926
June 1976 TheJournalofPEDIATRICS
Histocompatibility antigens in childhood-onset arthritis One hundred and twelve well-studied patients with a prior diagnosis of juvenile rheumatoid arthritis were differentiated into seven clinically distinct subgroups, including a group in whom recognizable ankylosing spondylitis had developed by time of follow-up. An apparent increased prevalence of HLA-B27 in the entire series (26%) was clearly related to its inereased prevalence in only two subgroups: patients whose disease had progressed to overt ankylosing spondylitis (five o f five patients) and boys with pauciarticular arthritis whose disease wouM be consistent with early ankylosing spondy6tis (11 of 18 patients). There were no significant associations of B27 with systemic onset JRA, polyarticular JR_d, pauciarticular JRA in girls, or JRA with chronic iridocyelitis. The only other significant alterations found were increased prevalences of HLA-A2 and HLA-B W15 in patients with polyarticular disease without identifiable rheumatoid factor. This study emphas&es that the c6nical disorders included under the category of juvenile rheumatoid arthritis represent more than a single disease and that this heterogeneity must be considered in interpreting studies such as those of histocompatibility typing.
Jane G. Sehaller,* Hans D. Oehs, E. Donnall Thomas, Brenda Nisperos, Polly Feigl, and Ralph J. Wedgwood, S e a t t l e , W a s h .
THE HISTOCOMPATIBILITY ANTIGEN, HLA-B27 (formerly W27)t has been associated with ankylosing spondylitis, 1,2 acute iridocyclitis, ~ the spondylitis of psoriasis and bowel disease, 4 Reiter syndrome, 5,G and certain types of postinfectious arthritis2 8 Several studies From the Departments of Pediatrics (Division of Arthritis and Immunology), Medicine (Division of Oncology, Hutchinson Cancer Research Center), and Biostatistics, University of Washington, School of Medicine. Supported in part by grants from the National Institute of Allergy and Infectious Diseases, A I 07073 and National Institute of Cancer, CA 18047. Presented before the VIII European Rheumatology Congress, HelsinkL June 1-7, 1975. Dr. Ochs is an investigatorof the Howard Hughes Medical Institute. Dr. Thomas is a recipient of Research Career Award A1 02425 from the National Institute of A llergy and Infectious Diseases. Dr. Sehaller is a Clinical Scholar of the Arthritis Foundation. *Reprint address: Department of Pediatrics RD-20, University of Washington, School of Medicine, Seattle, Wash. 98195.
"~Nomenclatureof HLA-typingis in accordancewith recommendations of the Sixth International Histocompatibility Testing Workshop ,in Histocompatibility Testing, 1975, Kissmeyer-Nielsen F, editor: Copenhagen, 1976, Ejnar MunksgaardForlag (in press).
Vol. 88, No. 6, pp. 926-930
have also suggested an association of B27 with .juvenile rheumatoid arthritis~-~; however, reported prevalences have varied greatly? -1~
See related article, p. 913. Abbreviations used JRA: juvenile rheumatoid arthritis AS: ankylosing spondylitis RF: rheumatoid factor HLA: histocompatibility antigen Juvenile rheumatoid arthritis is a disease which can be
readily differentiated into several clinically distinct syndromes, 17 We undertook this study to define the prevalence of B27 and other histocompatibitity antigens in a well-studied group of patients with JRA, and to determine whether the reported differences in prevalence of B27 might be explained by associations of this histocompatibility antigen with particular disease subgroups.
P A T I E N T S AND M E T H O D S One hundred and twelve patients with a prior diagnosis of JRA who visited the Children's Arthritis Clinic at the
Volume 88 Number 6
University of Washington between July, 1974, and January, 1975, were studied. Some patients were new referrals; most were regular clinic patients. All of them had well-documented medical histories. Criteria for original diagnosis of JRA were onset o f disease before the sixteenth birthday, persistent (three months or longer) arthritis of one or more joints, or typical high intermittent fever and rheumatoid rash, and exclusion Of other diseases associated with arthritis.* Diagnosis of AS requires characteristic radiographic sacroiliac joint changes 18 as well as definite clinical or radiographic changes of AS in the lumbar spine. TM Patients were differentiated into seven disease subgroups: (1) systemic onset disease (high intermittent fever and rheumatoid rash with or ffithout other extraarticular manifestations, usually associated with polyarthritis); (2) polyarticular disease without prominent systemic manifestations and with negative tests for rheumatoid factor (RF negative); (3) rheumatoid factor positive (RF positive) polyarticular disease; (4) chronic iridocyclitis and arthritis; (5) and (6) pauciarticular arthritis (arthritis limited to one to five joints throughout the course of disease) without iridocyclitis (pauciarticular arthritis in girls and pauciarticular arthritis in boys were considered as separate subgroups because of differences which became apparent on analysis of clinical and laboratory data); ( 7 ) a n k y l o s i n g spondyfitis (this group included patients with a prior diagnosis of JRA whose disease had evolved to recognizable AS by time of this follow-up). Standard evaluations included detailed review of history, complete physical examination, pelvic radiograph, determination of serum latex agglutination 1~ and antinuclear antibodies, s~ and histocompatibility typing. HLA typing was carried out using a modification of the method of Singal and associates. 2~ Each patient was studied with University of Washington trays containing 144 antisera determining 23 specificities. In addition, a number of B27 positive patients and their families were evaluated with NIH-trays, consisting of 68 antisera. All trays contained five sera monospecific for B27. The prevalences of HLA types in patients were compared with reported prevalences in a series of 1863 white North Americans. 6 Pelvic radiographs were read by two observers without knowledge of the patient's identity. All patients had presence or absence of iridocyclitis documented by slit*Patients in whom the following diseases were recognized were excluded: lupus erythematosus, rheumatic fever, dermatomyositis, scleroderma, ankylosingspondylitis,inflammatorybowel disease, Reiter syndrome, psoriasis,infectious arthritis, malignancies, and orthopedic conditionssuch as Legg-Perthesdisease.
Histocompatibility antigens in arthritis
927
lamp examinations at yearly or more frequent intervals. Acute and chronic iridocyclitis were distinguished on clinical grounds2~: acute iridocyclitis was characterized by prominent early symptoms and signs with attacks of brief duration; chronic iridocyclitis was an insidious process of long, active inflammation and frequent ocular sequelae. The observed prevalence of each HLA antigen in the patients was tested for equality against the prevalence in a controlled series of 1,863 subjects 6 using a chi square test without Yates correction. Resultant p values were multipried by 25, the number of tests, to adjust them for the multiple independent tests performed. *~3 Observed prevalences of B27, A2, BW15, and all other HLA antigens in each of the seven subgroups were tested separately for equality against their prevalences in the control group (1,863 subjects) by determining the exact binomial probabilities associated with differences as great as observed.t Probability values (corrected) of greater than 0.05 (p > 0.05) were considered not significant. RESULTS The 112 patients fell into the disease subgroups shown in Table I. There were several differences between subgroups with respect to sex, age at onset, and duration (Table !)- Higher proportions of boys were present in the systemic onset, pauciarticular, and AS groups than in the other subgroups. Patients with AS, boys with pauciarticular arthritis, and RF-positive patients were generally older at onset of disease than were girls with pauciarticular arthritis or patients from the systemic onset, polyarticular R F negative, and chronic iridocyclitis groups. Boys with pauciarticular arthritis had a relatively short duration of disease at time o f this follow-up as compared to patients with identifiable AS. Only one patient in the series had had acute iridocyclitis, a boy with AS. Table II shows the prevalences of histocompatibility antigens in the total patient group as compared to the control group. Significant associations were found for HLA-B27, HLA-A2, and HLA-BW15. Analyses of disease subgroups for these three antigens are shown in Table III. T h e prevalence of B27 was significantly increased in only two subgroups of patients: B27 was present in 11 of 18 boys with pauciarticular *This procedure,based on Bonferroni'sinequality,assuresthat if there are really no differences between the two groups, the chance of erroneously declaring one or more of the observed differences "significant"will be less than the adjusted p value.2~ tThe exactbinomialwas used because of small subgroup samplesizes, the controlprevalenceswere treated as known values,and the tabled onetailedprobabilitieswere doubledto givea two-sidedtest and multipliedby seven, the number of comparisons, to adjust for multiple comparisons (Tables of the Binomial Probability Distribution. National Bureau Standards, Applied Mathematics 6. United States Government Printing Office, Washington, 1949).
928
Schaller et al.
The Journal of Pediatrics June 1976
T a b l e I. Clinical characteristics o f 112 c h i l d r e n with c h i l d h o o d - o n s e t arthritis ( J R A a n d A S )
Age at onset
Systemic onsett polyarticular R F negative Polyarticular R F positive Chronic iridocyclitis3~ Pauciarticular, girls Pauciarticular, boys Ankylosing spondylitis
Female
Male
Total
Mean
10 24 7 11 18 -1
12 2 1 4 18 4
22 26 8 15 18 18 5
5.7 5.1 11.0 2.5 4.3 10.2 8.8
Duration
Median[Range* 3.8 2.4 12.0 2.2 2.2 9.8 9.0
2.3-10.6 2.0-9.1 6.6-15.3 1.9-3.3 1.8-6.0 7.4-12.6 6.0-12.5
Mean l Median 5.8 7.3 4.6 7.5 4.8 2.6 8.8
Range*
5.5 6.3 2.8 5.5 3.0 1.4 8.5
2.4-9.6 2.4-12.6 1.3-7.8 4.0-11.5 2.4-6.8 1.1-3.8 4.5-13.5
*Twenty-fifth tO seventyLfifth percentile. tEighteen of 22 paiients with systemic onset disease had arthritis of multiple joints. ~:Fourteen of 15 patients with chronic iridocyclitis had arthritis of 1-5 joints; one patient had arthritis of more than five joints. T a b l e II. P r e v a l e n c e o f h i s t o c o m p a t i b i l i t y a n t i g e n s in p a t i e n t s a n d c o n t r o l s ~
% Total patients studied (n = 112)
% Controls~ (n = 1,863)
X2
HL-A 1 HL-A 2 HL-A 3 HL-A 9 HL-A 10 HL-A 11 W 19] W28
26 66 24 22 10 4 18 2 28
32 50 27 20 13 11 23 7 17
1.82 10.88 .47 .14 1.56 4.81 1.06 4.57 9.81
ns 0.025 ns ns ns ns ns ns 0.05
HL-A 5 HL-A 7 HL-A 8 HL-A 12 HL-A 13 W14 W 15 W 16 W17 W18 W21 W22 W27 W5 W 10
10 22 18 26 2 3 20 4 7 3 0 4 26 11 19 25
10 24 20 28 5 8 9 7 10 8 4 4 8 18 12 25
0.003 .17 .30 .24 2.32 4.15 16.04 1.94 .96 4.27 4.62 .07 40.80 3.0 4.44 0
ns ns ns ns ns ns < 0.002 ns ns ns ns ns
June 1976 TheJournalofPEDIATRICS
Histocompatibility antigens in childhood-onset arthritis One hundred and twelve well-studied patients with a prior diagnosis of juvenile rheumatoid arthritis were differentiated into seven clinically distinct subgroups, including a group in whom recognizable ankylosing spondylitis had developed by time of follow-up. An apparent increased prevalence of HLA-B27 in the entire series (26%) was clearly related to its inereased prevalence in only two subgroups: patients whose disease had progressed to overt ankylosing spondylitis (five o f five patients) and boys with pauciarticular arthritis whose disease wouM be consistent with early ankylosing spondy6tis (11 of 18 patients). There were no significant associations of B27 with systemic onset JRA, polyarticular JR_d, pauciarticular JRA in girls, or JRA with chronic iridocyelitis. The only other significant alterations found were increased prevalences of HLA-A2 and HLA-B W15 in patients with polyarticular disease without identifiable rheumatoid factor. This study emphas&es that the c6nical disorders included under the category of juvenile rheumatoid arthritis represent more than a single disease and that this heterogeneity must be considered in interpreting studies such as those of histocompatibility typing.
Jane G. Sehaller,* Hans D. Oehs, E. Donnall Thomas, Brenda Nisperos, Polly Feigl, and Ralph J. Wedgwood, S e a t t l e , W a s h .
THE HISTOCOMPATIBILITY ANTIGEN, HLA-B27 (formerly W27)t has been associated with ankylosing spondylitis, 1,2 acute iridocyclitis, ~ the spondylitis of psoriasis and bowel disease, 4 Reiter syndrome, 5,G and certain types of postinfectious arthritis2 8 Several studies From the Departments of Pediatrics (Division of Arthritis and Immunology), Medicine (Division of Oncology, Hutchinson Cancer Research Center), and Biostatistics, University of Washington, School of Medicine. Supported in part by grants from the National Institute of Allergy and Infectious Diseases, A I 07073 and National Institute of Cancer, CA 18047. Presented before the VIII European Rheumatology Congress, HelsinkL June 1-7, 1975. Dr. Ochs is an investigatorof the Howard Hughes Medical Institute. Dr. Thomas is a recipient of Research Career Award A1 02425 from the National Institute of A llergy and Infectious Diseases. Dr. Sehaller is a Clinical Scholar of the Arthritis Foundation. *Reprint address: Department of Pediatrics RD-20, University of Washington, School of Medicine, Seattle, Wash. 98195.
"~Nomenclatureof HLA-typingis in accordancewith recommendations of the Sixth International Histocompatibility Testing Workshop ,in Histocompatibility Testing, 1975, Kissmeyer-Nielsen F, editor: Copenhagen, 1976, Ejnar MunksgaardForlag (in press).
Vol. 88, No. 6, pp. 926-930
have also suggested an association of B27 with .juvenile rheumatoid arthritis~-~; however, reported prevalences have varied greatly? -1~
See related article, p. 913. Abbreviations used JRA: juvenile rheumatoid arthritis AS: ankylosing spondylitis RF: rheumatoid factor HLA: histocompatibility antigen Juvenile rheumatoid arthritis is a disease which can be
readily differentiated into several clinically distinct syndromes, 17 We undertook this study to define the prevalence of B27 and other histocompatibitity antigens in a well-studied group of patients with JRA, and to determine whether the reported differences in prevalence of B27 might be explained by associations of this histocompatibility antigen with particular disease subgroups.
P A T I E N T S AND M E T H O D S One hundred and twelve patients with a prior diagnosis of JRA who visited the Children's Arthritis Clinic at the
Volume 88 Number 6
University of Washington between July, 1974, and January, 1975, were studied. Some patients were new referrals; most were regular clinic patients. All of them had well-documented medical histories. Criteria for original diagnosis of JRA were onset o f disease before the sixteenth birthday, persistent (three months or longer) arthritis of one or more joints, or typical high intermittent fever and rheumatoid rash, and exclusion Of other diseases associated with arthritis.* Diagnosis of AS requires characteristic radiographic sacroiliac joint changes 18 as well as definite clinical or radiographic changes of AS in the lumbar spine. TM Patients were differentiated into seven disease subgroups: (1) systemic onset disease (high intermittent fever and rheumatoid rash with or ffithout other extraarticular manifestations, usually associated with polyarthritis); (2) polyarticular disease without prominent systemic manifestations and with negative tests for rheumatoid factor (RF negative); (3) rheumatoid factor positive (RF positive) polyarticular disease; (4) chronic iridocyclitis and arthritis; (5) and (6) pauciarticular arthritis (arthritis limited to one to five joints throughout the course of disease) without iridocyclitis (pauciarticular arthritis in girls and pauciarticular arthritis in boys were considered as separate subgroups because of differences which became apparent on analysis of clinical and laboratory data); ( 7 ) a n k y l o s i n g spondyfitis (this group included patients with a prior diagnosis of JRA whose disease had evolved to recognizable AS by time of this follow-up). Standard evaluations included detailed review of history, complete physical examination, pelvic radiograph, determination of serum latex agglutination 1~ and antinuclear antibodies, s~ and histocompatibility typing. HLA typing was carried out using a modification of the method of Singal and associates. 2~ Each patient was studied with University of Washington trays containing 144 antisera determining 23 specificities. In addition, a number of B27 positive patients and their families were evaluated with NIH-trays, consisting of 68 antisera. All trays contained five sera monospecific for B27. The prevalences of HLA types in patients were compared with reported prevalences in a series of 1863 white North Americans. 6 Pelvic radiographs were read by two observers without knowledge of the patient's identity. All patients had presence or absence of iridocyclitis documented by slit*Patients in whom the following diseases were recognized were excluded: lupus erythematosus, rheumatic fever, dermatomyositis, scleroderma, ankylosingspondylitis,inflammatorybowel disease, Reiter syndrome, psoriasis,infectious arthritis, malignancies, and orthopedic conditionssuch as Legg-Perthesdisease.
Histocompatibility antigens in arthritis
927
lamp examinations at yearly or more frequent intervals. Acute and chronic iridocyclitis were distinguished on clinical grounds2~: acute iridocyclitis was characterized by prominent early symptoms and signs with attacks of brief duration; chronic iridocyclitis was an insidious process of long, active inflammation and frequent ocular sequelae. The observed prevalence of each HLA antigen in the patients was tested for equality against the prevalence in a controlled series of 1,863 subjects 6 using a chi square test without Yates correction. Resultant p values were multipried by 25, the number of tests, to adjust them for the multiple independent tests performed. *~3 Observed prevalences of B27, A2, BW15, and all other HLA antigens in each of the seven subgroups were tested separately for equality against their prevalences in the control group (1,863 subjects) by determining the exact binomial probabilities associated with differences as great as observed.t Probability values (corrected) of greater than 0.05 (p > 0.05) were considered not significant. RESULTS The 112 patients fell into the disease subgroups shown in Table I. There were several differences between subgroups with respect to sex, age at onset, and duration (Table !)- Higher proportions of boys were present in the systemic onset, pauciarticular, and AS groups than in the other subgroups. Patients with AS, boys with pauciarticular arthritis, and RF-positive patients were generally older at onset of disease than were girls with pauciarticular arthritis or patients from the systemic onset, polyarticular R F negative, and chronic iridocyclitis groups. Boys with pauciarticular arthritis had a relatively short duration of disease at time o f this follow-up as compared to patients with identifiable AS. Only one patient in the series had had acute iridocyclitis, a boy with AS. Table II shows the prevalences of histocompatibility antigens in the total patient group as compared to the control group. Significant associations were found for HLA-B27, HLA-A2, and HLA-BW15. Analyses of disease subgroups for these three antigens are shown in Table III. T h e prevalence of B27 was significantly increased in only two subgroups of patients: B27 was present in 11 of 18 boys with pauciarticular *This procedure,based on Bonferroni'sinequality,assuresthat if there are really no differences between the two groups, the chance of erroneously declaring one or more of the observed differences "significant"will be less than the adjusted p value.2~ tThe exactbinomialwas used because of small subgroup samplesizes, the controlprevalenceswere treated as known values,and the tabled onetailedprobabilitieswere doubledto givea two-sidedtest and multipliedby seven, the number of comparisons, to adjust for multiple comparisons (Tables of the Binomial Probability Distribution. National Bureau Standards, Applied Mathematics 6. United States Government Printing Office, Washington, 1949).
928
Schaller et al.
The Journal of Pediatrics June 1976
T a b l e I. Clinical characteristics o f 112 c h i l d r e n with c h i l d h o o d - o n s e t arthritis ( J R A a n d A S )
Age at onset
Systemic onsett polyarticular R F negative Polyarticular R F positive Chronic iridocyclitis3~ Pauciarticular, girls Pauciarticular, boys Ankylosing spondylitis
Female
Male
Total
Mean
10 24 7 11 18 -1
12 2 1 4 18 4
22 26 8 15 18 18 5
5.7 5.1 11.0 2.5 4.3 10.2 8.8
Duration
Median[Range* 3.8 2.4 12.0 2.2 2.2 9.8 9.0
2.3-10.6 2.0-9.1 6.6-15.3 1.9-3.3 1.8-6.0 7.4-12.6 6.0-12.5
Mean l Median 5.8 7.3 4.6 7.5 4.8 2.6 8.8
Range*
5.5 6.3 2.8 5.5 3.0 1.4 8.5
2.4-9.6 2.4-12.6 1.3-7.8 4.0-11.5 2.4-6.8 1.1-3.8 4.5-13.5
*Twenty-fifth tO seventyLfifth percentile. tEighteen of 22 paiients with systemic onset disease had arthritis of multiple joints. ~:Fourteen of 15 patients with chronic iridocyclitis had arthritis of 1-5 joints; one patient had arthritis of more than five joints. T a b l e II. P r e v a l e n c e o f h i s t o c o m p a t i b i l i t y a n t i g e n s in p a t i e n t s a n d c o n t r o l s ~
% Total patients studied (n = 112)
% Controls~ (n = 1,863)
X2
HL-A 1 HL-A 2 HL-A 3 HL-A 9 HL-A 10 HL-A 11 W 19] W28
26 66 24 22 10 4 18 2 28
32 50 27 20 13 11 23 7 17
1.82 10.88 .47 .14 1.56 4.81 1.06 4.57 9.81
ns 0.025 ns ns ns ns ns ns 0.05
HL-A 5 HL-A 7 HL-A 8 HL-A 12 HL-A 13 W14 W 15 W 16 W17 W18 W21 W22 W27 W5 W 10
10 22 18 26 2 3 20 4 7 3 0 4 26 11 19 25
10 24 20 28 5 8 9 7 10 8 4 4 8 18 12 25
0.003 .17 .30 .24 2.32 4.15 16.04 1.94 .96 4.27 4.62 .07 40.80 3.0 4.44 0
ns ns ns ns ns ns < 0.002 ns ns ns ns ns
