Reduction of corticosteroids after the acute phase of giant cell arteritis (GCA) is a complex exercise. Corticosteroid reduction ideally should be rapid enough to minimize bone loss and other corticosteroid-related complications and yet not result in a relapse. Recurrent ischemic optic neuropathy (ION) is the most feared type of relapse. We conducted a two-center retrospective study of a patient population with ION in GCA to determine if any factors were predictive of recurrent ION.
We retrospectively reviewed 100 consecutive patients with biopsy-proven GCA treated at the Royal Victorian Eye and Ear Hospital (RVEEH) and St. Vincent's Hospital (SVH), Melbourne, Australia between 1988 and 1998. Of the original cohort, 67 patients met the following inclusion criteria: (1) treatment was initiated at the RVEEH or SVH (to have access to adequate medical records); (2) ION was present on presentation; (3) temporal artery biopsy result was positive; and (4) there was a minimum follow-up of 18 months. We hunted for evidence of recurrent ION over the duration of follow-up, defined as further visual loss and optic disc swelling in the originally affected eye or new ION in the fellow eye. We excluded one episode of recurrent ION that occurred within the first month after initial presentation with ION because this may have represented progression rather than true recurrence.
Seven patients had recurrent ION. Their characteristics were compared with those of the 60 patients without recurrent ION to determine if any factors were predictive of recurrent ION. These factors were: (1) age; (2) gender; (3) indicators of the severity of the initial episode of ION, including mode of corticosteroid delivery, initial visual acuity in the affected eye (worse eye if bilateral), bilateral ION, and initial erythrocyte sedimentation rate (ESR) level; and (4) rate of corticosteroid dose taper. We also examined the corticosteroid dose, the acute phase reactant (ESR/C-reactive protein [CRP]) level, and the presence or absence of presence of systemic symptoms consistent with giant cell arteritis at the time of recurrent ION.
Tests of significance were not performed because of the retrospective nature of the study and the low patient numbers.
The median patient age was 76 years, with a female-to-male preponderance of 2:1 (44 women, 23 men). At outset, 52 of 67 patients presented with unilateral visual loss and 15 with bilateral visual loss. Visual loss was caused by anterior ischemic optic neuropathy in all patients. In 40 of 67 patients, initial treatment was high-dose intravenous methylprednisolone. In the remaining 27 patients, oral prednisone was administered. Overall, the patients treated with intravenous methylprednisolone had more severe ION as determined by a worse median visual acuity at presentation.
In all seven patients with recurrent ION, the recurrence occurred in the originally affected eye. The vision did not improve in any of these patients even with an elevation of the corticosteroid dose. Five of these seven patients had an associated increase in ESR and/or CRP at the time of recurrent ION but in only one case (Patient 3) did the elevation precede the recurrent ION by enough time to act as a warning to caregivers. In this patient, in the 5 months before recurrent ION, the ESR had been between 7 and 9. The prednisone dose had been gradually reduced from 15 mg/d to 5 mg/d over the same period. One month before recurrence, the ESR had increased to 29 (from a level of 72 months earlier). The prednisone dose was kept at 5 mg/d. At the time of recurrence, the ESR was 54. In the remaining six patients with recurrent ION, the ESR level up to 1 month before the episode of recurrent ION had been normal on gradually declining doses of prednisone. Three of seven patients (Patients 2, 3, and 6) had new headache consistent with giant cell arteritis at the time of recurrence, but in only one patient (Patient 3) did the headache precede the recurrence (Table 1). Only one of seven patients with ION recurrence (Patient 7) had neither an elevation of acute phase reactants nor a relapse in systemic symptoms at the time of recurrent ION.
The longest latency for recurrent ION was 36 months after initial diagnosis (Patient 7). That patient was still being treated with prednisone 3 mg/d. The earliest recurrence occurred 3 months after presentation in two patients. The other four patients had recurrent ION at 4, 8, 12, and 20 months. Three patients with recurrent ION had been initially treated with intravenous methylprednisolone and four with high-dose oral prednisone. Only one patient (Patient 5) with ION recurrence underwent an additional biopsy (3 months after the initial attack). It was positive for giant cell arteritis.
Comparison of the recurrent ION and nonrecurrent ION groups (Table 2) did not show any major differences in gender or age. The recurrent ION group did not have a worse median visual acuity in the affected eye initially, a higher mean ESR at initial diagnosis, a higher incidence of bilateral involvement, a higher incidence of treatment with intravenous corticosteroids, or a faster reduction in corticosteroid dose than the nonrecurrent ION group. Mean doses in the recurrent ION group at 1, 3, 6, 12, and 18 months were 43 mg/d, 23 mg/d, 15 mg/d, 16 mg/d, and 12 mg/d, respectively; in the nonrecurrent ION group, mean doses were 43 mg/d, 22 mg/d, 16 mg/d, 11 mg/d, and 8 mg/d at those times (Fig. 1). Because the mean dose of the recurrent ION group may have been artificially raised by increased doses after recurrence, we also compared the mean dose of corticosteroids in the nonrecurrent ION group to corticosteroid doses in the recurrent ION group at the time of the recurrence. Of the seven patients who had recurrent ION, four were being treated with doses lower than the mean and three with doses higher than the mean (Fig. 2).
In our study of 67 patients with ION in GCA, we found recurrent ION in seven patients (10%). This is a higher rate than that found by Hayreh (1), who reported no recurrences in 145 patients, and by Liu et al. (2), who reported two (4%) cases among 45 patients. This relatively high prevalence rate presented us with a chance to study the characteristics of patients with recurrent ION.
It is disturbing that only one of our patients (Patient 3) showed warning signs in the form of an increase in ESR or new headache in the days or weeks before recurrence of ION. In the remaining six patients, an increase in ESR or presence of headache occurred only at the time of recurrent ION. One patient had no increase in acute phase reactants or a relapse of systemic symptoms even at the time of recurrent ION. It had been 36 months since the initial episode of ION. This last patient is like the one reported by Kim et al. (3) who had recurrent ION at 5 and 14 months after initial presentation without any rise in inflammatory markers or systemic symptoms.
Contrary to expectations, our patients with recurrent ION were not older, did not have a more severe attack of GCA as evidenced by higher mean initial ESR, a higher degree of visual acuity depression, a higher prevalence of bilateral ION, or a higher dose of initial corticosteroid treatment. Nor were they on a relatively low corticosteroid treatment dose at the time of recurrent ION. The average prednisone dose among our seven patients at time of recurrent ION was 16 mg/d.
The implications of our findings are that physicians caring for patients with ION in GCA cannot use any demographic or clinical features to predict whether ION will recur. They must taper the corticosteroid dose gradually and monitor acute phase reactants at close intervals, watching for an elevation in acute phase reactants or a relapse in systemic symptoms typical of GCA, which should signal the possibility of an imminent recurrence of ION and trigger an elevation in corticosteroid dose. Even with such close monitoring, our study demonstrates that the increase in acute phase reactants or occurrence of systemic symptoms may occur with insufficient lead time to allow the treating physician to act preemptively to prevent recurrent ION.