Treatment of CMV retinitis with intravitreal ganciclovir in the HAART era

Department of Ophthalmology, University of Ottawa Eye Institute; Departments of Medicine and Infectious Diseases, The Ottawa Hospital, Ottawa, Ontario Correspondence: Dr WG Hodge, Department of Ophthalmology, University of Ottawa Eye Institute, 501 Smyth Road, Ottawa, Ontario K1H 8L6. Telephone 613-737-8218, fax 613-737-8836, e-mail rcamp035@uottawa.ca Received for publication August 9, 2000. Accepted January 29, 2001 RJ Campbell, B Chow, G Victor, S Kravcik, WG Hodge. Treatment of CMV retinitis with intravitreal ganciclovir in the HAART era. Can J Infect Dis 2001;12(5):300-304.

Before HAART, treatment of CMV retinitis consisted of lifelong treatment with specific anti-CMV compounds, with relapse of retinitis occurring within two to three weeks after discontinuation of maintenance therapy (6).However, in the HAART era, numerous reports have demonstrated that the improved immune function induced by HAART makes possible the cessation of all specific anti-CMV therapy without reactivation of quiescent CMV retinitis (7)(8)(9)(10).Moreover, regression of active CMV retinitis has been reported in patients receiving HAART without specific anti-CMV therapy (11).
Despite such reports, standard initial treatment of active CMV retinitis still involves anti-CMV chemotherapeutics.Three antiviral drugs -ganciclovir, foscarnet and cidofovirare currently in use (12)(13)(14).Ganciclovir, a guanosine analogue, was the first of these proven to be effective in halting the progression of CMV retinitis (15).Initial therapy with this drug generally involves intravenous administration; however, maintenance therapy has been effective via intravenous, oral and intravitreal routes.
Intravitreal drug delivery can be accomplished via an intravitreal implant device or through repeated intravitreal injection (16,17).In Canada, the injection option has been widely employed, because intravitreal implants are not funded by provincial health insurance programs.Therefore, the primary objective of the present study was to describe the course and outcome of CMV retinitis among patients treated with intravitreal ganciclovir injections and HAART.A secondary objective was to evaluate the impact of HAART on such outcomes as CMV retinitis progression, ocular complications of intravitreal treatment and mortality among CMV retinitis patients receiving intravitreal ganciclovir.

PATIENTS AND METHODS
Inclusion and exclusion criteria: Medical records were reviewed from all AIDS patients diagnosed with CMV retinitis at the University of Ottawa Eye Institute (UOEI), Ottawa, Ontario between January 1996 and August 1999.The UOEI is the only site providing tertiary ophthalmology service for this patient group in the greater Ottawa region, and thus, all patients diagnosed with CMV retinitis in this region were evaluated at the UOEI.Throughout this time, HAART was available to HIV patients at the Ottawa Hospital, Ottawa, Ontario, which provides tertiary care for the same population covered by the UOEI.Patients younger than 18 years of age at the time of diagnosis and those followed for less than two weeks were excluded from the present study.Hence, the study included 21 eyes from 16 patients.Diagnosis and covariates measured: All ophthalmological data were obtained from detailed ophthalmology clinic notes and drawings, with the additional support of fundus photographs.The diagnosis of CMV retinitis was based on well established clinical criteria (18).The baseline threat to central vision was assessed by determining the invasion of retinitis into three distinct retinal zones as previously delineated (19).Other covariates measured were age at diagnosis, sex, CD4+ T cell count immediately preceding diagnosis of CMV retinitis, CD4+ T cell count at time of cessation of intravitreal ganciclovir, the number of AIDS-related opportunistic infections or malignancies suffered before the onset of CMV retinitis and final visual acuity.Follow-up and outcomes measured: All patients underwent fundus examination at one-month intervals during intravitreal ganciclovir therapy.If ganciclovir was discontinued after HAART-induced immune recovery, the frequency of fundus evaluation was slowly decreased, with a maximum time between examinations of six months.Subjects were followed until August 1999.The primary outcomes were retinitis progression, ocular complications and death.Subjects were censored at the time of discontinuing intravitreal injections, except in the case of survival measurements, in which all patients were included until the end of the study period.
Disease progression was defined using previously established criteria -specifically, new regions of opacity or hemorrhage, or advancement of the leading edge of necrosis by more than 750 µm (19).Ocular complications consisted of retinal detachment and endophthalmitis.Date of death was established using hospital patient files and records from the infectious diseases clinic at the Ottawa Hospital.Using these resources, all subjects were accounted for at the end of the study period.Because the impact of intravitreal ganciclovir is local, each eye was evaluated independently for the ophthal-

RESULTS
In the present study, 12 patients were diagnosed with CMV retinitis in 1996, three in 1997, one in 1998 and none in the first seven months of 1999.Among patients receiving HAART, five of eight subjects discontinued all specific anti-CMV therapy and returned for regular follow-up.These patients underwent intravitreal ganciclovir treatment for a mean period of 428 days, during which time, mean CD4+ T cell count rose from 7.5 to 190/µL.Subsequently, after cessation of intravitreal ganciclovir, none of these patients experienced CMV retinitis recurrence during a mean follow-up period of 617 days (range 408 to 820 days).Among the other three individuals in the HAART group, one developed bacterial endophthalmitis and, after resolution, failed to return for ongoing ophthalmic assessment.This complication was felt to be secondary to the intravitreal injections.The other two patients did not experience significant recovery in CD4+ T cell count during the study period.One of these patients was eventually switched to intravitreal foscarnet because of ongoing CMV retinitis progression, which was felt to be secondary to ganciclovir resistance.The other began HAART 298 days after the development of CMV retinitis and then only tolerated HAART for 121 days.This subject continued to receive intravitreal ganciclovir for a further 158 days, during which time no CMV retinitis progression occurred.The patient subsequently entered a palliative care phase and died one month later.
Overall, three patients who originally presented with uniocular disease subsequently developed CMV retinitis in the second eye with a mean time of 175 days between diagnoses.Mean CD4+ count at the time of diagnosis of retinitis in the second eye was 21/µL.Two of these patients were not on HAART, while the third had started HAART less than four months before the diagnosis of CMV retinitis in the second eye and had not experienced a measured rise in CD4+ T cell count.
The secondary objective of this study was to compare the course of CMV retinitis in HAART patients and NOHAART patients.CD4+ counts were not different at diagnosis of CMV retinitis, with means of 22.6/µL and 16.5/µL for the HAART and NOHAART groups, respectively (P=0.33).However, there was a wide gap in CD4+ counts at the end of intravitreal ganciclovir therapy, with means of 115/µL and 43.3/µL for the HAART and NOHAART groups, respectively (P=0.062).The proportions of initially affected retinal zones were significantly different, with the HAART group displaying more sight-threatening central lesions (P=0.023;χ 2 test).However, final visual acuities were not different (P=0.66).While all patients in both groups displayed initial stabilization of CMV retinitis with intravenous ganciclovir induction therapy, subsequent progression during intravitreal ganciclovir therapy was 11.4 times more likely among those not receiving HAART (P=0.049).Median time to progression was 98 days for the NOHAART group, while the median progression event in the HAART group did not occur (Figure 1).

) Cytomegalovirus (CMV) retinitis progression during intravitreal ganciclovir therapy; comparison of group receiving highly active antiretroviral therapy ( ) and group not receiving highly active antiretroviral therapy ( )
therapy were not different in the HAART and NOHAART groups (P=0.92; Figure 2).Finally, patients not receiving HAART were 4.9 times more likely to die than those receiving HAART; however, this result was not highly significant (P=0.22)(Figure 3).

DISCUSSION
During recent years, HAART has greatly altered the morbidity and mortality associated with AIDS (2,21).Indeed, our data show the marked decrease in the incidence of CMV retinitis among AIDS patients during the HAART era.However, HAART not only helps prevent CMV retinitis, but also influences treatment of this retinitis.Numerous reports have demonstrated that after HAART-mediated immune system reconstitution, specific anti-CMV therapy may be discontinued without reactivation of CMV retinitis during follow-up periods of three to 18 months (7)(8)(9)(10)(11).Our data show that the potential duration of disease quiescence after discontinuation of ganciclovir can be as long as 820 days.These results are contributing to the evolution of CMV retinitis therapy.For instance, the frequency of follow-up evaluations is gradually being decreased, and in some patients with stable CD4+ T cell counts, yearly ophthalmic examinations may be appropriate.
One of the risks of local therapy for CMV retinitis is the lack of prophylaxis against second eye involvement.It is interesting to note that second eye involvement occurred in only one of eight patients receiving HAART.In that patient, second eye involvement occurred soon after initiation of HAART and before a significant improvement in CD4+ count, which was only 39/µL at the time of diagnosis in the second eye.This is consistent with an earlier report that suggested that for patients responding to HAART, systemic ganciclovir may not provide additional benefit in the prevention of second eye disease (22).
As anticipated, during both the early and later phases of CMV retinitis, those receiving HAART were much less likely to suffer progression of this disease.This was seen despite the fact that two patients included in the HAART group did not begin HAART until months after diagnosis of CMV retinitis.This leads to an underestimation of the impact of HAART in preventing CMV retinitis progression in the early phases of retinitis.This correlates closely with previous observations in patients with recurrent CMV retinitis treated with ganciclovir implants (23).A recent report showing the elimination of CMV viremia and prevention of retinitis in patients with documented CMV viremia and low initial CD4+ T-cell counts further supports the concept of HAART possessing the capability of halting CMV disease (24).
While a trend was observed indicating a decreased mortality rate among those receiving HAART, our results did not reach a high level of significance.This was likely a consequence of relatively small cohort sizes combined with the 'intent-to-treat' analysis, which resulted in the only death in the HAART group -in a patient who had tolerated HAART for only 121 days.
A potential bias in this study arose from the process of deciding whether a patient would initiate HAART.This clinical decision involved many patient and physician factors, with the ensuing possibility of systematic differences existing between the HAART and NOHAART groups.Moreover, while an effort has been made to adjust statistically for numerous potential confounding variables, the existence of unrecognized confounders cannot be ruled out.
In the HAART era, the incidence of CMV retinitis has declined significantly.However, new or reactivated cases continue to occur among patients who are HAART-naive or for whom HAART has failed.With the potential development of resistance to HAART, we may witness a rise in CMV retinitis incidence, necessitating temporary or long-lasting reintroduction of intravitreal ganciclovir to control CMV disease.This raises the possibility that intravitreal injections may play an expanding role as a local treatment option for active CMV retinitis.
mological outcomes.However, each subject was counted only once in survival calculations.Treatment: Immediately after the diagnosis of CMV retinitis, all patients received induction therapy consisting of 14 days of daily intravenous ganciclovir infusions.After this period, patients received maintenance therapy in the form of weekly intravitreal injections of ganciclovir (2 mg in 0.05 mL balanced saline[20]).HAART consisted of the combination of at least one protease inhibitor with at least two reverse transcriptase inhibitors.Patients in the HAART group typically began HAART within 30 days of the diagnosis of CMV retinitis.However, two patients in this group began HAART much later, commencing at 298 and 313 days after the diagnosis.Before the diagnosis of CMV retinitis, some patients in both the group receiving HAART and the group not receiving HAART (NOHAART) groups had received antiretroviral therapy in the form of mono or dual reverse transcriptase inhibitors (none had received protease inhibitors), while others were not receiving any anti-HIV medications at the time of retinitis diagnosis.Statistical analysis: Using Student's unpaired t test, covariates such as age, CD4+ T cell count at the time of CMV retinitis diagnosis, CD4+ T cell count at the time of cessation of intravitreal ganciclovir, number of AIDS-related illnesses before CMV retinitis and final visual acuity were compared between the HAART group and the NOHAART group.Affected retinal zone proportions were compared using a χ 2 test.Finally, progression, ocular complication and mortality rates were compared using a censored Cox proportional hazards analysis.The instantaneous hazard for each outcome was calculated while controlling for age, CD4+ T cell count at the time of diagnosis of CMV retinitis and HIV disease severity, as estimated by the number of AIDS-related illnesses suffered before the onset of CMV retinitis.

Figure 1
Figure 1) Cytomegalovirus (CMV) retinitis progression during intravitreal ganciclovir therapy; comparison of group receiving highly active antiretroviral therapy ( ) and group not receiving highly active antiretroviral therapy ( ) CMV retinitis in the HAART eraCan J Infect Dis Vol 12 No 5 September/October 2001 303

Figure 2 )Figure 3 )
Figure 2) Ocular complications during intravitreal ganciclovir therapy for cytomegalovirus (CMV) retinitis; comparison of group receiving highly active antiretroviral therapy ( ) and group not receiving highly active antiretroviral therapy ( )