Marfan syndrome is a heritable disorder of connective tissue that is caused
by mutations in the gene for the protein
fibrillin-11. It is
characterized by variable manifestations in multiple organ systems, including
ectopia lentis, aortic root dilatation, dural ectasia, and skeletal
anomalies2.
Protrusio
acetabuli3,4
was first recognized as a feature of Marfan syndrome by Steel in
19785. Since then,
others have reported protrusio in small populations of patients with Marfan
syndrome6-8.
The presence of protrusio is now considered to be one of the skeletal criteria
for a clinical diagnosis of Marfan syndrome according to the most recent Gent
nosology9. Yet, the
prevalence of protrusio has not been studied in a large population of patients
with Marfan syndrome.
Furthermore, the clinical importance of protrusio acetabuli in patients
with Marfan syndrome is not well understood. Several case reports have
described pain and degenerative changes in the hips of adults with protrusio
acetabuli and Marfan
syndrome8,10-13,
and some authors have recommended prophylactic triradiate epiphyseodesis in
pediatric patients with Marfan syndrome and protrusio to prevent early hip
degeneration in
adulthood5,8,14.
This cross-sectional study was undertaken to establish the prevalence of
protrusio acetabuli in a population of patients with Marfan syndrome and to
determine its effects on hip function.
The present study was approved by our institutional review board. Between
1998 and 2001, 237 consecutive patients with a diagnosis of Marfan syndrome,
confirmed by a clinical medical geneticist according to the Gent
criteria9, were
identified when they appeared for routine follow-up appointments at our
hospital's medical genetics clinic (Table
I). The sole inclusion criterion was a diagnosis of Marfan
syndrome with or without known musculoskeletal involvement at the time of
recruitment. One hundred and seventy-three patients (73%) (346 hips) agreed to
participate in the study. Each patient provided informed consent in a document
that was approved by the institutional review board of the hospital. Patients
who chose not to participate cited reasons such as busy schedules (these
patients usually have multiple departments to visit on their clinic days), a
desire to avoid the risks of radiography, and the somewhat arduous consent
process, which is institutionally required.
All patients were questioned about hip pain. On the basis of the physical
examination, we recorded the arc of painless hip flexion, limb-length
discrepancy, and gait abnormalities. On the basis of the physical examination
findings and the patient history, an Iowa hip
score15 was
calculated for each hip. In addition, the sum of the ranges of hip motion from
maximum flexion to maximum extension, maximum internal rotation to maximum
external rotation at both 90° of flexion and full extension, and maximum
abduction at both 90° of flexion and full extension was calculated.
A supine anteroposterior pelvic radiograph was made with a standard
tube-to-film distance of 40 in (101.6 cm) and was qualitatively analyzed for
the presence of subchondral sclerosis, degenerative cysts, and osteophytes
about the hip.
Each radiograph was oriented on the basis of a horizontal line between the
paired inferior points of the sacroiliac joints. Measurement of the vertical
superior hip joint space, in millimeters, was made on a line perpendicular to
the horizontal reference line, directly superior to the center of the femoral
head. The maximum horizontal distance between the medial acetabular line and
the ilioischial line (Kohler line) was measured as described by Armbuster and
colleagues16,17
and denoted the acetabular-ilioischial distance (Figs.
1-A, 1-B, and 1-C). When the
acetabular line protrudes medial to the ilioischial line, the
acetabular-ilioischial distance is called positive; when the acetabular line
is lateral to the ilioischial line, the acetabular-ilioischial distance is
called negative. The center-edge angle of Wiberg was measured as the angle
between a vertical line through the center of the femoral head and another
line connecting the center of the femoral head with the superolateral margin
of the
acetabulum18; an
increasingly positive centeredge angle of Wiberg indicates deepening of the
acetabular coverage over the femoral head (Figs.
1-A, 1-B, and 1-C).
Linear regression analyses were performed and Pearson correlation
coefficients were calculated to determine the relationships between the
center-edge angle of Wiberg and the acetabular-ilioischial distance and
between each of these radiographic measures of protrusio and patient age, the
Iowa hip score, the Iowa hip pain subscore, the summed range of motion of the
hip, and the superior joint space measured in millimeters. As multiple
simultaneous comparisons were to be performed (with five variables being
tested against each radiographic measure), the Bonferroni procedural
correction was used and the level of alpha was adjusted from 0.05 to 0.01 by
dividing it by the number of comparisons.
The relationship between radiographic protrusio measurements and age was
further investigated by stratifying patients by age according to decade and
using analysis of variance followed by the Tukey Studentized range test for
individual comparisons, with the level of alpha set at 0.05.
The presence of protrusio acetabuli was then defined according to two
previously described criteria: (1) a center-edge angle of Wiberg of
>50° (the Steel
method8), and (2) an
acetabular-ilioischial distance of =3 mm in male patients or =6 mm in
female patients (the Armbuster
method16). Effects
of the presence of protrusio on the Iowa hip score and the pain subscore were
tested with Student t tests.
Patients were then stratified according to age by decades. Chi-square
analysis was used to detect a significant difference in the prevalence of
protrusio according to the center-edge angle of Wiberg and
acetabular-ilioischial distance criteria. The Mantel-Haenszel chi-square test
was used to detect a rising trend. Follow-up comparisons to more precisely
define differences were tested with the Fisher exact test.
A power analysis was performed to determine adequate sample size for the
group of patients more than forty years of age. With the level of alpha set at
0.05, 80% power to detect a difference of 5 points on the Iowa hip score
between patients with and without radiographic protrusio was achievable with
152 hips, assuming a normal mean score of 95, a standard deviation of 10, and
a prevalence of protrusio of 30% (on the basis of the lowest estimate in the
literature6). We
therefore planned to include at least seventy-six patients with an age of more
than forty years.
The mean age (and standard deviation) of the 173 patients was 29.4 ±
15.5 years (range, three to sixty-four years). Eighty-three patients were
female, and ninety were male. One patient had a history of hip dysplasia that
had been treated with bracing during infancy and had radiographic evidence of
coxa breva. One patient previously had sustained a femoral neck fracture,
which had healed without complication. One patient had substantial hip pain
and had been recently told that she was a candidate for total hip arthroplasty
(Fig. 1-C). These three
patients were included in the analysis.
The mean center-edge angle of Wiberg for all patients was 46.5°
± 7.7°. The mean acetabular-ilioischial distance was 1.55 ±
2.69 mm. The mean Iowa hip score was 97.3 ± 7.5 (possible score, 100).
The mean Iowa hip pain subscore was 38.2 ± 5.0 (possible score,
40).
According to the Steel definition (a center-edge angle of >50°),
ninety-four hips (27%) had protrusio. According to the Armbuster definition
(an acetabular-ilioischial distance of =3 mm in male patients or =6 mm
in female patients), fifty-six hips (16%) had protrusio.
The acetabular-ilioischial distance was significantly and positively
correlated with the magnitude of the center-edge angle of Wiberg (r = 0.35; p
< 0.0001.)
Linear regression analyses for the entire study group demonstrated that
there was a significant, positive correlation between the magnitude of the
center-edge angle of Wiberg and age (p < 0.0001)
(Fig. 2) and between the
acetabular-ilioischial distance and age (p = 0.0003)
(Fig. 3). Although the
associations across all ages were positive and significant, the r values were
modest (0.21 and 0.19, respectively).
When the patients were stratified according to age in decades and the
presence of protrusio was defined according to the Steel method as a
center-edge angle of Wiberg of >50°, the Mantel-Haenszel chi-square
test demonstrated a significant trend of increasing prevalence with increasing
age (p = 0.0009) (Table II).
Subjective evaluation of the distribution table as well as the correlation
charts suggested a plateau in protrusio prevalence beginning in the third
decade. The Fisher exact test confirmed that, with the numbers available, no
differences in the prevalence of protrusio as defined with the Steel method
were detectable among the third through seventh-decade groups (p = 0.59) but
that the prevalence increased significantly from the first to the
second-decade groups and again to a pooled, third-through-seventh-decade group
(p = 0.0000049). This finding statistically confirmed the subjective
assessment that the prevalence of protrusio increased during the first two
decades and plateaued in the third decade. Identical analyses for protrusio as
defined with the Armbuster acetabular-ilioischial distance criteria
demonstrated nonsignificant trends in all of the same directions.
A slight but significant negative correlation was found between the Iowa
hip score and both the acetabular-ilioischial distance (p = 0.0001, r = -0.22)
(Fig. 4) and the center-edge
angle (p = 0.001, r = -0.18) (Fig.
5). These modest r values confirmed that while the degree of
protrusio as seen radiographically was associated with worsening hip function,
it accounted for only about 20% of the variance in the Iowa hip score, with
the other 80% of the variance being associated with other factors. Neither
radiographic measure of protrusio had a significant correlation with the pain
subscore of the Iowa hip rating, but both trended toward weak negative
correlations.
As significant but weak correlations were noted between age and the Iowa
hip score (r = -0.33, p < 0.0001) and between age and the pain subscore (r
= -0.19, p = 0.0008), each of the radiographic measures of protrusio was
tested again for correlation with the function and pain scores after
adjustment for age. Linear regression corrected for age yielded significant
but weaker negative Pearson partial correlation coefficients between age and
both the acetabular-ilioischial distance (p = 0.019, r = -0.13) and the
center-edge angle of Wiberg (p = 0.027, r = -0.12).
When protrusio was defined according to the Armbuster method, patients with
protrusio had a nonsignificant reduction in the Iowa hip score and the pain
subscore (p = 0.12 and 0.27, respectively). When protrusio was defined
according to the Steel method, patients with protrusio had a significant (p =
0.037) but clinically unimportant reduction in the Iowa hip score (difference,
2.6 points; 95% confidence interval, 0.8 to 4.4 points) and a nonsignificant
reduction in the pain subscore (p = 0.31).
When the comparison was repeated with the inclusion of only patients who
were more than forty years of age and protrusio was defined according to the
Armbuster method, hips with and without protrusio were not significantly
different in terms of either the Iowa hip score (p = 0.38) or the Iowa hip
pain subscore (p = 0.46). When the comparison was repeated with the inclusion
of only patients who were more than forty years of age and protrusio was
defined according to the Steel method, hips with and without protrusio were
not significantly different in terms of the pain subscore (p = 0.28) and hips
with protrusio had a nonsignificant trend toward a reduction in the Iowa hip
score (p = 0.078). Again, however, the difference between the means was only
4.7 points (95% confidence interval, 0.2 to 9.2 points).
Data regarding the radiographic measurements of protrusio for all twelve
patients with Iowa hip scores of <90 points are shown in
Table III. The summed range of
motion of the hip demonstrated a significant negative correlation with the
acetabular-ilioischial distance (p = 0.017, r = -0.14) and the center-edge
angle of Wiberg (p < 0.0001, r = -0.25) after correction for age. Joint
space also was negatively correlated with both the acetabular-ilioischial
distance (p < 0.0001, r = -0.30) and the center-edge angle of Wiberg (p
< 0.0001, r = -0.28) after correction for age. As for other signs of joint
degeneration, no patient had a cyst associated with either hip. Subchondral
sclerosis was not more frequent in hips with protrusio than in hips without
protrusio as defined according to either the Armbuster method or the Steel
method (p = 0.99 for both). Osteophytes were significantly more frequent in
the presence of protrusio as defined according to both the Armbuster method (p
= 0.012) and the Steel method (p = 0.0095).
Both the radiographic degree of protrusio and the prevalence of hip
symptoms were generally highly concordant between right and left hips in these
patients. The mean difference in acetabular-ischial distance between the two
hips was 1.7 ± 1.9 mm, with all but seventeen patients having a
difference of =3 mm. The center-edge angle of Wiberg differed between sides
by an average of 4.5° ± 4.0°, with all but nine patients having
a difference of =10°. The Iowa hip scores differed from one side to the
other by an average of 0.4 ± 2.0 point, with all but ten patients
having no difference at all in scores.
In previous studies, the radiographic prevalence of protrusio acetabuli in
patients with Marfan syndrome has ranged from 31% to 100%, but the largest of
those series included only twenty-two
patients6-8.
The present study of a much larger population of patients with Marfan syndrome
demonstrated a somewhat lower prevalence of 27% according to the criteria of
Steel8 and of only
16% according to the criteria of
Armbuster16.
The prevalence of protrusio acetabuli (as defined by a center-edge angle of
Wiberg of >50°) was found to increase in the first two decades of life
and then reached a plateau at a mean of 35.9% in patients with Marfan syndrome
who were twenty years of age or older. The relationship between the prevalence
of protrusio and age likely reflects the finding that protrusio develops
during growth and ceases to progress substantially after maturity. Just as
surgical closure of the triradiate cartilage retards the progression of
protrusio, it is likely that the natural progression of protrusio also is
arrested with skeletal maturity. As this was a cross-sectional study, we
cannot state whether individual patients with protrusio have progression with
age after skeletal maturity, but our data suggest that there is a plateau in
the prevalence and magnitude of protrusio with advancing age beyond twenty
years. While the age of twenty years represents an arbitrary plateau point
determined on the basis of our decade-stratified data, this age may
approximate the timing of skeletal maturity in this population as patients
with Marfan syndrome continue to have rapid growth later into the teenage
years than is the case in the general
population19.
Previous case reports have suggested that patients with Marfan syndrome are
at increased risk for hip degeneration and that this risk is associated with
protrusio
acetabuli7,8,10-13,20.
Prophylactic epiphyseodesis of the triradiate cartilage has been recommended
for patients with Marfan syndrome between the ages of eight and ten years who
have had previous progression of the center-edge angle of Wiberg, teardrop
alteration, diminishing range of motion of the hip, or any acetabular
medialization beyond the Kohler
line8. Others have
discussed combining triradiate physeal closure with a valgus intertrochanteric
osteotomy to prevent worsening of
protrusio14. In
contrast, the only previously reported screening study for protrusio in
individuals with Marfan syndrome, albeit one that involved only twenty-one
patients, demonstrated no correlation between the presence of protrusio
radiographically and clinical
symptoms6.
The present study provides limited information about the natural history of
protrusio as it was cross-sectional, not longitudinal, in design. Age-related
prevalence is a surrogate for true natural history, which can only be studied
directly in lengthy, expensive cohort studies. Nonetheless, we believe that
our study group was representative of the general population of patients with
Marfan syndrome because the patients were recruited from a medical genetics
clinic that serves as a primary-care resource for a large population. While
patients with orthopaedic complaints and manifestations of Marfan syndrome do
frequently present to our orthopaedic clinics first, all are routinely
referred for evaluation at the medical genetics clinic as well.
Our cross-sectional data did not demonstrate a strong association between
the presence of protrusio and hip problems. The significant correlations
between radiographic measures of protrusio and the Iowa hip score had very
small r values, suggesting that the presence of protrusio contributes little
to the prediction of hip symptomatology. Although we could not specifically
identify many of the factors that were responsible for the variance in hip
scores, we did find that age contributed significantly to declining hip
scores, as would be expected. The finding that hip scores varied minimally
from side to side in individual patients despite radiographic differences in
protrusio between the two hips reinforces the observation that the variance in
hip function is heavily affected by patient factors distinct from the degree
of protrusio.
Although the reduction in the mean Iowa hip score for hips with protrusio
as defined by a center-edge angle of Wiberg of >50° was significant, it
was not clinically important, with a magnitude of 2.6 points. Furthermore, the
95% confidence interval for this difference extended it to a maximum of 4.4
points. Therefore, this lack of a clinically meaningful difference is not
secondary to sampling error alone. A larger sample size should not change the
conclusion. This finding was reconfirmed in the group of patients who were
more than forty years of age. A power analysis was performed to determine the
sample size that would be required to permit the detection of a 5-point
difference in the Iowa hip score in this group. However, the confidence with
which we rejected the hypothesis that patients with protrusio have relatively
poor hip function was strengthened because even the 95% confidence interval
for the difference was <10 points (0.2 to 9.2 points). Specific clinically
important differences for the Iowa hip score have not been validated, but a
minimum difference of 10 points is probably conservative. For example,
patients without Marfan syndrome who are candidates for hip arthroplasty often
have scores of <60 on the Iowa
rating21.
Procedural interventions have demonstrated 25 to 30-point differences between
the preoperative and postoperative mean Iowa hip
scores22,23.
Some patients in this study had less than excellent hip function
(Table III). Many (but not all)
of these patients had protrusio acetabuli according to one of the two
radiographic criteria. However, there did not appear to be a threshold degree
of protrusio as measured radiographically that was associated with
consistently poor Iowa hip scores. Furthermore, some of these patients with
lower Iowa hip scores had little or no protrusio, suggesting that their hip
symptoms were more likely due to an unrelated process, degenerative or
otherwise.
As radiographic measures of protrusio did not correlate with the Iowa hip
pain subscore in either the entire study population or the subgroup of
patients who were more than forty years old, the slight differences in the
Iowa hip score are probably best explained by the reduced range of motion of
the hip associated with protrusio. Indeed, radiographic measurements of
protrusio correlated more strongly with the summed range of motion of the hip
than with the Iowa hip score overall. This relationship matches the anatomic
rationale that a deepening acetabulum will yield a reduction in the range of
motion but may be asymptomatic in these patients.
Similarly, the presence of a narrowed joint space and osteophytes usually
was not accompanied by subchondral sclerosis and degenerative cysts. Sclerosis
and cysts may be caused by the development of protrusio itself or by
femoroacetabular impingement on the neck, or they may be another manifestation
of deficient fibrillin in the cartilage or osseous tissues of the hip.
Alternatively, as forces are distributed over a larger area in a protruded
acetabulum, even osteoarthritic degeneration that develops may not be
associated with enough point and edge-loading to result in the development of
sclerosis and cysts.
The possibility that severe degrees of protrusio may cause substantial pain
and limitation that is worthy of early operative treatment in some patients is
corroborated by our preoperative analysis of patients with Marfan syndrome
undergoing total hip arthroplasty (unpublished data). However, the
population-based data from the present study suggest that such patients are
uncommon. We therefore do not support the practice of widespread prophylactic
epiphyseodesis for a condition that appears to be generally asymptomatic.
In summary, we found that medial acetabular protrusio is demonstrable in
nearly one-third of adults with Marfan syndrome, depending on the radiographic
criteria used. Its prevalence increases during skeletal growth and seems to
progress minimally (if at all) after skeletal maturity. Although increased
acetabular deepening was associated with a slight decrease in the Iowa hip
score, the great majority of hip scores in adults were very good. The decision
to perform hip surgery in patients with Marfan syndrome and protrusio should
be made on an individual basis and should be based on both clinical and
radiographic parameters. ?