Genetic differences between b27 positive patients with ankylosing spondylitis and b27 positive healthy controls.

1460
GENETIC DIFFERENCES BETWEEN B27 POSITIVE
PATIENTS WITH ANKYLOSING SPONDYLITIS AND
B27 POSITIVE HEALTHY CONTROLS
ANDRE1 CALIN, AMY MARDER, ELENA BECKS, and THOMAS BURNS
In a controlled study of the 499 available first
degree relatives of 79 consecutive HLA-B27 positive
patients with ankylosing spondylitis and 69 HLA-B27
positive healthy blood donors, 19 cases of ankylosing
spondylitis were found: 16 (15 B27 positive) among the
282 relatives of the patients with ankylosing spondylitis,
and 3 (1 B27 positive, 1 B27 negative, 1 unknown)
among the 217 relatives of healthy donors y
(,:
= 5.11;
P < 0.025). However, if all cases of possible spondylarthritis are included, 48 cases of ankylosing spondylitis
were found: 37 of 282 relatives of patients with ankylosing spondylitis and 11 of 217 relatives of healthy donors
y
,(:
= 8.29; P < 0.01). Assuming that 50% of relatives
of B27 positive individuals carry this antigen, 15 of 142
(10.6%) B27 positive relatives of patients and 2 of 108
=
(1.9%) B27 positive relatives of healthy subjects y(,:
5.91; P < 0.025) have ankylosing spondylitis. The
relative risk of spondylarthropathy for B27 positive
relatives of B27 positive patients compared with relatives of B27 positive healthy subjects is 5.6. Assuming all
subjects were evaluated in a similar manner, analysis of
these data suggests genetic differences between B27
positive diseased individuals and B27 positive healthy
subjects.
From the Division of Rheumatology, Veterans Administration Hospital, Palo Alto, CA, and Stanford University Medical
Center, Stanford, CA.
Supported by the Veterans Administration Research Service and the Northern California Chapter Arthritis Foundation.
Andrei Calin, MD, MRCP; Amy Marder, BA; Elena Becks,
BA; Thomas Burns, MD.
Address reprint requests to Dr. Andrei Calin, Division of
Immunology, Stanford University Medical Center, 300 Pasteur
Drive, Stanford, CA 94305.
Submitted for publication January 31, 1983; accepted in
revised form July 22, 1983.
Arthritis and Rheumatism, Vol. 26, No. 12 (December 1983)
In spite of the recognized association between
ankylosing spondylitis (AS) and HLA-B27 (1,2), the
precise nature of the predisposition to this disorder
remains an enigma. About 20% of whites with HLAB27 develop sacroiliitis (SI) (3-5) following an unknown inciting event. The figure reaches 50% for Pima
Indian males (6). The environmental agent or agents
that precipitate AS may be ubiquitous since the disease is prevalent with a worldwide distribution, the
frequency of which mirrors that of B27 (7). It is also
recognized that approximately 20% of subjects with
HLA-B27 develop Reiter’s syndrome (RS) after nonspecific urethritis (8), shigellosis (9), or other triggering events. The explanation of why the remaining 80%
of people with HLA-B27 remain healthy is unclear.
One possibility is that B27 positive patients with AS
have an additional susceptibility gene o r genes, that
are lacking in healthy B27 positive subjects. Alternatively, B27 may not be homogeneic, that is, healthy
B27 positive subjects may lack a hypothetical disease
epitope present on B27 molecules of diseased individuals.
To analyze these possibilities, studies can be
performed both at epidemiologic and molecular levels.
The former require in-depth family studies, whereas
the latter rely on monoclonal antibody and biochemical investigations.
We report a controlled study of the prevalence
of AS among first degree relatives of B27 positive
diseased individuals and among relatives of B27 positive healthy controls. A comparable prevalence of
disease in both groups would provide evidence against
genetic differences between the two B27 positive populations. Alternatively, if disease were found more
frequently in relatives of B27 positive patients than in
INHERITANCE OF AS
relatives of B27 positive controls, one could assume
that either additional gene or genes were operative, or
that the B27 molecules themselves were distinct.
Moreover, if a putative susceptibility gene in juxtaposition to B27 or a diseased B27 epitope were operative,
the majority of B27 positive relatives of patients would
be expected to have disease. By contrast, if the
additional genes were distant from B27, or on a
different chromosome, one would expect the frequency of AS among relatives of B27 positive patients t o
range from that found in a random B27 positive
population t o a higher prevalence. The actual figure
would depend on number and location of genes, crossover frequency, and other unknown factors.
PATIENTS AND METHODS
Ankylosing spondylitis (AS) patients with symptornatic bilateral sacroiliitis (SI) grade I1 or greater (10) were
selected for this study.
One hundred twenty consecutive patients with AS
(95% whites) who were seen at the Stanford University
Medical Center and Palo Alto Veterans Administration
Hospital were evaluated. Sixteen patients were excluded
either because they were HLA-B27 negative (9 patients) or
because their HLA type was unknown (7 patients). Of the
remaining I04 patients, 25 were unsuitable for study because
they were adopted, the only surviving member of the family,
or unwilling to release names of their relatives. Therefore, of
the original patient population, 79 (65.8%) were appropriate
for study. Likewise, 99 consecutive HLA-B27 positive
blood donors (97% whites) located through the Stanford
University, Irwin Memorial, and Midpeninsula Blood Banks
and the San Jose Red Cross agreed to take part in the study.
Nineteen were excluded because at the initial survey they
were found to have symptoms suggestive of spondylarthropathy or uveitis. Eleven olhps were excluded, again because
they were adopted, the only member of the family, or
unwilling to release names of relatives. Thus, of the original
99 donors, 69 (69.7%) were suitable for study.
The 79 patients with AS provided us with names of
330 living first degree relatives (mother, father, brother,
sister, son, or daughter); 282 (85.5%) of these relatives
signed a consent form and provided personal data. These
represented an average of 3.6 relatives per propositus.
Comparable figures for the 69 controls were 275 first degree
relatives of whom 217 (78.9%) consented to take part, which
is an average of 3.1 relatives per propositus. The male to
female ratio was 55:24 for the patients and 38:31 for the
donors. The mean ages for the propositi were 40.0 years and
37.9 years, respectively. The male to female ratio of the
relatives of the spondylitics was 129: 153 and for the relatives
of the donors 102: 1 15, and mean ages of the relatives were
51.0 and 48.4 years, respectively.
Screening questionnaires were sent to the 605 relatives, 79 spondylitics, and 69 controls. All individuals were
evaluated without knowledge of their study status. Analysis
included a standardized evaluation of questionnaires pertain-
1461
ing to inflammatory spinal disease (i.e., insidious onset,
duration >3 months, age at onset <40 years, association
with morning stiffness, improvement with exercise) (1 1,12),
and the peripheral stigmata of spondylarthropathy (uveitis,
keratodermia, mouth ulcers, urethritis, and peripheral joint
disease) (13).
Pelvic radiographs, at no cost to the subject, were
sought from individuals who gave a history of back pain that
included 4 or more of the 5 “inflammatory” symptoms.
Those subjects who had less than 4 of the 5. but at least a
total of 6 features, including those of the stigmata of spondylarthropathy, were also asked to have pelvic radiographic
evaluations. All 79 spondylitis patients had a diagnosis of AS
confirmed by radiograph. The 69 controls were considered
disease free (i.e., the absence of the above criteria or normal
pelvic radiograph results in symptomatic individuals). We
chose not to request radiographs from individuals with fewer
criteria since the sensitivity of the questionnaire was found
to be accurate to 95% in a prior study (1 I , 12). Specificity was
of less concern since a definitive diagnosis of AS would
depend on the “blind” reading of the pelvic radiograph.
Completed questionnaires were received from 282
(85.5%) of the 330 relatives of the spondylitic patients and
from 217 (78.9%) of the 275 relatives of the 69 controls. The
sacroiliac radiographs were read “blindly” according to the
New York criteria for AS (14) (grade 0 = normal, I =
suspicious change, I1 = minimal SI, I11 = moderate SI, IV =
fused) by 2 independent observers. If the observers did not
agree on individual readings, they reached a consensus.
Statistical analysis was performed using chi-square
with Yates correction.
RESULTS
One hundred seven (37.9%) of the 282 relatives
of the patients with ankylosing spondylitis gave a
positive response t o the questionnaire (i.e., 1 4 of 5
“inflammatory” back pain symptoms or a t least 6
features, including those suggesting peripheral stigmata of spondylarthritis), compared with 55 (25.3%) of
217 relatives of the controls (xc2 = 8.25; P < 0.01).
Seventy (65.4%) of the 107 “symptomatic” relatives
of the patients with AS agreed to have a pelvic
radiograph evaluation as did 41 (74.5%) of the 55
relatives of the controls (not significant).
Twenty (28.5%) of the 70 radiographs of the
symptomatic relatives of t he patient group showed
abnormal results (sacroiliac grade 211) compared with
4 (9.8%) of the 41 radiographs of the symptomatic
relatives of the healthy controls (20 of 282 versus 4 of
217; xc2 = 6.33; P < 0.025). When only bilateral
disease was considered, the numbers were 16 (22.9%)
and 3 (7.3%), respectively (xc’ = 5.11; P < 0.025).
When questionable radiograph results were included
(grades I and above), the total numbers were 37
(52.9%) and I 1 (26.8%), respectively, for unilateral
CALIN ET AL
1462
and bilateral SI (37 of 282 versus I 1 of 217;
xc2 = 8.29; P < 0.01).
The abnormal radiograph results of relatives of
the patients included 7 with grade IV :IV (bilateral
fused joints), 3 with 111: 111 (moderate sacroiliitis), 4
with II:III, 2 with II:II, 2 with II:I, and 1 each with
1:III and 0:III. The 4 relatives of the controls with
abnormal radiograph results included 1 each with
II:III, III:IV, II:II, and 1I:O.
The ma1e:female distribution of the 16 diseased
relatives (AS, symptomatic bilateral SI) of the patient
group was 10:6 compared with 2: 1 for the 3 patients in
the control group. Together, these 12 men and 7
women were derived from 231 male and 268 female
relatives: 12 of 231 (5.2%) and 7 of 268 (2.6%), a
distribution of 2: 1 in favor of males. Of 282 relatives,
20 had grade I1 or more sacroiliitis, at least on one
side. Assuming that spondylarthritis occurs predominantly in B27 positive subjects, and that 50% of our
study relatives are B27 positive, 20 of 141 (14.2%) B27
positive relatives are affected. Comparable figures for
symmetric disease and suspicious change are 16 of 241
( I 1.3%) and 37 of 241 (26.2%) respectively. Since none
of the propositi had more than 1 relative with symptomatic bilateral SI, 16 (20.3%) of 79 patients with AS
had a diseased relative compared with 3 (4.3%) of the
69 controls. When the 79 propositi with AS were
separated into male and female subgroups, 10 (18.2%)
of the 55 men had a relative with AS compared with 6
(25%) of the 24 women (not significant).
Fifteen (93.8%) of the 16 relatives of patients
with AS who had bilateral SI were B27 positive. The
single B27 negative individual had a 111:I1 reading for
SI. Of the 3 diseased relatives of the healthy controls,
1 was B27 positive, 1 was B27 negative (sacroiliitis
III:II), and 1 was HLA type unknown.
DISCUSSION
Analysis of the data suggests that there is a
highly significant difference in the prevalence of ankylosing spondylitis among the relatives of B27 positive
patients with AS compared with occurrence of AS
among the relatives of B27 positive healthy controls ( P
< 0.025). There are three potential explanations for
these findings: 1) a systematic bias has occurred, 2) a
difference in the environment between the families of
healthy B27 positive subjects and the families of B27
positive patients, or 3) genetic differences between
HLA-B27 positive diseased individuals and HLA-B27
positive healthy subjects.
Inevitably, biases are inherent in a study of this
nature. To some extent, these can be anticipated and
avoided. However, many problems remain and their
effect and direction can only be surmised. The ideal
study would require that patients be selected randomly
from B27 positive individuals who are found to have
symptomatic SI in a population survey. Patients referred to clinics should not be used in these studies
since such individuals may be more likely to have
diseased relatives than a sick individual found during a
routine survey. However, the perfect study could not
be attempted for logistic and ethical reasons. The use
of blood donors for “healthy” controls is suspect
since it is recognized that donors are often not healthy
individuals (3). Moreover, relatives of diseased individuals may be more willing to donate blood than
subjects in the general population.
We have defined AS as symptomatic bilateral
SI (10). Ideally, all 605 relatives would have received a
pelvic radiograph evaluation since asymptomatic SI
can occur (15). From a pathogenic point of view,
asymptomatic SI is as important as symptomatic disease. How many such individuals have been missed
remains unknown, though we have previously shown
that most patients with SI do have symptoms (15).
However, symptoms, by definition, are subjective and
our initial screening technique may have been too
stringent. Apart from ethical considerations, cost precluded taking additional radiographs. Nevertheless, it
seems likely that any under-defining of sacroiliac
disease would have affected both the patient and
control groups equally.
Had we used the New York criteria for the
definition of AS, 10 relatives of patients with AS
would have had definite AS (grade 111-IV bilateral SI
with any clinical criterion) compared with 1 relative of
healthy controls. The remaining 6 relatives of patients
with AS in whom the pelvic radiographs revealed
bilateral SI equal to or greater than grade I1 and the 2
subjects with unilateral SI grade 111, may also have
had definite AS, although this diagnosis would have
required evaluation of spinal mobility. Patients with
“limitation of back movement in all three planes” in
addition to grade 111-IV unilateral SI or grade I1
bilateral SI are said to have definite disease (14). Our
concern with this definition is that limitation of movement is not adequately defined. Regardless of the
definition of AS, the pelvic radiographs were read
“blindly” and therefore, over or under reading would
have affected both groups equally.
At each step data collection was less than
INHERITANCE OF AS
complete and the completion rate differed for the 2
groups. For example, 282 (85.5%) of first degree
relatives of patients returned questionnaires while
only 217 (78.9%) of control relatives did so. By contrast, radiographs were obtained from more of the
selected control groups, 41 of 55 (74.5%) than from the
patient group. 70 of 107 (65.4%). Although our screening questionnaire was developed in a controlled setting
(11) and used in other studies (12,13), its precise
sensitivity and specificity in other populations remain
unknown. Nevertheless, it should be equally applicable to both patient and control groups. However, we
are aware that relatives of “healthy” people may
answer questions differently from those of patients.
Evaluation of radiographs is notoriously unreliable (3,lO). But any systematic under or over reading
would not bias the central issue. Yet, from the point of
view of derived prevalence data, the numbers are
probably underestimated since the completion rates
were not 100%.
There were more women than men among
relatives of both groups (53.0% and 54.3%, respectively). Since the percentages are comparable, there appears to be no additional bias. The preponderance of
women would be expected to decrease the number of
diagnosed cases, since spondylarthritis is more common in men than women.
No attempt was made to control for duration of
familial cohabitation. Since the putative infective incident may occur at any time, such an exercise seemed
unwarranted. Nevertheless, the probands with AS and
their diseased relatives often developed illness at
different times and places. Although an environmental
factor must be operative, homozygotic twin studies
have shown a concordance rate of no more than 70%
(16). The possibility remains that the clustering of
cases within family units related to diseased probands
could be explained by a common infective or toxic
etiology. However, we consider this unlikely since AS
occurs in about 20% of random cases of B27 positive
individuals worldwide, indicating a prevalent environmental trigger (3-5).
The initial finding that 19 of the 99 B27 positive
donors had to be excluded since they had underlying
uveitis or signs of spondylarthropathy was expected
and confirmed previous data (3,4). The male to female
ratio of 2 : l for spondylarthritis among the relatives
approaches the figure suggested by more recent studies (3,17). If the differences between B27 positive
diseased individuals and B27 positive healthy subjects
that were suggested by our study can be confirmed.
1463
two major hypotheses could explain the findings: 1)
differences exist between the B27 molecules of healthy
and diseased individuals (i.e., B27 is heterogeneic), or
2) there must be one or more additional genes present
in those individuals who develop AS. The location of
such genes could be (a) in juxtaposition to HLA-B27,
or (b) distant from B27 on the same or a different
chromosome. If we accept that about 70% of subjects
encounter the infective trigger, a figure derived from
the twin series (16), our family studies should reveal a
similar prevalence of AS in the B27 positive relatives
of B27 positive patients if hypothesis 1 or 2a was
operative. In addition, a much lower figure among
relatives of B27 positive healthy subjects would be
expected. Alternatively, if 2b were the genetic explanation, the prevalence of disease among relatives of
B27 positive patients would be greater than among
relatives of random B27 positive individuals, but less
than the hypothetical 70%. Our data are more consistent with the latter possibility.
That B27 is the susceptibility gene has been
suggested by the reported cross-reactivity between
antibodies directed toward Klebsiella K43 and B27
positive lymphocytes of patients with AS but not of
control subjects or B27 negative patients (18). Recent
evidence that Yersinia enterocoliticn serotype 3 has an
intimate relationship with HLA molecules in patients
with Reiter’s syndrome also directs attention toward
B27 (19). Moreover, the use of monoclonal antibodies
in preliminary studies has differentiated two classes of
B27 molecules designated B27M2 positive and B27M2
negative (20). However, a contrasting biochemical
study found no differences between B27 molecules of
healthy and diseased individuals (21).
AS and Reiter’s syndrome (RS) are closely
related spondylarthropathies. Our own controlled data
and previously reported studies suggest that the nature
of the diseases AS and RS is comparable within
families to the substantial exclusion of the other disorders (22,23). This would suggest that phenotypic
expression is modified by an additional gene or genes.
In each group one relative with AS was found to be
B27 negative. The explanation for familial disease in
the absence of B27 is unknown. One possibility is that
a laboratory “error” has been made (24). Alternatively, genes related to clinically silent psoriasis or inflammatory bowel disease may be relevant (16). Assuming
that the one diseased relative of unknown HLA type
was B27 positive, and that 50% of relatives of B27
positive individuals carry this antigen, 15 of 142
(10.6%) B27 positive relatives of patients will have AS
CALIN ET AL
compared with 2 of 108 (1.9%)B27 positive relatives of
healthy subjects (x: = 5.91; P < 0.025). T h e relative
risk of AS for relatives of B27 positive patients compared with relatives of B27 positive healthy subjects is
5.6.
In summary, our findings suggest that genetic
differences exist between B27 positive healthy subjects and B27 positive patients with AS. Taken togethe r , the molecular a n d epidemiologic data have not yet
defined these putative differences.
ACKNOWLEDGMENTS
We thank the Stanford University Blood Bank, Irwin
Memorial Blood Bank, Midpeninsula Blood Bank, and the
San Jose Red Cross for allowing us to contact their B27
positive blood donors. We also wish to thank Doni Saunders
who was instrumental in preparing this manuscript and Dr.
G. Thomson for her gracious assistance.
REFERENCES
I. Brewerton DA, Caffrey M, Hart FD, James DCO, Sturrock RD: Ankylosing spondylitis and HL-A27. Lancet
1:904-906, 1973
2. Schlosstein L, Terasaki PI, Bluestone R, Pearson C:
High association of an HL-A antigen W27, with ankylosing spondylitis. N Engl J Med 288:704-705, 1973
3. Calin A, Fries JF: Striking prevalence of ankylosing
spondylitis in “healthy” W27 positive males and females: a controlled study. N Engl J Med 2932335-839,
1975
4. Cohen LM, Mittal KK, Schmid FR, Rogers LF. Cohen
K: Increased risk for spondylitis stigmata in apparently
healthy HLA W27 men. Ann Intern Med 84:l-7, 1976
5 . HLA-B27 and risk of ankylosing spondylitis (editorial).
Br Med J 2:650-65 I , 1978
6. Lisse JR, Kuberski TT, Bennett PH, Knowler WC,
Gofton JP, Calin A, Mann DL: High risk of sacroiliitis in
HLA-B27-positive Pima Indian men (letter). Arthritis
Rheum 25:236-238, 1982
7. Calin A: The relationship between genetics and environment in the pathogenesis of rheumatic diseases. West J
Med 131:205-218, 1979
8, Keat AC, Maini RN, Nkwazi GC, Pegrum GD,
Ridgeway GL, Scott JT: Role of chlamydia trachomatis
and HLA B27 in sexually acquired reactive arthritis. Br
Med J 1:605-607, 1978
9. Calin A, Fries JF: An “experimental” epidemic of
Reiter’s syndrome revisited: follow-up evidence on genetic and environmental factors. Ann Intern Med
84~564-566, 1976
10. Calin A: Ankylosing spondylitis, Textbook of Rheumatology. Vol. 11. Edited by WN Kelly, ED Harris, S
Ruddy, CB Sledge. Philadelphia, WB Saunders, 1981,
pp 1017-1032
11. Calin A, Porta J, Fries JF, Schurman DJ: The clinical
history as a screening test in ankylosing spondylitis.
JAMA 237:2613-2614, 1977
12. Calin A, Kaye B, Sternberg M, Antell B, Chan M: The
prevalence and nature of back pain in an industrial
complex. Spine 5:201-205, 1980
13. Simon DG, Kaslow RA, Rosenbaum J, Kaye RL, Calin
A: Reiter’s disease following epidemic shigellosis. J
Rheumatol 8:969-973, 1981
14. Bennett PH, Burch TA: Population Studies of the Rheumatic Diseases. Amsterdam, Excerpta Medica Foundation (Int. Long. Sev. No. 148) 1968, pp 305-313
15. Calin A, Fries JF, Schurman D, Payne R: The close
correlation between symptoms and disease expression
in HLA B27 positive individuals. J Rheurnatol
41277-281, 1977
16. Eastmond CJ, Woodrow JC: Discordance for ankylosing
spondylitis in monozygotic twins. Ann Rheum Dis
361360-364, 1977
17. Hill HFH, Hill AGS, Bodrner JG: Clinical diagnosis of
ankylosing spondylitis in women and relation to presence of HLA B27. Ann Rheum Dis 35:267-270, 1976
18. Seager K, Mashir HV, Geczy A: Evidence for a specific
B27 associated cell surface marker on lymphocytes of
patients with ankylosing spondylitis. Nature 277:68-70,
1979
19. Wiesenhutter CW, Brenner MB, Kobayashi S, Huberman A, Yu DTY: Participation of HLA molecules in the
interaction between human lymphocytes and a Reiter’sdisease-causing-bacterium (abstract). Arthritis Rheum
(supp1)25:S38, I982
20, Grumet FC, Fendly BM, Fish L, Foung S, Engleman
EG: Monoclonal antibody (B27M2) subdividing HLAB27. Hum Immunol 5:61-64. 1982
21. Karr RW, Hahn Y, Schwartz BD: Structural identity of
human histocompatibility leukocyte antigen-B27 molecules from patients with ankylosing spondylitis and
normal individuals. J Clin Invest 69:443-450, 1982
22 Calin A, Barnett M. Marder A, Marks S: The nature and
prevalence of spondylarthritis among relatives of probands with ankylosing spondylitis and Reiter’s syndrome (abstract). Arthritis Rheum (supp1)24:S78, 1981
23. Hochberg M, Bias W. Arnett FC: Family studies in
HLA-B27 associated arthritis. Medicine 57:463-475,
1978
24. Calin A, Barnett M: HLA-27 positive, negative or both?
N Engl J Med 302:1366-1367, 1980