TissueAntigens(1978), 11, 147-152 Published b y Munksgaard, Copenhagen, Denmark N o part may be reproduced by any process without written permission from the author(s)
HLA Antigens in Inflammatory Bowel Disease J. C. Woodrow,* R. M. Lewkonia,* R. B. McConnell,* Ella M. V. D. Berg-Loonen,* * *** S. G. M. Meuwissen,** Betty J. Dekker-Saeys,** L. E. Nijenhuis,** J. F. Mowbray, and N. McI. Johnson*** *Department of Medicine, University of Liverpool **Division of Gastroenterology and Laboratory for Clinical Experimental Medicine, University of Amsterdam, The Netherlands *** St. Mary’s Hospital, London, England Received for publication 10 August, accepted 19 August 1977
Previous studies of the relative frequencies of HLA antigens in patients with ulcerative colitis and Crohn’s disease have not demonstrated any clear-cut positive associations with any particular phenotypes. (Gleeson et al. 1972, Asquith et al. 1974, Jacoby & Jayson 1974, Bergman et al. 1976, Mallas et al. 1976). The three studies detailed here were carried out independently and because there was evidence at two of the centres (London and Amsterdam) that the frequency of HLA B27 was increased in patients with ulcerative colitis, it was decided to combine the data, to test this and t o see if there were any other disturbances in the frequencies of HLA antigens.
and the panels of antisera used for typing varied in detail from centre to centre.
Material and Methods At each centre the patients were examined at specialized gastroenterological units and the diagnosii was based on clinical, sigmoidoscopic, radiological and biopsy findings. The lymphocytotoxic technique used
HLA -B2 7 It is seen (Table 2) that the overall relative risk for ulcerative colitis in HLA B27 individuals is 2.15, x2 = 9.47, P = 0.002, with no evidence of heterogeneity. This is easily the lowest value for P in the total distrib-
Results Table 1 details the HLA antigen frequencies in ulcerative colitis found at the three centres and Table 2 gives the results of combining the data using the Woolf method (Wodf 1955) with the Hddane (1955) modification in those instances where in any of the three series n o patient had a particular HLA antigen. Table 3 gives corresponding frequencies for Crohn’s disease patients with a combined analysis (the London series constituted only 14 patients and is not documented in detail).
Patients (51) os.
20 30 15 5 3 7 1 2
Table 1 Frequencies of H L A anti.gens in patients with ulcerative colitis and in controls Liverpool Controls (375)
%Pm.
Pos.
37 111 110 106 20 28 19 29 37 27 19 45 43
11.8 11.8 5.9 59 21.6 9.8
5.9 25.5 21.6 37.3 3.9 3.9
3 13 11 19 2 2 0 6 6 3 3 11 5
23
2.o
1
18
7.8
128 170 110 73 32 46 18 27
39.2 58.8 29.4 9-8 59 13.7 2.o 39
0
% Pos.
34.3 45.3 29.3 19.5 8.5 12.3 4.8 7.2
Amsterdam Patients (58) Pos.
19 27 15 8 7 14
4 3 1
% Pos.
32.8 46.6 25.9 13.8 12.1 24.1 69 5.2 1.7
Controls (478) POS.
146 251 137 98 20 45 46 32 23
%PO%
30.5 52.5 28.7 20.5 4.2 9.4 9.6 6.7 4.8
London Patients (36)
Pos. 13 15 8 3 7 3 3 3 -
%PO%
36.1 41.7 22.2 8.3 19.4 8.3 8.3 8.3 -
Po 5 8 4 3 1 2 2
4.8 6.1 9.9 29.6 29.3 28.3 5.3 7.5 5.1 7.7 99 72 5.1 12.0 11.5
2 3 6 9 15 14 3 1 2 11 6 7 5 13 4
3.5 5-2 10.3 15.5 25.9 24.1 5.2 1.7 3.5 19.0 10.3 12.1 8.6 22.4 6.9
15 67 69 133 120 109 22 22 33 47 76 39 24 82 59
3.1 6.7
14.4 27.8 25.1 22.8 4.6 4.6 69 9.8 15.9 8.2 5 .o 17.2 12.3
3 4 11 14 10 1 2 0 8 3 1 0 4 5
8.3 11.1 30.6 38.9 27.8 2.8 5.6
1 1 4 4 5 1
0
22.2 8.3 2.8 0 11.1 13.9
1 2 1 2 2
HLA ANTIGENS IN INFLAMMATORY BOWEL DISEASE
149
Table 2 Analysis of the combined data from Table 1 using Woolfs method with Haldane’s modification (1955) where there were no patients with a particular antigen HLA
Combined X
Heterogeneity+
P
X2
X2
0.076 4.283 0.122 0.489 4.493 6.208 (P< 0.05) 0.272 3.280
A1 A2 A3 A9 A10 All A28 A29 A30 A31 A32
1.18 1.04 0.95 0.51 2.13 1.74 0.65 1.03
0.759 0.047 0.07 7 5.841 6.370 4.978 1.213 0.005
1.07.
0.025
0.87
1.362
0.83
0.210
0.64
1.541
B5 B7 B8 B12 B13 B14 B18 B27 BW15 BW17 BW22 BW35 BW40
0.76 0.82 1.06 1.16 0.95 0.53 0.39. 2.15 0.84 1.07 1.59 1.43 0.83
0.757 0.869 0.062 0.596 0.013 1.744 2.032 9.465 0.447 0.044 1.451 2.464 0.385
0.38 0.35 0.80
0.867 4.514 3.644 1.621 0.206 0.261 0.380 0.818 1.182 2.44 3 0.496 2.361 1.398
0.38 0.82 0.78 0.02
0.01 0.02 0.27 > 0.99
0.44 0.91 0.19 0.15 0.002 0.50 0.83 0.23 0.12 0.53
Liverpool and Amsterdam only.
t 2 degrees of freedom except A30/31 and B18 (1d.f.)
ution of P values and if multiplied by 23, the number of antigens included in the combined analysis gives a value of 0.046. The combined data therefore support the hypothesis that HLA B27 is increased in frequency in patients with ulcerative colitis. Combination of these data with those of Bergman et al. (1976) and Mallas et al. (1976) gives a relative risk of 2.03, x2 = 15.82, P = 6 . 9 x lo-’ with no evidence of heterogeneity. The high frequency of HLA B27 in patients with both ankylosing spondylitis and ulcerative colitis (Brewerton et al. 1973) could result in a positive association
between ulcerative colitis and B27 if there were bias in the selection of the patients, so that an undue proportion of patients with spondylitis was included. At each centre patients were referred and continued t o attend the clinics because of the gastrointestinal symptoms. In the three series there was a total of four patients with definite ankylosing spondylitis, all with HLA B27. If these are removed from the analysis, there were 20 out of 141 (14.2%)patients who had HLA B27. Comparison with the combined control series gives x2 = 4.00, P = 0.045. It is of interest to compare these results
Table 3 Frequencies of HLA antigens in patients with Crohn 5 diesase with a combitzed analysis. (Controls documented in Table I Liverpool n=43
No. Pos. 17 25 10 5 5 4 3 6 4 1
No.Pos.
% Pos.
6
11.6
1 9 10 14 4 3 14 2 15 4 2
2.3 7 .O 7 .O 7.O 7 .O
13.9 34.9 20.9 30.2 4.7
11.6
1
Amsterdam n=57
17 26 15 17 5 5 7 0 6 5
9.3 7 .O 13.9 9.3 2.3
39.5 58.1 23.3 11.6 11.6
6 15 9 13 2 5
7
2.3
1
3 3 3 3 5
% Pos.
29.8 45.6 26.3 29.8 8.8 8.8 12.3 0.o
10.5 8.8 1.8 15.8 17.5 24.6 8.5 5.3 24.6 3.5 26.3 8.5 3.5 10.5 12.3
X 1.26 1.06 0.82 1.20 1.75 0.83 1.37 1.56 1.59 1.05 0.93 0.83 0.64 1.26 1.27 1.42 3 A6 0.59 1.52 0.90 1.09 0.72 0.78
XZ
0.50 0.08
0.67 0.49 2.38 0.25 0.75 0.94 1.66 0.01 0.08
0.56 2.88 0.93 0.28 0.77 13.47 1.22 2.10 0.07 0.01 0.98 0.42
Combined Analysis
P
x2
0.48 0.77 0.41 0.48 0.12 0.62 0.38 0.33 0.20 0.93 0.78 0.45 0.09 0.34 0.60 0.38 2 x 10-~ 0.27 0.15 0.80
0.94 0.32 0.51
HLA ANTIGENS IN INFLAMMATORY BOWEL DISEASE
with those for patients with Crohn’s disease studied at the three centres. In Liverpool the incidence of HLA B27 in such patients was 3 out of 43, in Amsterdam 2 out of 57 and in London 0 out of 14. Combining the data using Haldane’s correction, the relative risk of Crohn’s disease in HLA B27 positive individuals is 0.59, but the difference from controls is not significant (x2 = 1.56). Comparing the frequency of HLA B27 in patients with ulcerative colitis and those with Crohn’s disease in the present study, the relative risk is 3.26, x2 = 6.14, P = 0.013 (x2 for heterogneity = 1.85, 2 d.f. P > 0.3). Further combination with corresponding data from two other studies (Bergman et al. 1976, Mallas et al. 1976) gives x2 = 10.81, P = 0.001 with no evidence of heterogeneity. There thus appears to be a higher frequency of HLA B27 in patients with ulcerative colitis than in those with Crohn’s disease. This is of interest in regard to the association of spondylitis with inflammatory bowel disease. One explanation of the latter association is that there are genes common to spondylitis and inflammatory bowel disease.
Other HLA Antigens Ulcerative colitis. The data in Table 2 give
x 2 for the combined data in respect of the A series of antigens as 19.53, 10d.f., p
< 0.05.
The relative risk for HLA A9 is low at 0.51 and for HLA A10 and HLA A l l is high (2.13 and 1.74 respectively). None of these differences can be considered individually significant and in the case of HLA A l l there is evidence of heterogeneity between the three centres. For the B series combined data, x 2 = 20.33, 13 d.f.,P > 0.05.
151
Crohn’s disease. For Series A antigens the combined analysis gives x2 = 7.73, 10 d.f., P > 0.5. Noteworthy is the increased frequency of B18 (24.6%) in patients in the Amsterdam series compared with controls (6.9%). However this was not seen in the Liverpool study and there appears t o be significant heterogeneity between the two studies ( P = 0.036). Combination with three other studies (Jacoby & Jayson 1974, Bergman et al. 1976, Mallas et d. 1976) gives a relative risk of 2.0, x2 = 10.1, P = 0.001 but with significant heterogeneity (P = 0.003). For Series B antigens x2 = 23.77, 13 d.f., P
HLA Antigens in Inflammatory Bowel Disease J. C. Woodrow,* R. M. Lewkonia,* R. B. McConnell,* Ella M. V. D. Berg-Loonen,* * *** S. G. M. Meuwissen,** Betty J. Dekker-Saeys,** L. E. Nijenhuis,** J. F. Mowbray, and N. McI. Johnson*** *Department of Medicine, University of Liverpool **Division of Gastroenterology and Laboratory for Clinical Experimental Medicine, University of Amsterdam, The Netherlands *** St. Mary’s Hospital, London, England Received for publication 10 August, accepted 19 August 1977
Previous studies of the relative frequencies of HLA antigens in patients with ulcerative colitis and Crohn’s disease have not demonstrated any clear-cut positive associations with any particular phenotypes. (Gleeson et al. 1972, Asquith et al. 1974, Jacoby & Jayson 1974, Bergman et al. 1976, Mallas et al. 1976). The three studies detailed here were carried out independently and because there was evidence at two of the centres (London and Amsterdam) that the frequency of HLA B27 was increased in patients with ulcerative colitis, it was decided to combine the data, to test this and t o see if there were any other disturbances in the frequencies of HLA antigens.
and the panels of antisera used for typing varied in detail from centre to centre.
Material and Methods At each centre the patients were examined at specialized gastroenterological units and the diagnosii was based on clinical, sigmoidoscopic, radiological and biopsy findings. The lymphocytotoxic technique used
HLA -B2 7 It is seen (Table 2) that the overall relative risk for ulcerative colitis in HLA B27 individuals is 2.15, x2 = 9.47, P = 0.002, with no evidence of heterogeneity. This is easily the lowest value for P in the total distrib-
Results Table 1 details the HLA antigen frequencies in ulcerative colitis found at the three centres and Table 2 gives the results of combining the data using the Woolf method (Wodf 1955) with the Hddane (1955) modification in those instances where in any of the three series n o patient had a particular HLA antigen. Table 3 gives corresponding frequencies for Crohn’s disease patients with a combined analysis (the London series constituted only 14 patients and is not documented in detail).
Patients (51) os.
20 30 15 5 3 7 1 2
Table 1 Frequencies of H L A anti.gens in patients with ulcerative colitis and in controls Liverpool Controls (375)
%Pm.
Pos.
37 111 110 106 20 28 19 29 37 27 19 45 43
11.8 11.8 5.9 59 21.6 9.8
5.9 25.5 21.6 37.3 3.9 3.9
3 13 11 19 2 2 0 6 6 3 3 11 5
23
2.o
1
18
7.8
128 170 110 73 32 46 18 27
39.2 58.8 29.4 9-8 59 13.7 2.o 39
0
% Pos.
34.3 45.3 29.3 19.5 8.5 12.3 4.8 7.2
Amsterdam Patients (58) Pos.
19 27 15 8 7 14
4 3 1
% Pos.
32.8 46.6 25.9 13.8 12.1 24.1 69 5.2 1.7
Controls (478) POS.
146 251 137 98 20 45 46 32 23
%PO%
30.5 52.5 28.7 20.5 4.2 9.4 9.6 6.7 4.8
London Patients (36)
Pos. 13 15 8 3 7 3 3 3 -
%PO%
36.1 41.7 22.2 8.3 19.4 8.3 8.3 8.3 -
Po 5 8 4 3 1 2 2
4.8 6.1 9.9 29.6 29.3 28.3 5.3 7.5 5.1 7.7 99 72 5.1 12.0 11.5
2 3 6 9 15 14 3 1 2 11 6 7 5 13 4
3.5 5-2 10.3 15.5 25.9 24.1 5.2 1.7 3.5 19.0 10.3 12.1 8.6 22.4 6.9
15 67 69 133 120 109 22 22 33 47 76 39 24 82 59
3.1 6.7
14.4 27.8 25.1 22.8 4.6 4.6 69 9.8 15.9 8.2 5 .o 17.2 12.3
3 4 11 14 10 1 2 0 8 3 1 0 4 5
8.3 11.1 30.6 38.9 27.8 2.8 5.6
1 1 4 4 5 1
0
22.2 8.3 2.8 0 11.1 13.9
1 2 1 2 2
HLA ANTIGENS IN INFLAMMATORY BOWEL DISEASE
149
Table 2 Analysis of the combined data from Table 1 using Woolfs method with Haldane’s modification (1955) where there were no patients with a particular antigen HLA
Combined X
Heterogeneity+
P
X2
X2
0.076 4.283 0.122 0.489 4.493 6.208 (P< 0.05) 0.272 3.280
A1 A2 A3 A9 A10 All A28 A29 A30 A31 A32
1.18 1.04 0.95 0.51 2.13 1.74 0.65 1.03
0.759 0.047 0.07 7 5.841 6.370 4.978 1.213 0.005
1.07.
0.025
0.87
1.362
0.83
0.210
0.64
1.541
B5 B7 B8 B12 B13 B14 B18 B27 BW15 BW17 BW22 BW35 BW40
0.76 0.82 1.06 1.16 0.95 0.53 0.39. 2.15 0.84 1.07 1.59 1.43 0.83
0.757 0.869 0.062 0.596 0.013 1.744 2.032 9.465 0.447 0.044 1.451 2.464 0.385
0.38 0.35 0.80
0.867 4.514 3.644 1.621 0.206 0.261 0.380 0.818 1.182 2.44 3 0.496 2.361 1.398
0.38 0.82 0.78 0.02
0.01 0.02 0.27 > 0.99
0.44 0.91 0.19 0.15 0.002 0.50 0.83 0.23 0.12 0.53
Liverpool and Amsterdam only.
t 2 degrees of freedom except A30/31 and B18 (1d.f.)
ution of P values and if multiplied by 23, the number of antigens included in the combined analysis gives a value of 0.046. The combined data therefore support the hypothesis that HLA B27 is increased in frequency in patients with ulcerative colitis. Combination of these data with those of Bergman et al. (1976) and Mallas et al. (1976) gives a relative risk of 2.03, x2 = 15.82, P = 6 . 9 x lo-’ with no evidence of heterogeneity. The high frequency of HLA B27 in patients with both ankylosing spondylitis and ulcerative colitis (Brewerton et al. 1973) could result in a positive association
between ulcerative colitis and B27 if there were bias in the selection of the patients, so that an undue proportion of patients with spondylitis was included. At each centre patients were referred and continued t o attend the clinics because of the gastrointestinal symptoms. In the three series there was a total of four patients with definite ankylosing spondylitis, all with HLA B27. If these are removed from the analysis, there were 20 out of 141 (14.2%)patients who had HLA B27. Comparison with the combined control series gives x2 = 4.00, P = 0.045. It is of interest to compare these results
Table 3 Frequencies of HLA antigens in patients with Crohn 5 diesase with a combitzed analysis. (Controls documented in Table I Liverpool n=43
No. Pos. 17 25 10 5 5 4 3 6 4 1
No.Pos.
% Pos.
6
11.6
1 9 10 14 4 3 14 2 15 4 2
2.3 7 .O 7 .O 7.O 7 .O
13.9 34.9 20.9 30.2 4.7
11.6
1
Amsterdam n=57
17 26 15 17 5 5 7 0 6 5
9.3 7 .O 13.9 9.3 2.3
39.5 58.1 23.3 11.6 11.6
6 15 9 13 2 5
7
2.3
1
3 3 3 3 5
% Pos.
29.8 45.6 26.3 29.8 8.8 8.8 12.3 0.o
10.5 8.8 1.8 15.8 17.5 24.6 8.5 5.3 24.6 3.5 26.3 8.5 3.5 10.5 12.3
X 1.26 1.06 0.82 1.20 1.75 0.83 1.37 1.56 1.59 1.05 0.93 0.83 0.64 1.26 1.27 1.42 3 A6 0.59 1.52 0.90 1.09 0.72 0.78
XZ
0.50 0.08
0.67 0.49 2.38 0.25 0.75 0.94 1.66 0.01 0.08
0.56 2.88 0.93 0.28 0.77 13.47 1.22 2.10 0.07 0.01 0.98 0.42
Combined Analysis
P
x2
0.48 0.77 0.41 0.48 0.12 0.62 0.38 0.33 0.20 0.93 0.78 0.45 0.09 0.34 0.60 0.38 2 x 10-~ 0.27 0.15 0.80
0.94 0.32 0.51
HLA ANTIGENS IN INFLAMMATORY BOWEL DISEASE
with those for patients with Crohn’s disease studied at the three centres. In Liverpool the incidence of HLA B27 in such patients was 3 out of 43, in Amsterdam 2 out of 57 and in London 0 out of 14. Combining the data using Haldane’s correction, the relative risk of Crohn’s disease in HLA B27 positive individuals is 0.59, but the difference from controls is not significant (x2 = 1.56). Comparing the frequency of HLA B27 in patients with ulcerative colitis and those with Crohn’s disease in the present study, the relative risk is 3.26, x2 = 6.14, P = 0.013 (x2 for heterogneity = 1.85, 2 d.f. P > 0.3). Further combination with corresponding data from two other studies (Bergman et al. 1976, Mallas et al. 1976) gives x2 = 10.81, P = 0.001 with no evidence of heterogeneity. There thus appears to be a higher frequency of HLA B27 in patients with ulcerative colitis than in those with Crohn’s disease. This is of interest in regard to the association of spondylitis with inflammatory bowel disease. One explanation of the latter association is that there are genes common to spondylitis and inflammatory bowel disease.
Other HLA Antigens Ulcerative colitis. The data in Table 2 give
x 2 for the combined data in respect of the A series of antigens as 19.53, 10d.f., p
< 0.05.
The relative risk for HLA A9 is low at 0.51 and for HLA A10 and HLA A l l is high (2.13 and 1.74 respectively). None of these differences can be considered individually significant and in the case of HLA A l l there is evidence of heterogeneity between the three centres. For the B series combined data, x 2 = 20.33, 13 d.f.,P > 0.05.
151
Crohn’s disease. For Series A antigens the combined analysis gives x2 = 7.73, 10 d.f., P > 0.5. Noteworthy is the increased frequency of B18 (24.6%) in patients in the Amsterdam series compared with controls (6.9%). However this was not seen in the Liverpool study and there appears t o be significant heterogeneity between the two studies ( P = 0.036). Combination with three other studies (Jacoby & Jayson 1974, Bergman et al. 1976, Mallas et d. 1976) gives a relative risk of 2.0, x2 = 10.1, P = 0.001 but with significant heterogeneity (P = 0.003). For Series B antigens x2 = 23.77, 13 d.f., P
