Abstract
The anti-CCR4 mAb, mogamulizumab, offers therapeutic benefit to patients with adult T-cell leukemia-lymphoma (ATL), but skin-related adverse events (AE) such as erythema multiforme occur frequently. The purpose of this study was to determine the mechanisms by which mogamulizumab causes skin-related AEs in patients with ATL.
We investigated whether autoantibodies were present in patients’ sera using flow cytometry to determine binding to keratinocytes and melanocytes (n = 17), and immunofluorescence analysis of tissue sections. We analyzed the IgM heavy chain repertoire in peripheral blood mononuclear cells before and after mogamulizumab or other chemotherapy by next-generation sequencing (NGS; n = 16).
Autoantibodies recognizing human keratinocytes or melanocytes were found in the sera of 6 of 8 patients suffering from mogamulizumab-induced erythema multiforme. In one patient, complement-dependent cytotoxicity (CDC) mediated by autoantibodies against keratinocytes or melanocytes was proportionally related to the severity of the erythema multiforme. The presence of autoantibodies in the epidermis was confirmed in all biopsy specimens of mogamulizumab-induced erythema multiforme (n = 12). Furthermore, colocalization of autoantibodies and C1q, suggesting the activation of CDC, was observed in 67% (8/12). In contrast, no autoantibody or C1q was found in ATL tumor skin lesions (n = 13). Consistent with these findings, NGS demonstrated that IgM germline genes had newly emerged and expanded, resulting in IgM repertoire skewing at the time of erythema multiforme.
Mogamulizumab elicits autoantibodies playing an important role in skin-related AEs, possibly associated with regulatory T-cell depletion. This is the first report demonstrating the presence of skin-directed autoantibodies after mogamulizumab treatment.
The anti-CCR4 mAb, mogamulizumab, offers therapeutic benefit to patients with adult T-cell leukemia-lymphoma, but skin-related adverse events (AE), such as erythema multiforme, occur frequently. This study has demonstrated that mogamulizumab treatment elicits autoantibodies binding keratinocytes and melanocytes, and that these autoantibodies have an important role in the pathogenesis of mogamulizumab-induced skin-related AEs. This is possibly associated with regulatory T-cell depletion and the consequent breakdown of peripheral immunologic checkpoints regulating autoantibody production. It is important to identify the target molecules recognized by these mogamulizumab-induced autoantibodies, which will lead to the development of novel diagnostic methods. Furthermore, this study indicates that B-cell–targeting therapies, such as rituximab, should be effective for treating mogamulizumab-induced skin-related AEs.
Introduction
Adult T-cell leukemia-lymphoma (ATL) is a peripheral T-cell neoplasm caused by human T-cell lymphotropic virus type-1 (HTLV-1), which has a miserable prognosis (1–4). Because CCR4 is expressed on tumor cells from most patients with ATL (5, 6), a therapeutic anti-CCR4 mAb, mogamulizumab, has been developed to target them (7, 8). Although this antibody offers clinical benefit to patients with ATL (9–11), skin-related adverse events (AE) such as erythema multiforme are frequent, and severe AEs including Stevens–Johnson Syndrome/toxic epidermal necrolysis are occasionally observed (12). On the other hand, moderate skin-related AEs after mogamulizumab may be associated with a favorable prognosis (9, 13). The skin-related AE may be due to regulatory T (Treg) cell depletion by mogamulizumab (9, 12, 14–16), but the detailed mechanisms have not been fully established. Treatment with mogamulizumab has been extended to cutaneous T-cell lymphoma (17, 18), and is likely to be extended to other diseases in future, such as many types of solid cancer (14–16), as well as HTLV-1–associated myelopathy (19). Therefore, it is a matter of some urgency to determine the most appropriate measures against the skin-related AEs caused by mogamulizumab.
Immune dysregulation, polyendocrinopathy, enteropathy, and X-linked (IPEX) syndrome are caused by mutations of FOXP3, a master gene of Treg cells, and are characterized by various sometimes fatal autoimmune disorders including severe dermatitis (20–22). Importantly, IPEX syndrome and the immune-related AEs caused by mogamulizumab share a fundamental characteristic, namely Treg cell depletion, although the former is congenital, the latter iatrogenic. Earlier studies indicate that autoantibodies play a significant role in the complicated autoimmune phenotypes of IPEX (23–25). Furthermore, autoantibody elicitation was observed in a clinical trial of mogamulizumab (14). These important observations prompted us to investigate whether autoantibodies were elicited that attacked the skin following Treg cell depletion by mogamulizumab.
Materials and Methods
Human samples
Sera were obtained before and after treatment from 8 patients with ATL (nos. 1–8) who suffered from skin-related AEs after mogamulizumab, as well as from 5 patients (nos. 9–13) who received mogamulizumab, but did not suffer from accompanying skin-related AEs, and finally, four patients (nos. 14–17) who received chemotherapy without mogamulizumab. Peripheral blood mononuclear cells (PBMC) were also obtained before and after treatment from 7 patients with ATL (nos. 2–6, 8, and an additional patient no. 18) who suffered from skin-related AEs after mogamulizumab. PBMCs were also obtained from 5 patients (nos. 10–13, and an additional no. 19) who received mogamulizumab, but did not suffer from accompanying skin-related AEs, and finally 4 patients (nos. 14–17) who received chemotherapy for ATL but no mogamulizumab. Biopsy specimens were obtained from mogamulizumab-induced skin lesions of 12 patients with ATL (nos. 3–5, 18, and additional nos. 20–27). All skin-related AEs observed in these 17 patients with ATL (nos. 1–8, 18, 20–27) after mogamulizumab were diagnosed as erythema multiforme according to the National Cancer Institute Common Terminology Criteria for AEs version 4.0. Skin biopsy specimens from ATL tumor lesions were obtained from 13 patients (nos. 3, 9, 14, 19, 20, and additional patients nos. 28–35) before mogamulizumab treatment. The characteristics of the patients with ATL enrolled in this study are shown in Supplementary Table S1. Sera were obtained from a patient with B-cell lymphoma who developed dermatitis after allogeneic hematopoietic stem cell transplantation. This patient had an anti-Bullous Pemphigoid 180 (BP180) antibody (as determined by SRL, Inc.) after the onset of dermatitis. All donors provided written informed consent prior to sampling according to the Declaration of Helsinki. This study was approved by the institutional ethics committees of the Nagoya City University Graduate School of Medical Sciences and Kyowa Hakko Kirin Co., Ltd.
Flow cytometry
Sera were diluted 1:20 in washing buffer (FBS in PBS), and incubated with normal human epidermal keratinocytes (D10006; Takara Bio Inc.) or normal human epidermal melanocytes (D10069; Takara Bio Inc.) at 4°C for 30 minutes. The patient's serum before treatment (with mogamulizumab or chemotherapy) was set as the primary baseline control antibody, and the same patient's serum after treatment was taken as the primary test antibody. After washing, goat anti-human IgG-PE (2040-09; Southern Biotech) or goat anti-human IgM-PE (2020-09; Southern Biotech) was added as the secondary antibody to the keratinocyte or melanocyte suspensions. Cells were analyzed on a FACSCalibur or FACSVers flow cytometer (BD Biosciences) with the aid of FlowJo software (Tree Star).
Complement-dependent cytotoxicity assay
Keratinocytes and melanocytes were labeled with calcein AM (C1430; Thermo Fisher Scientific) according to the manufacturer's instructions. Sequential sera from patient no. 1 were heat-inactivated at 55°C for 30 minutes. The calcein-labeled keratinocytes or melanocytes, heat-inactivated serum at a final concentration of 20%, and complement sera human lyophilized powder (S1764; Millipore Sigma) at a final concentration of 20% were incubated together at 37°C for 2 hours. Cell lysis was evaluated by measuring the calcein released into the culture medium by means of a Spectra Max Gemini Microplate Reader (Molecular Devices). The percentage cell lysis was calculated according to the following formula: percentage cell lysis = (E − S)/(M − S) × 100, where E is the experimental release, S is the spontaneous release, and M is the maximum release by 1.5% Triton X-100. All values are given as averages of triplicate experiments.
ELISA
Anti-skin-specific antibodies [anti-Dsg1, anti-Dsg3, anti-Bullous Pemphigoid (BP) 180, anti-BP230, and anti-type VII collagen] in human serum were detected using an Anti-Skin Profile ELISA Kit (RG-7115R; MBL Co., Ltd.) according to the manufacturer's instructions. Sera from a B-cell lymphoma patient before allogeneic hematopoietic stem cell transplantation and at the time of dermatitis, and sera from ATL patient no. 1 before mogamulizumab and at the time of erythema multiforme (day 78, where day 0 was the day of the first dose of mogamulizumab) were tested using this kit. Similarly, sera from patients nos. 2, 5, and 7 before mogamulizumab and at the time of erythema multiforme (day 166, 126, and 168, respectively) were tested with this kit.
Immunofluorescence analysis
Biopsy tissues were fixed in 10% buffered formalin, embedded in paraffin, and serially sectioned. Tissue sections were deparaffinized, rehydrated, and antigen retrieval was carried out in pH 6.0 buffer in a microwave oven for 15 minutes. Immunofluorescence analyses were performed using mouse anti-human C1q mAb (#9A7; Abcam), polyclonal rabbit anti-human IgM antibody (IR513; DAKO), polyclonal rabbit anti-human IgG (IR512; DAKO), and appropriate isotype controls as the primary antibodies. Alexa Fluor 488 AffiniPure F(ab')2 Fragment donkey anti-mouse IgG (H+L) (715-546-151; Jackson Immunoresearch Laboratories, Inc.), and goat anti-rabbit IgG (H+L) cross-adsorbed secondary antibody labeled with Alexa Fluor 594 (A-110102; Thermo Fisher Scientific) were used. Nuclei were stained by VECTASHIELD Antifade Mounting Medium with DAPI (H-1200; Vector Laboratories). Slides were viewed using a fluorescence microscope (ZEISS LSM 700; Carl Zeiss MicroImaging GmbH), and images were acquired using ZEN 2009, Service Pack 2, Build 5.5.0.443 (Carl Zeiss MicroImaging GmbH). Human normal skin, formalin-fixed, paraffin-embedded block (CB703998; OriGene Technologies, Inc.) was used as a healthy control. Immunofluorescence analyses were performed under experimental conditions where isotype control antibodies did not generate nonspecific signals in the epidermis. Neither the dermis nor the basement membrane was analyzed due to nonspecific signals by isotype control antibodies.
Immunoglobulin M repertoire analysis by next-generation sequencing
Total RNA was extracted from PBMCs using RNeasy Mini Kits (QIAGEN, Hilden, Germany). First-strand cDNA was synthesized from 100 ng of total RNA using SMARTer RACE cDNA Amplification Kits (Takara Bio Inc., Shiga, Japan). Immunoglobulin M (IgM) heavy chain variable regions were amplified by the SMARTer 5′ RACE method and touch-down PCR using Universal Primer A Mix (Takara Bio Inc.) and the following primer specific for the constant region: IgM_H_1st: “TCGTATCCGACGGGGAATTCTCAC”. The PCR conditions were: 94°C for 2 minutes, followed by five cycles of 98°C for 10 seconds, 72°C for 30 seconds, five cycles of 98°C for 10 seconds, 70°C for 5 seconds, 72°C for 30 seconds, 20 cycles of 98°C for 10 seconds, 68°C for 10 seconds, 72°C for 30 seconds. Nested PCR was performed to obtain the amplicons of IgM heavy-chain variable regions for next-generation sequencing (NGS) using Nested Universal Primer A (Takara Bio Inc.) and the following primer: IgM_H_nest#1: “AGGAGACGAGGGGGAAAAGGGTT.” The PCR conditions were: 94°C for 2 minutes, followed by 25 cycles of 98°C for 10 seconds, 68°C for 5 seconds, 72°C for 30 seconds. Nested PCR products were purified using 1.8× volumes of Agencourt AMPure XP (Beckman Coulter, Brea, CA) and 500 to 700 bp sequences were size-selected by Blue Pippin (Sage Science, Beverly, MA). The sequencing of the PCR product of IgM heavy-chain variable regions was performed with PacBio RS II (Pacific Biosciences, Menlo Park, CA) and high-quality reads of circular consensus sequencing (CCS) were obtained by TAKARA BIO Inc. Read length trimming was performed with original perl script and shorter (below 400 bp) and longer (over 700bp) CCS reads were excluded. Germline usage [Variable (V), Diversity (D), and Joining (J) gene segments] of each sequence was identified using IgBlast (ncbi-igblast-1.4.0) by aligning to reference germline sequences of the international ImMunoGeneTics information system (IMGT; ref. 26) and the frequency of each germline was calculated and visualized using TIBCO Spotfire (TIBCO Software, Palo Alto, CA). Inverse Simpson's diversity index (27) was calculated by R and the “Vegan” package. The change in diversity index was calculated according to the following formula: inverse Simpson's diversity index for the IgM repertoire in PBMCs after treatment (mogamulzumab or chemotherapy)/the same patient's inverse Simpson's diversity index for the IgM repertoire in PBMCs before treatment (mogamulzumab or chemotherapy).
Results
IgG and IgM binding to keratinocytes and melanocytes in sera from patients with mogamulizumab-induced erythema multiforme
Relative to baseline before allogeneic hematopoietic stem cell transplantation, it was found that serum of a patient with B-cell lymphoma contained IgG which bound to human keratinocytes at the time of dermatitis (Supplementary Fig. S1). Because the patient was positive for anti-BP180 IgG antibody that is known to be the autoantibody responsible for Bullous Pemphigoid, the above observation assured us that the flow cytometry system used here functioned appropriately for detecting autoantibodies against skin cells.
In this manner, we investigated whether autoantibody was present in the sera of 8 patients who developed mogamulizumab-induced erythema multiforme. It was determined that relative to baseline before mogamulizumab treatment, IgG and IgM binding to human keratinocytes and melanocytes was present in sera from ATL patient no. 1 at the time of the occurrence of erythema multiforme (Fig. 1A–C). In addition, complement-dependent cytotoxicity (CDC) activity against keratinocytes (Fig. 1D) and melanocytes (Fig. 1E) mediated by autoantibodies in this patient's serum was observed. As shown in the figures, CDC against either keratinocytes or melanocytes was proportionally related to the severity of erythema multiforme. Of the remaining 7 patients, IgG binding to human keratinocytes was found in the serum of patient no. 6 at the time of erythema multiforme (Supplementary Fig. S2A). Keratinocyte-binding IgM was found in three additional patients (no. 2, Fig. 2A; no. 3, Fig. 2B; and no. 5, Fig. 2D). Also, in these 7 patients, IgG binding to human melanocytes was found in the serum of a patient at the time of erythema multiforme (no. 4, Fig. 2C) and IgM in 5 (no. 2, Fig. 2A; no. 3, Fig. 2B; no. 4, Fig. 2C; no. 5, Fig. 2D and no. 6; Supplementary Fig. S2A). Collectively, relative to baseline values in sera before mogamulizumab treatment, IgG binding to human keratinocytes was present at the time of mogamulizumab-induced erythema multiforme in 2 of 8 (25%) patients, whereas IgM was found in 4 (50%). Similarly, IgG binding to human melanocytes was seen at the time of erythema multiforme in 2 (25%) patients, and IgM in 6 (75%). In the remaining two patients, there was no detectable IgG or IgM binding to human keratinocytes or melanocytes (no. 7, Supplementary Fig. S2B and no. 8, Supplementary Fig. S2C). Some patients with anti-melanocyte antibodies suffering from mogamulizumab-induced erythema multiforme developed vitiligo, which persisted over a long period of time (Fig. 3A and B).
IgG and IgM binding to keratinocytes and melanocytes in sera from patients who received mogamulizumab treatment, but did not suffer from erythema multiforme
No keratinocyte-binding IgG or IgM was found to be present in any of the 5 patients tested (Supplementary Fig. S3). Melanocyte-binding IgG was found in two patients (no. 11, Supplementary Fig. S3C; and no. 13, Supplementary Fig. S3E), but melanocyte-binding IgM was not found in any of the 5 patients (Supplementary Fig. S3). Collectively, relative to baseline values in sera before mogamulizumab treatment, IgG or IgM binding to human keratinocytes was not present in sera after mogamulizumab in any of the patients without erythema multiforme (IgG and IgM, both 0%). On the other hand, IgG binding to human melanocytes was seen after mogamulizumab in 2 (40%) patients, and IgM in zero (0%), in the absence of erythema multiforme.
IgG and IgM binding to keratinocytes and melanocytes in sera from patients who received chemotherapy but no mogamulizumab
Neither keratinocyte- nor melanocyte-binding IgG or IgM was found in any of the 4 patients tested (Supplementary Fig. S4). Collectively, relative to baseline values in sera before chemotherapy, IgG or IgM binding to human keratinocytes or melanocytes was not present in sera after chemotherapy in any of the patients (IgG or IgM to keratinocytes or melanocytes, all 0%).
Anti-skin–specific antibodies in sera
Anti-BP180 antibody was present in the serum of a patient with B-cell lymphoma at the time of dermatitis, and in the serum of ATL patient no. 1 at the time of erythema multiforme (day 78), but was absent before mogamulizumab (day −5), as assessed by the Anti-Skin Profile ELISA Kit. However, these two anti-BP180–positive sera were negative for anti-Dsg1, anti-Dsg3, anti-BP230, or anti-Type VII collagen antibodies. The other sera tested were negative for all of these antibodies. In this context, patient no. 1 exhibited bullous lesions as described previously (see Fig. 2A in ref. 12), and as observed in patients with Bullous Pemphigoid. Such lesions were sometimes observed as skin-related AE after mogamulizumab treatment. The bullous lesions observed in another patient with ATL after mogamulizumab treatment are shown in Fig. 3C.
Immunofluorescence analyses of mogamulizumab-induced erythema multiforme and ATL skin lesions
In mogamulizumab-induced erythema multiforme, deposition of IgM in the epidermis was observed in 12 of 12 affected tissues. Deposition of C1q was seen in 8 of 12, with colocalization of IgM and C1q in all tissues (Fig. 4A–C; Table 1). Similarly, deposition of IgG was also observed in 12 of 12 epidermal tissues, with C1q colocalization in 8 of them (Fig. 4A–C; Table 1).
In ATL skin lesions, in contrast, IgM, IgG, or C1q deposition in the epidermis was absent in all 13 tissues from 13 individual patients with ATL (Fig. 4D–F; Table 1). IgM, IgG, and C1q were also absent from the epidermis in commercially obtained normal skin (data not shown and Table 1). Immunofluorescence analyses of mogamulizumab-induced erythema multiforme, ATL skin lesions, and normal skin are summarized in Table 1.
IgM repertoire analysis in PBMCs by NGS
We analyzed the IgM heavy-chain repertoire by deep sequencing of the variable region in PBMCs of 16 patients with ATL. To visualize the entire IgM repertoire, each germline gene and its frequency is depicted by individual color and height, respectively, in the bar chart. The total number of sequence reads, the number of sequence reads after trimming, the number of sequence reads aligned to a reference germline, and the inverse Simpson diversity index of each sample are shown in Supplementary Table S2. Inverse Simpson diversity indices for the IgM repertoire were decreased at the time of erythema multiforme in 6 of 7 patients (85.7%, nos. 2–4, 6, 8, and 9), and mean and median values were 106.7 and 11.3, respectively. The changes in diversity index were 0.50 and 0.25 (mean and median; Fig. 5A and B; Supplementary Table S2). In addition, in 6 of 7 patients (85.7%, nos. 2–6 and 18), more than half of all IgM germline genes were newly emergent at the time of erythema multiforme (Fig. 5A). However, inverse Simpson's diversity indices for the IgM repertoire were decreased after mogamulizumab without erythema multiforme in only 2 of 5 patients (40.0%, nos. 11 and 19), with mean and median values of 251.5 and 164.0, respectively. The changes in diversity index were 7.97 and 2.54 (mean and median; Supplementary Fig. S5A and S5B; Supplementary Table S2). Inverse Simpson's diversity indices for the IgM repertoire were decreased after chemotherapy without mogamulizumab in 2 of 4 patients (50.0%, nos. 14 and 15), and mean and median values were 295.9 and 111.1, respectively (Supplementary Fig. S6A and S6B; Supplementary Table S2).
Discussion
This study demonstrated that mogamulizumab treatment can elicit autoantibodies recognizing normal skin cells, confirmed by both flow cytometry and immunofluorescence analysis on tissues. In addition, NGS analysis demonstrated that IgM germline genes were newly emergent and expanded, resulting in IgM repertoire skewing at the time of occurrence of erythema multiforme after mogamulizumab. These autoantibodies are expected to play an important role in the skin-related AEs observed. To the best of our knowledge, this is the first report demonstrating the presence of skin-directed autoantibodies after mogamulizumab treatment.
In ATL patient no. 1, the degree of CDC activity against keratinocytes or melanocytes, mediated by antibodies present in the serum after mogamulizumab treatment, appeared to correlate with the severity of the erythema multiforme, suggesting a causal association. Consistent with this, deposition of IgG and IgM was observed in the epidermis of all 12 patients with mogamulizumab-induced erythema multiforme. Furthermore, colocalization of IgM or IgG with C1q was observed in 66.7% (8/12) and 66.7% (8/12) respectively, of patients with mogamulizumab-induced erythema multiforme. Because the binding of C1q to the Fc region of the antibody triggers one of the three representative complement activation pathways (28), these observations of colocalization are also consistent with the findings of CDC mediated by the serum of patient no. 1. We also surmise that some autoantibodies induce skin damage directly, and in addition, autoantibody-dependent cell-mediated cytotoxicity, especially in the case of IgG, could contribute to erythema multiforme after mogamulizumab treatment.
The immunoglobulin heavy chain is formed by recombination of Variable (V), Diversity (D), and Joining (J) gene segments, and the light chain by recombination of V and J gene segments, which leads to enormous antibody diversity (29). However, a potential disadvantage of this process is that some of the antibodies produced are self-reactive (30). In fact, the majority (55%–75%) of all antibodies produced by early immature B cells in humans may display self-reactivity, that is, they are autoantibodies (30). In addition to a central checkpoint, which is controlled by B-cell–intrinsic factors (31–34), a peripheral checkpoint controlled by Treg cells plays an important role in regulating autoantibody production (35). Therefore, Treg cell depletion by mogamulizumab might abrogate this latter peripheral checkpoint, leading to the elicitation of autoantibodies and resulting in the observed AEs.
The presence of anti-BP180 antibody in the serum of patient no. 1 at the time of erythema multiforme seems to be important in this context. In patients with IPEX, the presence of anti-BP180 antibody was also reported (25, 36). Collectively, these data might indicate that anti-BP180 antibody has certain characteristics by which it likely pass through the central checkpoint. This is also consistent with the reports that Treg cell dysfunction is associated with the pathogenesis of Bullous Pemphigoid (37–39), which allows the release of autoantibodies including anti-BP180 antibody in the host.
Vitiligo is a depigmenting disorder resulting from the loss of melanocytes in the skin. Although the pathogenesis of the disease remains obscure, autoimmune mechanisms are thought to be involved. Indeed, autoantibodies and autoreactive T lymphocytes that target melanocytes have been reported in some patients with vitiligo (40–42). Thus, the elicitation of antimelanocyte antibodies after mogamulizumab treatment leading to vitiligo, as observed in this study, is not unexpected. Dysfunction of Treg cells has been reported to be associated with vitiligo pathogenesis (43–45). Considering these findings together, presumably some of the antimelanocyte antibodies might also have certain characteristics by which they likely pass through the immunologic central checkpoint, in the same manner as the anti-BP180 antibody.
Notably, antibodies recognizing normal skin cells were also observed in 2 of 5 patients without mogamulizumab-induced erythema multiforme, but in none of 4 patients who received chemotherapy but no mogamulizumab. This is consistent with our hypothesis that mogamulizumab elicits autoantibodies. However, the reason why these patients did not suffer from erythema multiforme even though they possessed skin-directed autoantibodies is not clear. However, this is consistent with the generally accepted finding that autoantibodies are not necessarily pathogenic (46, 47). In addition, in our RNA sequencing (RNA-Seq) analysis using the Ion Proton Sequencer with Oncomine Immune Response Research Assay (Thermo Fisher Scientific) in PBMCs from the present 12 patients with ATL who received mogamulizumab, there were no significant correlations between the degree of Treg depletion estimated by FOXP3 gene expression, the kinetics of B-cell dynamism estimated by CD19 gene expression, or the presence or absence of autoantibody elicitation (data not shown). We found it interesting that mRNA expression ratios (after-vs.-before mogamulizumab) for TNF receptor superfamily member 17 (TNFRSF17, NCBI Reference Sequence: NM_001192.3), joining chain of multimeric IgA and IgM (JCHAIN, NCBI Reference Sequence: NM_144646.4), and POU class 2 associating factor 1 (POU2AF1, NCBI Reference Sequence: NM_144646.4) were 34.2 and 27.3 (mean, median), 29.6 and 10.6, and 16.7 and 10.0, respectively, in PBMCs from 7 patients with mogamulizumab-induced erythema multiforme. In contrast, these values were only 3.4 and 3.3, 2.3 and 2.2, and 2.0 and 1.2, respectively, in PBMCs from 5 patients without mogamulizumab-induced erythema multiforme (data not shown). Because these three genes (TNFRSF17, JCHAIN, and POU2AF1) are all B-cell-related, the present finding of increased expression thereof, especially in the patients with mogamulizumab-induced erythema multiforme, as assessed by RNA-Seq, is consistent with our hypothesis that the autoreactive B-cell clones were emergent and expanded at the time of occurrence of the disease. On the other hand, the important question of which patients who received mogamulizumab will develop pathogenic autoantibodies and with which clinical consequences has not yet been determined, and requires further investigation.
In this study, skin-reactive IgM was detected in 6 of 8 patients (75%), whereas IgG was present in 2 of 8 (25%) by flow cytometry. Although the reason why skin-directed IgM was predominant, rather than IgG, at the time of skin-related AEs caused by mogamulizumab is not clear, we accordingly focused on the IgM repertoire in PBMCs. In our NGS analysis, IgM repertoire skewing, as evidenced by a decreased inverse Simpson's diversity index, was observed at the time of erythema multiforme occurrence in 6 of 7 patients, and there was a clear trend toward lower indices in patients at the time of erythema multiforme than in patients after mogamulizumab or chemotherapy who did not suffer from erythema multiforme. In addition, the degree of alteration in IgM repertoire skewing was more obvious in patients with mogamulizumab-induced erythema multiforme compared with those without, as evidenced by the changes in diversity index. Furthermore, more than half of the IgM germline genes were newly emergent at the time of erythema multiforme occurrence in 6 of 7 patients. These data also collectively indicate that several autoreactive B-cell clones were newly emergent and expanded at the time of erythema multiforme after mogamulizumab. These findings are also consistent with a report that Treg cells have an important role in regulating autoantibody production (35). It must also be kept in mind that these newly emergent and expanded autoreactive B-cell clones certainly include not only those secreting skin-reactive autoantibodies but also against other organs, such as the thyroid (14).
Although we did not investigate any direct contributions of CD8 T cells in this study, we have previously reported a high degree of CD8 T-cell infiltration into the affected lesions of mogamulizumab-induced skin-related AE (12). Therefore, we should also keep in mind the strong possibility that the activation of skin-directed CD8 T cells, and the accompanying hyper-cytokinemia, may also be involved in the pathogenesis of such AE.
In conclusion, this study has demonstrated that mogamulizumab treatment elicits autoantibodies binding keratinocytes and melanocytes, and that these autoantibodies have an important role in the pathogenesis of mogamulizumab-induced skin-related AEs. This is possibly associated with Treg cell depletion by mogamulizumab and the consequent breakdown of peripheral immunologic checkpoints regulating autoantibody production. It will be important to identify the target molecules recognized by these mogamulizumab-induced autoantibodies, which will contribute to the further understanding of the fundamental role of the human immune system in this context. Furthermore, if we could guide the elicitation of autoantibodies recognizing tumor-related antigens, such as so-called shared antigens, this would lead to marked enhancement of the antitumor activity of mogamulizumab. Importantly, this study indicates that B-cell-targeting therapies, such as rituximab, and others, should be effective for treating mogamulizumab-induced skin-related AEs (24). Further investigations regarding the AEs associated with Treg cell depletion are warranted.
Disclosure of Potential Conflicts of Interest
M. Saito holds ownership interest (including patents) in Kyowa Hakko Kirin Co., Ltd. H. Inagaki reports receiving commercial research grants and speakers bureau honoraria from Kyowa Hakko Kirin Co., Ltd. S. Iida reports receiving commercial research grants from Kyowa Hakko Kirin Co., Ltd., Chugai, Ono, Takeda, Janssen, Celgene, Bristol-Myers Squibb, Daichi-Sankyo, Novartis, MSD, AbbVie, and Gilead, and speakers bureau honoraria from Takeda, Celgene, Janssen, Daichi-Sankyo, Ono, and Bristol-Myers Squibb. T. Ishida reports receiving commercial research grants from Kyowa Hakko Kirin Co., Ltd., Bayer AG, and Celgene K.K., and speakers bureau honoraria from Celgene K.K., Kyowa Hakko Kirin Co., Ltd., and Mundipharma K.K. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: T. Ishii, T. Ishida
Development of methodology: Y. Suzuki, T. Ishii, T. Ishida
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Suzuki, M. Saito, I. Urakawa, A. Matsumoto, A. Masaki, A. Ito, S. Kusumoto, M. Hiura, T. Takahashi, A. Morita, T. Ishida
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Suzuki, M. Saito, T. Ishii, A. Matsumoto, A. Ito, S. Suzuki, T. Takahashi, A. Morita, H. Inagaki, T. Ishida
Writing, review, and/or revision of the manuscript: Y. Suzuki, M. Saito, T. Ishii, I. Urakawa, A. Matsumoto, A. Masaki, S. Kusumoto, M. Hiura, T. Takahashi, A. Morita, H. Inagaki, S. Iida, T. Ishida
Study supervision: T. Ishii, S. Iida, T. Ishida
Acknowledgments
We are sincerely grateful to Professor Ryuzo Ueda, Aichi Medical University School of Medicine, for his great advice on this study. We also thank Ms. Chiori Fukuyama for her excellent technical assistance, and Ms. Naomi Ochiai for her expert secretarial assistance. This work was funded by Kyowa Hakko Kirin Co. Ltd., and also by grants-in-aid for scientific research (B) (no. 16H04713 to T. Ishida) and from the Japan Agency for Medical Research and Development (nos. 17ck0106287h0001 to T. Ishida, 16cm0106301h0001 to T. Ishida, and 15ck0106132h0002 to S. Iida and T. Ishida).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.