrete evidence.

The primary aim was to compare the onset of action and clinical remission (mucosal/cutaneous) of azathioprine or cyclophosphamide as adjuvants in pemphigus vulgaris patients. The secondary aim was to assess the effect of these adjuvant drugs on the serological response.

Subjects and Methods

This prospective study was carried out in a tertiary referral hospital after due Institutional Review Board approval. As this work was part of the fulfillment of a postgraduate thesis of the State University, Ethical approval was taken at the institution, department, and the university.

Participants

A total of 30 patients of pemphigus between the age group 13–69 years attending the dermatology outpatient department were included, with the following inclusion criteria: (i) Presence of mucosal erosions and⁄or superficial cutaneous blisters with histological confirmation and DIF; (ii) serial anti-Dsg1 and anti-Dsg3 antibody levels; (iii) patients had to ensure a minimal follow-up at 3 and 6 months after the initiation of therapy. The exclusion criterion were: (i) Pregnancy or lactation, (ii) the presence of liver diseases, chronic renal failure, chronic heart failure or an ischemic heart disease (iii) any contraindications to cytotoxic drugs.

Apart from new cases, 20 controls were taken to validate the sensitivity of the enzyme-linked immunosorbent assay (ELISA). The ELISA used was the EUROIMMUN kit (EUROIMMUN, Lubeck, Germany) with a cut-off value of Dsg1 and Dsg3 >20 IU index value. The test was performed at baseline, 3 months and 6 months. After diagnosing the cases clinically, the disease activity of skin lesions and mucosal were noted as per the classification given in Table 1.[13]

Table 1

Grading of clinical severity

Interventions

To evaluate the effect of therapy, complete remission on therapy (RonT) was the end point, defined as the absence of new or established lesions while the patient was receiving minimal therapy. Minimal therapy was defined as prednisone (or the equivalent) at ≤10 mg/day and/or minimal adjuvant therapy for at least 2 months.[14]

All patients were started on prednisone (1–1.5 mg/kg/day) and the adjuvant ISA (azathioprine/cyclophosphamide) from the onset. This was to ensure that the effect of the individual ISA could be assessed on the clinical and serological parameters. After control of disease activity,[14] the maintenance therapy stage was initiated and tapering of steroid dose was initiated. The dose reduction was by 10–20 mg/month, depending on the severity of disease till a dose of 20 mg a day. Subsequently, it was reduced to 20 mg alternate day and reduced by 10 mg till a dose of 5 mg was achieved. The goal was to achieve a maintenance dose of 0.2 mg/kg/day or below which was maintained for a further 3 months. At this dose if the disease was well controlled for a month a serum cortisol level was done at 8 a.m. and if the level was >10 mg/dl the steroid was stopped. In case of a breakthrough in the disease the existing prednisolone dose was doubled and after control was tapered down as per the above protocol.

The use of adjuvant was based largely on the reproductive status of the patient, thus patients who had not completed their family or were unmarried were given azathioprine 1–1.5 mg/kg/day while the rest were given cyclophosphamide in a dose of 1–1.5 mg/kg/day. This dose was continued till the patient was off steroids and further maintained for a further 6 months. The monitoring of the adjuvant drugs was based on standard protocols.[2]

Follow-up

All patients were examined weekly in the 1st month and then every 2 weeks for 4 months and then monthly thereafter. Clinical assessment for severity and ELISA were performed at baseline, 3 months and 6 months.

Statistical analysis

The statistical analysis was carried out using Statistical Package for Social Sciences (SPSS Inc., Chicago, IL, USA version 17.0 for Windows). Normality of data was checked by measures of skewness. For normally distributed data, means were compared using paired t-test and ANCOVA for two groups, and one-way analysis of variance was used for more than two groups. For skewed data Mann–Whitney test was applied. Spearman rank correlation coefficient and Pearson Chi-square tests were applied wherever needed. All statistical tests were two-sided and performed at a significance level of P ≤ 0.05.

Results

Patients

Of the total of 30 clinically diagnosed cases of pemphigus vulgaris, 21 patients [Figure 1] met the inclusion criterion (intention-to-treat). The age varied from 13 to 68 years with a mean age of 40.13 ± 13 years, the majority of patients belonged to the age group of 31–40 years (33.33%), followed by 51–60 and 21–30 years age group (20.00%). A summary of the baseline profile is given in Table 2a.

Figure 1

Flow Diagram of the patients during the course of the study

Table 2a

A comparison of the two treatment groups*

Disease characteristics

In the majority of patients (46%) lesions started from oral mucosa, followed by the trunk (42%). Of the 21 patients, 16 had both cutaneous and mucosal involvement (76%), 2 patients (9.5%) had only cutaneous involvement while three patients (14.3%) had only mucosal involvement. The oral mucosa was most commonly involved (54.2%), followed by involvement of both oral and genital mucosa (20.8%). Involvement of all three mucosae was seen in three patients (14.3%).

We classified the patients as a mild, moderate or severe disease based the criteria given by Belloni-Fortina et al.[13] [Table 1]. The majority of patients had severe cutaneous involvement (60.9%) followed by mild (30.4%) and moderate involvement (8.7%). A similar scoring was used for mucosal severity and again majority had severe involvement (54.2%) followed by the mild involvement (25.0%) and moderate involvement (20.8%).

Enzyme-linked immunosorbent assay values against Dsg1 and Dsg3

Of the blood samples evaluated, receiver operating characteristic curves were plotted to get the values for anti-Dsg3 and anti-Dsg1. The area under the curve for anti-Dsg1 was 0.484 and for Dsg3 it was 0.923. At values >32.30 U/ml, the Dsg3 ELISA had 85.1% sensitivity and 100% specificity. The positive and negative predictive values calculated were 100% and 37.5% respectively with a diagnostic accuracy of 86.3%.

A statistically significant difference [Figure 2] was seen in the anti-Dsg3 levels between 1st and 2nd visit (P = 0.000), 1st and 3rd visit (P = 0.003) and 2nd and 3rd visit (P = 0.002). The anti-Dsg1 levels showed a significant difference only between 1st and 2nd visit (P = 0.000). There was no statistical difference between the baseline Dsg values between the two treatment groups [Figure 3].

Figure 2

A depiction of Dsg1 and 3 levels at different periods of evaluation. A significant difference was observed in the Dsg3 levels between different time periods but Dsg1 levels were significantly reduced only between the first and second visit

Figure 3

A comparison of Dsg enzyme-linked immunosorbent assay levels between the two treatment groups (A = Azathioprine, B = cyclophosphamide). Mann–Whitney test did not reveal any difference between the two treatment groups at baseline (P = 0.205 and 0.791; Dsg3/Dsg1) or at 6 months (P = 0.085, 0.263; Dsg3/Dsg1)

Effect of immunosuppressant agent on clinical severity

A total of nine patients received oral steroids with azathioprine, and seven received oral steroids with cyclophosphamide [Figure 1]. There was a progressive decrease in the cutaneous and mucosal severity in the azathioprine group [Figure 4] and the RonT at the end of 6 months was 44.4% and 33.3% for cutaneous and mucosal disease [Table 2b]. In the azathioprine group [Figure 4], significant improvement was seen between both cutaneous and mucosal severity between the 1st and 3rd visit (P = 0.016 in both).

Figure 4

A depiction of the effect of azathioprine and prednisolone on the clinical severity score of the disease *Wilcoxon signed rank test shows significant improvement between both cutaneous and mucosal severity of the 1st and 3rd visit (P = 0.016). C = cutaneous severity; M = mucosal severity, RonT: remission on therapy

Table 2b

Comparison of cutaneous and mucosal severity scores between the treatment groups

Patients on cyclophosphamide also had a progressive decrease in severity [Figure 5] and the RonT was 42.9% and 28% for the cutaneous and mucosal disease at the end of 6 months [Table 2b]. In the steroid and cyclophosphamide group [Figure 5], significant improvement was seen in cutaneous severity between the 1st and 3rd visit (P = 0.015) and 2nd and 3rd visit (P = 0.025) and in mucosal severity between 1st and 2nd (P = 0.038) and 1st and 3rd visits (P = 0.041). There was no difference in the severity at the third visit between the two groups (P = 0.982; P = 0.456) [Table 2b].

Figure 5

A depiction of the effect of cyclophosphamide plus prednisolone on disease severity. *Wilcoxon signed rank test shows significant improvement between cutaneous severity of 1st and 3rd visit (P = 0.015) and 2nd and 3rd visit (P = 0.025) and in mucosal severity between 1st and 2nd (P = 0.038) and 1st and 3rd visits (P = 0.041). C = cutaneous severity; M = mucosal severity RonT: remission on therapy

Effect of immunosuppressant agent on enzyme-linked immunosorbent assay values

In the azathioprine group [Table 3] a significant change was seen only in the Dsg1 levels between the 1st and 2nd visit (P = 0.001), while in the cyclophosphamide group there was a significant change in both the Dsg3 and Dsg1 levels between the 1st and 2nd visit (P = 0.001). There was no difference in the Dsg3 and Dsg1 values at the third visit [Figure 3]. There were seven patients who had clinical remission in the form of the absence of either cutaneous or mucosal lesions but still their Dsg levels were raised. Moreover, patients with cutaneous remission showed raised Dsg1 while those in mucosal remission showed raised Dsg3.

Table 3

Correlation of Dsg ELISA levels at different visit in treatment groups

Side effects

There was no major toxicity seen in either of the treatment group at the end of the study though two patients died in the cyclophosphamide group. The mortality, in the 1st month, was due to septic mycocarditis and cardiac arrest respectively.

Discussion

Our study revealed that the severity of the disease decreased during the study period [Figures 4 and 5] with a remission on therapy (RonT) of 42.9% to 44.4% for cutaneous disease and 28.8% to 33.3% for mucosal disease at 6 months [Table 2b]. Also, a statistically significant decrease in the Dsg3 levels was seen between the first (100.8 U/ml), second (82.9 U/ml) and third (70.3 U/ml) visits with the anti-Dsg1 levels showing a significant change only between first (74.1 U/ml) and third visit (60.6 U/ml) [Figure 2]. The group on azathioprine showed a significant clinical improvement between 1st and 3rd visit [Figure 4] while the cyclophosphamide group showed an improvement throughout the study period [Figure 5] with a faster onset of action (after 3 months). The serological improvement was not uniform and peaked at 3 months in both the treatment groups [Table 3]. Significantly, at the third visit there was no appreciable difference between the two ISA either in the serological values of Dsg1 (P = 0.263) or anti-Dsg3 (P = 0.085) ELISA and the cutaneous (P = 0.982) or mucosal severity score (P = 0.456).

Pemphigus patients require steroids with concomitant adjuvant therapy[1234] and while comparisons between ISA have been performed [Table 4],[56789] our focus was to compare azathioprine and cyclophosphamide. To achieve this goal the ISA were initiated at the outset with prednisolone. Azathioprine is one of the most frequently used adjuvants in pemphigus with strong evidence from multiple case series[123] while cyclophosphamide is reserved for severe recalcitrant case, due to the gonadotoxic side effects, secondary infections, and carcinomas.[123] Our study revealed that though cyclophosphamide had an earlier onset of action (3 months) than azathioprine, [Figures 4 and 5] at the end of the study [Table 2b] there was no difference in the clinical severity between the two drugs. Furthermore, there was no significant difference in the mean steroid dose between the two ISA or the time required for initial control of disease [Table 2a]. Though the study was not randomized, there was a better response in patients on azathioprine in terms of RonT for mucosal and cutaneous disease as compared to cyclophosphamide [Table 2], which is analogous to previous studies.[56] This is as there is a delay in healing of mucosal disease on cyclophosphamide.[1011] Chams-Davatchi et al.[5] in their work noted that the most efficacious cytotoxic drug was azathioprine, followed by cyclophosphamide, and mycophenolate mofetil[5] a finding that was also noted by Akhtar and Hasan.[6] Olszewska et al.[7] though found cyclophosphamide to be superior to azathioprine.

Table 4

Summary of previous studies on various immunosupressants for pemphigus

In pemphigus consequent to antigenic stimulation, a T helper 2 cell response is initiated[1516] wherein the released cytokines stimulate B cells that release antibodies and cause acantholysis. Though alternative pathogenetic models exist,[1617] this model is a reasonable hypothesis to understand the role of ISA. Azathioprine is a cell-specific drug[1819] and affects the T helper and CD4 naïve cells[1819] and affects both the cell-mediated and humoral immunity.[181920] Cyclophosphamide suppresses all the proliferating lymphoid cells and alkylates some resting cells.[1819] Thus, it causes a decrease in T and B lymphocytes, decreases antibody production and suppresses delayed hypersensitivity to new antigens.[1819] This cell cycle nonspecific action causes suppression of cells that are not directly involved in pemphigus and makes it a toxic and an inappropriate choice for all cases of pemphigus. Thus the use of cyclophosphamide as an adjuvant in the treatment of pemphigus vulgaris remains controversial and the use of the drug as a first-line adjuvant in pemphigus vulgaris is not supported by evidence of statistically significant benefit over azathioprine [Table 4].

Only two studies[412] have studied the effect of therapy on mucosal disease exclusively, and no study has compared the effect of cyclophoshamide and azathioprine on the mucosal and cutaneous disease severity of pemphigus. In one study, 94.6% of cases achieved clinical remission, with 56.8% who were on minimal therapy (RonT).[4] This is similar to our results where 31.5–43.8% of cases were in remission (RonT) at 6 months. There was no significant difference in terms of duration of treatment, severity of disease and attainment of clinical remission with the various adjuvants.[4] The other study compared cyclosporine with cylophosphamide and found the two to be equally effective.[12]

The sensitivity and specificity of ELISA for diagnosing PV was 96.1% and 85% (Dsg3) and 92.3% for Dsg1.[17] Though there was a decrease in the levels of Dsg3 throughout the study, [Figure 2] the same decrease was not seen in Dsg1 levels. Moreover, there was no correlation between anti-Dsg3 and anti-Dsg1 and cutaneous or mucosal severity.[17] The mean values after 6 months of Dsg1 ELISA ranged from 57.5 to 88.9 U/ml while for Dsg3 ranged from 59.0 to 125.3 U/ml. Even with these high levels, 43.8% of patents with cutaneous disease and 31.5% of cases with mucosal disease were in remission (RonT). This reiterates the previous observation about the lack of correlation of Dsg ELISA levels with disease severity.[132122232425] The discordance between the clinical phenotype of pemphigus and the serology ranges from 33% to 46%.[17] Moreover, about 39% of mucosal pemphigus cases have Dsg1 antibodies while 49% of cases with cutaneous pemphigus have Dsg3 antibodies.[17] Thus, while Dsg ELISA may be useful to diagnose pemphigus, it does not have a sound predictive ability to either determine clinical phenotype, mucosal or cutaneous nor monitor therapy.[13172122232425]

The 2 deaths in the cyclophosphamide group are a cause of concern and similar reports[10] focus on the cardiotoxicity of the drug which is a serious issue with the use of cyclophosphamide, specially the pulse method of administration. Studies on cyclophosphamide have found secondary infection, gonadal toxicity, reactivation of tuberculosis and cardiac arrest amongst other complications.[1011] Also, patients may develop transitional cell carcinoma after a period of 15 years,[11] when they may not present to dermatologists, and this is an often overlooked and underreported complication. Thus we cannot justify its use in pemphigus as a first-line immunosuppressant.

There are drawbacks in our study, including the lack of proper randomization. It is difficult to randomize ISA in pemphigus as there can be no justification to administer cyclophosphamide in patients who are unmarried or have not completed their family. In spite of this, our baseline comparison between the two groups revealed no difference [Table 2a]. There are other issue including the small sample size, short treatment course (6 months) and lack of long term follow-up. Thus, we are unable to comment on the remission off therapy or relapse rates. But, the remission on minimal therapy shows that there is no difference in the efficacy of the two drugs on the mucosal and cutaneous manifestations of pemphigus vulgaris.

It is important at this juncture to emphasize the difference between a randomized controlled trial (RCT) and comparative effectiveness research (CER) and the advantages of the latter as this was our study model.[26] RCTs have long been considered the gold standard of medical intervention studies.[26] However, there is increasing recognition that the “average” effect that is measured in RCTs does not necessarily apply to all participants within a trial, nor is it reflective of how an intervention performs in the “real world.”[27] The strengths of an RCT is its strong internal validity, but they have generally had poor external validity (i.e., are not generalizable to a larger patient group). This is as RCTs have poor generalizability, with short study periods (typically on the scale of weeks to months), specialized study populations with strict inclusion and exclusion criteria, comparison with placebo as opposed to a clinically relevant active comparator, and measurement of outcomes that may not be important to patients or relevant to clinical practice.[27] Results from RCTs, therefore, measure efficacy or how an intervention performs in an ideal, carefully controlled setting rather than the real-world clinical setting. Explanatory trials are designed to determine the effects of an intervention under ideal circumstances and are most relevant to answering a specific research question about the benefit of medical intervention. Pragmatic trials, like ours, on the other hand, are designed to determine the effects of an intervention under the conditions in which it will be applied in the clinical setting and are, thus, the design of choice for CER.[27]

Our results have shown that though both azathioprine and cyclophosphamide had similar clinical and serological outcomes, the former should be preferred as it is markedly safer than cyclophophosphamide. A longer follow-up and assessment of relapse rates may be able to answer the question of longevity of response.

What is new?

Though cyclophosphamide has a faster onset of action (3 months), the results at the end of 6 months were similar to azathioprine

The remission on therapy was better for azathioprine though there was no statistical difference between the two drugs

Enzyme-linked immunosorbent assay is a good tool for diagnosis of pemphigus vulgaris but may not be a good tool for monitoring disease response.