Mucous Membrane Pemphigoid, Bullous Pemphigoid, and Anti-programmed Death-1/ Programmed Death-Ligand 1: A Case Report of an Elderly Woman With Mucous Membrane Pemphigoid Developing After Pembrolizumab Therapy for Metastatic Melanoma and Review of the Literature

An 83-year-old patient developed erosions and a blister of the gingival mucous membrane, 6 months after discontinuation of the anti-programmed death-1 (anti PD-1) pembrolizumab therapy administered for 10 months for a metastatic melanoma. A diagnosis of mild mucous membrane pemphigoid (MMP) was made. Complete remission of MMP was rapidly obtained with minimal therapy (doxycycline). MMP remained in complete remission after a 3-month follow-up since discontinuation of the doxycycline therapy and no evidence of relapse of the melanoma was observed after a 14-month follow-up since discontinuation of the pembrolizumab therapy. The widespread use of anti PD-1 and anti-programmed death-ligand-1 (PD-L1) in several malignancies reveals new adverse events. MMP describes a group of chronic, inflammatory, mucous membrane-predominant, subepithelial auto-immune blistering diseases. It is clinically distinct from bullous pemphigoid another autoimmune blistering disease but shares some immunological similarities with it. Twenty-nine cases of bullous pemphigoid associated with anti PD-1/PD-L1 have been reported in the literature and one of MMP. Here, we described the case of a MMP developed after pembrolizumab and discussed the accountability of anti PD-1/PD-L1 in our case and the previous reported bullous pemphigoid and MMP cases using the Begaud system scoring.


BACKGROUND
Immune checkpoint inhibitors against programmed death-1 (anti PD-1) and programmed death-ligand 1 (anti-PD-L1) agents have revolutionized the treatment of metastatic melanoma and have shown encouraging promise in advanced solid tumors and hematological malignancies. However, these agents are associated with immune-related adverse events (IrAEs) that affect mainly the skin, hormone glands, liver and gastrointestinal tracts.
Indeed, up to 20% of treated patients may develop dermatological IrAEs. They are predominantly non-specific rashes and pruritus (1). Toxicities on buccal mucous membrane (MM) have also been described, including xerostomia, lichenoid reactions, and dysgeusia (2). Since 2015, an association between a treatment with anti PD-1/PD-L1 and bullous pemphigoid (BP) has been reported in 29 cases. One mucous membrane pemphigoid (MMP) case, an autoimmune bullous disease (AIBD) similar to BP, has also been described after pembrolizumab therapy.
Here, we report a second case of MMP that occurred 16 months after initiation of pembrolizumab therapy for a metastatic melanoma, discuss the association between MMP and melanoma, and review the literature on BP and MMP associated with anti PD-1/PD-L1.

CASE PRESENTATION
In 2014, an 83-year-old woman with no history of known autoimmune disease was diagnosed as having a right leg superficial spreading melanoma, initially T2b N0 M0. Eight months later, she developed iterative local and in transit cutaneous metastases on the same leg and she underwent four times surgical excision.
In 2016, at the fourth recurrence, surgery was not chosen. Baseline full-body computed tomography revealed no other metastasis (T2 N0 M1a). Mutation tested on a tumor sample excluded the presence of any BRAF mutation. Administration of pembrolizumab therapy was started at 2 mg/kg every 3 weeks, resulting in complete remission (CR) within 3 months (cycle 4). In March 2017, after 14 cycles, she remained in CR, and the pembrolizumab therapy was stopped at her request.
Doxycycline therapy (100 mg/day) and mouth washes with corticosteroid (betamethasone 2 mg) three times daily were initiated, which led to the control of the MMP within 2 weeks and CR under this minimal therapy in 6 weeks. After 3 months of treatment with doxycycline therapy, the patient decided on her own to discontinue it, and no MMP relapse had

DISCUSSION
Our case raises the question of an association between MMP and melanoma, or between MMP/BP and pembrolizumab administration.
MMP encompasses a group of AIBDs clinically defined by the predominance of MM lesions over skin lesions (3,4), and healing of its lesions leads to characteristic cicatricial scarring. The buccal involvement is the most frequent, followed in order of decreasing frequency by ocular, nasal, nasopharyngeal, anogenital, skin, laryngeal, and oesophageal involvements. The ocular, laryngeal, and oesophageal involvements can cause severe impairment or even death.
Our patient had a typical MMP, except the age at MMP onset, in a mild form because of the purely buccal involvement. Thus she was considered as a "low-risk patient" with few tendencies of scarring and required minimal therapy with doxycycline and topical steroids (3). Her MMP was controlled in 2 weeks, in CR on minimal therapy in 6 weeks, and in CR off treatment in 3 months. She developed mild cicatricial lesions of her gingival MM and did not relapse during follow-up. Rapid clinical improvement after only a short course of treatment is unusual in MMP.
MMP results from the activity of autoantibodies directed against BMZ antigens. The main autoantibody target is BP180, with the sera of most MMP patients reacting with its C-terminal domain (BP180-C term), combined or not with reactivity against the BP180-NC16A epitope and BP230 (5,6). Other target antigens associated with a clinical MMP phenotype have been characterized molecularly, including the following: laminin 332, both α6β4-integrin subunits, and type VII collagen (7), respectively defining laminin 332-MMP, α6β4-integrin MMP, and MM epidermolysis bullosa acquisita. Autoreactive T lymphocytes are thought to also play a key role in the pathogenesis of MMP, particularly in the fibrosing process (8)(9)(10)(11)(12)(13)(14)(15).
Our patient had a linear deposition of IgG/C3 along the epithelial BMZ on DIF microscopy and on the lamina densa on direct IEM, a location consistent with targeting of the C-terminal extremity of BP180 or laminin 332 by autoantibodies (16,17). No circulating antibodies against BMZ antigens were detected by serological studies, notably BP180-NC16A ELISA, as in 49% of MMP in a recent series (18).
An association between laminin 332-MMP and malignancy was first reported in 1998 but is currently controversial. On one hand, 21 cases of laminin 332-MMP have been reported in association with cancer, including 15 reviewed by Sadler in 2007 (19) and six case reports after 2007 (20)(21)(22)(23)(24)(25). An increased risk of solid cancers as compared with the general population was reported by two authors (26,27), higher in the first year following the laminin 332-MMP diagnosis. On the other hand, no significant correlation was found between laminin 332 reactivity and the proportion of patients with an associated internal cancer in three recent serological studies of MMP (28)(29)(30).
Anyway, no association between laminin 332-MMP and melanoma has been reported (31).
In our patient, who had a possible laminin 332-MMP, a link between MMP and melanoma seems unlikely, as the first incidence occurred 3 years after the second and the latter was in CR.
Although the clinical characteristics of MMP differ from those of BP typified by the absence of MM lesions, absence of predominant head-and-neck involvement, absence of scars, and older age at onset (>70 years) (53), MMP and BP share physiopathological features; BP result from the activity of autoantibodies directed against BP230 and BP180, such as most MMPs. However, the sera of most patients with BP react with the BP180-NC16A epitope, contrary to those of patients with MMP.
The melanoma treatment has been revolutionized by innovative immunomodulation drugs that break tolerance. The main treatment targets for melanoma are cytotoxic Tlymphocyte antigen-4 and PD-1. Ipilimumab, a monoclonal antibody against cytotoxic T-lymphocyte antigen-4, was the first drug to demonstrate a benefit in overall survival in a randomized controlled phase 3 study of patients with advanced melanoma (54) and to be approved by the Food and Drug Administration (FDA) and European Medicines Agency (EMA). The anti-PD-1s and anti-PD-L1s are remarkably more effective in terms of response and overall survival rates than ipilimumab. The response rate with anti-PD-1 reaches 40% in melanoma, and the 5-year overall survival rate for naive patients is close to 40% (55). Moreover, anti-PD-1s are less toxic in terms of IrAEs. In 2014, the FDA approved the anti-PD-1 antibodies pembrolizumab and nivolumab for advanced melanoma and, in 2015, the combination of ipilimumab and nivolumab.
PD-1 is a negative co-stimulatory receptor, which downregulates excessive immune responses by binding to its ligands, PD-L1 and PD-L2. This receptor is expressed mainly on activated T cells. In tumor tissue, binding of PD-1 inhibits effector T-cell function, which leads to exhausted T cells and suppression of the antitumor immune response (56). Pembrolizumab and nivolumab on one hand and atezolizumab and durvalumab on the other hand are selective humanized monoclonal antibodies that bind respectively to PD-1 and PD-L1 and thus block the interaction between PD-1 and its ligands, which leads to stimulatory effects on T cells.
B cells also express PD-1, and inhibition of PD-1 expression can directly activate B cells in a T-cell-independent manner (57). Lastly, it appears that anti PD-1 reduce regulatory T-cell activity (58), resulting in decreased tolerance and development of autoimmunity. All these mechanisms can be involved in the induction of MMP as soon as B and T cells played a central role in the MMP pathogenesis.
The potential drug induction of AIBD has been known for decades (18,(59)(60)(61)(62)(63)(64)(65)(66). Recently, between 2015 and 2018, 18 case reports or small series described 27 patients who developed BP while receiving anti PD-1/PD-L1 therapy, including three with negative DIF (35,37) or without DIF confirmation (47)  Six of the 27 patients received ipilimumab therapy before anti PD-1, one received an association of ipilimumab and nivolumab therapy, and one received ipilimumab therapy after treatment with nivolumab. Two additional BP cases within a large series were simply mentioned, without any clinical description (50,51). Recently, one MMP case treated with pembrolizumab therapy for a metastatic Merkel cell carcinoma has also been reported (53). Notably, the outcomes of the patients with melanoma were better [only 5 (33%) out of 15 cases in which the information were available, had a progressive disease] than reported in general anti-PD-1 treatment of melanoma (55). Other studies reported that objective response rate was significantly higher in patients with melanoma who experienced nivolumab-related adverse events (67).
The overall characteristics of these patients with anti PD-1/PD-L1-associated BPs were as follows: eight women and 19 men (female-to-male sex ratio, 0.4), median age of 68 years (35-90 years) at the time of BP diagnosis, and a median interval of 24  weeks between starting anti PD-1/PD-L1 therapy (challenge) and BP onset. This time was significantly shorter (median, 16 weeks; range, 6-80 weeks) with nivolumab than with pembrolizumab (median, 27 weeks; range, 16-84 weeks) (Wilcoxon rank sum test with continuity correction: p = 0.023).
Pruritus was a prominent feature of most cases (68). Administration of anti PD-1/PD-L1 agents was discontinued (dechallenge) in 21 of the 27 patients (patients 1-14, 16-20, 22, 23, and 27) because of the evolution of BP and/or cancer and continued in five patients (no dechallenge, patients 1, 15, and [24][25][26]. This information is unknown in one patient (patient 21). BP treatment with systemic steroids was required in all but four patients (patients 3, 5, 14, and 22). AntiPD-1/PD-L1 was reintroduced in two patients (rechallenge; patients 7 and 13). The reasons of the dechallenge, and the outcome after dechallenge or rechallenge are detailed in Table 2.
The patient with pembrolizumab-associated MMP reported by Haug et al. was a 62-year-old man who developed a pure buccal MMP after 13 weeks of pembrolizumab therapy. He had circulating autoantibodies targeting the C-terminal extremity of BP180. Pembrolizumab was discontinued at the time of MMP onset, and he was successfully treated with doxycycline and topical steroid, as in our case.
Lastly, we assessed the intrinsic accountability score of anti PD-1/PD-L1 in AIBD induction for our patient with MMP, and a posteriori for the 27 BPs and MMP cases that have been reported using the Begaud scoring system (terms in bold type), updated in 2011 (72). The present challenge was the treatment of a malignancy by using an anti PD-1/PD-L1. Patients with a malignancy who started treatment with anti PD-1/PD-L1 before MMP/BP onset may have a suggestive or compatible challenge. The dechallenge was the discontinuation of the treatment. Outcome after dechallenge or no dechallenge may be suggestive (if BP is controlled with dechallenge or worsened without dechallenge), conversely non-suggestive (if BP worsened after dechallenge or controlled unless without dechallenge), and inconclusive (without details on BP evolution or continued treatment). The rechallenge was the reintroduction of the treatment. It may be positive (R+) or negative (R-) or not done (R0). The chronological scoring (combining status of challenge, dechallenge, and rechallenge) may be C1, doubtful; C2, plausible; and C3, likely. The symptomatological scoring may be S1, doubtful; S2, plausible; and S3, likely. Lastly, the intrinsic accountability scoring [combining chronological (C) and symptomatological (S) scores] may be I1 (C1S1), I2 (C1S2 or C2S1), I3 (C2S2), I4 (C1S3 or C3S1), I5 (C2S3 or C3S2), or I6 (C3S3) (see detailed results in Supplementary Data, Table S1).
The Begaud system scoring indicates that the intrinsic accountability score was I4 for the MMP case that was reported but I1 for our patient with MMP.
In all the MMP/BP patients, the low score was essentially due to the long time between anti PD-1 introduction and AIBD onset (compatible challenge) and/or a non-suggestive dechallenge. Indeed, it could be the consequence of long delay of action of anti PD-1/PD-L1. A non-suggestive dechallenge may not be an argument against the accountability of a long-acting drug, and at the end, the Begaud scoring system may not be suitable for assessing the accountability of drugs with prolonged therapeutic effect.
In our patient, the eventuality of a rechallenge did not occur because the melanoma and MMP remained in CR. As an anti PD-1-induced MMP is possible, a rechallenge would theoretically expose her to a risk of MMP relapse in a potentially more serious form. Indeed, MMP can involve ocular, nasopharyngeal, laryngeal, esophageal, genital, or anal MM, sites that have a high likelihood of scarring, which is associated with loss of function. On the other hand and contrary to literatures on adverse drug reactions, a negative rechallenge with anti PD-1 has already been reported (38,73).
In conclusion, we report the case of a patient who developed a mild MMP, possibly induced by anti PD-1 rather than by melanoma. We cannot also exclude that the MMP could be triggered by aging, malignancy, and pembrolizumab acting in concert or developed quite independently. MMP was rapidly controlled by a minimal treatment, raising the question of reintroduction of anti PD-1 if the melanoma relapses. With the increasing use of immunotherapies for various malignancies, clinicians should be alert for this new anti PD-1-induced IrAE, which is related to BP but potentially more severe. Long-term clinical follow-up is warranted owing to delayed adverse events, even after discontinuation of anti PD-1 inhibitors. Lastly the Begaud system scoring applied to our patient and previous reported cases with anti-PD-1/PD-L1 related BP/MMP indicates a low intrinsic accountability score in most of the patients suggesting it may not be suitable for assessing the accountability of drugs with prolonged therapeutic effect. Development of another specific assessment might be necessary.

ETHICS STATEMENT
A written consent has been obtained from the patient.

AUTHOR CONTRIBUTIONS
CZ and CP-S conceived and designed the study. CZ, MA, CL, PW, AG, and EM collected clinical data. AL and CP-S conducted the histological studies and FA and SM-G the immunological ones. CZ wrote the first draft of the manuscript. CP-S rewrote sections of the manuscript. EM and FC corrected the final version. All authors contributed to manuscript revision, and read and approved the submitted version.