MECHANISMS OF DISEASE
Mechanisms of disease
Castleman’s tumours and production of autoantibody in paraneoplastic pemphigus
Liangchun Wang, Dingfang Bu, Yong Yang, Xixue Chen, Xuejun Zhu
Summary
Background Paraneoplastic pemphigus is an autoimmune mucocutaneous disease associated with Castleman’s tumours, which when surgically removed often result in great improvement of mucocutaneous lesions. An IgG autoantibody against epidermal proteins is often used as a diagnostic marker for disease. Our aim was to ascertain the role of Castleman’s tumours in production of the autoantibody and pathogenesis of paraneoplastic pemphigus. Methods We enrolled seven patients with paraneoplastic pemphigus associated with Castleman’s disease and assessed the effect of removal of tumours on mucocutaneous lesions in six individuals and on autoantibody titre with indirect immunofluorescence in four patients. We cultured tumour cells from one patient and assayed the secreted autoantibody. Finally, we characterised the gene sequence and expression of the variable region of the immunoglobulin heavy chain (IgVH) in tumour B cells from all patients by reverse transcription-PCR, DNA sequencing, and in-situ hybridisation. Findings Cutaneous lesions disappeared within 6–11 weeks after resection of tumours. Mucosal lesions also improved in this period, but lasted for 5–10 months overall. Autoantibody titre decreased and became undetectable within 5–9 weeks in three of four patients assessed. We identified secreted autoantibody, similar to that identified in patients’ serum, in cultured tumour cells. The tumour B-cells of the seven patients shared and expressed two rearrangement patterns of complementarity determining region 3 (CDR3) of IgVH. Interpretation Secreted autoantibody from Castleman’s tumours, which reacts against epidermal proteins, could be an essential factor in the pathogenesis of paraneoplastic pemphigus. We noted clonal rearrangement, resulting in similar variable regions of IgVH, in tumour B cells isolated from all seven patients. However, whether this pattern is associated with autoimmunity remains to be ascertained. Lancet 2004; 363: 525–31
Introduction
Paraneoplastic pemphigus is an autoimmune mucocutaneous disorder, characterised by an associated neoplasm and the presence of a unique IgG autoantibody that recognises epidermal proteins.1–5 Neoplasms associated with the disease are often of lymphocytic origin and include non-Hodgkin’s lymphoma,6,7 chronic lymphocytic leukaemia,1,8 and Castleman’s disease.9–13 One of the characteristic features of paraneoplastic pemphigus associated with Castleman’s disease and other benign tumours is the remarkable improvement in symptoms noted after removal of the tumour.11–14 In some individuals accompanied bronchiolitis obliterans occasionally may result in respiratory failure and death.9,10 The role of the neoplasm in the pathogenesis of paraneoplastic pemphigus is controversial, with detailed data unavailable possibly because of the limited number of affected individuals. Clinically, an IgG autoantibody against epidermal proteins is often used as a diagnostic marker for the disease,15,16 but might also be involved in autoimmune injuries.1,5 Our aim was to ascertain the mechanism by which the autoantibody is generated and to investigate the possible role of Castleman’s tumours in the process.
Methods
Participants We included in our analyses seven patients with paraneoplastic pemphigus associated with localised hyaline vascular-type Castleman’s tumours who were diagnosed and treated at the Department of Dermatology, Peking University First Hospital between May, 1999, and December, 2002. The patients all had clinically similar lesions of pemphigus vulgaris, pemphigus erythematosus, and lichen planus (figure 1). Histological examination of the cutaneous lesions indicated epidermal acantholysis or intraepidermal acantholytic blisters, individual keratinocyte necrosis, interface vacuolar degeneration, and diffuse dermal lymphocyte infiltration (figure 1). Six patients were refractory to treatment, including corticosteroids and other adjuvant therapies, and one patient (patient 6) showed transient improvement after infusion of gamma globulin (10 g daily for 3 days) during preparation for surgery. We undertook further examinations because of the persistence of mucocutaneous lesions and the skin biopsy findings. CT scans disclosed the tumours; five located in the retroperitonium and two in the mediastinum. Protein extract of epidermis from healthy human skin blotted by patients’ sera revealed positive bands similar in number, size, and intensity among the seven patients, all of whom had three positive bands of about 250 kDa, 210 kDa, and 190 kDa, with the 210 kDa band being the most intense. Indirect immunofluorescence with monkey tongue and mouse urinary bladder sections, sera from the seven 525
Department of Dermatology, Peking University First Hospital, Beijing, China (L Wang MD, Prof D Bu MD, Y Yang MD, X Chen MD, Prof X Zhu MD) Correspondence to: Dr Xuejun Zhu, Department of Dermatology, Peking University First Hospital, Beijing 100034, China (e-mail: ZHUXJ@public.bta.net.cn)
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
MECHANISMS OF DISEASE
GLOSSARY
COMPLEMENTARITY DETERMINING REGION
The heavy and light chains of antibodies each contain three regions of hypervariability, termed complementarity determining regions (CDR), which interact with antigen.
patients, and fluorescein isothiocyanate-conjugated antibody against human IgG revealed IgG deposition on the cell surface and in the cytoplasm of the squamous and transitional epithelia. We surgically removed the tumours of the seven patients, and routine histopathological examination indicated that all had hyaline vascular-type Castleman’s disease. We also included as controls, six individuals admitted to hospital with hyaline vascular-type Castleman’s disease with no mucocutaneous involvement and three patients with reactive lymphadenopathy. We did not obtain informed consent from participants, since clinical observation and photography of skin lesions are routine procedures for dermatological patients, skin biopsy is thought to be essential for diagnosis, and observation of autoantibody change is needed for adequate therapy after surgery. Furthermore, there are no laws to prevent the use of surgical specimens for medical research in China. We therefore concluded that publication of these data was in accord with the guidelines issued by the Committee of Medical Ethics, Ministry of Health, China. Procedures We measured IgG autoantibody titres after resection of tumours in blood samples obtained from four patients at intervals of 1–2 weeks for up to 9 weeks after surgery, using indirect immunofluorescence with mouse urinary bladder section to measure the antibody. To identify surface markers on lymphocytes in the tumours, we stained paraffin sections available from six patients with monoclonal antibodies against human CD20 and CD45RO (Santa Cruz Biotech, CA, USA).
We visualised the antibody with antimouse IgG conjugated with peroxidase and diaminobenzidine staining. To ascertain whether Castleman’s tumours produced autoantibody, we cut the tumour from one patient into small pieces with scissors immediately after resection, and filtered the tissue through several layers of gauze. We washed the cells three times with culture medium, and then cultured them in Iscove’s modified Dulbecco’s medium (IMDM), containing 10% human serum of AB type blood, which we changed once a week. We did indirect immunofluorescence and western blot analyses with the culture medium used in the third week of incubation and concentrated by ultrafiltration (Centricon plus 80; Millipore, Bedford, MA, USA). We included in the western blot, a sample of epidermis taken from a foreskin obtained surgically from a healthy man, which had been incubated in saline at 4°C for 36–48 h, homogenised in a protein extract buffer, ultrasonicated, and centrifuged. We seperated the proteins in the supernatant, using sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE), and transferred them onto a nitrocellulose membrane. The membrane strips were incubated with the concentrated culture medium or with one of various control solutions, including unused IMDM, a serum sample from a healthy individual, and a culture medium that had contained tumour B cells from a patient with multiple myeloma. The blotted IgG was identified by a sheep antibody against human IgG conjugated with horseradish peroxidase and developed by diaminobenzidine. To ascertain the genetic characteristics of the B cells of Castleman’s tumours, with respect to the variable region of the immunoglobulin heavy chain (IgVH), we used reverse transcription-PCR, DNA sequencing of cloned PCR product, and in-situ hybridisation. The panel shows the 14 primer pairs used for amplification of IgVH— forward primers locate in the leader sequence, framework region 1, and framework region 3 of IgVH genes,17–19 and reverse primers in the joining region or constant region of immunoglobulin heavy chain. We treated total RNA
Primer pairs used for amplification of IgVH
VL family specific primers/C primer17 VL1 5 -CCATGGACTGGACCTGGA VL2 5 -ATGGACATACTTTGTTCCAC VL3 5 -CCATGGAGTTTGGGCTGAGC VL4 5 -ATGAAACACCTGTGGTTCTT VL5 5 -ATGGGGTCAACCGCCATCCT VL6 5 -ATGTCTGTCTCCTTCCTCAT C 5 -CAGGAGACGAGGGGGAA FR1 family primers/JH consensus primer18 FR1C 5 -AGGTGCAGCTG(G/C)(A/T)G(G/C)AGTC(G/A/T)GG FR1-1 5 -CCTCAGTGAAGGTCTCCTGCAAGG FR1-2 5 -TCCTGCGCTGGTGAAGCCACACA FR1-3 5 -GAAGATCTCTCCTGTGCAGCCTCTGG FR1-4 5 -GTGGATCCTGTCCCTCACCTGC(A/G)(T/C)TG FR1-5 5 -GAAAAAGCCCGGGGAGTCTCTGA FR1-6 5 -GAAGATCTCTCACTCACCTGTGCCATC JH 5 -ACCTGAGGAGACGGTGACC(A/G)(G/T)(G/T)GT Figure 1: Mucocutaneous lesions (A, B) and histological findings (C, D) of paraneoplastic pemphigus
Haematoxylin and eosin staining of oral mucosa from patient 4 (C; 1, 2=necrotic keratinocytes, 3=infiltration of mononuclear cells in a lichenoid pattern) and skin from trunk of patient 6 (D; 1, 2=necrotic keratinocytes, 3=suprabasilar acantholysis and intraepidermal cleavage, 4=interface vacuolar degeneration).
FR3 consensus primer/JH consensus primer19 FR3 5 -CTGTCGACACGGCCGTGTATTACTG JH 5 -AACTGCAGAGGAGACGGTGACC
VL=leader sequences of heavy chain. FR1=framework region 1 in variable region of heavy chain. C=constant region of heavy chain. JH=joining region of heavy chain.
526
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
MECHANISMS OF DISEASE
Lesions and distribution
Respiratory involvement
Castleman’s tumours Location Size (cm)
Patient number/age (years)/sex 1/39/male Erosions and ulcers on lips, oral mucosa, tongue, and penis; papulosquamous and lichen planus-like eruptions on upper body; palmaplantar erythema 2/17/male Erosions and ulcers on cheek, tongue conjunctivae, and penis; lichenoid papules and plaques on hands; widespread erythematous macules and papules on body 3/30/male Erythema and ulcers on cheek and tongue; vesicopapules and erythematous patches on chest and upper back
4/35/female
Oral lesions only
5/17/female
Vesicles and bleeding erosions in vagina and on oral mucosa and lips; lichenoid papules on forearms, hands, and soles of feet Ulcers and erosions on tongue, lips, conjunctivae, and penis; pemphigus vulguis-like lesions over whole body Bleeding erosions in vagina and on oral mucosa and lips; dark brown patches on whole body; ulcers and purulent crusts on palms of hands and soles of feet
6/29/male 7/22/female
Retroperi- 7·0 5·0 6·0 toneum None Retroperi- 8·0 5·0 5·0 toneum Dyspnoea and cough Retroperi- 12·0 6·0 6·0 developed 4–5 months toneum after surgery due to lung abscess Dyspnoea and cough Mediastinum 4·5 6·0 6·5 still present 21 months after surgery Dyspnoea and cough Mediastinum 9·0 6·0 3·0 recovered 7–8 months after surgery None Retroperi- 9·0 8·1 7·5 toneum None Retroperi- 8·0 6·0 6·0 toneum
None
Clinical characteristics
Time since surgery (weeks)
extracted from tumours with RNase-free DNase I before undertaking reverse transcription. We used tumour cDNA samples from the seven patients for PCR amplification. We used as controls, genomic DNA samples extracted from paraffin-embedded tissues of the six patients with Castleman’s disease without mucocutaneous involvement and from the three individuals with reactive lymphadenopathy. PCR templates, using water as a negative control, cDNA from human peripheral lymphocytes from healthy individuals as a normal control, and DNA from Namalwa cells (a Burkitt’s lymphoma-derived cell line) as a monoclonal control were always included in the PCR experiment to exclude any possible carry-over contamination. We purified and cloned PCR products, showing as a sharp band in polyacrylamide gel after electrophoresis, into pBluescript SK (+) vector (Stratagene, La Jolla, CA, USA). We picked up and sequenced more than seven recombinant clones from each patient with paraneoplastic pemphigus. We linearised and in-vitro transcribed the plasmid that contained COMPLEMENTARITY DETERMINING REGION 3 (CDR3) of the IgVH sequence shared by the seven patients with paraneoplastic pemphigus to make antisense and sense RNA probes for in-situ hybridisation. Role of the funding source The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, writing of the report, or in the decision to submit the paper for publication.
decreased rapidly in the third week after surgery and became undetectable within 5–9 weeks in three patients— data for one patient are incomplete because of the unavailability of blood samples (figure 2). Staining of tumour sections from six patients (tumour of patient 1 not available) showed that CD45RO-positive T lymphocytes scattered in interfollicular areas, and CD20-positive B lymphocytes predominated in lymphoid follicles of various sizes and shapes. This distribution pattern is similar to that noted in lymph nodes of healthy individuals. On western blot, protein extract of healthy human epidermis blotted by the culture medium used to grow
A 12 10 8 6 4 2 0 1 B 2 3 5 Patient number* 6 7
Results
The table shows the characteristics of the seven patients, most of whom responded badly or only partially to routine treatment, including corticosteroids and other immunosuppressive drugs. After resection of Castleman’s tumours, however, cutaneous lesions improved gradually and disappeared within 6–11 weeks without recurrence in the six patients with paraneoplastic pemphigus (figure 2), and erosions of the mucosa and ulcers also improved. However, mucosal lesions recovered only slowly, and one or two oral ulcers remained in most patients until 5–10 months after surgery. Dyspnoea and cough, present in two patients before surgery, improved gradually after surgery in patient 5 and disappeared without sequelae within 7–8 months, and remained unchanged despite surgery during 21 months’ follow-up in patient 4. Autoantibody titres, assessed in patients 3, 5, 6, and 7,
1:640 Antibody titre 1:320 1:160 1:80 1:40 0 3 2 1 0 1 2 3 4 5 6 7 8 Time since resection of tumour (weeks) 9
Patient 3 Patient 5 Patient 6 Patient 7
Figure 2: Duration of cutaneous lesions (A) and change in autoantibody titre (B) after resection of Castleman’s tumours
*Patient 4 had no cutaneous lesions.
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
527
For personal use. Only reproduce with permission from The Lancet.
MECHANISMS OF DISEASE
A
250
B
210 190
Similarly, 11 nucleotide sites are different among the homologous clones of the 122-bp sequences. Most clones, however, differed from the consensus sequence by only one or two nucleotides; only the clone from patient 4 has multiple nucleotide differences. To investigate the expression of the rearranged gene containing the 128-bp CDR3 of IgVH in tumour B cells, we did in-situ hybridisation with an antisense RNA probe transcribed from a clone of 128 bp. Tumours from six patients (tumour of patient 1 not available) showed similar patterns of signal distribution, with gene expression arising in the cytoplasm of the cells present in follicles of the tumours. These findings suggest that this rearrangement of CDR3 constitutes one of the predominant B-cell clones in tumours and this mRNA is presented in cytoplasm. By contrast, we noted no signal in the cytoplasm of tumours from the six participants with Castleman’s disease without mucocutaneous lesions or in the lymph nodes of the three individuals with reactive lymphadenopathy.
nt tie pa fro m ls ltu re m ro ls er um ed
7 iu m sa m pl e 7 rc ou ye le m us Un nt tie lo ed Co m nt IM DM pa of m DM of ru Se IM
C
M
ul
tip
tu
m
a
cu
el
D
Discussion
Our findings suggest that Castleman’s tumours are involved in the production of autoantibodies in patients with paraneoplastic pemphigus, and that this production is important in the pathogenesis of the disease. As previously reported,20,21 our findings indicate a close relation between the presence of a Castleman’s tumour and the mucocutaneous manifestations seen in patients with paraneoplastic pemphigus, which do not respond to routine immunosuppressive treatment but resolve rapidly after resection of the cancer. The presence of autoantibodies, which is characteristic of the disease, seems to be an essential factor in the pathogenesis of the disease.1,5,10 In our patients, after resection of the tumours, complete recovery from cutaneous lesions and great improvement of mucosal involvement were accompanied by a decline in concentrations of autoantibody in a period roughly corresponding to the half-life time of circulating IgG.22 Paraneoplastic pemphigus, therefore, seems to be an autoantibody-mediated mucocutaneous disorder. But, how do Castleman’s tumours in these patients stimulate the production of autoantibody? One hypothesis is that the tumour proteins function as antigens and induce the development of autoantibody against epithelial proteins.23 However, sera from patients did not recognise tumour protein extract in our western blot experiment (data not shown). Furthermore, the disappearance of cutaneous lesions as well as autoantibody within 11 weeks of removal of the tumours does not support this hypothesis. There is also no direct evidence that Castleman’s tumours alter the response of antibody synthesis in the immune system via tumorigenic cytokines or helper T cells.23 The epitopespreading hypothesis suggests that lichenoid dermatitis initiates primary autoimmune responses by the release of intracellular antigens from keratocyte necrosis.24,25 However, we found no lichenoid dermatitis in the course of the disease in four of our seven patients (patients 3, 4, 6, and 7). Our findings suggest instead that several tumour B-cell clones produce autoantibodies against desmosomal and hemidesmosomal proteins, such as envoplakin and periplakin.3,4 The direct evidence comes from IgG autoantibody secreted by in-vitro cultured tumour cells. This autoantibody also recognises normal human epidermal proteins of 210 kDa and 190 kDa on western blot and proteins in mouse bladder epithelia in indirect immunofluorescence, indicating the same or similar specificity as that of autoantibodies in patients’ sera.
Figure 3: Identification of IgG antibody secreted by cultured tumour cells from patient 7
A=western blot. B, C, D=indirect immunofluorescence with mouse urinary bladder and culture medium of tumour cells (titre 1:80; B) and serum (C) from patient 7, and with transitional epithelia and culture medium of tumour cells from patient with multiple myeloma (D).
tumour cells from patient 7 showed two bands of about 210 kDa and 190 kDa (figure 3); no bands were noted in control lanes. Indirect immunofluorescence with mouse urinary bladder sections and the culture medium used to grow tumour cells from patient 7 also indicated IgG deposition on cell surface and in cytoplasm of the transitional epithelia (figure 3). When we used the 14 pairs of PCR primers for amplification of various regions of IgVH, PCR with primers located in framework region 3 and joining region for the amplification of CDR3 of IgVH produced one discrete band of about 128 bp in every patient with paraneoplastic pemphigus, whereas PCR with the other primer pairs showed multiple bands, a smeared band, or no product (data not shown). We noted no discrete band when we did PCR with genomic DNA from the six Castleman’s tumours taken from patients without mucocutaneous involvement or from the three lymph nodes from individuals with reactive lymphadenopathy. After cloning the PCR product of the discrete 128-bp band into plasmid, we picked up and sequenced between seven and ten clones for each patient. Figure 4 shows the aligned sequences of the cloned PCR products and the number of clones identified. Only two kinds of highly homologous sequences were noted in the seven patients. As shown in figure 4, the 128-bp sequences are the major portion of CDR3 clones seen in all patients, and the 122-bp sequences are the relatively minor one seen in five patients. In the major 128-bp sequences, 15 nucleotide sites are different among the clones, and the variation of G to C at nucleotide number 41 is frequently seen.
528
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
A
Patient 1: 128 128 128 128 128 128 128 128
Patient 2:
CTGTCGACACGGCCGTGTATTACTGTGCGAGAGATAGGCGGGGGGAATGGCCTCCCTCGGATTACTACTACTACTACTACATGGACGTCTGGGGCAAAGGGACCACGGTCACCGTCTCCTCTGCAGTT 128 ----------------------------------------C--------------------------------------------------------------------------------------- 128
Patient 3:
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------C-----------------------------------------------------------------------------------------------------------------------------------G-------------------------------------------------------G----------------------------------------------------T----------------------------------------------------------------------------------------------------
Patient 4:
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------C--------------------------------T---------------------------------------------------------------------------------------------C----------------------------------------C-------------------------------------------------------------------------------------C---------------------------------------------------------------------------------------
Patient 5:
-------------------------------------------------------------------------------------------------------------------------------- 128 ----------------------------------------C--------------------------------------------------------------------------------------- 128 128 128 128 128 128 128 128 128 128
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
Patient 6:
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------C------------------------------------------------------------------------------------------------------------------C------------------------------------------------------------------------------------------------------------------------------C------------------------C--------------------------------------------------------------
Patient 7:
----------------------------------------C--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------C---------------------------T--------------------------------------------------------G-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------C---------------------------
-------------------------------------------------------------------------------------------------------------------------------- 128 --------------------------------------T-----------------------------------C----------------------------------------------------- 128 -------------------------C--------------C--------------------------------------------------------------------------------------- 128 Consensus: *************************.*.*******.**.*.****.******.************.*******..**.***.******************..***.**********************
B
Patient 1:
Patient 4
CTGTCGACACGGCCGTGTATTACTGTGCGATGACGACGATTTTTGGAGTGGTTATTAGGCCTAAGCTTGATGCTTTTGATATCTGGGGCCAAGGGACAATGGTCACCGTCTCCTCTGCAGTT 122
Patient 5:
-------------------------------------------------------------------------------------------------------------------------- 122 --------------------------A-------A---------------------------------A---------------AA-A---------------------------------- 122 -----------------------------------------------------------------------------------------------G-------------------------- 122
Patient 6:
-------------------------------------------------------------------------------------------------------------------------- 122 -----------------------------------G-------------------------------------------------------------------------------------- 122
Figure 4: Sequence alignment of CDR3 of IgVH amplified by PCR from Castleman’s tumours of patients with paraneoplastic pemphigus
Clones can be divided into two categories, according to their similarity of sequence: major homologous clones (A), and relatively minor homologous clones (B). For each kind of sequence, one sequence is shown and sequences different from the upper line are listed for other homologous clones. Sequences of PCR primer sites are underlined. C=alignment of two sequence types.
For personal use. Only reproduce with permission from The Lancet.
Patient 7:
-------------------------------------------------------------------------------------------------------------------------- 122 -------------------------------------------------------------G------------------------------------------------------------ 122 -----------------------------------T-------------------------------------------------------------------------------------- 122
-------------------------------------------------------------------------------------------------------------------------- 122 ---------------------------------------------T---------------------------------------------------------------------------- 122 Consensus: **************************.*******..*********.***************.******.***************..*.*******.**************************
C
MECHANISMS OF DISEASE
529
CTGTCGACACGGCCGTGTATTACTGTGCGA-GAGATAGGCGGGGGGAATGGCCTCCCTCGGATTACTACTACTACTACTACATGGACGTCTGGGGCAAAGGGACCACGGTCACCGTCTCCTCTGCAGTT 128 CTGTCGACACGGCCGTGTATTACTGTGCGATGACGACGATTTTTGGAGTGGÑ-TTATTAGGCCTAAGCTTGATGCTTTT-----GATATCTGGGGCCAAGGGACAATGGTCACCGTCTCCTCTGCAGTT 122 Consensus: ****************************** ** * ******* *** * * ** ** * * ** * ** ******** ******* * **********************
MECHANISMS OF DISEASE
RELEVANCE TO PRACTICE
BACKGROUND
Paraneoplastic pemphigus is an autoimmune disease frequently associated with some types of tumour. Patients with this disease have autoantibodies specific for epidermal proteins. Although in some patients symptoms have been found to improve when such tumours are removed, the exact role of the tumour in the pathogenesis of paraneoplastic pemphigus has remained unclear. This research found that in seven patients with paraneoplastic pemphigus and Castleman's tumour, the tumours all expressed similar complementary determining region 3 (CDR3) of immunoglobulin heavy chain. Furthermore, one tumour was found to secrete autoantibodies similar to those found in the patient’s serum. Serum autoantibodies became undetectable after resection of the tumours.
IMPLICATIONS
In patients with paraneoplastic pemphigus, Castleman's tumours may have an important role in the pathogenesis of the disease. As well as providing an important insight into the disease process in this group, these results may suggest an underlying mechanism in other forms of the disease.
Moreover, the Castleman’s tumours of our patients have integrated lymphoid follicle structure as well as B-cell and T-cell distribution comparable to normal lymph nodes, lending support to the notion that these lymphocytes are immunologically active. Autoantibody disappeared within 5–9 weeks of removal of the tumours in our patients, also indicating an association between the tumour and production of autoantibody. Western blot analyses with expressed plakin proteins as the antigens should be undertaken to confirm the specificity of the secreted autoantibody from cultured tumour cells. Although mucosal lesions improved rapidly after surgery, oral ulcers lasted for 5–10 months after the disappearance of autoantibody in most of our patients. Pulmonary involvement recovered 7–8 months after surgery in one patient, and remained unchanged in another. Consequently, mechanisms other than tumorigenic autoantibody-mediated impairment are involved in the mucosal and pulmonary injuries. However, based on clinical improvement after surgery, secreted autoantibodies from the tumours seem essential in the pathogenesis of paraneoplastic pemphigus. The genes that encode immunoglobulin heavy chain variable regions are assembled during B cell development from variable gene segments termed V, D, and J through a process of site-specific recombination. Individual immunoglobulin rearrangements are specific for a given B cell and its progeny. CDR3 of IgVH is the most diverse of the six CDRs in the immunoglobulin molecule, and is also the most important portion of the molecule for binding antigens and for determining its specificity.26 In view of the huge diversity of immunoglobulin, the identical or very similar sequence of CDR3 in IgVH noted in our patients is very unlikely to be a chance finding. Additionally, specific signals of in-situ hybridisation seen in the patients with paraneoplastic pemphigus but not in the controls confirm that the variable region was expressed in the tumours. One to two nucleotide substitutions but no deletions or insertions were noted in the homologous clones, indicating that there are one or two aminoacid differences in the homologous CDR3 regions. Normally, somatic mutations in the variable region arise after antigenic stimulation, involving affinity maturation of antibodies. Therefore, these differences could indicate the different affinity but the same specificity of these B-cell clones. The same or similar CDR3 of IgVH in Castleman’s tumours of the seven patients implies that clonal rearrangement, resulting in
identical or very similar variable regions, happened in tumour B cells, although we were unable to identify the whole variable region. The explanation for the identical or very similar CDR3 of IgVH could be that a common antigen of unknown origin reacted to the tumour B-cell precursors in these tumours at an early stage of development. However, why the antigen stimulated only the tumours B cells and not normal immunological cells is unclear. More work, such as enzyme-linked immunospot assay (ELISPOT) of cultured tumour cells and the specificity of cloned variable region of light and heavy chain, needs to be done to ascertain whether tumour B cells with the same or similar CDR3 of IgVH constitute the lymphocytes secreting autoantibody specifically against plakin proteins in epidermis. Our findings indicate that expansion of several immunologically active B-cell clones arises in Castleman’s tumours. These clones can secrete autoantibodies that react specifically with epidermal proteins, leading to impairment of cell-cell adhesion and subsequent development of pemphigus lesions in skin and mucosal membranes. In view of our results, we believe patients with refractory pemphigus should be examined as early as possible for the presence of a tumour, and we maintain that complete resection of tumours is the most effective treatment available for this disease.
Contributors
L Wang and D Bu designed and did the experimental work, interpreted the findings, and wrote the manuscript. Y Yang was involved in design of experiments, and X Chen in the collection and analysis of data. X Zhu proposed the study, organised and helped with the design of the research, and is responsible for the data reported.
Conflict of interest statement
None declared.
Acknowledgments
This work was funded by the National Natural Science Foundation of China (grant number 30371292).
References
Anhalt GJ, Kim SC, Stanley JR, et al. Paraneoplastic pemphigus: an autoimmune mucocutaneous disease associated with neoplasia. N Engl J Med 1990; 323: 1729–35. 2 Oursler JR, Labib RS, Ariss-Abdo L, Burke T, O’Keefe EJ, Anhalt GJ. Human autoantibodies against desmoplakins in paraneoplastic pemphigus. J Clin Invest 1992; 89: 1775–82. 3 Kim SC, Kwon YD, Lee IJ, Chang SN, Lee TG. cDNA cloning of the 210-kDa paraneoplastic pemphigus antigen reveals that envoplakin is a component of the antigen complex. J Invest Dermatol 1997; 109: 365–69. 4 Mahoney MG, Aho S, Uitto J, Stanley JR. The members of the plakin family of proteins recognized by paraneoplastic pemphigus antibodies include periplakin. J Invest Dermatol 1998; 111: 308–13. 5 Amagai M, Nishikawa T, Nousari HC, Anhalt GJ, Hashimoto T. Antibodies against desmoglein 3 (pemphigus vulgaris antigen) are present in sera from patients with paraneoplastic pemphigus and cause acantholysis in vivo in neonatal mice. J Clin Invest 1998; 102: 775–82. 6 Schlesinger T, McCarron K, Camisa C, Anhalt G. Paraneoplastic pemphigus occurring in a patient with B-cell non-Hodgkin’s lymphoma. Cutis 1998; 61: 94–96. 7 Plumb R, Doolittle GC. Paraneoplastic pemphigus in a patient with non-Hodgkin’s lymphoma. Am J Hematol 1996; 52: 58–59. 8 Marzano AV, Grammatica A, Cozzani E, Terracina M, Berti E. Paraneoplastic pemphigus: a report of two cases associated with chronic B-cell lymphocytic leukaemia. Br J Dermatol 2001; 145: 127–31. 9 Wolff H, Kunte C, Messer G, et al. Paraneoplastic pemphigus with fatal pulmonary involvement in a woman with a mesenteric Castleman tumour. Br J Dermatol 1999; 140: 313–16. 10 Nousari HC, Deterding R, Wojtczack H, et al. The mechanism of respiratory failure in paraneoplastic pemphigus. N Engl J Med 1999; 340: 1406–10. 11 Hsiao CJ, Hsu MM, Lee JY, Chen WC, Hsieh WC. Paraneoplastic pemphigus in association with a retroperitoneal Castleman’s disease 1
530
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
MECHANISMS OF DISEASE
12
13
14 15
16
17
18
presenting with a lichen planus pemphigoides-like eruption: a case report and review of literature. Br J Dermatol 2001; 144: 372–76. Jansen T, Plewig G, Anhalt GJ. Paraneoplastic pemphigus with clinical features of erosive lichen planus associated with Castleman’s tumor. Dermatology 1995; 190: 245–50. Lemon MA, Weston WL, Huff JC. Childhood paraneoplastic pemphigus associated with Castleman’s tumour. Br J Dermatol 1997; 136: 115–17. Anhalt GJ. Paraneoplastic pemphigus. Adv Dermatol 1997; 12: 77–96. Helou J, Allbritton J, Anhalt GJ. Accuracy of indirect immunofluorescence testing in the diagnosis of paraneoplastic pemphigus. J Am Acad Dermatol 1995; 32: 441–47. Liu AY, Valenzuela R, Helm TN, Camisa C, Melton AL, Bergfeld WF. Indirect immunofluorescence on rat bladder transitional epithelium: a test with high specificity for paraneoplastic pemphigus. J Am Acad Dermatol 1993; 28: 696–99. Campbell MJ, Zelenetz AD, Levy S, Levy S. Use of family specific leader region primers for PCR amplification of the human heavy chain variable region gene repertoire. Molec Immun 1992; 29: 193–203. Aubin J, Davi F, Nguyen-Salomon F, et al. Description of a novel FR1 IgH PCR strategy and its comparison with three other strategies for the detection of clonality in B cell malignancies. Leukemia 1995; 9: 471–79.
19 Yamada M, Hudson S, Tournay O, et al. Detection of minimal disease in hematopoietic malignancies of the B-cell lineage by using third-complementarity-determining region (CDR-III)-specific probes. Proc Natl Acad Sci USA 1989; 86: 5123–27. 20 Camisa C, Helm TN. Paraneoplastic pemphigus is a distinct neoplasia-induced autoimmune disease. Arch Dermatol 1993; 129: 883–86. 21 Berg WA, Fishman EK, Anhalt GJ. Retroperitoneal reticulum cell sarcoma: a cause of paraneoplastic pemphigus. South Med J 1993; 86: 215–17. 22 Turner M. Antibodies and their receptors. In: Roitt I, Brostoff J, Male D, eds. Immunology, 5th edn. London: Mosby Press, 1998: 73. 23 Robinson ND, Hashimoto T, Amagai M, Chan LS. The new pemphigus variants. J Am Acad Dermatol 1999; 40: 649–71. 24 Bowen GM, Peters NT, Fivenson DP, et al. Lichenoid dermatitis in paraneoplastic pemphigus: a pathogenic trigger of epitope spreading? Arch Dermatol 2000; 136: 652–56. 25 Chan LS. Epitope spreading in paraneoplastic pemphigus: autoimmune induction in antibody-mediated blistering skin diseases. Arch Dermatol 2000; 136: 663–64. 26 Bowne WB, Lewis JJ, Filippa DA, et al. The management of unicentric and multicentric Castleman’s disease: a report of 16 cases and a review of the literature. Cancer 1999; 85: 706–17.
THE LANCET • Vol 363 • February 14, 2004 • www.thelancet.com
531
For personal use. Only reproduce with permission from The Lancet.