To assess whether the intensity of the initial systemic inflammatory response is able to predict response to therapy in patients with giant cell arteritis (GCA).
To assess whether the intensity of the initial systemic inflammatory response is able to predict response to therapy in patients with giant cell arteritis (GCA).
Retrospective review of 75 patients (49 women and 26 men) with biopsy-proven GCA who had regular followup and were treated according to uniform criteria. Four parameters were used to evaluate the baseline inflammatory response at diagnosis: fever, weight loss, erythrocyte sedimentation rate ≥ 85 mm/hour, and hemoglobin < 110 gm/liter. Patients were considered to have a weak inflammatory response if they had 2 or fewer inflammatory parameters (group 1) and a strong inflammatory response if 3 or 4 parameters were present (group 2). Time required to achieve a maintenance dose of less than 10 mg prednisone/day was recorded and analyzed by the Kaplan–Meier survival analysis method. Tumor necrosis factor α (TNFα) and interleukin 6 (IL-6) serum levels were also determined in 62 patients and 15 controls.
Forty patients had a weak (group 1) and 35 had a strong (group 2) initial inflammatory response. Patients in group 2 had significantly higher levels of circulating TNFα (31.9 ± 16.8 versus 22.3 ± 9 pg/ml; P = 0.01) and IL-6 (28.2 ± 17.4 versus 16.6 ± 13 pg/ml; P = 0.004) than patients in group 1. In group 1, 50% of patients required a median of 40 weeks (95% CI 37–43) to reach a maintenance dose of <10 mg, whereas in group 2 a median of 62 weeks (95% CI 42–82) was necessary (P = 0.0062). Patients in group 2 experienced more flares than patients in group 1 (P = 0.01) and received higher cumulative steroid doses (8.974 ± 3.939 gm versus 6.893 ± 3.075 gm; P = 0.01).
GCA patients with a strong initial systemic inflammatory reaction have more elevated circulating levels of IL-6 and TNFα, have higher and more prolonged corticosteroid requirements, and experience more disease flares during corticosteroid therapy than patients with a weak systemic acute phase response.
Giant cell (temporal) arteritis (GCA) is a chronic granulomatous vasculitis preferentially targeting large- and medium-sized arteries (1). Most of the classic disease manifestations result from symptomatic involvement of the carotid artery branches. Typical symptoms include headache, jaw claudication, scalp tenderness, and a variety of aches in the craniofacial area (1, 2). Ischemic complications derived from vessel occlusion include visual loss mainly due to anterior ischemic optic neuritis and, less frequently, stroke or scalp necrosis (3).
In addition to the above-mentioned clinical manifestations, GCA is a disease characterized by a prominent systemic inflammatory reaction (1–3). The acute phase response to infection or injury is a complex and not completely understood phenomenon that, globally, is thought to be protective and meant to avoid excessive tissue destruction. It encompasses a series of reactions distant from the areas of inflammation in which many organs and systems participate. As a consequence, such clinical manifestations as fever, anorexia, weight loss, hematologic abnormalities (i.e., anemia and thrombocytosis), biochemical alterations (acute phase protein synthesis), and metabolic changes (i.e., increased lipolysis and muscle loss) characteristically occur. This systemic reaction to injury is driven by proinflammatory cytokines, mostly interleukin 1 (IL-1), tumor necrosis factor α (TNFα), and IL-6, which are produced mainly by macrophages at the sites of inflammation. Although most cytokines act in an autocrine/paracrine fashion, proinflammatory cytokines drive the acute-phase response in a distant, systemic way. They have pleiotropic effects on a variety of cells that, in turn, secrete a wide array of products. A complex network of stimulatory and inhibitory mediators determines, eventually, the intensity of the acute phase response (4, 5).
Approximately 50% of GCA patients experience fever and weight loss. Most patients have remarkably elevated erythrocyte sedimentation rates (ESR) and chronic anemia (1–3). Acute phase proteins such as C-reactive protein (CRP), orosomucoid, and haptoglobin are elevated in a substantial proportion of patients (3, 6, 7). However, the intensity of the acute phase response is highly variable among patients. Patients with no constitutional symptoms (3, 8) and normal or near-normal ESR have been reported repeatedly (3, 9–12). The mechanisms underlying this variability have not been investigated.
Corticosteroids are the treatment of choice for patients with GCA; they rapidly relieve most symptoms in the majority of cases. However, the duration of corticosteroid therapy is highly variable (13, 14). Some individuals achieve a sustained remission after a few months of treatment. Most patients require 1–2 years of therapy and some patients require long-term corticosteroid therapy. Some patients are able to maintain remission with less than 10 mg/day of prednisone, and other patients require at least 20 mg/day to remain asymptomatic (13–15). Corticosteroid-related morbidity is elevated in patients with GCA, and iatrogenic complications are heavily influenced by the intensity and duration of corticosteroid treatment (14, 16). To date, no clinical or analytic parameters have been identified that can consistently predict the intensity and duration of corticosteroid therapy in a large and homogeneous cohort of people with GCA.
The goals of our study were to determine whether circulating levels of proinflammatory cytokines are related to the intensity of systemic inflammatory response in GCA and to assess whether the intensity of the systemic inflammatory response may be an indicator of the magnitude and duration of corticosteroid treatment.
The study group consisted of 75 patients, 49 women and 26 men, with biopsy-proven GCA diagnosed and treated at our Internal Medicine Department over a 14-year period. Patients were selected consecutively among those who had regular followup. Patients who were transferred to another institution or treating physician, had low compliance, or died early (within 3 months) in the course of the disease were excluded. Although this study is retrospective in design, patients included were evaluated and treated by the authors according to uniform criteria. The treatment schedule began with an initial prednisone dose of 1 mg/kg/day (up to 60 mg/day) for 1 month. Subsequently, prednisone was tapered by 5 mg/week. Reductions below 20 mg/day were slower and individualized. A rate of 2.5 mg every 3 months was attempted. A disease flare was considered when ESR rose above 50 mm/hour and disease-related manifestations (cranial symptoms, polymyalgia rheumatica, fever, or malaise) appeared or hemoglobin fell below 110 gm/liter. When clear and worsening symptoms occurred with a normal or slightly elevated ESR, a flare was also considered. When ESR rose with no clinical symptoms or anemia, the maintenance dose was held until it went back to normal or a flare could be defined. When a disease flare was suspected, prednisone was increased to 10 mg above the previous effective dose; to be fully considered a flare, symptoms had to remit after adjusting the prednisone dose.
Data recorded at entry included age, sex, number and type of cranial symptoms, transient or permanent ischemic complications, polymyalgia rheumatica, fever (>37°C), weight loss (>5 kg), duration of clinically symptomatic disease before diagnosis and time (weeks) of followup. Laboratory parameters included ESR, CRP, hemoglobin (Hgb), haptoglobin, γ-glutamyl transpeptidase, alkaline phosphatase, albumin, α2-globulin, and platelet count.
To evaluate the initial systemic inflammatory response, the following 4 parameters were considered: fever, weight loss, ESR ≥ 85 mm/hour, and Hgb < 110 gm/liter because they have been previously demonstrated to be useful in discriminating patients at high and low risk of developing ischemic events (3). Patients were considered to have a weak systemic inflammatory response when they had 2 or fewer inflammatory parameters (group 1) and a strong systemic inflammatory response when they had 3 or 4 parameters (group 2). The time (weeks) required to achieve a maintenance dose of <10 mg prednisone per day and the cumulative dose of prednisone received at that point were recorded. The number of flares and the number of patients out of treatment at the end of followup were also included.
Sera was obtained from 62 patients (36 from group 1 and 26 from group 2) with active disease and from 15 age- and sex-matched healthy individuals. Aliquots were frozen and stored at −80°C until use. IL-1β, TNFα, and IL-6 concentrations were determined by enzyme-linked immunosorbent assay (ELISA). Commercially available ELISA kits for TNFα were obtained from Medgenix (Fleurus, Belgium) and kits for IL-1β and IL-6 from Genzyme (Minneapolis, MN). The assays were performed according to the manufacturer's instructions.
Fisher's exact test was used for qualitative comparisons. For quantitative comparisons among groups of individuals, Student's unpaired t test was employed. The Pearson's correlation coefficient was used. The time required to achieve a maintenance prednisone dose of <10 mg/day was compared between group 1 and group 2 by the Kaplan–Meier survival analysis method.
According to the previously mentioned criteria, 40 patients had a weak (group 1) and 35 had a strong (group 2) systemic inflammatory response at first evaluation. The main clinical findings in both groups of patients are summarized in Table 1. In accordance with previously reported data (3, 17, 18), ischemic events were significantly more frequent in patients with a weak systemic inflammatory response. No differences were found in the distribution of other major clinical manifestations between both groups of patients except for the clinical criteria (fever and weight loss) used to define them. No differences existed in the duration of clinically apparent disease between the groups, suggesting that differences in the intensity of the systemic inflammatory response may be constitutive rather than reflect early or late time points in the course of the disease. No differences existed in the duration of followup.
|Clinical characteristics||Group 1 (n = 40)||Group 2 (n = 35)|
|Age in years, mean (range)||76 (57–90)||73 (58–87)|
|Duration of symptoms in weeks, mean (range)||14 (1–80)||16 (2–104)|
|Followup time in months, mean (range)||31 (4–84)||40 (4–166)|
|Cranial symptoms (%)|
|Abnormal temporal arteries*||83||79|
|Ischemics events (%)||30†||9|
|Symptomatic involvement of other vascular territories (%)||2.5||3|
|Systemic manifestations (%)|
In addition to Hgb and ESR values employed as criteria to define both groups of patients, parameters related to the acute phase response, such as CRP, haptoglobin, platelet count, and α2 globulins, were more elevated in patients with a strong systemic inflammatory reaction (Table 2). In contrast, albumin, a negatively-regulated protein during the acute phase response, was lower in patients with a strong systemic inflammatory reaction.
|Parameter||Group 1 mean (range)||Group 2 mean (range)|
|ESR, mm/hour||80 (28–130)†||114 (65–147)|
|CRP, mg/dl||4.7 (0.5–25.5)‡||12 (1.9–33.3)|
|Haptoglobin, gm/liter||3.221 (0.079–6.770)§||4.877 (3.024–7.490)|
|Hemoglobin, gm/liter||120 (66–156)†||98 (75–119)|
|Platelets, × 109/liter||315 (105–493)¶||378 (130–768)|
|Albumin, gm/liter||35 (24–42)#||32 (25–43)|
|α2-globulin, gm/liter||9 (5–18)¶||10.4 (4.4–20)|
|Alkaline phosphatase, units/liter||219 (139–450)||304 (98–1682)|
|γ-glutamyl transpeptidase, units/liter||32 (8–140)||63 (10–383)|
Circulating TNFα levels were moderately but significantly higher in GCA patients (26.4 ± 13.7 pg/ml) than in controls (16 ± 9.5 pg/ml; P = 0.007) (Figure 1). As was shown previously (19–21), IL-6 concentrations were more elevated in patients than in healthy controls (21.4 ± 16 pg/ml versus 5 ± 11 pg/ml; P = 0.0004) (Figure 2). In addition, remarkable differences in the concentrations of both cytokines were observed between patients with a strong systemic inflammatory response and patients with a weak inflammatory reaction. TNFα concentrations were 22.3 ± 9 pg/ml in group 1 and 31.9 ± 16.8 pg/ml in group 2 (P = 0.01) and IL-6 concentrations were 16.6 ± 13 pg/ml in group 1 and 28.2 ± 17.4 in group 2 (P = 0.004) (Figures 1 and 2), indicating that TNFα and IL-6 may participate in the development of the acute phase response in GCA. TNFα concentrations correlated positively with ESR values (r = 0.364, P = 0.018) and haptoglobin levels (r = 0.448, P = 0.022), and negatively with hemoglobin concentration (r = −0.329, P = 0.033). Similarly, IL-6 levels significantly correlated with CRP (r = 0.378, P = 0.025). In contrast, IL-1β was below the detection threshold in most patients and controls.
Corticosteroid requirements were significantly higher in patients with a strong systemic inflammatory response. While in group 1, 50% of patients required a median of 40 weeks (95% CI 37–43) to reach a maintenance dose of prednisone lower than 10 mg/day; in group 2, 50% of patients required a median of 62 weeks to reach maintenance dose (95% CI 42–82; P = 0.0062) (Figure 3). The cumulative dose of prednisone received during this period was 6.893 ± 3.075 gm in group 1 and 8.974 ± 3.939 gm in group 2 (P = 0.01). During the followup, 22 of 40 (55%) patients in group 1 and 27 of 35 (77%) in group 2 experienced at least 1 disease flare (P = 0.054). The main manifestation of a GCA recurrance was headache in 29 flares, polymyalgia rheumatica in 37, fever in 7, malaise in 12, and anemia in 5. Headache was more frequent in flares of patients in group 1 (17 versus 12; P = 0.0046) and malaise was slightly more frequent in patients in group 2 (P = 0.0495). No additional differences in the nature of flares were observed between groups. Nine (22.5%) patients in group 1 and 18 (51.4%) in group 2 had more than 1 disease flare (P = 0.01). At the end of followup, 17 (42.5%) patients in group 1 and 6 (17%) patients in group 2 were out of treatment (OR 3.6; CI 95% 1.2–10.5; P = 0.02). Six (15%) patients in group 1 and 7 (20%) patients in group 2 had died by the end of followup (P = NS).
GCA is a disease characterized by a strong acute phase response, and proinflammatory cytokine transcripts, namely IL-1β, TNFα, and IL-6 have been detected in temporal artery lesions by reverse transcriptase–polymerase chain reaction and in situ hybridization (22, 23). However, when evaluating the influence of these cytokines in the development of the systemic inflammatory response, determining circulating levels may be more significant than their detection in tissue. First, most cytokines have instability sequences at their 3` untranslated region and are subjected to a strict postranscriptional regulation, and the amount of mRNA at a given time point may not represent the actual resulting protein synthesis. Second, activated circulating monocytes may also contribute to the production of cytokines in the bloodstream (19, 23); and third, circulating levels of cytokines may better reflect their overall systemic effects. With the exception of IL-6, previous attempts to determine circulating levels of cytokines have been performed mostly in patients with polymyalgia rheumatica, including just a few patients with GCA (24–28). In a homogeneous series of 19 and 20 patients with biopsy-proven GCA, Roche et al (19) and Roblot et al (20) found elevated levels of circulating IL-6 in patients with active disease. A trend towards elevated levels of TNFα in patients versus controls was also found, but the difference was not significant, probably due to the small number of patients included and to the fact that, according to our data, TNFα may only be elevated in a subset of patients. In an extended prospective followup study including 25 patients, Weyand et al (21) found IL-6 to be a sensitive marker of disease activity, but no correlation with clinical findings was investigated. In the present study of a large and homogeneous series of patients with biopsy-proven GCA, we found elevated levels of both TNFα and IL-6 in sera from active patients compared to controls. In addition, in active patients, IL-6 levels correlated with CRP concentrations; TNFα levels correlated positively with haptoglobin and ESR values and inversely with Hgb concentrations. Both TNFα and IL-6 levels were significantly higher in patients with strong overall systemic inflammatory reaction evaluated with previously established clinical and analytic parameters. Taken together, these data indicate an important role for circulating TNFα and IL-6 in the pathogenesis of the acute phase response in GCA. In contrast, although IL-1β mRNA can be detected in temporal artery lesions from patients with GCA (22, 23), serum IL-1β levels were below the detection threshold in most patients. This result supports the concept that tissue cytokine mRNA may not correlate with circulating cytokine concentrations and suggests a less relevant participation of IL-1β in the generation of the systemic inflammatory response in GCA.
Corticosteroid requirements are highly variable among patients with GCA (13–15). Although some patients treated for a few months sustain remission, others require long-term therapy and higher-than-desirable maintenance prednisone doses with their ensuing iatrogenic complications (13–16). To date, no clinical or analytic findings able to predict the outcome of patients with GCA have been identified in large and homogeneous series of patients. Our results indicate that the intensity of the initial systemic inflammatory reaction is a major predictor of disease duration and corticosteroid requirements. Patients with a strong initial systemic inflammatory response evaluated with previously established clinical and analytic parameters have more disease flares and require a significantly longer duration of corticosteroid therapy. A similar trend has been observed by other investigators. In this regard, Weyand et al (29) found that elevated pretreatment ESR was associated with longer duration of corticosteroid treatment in a series of 27 patients with polymyalgia rheumatica.
We have previously published that patients with a strong inflammatory reaction have a lower risk of developing ischemic complications (3). The reason the intensity of the initial systemic inflammatory response is able to delineate patient subpopulations with different prognoses is unknown. As suggested by our results, an intense systemic inflammatory reaction may reflect higher proinflammatory cytokine production. Higher cytokine production may be constitutive in some patients, caused by more widespread inflammatory lesions; by a more sustained, self-perpetuating, inflammatory response; or due to a combination of these or other factors. The intensity of the acute phase response probably reflects upstream cytokine and growth factor production, which influences vessel permeability and remodeling and determines the fate of inflammatory lesions in GCA (30–35). In this regard, we have previously shown that patients with strong systemic inflammatory reactions have more striking inflammation-induced angiogenesis and expression of endothelial adhesion molecules for leukocytes in their lesions (36, 37). Taken together, our data suggest that some patients would develop an obliterative, self-limiting disease with high risk of vessel occlusion and ischemic events, whereas other patients would develop a chronic self-perpetuating disease. In the latter, continuous release of unknown mediators would prevent vessel occlusion, and neovessels would, at distal sites, compensate for ischemia but, at the same time, would continue recruiting leukocytes through adhesion molecule expression. The intensity of acute phase response, although probably an epiphenomenon derived from more directly related upstream events, would be able to distinguish between these 2 disease patterns.
Our data suggest that different cytokine production might, at least partially, account for these 2 different disease patterns. Both IL-6 and TNFα levels correlated with the intensity of the inflammatory response in our patients. TNFα is upstream of IL-6 production in many macrophage responses and it is one of its major inducers (4, 5, 38, 39). Perhaps TNFα blockade, which appears to be promising in several chronic inflammatory disorders including rheumatoid arthritis and inflammatory bowel disease (40–43), and which is currently being tested in other vasculitides such as Wegener's granulomatosis (44), might also be of help for GCA patients, particularly those with strong systemic inflammatory response and high TNFα who, according to our results, have higher and longer corticosteroid requirements.
The authors would like to express appreciation to the nursing staff of the Internal Medicine and Emergency departments for their invaluable contributions.