Giant cell arteritis (GCA) is a vasculitis that primarily affects the elderly and may lead to permanent blindness if not detected and treated early. Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are 2 serum acute-phase reactants that can aid clinicians in diagnosing and monitoring patients with GCA. Although the results have been variable and studies have not analyzed patients with biopsy-proven GCA, previous reports suggest that the ESR and CRP level may be affected by nonsteroidal anti-inflammatory drugs (NSAIDs) (1-7) or statin use (8-10).
Tishler et al found that NSAID use had a weak but significant correlation with CRP levels in patients with rheumatoid arthritis (RA) (5,13) and other authors have reported that ESR and CRP levels are decreased by NSAID use in RA (7). In contrast, one study tested the effect of 12 weeks of NSAID use on ESR and CRP in patients with RA and found no difference (1). In another study of patients with RA, NSAIDs were found not to affect ESR, CRP, or haptoglobin (another acute-phase reactant) but corticosteroids did affect these markers (11).
Statin use has also been associated with decreased inflammatory marker levels, with CRP being the most studied marker, but if this is secondary to anti-inflammatory effects, then ESR should theoretically be lower in statin users as well. One study comparing statins and aspirin found a mean CRP of 2.78 mg/L in controls, 2.73 mg/L in patients on aspirin alone, 2.29 mg/L in patients on statins alone, and 2.03 mg/L in patients on both aspirin and statins (8). These results suggest that statins are associated with lower CRP values, and this may be synergistic with aspirin in patients without inflammatory diseases. Kinlay reported on the CRP-lowering effects of simvastatin vs simvastatin/ezetimibe. The author found that simvastatin lowered baseline CRP by 14.3% and that simvastatin/ezetimibe lowered CRP by 31.0% (9). A meta-analysis of studies analyzing the association between cholesterol-lowering therapies and CRP values found that across all studies (58% of which were statin-only treatment studies), there was a 28% (95% CI: 26%-30%) drop in CRP, with greater drops occurring in statin treatments than with other cholesterol-lowering therapies (10). However, 89%-98% of the decrease in CRP associated with using cholesterol-lowering agents was found to be related to the low-density lipoprotein (LDL)-lowering effects of the treatments (10). These findings suggest that while statin use is associated with lower CRP values, this association may be directly related to lower LDL levels produced by the statins.
For ophthalmologists considering the diagnosis of GCA in patients taking NSAIDs or statins, artificially low or normal inflammatory marker levels might lead to error in clinical judgment (increased false-negative rate). To the best of our knowledge, there has been no prior study that searches for an association between NSAID or statin use and ESR or CRP in patients with biopsy-proven GCA. Because these 2 serum markers are often pivotal in the evaluation of patients with suspected GCA, it is critical to know if these medications are associated with lower ESR and CRP values in this patient population. In this study, we report on the associations between use of statins or NSAIDs and serum ESR and CRP in patients with biopsy-proven GCA.
A retrospective chart review of all patients with the primary diagnosis of GCA seen in the H. Stanley Thompson Neuro-Ophthalmology Clinic at the University of Iowa Hospitals and Clinics from 1960 to 2008 was conducted. Institutional review board approval was obtained prior to data collection, and the collection was compliant with Insurance Portability and Accountability Act. This included patient age, date of diagnosis, initial serum ESR and/or CRP, temporal artery biopsy result, and medication profile, particularly the use of NSAIDs or statins. Patients were included only if they had adequate documentation of their medications, a serum ESR and/or CRP measurement prior to the initiation of corticosteroid treatment, and biopsy-proven GCA. Patients were excluded for the following reasons: 1) they had inadequate documentation of their medications; 2) they had inadequate documentation of serum ESR and/or CRP measurements; 3) they had initiation of steroid treatment before initial ESR and/or CRP measurements; 4) they had a history of other collagen vascular or inflammatory disease that could also produce an elevated ESR/CRP; or 5) they did not have biopsy-proven GCA.
All patients were categorized into 1 of 2 groups: those exposed to the study medications (NSAIDs and/or statins) and those not exposed to the study medications at the time of their first evaluation. The initial ESR and CRP in patients from both groups were compared using the Wilcoxon rank sum test. The proportion of patients with a normal initial ESR measurement (defined as ESR of 37 mm per hour or less) and those with a normal initial CRP measurement (defined as CRP of 0.5 mg/dL or less) were computed and then compared between patient groups using Fisher exact test. We defined a normal ESR at a cutoff of 37 mm per hour because the average patient age was 75 years and if divided by 2 this is 37.5, which is a method sometimes used as a reference for ESR since it varies by age (12). However, we do recognize that this is a somewhat arbitrary cutoff. We defined a normal CRP as <0.5 mg/dL because this was the laboratory's reference value. In addition to assessing the effect of either statin or NSAID use on a defined normal ESR, logistic regression analysis was performed with either NSAID use or statin use as the independent variable, and age as a covariate to account for the known effect of age on ESR. This same analysis was also performed for patients with a defined normal CRP. Power calculations were performed based on the number of subjects with recorded ESR and CRP values. Using Fisher exact test (P ≤ 0.05; power = 0.8), we estimated that a statistically significant difference between study group patients could be detected if 30% of statin users (n = 24) and ≤6% of statin nonusers (n = 137) had a normal ESR. Similarly, we estimated that a statistically significant difference between study group patients could be detected if 25% of NSAID users (n = 80) and ≤8% of NSAID nonusers (n = 81) had a normal ESR. For comparisons involving CRP levels, Fisher exact test (P ≤ 0.05; power = 0.8) was estimated to detect significant differences between study groups if 25% of the statin user group (n = 17) and ≤1.3% of the statin nonuser group (n = 90) had normal CRP levels. For NSAID users (n = 53) vs nonusers (n = 54), we estimated that a group difference could be detected with 0.80 power if CRP was normal in 20% of NSAID users and ≤2% or less of NSAID nonusers.
Of the 742 charts reviewed, 161 patients (22%) were included and 581 patients (78%) were excluded. Of the included patients, 161 had measurements for ESR and 107 of them had CRP measurements; the difference occurred because mostly all clinicians ordered ESR and not all ordered CRP. It was rare for clinicians to have only ordered CRP. The patients in the study were composed of 118 women (73%) and 43 men (27%). The mean patient age was 75.7 years (SD = 8.5 years, range = 53-95 years). Of the 161 patients included in our study, 81 patients used NSAIDs. Of these, 62 patients used NSAIDs but not statins, 19 used both NSAIDs and statins. Of the 24 patients using statins, 19 patients used both statins and NSAIDs and 5 patients used only statins. Of the total 161 patients, 75 patients used neither class of medication. NSAIDs used by patients in the study included aspirin, ibuprofen, naproxen, piroxicam, celecoxib, valdecoxib, salsalate, diclofenac, ketorolac, rofecoxib, ketoprofen, and nabumetone, with aspirin outnumbering all the other NSAIDs combined. The statins used by patients in this study included simvastatin, atorvastatin, pravastatin, lovastatin, fluvastatin, and rosuvastatin, with simvastatin being the most common statin used.
The median ESR of statin users was 57.5 mm per hour (interquartile range [IQR] = 35-85 mm per hour; IQR = 25th to 75th percentile; range = 17-136 mm per hour), which was significantly lower than the median ESR of statin nonusers (median = 85.0 mm per hour; IQR = 60-110 mm per hour; range = 6-150 mm per hour; P = 0.005). The median ESR of NSAID users was 75 mm per hour (IQR = 46-98.5 mm per hour; range = 6-143 mm per hour), which was significantly lower than the median ESR of NSAID nonusers (median = 98 mm per hour; IQR = 64-116 mm per hour; range = 11-150 mm per hour; P = 0.004).
Comparison of median CRP values showed no significant difference between statin users and nonusers (P = 0.127), with a median CRP of 2.4 mg/dL (IQR of 0.8-15.9 mg/dL; range = 0.5-18.8 mg/dL) for statin users and 8.7 mg/dL (IQR of 2.7-16.2 mg/dL; range = 0.4-61.5 mg/dL) for statin nonusers. There was also no significant difference in CRP (P = 0.524) between those who used NSAIDs (median of 8.0 mg/dL; IQR of 1.5-16.2 mg/dL; range = 6-143 mg/dL) and those who did not (median of 8.7 mg/dL; IQR of 2.1-16.2 mg/dL; range = 0.4-61.5 mg/dL).
In our population of patients with biopsy-proven GCA, ESR was normal in 29.2% (7 of 24) of statin users compared with 13.1% (18 of 137) of statin nonusers; however, this was not statistically significant (P = 0.064). After using logistic regression analysis to correct for the effect of age on ESR, this comparison was still not significant (P = 0.055). The odds for a normal ESR were 2.94 (95% CI: 0.89-9.15) times higher in statin users compared with statin nonusers. The ESR was normal in 17.5% of NSAID users and was normal in 13.6% of NSAID nonusers, and this difference was not significant (P = 0.521). The odds for a normal ESR were 1.33 (95% CI: 0.52-3.51) times higher in NSAID users compared with NSAID nonusers. The CRP was normal in 17.6% of statin users and was normal in 8.9% of statin nonusers, and this difference was not significant (P = 0.376). The odds for a normal CRP were 2.58 (95% CI: 0.38-13.57) times higher in statin users compared with statin nonusers. CRP was normal in 11.3% of NSAID users and was normal in 9.3% of NSAID nonusers, and this was not significant (P = 0.740). The odds for a normal CRP were 1.53 (95% CI: 0.34-7.96) times higher in NSAID users compared with NSAID nonusers.
Our study demonstrated that patients with biopsy-proven GCA who were taking NSAIDs or statins at the time of initial laboratory evaluation had a significantly lower median ESR than those not taking these medications. This finding was expected for NSAIDs given the previous similar findings by Helms et al (7). The finding that statins have an effect on ESR was expected on a theoretical basis in our patient population given the possible anti-inflammatory effects of statins. The finding that statin use was not associated with decreased CRP levels was unexpected, given the previous studies that found the opposite (8-10). As expected, patients with a higher ESR tended to be older and patients with a lower ESR tended to be younger. Our finding that statin or NSAID use was associated with lower median ESR values is important because this may mean that the ESR measurement may have lower sensitivity and specificity for patients with GCA. It should be emphasized that despite the focus on ESR and CRP in this study, the diagnosis of GCA is a clinical one and clinicians should interpret the results of ESR and CRP values in the context of the entire clinical picture when considering a diagnosis of GCA (14).
We recognize the limitations of our work. First, this study is retrospective and susceptible to the flaws of such a design, including recall, selection, and ascertainment bias. Patients may not have been asked specifically about over-the-counter use of NSAIDs, and thus, we may have excluded additional cases from our study. Second, patients on NSAIDs or statins may have been receiving those medications for the treatment of disorders that could influence the ESR and CRP values; however, we attempted to exclude all patients with diseases known to affect ESR and CRP. Third, we did not have independent verification (e.g., pharmacy records) that the patients were taking the reported medications consistently, and for many cases, we did not know the dosage. Thus, dose-related analysis of the association between NSAID or statin use and the reduction in acute-phase reactants seen in our study could not be performed. Finally, the sample size may limit the general applicability of our conclusions. Despite these limitations, we believe that the results of our study suggest that biopsy-proven GCA patients treated with NSAIDs or statins have a lower median ESR than those not treated with these medications.
1. Amos RS,
Crockson RA, Crockson AP, Walsh L, McConkey B. Rheumatoid arthritis: C-reactive protein and erythrocyte sedimentation rate during initial treatment. Br Med J. 1978;1:1396-1397.
2. McConkey B,
Crockson RA, Crockson AP, Wilkinson AR. The effects of some anti-inflammatory drugs on the acute phase proteins in rheumatoid arthritis. Q J Med. 1973;42:785-791.
3. Mallya RK,
deBeer FC, Berry H, Hamilton ED, Mace BE, Pepys MB. Correlation of clinical parameters of disease activity in rheumatoid arthritis with serum concentration of C-reactive protein and erythrocyte sedimentation rate. J Rheumatol. 1982;9:224-228.
4. Dixon JS,
Bird HA, Sitton NG, Pickup ME, Wright V. C-reactive protein in the serial assessment of disease activity in rheumatoid arthritis. Scand J Rheumatol. 1984;13:39-44.
5. Tishler M,
Caspi D, Yaron M. C-reactive protein levels in patients with rheumatoid arthritis: the impact of therapy. Clin Rheumatol. 1985;4:321-324.
6. Grindulis KA,
Calverley M, Constable TJ, Forster PJ, Ahmed ME, McConkey B. A comparison between clinical and laboratory tests in rheumatoid arthritis. Scand J Rheumatol. 1983;12:285-288.
7. Helms RA,
Quan DJ, Herfindal ET, Gourley DR. Rheumatoid arthritis. In: Boyce EG, ed. Textbook of Therapeutics, Drug and Disease Management, 8th edition. New York, NY: Lippincott Williams and Wilkins; 2006:1705-1731.
8. Fisher M,
Cushman M, Knappertz V, Howard G. An assessment of the joint associations of aspirin and statin use with C-reactive protein concentration. Am Heart J. 2008;156:106-111.
9. Kinlay S
. Low-density lipoprotein-dependent and -independent effects of cholesterol-lowering therapies on C-reactive protein. J Am Coll Cardiol. 2007;49:2003- 2009.
10. Pearson T,
Ballantyne C, Sisk C, Shah A, Veltri E, Maccubbin D. Comparison of effects of ezetimibe/simvastatin versus simvastatin, versus atorvastatin in reducing C-reactive protein and low-density lipoprotein cholesterol levels. Am J Cardiol. 2007;99:1706- 1713.
11. Constable TJ,
Crockson RA, Crockson AP, McConkey B. Drug treatment of rheumatoid arthritis. Lancet. 1975;1:1176-1180.
12. Miller A,
Green M, Robinson D. Simple rule for calculating normal erythrocyte sedimentation rate. Br Med J. 1983;286:866.
13. Tishler M,
Caspi D, Yaron M. C-reactive protein levels in patients with rheumatoid arthritis: the impact of therapy. Clin Rheumatol. 1985;4:321-324.
14. Hayreh SS,
Podhajsky PA, Zimmerman B. Occult giant cell arteritis: ocular manifestations. Am J Ophthalmol. 1998;125:521-526.