Abstract
Objective. To evaluate the usefulness of anticardiolipin antibodies (aCL) in identifying flares and relapses in giant‐cell arteritis.
Methods. We studied 58 consecutive patients with biopsy‐proven temporal giant‐cell arteritis. C‐reactive protein and aCL serum levels were measured simultaneously at the time of diagnosis and at each out‐patient visit until recovery. All observed episodes of a rise in C‐reactive protein attributable to a precise cause, for which the simultaneous measurement of aCL was available, were analysed.
Results. The mean duration of clinical observation and serum aCL assessment was 34 ± 18 and 24 ± 11 months, respectively. Anticardiolipin antibody positivity (IgG or total antibodies ≥20 U) before treatment was found before treatment in 27 cases (46.6%) (mean 45.6 ± 26 U/l, range 20–110 U). Levels of aCL decreased below 10 U with appropriate treatment in all patients except one, after a variable delay. No rise in aCL levels was recorded subsequently in any patient whose disease was controlled permanently. A significant rise in aCL was recorded in 20 of 27 (74%) of the flares or relapses of giant‐cell arteritis, including seven of 12 flares in seven patients whose initial aCL level was <20 U vs none of the 28 inflammatory episodes unrelated to giant‐cell arteritis (P < 0.0000001). IgM aCL, infrequently found at diagnosis, was not associated with signs of disease activity.
Conclusion. Serum aCL levels are useful in the detection of flares and relapses in giant‐cell arteritis, with fairly good sensitivity (74%) and a specificity of 100%, and can be of value in distinguishing subclinical flares from infection.
The course of giant‐cell arteritis (GCA) is characterized by the frequent occurrence of vascular complications and a high rate of flares or relapses [1], which occur during low‐dose steroid treatment or within the first year after cessation of therapy. Although some flares and relapses are of acute onset and have clinical symptoms suggestive of temporal arteritis (TA) or polymyalgia rheumatica (PMR), many episodes have a gradual, subclinical onset, and the almost invariably associated inflammatory response is therefore not readily ascribed to GCA. Notably, intercurrent sepsis, which is especially frequent in elderly patients being treated with corticosteroids, must be carefully searched for. For these reasons, delayed or inaccurate therapeutic decisions may occur and hospitalization, with its additional cost, may be required for further investigation.
The presence of anticardiolipin antibody (aCL), predominantly of the IgG isotype or a mixture of IgG and IgM, has been found in 32–50% of cases of untreated GCA [2–6]. Anticardiolipin antibody positivity correlates with a temporal artery biopsy positive for GCA changes [6] and a low haemoglobin level [5], but not with ischaemic complications [3–9]. These antibodies are generally present at moderate levels and are never found in association with anti‐β2‐glycoprotein I antibodies [10]. With corticosteroid treatment, aCL disappear within 3 months in approximately 80% of cases [4, 5, 10–13], indicating that the initial positivity might be linked to disease activity. In fact, the rate of aCL positivity at 6 months is still high (up to 44%) [4, 5, 12–14], possibly because of persistently active vasculitis, despite corticosteroid treatment, in some patients [4, 15]. However, there has been no sufficiently long study which has proven that a relationship exists between the reappearance of aCL and flares of GCA.
We undertook a prospective study to evaluate the fluctuations of aCL in patients with GCA and the temporal relationship of this antibody with flares and relapses of GCA.
Patients and methods
Patient characteristics and treatment regimens
We studied prospectively 58 consecutive patients (36 women and 22 men, mean age 74 yr) seen in two internal medicine departments since 1990. The diagnosis of GCA was established clinically and was proven in all cases by characteristic changes on temporal artery biopsy [16]. Thirty‐one patients presented with isolated TA, 13 with symptoms of both TA and PMR and eight with constitutional symptoms alone. In addition, involvement of upper limb arteries was demonstrated in seven patients and transient or permanent ophthalmological ischaemic complications (OIC) occurred in 14 patients (24.1%). No patient had a history of recurrent thromboses or shared any feature, such as livedo, thrombocytopenia or an atypical location of thrombosis, consistent rather with the primary antiphospholipid antibody syndrome than with TA. Patients without initial OIC received initially prednisone 0.35 mg/kg every 12 h for at least 2 weeks, until they were asymptomatic with a normal serum concentration of C‐reactive protein (CRP). The dose of prednisone was tapered to half the initial dose over the next 6 weeks and was then reduced slowly (by 1 mg per month); the total duration of treatment was at least 2 yr. Patients with initial OIC were administered pulse methylprednisolone 0.3–0.5 g per day for 3 days followed by prednisone 0.5 mg/kg every 12 h, and was then tapered as described above. After a mean follow‐up of 34 months (range 10–78 months), 32 patients had completed their treatment, 22 of them with a high probability of recovery (no treatment for more than 18 months), three patients had died (two deaths related to malignancy) and three had been lost to follow‐up.
Flares and relapses
A flare was defined as the recurrence of clinical symptoms suggestive of TA and/or PMR and/or an elevation of the erythrocyte sedimentation rate (ESR) and CRP (at least 3‐fold higher than the upper limit of normal) during therapy which remained unexplained after appropriate work‐up (including at least complete blood tests, uroculture and roentgenograms of the chest and sinuses) but disappeared promptly upon increasing the dose of corticosteroids. A relapse was defined similarly, but in the absence of therapy. Twenty‐one patients experienced at least one flare or relapse of GCA, with a total of 29 episodes (16 flares and 13 relapses). Six patients experienced more than one flare and/or relapse. Whereas 12 inflammatory events were easily attributable to GCA because suggestive signs and/or symptoms were present, the other 17 were not and required further investigation. The treatment of relapses and flares was not standardized.
Methods
Biological parameters were obtained in each patient at the time of diagnosis, before any treatment had been instituted, and were monitored throughout the course of the illness until recovery. CRP, because of its sensitivity and fast kinetics compared with ESR and other inflammatory parameters [17], was chosen as the standard for laboratory investigation. Other reasons justifying our choice were that, unlike ESR, normal values of CRP do not vary with increasing age or digestion, and that factors influencing plasma viscosity other than a specific inflammatory response have no effects on its serum level. Levels of both aCL and CRP were assessed in serum twice during the first month, monthly during the first trimester and every 6 months thereafter. In addition, the investigators attempted to assess promptly any persistent inflammatory episode reported by the patient's family physician.
Before 1993, only total aCL were investigated (10 patients); after that time, IgG and IgM aCL were determined in stored sera using an enzyme‐linked immunoassay technique according to Loizou et al. [18]. Since we found isolated IgM aCL positivity in only one of 95 untreated patients with GCA (1.1%), including the present series, we assigned arbitrarily the total aCL positivity to the presence of an IgG isotype. Polystyrene plates (Immunosorb, Nunc) were coated with cardiolipin purified from bovine heart (Sigma, St Louis, MO) and saturated with bovine serum albumin (Diamed, Cressier/Morat, Switzerland) for 1 h. After 2 h at 37°C, 1:100 dilutions of sera were also studied in triplicate with coated and uncoated wells to determine the correct optical density, and were reported with respect to the standard reference sera of Harris et al. [19] calibrated in IgM and IgG antibodies, in order to express the levels of aCL in IgM and IgG units (positive: >20 U).
Statistical analysis
The characteristics and outcome of patients with an initial serum aCL level >20 U (aCL+) and <20 U (aCL−) were compared using either the χ2 test or the Mann–Whitney U‐test. We decided to exclude the results of IgM aCL from statistical analysis for two reasons. First, as indicated above, their isolated positivity is seen infrequently in GCA [2, 5, 6]; secondly, this isotype of aCL is more liable to be triggered by an infectious cause or a drug side‐effect [20], and in GCA it does not seem to be linked to disease activity [2]. The relationship between the initial level of aCL and the time when the level became negative was assessed with Spearman's correlation test. The association of increased serum aCL positivity with a flare/relapse of GCA was determined with the χ2 test.
Results
For the whole series, the median initial level of aCL was 23.7 ± 27.4 U (range 0–110 U). The average IgG level was 24.1 U (range 0–110 U) and the average IgM level 5.6 U (range 0–67 U). Twenty‐seven patients (46.6%) were aCL+ (mean 45.6 ± 26 U; range 20–110 U); of these patients, only one (1.7%) had both IgG and IgM isotypes and the other 26 had total aCL and/or IgG isotypes. Thirty‐one patients (53.4%) were aCL− (mean 4.1 ± 3 U; range 0–16 U). No case of isolated IgM positivity was detected. The characteristics of aCL+ and aCL− patients are compared in Table 1. The two groups were statistically indistinguishable with respect to the frequency of OIC (P = 0.2), the occurrence of flares and relapses (P = 0.86) and the number of patients who recovered (P = 0.46). In addition, there was no relationship between the initial level of aCL and the risk of relapse (P = 0.18) or the probability of achieving recovery (P = 0.42).
Once corticosteroid treatment had been introduced, the level of aCL decreased below 10 U in all but one patient. Most cases (81%) became aCL− within 3 months (mean 38 days). However, aCL positivity persisted 6–14 months in four cases and indefinitely (up to 30 months) in a fifth case. All five patients had high aCL levels initially (mean 83 U; range 61–110 U). Furthermore, the higher the initial level of aCL, the longer was the time to its disappearance from serum (P = 0.0058). In the only patient with IgG and IgM aCL (47 and 67 U, respectively), both isotypes disappeared from the serum within 2 months. This patient did not relapse.
We did not detect a significant secondary increase in IgG and IgM aCL in any patient whose disease was controlled permanently, either during or off therapy. Conversely, a significant rise in aCL (total and/or IgG) was recorded in 20 of 27 (74.2%) analysable inflammatory episodes related to GCA, occurring in 21 patients (Table 2): 13 of 15 episodes occurred in 14 aCL+ patients and seven of 12 episodes occurred in seven aCL− patients. Moreover, a transient weak increase, close to the limit of positivity, was seen during a flare in another patient. Therefore, aCL levels did not increase in only 22.2% of the assessable inflammatory episodes related to GCA. Among the 19 episodes assessable for the kinetics of serum aCL, the observed increases in aCL and CRP were synchronous in 14 episodes, whereas the rise in aCL level in the others obviously preceded (two episodes) or succeeded (three episodes) that of CRP. After corticosteroid treatment had been reintroduced or modified, aCL decreased below 10 U/l within 2 months in five episodes and persisted 3 months or more in eight other assessable episodes.
In contrast with this finding, no aCL positivity was recorded in any of the 28 analysable GCA‐unrelated inflammatory flares (P < 0.0000001). These 28 episodes, summarized in Table 3, involved 22 cases of well‐documented infection in 17 patients (of whom 10 belonged to the aCL+ group) and six inflammatory episodes unrelated to GCA or sepsis in six patients.
Comparison of aCL+ and a CL− patients
| aCL+ patients (n = 27) | aCL− patientsa (n = 31) | Total (n = 58) | |
| Mean age (yr) | 74.5 | 73.8 | 74.1 |
| Female | 14 (51.9%) | 20 (64.5%) | 34 (58.6%) |
| TA alone | 16 (59.3%) | 20 (64.5%) | 36 (62.1%) |
| TA+PMR | 7 (25.9%) | 6 (19.4%) | 13 (22.4%) |
| PMR alone | 0 | 1 (3.2%) | 1 (1.7%) |
| Constitutional symptoms | 4 (14.8%) | 4 (12.9%) | 8 (13.8%) |
| alone | |||
| Subclavian–axillar | 4 (14.8%) | 3 (9.7%) | 7 (12.1%) |
| involvement | |||
| OIC | 9 (33.3%) | 5 (16.1%) | 14 (24.1%) |
| Mean follow‐up (months) | 36.4 | 33.7 | 34.5 |
| Relapse/flare | 13 (48.1%) | 9 (29%) | 22 (37.9%) |
| Treatment completed | 15 (55.6%) | 17 (54.8%) | 32 (55.2%) |
| Cure achieved | 10 (37%) | 12 (38.7%) | 22 (37.9%) |
| aCL+ patients (n = 27) | aCL− patientsa (n = 31) | Total (n = 58) | |
| Mean age (yr) | 74.5 | 73.8 | 74.1 |
| Female | 14 (51.9%) | 20 (64.5%) | 34 (58.6%) |
| TA alone | 16 (59.3%) | 20 (64.5%) | 36 (62.1%) |
| TA+PMR | 7 (25.9%) | 6 (19.4%) | 13 (22.4%) |
| PMR alone | 0 | 1 (3.2%) | 1 (1.7%) |
| Constitutional symptoms | 4 (14.8%) | 4 (12.9%) | 8 (13.8%) |
| alone | |||
| Subclavian–axillar | 4 (14.8%) | 3 (9.7%) | 7 (12.1%) |
| involvement | |||
| OIC | 9 (33.3%) | 5 (16.1%) | 14 (24.1%) |
| Mean follow‐up (months) | 36.4 | 33.7 | 34.5 |
| Relapse/flare | 13 (48.1%) | 9 (29%) | 22 (37.9%) |
| Treatment completed | 15 (55.6%) | 17 (54.8%) | 32 (55.2%) |
| Cure achieved | 10 (37%) | 12 (38.7%) | 22 (37.9%) |
aNone of the differences reached statistical significance.
Summarized data of cases of GCA whose flares or relapses were assessed with both CRP and aCL
| Age | aCL at diagnosis (U/l) | aCL: time to decrease below 10 U/l with | Relapse or flare of GCA (witha | Chronology | aCL at relapse (U/l) | aCL: time to | |||||||||
| Patient | (yr) sex | Total | IgG | IgM | treatment (days) | or without clinical symptoms) | between rises in aCL and CRP | Total | IgG | IgM | subsequent disappearance | ||||
| 1 | 76/F | 68 | 110 | 0 | 240 | Flare at 15 months | Synchronous | — | 76 | 0 | n.a. | ||||
| 2 | 67/M | 61 | — | — | 320 | Relapse at 1 month | Synchronous | 37 | — | — | 6 months | ||||
| 2nd relapse | n.a. | 22 | — | — | n.a. | ||||||||||
| 3 | 78/M | 21 | — | — | 45 | Relapse at 2.5 months | Synchronous | 20 | — | — | 2 months | ||||
| 4 | 72/F | — | 26 | 10 | 100 | Relapse at 2 months | Synchronous | — | 41 | 8 | n.a. | ||||
| 5 | 71/F | 53 | — | — | 21 | aRelapse at 3 months | aCL followed by | 20 | — | — | 45 days | ||||
| 1 month | |||||||||||||||
| 6 | 71/M | — | 64 | 9 | 70 | aFlare at 17 months | Synchronous | — | 34 | 8 | n.a. | ||||
| 7 | 76/F | 32 | 36 | 0 | 45 | Relapse at 6 months | aCL preceded by | — | 64 | 0 | 4 months | ||||
| 3 months | |||||||||||||||
| 8 | 67/F | 16 | — | — | 45 | aFlare at 14 months | n.a. | 5 | — | — | n.a. | ||||
| 9 | 78/F | 89 | — | — | 60 | Relapse at 2 months | aCL followed by | 28 | — | — | 5 months | ||||
| 8 months | |||||||||||||||
| 10 | 88/F | — | 32 | 4 | 25 | aRelapse at 3 months | Synchronous | — | 42 | 8 | 45 days | ||||
| 11 | 70/M | 25 | 21 | 6 | 60 | Flare at 21 months | Synchronous | — | 28 | 5 | >6 months | ||||
| 12 | 73/M | 23 | 28 | 0 | 30 | Relapse at 3 months | Synchronous | — | 21 | 0 | 1 month | ||||
| 13 | 63/M | — | 45 | 0 | 30 | Flare at 14 months | aCL followed by | — | 37 | 0 | n.a. | ||||
| 1 month | |||||||||||||||
| 14 | 78/F | — | 41 | 5 | 15 | aFlare at 9 months | n.a. | — | 8 | 4 | |||||
| 15 | 80/M | 7 | — | — | aRelapse at 2 months | Synchronous | 20 | — | — | 15 days | |||||
| 16 | 77/F | 5 | — | — | Flare at 15 months | Synchronous | — | 47 | 0 | 16 months | |||||
| 2nd flare at 23 months | aCL preceded by | 55 | 49 | 7 | >3 months | ||||||||||
| 4 months | |||||||||||||||
| 17 | 74/F | — | 6 | 0 | aFlare at 2 months | n.a. | — | 5 | 0 | ||||||
| 2nd flare at 5 months | n.a. | — | 7 | 0 | |||||||||||
| 3rd flare at 19 months | Synchronous | — | 54 | 0 | n.a. | ||||||||||
| 18 | 74/M | — | 0 | 1 | aRelapse at 2 months | n.a. | — | 2 | 0 | ||||||
| 19 | 78/F | — | 3 | 2 | aFlare at 9 months | Synchronous | — | 18 | 0 | n.a. | |||||
| 2nd flare at 24 months | Synchronous | — | 45 | 0 | >6 months | ||||||||||
| 20 | 76/F | — | 1 | 0 | Flare at 3 months | n.a. | — | 4 | 1 | ||||||
| a2nd flare at 24 months | Synchronous | — | 56 | 1 | n.a. | ||||||||||
| 21 | 67/M | — | 14 | 6 | aFlare at 7 months | Synchronous | — | 22 | 9 | n.a. | |||||
| Age | aCL at diagnosis (U/l) | aCL: time to decrease below 10 U/l with | Relapse or flare of GCA (witha | Chronology | aCL at relapse (U/l) | aCL: time to | |||||||||
| Patient | (yr) sex | Total | IgG | IgM | treatment (days) | or without clinical symptoms) | between rises in aCL and CRP | Total | IgG | IgM | subsequent disappearance | ||||
| 1 | 76/F | 68 | 110 | 0 | 240 | Flare at 15 months | Synchronous | — | 76 | 0 | n.a. | ||||
| 2 | 67/M | 61 | — | — | 320 | Relapse at 1 month | Synchronous | 37 | — | — | 6 months | ||||
| 2nd relapse | n.a. | 22 | — | — | n.a. | ||||||||||
| 3 | 78/M | 21 | — | — | 45 | Relapse at 2.5 months | Synchronous | 20 | — | — | 2 months | ||||
| 4 | 72/F | — | 26 | 10 | 100 | Relapse at 2 months | Synchronous | — | 41 | 8 | n.a. | ||||
| 5 | 71/F | 53 | — | — | 21 | aRelapse at 3 months | aCL followed by | 20 | — | — | 45 days | ||||
| 1 month | |||||||||||||||
| 6 | 71/M | — | 64 | 9 | 70 | aFlare at 17 months | Synchronous | — | 34 | 8 | n.a. | ||||
| 7 | 76/F | 32 | 36 | 0 | 45 | Relapse at 6 months | aCL preceded by | — | 64 | 0 | 4 months | ||||
| 3 months | |||||||||||||||
| 8 | 67/F | 16 | — | — | 45 | aFlare at 14 months | n.a. | 5 | — | — | n.a. | ||||
| 9 | 78/F | 89 | — | — | 60 | Relapse at 2 months | aCL followed by | 28 | — | — | 5 months | ||||
| 8 months | |||||||||||||||
| 10 | 88/F | — | 32 | 4 | 25 | aRelapse at 3 months | Synchronous | — | 42 | 8 | 45 days | ||||
| 11 | 70/M | 25 | 21 | 6 | 60 | Flare at 21 months | Synchronous | — | 28 | 5 | >6 months | ||||
| 12 | 73/M | 23 | 28 | 0 | 30 | Relapse at 3 months | Synchronous | — | 21 | 0 | 1 month | ||||
| 13 | 63/M | — | 45 | 0 | 30 | Flare at 14 months | aCL followed by | — | 37 | 0 | n.a. | ||||
| 1 month | |||||||||||||||
| 14 | 78/F | — | 41 | 5 | 15 | aFlare at 9 months | n.a. | — | 8 | 4 | |||||
| 15 | 80/M | 7 | — | — | aRelapse at 2 months | Synchronous | 20 | — | — | 15 days | |||||
| 16 | 77/F | 5 | — | — | Flare at 15 months | Synchronous | — | 47 | 0 | 16 months | |||||
| 2nd flare at 23 months | aCL preceded by | 55 | 49 | 7 | >3 months | ||||||||||
| 4 months | |||||||||||||||
| 17 | 74/F | — | 6 | 0 | aFlare at 2 months | n.a. | — | 5 | 0 | ||||||
| 2nd flare at 5 months | n.a. | — | 7 | 0 | |||||||||||
| 3rd flare at 19 months | Synchronous | — | 54 | 0 | n.a. | ||||||||||
| 18 | 74/M | — | 0 | 1 | aRelapse at 2 months | n.a. | — | 2 | 0 | ||||||
| 19 | 78/F | — | 3 | 2 | aFlare at 9 months | Synchronous | — | 18 | 0 | n.a. | |||||
| 2nd flare at 24 months | Synchronous | — | 45 | 0 | >6 months | ||||||||||
| 20 | 76/F | — | 1 | 0 | Flare at 3 months | n.a. | — | 4 | 1 | ||||||
| a2nd flare at 24 months | Synchronous | — | 56 | 1 | n.a. | ||||||||||
| 21 | 67/M | — | 14 | 6 | aFlare at 7 months | Synchronous | — | 22 | 9 | n.a. | |||||
n.a. = not assessable.
Summarized data on cases of non‐GCA‐related flares assessed with both CRP and aCL
| aCL | GCA‐ related | Non‐GCA‐related inflammatory flare | |||||
| Patient | level at diagnosis (U/l) | flare (aCL, U/l) | Origin of the flare | CRP (mg/l) | aCL (U/l) | ||
| 5 | 53 | Yes (20) | Metastatic carcinoma | 80 | 5 | ||
| 7 | 36 | Yes (60) | Cystitis | 30 | 4 | ||
| Pyelonephritis | 83 | 7 | |||||
| 12 | 28 | Yes (20) | Subacute bronchitis | 24 | 0 | ||
| 14 | 41 | Yes (6) | Pneumonitis | 58 | 5 | ||
| Cystitis | 21 | 4 | |||||
| Pneumonitis | 93 | 4 | |||||
| 15 | 7 | Yes (18) | Sinusitis | 24 | 18a | ||
| 16 | 5 | Yes (55) | Cystitis | 15 | 2 | ||
| 25 | 58 | No | Acute chondrocalcinosis | 49 | 6 | ||
| 26 | 0 | No | Pulmonary tuberculosis | 30 | 0 | ||
| Bronchopulmonary | 20 | 0 | |||||
| aspergillosis | |||||||
| 29 | 22 | No | Treated pyelonephritis | 37 | 4 | ||
| 31 | 16 | No | Urological sepsis | 18 | 0 | ||
| 33 | 46 | No | Cystitis | 16 | 4 | ||
| 35 | 0 | No | Metastatic carcinoma | 23 | 0 | ||
| 36 | 8 | No | Cystitis | 15 | 7 | ||
| 41 | 5 | No | Pulmonary embolism | 16 | 2 | ||
| Acute bronchitis | 27 | 3 | |||||
| 42 | 45 | No | Diverticulitis | 22 | 8 | ||
| 46 | 31 | No | Cystitis | 17 | 3 | ||
| 47 | 0 | No | Gout attack | 51 | 4 | ||
| 49 | 2 | No | Pyelonephritis | 32 | 12 | ||
| 50 | 5 | No | Acute colonic | 168 | 0 | ||
| diverticulitis | |||||||
| 51 | 3 | No | Cystitis | 27 | 1 | ||
| 53 | 20 | No | Subacute bronchitis | 25 | 6 | ||
| 55 | 0 | No | Pulmonary embolism | 366 | 0 | ||
| 56 | 6 | No | Pyelonephritis | 108 | 5 | ||
| aCL | GCA‐ related | Non‐GCA‐related inflammatory flare | |||||
| Patient | level at diagnosis (U/l) | flare (aCL, U/l) | Origin of the flare | CRP (mg/l) | aCL (U/l) | ||
| 5 | 53 | Yes (20) | Metastatic carcinoma | 80 | 5 | ||
| 7 | 36 | Yes (60) | Cystitis | 30 | 4 | ||
| Pyelonephritis | 83 | 7 | |||||
| 12 | 28 | Yes (20) | Subacute bronchitis | 24 | 0 | ||
| 14 | 41 | Yes (6) | Pneumonitis | 58 | 5 | ||
| Cystitis | 21 | 4 | |||||
| Pneumonitis | 93 | 4 | |||||
| 15 | 7 | Yes (18) | Sinusitis | 24 | 18a | ||
| 16 | 5 | Yes (55) | Cystitis | 15 | 2 | ||
| 25 | 58 | No | Acute chondrocalcinosis | 49 | 6 | ||
| 26 | 0 | No | Pulmonary tuberculosis | 30 | 0 | ||
| Bronchopulmonary | 20 | 0 | |||||
| aspergillosis | |||||||
| 29 | 22 | No | Treated pyelonephritis | 37 | 4 | ||
| 31 | 16 | No | Urological sepsis | 18 | 0 | ||
| 33 | 46 | No | Cystitis | 16 | 4 | ||
| 35 | 0 | No | Metastatic carcinoma | 23 | 0 | ||
| 36 | 8 | No | Cystitis | 15 | 7 | ||
| 41 | 5 | No | Pulmonary embolism | 16 | 2 | ||
| Acute bronchitis | 27 | 3 | |||||
| 42 | 45 | No | Diverticulitis | 22 | 8 | ||
| 46 | 31 | No | Cystitis | 17 | 3 | ||
| 47 | 0 | No | Gout attack | 51 | 4 | ||
| 49 | 2 | No | Pyelonephritis | 32 | 12 | ||
| 50 | 5 | No | Acute colonic | 168 | 0 | ||
| diverticulitis | |||||||
| 51 | 3 | No | Cystitis | 27 | 1 | ||
| 53 | 20 | No | Subacute bronchitis | 25 | 6 | ||
| 55 | 0 | No | Pulmonary embolism | 366 | 0 | ||
| 56 | 6 | No | Pyelonephritis | 108 | 5 | ||
aThis patient experienced a relapse of GCA within a few weeks.
Discussion
This is the first large‐scale longitudinal study to determine clearly the relationship between the kinetics of aCL in serum and the occurrence of flares or relapses in biopsy‐proven GCA. We found aCL positivity before treatment in 46.6% of patients, a frequency close to that found previously by ourselves and others [4–6, 9–11]. In agreement with previous studies [4–6, 8–10], we found similar rates of OIC in aCL+ and aCL− patients. Furthermore, the fact that a patient was aCL+ or aCL− had no bearing on the long‐term outcome. It is not impossible, however, that the adverse effect of initial aCL positivity may have been overlooked since only 38% of our patients achieved complete recovery.
We confirmed, in most cases, that IgG aCL disappear from the serum within a few weeks or months after the introduction of corticosteroid treatment, provided the patient has become asymptomatic with a continuously low CRP level [4, 10–13]. However, aCL positivity persisted 6–14 months in four patients and indefinitely in a fifth; it was unrelated to CRP level but was seemingly due to a more intense immune response, as all five patients had an initially high level of aCL (average 83 U/l). Overall we found a positive correlation between the duration of aCL positivity and its initial level. Therefore, in cases with a very high level of aCL initially, corticosteroid treatment may not clear from the serum for many months. It thus appears important to distinguish, among patients with high levels of aCL initially, those with a slow but consistent decrease in aCL after the introduction of corticosteroid treatment and those with a continuously high aCL level, regardless of the achievement of control of inflammation. In the former group, a significant rise from a moderately high baseline value might imply a flare of GCA, whereas in the latter group a raised aCL level is unlikely to be caused by GCA and is therefore of no value in monitoring disease activity.
The absence of elevated aCL level with infection is not surprising in any case, considering that aCL are most commonly generated by bacterial, viral or parasitic diseases, which are involved only infrequently in patients with treated GCA, such as the acquired immunodeficiency syndrome, hepatitis, cytomegalovirus, parvovirus B19 and Epstein–Barr viruses, syphilis, malaria, borreliosis and rickettsiosis [20–25]. However, since our sample size was relatively small, we cannot rule out an association of GCA with the more frequent infectious diseases from which aCL may arise, such as tuberculosis [23, 26], Mycoplasma pneumoniae infection [27] and infective endocarditis [28].
The most important finding of our study is that the IgG aCL level correlates with the outcome of GCA in a substantial proportion of cases and has a high specificity for vasculitis in this elderly population. In fact, we noted that a rise in aCL preceded, accompanied or succeeded 74% of the documented flares and relapses of GCA, whereas this finding was never observed in patients in whom GCA was permanently controlled and those with inflammatory flares unrelated to GCA, especially flares related to infection. In fact, the only borderline increase in aCL level recorded in 22 episodes of infection concerned a patient whose GCA flared up soon afterwards (Table 3). In this case, each inflammatory episode was diagnosed accurately and treated. Nevertheless, since a flare of GCA may occasionally occur within a short period after infection, any isolated inflammatory episode not responding well to corticosteroid dose adjustment should be investigated thoroughly for a hidden infection, even if a concurrent rise in aCL is noted.
We also demonstrated that aCL positivity can occur in previously aCL− patients with relapse and is not invariably present in initially aCL+ patients with relapses. Finally, we postulate that the majority of subclinical flares and relapses of GCA should benefit from the monitoring of aCL levels, which might modify the approach taken in both work‐up and therapy. However, prospective studies are still needed to confirm these results definitively, since in our study clinicians were not blinded to the aCL, which might, in theory, have led to significant bias.
One important issue raised in this paper is whether our findings on GCA apply to other vasculitides. Although aCL have also been found in association with rheumatoid vasculitis, polyarteritis nodosa, relapsing polychondritis, Takayasu's arteritis, Buerger's disease, Kawasaki disease and Behçet syndrome, its value for monitoring these vasculitides has been only partially addressed in a few individual patients and small cohort studies, with inconsistent conclusions. From our results and some selected published studies [29–34], aCL may be non‐specific but sensitive markers of vascular damage in various systemic vasculitides [35], and it would be useful to carry out prospective longitudinal studies evaluating the value of aCL in monitoring such patients.
The origin of aCL in GCA is as yet unknown. They may be induced by the exposure of anionic phospholipids at the outer leaflet membrane after the rearrangement of membrane phospholipids due to endothelial cell stimulation during inflammation. Therefore, aCL, which do not seem to be pathogenic in GCA, could be merely secondary, induced by local inflammation of the arterial walls. This hypothesis, which is supported by the dramatic decrease in aCL level after corticosteroid administration [4, 5, 10–13], is reinforced by our finding of a strong relationship between the subsequent kinetics of aCL and signs of disease activity. The alternative hypothesis, of a causal pathogenic role of aCL in GCA, has never been argued convincingly [11, 36–41] and conflicts with the finding that aCL negativity is constant in many patients. The surprisingly high frequency of aCL positivity in relapsing patients (74% vs 46.6% before treatment) could be related to either a time‐dependent effect on the immune system or cumulative lesions of endothelial cells leading tardily to a stronger autoimmune response. The significance of a rise in serum aCL as the sole abnormality, preceding by weeks or months the recurrence of the inflammatory response, is uncertain. Although this was observed in only a few patients, it might imply that endothelial cells, from which aCL may arise, are the primary targets of an as yet unknown type of attack that precedes the overt local inflammatory response.
In conclusion, the monitoring of serum aCL can be very useful during the follow‐up of patients with GCA until recovery, allowing the detection of flares and relapses of GCA with fairly good sensitivity (74%) and a specificity of 100%. We suggest that aCL levels should be measured at baseline, at 3 months, at 6 months (if levels remains high), during the final period of treatment, when the risk of experiencing a flare is high, and, finally, each time a subclinical flare is suspected. The presence of elevated IgG aCL in a patient with unexplained sustained inflammatory response during the course of GCA strongly suggests a flare of GCA rather than an infectious complication, allowing the simplification of investigatory procedures and the prompt taking of appropriate therapeutic decisions. Whether this convenient algorithm also applies to isolated polymyalgia rheumatica and other systemic vasculitides deserves specific prospective studies.
Correspondence to: E. Liozon, Service de Médecine Interne A, Hôpital Universitaire Dupuytren, 2 av. Martin Luther‐King, Limoges, France.
Comments