The standard of care for the treatment of cytomegalovirus retinitis is intravenous therapy with ganciclovir [2, 3] or foscarnet [4, 5]. Initial treatment with these drugs involves high-dose induction therapy for 2 to 4 weeks, followed by lower-dose maintenance therapy, which must be given indefinitely. These therapies are associated with many potential problems. First, the rate of reactivation and progression of retinitis is high; because the median time to progression is approximately 8 weeks, multiple inductions are required [6]. Reactivation presumably occurs because the virus develops some degree of drug resistance. Subsequent reactivations occur more rapidly than the initial one [6], and responses to later treatment are less favorable. Combination intravenous therapy with both foscarnet and ganciclovir may be effective for these cases of clinically resistant retinitis, but the treatment is time consuming for the patient and complex to administer [7, 8]. Second, ganciclovir and foscarnet are given systemically and over the long term and have serious potential side effects, most notably bone marrow suppression with ganciclovir [9] and renal toxicity with foscarnet [4, 5, 10]. Finally, the need for a permanent indwelling intravenous catheter is a major problem because of the deleterious effects of the catheter on the patient's quality of life and because of the potential medical complications, including sepsis [4].

All of these problems have led to the development of local (nonsystemic) therapies for cytomegalovirus retinitis. Serial intravitreous injections of ganciclovir have been shown to be effective, but they can be difficult to tolerate as sole therapy because they must be given at least once a week [11, 12]. Furthermore, the time to reactivation and progression with this method is 8 weeks, which is similar to the time with intravenous therapy [12]. Foscarnet has also been given in intravitreous injections, but weekly injections are required [13]. A sustained-release intravitreous ganciclovir implant has recently been shown to be effective as sole therapy for cytomegalovirus retinitis [14, 15]. The reported median time to progression was 226 days, compared with 15 days for untreated disease [14]. Because the patients in that study had recently received a diagnosis, it is not clear whether the implant will be as effective for reactivated retinitis. This treatment also requires surgical vitrectomy and may increase the rate of early retinal detachment. Moreover, the investigators [14] reported that five (of seven) retinal detachments occurred 30 to 64 days after vitrectomy. This period is shorter than the time to retinal detachment seen with intravenous therapy; with the ganciclovir implant, fewer than 10% of retinal detachments occur within the first 4 months [16].

Cidofovir is a relatively new nucleoside analogue that is highly active against cytomegalovirus [17-19]. In previous studies [20, 21], we showed that a single intravitreous injection of 20 µg of cidofovir is highly effective in healing cytomegalovirus retinitis in patients with AIDS. We did the current prospective study to further examine the efficacy and safety of multiple intravitreous injections of cidofovir as sole long-term maintenance therapy for cytomegalovirus retinitis in patients with AIDS. Because none of the eyes in the previous studies showed progression earlier than 35 days after a single injection, we decided that maintenance injections should be given every 5 to 6 weeks.

Methods

Top Methods Results Discussion Author & Article Info References

All patients were seen at the University of California, San Diego, AIDS Ocular Research Unit, which is a tertiary referral center for patients with retinal complications of AIDS. Twenty-two patients were enrolled in a prospective, nonrandomized, consecutive case series between April 1994 and January 1995. Thirty-two eyes were affected. Included patients had to have AIDS, be at least 18 years of age, have cytomegalovirus retinitis in at least one eye, and have signed an informed consent form approved by the institutional review board. Two groups of patients were analyzed. Group A consisted of patients with newly diagnosed cytomegalovirus retinitis; group B consisted of patients who had previously received intravenous systemic therapy for the disease and had been switched to intra vitreous cidofovir because of clinical resistance, intolerance, or preference. Therapy with all systemic or local anticytomegalovirus medications (other than intravitreous cidofovir) was discontinued before the beginning of the study period. Patients were allowed to receive systemic therapy if their primary care physician believed that extraocular cytomegalovirus infection was developing during the study. Those patients would then be censored at that point. Exclusion criteria were evidence of extraocular cytomegalovirus infection, the presence of silicone oil vitreous substitute in the affected eye, or a history of an unrelated inflammatory ocular condition (for example, uveitis).

All 22 patients were treated until they reached one of the following end points: death (5 patients [23%]), development of extraocular cytomegalovirus infection (2 patients [9%]), development of serious illness precluding adequate compliance with treatment (3 patients [14%]), detachment of the retina of the affected eye (1 patient [4%]), occurrence of substantial complications attributed to treatment (2 patients [9%]), poor compliance (1 patient [4%]), or the end of the study period (8 patients [36%]). No patients were lost to follow-up.

For the purposes of follow-up, the beginning of the study period was considered to be the day of the first cidofovir injection. Baseline and follow-up examinations included standardized Early Treatment for Diabetic Retinopathy Study (ETDRS) visual acuity, slit-lamp biomicroscopy, intraocular pressure measurement, indirect ophthalmoscopy of dilated pupils with detailed fundus drawing, and wide-angle fundus photography. Eyes were examined at 2-week intervals. Healing of retinitis was defined as resolution of retinal opacification, development of a scar, and cessation of progression. Reactivation and progression were determined clinically and by masked retrospective review of fundus photographs on the basis of the criteria described by the Studies of the Ocular Complications of AIDS Research Group [22]: 750 µm of border progression or the appearance of a new lesion at least 750 µm in diameter. Retinal photographs from all patients seen on a particular day were examined retrospectively by at least two investigators who had no knowledge of the patient's treatment protocol (for example, intravenous therapy, ganciclovir implantation, and intravitreous cidofovir). When the investigators' review of photographs disagreed with the clinical analysis of that day, the photographic reading was accepted.

Intravitreous injections were given every 5 to 6 weeks. Sterile cidofovir was synthesized as described elsewhere [19]. All injections were given systematically. With a lid speculum and topical anesthesia, povidone-iodine solution was used to clean the conjunctiva and lids. Lidocaine 2%, with epinephrine 1:1000, was injected subconjunctivally in the inferotemporal or supertemporal quadrant and was followed by a trans pars plana midvitreous injection of 20 µg of cidofovir in 0.1-mL normal saline given using a 30-gauge or a 28.5-gauge needle. Oral probenecid was given as prophylaxis to patients who had no known substantial allergy to sulfa medications (2 g given 3 hours before injection and 1 g given 2 hours and 8 hours after injection). Probenecid is known to competitively inhibit anion transport across epithelial barriers and has reduced renal toxicity in patients receiving intravenous cidofovir by decreasing renal tubular secretion of cidofovir ([23, 24], Gilead Sciences Investigators Brochure). We used probenecid in our patients in a similar way in an effort to decrease uptake of cidofovir into the ciliary body epithelium, where intraocular pressure may be adversely affected [20, 21]. In addition, prophylactic use of topical Polysporin Ointment (Glaxo Wellcome, Research Triangle Park, North Carolina) (twice daily for 3 days), topical 1% cyclopentolate (twice daily for 7 to 10 days), and topical 1% prednisolone acetate (four times daily for 7 to 10 days) was included in the care of all patients after injection.

The primary outcome measure was time to reactivation and progression of retinitis. Other outcome measures were visual acuity, response to initial treatment with intravitreous cidofovir, development of extraocular cytomegalovirus infection, involvement of the other eye in patients who began with unilateral retinitis, retinal detachment, survival, and ocular complications of treatment (including reactions or adverse events attributed to probenecid).

Statistical analysis included Kaplan-Meier estimates of the time to a particular outcome measure (such as progression, disease in the other eye, and death). We used chi-square analysis to calculate P values and calculated 95% CIs using the normal approximation.

Results

Top Methods Results Discussion Author & Article Info References

Twenty-one of the 22 patients (95%) were men, and the mean age was 37.7 years (range, 27 to 57 years). Group A consisted of 15 eyes (47%; 11 patients) in which intravitreous cidofovir was given as the initial therapy for active cytomegalovirus retinitis. Group B consisted of 17 eyes (53%; 11 patients); patients in this group had previously received intravenous therapy with ganciclovir or foscarnet or both. The disease in 8 of these 17 eyes (4 patients) was controlled and was not progressing at the time of initial injection; these patients switched therapies because of intolerance or preference. Nine of the 17 eyes (7 patients) had active progressive retinitis at the time of first injection.

A total of 108 injections were given, resulting in an average of 3.5 injections per eye and 4.9 injections per patient. The mean interval between injections was 5.6 weeks (median, 5.6 weeks). Each eye was examined a mean of 7.9 times (median, 7.5 times). The mean interval between examinations and fundus photography was 2.2 weeks (median, 2.0 weeks). Of eyes in which active cytomegalovirus retinitis was present at the time of initial cidofovir injection (n = 23), 100% (95% CI, 87% to 100%) healed in response to treatment (Figure 1). None of the eyes in group A had disease progression at any time during the study. In two of the eyes in group B (12%), one episode each of reactivation and progression developed. Thus, only 2 of the 32 eyes (6%; CI, 0% to 15%) or 2 of 254 visits (0.8%) showed any progression during the study period. The 2 eyes that showed progression were from the same patient, and each had one episode of progression at 5.6 weeks and 12.6 weeks, respectively. This patient was known to have retinitis that was resistant to both intravenous ganciclovir and intravenous foscarnet and that, despite treatment, had been progressing in both eyes at the time of study entry. Both eyes subsequently healed with no change in the protocol of intravitreous cidofovir injections every 5 to 6 weeks and with no further progression during the rest of the study, which included 21 weeks of follow-up for this patient. The mean follow-up of all patients was 15.3 weeks (median, 13.8 weeks; range, 5 to 44 weeks). The median time to disease progression could not be calculated using Kaplan-Meier analysis because so few episodes of reactivation and progression occurred. Among the 32 affected eyes, one retina (from a patient in group A) became detached during the study period (3%; CI, 0% to 9%).

View larger version (82K): [in this window] [in a new window]   Figure 1. Top. Fundus photograph of patient from group A showing active (newly diagnosed) cytomegalovirus retinitis. An old toxoplasmic retinitis scar (caused by currently quiescent disease) can be seen. Bottom. Fundus photograph of the same patient after three injections of intravitreal cidofovir (3 months later). Cytomegalovirus retinitis is completely healed and has not progressed. The toxoplasma scar is unchanged.

The mean initial ETDRS visual acuity for all eyes was 20/24 (median, 20/20); the mean final visual acuity was 20/28 (median, 20/31). These results were not significantly different. Only 3 of the 32 eyes (9%) lost visual acuity of two lines or more during the study period. The loss was caused by irreversible hypotonia in one eye, by transient hypotonia (with visual acuity improving at study end) in one eye, and by continued visual decline from macular retinitis despite rapid healing in one eye. The cases of hypotonia (one reversible and one irreversible) were the only treatment-related complications that resulted in discontinuation of the study protocol.

Iritis developed after 14 of the 101 injections (14%) that had been preceded by prophylaxis with oral probenecid. Of the 7 injections not preceded by prophylaxis, 4 (57%) led to iritis; this difference was statistically significant (P = 0.003). All cases of iritis responded quickly to topical steroids and cycloplegia (in higher doses than those given as prophylaxis) within 2 weeks. None of these episodes resulted in long-term adverse visual affects (that is, a reduction in visual acuity of two lines or more). Twenty of the 22 patients (91%) used probenecid for prophylaxis. Two patients (9%) did not use this agent because of a history of severe sulfa allergy. Only 1 of the 20 patients receiving probenecid discontinued this therapy at any time because of an adverse event (a mild, maculopapular reversible rash). Thus, 19 of 20 patients (95%) tolerated oral probenecid well throughout the study and reported receiving their dose for each injection.

No complications (such as cataract, endophthalmitis, and vitreous hemorrhage) attributable to the injection procedure itself developed after any of the 108 injections.

Because of possible extraocular cytomegalovirus, the maintenance cidofovir treatment protocol was discontinued in one patient in group A (9%) and one patient in group B (9%). These patients began receiving systemic therapy after discontinuation. Both of these patients had refractory diarrhea; one improved after receiving systemic intravenous therapy. The other did not improve, and the clinical diagnosis was retrospectively questioned. Neither patient had biopsy-proven extraocular disease. Of the seven patients in group A who had unilateral cytomegalovirus retinitis at study entry, three (43%) developed bilateral disease after a median interval of 18 weeks (measured from the date of diagnosis of cytomegalovirus retinitis). Only four patients in group B had unilateral disease at study entry; in three of the four, disease spread to the other eye after a median of 68 weeks.

By using Kaplan-Meier analysis, we determined that the median survival of patients who had never received any systemic therapy (group A) was 25 weeks.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study findings show the efficacy of a new drug, cidofovir, delivered intravitreously for the treatment and continued maintenance of cytomegalovirus retinitis in patients with AIDS. No patients in our study received any other form of anticytomegalovirus therapy (either local or systemic) at any time during the study period. Many patients had clinically resistant retinitis at study entry and stopped receiving other therapies (for example, intravenous ganciclovir or foscarnet). Our previous research [20, 21] showed the consistent and predictable healing of cytomegalovirus retinitis in response to intravitreous cidofovir. However, we did not determine whether the retinitis could be successfully maintained in a quiescent state for longer periods using serial injections; this maintenance is important because this disease requires indefinite treatment.

Only two episodes of reactivation and progression occurred among 32 eyes treated for as long as 44 weeks with serial intravitreous injections of cidofovir, 20 µg, given at 5- to 6-week intervals. Thirty of the 32 eyes (94%) had no progression throughout the study period. Disease did not progress in any eye in group A (which consisted of patients in whom intravitreous cidofovir was the first and only treatment of cytomegalovirus retinitis). Although we could not calculate the median time to progression because of the small percentage of eyes in which reactivation and progression occurred (6%), these results are superior to those obtained with conventional intravenous therapy [5, 6]. The interval between injections did not need to be shortened; thus, we believe that cytomegalovirus in these patients showed relatively little short-term resistance to cidofovir. The one patient in whom retinitis progressed during the study had a known history of clinically resistant retinitis that was unresponsive to ganciclovir and foscarnet. After the single episode of reactivation seen during our study, each of the patient's eyes subsequently healed, disease remained quiescent, and no loss of vision occurred (total duration of follow-up for this patient, 21 weeks). The low rate of retinal detachment that we observed may have been due to the minimal disruption of the vitreous caused by the small-bore needle (28.5- to 30-gauge) used to inject the drug as well as to the excellent control of the extent of retinitis and the degree of activity. These two measures are known risk factors for the development of retinal detachment in cytomegalovirus retinitis [25].

Although we have shown that intravitreous cidofovir had excellent efficacy in our patients, the drug also has some potential adverse effects. The iritis that was seen after 14% of injections caused transient and mild visual symptoms and photophobia. All cases of iritis resolved within 2 weeks with topical steroids and cycloplegia. One patient had a mild, clinically insignificant (asymptomatic) reduction in intraocular pressure that stabilized after subsequent injections. One patient, however, developed irreversible hypotonia that caused severe loss of vision. This adverse effect and another in which temporary visual reduction was caused by reversible hypotonia are disturbing and draw attention to cidofovir's potential for causing hypotonia. Systemic prophylactic probenecid appears to be useful in preventing iritis; it also appears to be well tolerated at the infrequent intervals used.

Our study was not designed to address mortality, extraocular cytomegalovirus infection, or involvement of the other eye when only local therapy is used. Indeed, local ocular therapy may ultimately be an adjuvant to a convenient form of systemic therapy if such a treatment (for example, oral ganciclovir) can be developed and shown to be effective. It is notable, however, that in our small, prospective, nonrandomized series, only 2 of 22 patients (9%) showed clinical evidence of extraocular cytomegalovirus infection. This percentage is lower than the previously reported 31% of patients in whom a tissue diagnosis of extra-ocular cytomegalovirus was established and in whom a ganciclovir implant was used as sole therapy [14]. Not surprisingly, in our study, involvement of the other eye was seen earlier in patients who received only local therapy (intravitreous cidofovir). None of these patients developed macular or optic nerve disease as the initial manifestation of fellow eye involvement, and none had subsequent reduction in visual acuity. We cannot directly compare the median 6-month survival rate in those of our patients who never received systemic anticytomegalovirus medication (group A) with that in those of our patients who received systemic therapy. Many of our patients were referred for intravitreal treatment because they were considered to be too sick to tolerate, comply with, or accept intravenous therapy. Furthermore, the frequently cited survival periods established by the Studies of the Ocular Complications of AIDS Research Group [4] are themselves based on studies of patients who have met certain standard Karnofsky scores; these patients may thus have been healthier. All of these issues can be appropriately addressed only in a larger, randomized, masked, controlled trial. Our pilot study shows the efficacy of this treatment protocol, but our results must be interpreted with caution because our sample size was small and we lacked a control group. Regardless of these limitations, the upper limit of the 95% CI for the percentage of eyes with progression (15%) is still impressive. We believe that our study does justify a larger, controlled clinical trial, but ethical considerations may limit the ability of investigators to do a double-masked trial because doing such a study would require a fairly invasive procedure (intravitreal injection) for delivery of placebo. Masking could be accomplished through the blinded review of retinal photographs, a strategy similar to that used in our study.

Presented in part at the Association for Research in Vision and Ophthalmology meeting held in Ft. Lauderdale, Florida, 14-19 May, 1995.

Drs. Arevalo, de la Paz, Munguia, and Freeman: University of California, San Diego, Shiley Eye Center, Gilman Drive, La Jolla, CA 92093-0946.

Dr. Azen: Division of Biometry, Department of Preventive Medicine, University of Southern California School of Medicine, 1540 Alcazar Street, Los Angeles, CA 90033.