Giant cell arteritis (GCA) is the most common systemic vasculitis in the American population older than 50 years. Symptoms are nonspecific and include temporal headache, malaise, jaw claudication, and transient or permanent vision loss and diplopia. High-dose steroid treatment can clear the affected arteries within hours, allowing return of blood flow to vital structures (1). GCA is definitively diagnosed with a biopsy of the superficial temporal artery. Temporal artery biopsy (TAB) as a procedure is time consuming and not a treatment modality. Therefore, much attention has been paid to maximizing the yield on diagnostic studies by identifying the most specific and sensitive clinical criteria (2–4).
Although there are many studies focusing on characteristics of GCA in the general population, there is a definite lack of studies focusing on the veteran population, which tends to have higher rates of physical and mental comorbidities (5,6). The goal of this study was to determine diagnostic considerations of GCA, with a specific focus on identifying differences in veteran vs. community demographics. We anticipated a lower than average positive biopsy rate in this population because of the higher rates of comorbidities that could mimic symptoms and laboratory values of GCA. We also hypothesized that the veteran population presented signs and symptoms parallel to that of the general population. This is the first study describing the diagnostic features of GCA in the US military veterans.
A retrospective chart review was performed in the Veterans Information Systems and Technology Architecture Computerized Patient Record System at a single Veterans Administration institution (VA Portland Health Care System). All anatomic pathology reports were collected using the terms “superficial temporal artery” and/or the diagnosis of “giant cell arteritis” or “temporal arteritis.”
A positive biopsy was considered to be the final confirmation of temporal arteritis diagnosis. Healed arteritis and remote arteritis were considered positive for GCA, whereas inconclusive biopsies (such as isolated small breaks in the internal elastic lamina) were considered negative. A subject was considered positive for polymyalgia rheumatica (PMR) if he or she had a formal diagnosis or presenting signs and symptoms of PMR (acute onset bilateral shoulder and hip pain, muscle or joint stiffness, or proximal muscle weakness).
Clinical characteristics, including age and gender, as well as laboratory values such as platelet level, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) levels were recorded for each patient when available. The normal value for ESR depends on gender and age. We used the calculation ESR = age/2 for men and ESR = (age + 10)/2 for women to determine the age-adjusted upper limit of normal (ULN) ESR value for each patient (7,8). We then divided each ESR value by the calculated age-adjusted ULN value to standardize the data. The ULN CRP value for systemic inflammation at the Portland VA laboratory is 11 mg/L. The length of the temporal artery specimen was also included because past studies have found that insufficient specimen length may be associated with a higher false negative biopsy rate (9).
Statistical analysis was performed using SPSS version 25.0 (IBM Corp, IBM SPSS Statistics for Macintosh, NY). Frequency, sensitivity, and specificity were calculated for the following variables: new-onset headache, jaw claudication, scalp tenderness, PMR, and vision changes. Vision changes included diplopia, ischemic vision loss/optic disc disease, and amaurosis fugax. Sensitivity and specificity were also calculated for elevated ESR. Because of the high rate of missing CRP values for patients that underwent a TAB before the year 2005, a subset of 241 patients from 2005 and after were selected to calculate the sensitivity and specificity of elevated CRP. In addition, descriptive statistics including mean and SD were calculated for ESR, CRP, platelets, and biopsy length.
Next, a logistic regression (LR) diagnostic prediction model was developed from the VA population data set. The following 10 independent variables were investigated for inclusion in the model: age, gender, new-onset headache, scalp tenderness, vision changes, jaw claudication, biopsy length, platelet levels, CRP, and ESR. The outcome variable was a positive TAB. The individual subject was the unit of analysis for the LR model, and Stata 15.1 was used for statistical analysis (StataCorp LLC, College Station, TX). A P value of <0.05 was used for statistical significance.
Each continuous independent variable was graphed and assessed for normal distribution. The variables CRP and platelets were both observed to have a right-skewed distribution, and a logarithmic transformation was used on each to obtain normal distribution. Missing values within the data set were identified and investigated. As stated above, before the year 2005, CRP laboratory results were not regularly performed on patients in the study population. Therefore, 18.8% of study subjects were missing results for this test, over half of which were biopsied before 2005. Because of the high amount of missing values within this time frame, a subset of the study population was selected to develop the diagnostic prediction model. This subset included 241 patient records and only included subjects who were biopsied during 2005 and after. Using this population subset, the percentage of missing CRP values was reduced to 9.5%, with 23 total missing values. In addition to the missing values within CRP, 2 patients were missing ESR values, 5 were missing platelet levels, and 4 were missing biopsy length. The Little Missing Completely at Random test was performed, and it was determined that the values were informatively missing (nonrandom). Covariate-dependent missingness was assessed and used to adjust for the missing values.
Model diagnostics were performed in Stata to determine the performance of the model. Receiver operating characteristic (ROC) curve analysis and classification statistics were used to determine sensitivity and specificity, and calibration was assessed using the Hosmer–Lemeshow test. Overall performance was reported using the overall misclassification rate and pseudo R (2).
The initial search resulted in 352 TABs. Fifty-one subjects from the years 1989–1999 were excluded because of limited information. An additional 2 subjects were also excluded because of unavailable chart notes. Two subjects were younger than 50 years, one subject refused to have a biopsy performed, and four biopsies did not have the temporal artery in the specimen; these were also excluded. Eight subjects had 2 separate TABs performed. We included only the first presentation of each subject. Of note, all of the patients who had multiple TABs performed and who also had an initial negative biopsy did not have subsequent biopsies that were positive. These screening criteria resulted in 292 subjects for final analysis.
Clinical characteristics and average laboratory values (ESR, CRP, platelets, and length of TAB) of patients are included in Table 1. Of the 292 subjects, 40 had positive biopsies (13.7%). Only 20 specimens were from women (6.8%), but 25% of them were positive for GCA vs. only 12.9% of male specimens. The average age of the subjects with a positive biopsy was 73 ± SD of 8.8 years, whereas the average age of patients with a negative biopsy was 71.2 ± 9 years (no significant difference, P = 0.2532). The average ESR, CRP, and platelet levels were 53.1 ± 34.6 mm/hr, 35.5 ± 62.4 mg/L (observed values ranged from 0.3 to 349.7 mg/L), and 268.5 ± 96 × 109/L, respectively. There was a significant difference between ESR, CRP, and platelet levels in patients with positive vs. negative biopsies. The average ESR in patients with a positive biopsy was 69.1 mm/hr and in patients with a negative biopsy was 50.5 mm/hr (P = 0.0016). The average CRP for patients with a positive biopsy was 56.6 mg/L and for patients with a negative biopsy was 32.2 mg/L (P = 0.0394). The mean platelet levels were 317.6 × 109/L for patients with a positive biopsy and 260.6 × 109/L for patients with a negative biopsy (P = 0.0005). The average biopsy length for patients with a positive biopsy was 2 ± 0.7 cm and for a negative biopsy was 1.9 ± 0.7 cm (P = 0.4145). Of note, 20% of patients with a positive TAB had normal ESR values for their age, at the time of diagnosis, vs. 40.5% of patients with a negative TAB. In the subset of patients from 2005 and onward, when CRP was commonly performed (see methods section for details), 16.7% of patients with a positive TAB had normal CRP values vs. 43.6% of patients with a negative TAB.
TABLE 1. -
Characteristics of Veterans Administration cases referred for a temporal artery biopsy
||Patients Referred for a TAB
||Patients With a Positive TAB
||Patients With a Negative TAB
|Age, years, mean (SD)
|Biopsy length, cm, mean (SD)
| Platelets, × 109/L, mean (SD)
| ESR, mm/hr, mean (SD)
| CRP, mg/L, mean (SD)
The normal values for ESR depend on age and gender.
CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; n, number of subjects; TAB, temporal artery biopsy.
The frequencies of the presenting symptoms in patients with positive and negative TABs are summarized in Figure 1. Headache was the most common presenting symptom overall (71.6%), followed by vision changes (50.3%), scalp tenderness (25.7%), jaw claudication (20.9%), and PMR-related symptoms (12.7%). Amaurosis fugax was the most common visual change (10.6%), closely followed by ischemic vision loss (10.3%). The difference between the number of patients in the positive vs. negative TAB groups was significant for the variables scalp tenderness (P = 0.0409), PMR (P = 0.0003), and amaurosis fugax (P = 0.0324).
The sensitivity and specificity of these symptoms are shown in Table 2 along with the sensitivity and specificity of elevated ESR and elevated CRP. We found that elevated CRP had the highest sensitivity at 83%, followed by elevated ESR with a sensitivity of 80% and headache with a sensitivity of 62.5%. Jaw claudication and PMR were most specific (81.3% and 89.3%, respectively). Scalp tenderness also had a high specificity at 75.8%.
TABLE 2. -
Sensitivity and specificity of the presenting symptoms and elevated ESR and CRP values for a positive temporal artery biopsy
*Vision changes included diplopia, ischemic vision loss/optic disc disease, and amaurosis fugax.
†Sensitivity and specificity of elevated CRP values were calculated for patients who underwent temporal artery biopsy during and after 2005 (n = 241, vs. 292 patients total).
CRP, C-reactive protein; ESR, erythrocyte sedimentation rate.
LR modeling showed that the independent variables of log-transformed platelets (odds ratio [OR] = 4.309, P = 0.049), log-transformed CRP (OR = 1.504, P = 0.022), and scalp tenderness (OR = 3.860, P = 0.016) were statistically significant predictors of a positive TAB (Table 3). Although age was not a statistically significant predictor in the final model (OR = 1.043, P = 0.096) (Table 4), it was retained because of its high clinical significance in the diagnosis of GCA. Similarly, vision change was not found to be statistically significant (OR = 0.938, P = 0.889) but was retained in the model because, combined with the variables age and scalp tenderness, it was informative with regard to the missing values in the data set.
TABLE 3. -
Multivariable logistic regression for a positive temporal artery biopsy, using complete-case analysis
||95% CI, OR
CI, confidence interval; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; PMR, polymyalgia rheumatica; OR, odds ratio.
TABLE 4. -
Multivariable logistic regression for a positive temporal artery biopsy, using variables from the final model
||95% CI, OR
CI, confidence interval; CRP, C-reactive protein; OR, odds ratio.
The final LR model (Table 4) contained the independent variables of scalp tenderness, vision changes, age, log (platelets), and log (CRP). This model had a chi-square P value of 0.0008, indicating high statistical significance. A one-unit increase in log (CRP) resulted in a 1.50 increase in the log-odds of a positive biopsy (95% confidence interval [CI] = 1.06–2.13), whereas a one-unit increase in log (platelets) resulted in a 4.31 increase in the log-odds of a positive biopsy (95% CI = 1.05–2.13). Patients who reported scalp tenderness had 3.09 times greater odds of having a positive biopsy than those who did not report this symptom (95% CI = 1.24–7.72). Although age (OR = 1.04, 95% CI = 0.99–1.10) and vision changes (OR = 0.94, 95% CI = 0.38–2.31) both showed negligible effects on the odds of a positive biopsy, they were retained because of their aforementioned importance to the overall model.
Diagnostics performed in Stata indicated fairly good predictive performance of this model, with an area under the ROC curve of 0.7440 (Fig. 2). A Hosmer–Lemeshow goodness of fit test resulted in a P value of 0.9220, which showed no evidence of poor fit. This model showed poor sensitivity (50%) and relatively high specificity (88.36%). Although the sensitivity can be improved by adjusting the cutoff score, these adjustments greatly reduce the model's specificity. The model had an overall misclassification rate of 16.28% and a pseudo R2 of 0.1319.
In summary, our data show a lower positive TAB rate than in previous nonveteran studies. Nevertheless, they are performed in our veteran patients for similar symptoms, and they carry similar demographics, as nonveteran populations. For instance, the female veteran population was proportionally smaller, but women were twice as likely to have positive TABs as men. This trend mirrors that of the general population; in Northern Europe, a ratio of 3:1 women to men has been observed (10).
We postulate that the low positive biopsy yield of 13.7% in veterans is related to the nonspecific nature of the presenting symptoms of temporal arteritis. Furthermore, with veterans having higher rates of medical comorbidities, elevations in ESR and CRP are not unexpected and would surely make diagnosis of GCA an even greater challenge. Our study shows that approximately 62% of patients who had a negative TAB had elevated ESR and/or CRP values. Even with negative laboratory results, patients with multiple comorbidities and multiple suggestive symptoms necessitated a biopsy. Recent studies have sought to rectify this issue by creating diagnostic algorithms (11).
Another possible explanation for the low biopsy yield is that the specimen was insufficient in length or did not include the portion of the blood vessel with the classic histological signs of GCA. GCA affects vessels segmentally, which confers a risk of missing the pathologic part of a biopsy specimen (so-called “skip lesions”). Some studies have recommended lengths that range from 0.5 to 1.5 cm to remedy this (9,12). In our study, the range of biopsy lengths was 0.5–5.5 cm (mean of 2 cm), which is in line with the authors' preference for samples of 2 cm or more.
The most common presenting symptom in patients that were sent for a TAB was new-onset headache. Other studies have found similar results with 66%–86% of patients presenting with headache (13). However, headache is not a symptom specific to GCA. In 2016, headache was the fifth most common presenting complaint in the emergency department (14). An elevated ESR was also found to be very sensitive for GCA but, like headache, it is not specific to GCA. An elevated ESR can be found in a myriad of other inflammatory, infectious, and neoplastic diseases. Furthermore, a normal ESR does not rule out GCA. A retrospective study from 2012 found that 4% of patients with a positive TAB had normal ESR values at the time of diagnosis (15). We found that 8 of the 40 patients with biopsy-proven GCA had normal ESR values, although this sample size is considerably smaller than other studies, limiting its generalizability.
Although the presenting symptoms of GCA may be ambiguous and mimic several other diseases, our findings do suggest that jaw claudication and PMR were highly specific for biopsy-proven GCA. Several past publications have also demonstrated strong associations between jaw claudication and TAB (3,16,17). An association between PMR and GCA is also well established in the current literature (18–22). PMR and GCA are closely related clinically. They cause significant elevations in acute phase reactants, respond well to corticosteroid therapy, and affect an older and predominantly female population. Thus, these 2 entities may indeed be parts of a pathologic spectrum.
Based on our LR model, CRP, platelets, and scalp tenderness were found to be significantly associated with a positive TAB in VA patients. Although CRP was not historically used as a method for diagnosing GCA, it has been a well-established independent risk factor for GCA diagnosis and has a high sensitivity for systemic inflammation (11,15). The significance of platelets and their association with positive TABs have also been demonstrated in the past literature (11,23). Furthermore, a 2019 study performed by Chan et al found that a combination of CRP and platelet tests may be the most useful laboratory values in screening for GCA, and our study supports these data (24).
The sensitivity and specificity of CRP for a positive TAB varies widely in the current literature, and the ULN also varies based on the institution/clinical laboratory where these tests are performed (2,15). To this point, the VA uses high-sensitivity CRP, which is measured in milligrams per liter and has a different systemic inflammatory threshold (over 11 mg/L) than cardiac threshold of disease (over 3 mg/L). These differences highlight the complexity of diagnosing an inflammatory condition within the veteran population.
Half of the patients in our study had ocular symptoms but half did not. It is possible that subtle visual signs or symptoms may have been missed in patients who were not referred for ophthalmological evaluation. Although the authors performed nearly all biopsies in the past 7 years, previous biopsies were subject to individual variability in this single institution study. In our hospital, it is currently the neuro-ophthalmologist who receives these biopsy consults, primarily from rheumatology, primary care, and the emergency department. In this setting, the job may fall appropriately to the ophthalmologist or neuro-ophthalmologist to elicit subtle visual symptoms and signs and to rule them in or out based on expert experience with this population.
There are some additional limitations with this study. Statistical limitations were addressed in the Methods section, and the resulting analyses accounted for these limitations. Nevertheless, certain variables were more difficult to identify in patients' charts; for instance, a clinician may have described jaw fatigue and pain without explicitly using the term “jaw claudication.” In this instance, the authors reviewed the chart notes further to make a clinical judgment regarding whether the variable should be rendered as positive or not. And as with any retrospective study design, one must rely on the recordkeeping of other individuals.
In conclusion, we assert that GCA diagnosis may be more difficult in veterans who have known comorbid disease, i.e., more common than in the general population. This observation raises the following questions: Should clinicians have a different threshold for concern for possible GCA in VA patients with abnormal laboratory results and nonspecific symptomatology? Should we be looking more closely at platelet counts in this specific population? The results of this retrospective analysis from a single institution warrant more comprehensive study to assess generalizability and to determine the best method to evaluate and treat veteran patients with regard to GCA.
STATEMENT OF AUTHORSHIP
Category 1: a. Conception and design: K. Winges and L. Selby; b. Acquisition of data: K. Winges and L. Selby; and c. Analysis and interpretation of data: K. Winges, L. Selby, and B. Park-Egan. Category 2: a. Drafting the manuscript: K. Winges, L. Selby, and Brenna Park-Egan and b. Revising it for intellectual content: K. Winges. Category 3: Final approval of the completed manuscript: K. Winges, L. Selby, and B. Park-Egan.
The authors thank Rebecca Rdesinski, MPH, MSW, Research Associate at Oregon Health & Science University and Jack Wiedrick, MS, Biostatistician in the Biostatistics and Design Program at Oregon Health & Science University for their assistance with statistical analysis. The authors also thank Tahnee Groat at the Portland VA for IRB and institutional help with this study.
1. Hayreh SS, Zimmerman B, Kardon RH. Visual improvement with corticosteroid therapy in giant cell arteritis. Report of a large study and review of literature. Acta Ophthalmol Scand. 2002;80:355–367.
2. Parikh M, Miller NR, Lee AG, Savino PJ, Vacarezza MN, Cornblath W, Eggenberger E, Antonio-Santos A, Golnik K, Kardon R, Wall M. Prevalence of a normal C-reactive protein with an elevated erythrocyte sedimentation rate in biopsy-proven giant cell arteritis. Ophthalmology. 2006;113:1842–1845.
3. Peral-Cagigal B, Perez-Villar A, Redondo-Gonzalez LM, Garcia-Sierra C, Morante-Silva M, Madrigal-Rubiales B, Verrier-Hernandez A. Temporal headache and jaw claudication may be the key for the diagnosis of giant cell arteritis. Med Oral Patol Oral Cir Bucal. 2018;23:e290–e4.
4. Rodriguez-Valverde V, Sarabia JM, Gonzalez-Gay MA, Figueroa M, Armona J, Blanco R, Fernandez-Sueiro JL, Martinez-Taboada VM. Risk factors and predictive models of giant cell arteritis in polymyalgia rheumatica. Am J Med. 1997;102:331–336.
5. Agha Z, Lofgren RP, VanRuiswyk JV, Layde PM. Are patients at Veterans Affairs medical centers sicker? A comparative analysis of health status and medical resource use. Arch Interm Med. 2000;160:3252–3257.
6. Yu W, Ravelo A, Wagner TH, Phibbs CS, Bhandari A, Chen S, Barnett PG. Prevalence and costs of chronic conditions in the VA health care system. Med Care Res Rev. 2003;60:146S–167S.
7. Gonzalez-Gay MA, Vazquez-Rodriguez TR, Lopez-Diaz MJ, Miranda-Filloy JA, Gonzalez-Juanatey C, Martin J, Llorca J. Epidemiology of giant cell arteritis and polymyalgia rheumatica. Arthritis Rheum. 2009;61:1454–1461.
8. Miller A, Green M, Robinson D. Simple rule for calculating normal erythrocyte sedimentation rate. Br Med J. 1983;286:266.
9. Ranganath VK, Elashoff DA, Khanna D. Age adjustment corrects for apparent differences in erythrocyte sedimentation rate and C-reactive protein values at the onset of seropositive rheumatoid arthritis in younger and older patients. J Rheumatol. 2005;32:1040–1042.
10. Taylor-Gjevre R, Vo M, Shukla D, Resch L. Temporal artery biopsy for giant cell arteritis. J Rheumatol. 2005;32:1279–1282.
11. Ing EB, Lahaie Luna G, Toren A, Ing R, Chen JJ, Arora N, Torun N, Jakpor OA, Fraser JA, Tyndel FJ, Sundaram AN, Liu X, Lam CT, Patel V, Weis E, Jordan D, Gilberg S, Pagnoux C, Ten Hove M. Multivariable prediction model for suspected giant cell arteritis: development and validation. Clin Ophthalmol. 2017;11:2031–2042.
12. Mahr A, Saba M, Kambouchner M, Polivka M, Baudrimont M, Brocheriou I, Coste J, Guillevin L. Temporal artery biopsy for diagnosing giant cell arteritis: the longer, the better? Ann Rheum Dis. 2006;65:826–828.
13. Gonzalez-Gay MA, Barros S, Lopez-Diaz MJ, Garcia-Porrua C, Sanchez-Andrade A, Llorca J. Giant cell arteritis: disease patterns of clinical presentation in a series of 240 patients. Medicine (Blatimore). 2005;84:269–276.
14. Rui P, Kang K, Ashman JJ. National Hospital Ambulatory Medical Care Survey: 2016 Emergency Department Summary Tables. 2016. Available at: https://www.cdc.gov/nchs/data/ahcd/nhamcs_emergency/2016_ed_web_tables.pdf
. Accessed December 10, 2019.
15. Kermani TA, Schmidt J, Crowson CS, Ytterberg SR, Hunder GG, Matteson EL, Warrington KJ. Utility of erythrocyte sedimentation rate and C-reactive protein for the diagnosis of giant cell arteritis. Semin Arthritis Rheum. 2012;41:866–871.
16. Gabriel SE, O'Fallon WM, Achkar AA, Lie JT, Hunder GG. The use of clinical characteristics to predict the results of temporal artery biopsy among patients with suspected giant cell arteritis. J Rheumatol. 1995;22:93–96.
17. Chacko JG, Chacko JA, Salter MW. Review of giant cell arteritis. Saudi J Ophthalmol. 2015;29:48–52.
18. Weyand CM, Goronzy JJ. Giant-cell arteritis and polymyalgia rheumatica. N Engl J Med. 2014;371:50–57.
19. Franzen P, Sutinen S, von Knorring J. Giant cell arteritis and polymyalgia rheumatica in a region of Finland: an epidemiologic, clinical and pathologic study, 1984–1988. J Rheumatol. 1992;19:273–276.
20. Salvarani C, Gabriel SE, O'Fallon WM, Hunder GG. Epidemiology of polymyalgia rheumatica in Olmsted County, Minnesota, 1970–1991. Arthritis Rheum. 1995;38:369–373.
21. Kermani TA, Warrington KJ. Polymyalgia rheumatica. Lancet. 2013;381:63–72.
22. Salvarani C, Cantini F, Hunder GG. Polymyalgia rheumatica and giant-cell arteritis. Lancet. 2008;372:234–245.
23. Costello F, Zimmerman MB, Podhajsky PA, Hayreh SS. Role of thrombocytosis in diagnosis of giant cell arteritis and differentiation of arteritic from non-arteritic anterior ischemic optic neuropathy. Eur J Ophthalmol. 2004;14:245–257.
24. Chan FLY, Lester S, Whittle SL, Hill CL. The utility of ESR, CRP and platelets in the diagnosis of GCA. BMC Rheumatol. 2019;3:14.