Incidence and predictors of large-artery complication (aortic aneurysm, aortic dissection, and/or large-artery stenosis) in patients with giant cell arteritis: A population-based study over 50 years

Authors


Abstract

Objective

To determine the incidence and predictors of large-artery complication (aortic aneurysm, aortic dissection, and/or large-artery stenosis) in patients with giant cell arteritis (GCA).

Methods

The cohort of all residents of Olmsted County, Minnesota, in whom GCA was diagnosed between January 1, 1950, and December 31, 1999, was followed up. The incidence of aortic aneurysm, aortic dissection, and large-artery stenosis was determined. Possible predictors and correlates of large-artery complication were assessed.

Results

Forty-six incident cases of large-artery complication (representing 27% of the 168 patients in the cohort) were identified. These included 30 incident cases (18%) of aortic aneurysm and/or aortic dissection. Of these cases, 18 (11%) involved the thoracic aorta, with aortic dissection developing in 9 (5%). There were 21 incident cases (13%) of large-artery stenosis. Fifteen patients (9%) had incident cervical artery stenosis, and 6 (4%) had incident subclavian/axillary/brachial artery stenosis. One patient (0.6%) had incident iliac/femoral artery stenosis attributable to GCA. Hyperlipidemia and coronary artery disease were associated with aortic aneurysm and/or dissection (P < 0.05 for both). Cranial symptoms (headache, scalp tenderness, abnormal temporal arteries) were negatively associated with large-artery stenosis (hazard ratio [HR] 0.10 [95% confidence interval (95% CI) 0.03–0.35, P < 0.0005]), as was a higher erythrocyte sedimentation rate (HR 0.80 [95% CI 0.67–0.95, P < 0.05] per 10 mm/hour).

Conclusion

Large-artery complication is common in GCA. Increased awareness of large-artery complication in GCA, particularly early-occurring aortic dissection, may decrease associated mortality.

Giant cell arteritis (GCA) is an inflammatory vasculopathy that primarily affects medium-sized and large arteries. It is the most common vasculitis in populations with predominantly Northern European ancestry, with an annual incidence of 15–33 cases per 100,000 persons age 50 years or older (1–5). Although classically the temporal arteries or other cranial arteries were known to be involved in GCA (hence the name temporal arteritis), the aorta, with any of its primary or secondary branches, can be affected (6–9). Aortic aneurysm (6, 10, 11), aortic dissection (7, 8, 10, 11), and large-artery stenosis of the arm or leg arteries (12–14) or the cervical arteries (15) have been described in GCA. Evans et al (10) demonstrated in a population-based study that patients with GCA were 17.3 times more likely to develop a thoracic aortic aneurysm and 2.4 times more likely to develop an abdominal aortic aneurysm compared with the general population (16, 17).

Large-artery stenosis in GCA (12, 18, 19) can lead to intermittent claudication in the arms if the subclavian/axillary/brachial arteries are critically narrowed, or to ischemic neurologic symptoms such as transient ischemic attack (TIA) or stroke if the cervical arteries are critically narrowed. Until now, no population-based studies have addressed large-artery stenosis in GCA. Estimates of the incidence of large-artery stenosis in GCA have ranged from 18–21% for arm arteries (19), to 9–14% for any large-artery complication (13), to 7% for cerebrovascular ischemia due to GCA (15).

Previous studies either were unable to identify factors predictive of large-artery complication (10) or yielded conflicting findings (13, 14, 19). The identification of factors that are associated with or predict the development of large-artery complication may guide clinicians in stratifying patients for the risk of developing large-artery complication in GCA and in the cost-effective use of resources (e.g., use of screening imaging studies in high-risk patients).

PATIENTS AND METHODS

A retrospective cohort study was performed by chart review of the medical records of all residents of Olmsted County, Minnesota, in whom GCA was first diagnosed between January 1, 1950, and December 31, 1999. Patients were followed up through March 1, 2002, for a median of 7.6 years (interquartile range [IQR] 3.9–13.5 years). Residency in Olmsted County for at least 1 year prior to the diagnosis of GCA was required to avoid migration bias. The population of Olmsted County, Minnesota, is characterized by a balanced urban and rural distribution. The predominantly white population is representative of the white population in the US. Therefore, this population represents an ideal group in which to study GCA, a disease that affects predominantly persons of Northern European ancestry (1–5).

The unique resource of the Rochester Epidemiology Project allowed essentially complete case ascertainment of patients with GCA who came to medical attention in Olmsted County during the time period of interest (20). A total of 166 patients who fulfilled the 1990 American College of Rheumatology (ACR) classification criteria for GCA (21) were identified, as well as 2 additional patients with classic large-artery complication of GCA who did not formally meet the 1990 ACR classification criteria. Approval was obtained from the institutional review boards of both the Mayo Clinic and the Olmsted Medical Group.

One investigator (DMN) reviewed all medical records. A second investigator (ELM) reviewed 10% of the records in order to validate the data abstraction. Interobserver agreement in the classification of patients with both GCA and the particular kind of large-artery complication was assured. Classification questions were reviewed and adjudicated with a third investigator (GGH).

Each patient was assessed for any evidence of large-artery complication at any time. An incident large-artery complication was defined as a large-artery complication that was detected within 1 year prior to the diagnosis of GCA or thereafter. A large-artery complication was defined as aortic aneurysm, aortic dissection, and/or large-artery stenosis, including aortic arch syndrome, as determined by evidence on imaging studies (radiography, angiography, computed tomography, magnetic resonance imaging, ultrasonography), or at the time of surgery or autopsy. Physical examination findings alone were not sufficient to establish a diagnosis of large-artery complication.

A patient was classified as having an aortic aneurysm only when results of imaging studies were determined by a radiologist to be consistent with aneurysm, or when there was clear-cut evidence present at the time of surgery or autopsy. Both aortic dissection and aortic rupture were categorized as aortic dissection. Large-artery stenosis was defined as stenosis of any primary branch of the aorta with a size larger than that of the radial arteries, including the innominate, subclavian, axillary, brachial, common carotid, vertebral, and basilar arteries, or as stenosis of any unbranched secondary branch of the aorta, including the unbranched internal and external carotid arteries. The carotid arteries, vertebral arteries, and basilar arteries were classified as cervical arteries. Lower-extremity arterial stenosis was excluded unless there was classic angiographic evidence suggestive of vasculitis.

The incidence of aortic aneurysm, aortic dissection, and/or large-artery stenosis was calculated as the ratio of the number of observed events to the number of person-years at risk for the development of incident large-artery complication. Baseline demographic, clinical, and laboratory variables that were abstracted for analysis of their predictive value for the development of large-artery complication included sex, age, signs and symptoms at the time of diagnosis of GCA, erythrocyte sedimentation rate (ESR) at the time of diagnosis of GCA, and starting dose of glucocorticoids. Signs and symptoms at the time of diagnosis of GCA included cranial symptoms (headache; scalp tenderness; tender, swollen, nodular temporal arteries), jaw claudication, transient vision loss, permanent vision loss, diplopia, transient ischemic event or stroke, polymyalgia rheumatica–type symptoms (polymyalgia rheumatica, joint swelling, tenosynovitis), cardiopulmonary symptoms other than cough (dyspnea, orthopnea, rales, peripheral edema, palpitations, angina, chest pain), cough, fever, weight loss (>10 pounds over the past 6 months), diminished pulse or blood pressure and/or claudication of an arm, and aortic insufficiency murmur.

Possible correlates of large-artery complication in GCA were abstracted, including lifetime prevalence of stroke, coronary artery disease (CAD), peripheral vascular disease, hypertension, hyperlipidemia, smoking (ever smoked/never smoked), and diabetes mellitus. Possible predictive variables for the development of large-artery complication and its subtypes (aortic aneurysm, aortic dissection, and/or large-artery stenosis) were assessed using Cox proportional hazards regression. Possible correlates of large-artery complication and its subtypes were examined using chi-square tests or Fisher's exact tests. Predictive variables were defined as variables that must have been present prior to or at the time of the diagnosis of GCA, while correlates were defined as variables that could have been present at any time.

Results of the analysis for predictive factors and correlates for the entire inception cohort of 168 patients in whom GCA was diagnosed between January 1, 1950, and December 31, 1999, were compared with those for the inception cohort of 107 patients in whom GCA was diagnosed between January 1, 1980, and December 31, 1999. This comparison was performed to assess for possible ascertainment bias related to the introduction and more widespread use of imaging studies over the decades and to increasing disease suspicion bias.

RESULTS

Incidence of any large-artery complication.

A total of 46 incident cases (representing 27% of the 168 patients in the cohort) of any large-artery complication were identified (Figure 1), accounting for an incidence of 30.5 per 1,000 person-years at risk. In 5 patients (3%), concomitant incident aortic aneurysm and/or dissection and large-artery stenosis affecting the cervical arteries developed after the diagnosis of GCA.

Figure 1.

Incidence of large-artery complications in patients with giant cell arteritis. Five patients had both incident aortic aneurysm and/or dissection and incident large-artery stenosis affecting the cervical arteries.

Incidence of aortic aneurysm and/or dissection.

The incidence of aortic aneurysm and/or dissection was 18.7 per 1,000 person-years at risk. The incidence of thoracic aortic aneurysm was 8.2, that of thoracic aortic dissection was 5.4, that of abdominal aortic aneurysm was 10.1, and that of abdominal aortic dissection was 0.6 per 1,000 person-years at risk. A total of 30 incident cases (18%) of either aortic aneurysm and/or aortic dissection were identified. Of these, 18 cases (11%) involved the thoracic aorta; aortic dissection developed in 9 of these patients (5%) and was the cause of death in 7 (4%). Active aortitis was documented in 5 of the 7 cases of thoracic aortic dissection in which an autopsy was performed (Figure 2). Four patients (2%) had both incident thoracic aortic aneurysm and incident abdominal aortic aneurysm. One patient survived surgery for dissection of an abdominal aortic aneurysm and died later of an unrelated cause. Giant cell aortitis was not found on examination of pathology specimens obtained from this patient at the time of surgery. Ten (56%) of the 18 patients with thoracic aortic aneurysm and/or dissection had at least mild aortic insufficiency, and 7 (39%) had at least moderate aortic insufficiency, as documented by imaging studies.

Figure 2.

Incidence of aortic aneurysm and/or dissection in patients with giant cell arteritis. Four patients had both incident thoracic aortic aneurysm and incident abdominal aortic aneurysm.

The median time from diagnosis of GCA to detection was as follows: for thoracic aortic aneurysm and/or thoracic aortic dissection, 5.1 years (IQR 0.2–12.3 years); for abdominal aortic aneurysm and/or dissection, 6.3 years (IQR 1.1–13.3 years); for thoracic aortic aneurysm, 10.9 years (IQR 4.5–13.3 years); for abdominal aortic aneurysm, 6.3 years (IQR 1.1–13.3 years); for thoracic aortic dissection, 1.1 years (IQR 0.2–2.1 years); and for abdominal aortic dissection, 7.6 years.

Incidence of large-artery stenosis.

The incidence of large-artery stenosis was 13.5 per 1,000 person-years at risk. The incidence of cervical artery stenosis was 9.7, that of subclavian/brachial/axillary artery stenosis was 3.9, and that of lower-extremity artery stenosis was 0.6 per 1,000 person-years at risk. Incident large-artery stenosis was identified in 21 cases (13%). Fifteen patients (9%) had incident cervical artery stenosis. Six patients (4%) had incident subclavian/axillary/brachial artery stenosis. One patient (0.6%) had incident iliac/femoral artery stenosis due to GCA, with classic angiographic changes (Figure 3).

Figure 3.

Incidence of large-artery stenosis in patients with giant cell arteritis. One patient had both incident subclavian/axillary/brachial artery stenosis and incident cervical artery stenosis.

The median time from the diagnosis of GCA to the detection of subclavian/axillary/brachial artery stenosis was 1.1 years (IQR 0–8.3 years), and the median time from the diagnosis of GCA to the detection of cervical artery stenosis was 5.1 years (IQR 1.1–8.4 years). In the single patient with incident iliac/femoral artery stenosis due to GCA, the complication was diagnosed 0.7 years after the diagnosis of GCA.

Predictors and correlates of large-artery complication in GCA.

The symptoms and disease and laboratory characteristics of patients with GCA and different types of large-artery complication are shown in Tables 1 and 2. The time from first symptom onset to diagnosis of GCA was longest for the patients in whom aortic dissection developed (median time 100 days [IQR 33–201 days]), followed by patients in whom aortic aneurysm and/or dissection developed (median time 77 days [IQR 33–181 days], the group with large-artery stenosis (median time 39 days [IQR 23-100 days]), and the group without large-artery complication (median time 34 days [IQR 19–68]). Results of multivariate Cox proportional hazards regression assessing the possible predictor variables for large-artery complications are displayed in Table 3. Results of chi-square tests or Fisher's exact tests assessing possible correlates of large-artery complication are shown in Table 4.

Table 1. First symptoms and symptoms at time of diagnosis with giant cell arteritis, by presence of large-artery complication*
 Overall (n = 168)Aortic aneurysm and/or dissection (n = 30)Large-artery stenosis (n = 21)No large-artery complication (n = 122)
  • *

    Values are the number (%) of patients. GCA = giant cell arteritis.

First symptom    
 Polymyalgia rheumatica45 (26.8)6 (20.0)6 (28.6)35 (28.7)
 Abnormal temporal artery8 (4.8)1 (3.3)2 (9.5)5 (4.1)
 Scalp tenderness18 (10.7)1 (3.3)2 (9.5)15 (12.3)
 Headache88 (52.4)14 (46.7)13 (61.9)65 (53.3)
 Visual disturbance16 (9.5)2 (6.7)2 (9.5)13 (10.7)
 Jaw claudication18 (10.7)4 (13.3)1 (4.8)14 (11.5)
 Fever12 (7.1)1 (3.3)1 (4.8)10 (8.2)
 Weight loss >10 pounds19 (11.3)6 (20.0)3 (14.3)11 (9.0)
Symptom at GCA diagnosis    
 Aortic insufficiency murmur4 (2.4)2 (6.7)0 (0)2 (1.6)
 Diminished pulse or blood pressure5 (3.0)0 (0)2 (9.5)3 (2.5)
 Arm claudication1 (0.6)0 (0)1 (4.8)0 (0)
 Jaw claudication61 (36.3)15 (50.0)6 (28.6)43 (35.2)
 Headache129 (76.8)24 (80.0)15 (71.4)95 (77.9)
 Scalp tenderness112 (66.7)18 (60.0)12 (57.1)86 (70.5)
 Temporal artery abnormality122 (72.6)20 (66.7)11 (52.4)94 (77.0)
 Temporal artery tenderness90 (53.6)16 (53.3)9 (42.9)68 (55.7)
 Temporal artery swelling95 (56.5)16 (53.3)10 (47.6)71 (58.2)
 Temporal artery nodularity61 (36.3)13 (43.3)3 (14.3)46 (37.7)
 Vision changes37 (22.0)4 (13.3)6 (28.6)27 (22.1)
 Transient vision loss14 (8.3)3 (10.0)1 (4.8)10 (8.2)
 Permanent vision loss18 (10.7)0 (0)2 (9.5)16 (13.1)
 Diplopia14 (8.3)1 (3.3)4 (19.0)9 (7.4)
 Transitory ischemic attack/stroke5 (3.0)1 (3.3)2 (9.5)3 (2.5)
 Polymyalgia rheumatica symptoms52 (31.0)10 (33.3)4 (19.0)39 (32.0)
 Cardiopulmonary symptoms other than cough36 (21.4)8 (26.7)0 (0)28 (23.0)
 Cough23 (13.7)5 (16.7)0 (0)18 (14.8)
 Systemic, any94 (56.0)17 (56.7)13 (61.9)66 (54.1)
 Fever, >100°F50 (29.8)6 (20.0)5 (23.8)39 (32.0)
 Weight loss, >10 pounds17 (10.1)4 (13.3)1 (4.8)13 (10.7)
 Anorexia52 (31.0)13 (43.3)9 (42.9)32 (26.2)
 Malaise74 (44.0)19 (63.3)9 (42.9)48 (39.3)
Table 2. Disease characteristics and laboratory parameters of patients with GCA, by presence of large-artery complication*
CharacteristicOverall (n = 168)Aortic aneurysm and/or dissection (n = 30)Large-artery stenosis (n = 21)No large-artery complication (n = 122)
  • *

    Except where indicated otherwise, values are the median (interquartile range). GCA = giant cell arteritis; ESR = erythrocyte sedimentation rate.

ESR at time of diagnosis with GCA, mm/hour82.5 (60–105)81.5 (64–116)79 (58–97)84.5 (60–103)
Starting dose of prednisone, mg/day60 (50–60)60 (60–60)60 (60–60)60 (50–60)
No. (%) male35 (20.8)6 (20.0)4 (19.0)25 (20.5)
No. (%) female133 (79.2)24 (80.0)17 (81.0)97 (79.5)
Age at diagnosis with GCA, years75.6 (70.3–82.2)74.7 (70.7–83.1)75.2 (68.8–80.9)75.8 (70.3–82.3)
No. (%) of patients with cranial artery biopsy at time of diagnosis with GCA155 (92.3)27 (90.0)18 (85.7)114 (93.4)
Percentage of positive cranial artery biopsy results at time of GCA diagnosis96.1100.094.495.6
Time from onset of symptoms to diagnosis with GCA, days40 (21–89)77 (33–181)39 (23–100)34 (19–68)
Table 3. Possible predictor variables for large-artery complications in giant cell arteritis*
VariableAny large-artery complicationAortic aneurysm and/or dissectionLarge-artery stenosis
HR95% CIPHR95% CIPHR95% CIP
  • *

    Assessments were performed using multivariate Cox proportional hazards regression. The final model was the result of a stepwise search, with P < 0.10 required for entry of variables into the model. Variables assessed (other than those shown) were jaw claudication, transient vision loss, permanent vision loss, cardiopulmonary symptoms (other than cough), cough, fever, weight loss, starting dose of glucocorticosteroids, male sex, and age (per 10 years). HR = hazard ratio; 95% CI = 95% confidence interval; NA = not applicable (variable not included in the final model); ESR = erythrocyte sedimentation rate.

Aortic insufficiency murmur7.021.61–30.70.0109.302.04–42.30.004NANANA
Diminished pulse or blood pressure, armsNANANANANANA7.601.33–43.40.023
Cranial symptoms0.320.16–0.660.002NANANA0.100.03–0.350.0004
DiplopiaNANANANANANA4.031.09–14.80.036
Transient ischemic attack and/or stroke3.720.87–15.90.076NANANA6.391.21–33.80.029
Polymyalgia rheumatica symptomsNANANANANANA0.240.06–1.020.053
ESR, per 10 mm/hourNANANANANANA0.800.67–0.950.012
Table 4. Possible correlates of large-artery complication in giant cell arteritis*
VariableNo large-artery complication (n = 122)Any large-artery complication (n = 46)Aortic aneurysm and/or dissection (n = 30)Large-artery stenosis (n = 21)
No.%No.%PNo.%PNo.%P
  • *

    Percentages were calculated based on the number of persons having data available pertaining to the respective correlate, not necessarily on the overall number in the group. P values are versus no large-artery complication.

  • By Fisher's exact test; all other P values were determined by chi-square test.

Stroke2319.01737.00.015723.30.5951361.9<0.001
Coronary artery disease4235.32044.40.2811655.20.049838.10.805
Peripheral vascular disease2218.61021.70.653413.30.495733.30.147
Hypertension9779.53678.30.8592376.70.7321781.01.000
Hyperlipidemia5859.23378.60.0282482.80.0201477.80.135
Smoking, ever smoked/never smoked5550.02963.00.1361860.00.3311466.70.161
Diabetes mellitus1411.548.70.782310.01.00029.51.000

Predictors and correlates of aortic aneurysm and/or dissection.

An aortic insufficiency murmur at the time of diagnosis of GCA was predictive of aortic aneurysm and/or dissection (hazard ratio [HR] 9.30 [95% CI 2.04–42.3, P < 0.005]). Hyperlipidemia and CAD at any time were associated with aortic aneurysm and/or dissection: 82.8% of patients with aortic aneurysm and/or dissection had hyperlipidemia, compared with only 59.2% of patients without large-artery complication (P < 0.05), and 55.2% of patients with aortic aneurysm and/or dissection had CAD, compared with only 35.3% of patients without large-artery complication (P < 0.05).

Predictors and correlates of large-artery stenosis.

Cranial symptoms (headache, scalp tenderness, and/or swollen or nodular arteries) at the time of diagnosis of GCA were negatively associated with large-artery stenosis (HR 0.10 [95% CI 0.03–0.35, P < 0.0005]). Diminished pulse or blood pressure and/or claudication of an arm, TIA or stroke, and diplopia at the time of diagnosis of GCA were associated with large-artery stenosis (for diminished pulse or blood pressure and/or claudication of an arm, HR 7.60 [95% CI 1.33–43.4, P < 0.05]; for TIA or stroke, HR 6.39 [95% CI 1.21–33.8, P < 0.05]; and for diplopia, HR 4.03 [95% CI 1.09–14.8, P < 0.05]). A high ESR was negatively associated with large-artery stenosis (HR 0.80 [95% CI 0.67–0.95, P < 0.05] per 10 mm/hour), and polymyalgia rheumatica–type symptoms showed a trend for being negatively associated with large-artery stenosis (HR 0.24 [95% CI 0.06–1.02, P = 0.05]).

Stroke at any time was the only comorbidity that was associated with large-artery stenosis (61.9% of patients with large-artery stenosis experienced a stroke, compared with only 19% of patients without large-artery complication [P < 0.001]).

Large-artery complication in GCA over the decades.

The apparent incidence of large-artery complication increased over the decades, reaching a plateau for any kind of large-artery complication in the 1980s and 1990s (Figure 4). The Cox proportional hazards model for predictors of large-artery complication and chi-square tests or Fisher's exact tests for correlates of large-artery complication showed similar results for both the 2-decade and the 5-decade inception cohorts (data not shown).

Figure 4.

Number of patients with large-artery complications per person-years of followup (f/u), by decade in which giant cell arteritis was diagnosed. Bars denoting any large-artery complication are independent (cases are represented only once). Five patients had both incident aortic aneurysm and/or dissection and incident large-artery stenosis affecting the cervical arteries; these cases are represented in both the aortic aneurysm and/or dissection and the large-artery stenosis bars, within the same decade.

DISCUSSION

To our knowledge, this is the first population-based study of a cohort of patients with GCA to determine the incidence and predictors of both aortic aneurysm and/or dissection and large-artery stenosis in GCA.

Aortic aneurysm and/or dissection or large-artery stenosis developed in a substantial proportion of patients (27%). No patient had both large-artery stenosis of the arm arteries and aortic aneurysm and/or dissection. The 5 patients in whom concomitant large-artery stenosis and aortic aneurysm and/or dissection developed had cervical artery stenosis. Concomitant large-artery stenosis of the arm arteries and aortic aneurysm and/or dissection have only rarely been reported in the literature (6, 13, 22), and have not been observed in other studies (11, 14). Our observations support the opinion that aortic aneurysm and large-artery stenosis affecting the arm arteries may represent distinct, almost mutually exclusive presentations of GCA.

Aortic aneurysm and/or aortic dissection developed in almost 1 of every 5 patients (18%). In 4 patients, thoracic aortic dissection developed in the absence of documented aortic aneurysm. Thoracic aortic dilatation or ectasia was documented in only 2 of these 4 patients. If the observation is true that aortic dissection can occur in GCA in the absence of aortic aneurysm or even dilatation or ectasia, this has important implications. It is likely that most, if not all, patients with GCA have aortic inflammation at some time point (19, 22, 23). Persistent giant cell aortitis may increase the risk of aortic dissection and rupture even in the absence of an aortic aneurysm. Assessment for persistent giant cell aortitis should then go beyond that for possible aneurysm development only, and should include assessment for possible inflammatory wall changes. Contrast-enhanced computed tomography, magnetic resonance imaging, or positron emission tomography may be of use to evaluate for possible inflammatory aortic wall changes, although any of these techniques may lack sensitivity and specificity (23–27). Adequate glucocorticoid therapy may need to be administered in cases of persistent giant cell aortitis, even in the absence of an aneurysm, to prevent the occurrence of aortic dissection and rupture.

Whereas thoracic aortic dissection appears to be an early complication in GCA, thoracic aortic aneurysm is usually detected later. This suggests that the observed increased mortality related to aortic dissection is likely mostly attributable to inadequate suppression of giant cell aortitis earlier in the course of the disease.

Large-artery stenosis developed in 1 of every 8 patients (13%). In this series, large-artery stenosis affected the cervical arteries in the majority of patients (9%). The incidence of large-artery stenosis affecting the arm arteries in our study (4%) was lower than that reported by either Hamrin (18–21%) (19) or Klein et al (9–14%) (13). Part of the difference may have been related to referral bias in the respective studies, and, in Hamrin's study (19), may additionally be attributable to a prospective study design that could have resulted in more complete case ascertainment.

Subclavian/axillary/brachial artery and iliac/femoral artery stenosis appear to be early complications of GCA. The later diagnosis of cervical artery stenosis may reflect the fact that some patients with stenosis related to arteriosclerosis rather than to GCA may have been included in this group.

The longer time from onset of symptoms to diagnosis of GCA in patients in whom aortic dissection developed, with the associated delay in adequate treatment of GCA with glucocorticoids, likely contributed to the development of aortic dissection. A delay in diagnosis with GCA and delay in treatment were also seen in patients in whom aortic aneurysm developed, although to a lesser extent. Aortic insufficiency related to aortic root dilatation with thoracic aortic aneurysm explains the association of an aortic insufficiency murmur at the time of the diagnosis of GCA with aortic aneurysm and/or dissection.

Although hyperlipidemia at any time was associated with aortic aneurysm and/or dissection, it is unclear whether hyperlipidemia is an independent risk factor for the development of aortic aneurysm in GCA. Because only the lifetime prevalence of hyperlipidemia was ascertained, hyperlipidemia was not included in the Cox proportional hazards model and was not assessed for possible interactions. The higher degree of atherosclerosis related to hyperlipidemia resulting in vessel wall weakening may explain the association of hyperlipidemia with aortic aneurysm and/or dissection. Statin drugs, because of their antiinflammatory and vasoprotective properties (28), may have a role in the treatment of GCA, but this has not yet been the subject of a clinical trial.

The reason for the association of CAD with aortic aneurysm and/or dissection is unknown, but the association could be related to an increased incidence of coronary arteritis in GCA. This association may, however, reflect only an increased prevalence of arteriosclerosis in patients in whom aortic aneurysm and/or dissection develop. The use of low-dose aspirin in all patients with GCA needs to be considered, in order to decrease the risk of ischemic complications related to GCA. Aspirin use in GCA has not been rigorously studied in a clinical trial.

The finding that cranial symptoms at the time of the diagnosis of GCA were negatively associated with large-artery stenosis was in concordance with a previous report by Brack et al (14). Because the presence of cranial symptoms increases the likelihood that patients will seek medical attention, it is likely that GCA will be diagnosed earlier in patients in whom cranial symptoms develop, and that these patients will receive prompt treatment with adequate doses of glucocorticoids. Such early diagnosis and treatment may prevent the development of large-artery stenosis.

The association between TIA or stroke at the time of diagnosis of GCA and large-artery stenosis can be explained by large-artery stenosis affecting the cervical arteries, resulting in ischemic cerebrovascular symptoms. Similarly, the association of diplopia with large-artery stenosis is explained by symptomatic large-artery stenosis affecting cervical and/or cranial arteries. Symptoms related to large-artery stenosis of the arm arteries explain the association between a diminished pulse or blood pressure and/or claudication of an arm at the time of diagnosis of GCA and large-artery stenosis.

The finding that a higher ESR was negatively associated with large-artery stenosis is concordant with the observations of Klein et al (13) and Brack et al (14), and is contrary to the findings reported by Hamrin (19). Because patients in whom large-artery stenosis developed were less likely to present with cranial symptoms, they were probably diagnosed as having GCA later in the course of the disease, when maximum inflammation had already subsided or had been suppressed by glucocorticoid use. In the current study, this was not related to possible glucocorticoid use at the time of the diagnosis of GCA. There was no interaction between glucocorticoid use at the time of diagnosis of GCA and the ESR, based on examination of predictors of large-artery stenosis in the Cox proportional hazards model (data not shown). The discordance with Hamrin's observations (19) may be attributable to a different proportion of patients with classic presentation of GCA and patients who came to medical attention primarily because of symptoms caused by large-artery stenosis. Hamrin also recorded the highest ESR for each patient, regardless of when during followup this value was obtained (i.e., not only at the time of diagnosis of GCA), which may further explain the differences.

Similar to cranial symptoms, polymyalgia rheumatica–type symptoms may lead to earlier detection and treatment of GCA, which explains the trend toward a negative association of polymyalgia rheumatica–type symptoms with development of large-artery stenosis, a finding that contrasts with the report by Brack et al (14). The association of stroke at any time with large-artery stenosis can be explained by the same mechanism as that involved in the association of TIA or stroke and diplopia at the time of diagnosis of GCA with large-artery stenosis. These complications can be caused by symptomatic stenosis of cervical and cranial arteries related to GCA.

The increasing apparent incidence of large-artery complications over the decades, which reached a plateau for any kind of large-artery complication in the 1980s and 1990s, is likely explained by ascertainment bias due to the development and more widespread use of new imaging modalities. Because the separate analysis of possible predictive factors and correlates of large-artery complication in GCA for the last 2 decades revealed associations similar to those for the full 5-decade period, ascertainment bias does not appear to have significantly influenced the respective results and conclusions.

Our study was not designed to clearly distinguish the etiology of large-artery complications, although active aortitis was documented in 5 of 7 cases of thoracic aortic dissection in which autopsy was performed. Particularly in the cases of cervical artery stenosis, it is uncertain which proportion was related to GCA, because cervical artery arteriosclerosis is common among elderly persons. Histopathologic documentation of GCA was available in only a few patients (e.g., for 1 patient who underwent a carotid endarterectomy). Even if biopsy specimens had been available for each patient in whom cervical artery stenosis was diagnosed, the absence of GCA at the time tissue was obtained would not have entirely excluded GCA as the cause of prior stenosis. The closer the time of detection of large-artery stenosis was to the date of the diagnosis of GCA, the more likely it appears that this complication was, in fact, related to GCA. The same appears to be true for the detection of aortic aneurysm. For this reason, a Cox proportional hazards model was chosen to explore possible predictive factors at the time of diagnosis of GCA. This model led to higher HRs for possible predictive factors, the closer the detection of large-artery complication was to the time of diagnosis of GCA.

This study was further limited by possible diagnostic accuracy bias in assessing for the presence of either aortic aneurysm or large-artery stenosis. Inherent to the nature of a retrospective cohort study, there was no standardized evaluation process to assess for the presence of either aortic aneurysm or large-artery stenosis in these patients with GCA. This lack of standardization will have led to underascertainment of subclinical cases. No uniform criteria (29) were used by the multiple evaluating radiologists or pathologists over the decades, which will have led to intraobserver and interobserver variability. The incidence of large-artery stenosis of the lower extremities due to GCA was certainly underestimated, because classic angiographic changes suggestive of vasculitis were required for its classification. We chose this approach because arteriosclerotic peripheral vascular disease of the lower extremities is common in the elderly.

Most of the factors predictive of large-artery stenosis that were identified in this study were expected. Diminished pulse or blood pressure and/or claudication of an arm, TIA or stroke, and diplopia were all predictive of large-artery stenosis, and an aortic insufficiency murmur was predictive of aortic aneurysm and/or aortic dissection. The presence of any of these symptoms should prompt further evaluation for possible large-artery complication. The negative association of cranial symptoms and a higher ESR at the time of diagnosis of GCA with large-artery stenosis, and the borderline association of polymyalgia rheumatica–type symptoms with large-artery stenosis might somewhat help in risk-stratifying patients with GCA for the development of large-artery complication. However, these predictors were not exclusive enough to establish a clear-cut risk stratification tool and guidelines for monitoring of patients with GCA for large-artery complications. Standardized, prospective, large-scale studies are needed to better determine factors predictive of large-artery complications.

Ancillary