SEARCH

SEARCH BY CITATION

Keywords:

  • protein microarray;
  • inflammatory bowel disease;
  • antibodies;
  • family with sequence similarity 84 member A

Abstract

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Background:

Patients with inflammatory bowel disease (IBD) display immunoreactivity to self-antigens and microbial antigens. We used a protein microarray approach to identify novel autoantigens in IBD.

Methods:

ProtoArray Human Protein Microarray v4.0 containing 8268 human proteins from Invitrogen (La Jolla, CA) was used.

Results:

Twenty-five IBD patients and five healthy controls were screened for candidate autoantigens. For 256 antigens, IBD patients had a higher seroreactivity than controls. Twenty antigens were selected for further evaluation in a larger cohort (60 ulcerative colitis [UC] patients, 60 Crohn's disease [CD] patients, 60 healthy controls, and 60 gastrointestinal-diseased controls) by means of a customized protein microarray. Out of these 20 antigens, one antigen, family with sequence similarity 84 member A (FAM84A), was identified as a target antigen in IBD. Antibodies to FAM84A were significantly more prevalent in IBD patients (19%) than in gastrointestinal-diseased controls (1.7%) (P = 0.0008) and healthy controls (5%) (P = 0.01). Anti-FAM84A antibodies were found in 26.6% of UC patients and in 11.7% of CD patients. FAM84A was confirmed as target antigen in IBD by means of Western blotting in a large independent cohort (100 UC patients, 106 CD patients, 102 healthy controls, and 100 gastrointestinal-diseased controls). Antibodies to FAM84A were significantly more prevalent in IBD patients (20%) than in gastrointestinal-diseased controls (5%) (P = 0.0004) and healthy controls (0%) (P < 0.0001). Anti-FAM84A antibodies were found in 18% of UC patients and in 22% of CD patients.

Conclusions:

We identified FAM84A as a novel autoantigen in IBD. (Inflamm Bowel Dis 2011;)

Inflammatory bowel disease (IBD) is a group of gastrointestinal-disorders, with Crohn's disease (CD) and ulcerative colitis (UC) as the most prominent phenotypes. These diseases are characterized by a chronic inflammation of the gastrointestinal tract.1 Although UC and CD have distinct disease characteristics, in about 10% of patients with colonic disease the differential diagnosis is difficult and patients are (temporarily) diagnosed with indeterminate-type colitis.2

The main hypothesis regarding the pathogenesis of IBD is that this disease is caused by an aberrant immune response directed against the normal intestinal flora in a genetically susceptible host.3 In immune-mediated disorders, antibodies can be important biomarkers for diagnosis and prognosis in the clinical setting. Several antibodies to autoantigens and to microbial antigens have been described in IBD.4 Perinuclear antineutrophil cytoplasmatic antibodies (or pANCA) have been associated with UC patients (prevalence: 60%–80% of UC, 5%–25% of CD, 6% of gastrointestinal-diseased controls, 4% of healthy controls), whereas anti-Saccharomycescerevisiae antibodies (ASCA) and antipancreatic antibodies have been associated with CD patients (prevalence: less than 10% of UC, 50%–80% of CD, less than 5% of gastrointestinal-diseased controls, less than 5% of healthy controls).5–11 More recently described microbial antigens are I2, CBir1, and OmpC. Antibodies to these antigens have been associated with CD patients, but have also been found in a substantial number of gastrointestinal-diseased controls.4 In this study, ProtoArray Human Protein Microarrays v4.0 from Invitrogen (La Jolla, CA), which contain 8268 human proteins immobilized on hydrophobic surfaces, were used to screen for novel autoantibodies in IBD.

SUBJECTS AND METHODS

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Subjects

A first cohort of patients and controls consisted of 10 UC patients, 15 CD patients, and 5 healthy controls. The UC patients were selected based on pANCA-positivity (titer >1/80) (female/male [F/M] ratio: 4/6; mean age 47.3 years). The CD patients all suffered from ileocolitis; 10 of them were selected based on ASCA-positivity, and 5 were selected based on pancreatic antibody-positivity. (F/M ratio: 9/6; mean age 45.5 years). Five healthy controls were negative for pANCA and ASCA (F/M ratio: 3/2; mean age 35.8 years). The patient characteristics are given in the Supplemental Data, Table 1.

Table 1. Protoarray Human Protein Microarrays v4.0 Analysis for Autoantibodies in UC, CD, and Healthy Controls
  UCCDIBDHC
Protein n=10n=15n=25n=5
  • Results for a selection of 20 proteins.

  • A protein microarray containing 8268 human proteins was used to screen for autoantibodies in UC patients, CD patients, and healthy controls (HC). The number of antibody-positive subjects per group is shown for a selection of 20 proteins. Prevalence of the antibodies in patients was compared to the prevalence of antibodies in controls by means of two statistical software packages (prospector and proCAT, see Materials and Methods).

  • a

    Significant difference between patients and controls by means of prospector software.

  • b

    Significant difference between patients and controls by means of proCAT software.

ABRActive BCR-related gene, transcript variant 1310a131
CDC2Cell division cycle2, transcript variant 108a8a1
DDX19BDEAD box polypeptide 19B, transcript variant 105ab5ab0
FAM21CFamily with sequence similarity 21, member C71017a0
FAM84AFamily with sequence similarity 84, member A413a17a1
HIST3H2BBHistone cluster 3, H2bb2a461
LIMCH1LIM and calponin homology domains 15b914a0
MAP2K6Mitogen-activated protein kinase kinase 6101424a1
MAPK8IP2Mitogen-activated protein kinase 8 interacting protein 24a041
NECAP1NECAP endocytosis associated 131013a0
PABPC3Poly(A) binding protein, cytoplasmic 391019a1
PAK6p21 (CDKN1A)-activated kinase 65a8b13ab0
PDHPyruvate dehydrogenase711a18a0
PHLDA1Plekstrin homology-like domain, family A, member 17815a0
RBM8ARNA binding motif protein 8A06ab6ab0
STK40Serine/threonine kinase 4022ab4a2
TOM1Target of Myb109a9a0
TTKTTK protein kinase6a7130
TUBG1Tubulin gamma 106a60
IOH14840Homo sapiens cDNA clone MGC:271524b9b13ab2

A second cohort of patients and controls consisted of 60 UC patients, 60 CD patients, 60 healthy controls, and 60 gastrointestinal-diseased controls. The UC patients were pANCA-positive (F/M ratio: 34/26; mean age 45.0 years). The CD patients were ASCA-positive (F/M ratio: 37/23; mean age 47.5 years). The healthy controls were negative for pANCA and ASCA (F/M ratio: 32/28; mean age 25.5 years). The gastrointestinal-diseased controls included patients suffering from different gastrointestinal inflammatory diseases such as diverticulitis, infectious gastroenteritis (campylobacter, salmonella, and clostridium), culture-negative gastroenteritis, ischemic colitis, periappendicular abscess, and irritable bowel syndrome (F/M ratio: 25/35; mean age 62.0 years). These gastrointestinal-diseased controls were negative for pANCA and ASCA.

A third cohort of patients and controls consisted of nonselected but well-defined IBD patients (106 CD patients (F/M ratio: 44/62; mean age 42.2 years) and 100 UC patients (F/M ratio: 53/47; mean age 43.4 years)), 100 gastrointestinal controls (F/M ratio: 50/50; mean age 56.5 years), and 102 healthy controls (F/M ratio: 49/53; mean age 43.8 years). The gastrointestinal controls included patients suffering from different gastrointestinal inflammatory diseases.

The diagnosis of UC or CD was based on accepted clinical and endoscopic criteria.12 Serum was stored at −20°C until use. Institutional Review Board approval was obtained from the University Hospitals Leuven, Belgium, and all individuals signed an informed consent.

Detection of pANCA, ASCA, and Pancreatic Antibodies

pANCA were determined by indirect immunofluorescence using substrate and reagents from Inova Diagnostics (San Diego, CA).13 ASCA were determined by enzyme-linked immunosorbent assay (ELISA) from Medipan Diagnostica (Berlin, Germany). Antipancreatic antibodies were determined by indirect immunofluorescence as previously described.11 Antimitochondrial antibodies were determined by indirect immunofluorescence using mouse liver, kidney, and stomach slides from the Binding Site (Birmingham, UK). Sera were diluted 1/20.

Protein Microarray to Screen for Autoantibodies

The ProtoArray Human Protein Microarray v4.0 containing 8268 human proteins, reagents and other equipment were from Invitrogen. Assays, including appropriate controls, were performed according to the instructions of the manufacturer. In short, protein microarray slides were blocked with phosphate-buffered saline (PBS) containing 1% bovine serum albumin (BSA) and 0.1% Tween-20 before incubation with serum (1:500) and Alexa Fluor 647-conjugated antihuman IgG (1.0 μg/ml buffer). The arrays were dried and scanned using an Axon Genepix 400B fluorescent microarray scanner. Genepix 6.0 software was used to align the scanned image to the template and to determine the pixel intensities for each spot on the array. The reported pixel intensity was calculated as the average of duplicate signals after background subtraction.

Two types of software were used for statistical analysis of the microarray data. The first software tool was the Prospector software (Invitrogen), which is based on M-statistics. This software performs background subtraction, normalization of the signals, and analysis of the differences between two patients or two groups of patients. When comparing two groups, a cutoff for positivity is calculated for each protein using M-statistics. For both groups the proportion of subjects with an immune response above the cutoff value is counted and a P-value representing the significance of the difference between both groups is calculated. A second statistical software was the protein chip analysis tool (proCAT), developed at Yale University. This software corrects for background using neighborhood background correction, it identifies significant signals, filters nonspecific spots by comparison with the negative control assay, and normalizes the resulting signal to protein amount. This software analyzes each patient separately and reports the significant signals per patient.14

Customized Protein Microarray

A customized protein microarray containing 20 selected proteins was constructed by Invitrogen in order to confirm the data obtained with the screening microarray. The experiments were performed as described above. A Mann–Whitney test was used to evaluate differences in signals, which represent antibody-levels, between patients and controls. The cutoff was set at a specificity of 95% for the healthy controls. Statistical evaluation for differences in prevalence of antibodies (proportions) were calculated by means of Fisher's exact test. P-values <0.05 were considered significant. Statistical analyses were performed using SPSS 15.0 statistical software package (Chicago, IL).

Western Blotting

All materials for gel electrophoresis and Western blotting were purchased from GE Healthcare (Milwaukee, WI). Family with sequence similarity 84, member A (FAM84A) was from Abnova (Taiwan). Goat-antihuman IgG and peroxidase-conjugated rabbit-antigoat IgG were purchased from Nordic (The Netherlands) and Dako (Glostrup, Denmark), respectively.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS PAGE), Western blotting, and immunodetection were performed as described earlier.15 In short, FAM84A was diluted in sample buffer, heated, and loaded onto the gel. Simultaneously, a molecular weight marker was loaded. After separation, the proteins were transferred to a PVDF membrane by semidry blotting. Subsequently, membranes were blocked with 5% (wt/vol) BSA in Tris-saline buffer (TSB) (10 mM Tris, 150 mM NaCl, and 0.1% [vol/vol] Triton-X 100, pH 7.6) followed by overnight incubation with serum (diluted 1/500 in TSB). After an intermediate wash step, Western blots were incubated with goat-antihuman IgG (1/2500 in TSB) and peroxidase-conjugated rabbit-antigoat IgG (1/5000 in TSB). For detection of antibody-antigen binding, 0.7 mM 3,3′-diamino-benzidine-tetrahydrochloride dehydrate containing 0.01% (vol/vol) H2O2 was added as substrate. Each blot was scanned and loaded into ImageJ (developed at the National Institutes of Health and available at http://rsb.info.nih.gov/ij/). For each band the area under the curve was calculated using the “Gel Analysis” function. We included a positive and negative control serum sample in each run. The coefficient of variation of the positive control was 5% over 18 runs.

Statistical Analysis

Statistical analysis for the microarray studies is described above. Results of the Western blotting and ELISA experiments were evaluated by means of Mann–Whitney tests for differences in the central location (median signal intensities) between two independent groups. Fisher's exact analysis was performed to evaluate differences in proportion of antibody-positive subjects per group.

RESULTS

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Profiling Autoimmune Signatures in IBD Patients

Sera from 10 UC patients, 15 CD patients, and 5 healthy controls were screened for the presence of autoantibodies using ProtoArray Human Protein Microarrays v4.0.

M-statistics (Prospector software) was used to reveal differences between groups of patients. Antibodies to 224 antigens were significantly more prevalent in IBD patients than in controls. Antibodies to 41 antigens were more prevalent in UC and CD patients than in controls. Antibodies to 110 antigens were more prevalent in UC patients than in controls and antibodies to 73 antigens were more prevalent in CD patients than in controls. Antibodies to 101 antigens were significantly more prevalent in controls than in IBD patients.

A second analysis of the results was done using the proCAT software. There were 14 antigens to which more than 30% of the UC patients and no or only one of the five healthy controls reacted. There were 18 antigens to which more than 30% of the CD patients and no or one of the five healthy controls reacted.

Validation of Novel Autoantigens by Customized Protein Microarray

The screening experiment resulted in 224 antigens to which antibodies were significantly more prevalent in IBD patients than in controls (Prospector software) and 32 antigens to which more than 30% of the UC/CD patients and less than 20% of the controls reacted (PROCAT software). Out of these 256 antigens revealed by the screening experiment, we selected 20 antigens for further analysis (Table 1). The antigens were selected based on 1) a highly statistically significant difference of prevalence of antibodies in patients to prevalence of antibodies in controls, and 2) on a high ratio (>3) of antibody levels in patients to antibody levels in controls (>3). A customized protein microarray that contained the 20 selected antigens in duplicate was constructed and used to screen 240 subjects for the presence of antibodies.

One of the 20 antigens gave a saturated signal in all patients and was not evaluated further. The median fluorescence signal intensities (representing antibody-levels) for the 19 other antigens are summarized in Table 2. Mann–Whitney statistics was used to evaluate differences in fluorescence signal intensity between the groups (UC, CD, gastrointestinal-diseased controls, and healthy controls).

Table 2. Results of the Custom-made Protein Microarray
     P-value (Mann-Whitney Analysis,Corrected for Multiple Comparisons)
ProteinUC Median (IQR)CD Median (IQR)HC Median (IQR)DC Median (IQR)UC vs. CDUC vs. HCUC vs. DCCD vs. HCCD vs. DC
  • A custom-made protein microarray containing 20 proteins was used to screen for autoantibodies in UC patients, CD patients, healthy controls (HC), and gastrointestinal diseased controls (DC). Data represent the median and interquartile range of the signal intensities. Mann-Whitney analysis was used to evaluate differences between the groups.

  • a

    P-values that remain significant after Bonferroni correction for testing 20 variables.

ABR5284 (3406.3-8525.7)3786.2 (2079-6041.7)3845 (2413.5-7170.6)4798.5 (2668-7616)n.sn.s.n.s.n.s.n.s.
CDC22025.5 (1292.5-2956.5)1735 (1334-2476.2)1686.5 (1158.3-2043.6)2371.5 (1563.7-3777.2)n.s.n.sn.s.n.s.n.s
DDX19B3462 (1800.3-4587.8)2366 (1668.2-3543)2743.2 (1789.2-3674.2)2897.2 (1468.1-4066)n.s.n.s.n.s.n.s.n.s.
FAM21C4405 (3271.5-7726.7)3602.2 (2289.1-6211.7)4065.2 (2934.5-7744.8)5195.5 (2699.5-7482.6)n.s.n.s.n.s.n.s.n.s.
FAM84A21051.7 (14388.1-39912.3)14056 (6760.7-23365.2)15269.2 (8435.1-233394.6)13545.7 (5876.5-19577.8)n.sn.s0.0012an.s.n.s.
HIST3H2BB3300.7 (2186.3-4863.6)3052 (2006.8-4577.3)2763.2 (1974.5-3924)4136.5 (1947.6-5783.8)n.s.n.s.n.s.n.s.n.s.
LIMCH15456.7 (3124.5-8729.8)4795 (3188.1-7160.8)4862.5 (2975.2-7262.5)6125.7 (2832.2-8239.6)n.s.n.s.n.s.n.s.n.s.
MAP2K61344 (897-1956.2)1105 (899.1-1429.3)1106 (879.5-1652)1395.2 (952.3-1773)n.s.n.s.n.s.n.s.n.s
MAPK8IP25610.2 (3955.8-7738)5380.7 (3745.6-7050.2)4711.2 (3457.7-6269.5)5644.5 (3318.1-8717.7)n.s.n.s.n.s.n.s.n.s.
NECAP17406.2 (4628.6-11717.3)5263.5 (2866.1-8988.8)5154.7 (2839.2-8998.6)5882 (3313.3-9000.2)n.sn.sn.s.n.s.n.s.
PABPC32204 (1341.8-2925.3)1790 (1353.3-2574.8)1846 (1233.2-2588)2134.5 (1328.8-2782.7)n.s.n.s.n.s.n.s.n.s.
PAK62505.5 (1591.1-4431)2273 (1549.7-3058.5)2472 (1591-3508.2)3237 (2084.5-4553.7)n.s.n.s.n.s.n.s.n.s
PHLDA19993.5 (5042.12-17595.6)7407.5 (3179.2-16795)4749.5 (2532.2-10630.3)9046.2 (4083.8-29802.3)n.s.0.02an.s.n.s.n.s.
PDH2800 (991.6-5639.2)1801.5 (922.5-3325)1095.5 (597.3-2334.7)1979.5 (890.5-4243.2)n.s.0.02an.s.n.sn.s.
RBM8A4521 (2784.3-6605.8)3754.7 (2452.8-5409.3)3887.5 (2422.5-5339)4609.7 (2228.6-6745.8)n.s.n.s.n.s.n.s.n.s.
STK401362.7 (1023.6-1720.5)1271 (1005.5-1648.7)1275.2 (952.3-1514.8)1387.5 (924.1-1708.7)n.s.n.s.n.s.n.s.n.s.
TOM15421.7 (3394.5-8610.7)4607.5 (2728.1-6340.5)4741.7 (3022.1-6914.5)5307.25 (2731.2-7939.5)n.sn.s.n.s.n.s.n.s.
TTK5392.7 (2193.5-7724.2)3495.2 (2198-4767.8)3915.5 (1916.6-6825.2)4900.5 (2391.8-7241.3)n.sn.s.n.s.n.s.n.s
TUBG13031.7 (2076.3-4102.8)2690.5 (1944-3624.1)3021 (1952.7-3639.1)2913.7 (1721.7-4341.8)n.s.n.s.n.s.n.s.n.s.

FAM84A was the only antigen for which there was a statistically significant difference between patients and gastrointestinal-diseased controls. Therefore, we concentrated on FAM84A. The fluorescence signal intensities are shown in Figure 1. A cutoff was set at a specificity of 95% in healthy individuals. The number of positive patients per group was calculated and compared by means of Fisher's exact test. Antibodies to FAM84A were significantly more prevalent in IBD patients (19%) than in healthy controls (5%) (P = 0.01) and than in gastrointestinal-diseased controls (1.7%) (P = 0.0008). Anti-FAM84A antibodies were found in 26.6% of UC patients and in 11.7% of CD patients. Anti-FAM84A antibodies were not associated with age at presentation, disease duration, disease location, smoking habits, colectomy (UC), disease behavior (CD), perianal-disease (CD), surgery (CD), or ASCA titer (CD) (Tables 3, 4).

thumbnail image

Figure 1. Results of the protein microarray experiments for FAM84A antibodies. The intensity of the microarray signal, expressed in relative fluorescence units (RFU), is shown for all patients (UC and CD) and healthy (HC) and gastrointestinal-diseased controls (GICO).

Download figure to PowerPoint

Table 3. Anti-FAM84A Antibodies Measured by Microarray and Clinical Characteristics of UC Patients
 FAM84A Antibody PositiveFAM84A Antibody NegativeTotal
  • a

    Information on smoking habits was known in 48 subjects.

  • Antibodies to FAM84 were measured using ProtoArray Human Protein Microarrayv4.0. The table summarizes the clinical information on UC patients according to the presence or absence of antibodies to FAM84A.

(n)164460
Male/female9/717/2726/34
Age (years) (mean ± SD)41.5 ± 14.546.2 ± 13.245.0 ± 13.6
Age at presentation (mean ± SD)28.5 ± 11.531.4 ± 13.930.6 ± 13.3
Disease duration (mean ± SD)17.0 ± 6.3413.3 ± 7.014.3 ± 7.0
Disease location:
Proctitis  0
 Left-sided colitis62228
 Pancolitis (extended)102232
 Colectomy31316
Smoking habits (n)a
 Never92231
 Yes41317
Table 4. Anti-FAM84A Antibodies Measured by Microarray and Clinical Characteristics of CD Patients
 FAM84A Antibody PositiveFAM84 Antibody NegativeTotal
  • a

    Information on smoking habits was known in 48 subjects.

  • Antibodies to FAM84 were measured using ProtoArray Human Protein Microarrayv4.0. The table summarizes the clinical information on CD patients according to the presence or absence of antibodies to FAM84.

  • Location CD: L1= ileum, L2 =colon, L3 = ileum + colon, L4 = upper gastrointestinal.

  • Behavior CD: B1= nonstricturing and nonpenetrating, B2 = structuring, B3 = penetrating.

(n)75360
Male/female3/420/3322/37
Age (years) (mean ± SD)55.6 ± 10.546.8 ± 10.847.8 ± 11.0
Age at presentation (mean ± SD)28.1 ± 6.024.5 ± 9.524.8 ± 9.2
Duration (mean ± SD)27.5 ± 11.221.6 ± 7.222.3 ± 7.9
Disease location
 L1  0
 L2  0
 L375360
 L4  0
Disease behavior
 B101212
 B221214
 B352934
Perianal disease42933
Surgery63440
Smoking habits (n)a
 Never43135
 Yes32225
ASCA titer101.3 ± 35.6122.7 ± 42.0120.2 ± 41.6

Validation of FAM84A as a Novel Autoantigen

In order to confirm the data obtained with the customized protein microarray, we screened 100 UC patients, 106 CD patients, 102 healthy controls, and 100 gastrointestinal-diseased controls for antibodies to FAM84A by Western blotting. The signal intensities are shown in Figure 2. At a cutoff value that corresponded to a specificity of 95% in gastrointestinal-diseased controls, antibodies to FAM84A were found in 18% of UC patients, 22.6% of CD patients, and 0% of the healthy controls. At a cutoff value that corresponded to a specificity of 97% in gastrointestinal-diseased controls, antibodies to FAM84A were found in 11% of UC patients, 20% of CD patients, and 0% of the healthy controls. At both cutoffs, anti-FAM84A antibodies were significantly more prevalent in CD patients than in gastrointestinal-diseased controls (P < 0.0004) and in UC patients than in gastrointestinal-diseased controls (P < 0.05). The area under the curve of the receiver operating characteristics curve was 0.59 for using FAM84A for discriminating IBD from gastrointestinal-diseased controls and 0.86 for discriminating IBD from healthy controls. The area under the curve was 0.55 for discriminating UC from gastrointestinal-diseased controls, 0.62 for discriminating CD from gastrointestinal-diseased controls, 0.85 for discriminating UC from healthy controls, and 0.86 for discriminating UC from healthy controls.

thumbnail image

Figure 2. Results of the Western blotting for FAM84A antibodies. The density of the band on Western blotting is plotted for all patients (UC and CD) and healthy (HC) and gastrointestinal-diseased controls (DC).

Download figure to PowerPoint

Anti-FAM84A antibodies were not associated with age at presentation, disease duration, disease location, smoking habits, colectomy (UC), disease behavior (CD), perianal-disease (CD), surgery (CD), or ASCA titer (CD) (Tables 5, 6). We are unaware how FAM84A antibodies behave in active and quiescent IBD.

Table 5. Anti-FAM84A antibodies measured by Western blotting and clinical characteristics of UC patients
 FAM84A Antibody PositiveFAM84A Antibody NegativeTotal
  • a

    Information on smoking habits was known in 48 subjects.

  • Antibodies to FAM84 were measured using ProtoArray Human Protein Microarrayv4.0. The table summarizes the clinical information on 74 UC patients according to the presence or absence of antibodies to FAM84A.

(n)86674
Male/female2/638/2840/34
Age (years) (mean ± SD)35.8 ± 1343.5 ± 13.042.7 ± 13
Age at presentation (mean ± SD)31.5 ± 1230.0 ± 1230.2 ± 12
Disease duration (mean ± SD)12.6 ± 6.3413.2 ± 713.2 ± 7
Disease location:
Proctitis21618
Left-sided colitis32427
Pancolitis (extended)32629
Surgery12425
P-ANCA52025
C-ANCA11213
Smoking habits (n)a
Unknown54045
Never smoked31417
Active smoker099
Past smoker033
Table 6. Anti-FAM84A Antibodies Measured by Western Blotting and Clinical Characteristics of CD Patients
 FAM84A Antibody PositiveFAM84 Antibody NegativeTotal
  • a

    Information on smoking habits was known in 48 subjects.

  • Antibodies to FAM84 were measured using ProtoArray Human Protein Microarrayv4.0. The table summarizes the clinical information on CD patients according to the presence or absence of antibodies to FAM84.

  • Location CD: L1= ileum, L2 =colon, L3 = ileum + colon, L4 = upper gastrointestinal.

  • Behavior CD: B1= nonstricturing and nonpenetrating, B2 = structuring, B3 = penetrating.

(n)156075
Male/female5/1031/2936/39
Age (years) (mean ± SD)44.4 ± 16.742.5 ± 12.942.9 ± 13.7
Age at presentation (mean ± SD)26.1 ± 11.725.5 ± 10.325.6 ± 10.5
Duration (mean ± SD)20.4 ± 11.119.3 ± 9.219.5 ± 9.6
Disease location
 L182331
 L22810
 L342024
 L1+L4134
 L2+L4011
 L3+L4011
 Unknown044
Disease behavior   
 B172633
 B241620
 B32911
 B2+B32911
Perianal disease61622
Surgery103949
ASCA102636
Smoking habits (n)a   
 Unknown52328
 Never71825
 Active11617
 Past smoker at diagnosis235

DISCUSSION

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Protein microarrays have previously been used for the detection of antibodies in patients suffering from connective tissue disorders, ovarian cancer, type I diabetes, rheumatoid arthritis, and multiple sclerosis.15–19

In an attempt to identify novel autoantibody biomarkers in IBD patients, we evaluated the autoimmune profile in a limited cohort of patients (10 UC and 15 CD) and controls (n = 5) by means of a protein microarray containing 8268 human proteins. In all, 256 antigens were identified to which IBD patients had a higher seroreactivity than controls. Twenty antigens were selected for further evaluation in a larger cohort (n = 240) by means of a customized protein microarray. Out of these 20 antigens, one antigen (FAM84A) was identified to which IBD patients had an appreciably higher seroreactivity than gastrointestinal controls.

The study suffers from a selection bias, as only ANCA-positive UC patients and ASCA- or pancreatic antibody-positive CD patients were included in the screening experiments. The reason for this is that we initially opted to search for the target antigen of ANCA and pancreatic antibodies. Therefore, a validation experiment was performed in which a large nonselected but well-defined IBD cohort was tested.

The validation experiment using Western blotting confirmed the microarray data and showed an increased prevalence of antibodies to FAM84A in IBD patients compared to controls. However, the percentages of antibody-positive UC and CD patients differed. This can be explained by differences in the conformation of the target antigen (native versus denaturated), differences in sensitivity of the detection reagents, and differences in defining cutoff values.

The protein microarray does not contain the GP2 zymogen granule protein that was recently identified as the pancreatic autoantigen, and therefore our study did not reveal this protein as the target of pancreatic antibodies.20 Neither did the protein microarray contain ubiquitination factor E4A, tropomyosin isoform 5, or complement factor C3. Antibodies to these antigens have been described in CD, UC, and IBD, respectively.21–23

To our knowledge, antibodies to FAM84A have not been described previously. FAM84A is a gene whose expression has been reported to be upregulated in colorectal cancer.24

The diagnostic accuracy of antibodies to FAM84A was less compared to the diagnostic accuracy of pANCA and ASCA antibodies, mostly because of lower sensitivity. However, the fact that anti-FAM84 antibodies had a low prevalence in gastrointestinal-diseased controls makes them an attractive marker for IBD. The diagnostic accuracy of antibodies to FAM84A was comparable to the diagnostic accuracy of some other antibodies to autoantigens and bacterial antigens described in IBD. Antibodies to ubiquitination factor E4A were found in 46.2% of CD patients, in 7.1% of UC patients, and in 3.3% of healthy controls.21 Antibodies to human tropomyosin isoform 5 were found in 19% of CD patients, in 64% of UC patients, and in 4% of healthy controls.22 Antibodies to complement C3 were found in 23% of IBD patients, but not in healthy controls.23 Antibodies to pancreatic tissue were found in 39% of CD patients, 4% of UC patients, but not in healthy controls or patients with celiac disease.25, 26 Goblet cell autoantibodies (GAB) were detected in 30%–33% of CD patients, compared to 28%–39% of UC patients and 3% of healthy subjects, but also in 20% of first-degree relatives of patients with UC and CD, and in 3% of inflammatory controls.27, 28 Anti-I2 IgA antibodies were found in 10% of the UC patients and 54% of the CD patients, compared to 3.8% of healthy subjects and 18.9% of diseased inflammatory controls.29 Anti-CBir1 antibodies were detected in 6% of the UC patients and 50% of CD patients, compared to 8% of healthy controls and 14% of gastrointestinal-diseased controls.30 Anti-OmpC antibodies were detected in 19.5% of UC patients, 29.1% of CD patients, 6.5% of healthy controls, and 24.8% of gastrointestinal-diseased controls.31 Several of these markers suffer from a low specificity in gastrointestinal-diseased controls and a low likelihood ratio (varying between 0.97 and 12) (likelihood in diseased IBD people divided by likelihood in gastrointestinal-diseased controls). For antibodies to FAM84A, the likelihood ratio for UC or CD varied between 3.6 and 15.7, depending on the cohort that was tested. These data are summarized in Table 7.

Table 7. Prevalence of Antibodies in IBD Patients
 LikelihoodLR Compared to HCLR Compared toGiCReference
UCCDHCGiCLR(UC)LR(CD)LR(UC)LR(CD)
  1. The table summarizes the likelihoods for UC, CD, healthy controls (HC) and gastrointestinal diseased controls (GiC). The likelihood ratio (LR) for UC and CD compared to HC andGiCis given as well.

  2. ALCA: laminaribioside; ACCA: chitobioside.

pANCA60-80%5-25%4%6%15-201.2-6.210-13.30.8-4.2(5,6&&rpar;
ASCA<10%50-80%<5%2%210-16525-40(7)
ALCA4%27%2%9%22.4-13.523(32, 33)
ACCA5%25%12%9%0.42.1-2.70.42.8(32, 33)
OmpC Ab19.5%29.1%6.5%24.8%34.40.81.2(31)
I2 Ab10%54%3.8%18.9%2.614.20.52.8(29)
CBir1 Ab6%50%8%14%0.756.30.43.6(30<)
Pancreas Ab4%39%0%0%(25,26)
Goblet cell Ab18-39%30-33%3%3%6-1310-116-1310-11(27,28)
Ubiquination factor E4A Ab7.1%46.2%3.3% 2.114  (21)
Tropomyosin 5 Ab64%19%4% 164.7  (22)
Complement C323%0%(23)
FAM84A Ab26.7%11.7%5%1.7%5.32.3415.76.8 
 18%22%0%5%3.64.4 

In conclusion, using a protein microarray approach we identified FAM84A as a novel autoantigen in IBD. Further evaluation in clinically well-characterized patients and controls is needed to explore its clinical utility.

REFERENCES

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. SUBJECTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  7. Supporting Information

Additional supporting information may be found in the online version of this article.

FilenameFormatSizeDescription
IBD_21508_sm_suppInfo.doc38KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.