Discordance Rate Among Bilateral Simultaneous and Sequential Temporal Artery Biopsies in Giant Cell Arteritis: Role of Frozen Sectioning Based on the Mayo Clinic Experience

Key Points

Question  What is the role of frozen sectioning in the determination to perform a contralateral temporal artery biopsy (TAB)?

Findings  In this cohort study including 795 patients, frozen section TAB had a sensitivity of 83.2% and specificity of 99.4% for detecting vascular inflammation consistent with active giant cell arteritis. The discordance rate of bilateral TABs was 5% to 5.5%.

Meaning  A negative frozen section result obtained after a unilateral TAB guides the surgeon to perform a contralateral biopsy in the same sitting, whereas a positive frozen section result can reliably indicate giant cell arteritis without the need for a contralateral biopsy.

Importance  Frozen section temporal artery biopsy (TAB) may prevent a contralateral biopsy from being performed.

Objective  To evaluate the sensitivity and specificity of TAB frozen vs permanent section pathology results for giant cell arteritis (GCA) and determine the discordance rate of bilateral TABs.

Design, Setting, and Participants  In this retrospective cohort study, medical records were reviewed from 795 patients 40 years or older who underwent TAB from January 1, 2010, to December 1, 2018, treated at a single tertiary care center with the ability to perform both frozen and permanent histologic sections. Data were analyzed from January 2019 to December 2020.

Main Outcomes and Measures  Sensitivity and specificity of frozen section TAB for detecting GCA, and discordance rates of bilateral permanent section TAB.

Results  Of the 795 included participants, 329 (41.4%) were male, and the mean (SD) age was 72 (10) years. From the 795 patients with 1162 TABs, 119 patients (15.0%) and 138 TABs had positive findings on permanent section. Of these 119 patients, 103 (86.6%) also had positive results on the frozen section, with 4 false-positives (0.6%) and 20 false-negatives (16.8%). Frozen section had a specificity of 99.4% (95% CI, 98.5-99.8), sensitivity of 83.2% (95% CI, 75.2-89.4), positive predictive value of 96.1% (95% CI, 90.4-98.9), negative predictive value of 96.6% (95% CI, 94.9-97.8), positive likelihood ratio of 140.6 (95% CI, 72.7-374.8), and a negative likelihood ratio of 0.17 (95% CI, 0.11-0.25). Simultaneous bilateral TABs were performed in 60 patients (7.5%) with a 5% discordance rate on permanent section. In comparison, bilateral frozen section–guided sequential TABs were performed in 307 patients (38.6%) with 5.5% discordance based on permanent section. In multivariate models, there was a greater odds of positive findings with age (odds ratio [OR], 1.04; 95% CI, 1.01-1.07; P = .008), vision loss (OR, 2.72; 95% CI, 1.25-5.75; P = .01), diplopia (OR, 3.33; 95% CI, 1.00-10.29; P = .04), headache (OR, 2.32; 95% CI, 1.25-4.53; P = .01), weight loss (OR, 2.37; 95% CI, 1.26-4.43; P = .007), and anorexia (OR, 5.65; 95% CI, 2.70-11.89; P < .001).

Conclusions and Relevance  These results support the hypothesis that negative findings from frozen sections should not be solely relied on to refute the diagnosis of GCA, whereas positive findings from frozen sections can be reliably used to defer a contralateral biopsy pending the permanent section results.

Temporal artery biopsy (TAB) is the criterion-standard diagnostic test for giant cell arteritis (GCA), but determining the pretest probability of a positive result relies on clinical and paraclinical manifestations.^1-3 Bilateral simultaneous TABs have been advocated by some to improve the sensitivity of detecting GCA.^4,5 Studies have shown a very wide range of discordance rates (3% to 45%) in patients undergoing bilateral TABs.^3,6-13 In clinical practice, this has resulted in 2 general strategies for performing a TAB in patients with a high clinical index of suspicion for GCA. The first strategy is to perform a second TAB on the contralateral side at some point after the first TAB is found to be negative for GCA, which has the potential to be detrimental in patients with comorbid conditions, such as diabetes. This strategy may result in unnecessary interval glucocorticoid treatment if the second TAB has negative findings. The second strategy is to perform bilateral TABs at the same time (ie, simultaneous). The disadvantage of this strategy is an unnecessary second procedure if the first TAB is found to be positive for GCA on review of the permanent histology slides.

Unique to our institution, TAB specimens are examined initially with a procedural frozen section at the time of the biopsy followed by, within a few days, the final pathology confirmation based on evaluation of formalin-fixed tissue permanent section.¹⁴ Therefore, a frozen section–guided sequential TAB is commonly performed, with the results of the first biopsy obtained within minutes, which determines the need for evaluation of the contralateral side. If the first side is positive for GCA, the contralateral side is not biopsied. However, if the frozen section diagnosis is negative for the initial side, then the contralateral artery is immediately biopsied. In both cases, the permanent section still provides the final biopsy result, while the frozen section helps guide the decision to perform a contralateral biopsy in conjunction with clinical judgment.

Frozen section diagnosis is a technique used to provide intraoperative diagnosis during surgical procedures, although this technique is not uniformly used in the workup of GCA. The procedure was first described in 1895, but the 1905 landmark publication by Louis B. Wilson, MD, at Mayo Clinic¹⁵ is credited with heralding in the “new era in intraoperative diagnosis.”¹⁶ Dr Wilson’s standardized frozen section technique facilitated intraoperative diagnoses for surgeons within 5 minutes. A modified version is still used today at Mayo Clinic.^17-19 Our institution’s history and experience with the frozen section technique has influenced our unique practice of evaluating TABs. It has been suggested that frozen sections might not have a role in the evaluation of a patient with GCA due to high false-negative results.^12,20,21 Because this impression is not based on any large-scale study, we set out to systematically assess the sensitivity and specificity of frozen section TAB pathology results to determine its role in the evaluation of a patient with suspected GCA. In addition, we determined the discordance rate of bilateral simultaneous and frozen section guided sequential TABs.

The Mayo Clinic institutional review board approved this retrospective medical record review of patients 40 years or older who underwent TABs from January 1, 2010, to December 1, 2018, using both frozen section and permanent section techniques. Informed consent was waived given minimal risk to included participants. Participants were not included in the study if they previously withdrew consent to participate in research studies.

A total of 795 patients with 1162 TABs (428 patients with unilateral biopsies and 367 with bilateral biopsies) were analyzed (Figure 1). To capture all patients, a clinical data repository (Advanced Cohort Explorer) was used that contains patient demographic characteristics, diagnoses, and laboratory, clinical, and pathology notes using the search terms temporal artery biopsy and giant cell arteritis. We used the Advanced Cohort Explorer to identify patients meeting the inclusion criteria of age (40 years or older), TAB procedure, and time frame (January 1, 2010, to December 1, 2018). We then abstracted the clinical medical records for the following parameters: demographic characteristics, inflammatory markers, symptoms, steroid treatment preceding biopsy (dose and duration), TAB specimen length, and frozen section and permanent section pathology reports for both the right and left side if bilateral TABs were performed. Rereview of the slide material from frozen and permanent sections was performed in select cases by either J. J. M. or D. R. S.

For frozen sections, the temporal artery segment is bisected in cross-section, and using a freezing microtome, multiple 10-μm cross-sections are cut sequentially for histologic evaluation. A metachromatic dye (toluidine blue) is used to stain the specimen, which is placed on a glass slide and then evaluated by a board-certified surgical pathologist with experience in the frozen section technique. Between 2 to 3 levels of the specimen are examined. Results of the frozen section evaluation are typically available 15 minutes after specimen arrival in the laboratory. Following frozen section, the remaining artery is fixed in 10% buffered formalin and processed according to standard clinical procedures. Multiple cross-sections of the artery are placed on glass slides and stained following this sequence: a 4-μm hematoxylin-eosin–stained slide, a 4-μm Verhoeff–Van Gieson slide, and then 4 other hematoxylin-eosin–stained slides. This technique ensures that arteries are examined at 20 to 125 different levels (depending on the length of artery provided), and all biopsies were performed in this manner.

A positive result (either on frozen or permanent section) is substantiated by active arteritis characterized by medial inflammation with or without giant cells, disruption/duplication of the internal elastic membrane, intimal fibroplasia, and medial attenuation (Figure 2). With the toluidine blue dye, a metachromatic stain, the elastic layer is highlighted by its magenta color and easily identified on the frozen section slides. Only TABs with active arteritis were considered positive for the purpose of our study. Healed arteritis, defined by the presence medial scarring (eg, fibrosis with or without thinning) with or without the supporting findings of intimal fibroplasia and disruption and duplication of the internal elastic membrane, was not considered to be a positive TAB. Some authors consider the presence of histiocytes in the elastic lamina even in the absence of fibrosis or thinning of the media, demonstrated by immunostains, as evidence of healed arteritis.^22-24 McDonnell et al²⁵ found healed arteritis to have the greatest interobserver variability (up to 50%) in their study on the histopathologic diagnosis of GCA. Adventitial or intimal inflammation only were not considered diagnostic of GCA and triggered examination of multiple extra levels to evaluate for inflammation extending into the vascular wall. If a confirmed diagnosis of GCA was not achieved, these cases were not included in this study, although some cases might have been reported with a descriptive diagnosis.^26-28 A negative result reflects the evaluation of multiple cross-sections and levels through the artery specimen without medial inflammation or medial scarring.

TABs were either performed in the minor procedure room in clinic or operating room of a hospital. We considered bilateral TABs to be simultaneous if the procedure was performed at the same time, without obtaining the results of the frozen section before performing the contralateral side. Bilateral frozen section–guided sequential cases were performed based on the results of the frozen section, meaning that if the initial side was negative on frozen section then the contralateral side was biopsied in the same sitting.

Overall sensitivity and specificity were estimated for frozen sections. Within each of the positive permanent and negative permanent cases, the probability of a mismatch between the frozen and permanent result were investigated using logistic regression analysis. Univariate results are provided as well as multivariate models adjusting for erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), platelet count, jaw claudication, and scalp tenderness. In each case, the probability of a mismatch was the outcome of interest. P values were all 2-sided, and no adjustment was made for multiple comparisons in the results. Significance was set at P < .05. All analysis was completed using R version 3.6.2 (The R Foundation).

Of the 795 included participants, 329 (41.4%) were male, and the mean (SD) age was 72 (10) years. From the 795 patients with 1162 TABs, 119 patients (15.0%) and 138 TABs had positive findings on permanent section. Of these 119 patients, 103 (86.6%) also had positive results on the frozen section, with 4 false-positives (0.6%) and 20 false-negatives (16.8%) (Table 1). Frozen section had a specificity of 99.4% (95% CI, 98.5-99.8), sensitivity of 83.2% (95% CI, 75.2-89.4), positive predictive value of 96.1% (95% CI, 90.4-98.9), and a negative predictive value of 96.6% (95% CI, 94.9-97.8). Frozen section TAB had a positive likelihood ratio of 140.6 (95% CI, 72.7-374.8) and a negative likelihood ratio of 0.17 (95% CI, 0.11-0.25), indicating that a patient with a positive frozen section has a greater likelihood of having a positive permanent section, whereas a patient with a negative frozen section has decreased likelihood of having a positive permanent section.

Of the 119 patients with positive permanent section TAB, 77 (64.7%) were women, and the median (interquartile range [IQR]) age was 73 (68-79.5) years (Table 2). The median (IQR) arterial length was 2.95 (2.50-3.45) cm, median (IQR) ESR was 56 (37-100) mm/h, median (IQR) CRP was 5.4 (2.1-10.0) mg/dL (to convert to milligrams per liter, multiply by 10), and median (IQR) platelet count was 341.5 (267.8-412.3) ×10³/μL. A total of 61 patients (51.3%) with positive permanent section TABs had jaw claudication, the only statistically significant symptomatic manifestation of GCA. Glucocorticoids were administered prior to TAB in 78 patients (68.3%), with a median (interquartile range) duration of 4 (2-8) days, which was previously shown not to be a significant confounder given the persistence of the histopathological changes in GCA despite immunosuppressant therapy (Table 2).²⁹

In univariate regression analyses, all clinical symptoms except headache were associated with greater statistical odds of a positive permanent section TAB. The presence of anorexia (odds ratio [OR], 7.11; 95% CI, 4.19-12.07; P < .001) and jaw claudication (OR, 5.28; 95% CI, 3.49-8.02; P < .001) were associated with the greatest odds of a positive section TAB. Initiating glucocorticoids prior to the TAB (OR, 2.01; 95% CI, 1.32-3.10; P = .001), but not chronic glucocorticoids use prior to biopsy, was associated with a positive section TAB. Age, ESR, and CRP were associated with increased odds of a positive section TAB but were less strongly associated with odds of a positive section TAB than diplopia, jaw claudication, weight loss, and anorexia. In a multivariate analysis adjusting for ESR, CRP, platelets, presence of jaw claudication, or scalp tenderness, we found that there was a greater odds of a positive permanent section TAB with age (OR, 1.04; 95% CI, 1.01-1.07; P = .008), vision loss (OR, 2.72; 95% CI, 1.25-5.75; P = .01), diplopia (OR, 3.33; 95% CI, 1.00-10.29; P = .04), headache (OR, 2.32; 95% CI, 1.25-4.53; P = .01), weight loss (OR, 2.37; 95% CI, 1.26-4.43; P = .007), and anorexia (OR, 5.65; 95% CI, 2.70-11.89; P < .001). Table 2 details the results of univariate and multivariate logistic regression analyses and corresponding ORs for each of the variables of interest.

A total of 367 patients (46.2%) underwent bilateral TABs. A total of 60 patients (7.5%) underwent simultaneous TABs with a 5% discordance rate between the 2 sides, and 307 patients (38.6%) underwent frozen section–guided sequential TABs with a 5.5% discordance rate. Among the 27 bilateral TABs that were positive for GCA on permanent section, 4 were simultaneous and 23 were sequential. There were 7 patients whose bilateral TAB specimens were only positive on permanent section on one side, while the other side was normal (Table 3).^3,6-11,13 In total, there were 49 patients with bilateral TABs that had an abnormal permanent section result on either one or both sides; of these, 22 patients had no evidence of active inflammation, including 1 patient with adventitial inflammation reported on only one side on permanent TAB, and 21 patients with healed arteritis, including 12 in which only one side was positive for healed arteritis while the contralateral side was normal (57.1% discordant) and 9 bilaterally positive for healed arteritis.

We systematically evaluated the utility of frozen section for TAB and found it has a sensitivity of 83.2% and specificity of 99.4% for detecting active inflammation consistent with GCA. Based on these results, a negative frozen section result should not be relied on solely to indicate that TAB is negative for GCA. However, if the frozen section is positive (positive predictive value of 96.1%), the surgeon can confidently defer performing a contralateral TAB and wait for the permanent section results, which are the most reliable.^1,30 Based on the calculated sensitivity of 83.2%, 83 of 100 patients with positive frozen sections would be saved from getting a contralateral TAB if bilateral simultaneous section is planned. It is important to point out that our study was not designed to look at the overall sensitivity of diagnosing GCA with frozen section but rather comparing head-to-head frozen section with permanent section.

A prior retrospective and prospective examination of TAB specimens performed at our institution concluded that a contralateral TAB should be performed when the frozen section from the initial side is negative, based on isolated foci of arteritis (skip lesions) found in 17 of 60 patients (28%).²⁰ Although prior studies have systematically evaluated the histopathologic reasoning behind bilateral TABs, only clinical opinion has reviewed the value of frozen sections in the literature.^7,20,21,31 In fact, Chévez-Barrios stated, “I recommend, as do others in the literature, that the diagnosis of temporal arteritis be made on permanent sections.… [A] negative frozen section may necessitate cutting numerous sections until arteritis is identified or the specimen is exhausted, which is a time-consuming and expensive procedure.”^6,20,32-34 Although we do not entirely disagree, this was not based on a systematic review, and our study supports a role of frozen section when deciding if a unilateral or bilateral simultaneous TAB should be performed.

Among our cohort of 60 patients who underwent bilateral simultaneous TABs, there was a 5% discordance rate. Of the 307 patients who underwent bilateral frozen section–guided sequential TABs—with the contralateral side performed immediately after the first side was found to be negative on frozen section—the discordance rate was 5.5%. These results are on the lower side of the range that has been reported in the literature, which have referred to positive cases (Table 3).^3,6-13

Hall and Hunder³ calculated that 14% of patients with GCA undergoing bilateral sequential TABs were diagnosed only because the contralateral side was biopsied and indicated a discordance rate of 29% of 41 total cases. Boyev et al¹¹ reported a 3.3% discordance rate based on 300 bilateral simultaneous and 72 sequential TABs in 186 patients. Breuer et al¹³ reviewed 51 TAB-positive GCA cases (of 132 total bilateral simultaneous TABs), of which only 13 were unilaterally positive (discordance rate of 25.4%). They inferred that up to 12.7% of the patients (one-half of the discordance rate) could have been misdiagnosed as TAB-negative cases had a unilateral TAB been performed (Table 3).

Niederkohr and Levin³⁵ used bayesian analysis to explain that discordance rates from bilateral TABs should not be used to estimate their increased sensitivity (over unilateral TABs), based on the assumption that there is a 50% chance that the positive side of a discordant pair would be randomly chosen in a unilateral TAB. However, determining which side to sample in a unilateral TAB is usually based on clinical presentation (ie, side of visual loss, temporal artery tenderness, decrease pulse, or nodularity).³⁶ Pretest probability of the disease therefore varies with the side chosen and influences the discordance rate.

At our institution, if there is a high pretest probability of GCA, bilateral simultaneous TABs are planned, but frozen section for the first side is submitted and the results are obtained immediately prior to deciding whether to proceed to the contralateral side. If the frozen section is positive on the first TAB, a contralateral TAB is deferred, given the very low false-positive rate (0.6%). However, if the frozen section does not align with the permanent section result, in particular if the frozen section is positive but permanent section is negative, the patient returns for a TAB on the contralateral side if the GCA suspicion remains high. Based on our study, this would occur in less than 1% of patients. This frozen section–guided sequential TAB strategy has the advantage of saving some patients from an unnecessary second TAB (and its associated risks) as well as saving time by avoiding a second procedure visit. At our institution, this method can additionally reduce the overall cost of care, but we recognize that this is not generalizable to other institutions.^36,37

We found that older age, higher baseline inflammatory markers, and presence of vision loss, diplopia, headache, weight loss, and anorexia were associated with higher odds of GCA. These results are consistent with the findings of previous reports.³⁸ The association of increased odds of GCA positivity in patients treated with glucocorticoids prior to the TAB is likely due to the higher index of clinical suspicion, necessitating earlier therapeutic intervention.

This study had limitations. This is a single-center retrospective analysis. We did not rereview all of the TABs but rather relied on the original pathology reports for most of these cases. We recognize the uniqueness of our institution to use frozen sections, where the additional 15 minutes for the frozen section result is ultimately considered to be a time-saving and cost-saving procedure compared with bringing the patient back for a subsequent procedure visit to conduct a contralateral biopsy. We understand that this may not be feasible at other institutions that are not set up for both rapid and accurate frozen sectioning, where this procedure could instead be cost prohibitive. The pathology laboratory at our institution is well equipped for frozen section as it is available at all times, allowing for very rapid TAB processing turnaround. This is in part because fewer sections are examined with frozen vs permanent sections (2 to 3 levels vs 20 to 125 levels, respectively), suggesting inherently decreased sensitivity with frozen sections. Our frozen section procedure and experience (evaluate multiple and use toluidine blue staining that highlights the elastic layer) may not be similar to other institutions and may lead to variable results to our results. Additionally, we acknowledge that there is controversy regarding the clinical interpretation of healed arteritis. At our institution, healed arteritis is interpreted in the context of patient clinical characteristics and radiographic findings, which may differ from other institutions and may impact the results of this study.

Our study has shown that frozen section TAB has a very good sensitivity and excellent specificity for GCA and therefore supports the role of frozen section in performing TAB. However, more studies with a comparative analysis of laboratory results, clinical symptoms, and patient demographic characteristics between positive and negative frozen and permanent TAB results are needed to confirm our findings.

Accepted for Publication: December 3, 2020.

Published Online: February 18, 2021. doi:10.1001/jamaophthalmol.2020.6896

Corresponding Author: M. Tariq Bhatti, MD, Department of Ophthalmology, Mayo Clinic College of Medicine & Science, 200 First St SW, Rochester, MN 55905 (bhatti.muhammad@mayo.edu).

Author Contributions: Drs Cohen and Bhatti had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Cohen, Chen, Neth, Fillmore, Bhatti.

Acquisition, analysis, or interpretation of data: Cohen, Chen, Neth, Sabbagh, Hodge, Warrington, Maleszewski, Salomao, Bhatti.

Drafting of the manuscript: Cohen, Neth, Sabbagh, Salomao.

Critical revision of the manuscript for important intellectual content: Cohen, Chen, Hodge, Warrington, Fillmore, Maleszewski, Salomao, Bhatti.

Statistical analysis: Cohen, Hodge.

Administrative, technical, or material support: Neth, Sabbagh, Fillmore, Bhatti.

Study supervision: Warrington, Fillmore, Maleszewski, Bhatti.

Conflict of Interest Disclosures: Dr Warrington has received grants from Eli Lilly and Company and Kiniksa Pharmaceuticals as well as personal fees from Genentech-Roche and Sanofi. No other disclosures were reported.

Meeting Presentation: These data were presented as a poster at the 46th annual North American Neuro-Ophthalmology Society Meeting; March 7-12, 2020; Amelia Island, Florida.