To define further the effect of previous corticosteroid treatment on temporal artery biopsy results, we reviewed the clinical and histologic findings in a consecutive cohort of 535 patients who had temporal artery biopsy at Mayo Clinic during a 4-year period. We compared biopsy positivity rates among those who received corticosteroid therapy before temporal artery biopsy with rates among those who did not, adjusting for clinical and laboratory characteristics.

Methods

Top Methods Statistical Analysis Results Discussion Author & Article Info References

Patients

We reviewed the medical records and histologic temporal artery biopsy specimens of all patients who had temporal artery biopsy at the Mayo Clinic, Rochester, Minnesota, between 1 January 1988 and 31 December 1991. During this time, 545 patients had temporal artery biopsy. Ten patients were excluded for the following reasons: unavailable records (n = 1), unavailable slides (n = 4), systemic non-giant cell arteritis (n = 3), juvenile (at age 8 years) temporal arteritis with dissection (n = 1), and first temporal artery biopsy done at the Mayo Clinic before 1 January 1988 (n = 1).

Therefore the study sample consisted of 535 patients. We used a standardized data collection form to record information, including the date of biopsy and the dose and duration of corticosteroid treatment received before biopsy. Information recorded from the history included age, sex, duration of symptoms, and the presence (during the month preceding temporal artery biopsy) of general malaise, fever, chills, sweats, weight loss, new headache, jaw or tongue claudication, blurred vision, amaurosis fugax, diplopia, polymyalgia rheumatica, morning stiffness, arthralgias, sore throat, and cough. Information recorded from the physical examination included the presence of ischemic optic neuritis, optic atrophy, decreased visual acuity, tender temporal artery, decreased or absent temporal artery pulse, scalp nodules or tenderness, synovitis, and large-vessel bruits. Recorded laboratory information included the erythrocyte sedimentation rate, hemoglobin concentration, platelet count, and the aspartate aminotransferase, alkaline phosphatase, albumin, and serum -2 globulin levels.

Temporal Artery Biopsies

The usual practice at the Mayo Clinic is to biopsy the temporal artery in patients with suspected giant cell arteritis when that diagnostic question arises. A 3- to 4-cm specimen of the more clinically suspicious temporal artery is removed and frozen sections are examined at multiple levels. If frozen sections of this temporal artery show no vasculitis, a biopsy specimen from the opposite temporal artery is usually taken at the same time and examined at multiple levels. Permanent sections are also examined 24 hours later. An experienced pathologist who was blinded to the clinical data, previous corticosteroid treatment information, and the original pathologic diagnosis re-examined all biopsy slides. We classified the biopsy results as negative (no evidence of arteritis) or positive (histologic evidence of arteritis). We further classified positive biopsy results as typical temporal arteritis or atypical temporal arteritis [12, 13]. Typical temporal arteritis Figure 1 denotes granulomatous arteritis with one or more giant cells present in a cross section of the artery. The inflammatory infiltrate is a mixed cell type with mononuclear cells (lymphocytes, histiocytes, and plasma cells) predominating. Atypical temporal arteritis Figure 2 denotes the presence of inflammation (consistent with giant cell arteritis) but with atypical features such as the absence of giant cells or the occurrence of the inflammatory infiltrate mainly in the adventitia rather than in the media, the more typical location.

View larger version (116K): [in this window] [in a new window]   Figure 1. Photoµgraph of typical granulomatous temporal arteritis (left) with giant cells (arrow) seen in a close view (right) (hematoxylin and eosin; left, x 64; right, x 400). (Reproduced with permission of the Journal of Rheumatology [13].).

View larger version (162K): [in this window] [in a new window]   Figure 2. Hematoxylin and eosin (top left) and elastic stain (top right) photoµgraphs of a positive temporal artery biopsy specimen without identifiable giant cells. The predominantly lymphocytic inflammatory infiltrate is evident in a close view (bottom) (hematoxylin and eosin: top left, x 64; bottom, x 160; elastic stain: top right x 64). (Reproduced with permission of the Journal of Rheumatology [13].).

Statistical Analysis

Top Methods Statistical Analysis Results Discussion Author & Article Info References

We assessed the reliability of the histologic interpretation by comparing the repeated histologic interpretation used in the study with that of the original pathologist using the statistic [14]. We used a chi-square analysis (or Fisher exact test in cases of small cell sizes) to compare the biopsy positivity rates among patients with various doses and durations of corticosteroid therapy before temporal artery biopsy. To assess the influence of multiple independent variables, univariably and multivariably, we used logistic regression, incorporating the temporal artery biopsy result (positive or negative) as the dependent variable [15]. We analyzed all recorded clinical and laboratory variables univariably. We used stepwise methods to develop a "final" logistic model in which only statistically significant (P < 0.05) independent variables and interactions among them were retained. Then we used the final best logistic model to test the null hypotheses that previous corticosteroid treatment does not affect the temporal artery biopsy positivity rate, adjusting for the influence of the independent variables in the model.

Results

Top Methods Statistical Analysis Results Discussion Author & Article Info References

Study Sample

Of 535 patients who had biopsies, 345 (64%) were women and 190 (36%) were men. Their mean age was 71.7 years (range, 31 to 93 years). The study sample included 46 (9%) patients from Olmsted County, Minnesota (local), and 489 (91%) patients from other areas. Many patients had been given corticosteroid therapy before referral to the Mayo Clinic, especially those receiving treatment for longer periods before temporal artery biopsy. There were 286 (53%) untreated patients and 249 (47%) patients who received corticosteroid therapy before temporal artery biopsy.

Patients in the study generally had some clinical indication suggesting arteritis that prompted the clinician to order a temporal artery biopsy. For example, 60% (322 of 535 patients) had a new headache, and 73% (393 of 535 patients) satisfied American College of Rheumatology Criteria for classification as giant cell arteritis [16]. Patients who did not satisfy these criteria generally had symptoms and findings of a systemic illness that suggested possible giant cell arteritis.

Table 1 shows clinical and laboratory features present in corticosteroid-treated and untreated patients before temporal artery biopsy. Many features occurred frequently among both corticosteroid-treated and untreated patients, including headache, visual symptoms or ocular findings, temporal artery abnormalities, polymyalgia rheumatica, and an erythrocyte sedimentation rate greater than 40 mm/h. Patients who received more than 15 mg/d of prednisone for more than 14 days before temporal artery biopsy, when compared with untreated patients, had equal or higher frequencies of many symptoms, including headache (69% compared with 55%) and visual symptoms or ocular findings (63% compared with 25%). Patients in this corticosteroid subgroup had slightly less frequent temporal artery abnormalities (38% compared with 49%) and a lower median erythrocyte sedimentation rate (40 mm/h compared with 72 mm/h) than did untreated patients.

Overall Biopsy Results

Positive biopsy results were found in 175 (33%) patients, and negative results were found in 360 (67%) patients. Of the 175 patients with positive biopsy results, 86 (49%) had typical temporal arteritis and 89 (51%) had atypical temporal arteritis according to histologic examination. The reviewed histologic diagnosis used in the study correlated with the original pathologist's interpretation in 94% of cases (estimated = 0.87). Mean biopsy specimen length was similar in patients with positive (3.7 cm) and negative (3.6 cm) results. Bilateral temporal artery biopsy specimens were taken in 314 (59%) patients.

Relation between Previous Corticosteroid Treatment and Biopsy Results

Table 2 shows the relation between corticosteroid treatment before temporal artery biopsy and biopsy results. The biopsy results were positive for 31% (89 of 286) of patients who did not receive corticosteroids before biopsy and for 35% (86 of 249) of those who did receive corticosteroids before biopsy (P = 0.4; 95% CI for the difference, –4.7% to 11.5%). We grouped patients treated with more than 15 mg/d of prednisone or an equivalent dose of a different corticosteroid (n = 149) according to treatment duration before biopsy, and biopsy results were positive in 43% at 1 to 7 days (46 of 107 patients), 30% at 8 to 14 days (3 of 10 patients), and 28% at more than 14 days (9 of 32 patients).

Patients who received more than 15 mg of prednisone per day for 1 to 7 days before temporal artery biopsy more often had a positive result than did untreated patients (43% compared with 31%; P = 0.027) (Table 2). Thus, the ratio of the odds of a positive biopsy result for patients in this treatment subgroup to the odds of a positive result for those who received no previous corticosteroid treatment was 1.67 (CI, 1.06 to 2.64). However, this 1- to 7-day treatment subgroup tended to have more frequent classic giant cell arteritis symptoms or signs (such as jaw claudication or tender or pulseless temporal artery) than did untreated patients (Table 1). Among the clinical and laboratory findings, the final logistic model (see Statistical Analysis) included synovitis, claudication (involving jaw, arm, or tongue), temporal artery abnormalities (palpably abnormal), increased platelet count, elevated erythrocyte sedimentation rate, and an interaction between these last two variables. After adjusting for these variables, the adjusted odds ratio comparing this treatment subgroup (>15 mg/d for 1 to 7 days) with patients who received no previous corticosteroid treatment became 1.23 (CI, 0.72 to 2.12). Similar analyses comparing other subgroups of patients who received corticosteroid treatment with untreated patients also showed no significant differences in biopsy positivity rate.

Notably, of the 32 patients who received more than 15 mg of prednisone per day for more than 14 days before temporal artery biopsy, 9 (28%) had a positive biopsy result. The 9 patients with positive biopsy results received prednisone (prescribed elsewhere before referral) for more than 14 days before biopsy because of suspected giant cell arteritis (n = 6), polymyalgia rheumatica (n = 2), and "orbital pseudotumor" (n = 1). The longest duration of corticosteroid treatment (> 15 mg/d of prednisone) before a positive biopsy was 11 months.

Table 2 also shows the proportions of positive biopsy results that had atypical temporal arteritis histologic characteristics, determined for each corticosteroid treatment subgroup. There was a higher proportion of atypical temporal arteritis histologic features among the 9 biopsy-positive patients who received more than 15 mg/d of corticosteroid for more than 14 days than in untreated patients (8 of 9 compared with 40 of 89; P = 0.012). In addition, these data suggested a trend toward an increasing proportion of atypical biopsy results with longer duration of corticosteroid treatment.

Discussion

Top Methods Statistical Analysis Results Discussion Author & Article Info References

Although these results do not prove that histologic features are unaffected by corticosteroids, they show that, in this large consecutive sample, the positivity rate for temporal artery biopsy was similar in untreated and corticosteroid-treated patients. Temporal artery biopsy specimens may show arteritis even after more than 14 days of corticosteroid therapy in the presence of clinical indications of active disease.

Patients who received corticosteroids before biopsy tended to have clinical characteristics more suggestive of arteritis. The frequency of new headache, jaw claudication, visual symptoms or ocular findings, and palpable temporal artery abnormalities was greater in corticosteroid-treated than in untreated patients (see Table 1), which may reflect clinician selection of patients for corticosteroid treatment before biopsy. We used a multiple logistic regression analysis model to control for differences in presenting clinical and laboratory features in untreated compared with corticosteroid-treated patients and found that biopsy positivity rate was still not related to previous corticosteroid treatment.

Patients who received less than 15 mg/d of prednisone before temporal artery biopsy were generally being treated for polymyalgia rheumatica or other conditions during the period immediately before biopsy. Patients receiving treatment of more than 15 mg/d received different and sometimes fluctuating corticosteroid doses before biopsy, but these were usually about 40 to 60 mg/d. These patients were most often being treated for presumed giant cell arteritis. Because low doses of prednisone are thought to be inadequate treatment for giant cell arteritis, we separately analyzed patients who received more than 15 mg/d of prednisone before biopsy.

Histologic arteritis could still be detected in some patients who received more than 15 mg/d of prednisone for more than 14 days before biopsy. In fact, the biopsy positivity rate was not substantially different between this subgroup (> 15 mg/d for more than 14 days) and untreated patients. However, the proportion of positive biopsy specimens that showed atypical temporal arteritis histologic features was significantly higher (P = 0.012) (see Table 2) in this subgroup of patients who received higher doses for more than 14 days than in untreated patients. In addition, there was a trend toward a higher proportion of atypical biopsy specimens [such as those without giant cells] with longer duration of corticosteroid treatment. These data suggest that the histologic features of temporal artery biopsy specimens are probably modified by higher-dose, longer-duration corticosteroid therapy, even though the histologic evidence of arteritis was still detectable.

Patients who received more than 15 mg/d of prednisone for 1 to 7 days before temporal artery biopsy had a higher biopsy positivity rate and more frequent classic symptoms and signs of giant cell arteritis (jaw claudication, pulseless or tender temporal artery) than did untreated patients. When differences in clinical and laboratory features were accounted for using the multiple logistic regression analysis model, no significant difference occurred in the biopsy positivity rate between this 1- to 7-day treatment subgroup and untreated patients.

Our study had the limitations of a retrospective study. Probably, differences occurred in untreated compared with corticosteroid-treated patients that were not controlled for by randomization. We used a logistic model to control for differences in presenting symptoms in untreated compared with corticosteroid-treated groups. However, differences may remain between these groups that are not controlled for by such an analysis. In addition, whenever information is derived from subgroups of any study sample (such as in our study, patients treated with more than 15 mg/d of prednisone before biopsy), this information generally should be interpreted with caution.

Our data do not exclude the possibility that a given biopsy-negative patient who received previous corticosteroids might have shown histologic features of arteritis if a biopsy specimen had been obtained before initiation of corticosteroids. Such information would require biopsy specimens taken before and after treatment. Even then, the results may be difficult to interpret because patients with positive pretreatment biopsy results may have negative post-treatment results that are not due to treatment but rather to sampling of "skip areas" [9, 10].

For this study, we did not collect data on postbiopsy treatment and outcome. However, this information was studied previously at our institution [17]. In that population-based, retrospective cohort study, patients with negative biopsy results were followed for a median of 6 years after biopsy. Only 9% of all patients with negative biopsy results later required corticosteroids for biopsy-negative temporal arteritis. None of those patients lost his or her vision. At our institution, when patients have long bilateral biopsy specimens examined in multiple sections that yield negative results, corticosteroids are usually withheld. In these instances, patients are monitored and other causes for their clinical symptoms are sought. However, some patients with negative biopsy results and clinical findings suggestive of arteritis (without other explanations), especially vision loss, are treated with corticosteroids.

Previous studies provided conflicting results regarding the relation between previous corticosteroid treatment and histologic features of temporal artery biopsy specimens. Allison and Gallagher [5] found a positive biopsy result in 82% of untreated patients, in 60% of patients who received corticosteroid therapy for 1 week or less before temporal artery biopsy, and in 10% of patients who received corticosteroid therapy for more than 1 week before temporal artery biopsy. These investigators questioned the value of the biopsy in patients who received high doses of corticosteroid for more than 1 week before biopsy. The biopsy specimens in their series were smaller (mean length, 0.7 cm) than in the current study (mean length, 3.6 cm). Allison and Gallagher [5] speculated that examining larger biopsy specimens "might yield evidence of more residual foci of active inflammation or the changes of healed arteritis." This correlates with the earlier observation that "skip lesions" are common in giant cell arteritis [9, 10]. Although 59% of the patients in our study had bilateral biopsy specimens, no information regarding bilateral specimens was reported by Allison and Gallagher [5]. They included in their study only "clinically genuine" cases (n = 132) of giant cell arteritis (in all cases, patients were subsequently treated as having giant cell arteritis by their clinicians despite a negative biopsy result in 48 patients). In contrast, the current study included all patients having temporal artery biopsy during the study period (n = 535), regardless of the subsequent diagnosis or management. This probably explains the higher overall biopsy positivity rate (84 of 132 patients [64%]) in the study by Allison and Gallagher compared with our study (175 of 535 patients [33%]) and with previously published population-based estimates of temporal artery biopsy positivity rates (41% in women and 26% in men, with a weighted average of 38%) [18].

McDonnell and coworkers [6] reviewed 250 temporal artery biopsy specimens from 237 patients, of which 42 (17%) were positive. They found that temporal artery biopsy specimens showed evidence of "active arteritis" after a mean of 7 days (n = 32) of corticosteroid therapy and "healed arteritis" after a mean of 82 days (n = 10) of corticosteroid therapy. In their study, fewer patients had bilateral biopsies (5%) than in our study (59%). Also, smaller biopsy specimens were obtained in their study (mean length, 1.5 cm) than in our study (mean length, 3.6 cm).

Chmelewski and associates [7] reviewed temporal artery biopsy results from 98 patients, of which 30 (31%) were positive. In their series, 6 of 16 patients (37%) who received more than 2 weeks of corticosteroid treatment before biopsy had a positive biopsy result compared with a biopsy positivity rate of 19 of 65 (29%) in patients who received 0 to 2 days of corticosteroid before biopsy (P > 0.05). The proportion of positive biopsy specimens showing atypical temporal arteritis histologic characteristics was not reported. In their study, 5% of patients had bilateral biopsy specimens that were smaller (mean length, 2 cm) than those obtained in our study.

In addition, other cases are reported [8, 19, 20] in which a repeated biopsy was still positive several months to several years after an initial positive temporal artery biopsy. However, the corticosteroid regimens in these patients were not consistently described and appear to have included periods without corticosteroid therapy and other periods with low corticosteroid doses.

Neither our study nor previous studies that provided data on effects of corticosteroid therapy on temporal artery biopsy histologic characteristics were population based. In our study sample, no patients from Olmsted County, Minnesota (the population-based subset of our study), received more than 15 mg/d of prednisone for more than 7 days before biopsy (data not shown). Thus, we had no information about how prolonged corticosteroid treatment affected biopsy findings in our population-based patient subset. Our study included larger biopsy specimens and more bilateral specimens than did previous studies, thus ensuring against false-negative biopsy results.

Also, to our knowledge, our study is the largest of its kind in the reported literature. The sample size yielded a 95% CI of –4.7% to 11.5% for the difference in biopsy positivity rates between corticosteroid-treated patients and untreated patients. Although this CI includes nontrivial differences, it suggests that the positivity rate of biopsy results in patients who received previous corticosteroids is probably not substantially lower (that is, more than 4.7%) than that of untreated patients.

In summary, although corticosteroid treatment may modify temporal artery biopsy histologic characteristics, our data in this large consecutive sample show that the positivity rate for temporal artery biopsy was similar in untreated and corticosteroid-treated patients. These data show that if clinical indications of giant cell arteritis are present, it is reasonable to do a temporal artery biopsy to confirm arteritis, even in patients who received previous corticosteroid treatment.