Laboratory diagnosis of paraneoplastic pemphigus


  • Funding sources None.
  • Conflicts of interest None declared.



Paraneoplastic pemphigus (PNP) is a multiorgan disease characterized by antibodies against plakins, desmogleins and the α2-macroglobulin-like-1 (A2ML1) protein, in association with an underlying neoplasm. Accurate diagnosis relies on the demonstration of these autoantibodies in serum.


To evaluate the value of different laboratory techniques in the serological diagnosis of PNP.


We performed immunoblotting, envoplakin (EP) enzyme-linked immunosorbent assay (ELISA), indirect immunofluorescence (IIF) on rat bladder, radioactive immunoprecipitation and a nonradioactive combined immunoprecipitation-immunoblot assay. Additional assays included BP180 ELISA and BP230 ELISA. We included the sera of 19 patients with PNP and 40 control subjects.


The sensitivities were 63% for anti-EP ELISA, 74% for rat bladder IIF, 89% for immunoblotting, 95% for radioactive immunoprecipitation and 100% for nonradioactive immunoprecipitation. Specificities ranged from 86% to 100%. The BP180 and BP230 ELISAs had low sensitivity and specificity for PNP. The combination of rat bladder IIF and immunoblot showed 100% sensitivity and specificity. The analysis of sequential PNP sera showed that antibody titres may decrease over time, possibly resulting in negative outcomes for EP ELISA and rat bladder IIF studies.


The detection of autoantibodies against EP and periplakin, or A2ML1 by immunoprecipitation is most sensitive for PNP. The combination of rat bladder IIF and immunoblotting is equally sensitive and highly specific, and represents an alternative valuable and relatively easy approach for the serological diagnosis of PNP.

Paraneoplastic pemphigus (PNP) is a severe autoimmune multiorgan disease, first described by Anhalt et al.[1] in 1990. It is characterized clinically by painful stomatitis, polymorphic cutaneous manifestations and sometimes also pulmonary involvement, in patients with underlying neoplasia.[2, 3] Mortality rates are high, with 1-, 2-, and 5- year survival rates of 49%, 41% and 38%, respectively, the main causes of death being infections and progression of neoplasia.[4] Histological changes are numerous, including intraepidermal acantholysis, vacuolar interface dermatitis and keratinocyte necrosis.[5, 6] Direct immunofluorescence (DIF) studies of skin or mucosa biopsies often show deposition of IgG and complement component C3 on the epithelial cell surface, sometimes accompanied by linear depositions along the epithelial basement membrane zone (BMZ).[6-8] The manifestations of PNP are therefore diverse, resembling pemphigus vulgaris (PV), erythema multiforme, Stevens–Johnson syndrome, toxic epidermal necrolysis (TEN) or a lichenoid dermatosis.[8-12] This is why diagnosis is often challenging.

PNP is also characterized by the presence of autoantibodies against multiple antigens, among which are the proteins of the plakin family, which are part of the intracellular plaque of desmosomes and/or hemidesmosomes.[1, 13-23] These include the 210-kDa envoplakin (EP), the 190-kDa periplakin (PP), the 230-kDa bullous pemphigoid antigen BP230, 250- and 210-kDa desmoplakins I and II (DSPI and DSPII), and the 500-kDa plectin. The previously described 170-kDa autoantigen has recently been identified as the protease inhibitor α2-macroglobulin-like-1 (A2ML1).[24, 25] Other antigens include the desmogleins (Dsg) 1 and 3.[16] Antibodies to plakophilin 3 and desmocollins 1–3 have also been reported.[26, 27] The demonstration of PNP specific antibodies is an important tool in diagnosing PNP. Several laboratory tests are available, including indirect immunofluorescence microscopy (IIF) studies on rat bladder sections. Rat bladder urothelium does not contain Dsg1 and Dsg3 but does contain DSP, EP and PP. A positive staining of the urothelium is therefore considered an indication of PNP.[28-30] In addition, immunoprecipitation (IP) and immunoblot (IB) are often used to demonstrate the characteristic PNP antibody response.[9, 14, 18] A relatively new test is an anti-EP enzyme-linked immunosorbent assay (ELISA) that has recently become commercially available.[31] Tests to characterize the anti-BMZ antibody response include a BP230 and BP180 ELISA.[32] Although these techniques have been individually validated, comparative studies on their usefulness in diagnosing PNP are scarce. As PNP is rare, with only approximately 450 cases reported so far,[33] and as it only comprises 3–5% of all pemphigus cases (personal estimation based on Dutch patients), most laboratories are not familiar with its diagnostics. In the present study we therefore compared the diagnostic value of the currently available immunoserological techniques.

Materials and methods

Patient selection

Nineteen patients with PNP were included, with a median age of 57 years (age range 29–86 years). The male to female ratio was 2 : 1. The diagnosis of PNP was made if patients fulfilled the following criteria, adapted from the revised criteria proposed by Anhalt in 2004 and Zimmerman et al. in 2010:[6, 8] (i) painful and persistent stomatitis, with or without a polymorphic skin eruption; (ii) the demonstration of plakin autoantibodies; (iii) the presence of an underlying neoplasm. As control, 24 patients with PV were included, one of whom had an underlying neoplasm (lung carcinoma) but no other PNP characteristics. Furthermore, three patients who had additional anti-DSP antibodies (a-DSP) (one pemphigus foliaceus, one PV and one bullous pemphigoid), and 13 patients with TEN were included as controls. The diagnosis of PV was based on the clinical and histological features of suprabasal acantholysis and the presence of anti-Dsg3 with or without anti-Dsg1 antibodies. The diagnosis of TEN was based on clinical and histological features. If DIF biopsies of PNP and PV were available these showed the typical pemphigus epidermal anticell surface (ACS) deposition pattern of IgG with or without C3 deposits. Most PNP patients had additional depositions of IgG or C3 along the BMZ. The characteristics of patients with PNP are summarized in Table 1.

Table 1. Characteristics of patients with paraneoplastic pemphigus
Patient no.Age (years)/sexNeoplasmDsg1 ELISA indexDsg3 ELISA indexDIF
  1. Dsg, desmoglein; CLL, chronic lymphocytic leukaemia; NHL, non-Hodgkin lymphoma; ELISA, enzyme-linked immunosorbent assay; DIF, direct immunofluorescence microscopy; ACS, anticell surface; BMZ, basement membrane zone; G+, positive for IgG deposits; C+, complement deposits; M+, IgM deposits; A+, IgA deposits; n.a., biopsies not available; −, negative.

161/mCLL98> 150n.a.n.a.
2?/?Castleman tumour2> 150n.a.n.a.
354/fB-cell NHL22G+ C+G+ C+
453/mNHL57145G+ C+M+
529/fCastleman disease3158G+G+ M+ A+ C3c+
675/mNHL1> 150n.a.n.a.
756/fEpithelioid leiomyosarcoma> 150> 150n.a.n.a.
861/mMalignant fibriohistiocytoma242n.a.n.a.
964/mLow-grade B-cell NHL9150G+G+ A+ C+
1043/mFollicular NHL7114InconclusiveG+
1186/mLarynx carcinoma> 150136G+ C+
1270/mFollicular NHL2> 150G+ C+C+
1339/mB-cell NHL2142G+ C+C+
1469/mCentrocytic lymphoma46115n.a.n.a.
1549/fLow-grade NHL354G+ C+G+ C+
1856/mCarcinoid150G+ C+M+
1973/fFollicular NHL094G+ C+

Immunoblot, radioactive and nonradioactive immunoprecipitation analyses

Routine IB analysis was performed as described previously.[34] Radioactive IP (r-IP) analysis was performed using radiolabelled extracts from cultured differentiated human keratinocytes, as previously described.[24] For the nonradioactive IP (nr-IP) analysis, keratinocytes grown to near confluence were differentiated for 6–10 days to induce A2ML1 by adding 1 mmol L−1 CaCl2, 0·1 mmol L−1 isoproterenol and 0·4 μg mL−1 hydrocortisone to CnT keratinocyte medium (Cellntech, Bern, Switzerland). Cells were then extracted for 10 min in ice cold 1% (v/v) Triton-X100, 50 mmol L−1 MOPS, 150 mmol L−1 NaCl, 5 mmol L−1 EDTA with protease inhibitor Complete (Roche, Almere, the Netherlands). Debris was removed by centrifugation and the extract was stored at −80 °C. Patient serum (20 μL) was incubated for 1 h at room temperature with 60 μL GammaBind G Sepharose (GE Healthcare, Uppsala, Sweden) and washed over a 1 mol L−1 sucrose plus 150 mmol L−1 NaCl solution. One hundred microlitres keratinocyte extract was incubated with the beads for 1 h at room temperature, and washed over sucrose solution. Twenty microlitres Laemmli sample buffer was added to the beads and heated at 100 °C for 5 min. This was run on a 5% SDS-PAGE slab gel and then electrophoretically transferred to an Immobilon-P PVDF-membrane (Millipore, Billerica, MA, U.S.A.). The membrane was blocked with 2% (w/v) low-fat milk powder TBST for 1 h and incubated overnight with a cocktail of 1 : 500 diluted antibodies against DSP (DSPII.15; Abcam, Cambridge, U.K.), EP (CRENV-1; Abcam), PP (C20; Santa Cruz Biochemicals, CA, U.S.A) and a 1 : 250 diluted anti-A2ML1 antibody (B01P; Abnova, Taipei City, Taiwan). Subsequent incubations consisted of 1 : 2000 goat antirabbit IgG (Nordic Immunologic Laboratories; Eindhoven, the Netherlands) and 1 : 500 AffiniPure goat antimouse IgG (Jackson Immunoresearch, West Grove, PA, U.S.A), followed by 1 : 500 alkaline phosphatase conjugated AffiniPure rabbit antigoat IgG (Jackson Immunoresearch). Antibody binding was visualized by incubation with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium.

Enzyme-linked immunosorbent assay

ELISAs for the detection of anti-Dsg1, anti-Dsg3, the NC16A ectodomain of BP180, BP230 (all MBL, Nagoya, Japan) and the N-terminal domain of EP (EUROIMMUN, Lübeck, Germany) IgG autoantibodies were performed according to the manufacturers' instructions.

Indirect immunofluorescence microscopy

IIF studies of patient sera was performed on rat bladder (1 : 10) as previously described.[35] As secondary antibody we used FITC-conjugated goat antihuman IgG (Protos Immunoresearch, Burlingame, CA, U.S.A.). Sections were viewed under a Leica DMRA fluorescence microscope and images were recorded with a Leica DFC350 FX digital camera (Leica, Wetzlar, Germany).

Statistical analysis

The sensitivities and specificities of the different immunoserological techniques to diagnose PNP were calculated as follows:

Sensitivity (%) = True positive/(True positive + False negative) × 100.

Specificity (%) = True negative/(True negative + False positive) × 100.


Immunoprecipitation using radioactively labelled keratinocyte extracts

When tested by r-IP, 14 PNP sera (74%) precipitated EP and PP and 15 PNP sera (79%) precipitated A2ML1 (Fig. 1a). Overall, 18 PNP sera (95%), but none of eight control sera, precipitated either EP and PP or A2ML1. Thirteen (68%) PNP sera precipitated DSP and four sera of patients with PNP (21%) precipitated BP230. One serum (#15) only precipitated DSP, with two bands of 250 kDa and 210 kDa representing DSP1 and DSPII, respectively, and could not be distinguished from the three a-DSP controls that also precipitated DSP. This serum however was positive on IB and by nr-IP-IB for antibodies against EP and PP.

Figure 1.

Autoantibodies to plakins and anti-α2-macroglobulin-like-1 (A2ML1) detected by r-IP, IP-IB and IB. (a) Radioactive immunoprecipitation (r-IP) results for 19 paraneoplastic pemphigus (PNP) sera. (b) Nonradioactive immunoprecipitation-immunoblot (IP-IB) results for 15 PNP sera. For patient #18 the A2ML1 band was weakly visible by eye, but barely visible after scanning to a digital image. We therefore used software to further enhance this image. (c) Immunoblot (IB) results for 19 PNP sera. (d) The prevalence of the different antiplakin and anti-A2ML1 antibodies in the PNP and control groups, as measured by r-IP, IP-IB and IB. DSP, desmoplakin; EP, envoplakin; PP, periplakin.

Nonradioactive immunoprecipitation-immunoblot analysis

To evaluate a nonradioactive combined IP-IB technique for PNP, we tested sera of 15 PNP (Fig. 1b), 12 PV, 13 TEN, and the three a-DSP control subjects. Four PNP sera (#1, 2, 3 and 13) could not be tested due to the limited amount of serum. Thirteen PNP sera (87%) precipitated EP and PP, seven (47%) of these 13 also precipitated A2ML1, and three (20%) precipitated DSP. Two sera (#18 and 19) tested positive for A2ML1 by nr-IP and IP-IB, but tested negative for EP and PP by all techniques. Overall, all 15 PNP sera tested positive for either anti-EP and anti-PP antibodies or anti-A2ML1 antibodies. None of the 28 controls precipitated EP, but two PV sera precipitated PP. Surprisingly, with four of the 13 TEN sera we observed a faint band at the position of A2ML1. The intensities of these bands were considerably weaker than those obtained with the PNP sera. Two of these four also weakly precipitated PP. DSP was precipitated by one PV and by one of the three a-DSP sera.


By IB, using protein extracts from undifferentiated keratinocytes, 17 of the 19 (89%) PNP sera bound to EP and PP (Fig. 1c). Binding to DSPI and DSPII was seen for, respectively, nine (47%) and seven (37%) PNP sera. Six sera (32%) bound to BP230 and one (5%) bound to plectin. In the control group, one of 24 PV sera recognized BP230 (4%), while all three a-DSP sera bound DSP. The 12 TEN sera all tested negative. An overview of the r-IP, IP-IB and IB results for the PNP and control group is shown in Figure 1d. The presence of EP and PP, or A2ML1 autoantibodies was valued as a positive outcome and used in the calculation of the sensitivity and specificity of each test (see Table 3).

Envoplakin enzyme-linked immunosorbent assay

Twelve PNP sera (63%) tested positive by EP ELISA (Fig. 2a). One of the 24 PV controls also tested positive, but no evidence for EP antibodies was found by either IB or IP-IB and rat bladder IIF was negative. All TEN sera tested negative. Consecutive sera were available from five patients with PNP. EP ELISA of these sera showed that anti-EP titres decreased during immunosuppressive therapy, and for two sera (#6 and 7) the EP ELISA index dropped below the cut-off of the ELISA (Fig. 2b). Nevertheless EP antibodies could still be demonstrated by IB, r-IP and IP-IB.

Figure 2.

Envoplakin autoantibody titres as measured by enzyme-linked immunosorbent assay (ELISA). (a) Twelve patients with paraneoplastic pemphigus (PNP) and one control patient had titres above the ELISA detection limit. (b) Titres decreased over time. Two of five tested PNP patients had titres decreasing below the ELISA detection limit.

Indirect immunofluorescence on rat bladder

Rat bladder IIF was considered positive if sera bound to the urothelial cell surface (Fig. 3a) with or without binding to the BMZ. Some PNP sera bound in a smudgy urothelial pattern, but as this was also seen for some controls we considered this a negative test. Two of three a-DSP sera gave subtle urothelial cell surface staining of luminal cells (Fig. 3b), which was different from the PNP pattern and was also interpreted as negative. Sera from 14 of the 19 patients with PNP (74%) tested positive and all PV, TEN, and a-DSP sera tested negative. The two sera (#18 and #19) that tested negative for PNP-specific antibodies by IB and EP ELISA, but positive for anti-A2ML1 antibodies by r-IP and IP-IB, tested positive on rat bladder. For one patient (#6) from whom we had consecutive sera, the test became negative during treatment.

Figure 3.

Indirect immunofluorescence on rat bladder urothelium. (a) Positive urothelial cell surface staining pattern typical for paraneoplastic pemphigus (PNP). (b) Antidesmoplakin control serum gives a subtle luminal staining pattern (white arrows) of rat bladder urothelium. Scale bars represent 20 μm.

Additional assays for basement membrane zone antigens

DIF studies of PNP patients' skin specimens sometimes demonstrate the presence of BMZ IgG deposition in addition to epidermal ACS deposition. Therefore, we tested all PNP and PV sera by BP230 and NC16A ELISA. Two PNP and three PV sera tested positive in the B230 ELISA, and five PNP sera and one PV serum tested positive in the NC16A ELISA. No clear correlation between DIF IgG BMZ depositions and ELISA was observed.

All data are summarized in Tables 2 and 3. Based on these data we formulated a diagnostic algorithm for PNP, as illustrated in Figure 4.

Table 2. Overview of test outcomes for paraneoplastic pemphigus
  1. Pat. #, patient number; Rb-IIF, rat bladder indirect immunofluorescence; ELISA, enzyme-linked immunosorbent assay; r-IP, radioactive immunoprecitation; IP-IB, immunoprecipitation-immunoblot assay; IB, immunoblot; DSP, desmoplakin; EP, envoplakin; PP, periplakin; A2ML1, alpha-2-macroglobulin-like-1; Nd, test not done; −, negative test outcome; +, positive test outcome.

2+  EPPPA2ML1Nd     EPPP 
3  EPPPA2ML1Nd     EPPP 
5+++    A2ML1 EPPPA2ML1  EPPP 
18+DSP   A2ML1  PPA2ML1    Plectin
19++DSP   A2ML1   A2ML1DSPBP230   
Table 3. Sensitivity and specificity of different laboratory techniques for paraneoplastic pemphigus (PNP)
TechniqueNo. of patients includedPositive outcomeSensitivity (%)Specificity (%)
  1. r-IP, radioactive immunoprecitation; IP-IB, immunoprecipitation-immunoblot; IB, immunoblot; IIF, indirect immunofluorescence; ELISA, enzyme-linked immunosorbent assay; PV, pemphigus vulgaris; a-DPK, control patients with anti-desmoplakin antibodies; TEN, toxic epidermal necrolysis; EP, envoplakin; PP, periplakin; A2ML1, α2-macroglobulin like-1.


19 PNP

8 Controls

Reactivity for EP and PP, or A2ML195100

15 PNP

28 Controls

(12 PV, 3 a-DPK, 13 TEN)

Reactivity for EP and PP, or A2ML110086

19 PNP

39 Controls

(24 PV, 3 a-DPK, 12 TEN)

Reactivity for EP and PP89100
IIF rat bladder

19 PNP

40 controls

(24 PV, 3 a-DPK, 13 TEN)

Positive urothelial cell surface staining74100
Envoplakin ELISA

19 PNP

40 Controls

(24 PV, 3 a-DPK, 13 TEN)

Figure 4.

Diagnostic algorithm for paraneoplastic pemphigus (PNP). A positive IIF-rat bladder is defined as positive urothelial cell surface staining. A positive immunoblot outcome is defined as serum reactivity to both envoplakin (EP) and periplakin (PP). IIF-rat bladder, indirect immunofluorescence on rat bladder; TEN, toxic epidermal necrolysis; A2ML1, α2-macroglobulin like-1.


In this study we compared different assays for the serological diagnosis of PNP. Historically, the IP assay using radioactively labelled keratinocyte extracts was the first technique used to identify PNP autoantibodies.[1] We show that the sensitivity of this technique is superior to IB, IIF on rat bladder, and EP ELISA. However, because of its radioactivity, this technique is not widely available. We therefore developed a nr-IP-IB assay. Although the comparison is limited by the smaller number of PNP patients we used for the nr-IP assay, our results suggest that this assay is slightly more sensitive than the r-IP and similarly specific. Another advantage of our nr-IP assay is that we confirm the identity of the autoantigens not only based on size but by using specific antiplakin and A2ML1 antibodies to visualize the blotted precipitates. Therefore, this assay adds to the other nr-IP assays that have been developed in previous PNP studies.[18] The sensitivity of IB analysis closely follows that of r-IP and IP-IB studies. By r-IP we missed one and by IB two of our 19 patients with PNP. The patients that were missed by IB had antibodies to A2ML1, in addition to other non-PNP-specific antibodies. The reason that these cases were missed is that our routine diagnostic blots are prepared with low-calcium grown cells and reduced substrate, while A2ML1 is produced by cells that are grown for several days at high calcium and is only recognized by autoantibodies under nonreducing conditions.[24] The serum that was false negative in the r-IP possibly had too low anti-EP and anti-PP titres.

The EP ELISA had a lower sensitivity and specificity when compared with IP and IB studies. Our results from sequential PNP sera further show that anti-EP antibody titres may decrease over time, sometimes resulting in a negative ELISA as titres drop below the cut-off. This decrease may be due to immunosuppressive therapies, or may be part of the natural disease course. Therefore, the possibility of seroreversion should be considered when interpreting EP ELISA results from patients with suspected PNP from whom serum is taken later in the course of disease.

Serum IF on rat bladder substrate is an accurate test to differentiate PNP from PV, as rat bladder epithelium does not contain Dsgs, but does express plakins.[28] In the present study rat bladder IIF was more sensitive than EP ELISA. Notably, two patients that tested negative for anti-EP and anti-PP antibodies by IB and ELISA tested positive by rat bladder IIF. The sera of these patients contained anti-DSP and anti-A2ML1 antibodies as shown by r-IP. Rat bladder does not contain an A2ML1 orthologue,[36] and anti-DSP antibodies give a different staining pattern than that seen for these two sera. Therefore, this positive urothelial staining suggests that more – still unknown – PNP antigens are present in rat bladder. In our group the combination of IB and rat bladder IIF led to 100% sensitivity and specificity. Both techniques are faster, less labour intensive, and require less patient's serum than IP assays. However, rare PNP cases with only anti-A2ML1 antibodies might be missed by the combination of IB and rat bladder IIF.[24] Therefore, for patients clinically suspect for PNP, but with negative IB and rat bladder IIF, IP might be useful. The detection of solitary anti-A2ML1 antibodies by IP may not always discriminate between PNP and TEN because TEN sera may also produce very weak bands at the position of A2ML1. However, our findings do suggest that anti–A2ML1 antibody titres are much lower in TEN than in PNP. Studies aimed at developing an A2ML1 ELISA might therefore prove fruitful to further quantify these differences. Similar to the EP ELISA, the use of medication led to a negative rat bladder test for one of the patients, indicating that concurrent use of medication might also reduce the sensitivity of rat bladder IIF.

Our results show that the antibody response in PNP is variable, as it can be directed to the complete spectrum of PNP antigens but can also be limited to part of it. We confirm that the combined presence of anti-EP and anti-PP antibodies is most specific and sensitive for PNP,[8, 23] and that solitary anti-A2ML1, anti-PP, anti-DSP and anti-BMZ antibodies are not.[37, 38] We therefore propose that the diagnosis of PNP should not be solely based on clinical, histological, DIF or nonspecific IIF findings, as suggested by various studies,[4, 9] but that the demonstration of antibodies to EP and PP, or a positive rat bladder IIF is necessary.

In conclusion, our findings indicate that IP studies are the most sensitive in detecting PNP-specific autoantibodies. Rat bladder IIF, although less sensitive, is a relatively simple technique, which may be of complementary value. In settings where IP studies are not available, the combination of IB and rat bladder IIF should be used as the first serological analyses for confirming the diagnosis of PNP. Finally, when evaluating laboratory results, one should consider that, possibly due to therapy or the natural course of the disease, antibody titres may decrease over time. Further studies should determine if specific antibody profiles or IIF staining patterns are related to specific PNP phenotypes and might predict disease outcome. This would provide more insight in PNP pathogenesis and treatment.