Cytomegalovirus Retinitis in the Era of Highly Active Antiretroviral Therapy

A 37-year-old man was diagnosed as having acquired immunodeficiency syndrome (AIDS) in 1996. His CD4⁺ cell count was 0.005 × 10⁹/L, and routine ophthalmologic examination revealed active cytomegalovirus (CMV) retinitis in both eyes. He immediately began taking ritonavir, didanosine, and zidovudine. The patient's CMV retinitis became inactive while he was receiving intravenous ganciclovir sodium, but his therapy was switched to intravenous foscarnet sodium because of leukopenia. His human immunodeficiency virus (HIV) load had been negative since starting antiretroviral therapy except for 1 sample in 1998.

When the patient was examined at the National Eye Institute in March 1998, his visual acuity was 20/25 in the right eye and 20/40 in the left eye (Snellen equivalent). Slit-lamp biomicroscopy showed mild inflammation in the anterior segments and moderate vitreous inflammation (vitritis) of both eyes. Dilated retinal examination showed inactive CMV retinitis involving the midperipheral retina but sparing the optic nerve and macula. Slight edema of the left macula was also noted. The patient's CD4⁺ cell count was 0.15 × 10⁹/L, and a decision was made to discontinue foscarnet therapy. The CMV retinitis has remained inactive during the past year. The patient's most recent CD4⁺ cell count was 0.30 × 10⁹/L, and his HIV load was undetectable. However, the patient's visual acuity has decreased to 20/40 in the right eye and 20/63 in the left eye, with worsening of the vitritis and cystoid macular edema in both eyes (Figure 1). The patient declined treatment with periocular injections of triamcinolone acetonide or oral prednisone.

We have described 28 patients with CMV retinitis who have demonstrated a similar clinical course.^{1 - 4} Cytomegalovirus retinitis remained inactive without maintenance anti-CMV therapy in all patients as long as the CD4⁺ cell count was higher than 0.05 × 10⁹/L (Figure 2), with many patients followed up for at least 1 year (Table 1). Three of our patients had resolution of CMV retinitis without receiving specific anti-CMV therapy. To date, CMV retinitis has reactivated in only 1 of our patients but only after the CD4⁺ cell count decreased to 0.02 × 10⁹/L. Similar to the patient described, we have noted intraocular inflammation in many patients with CMV retinitis and immune recovery while receiving highly active antiretroviral therapy (HAART). We have documented immune recovery uveitis in 21 of 23 eyes with quiet CMV retinitis in patients receiving HAART.⁵ Importantly, visual acuity was significantly worse in eyes with moderate-to-severe uveitis.

These cases illustrate a number of striking changes in the presentation and course of CMV retinitis in patients with AIDS who are receiving HAART. First, retinitis did not progress when specific anti-CMV therapy was discontinued. Second, a number of patients developed substantial intraocular inflammation that has resulted in decreased visual acuity. These 2 findings highlight both the benefits and the potential adverse effects of a rejuvenated immune response.

Before the use of HAART, CMV retinitis was the most common intraocular infection in patients with AIDS, occurring in up to 40% of patients during their disease.^{6 - 9} Cytomegalovirus retinitis typically occurs when CD4⁺ cell counts fall below 0.10 × 10⁹/L.^{7 ,10} Without specific therapy, CMV retinitis predictably progresses in the immunocompromised patient¹¹ in a median time of 2 to 3 weeks.¹² However, specific anti-CMV therapy can slow the destructive course of CMV retinitis. Median time to progression of CMV retinitis is approximately 2 months in patients treated with intravenous ganciclovir or foscarnet,^{12 - 13} 2 to 4 months with intravenous cidofovir, and up to 7 months with the sustained-release intravitreal ganciclovir implant.^{14 - 15}

HAART, which generally includes a protease inhibitor and at least 2 nucleoside analogs, has led to decreased HIV replication, elevations in CD4⁺ cell counts,^{16 - 17} and reduced mortality.¹⁸ The study of CMV retinitis provides a unique opportunity to investigate the important clinical question of whether the rejuvenated immune system that results from HAART can effectively control opportunistic infections in patients with AIDS. First, the clinical course of CMV retinitis in both treated and untreated patients has been detailed in well-controlled randomized clinical trials. Second, since progression of CMV retinitis can be viewed clinically, photographed, and graded objectively by independent observers, precise assessment of this opportunistic infection is possible.

A number of studies during the last 3 years suggest that HAART-induced immune recovery has had a significant impact on CMV retinitis: (1) a decrease in incidence, (2) a change in the clinical course of the disease, and (3) an altered clinical presentation. Soon after the widespread use of HAART, the annual number of new cases of CMV retinitis decreased by more than 50%.¹⁹ In a study¹⁸ of 1255 patients, each of whom had at least one CD4⁺ cell count below 0.10 × 10⁹/L, the incidence of any of 3 major opportunistic infections, including CMV retinitis, declined from 21.9 per 100 person-years in 1994 to 3.7 per 100 person-years in mid-1997.

Despite the decreased occurrence of CMV retinitis in patients receiving HAART, it remained unclear whether the restored immune response could effectively control the disease in the absence of specific anti-CMV therapy. Some predicted that since patients with CMV retinitis are severely immunosuppressed, they might not be able to regenerate an effective immune response against the virus because of loss of clones that recognize and fight the CMV. Clinical reports suggest that immune recovery in patients receiving HAART is sufficient to control CMV retinitis.

A number of case series^{1 - 3 ,20 - 23} have suggested that CMV retinitis may not reactivate after maintenance anti-CMV therapy is stopped in patients with HAART-induced immune recovery and that active CMV retinitis may resolve in some patients who never receive specific anti-CMV treatment. To date, 65 patients with CMV retinitis who have experienced immune recovery while receiving HAART but not receiving specific anti-CMV therapy have been followed up for a median of 5 to 16.4 months (Table 1). Reported CD4⁺ cell counts at the first examination after stopping specific anti-CMV therapy range from 0.06 to 1.3 × 10⁹/L. Seven of these patients never received any anti-CMV medications. Only 1 of the 65 patients had a reactivation of the CMV retinitis, which occurred after his CD4⁺ cell count decreased below 0.05 × 10⁹/L despite continued HAART.³ This lack of CMV retinitis reactivation is especially striking because before HAART the median time to reactivation of retinitis was about 2 months despite systemic anti-CMV therapy.

All patients whose CMV retinitis remained inactive without maintenance anti-CMV therapy responded to HAART with elevated CD4⁺ cell counts, but Jacobson et al²⁴ described 5 patients diagnosed as having active CMV retinitis within 7 weeks of starting HAART. Patients may need to receive HAART for a number of months before the immune response against opportunistic infections is effectively restored.²⁵ Researchers have shown that the proportion of naive T cells increases after about 3 months of therapy.²⁶ Therefore, specific anti-CMV therapy should be continued for at least 3 months after CD4⁺ cell counts have increased while the patient was receiving HAART.

The CD4⁺ cell count appears to be a better predictor of an effective immune response against CMV than other laboratory parameters, including HIV load and CMV culture data. Many patients have successfully discontinued maintenance anti-CMV therapy without reactivation of retinitis as long as the CD4⁺ cell count is greater than 0.10 × 10⁹/L, regardless of HIV load or CMV cultures of blood or urine. Although investigators have reported that decreased HIV load over time often parallels a rise in CD4⁺ cell counts,^{2 ,22 - 23} reactivation of CMV retinitis did not occur in patients with CD4⁺ cell counts above 0.10 × 10⁹/L, even in patients with a 1.0-log rise in HIV load with titers as high as 139,000 copies/mL.² Currently, researchers are investigating whether CMV viral load²¹ or CMV-specific CD4⁺ cell responses quantified by flow cytometry²⁷ can quantify an effective clinical immune response against CMV.

Although HAART-associated elevation in CD4⁺ cell count is effective in controlling CMV retinitis in some patients, the rejuvenated immune response has in many cases led to intraocular inflammation. Only mild vitritis was reported in eyes with CMV retinitis in severely immunosuppressed patients.²⁸ More pronounced uveitis, manifested as inflammation involving both the anterior and posterior segments of the eye, is now occurring with CMV retinitis in patients receiving HAART. Manifestations include anterior uveitis, cataract, vitritis, cystoid macular edema, epiretinal membrane, and disc edema. This immune recovery uveitis is characterized by both a significant vitritis (Figure 1, A) and retinal edema (Figure 1, B).^{3 ,19 ,22 ,29 - 31} However, the reported occurrences of immune recovery uveitis vary among investigators,^{32 - 33} and reasons for this variability remain unclear.

Similarly, the exact mechanism of immune recovery uveitis has not been delineated. Since immune recovery uveitis does not occur in eyes without CMV retinitis, the ocular inflammation appears to be related to the CMV infection. We previously showed that CMV retinitis leads to a breakdown in the blood-ocular barrier. Before the use of HAART, this breakdown was manifested by increased anterior chamber protein without significant inflammatory cell infiltration.³⁴ However, breakdown in the blood-ocular barrier may allow CMV antigen to leak out of the eye and give the antigen access to lymphoid organs. Previous studies suggest that antigen must reach outside lymphoid tissue to stimulate an antigen-specific immune response.³⁵ Interestingly, in our patient series, the one eye with CMV retinitis and no uveitis had a small area of retinitis. Both little available CMV antigen in the eye and minimal breakdown in the blood-ocular barrier could limit the development of an immune response against CMV and explain the lack of uveitis. This may be influenced by the anti-CMV regimen; intravitreal sustained-release ganciclovir implants may better limit CMV replication than systemically administered anti-CMV medications. Finally, it is unclear whether immune recovery uveitis is related to subclinical viral replication or whether stopping treatment with anti-CMV medications exacerbates the ocular inflammation. However, immune recovery uveitis does develop in eyes with CMV retinitis despite treatment with anti-CMV medications.

Immune recovery following initiation of HAART has resulted in inflammatory disease associated with other opportunistic infections. Patients with subclinical Mycobacterium avium complex infection develop fever, leukocytosis, and lymphadenitis following the initiation of HAART.³⁶ Patients with latent cryptococcal central nervous system infection similarly developed a clinically apparent meningitis following HAART.³⁷ Although HAART-associated immune recovery has led to improved survival and lower rates of opportunistic infections, these data suggest that the rejuvenated immune response can induce clinically significant inflammatory disease. Although treatment with corticosteroids may help control the inflammation,³⁰ additional studies are needed to determine the long-term effects of this immune recovery uveitis on vision.

How should the management of CMV retinitis change in the era of HAART? Current data suggest that maintenance anti-CMV therapy can be safely discontinued in some patients with stable CMV retinitis. Although published studies lack control groups, a number of well-controlled randomized clinical trials from before the use of HAART demonstrated that CMV retinitis progresses within a median of 3 weeks without specific anti-CMV therapy^{7 ,14 - 15 ,38} and within 2 months in patients receiving systemic anti-CMV therapy.¹⁵ The lack of reactivation of retinitis or the development of extraocular CMV in any of the published series is most likely due to immune recovery leading to control of CMV infection. We currently consider stopping specific anti-CMV therapy in patients with quiet CMV retinitis if CD4⁺ cell counts are stable or increasing and have been greater than 0.10 × 10⁹/L for at least 3 months. Freedom from anti-CMV therapy and potential adverse effects has led to improved quality of life for many of our patients. However, it is important to examine these patients closely for reactivation of retinitis or development of extraocular disease. If CD4⁺ cell counts fall below 0.05 ×10⁹/L, the clinician should consider restarting specific anti-CMV therapy, especially if the patient has retinitis close to sight-sensitive structures such as the optic nerve or the macula.

Finally, previous National Institutes of Health guidelines³⁹ on the therapy of HIV infection recommended that use of prophylactic medications be continued even if CD4⁺ cell counts increase above threshold levels. Because of the clonal nature of the antigen-specific immune response, it is possible that the increased T-cell numbers following HAART may not restore adequate protection against opportunistic pathogens like CMV. Studies provide clinical evidence that immune recovery in patients receiving HAART is effective in controlling opportunistic infections, even in patients with a history of severe immunosuppression. Therefore, more recent guidelines address discontinuation of maintenance anti-CMV therapy in patients receiving HAART.⁴⁰ However, this restored immune response may also have adverse effects on the eye. HAART-associated immune recovery uveitis can occur in eyes with CMV retinitis and may diminish vision in some patients. Further studies are needed to fully characterize the effects of the restored immune response on the eye and to develop the best therapeutic approach for the patient.