Cytomegalovirus resistance to ganciclovir and clinical outcomes of patients with cytomegalovirus retinitis☆☆☆

Methods 

The Cytomegalovirus Retinitis and Viral Resistance Study is a prospective cohort study of patients with AIDS and cytomegalovirus retinitis.17 At the time of enrollment, before the initiation of anticytomegalovirus therapy, patients undergo an interview for medical and ophthalmic history, an eye examination, fundus photography, and cultures of the blood and urine for cytomegalovirus. Patients return for monthly follow-up examinations, at which time an eye examination and fundus photographs are repeated. Cultures of the blood and urine for cytomegalovirus are repeated at 1 and 3 months after enrollment and every 3 months thereafter. All positive cultures are submitted for susceptibility testing. Patients are treated according to best medical judgment. Patients treated with intravenous ganciclovir were given induction therapy at a dosage of 5 mg/kg twice daily for 2 weeks, followed by maintenance therapy at a dosage of 5 mg/kg once daily. Patients treated with oral ganciclovir were given intravenous ganciclovir induction, followed by maintenance therapy with oral ganciclovir at a dosage of 1 g three times daily. Patients with a ganciclovir implant typically also received oral ganciclovir at a dosage of 1 g three times daily.

For cytomegalovirus cultures and resistance testing, blood and urine specimens for cytomegalovirus culture were inoculated onto cell lines and processed, as previously described.18 All positive cultures were tested for susceptibility to ganciclovir, using the Hybriwix Probe System/CMV Susceptibility Test Kit (Diagnostic Hybrids, Athens, Ohio, USA), as previously described.10, 19 For ganciclovir, blood culture isolates were considered phenotypically resistant if the IC50 was >6.0 μmol/l, and urine culture isolates were considered phenotypically resistant if the IC50 was >8.0 μmol/l.10, 20, 21, 22, 23 The slightly higher cutoff was used for urine culture isolates due to previous data demonstrating that the phenotype–genotype correlations were best for a blood threshold of 6.0 μmol/l and a urine threshold of 8.0 μmol/l.23 Patients with negative cultures were assumed to harbor a sensitive virus. DNA was extracted from cytomegalovirus isolates for genotypic testing, as previously described.13, 23 The UL97 and UL54 genes were amplified using polymerase chain reaction and sequenced using the dRhodamine Terminater Ready Reaction kit (PE Applied Biosystems, Branchburg, New Jersey, USA) to detect resistance-conferring mutations.13, 23 The UL97 and UL54 sequences from the patient’s isolates were aligned by computer analysis with that of AD169 to determine the presence of mutations.13, 23 Because all patients with UL54 mutations also had UL97 mutations (that is, there were no patients with a UL54 mutation without a UL97 mutation),13, 23 for the purposes of this analysis, genotypic resistance was defined as the presence of a UL97 mutation known to confer ganciclovir resistance.

To determine clinical outcome visual acuity was measured at each visit using logarithmic visual acuity charts.17 With these charts, 85 letters read is equivalent to a visual acuity of 20/20, and a loss of 5 letters of acuity is equivalent to a loss of 1 line of visual acuity. The loss of 3 lines of visual acuity (15 letters) represents a doubling of the visual angle (for example, 20/20 to 20/40). Duplicate sets of fundus photographs were taken at the monthly visits. One set was kept by the clinicians to provide clinical care to the patients; the other was forwarded to the Fundus Photograph Reading Center for evaluation of retinitis progression and retinal area involved by cytomegalovirus. Graders at the Fundus Photograph Reading Center, masked as to the patient’s resistance status, evaluated photographs from each visit and compared them to previous photographs from the same patient for retinitis progression and for the amount of retinal area involved by cytomegalovirus retinitis.18, 24 The location of cytomegalovirus lesions in the eye was recorded using the standard “zone system,” in which zone 1 refers to an area within 3,000 μm of the center of the fovea or 1,500 μm of the edge of the optic nerve; zone 2 extends from this region to a circle through the ampulae of the vortex veins; and zone 3 is anterior to zone 2.14, 17 Retinitis progression was defined using the standard definition for clinical trials: movement of a border of a lesion ≥750 μm along a front ≥750 μm in size or the occurrence of a new lesion ≥0.25 disk area in size.14, 17 The amount of retinal area involved by cytomegalovirus retinitis was measured using grids superimposed on the fundus photographs, as previously described by the Studies of the Ocular Complications of AIDS Research Group.24

The first patient was enrolled in the Cytomegalovirus and Viral Resistance Study in December 1994. This study includes information on patients followed up through January 31, 2001, who received ganciclovir therapy at some time. Time-dependent analyses of the association of resistance with clinical outcomes were performed for several reasons: (1) retinitis progression occurs in nearly all patients treated with systemic anticytomegalovirus therapy9, 14 owing to the limited intraocular penetration of these drugs, even without the occurrence of resistance14, 25; (2) resistant cytomegalovirus typically develops only after 3 months of therapy10, 21; (3) resistance status may change during follow-up13; and (4) treatment regimens may change over time.

Patient follow-up was divided into 3-month blocks of time centered around the collection of blood and urine specimens for culture for cytomegalovirus and subsequent resistance testing. For intervals during which ganciclovir was administered, patients were classified as resistant or sensitive within an interval based on the susceptibility testing and genotype testing results. Four separate analyses were performed, each using a different definition of resistance: (1) phenotypic resistance detected in the blood or urine; (2) phenotypic resistance detected only in the blood; (3) genotypic resistance detected in either the blood or urine; and (4) genotypic resistance detected only in the blood. Because our previous data suggested that most resistant isolates were detected in the blood and that the urine cultures added limited detection of resistant isolates to those detected by the blood culture,10 resistant isolates detected in the urine only were not analyzed.

Anticytomegalovirus treatment was classified as the predominant regimen used in a time interval: intravenous ganciclovir, oral ganciclovir, ganciclovir implant only, ganciclovir implant plus systemic therapy, combination systemic ganciclovir with other systemic treatment (for example, foscarnet), and other therapy (for example, foscarnet alone, cidofovir). Highly active antiretroviral therapy was defined as combination antiretroviral therapy including either a protease inhibitor or a non-nucleoside reverse transcriptase inhibitor. Because immune reconstitution as a consequence of highly active antiretroviral therapy can result in arrest of the progression of cytomegalovirus retinitis without concomitant anticytomegalovirus therapy,26, 27, 28 and because the ganciclovir implant is associated with longer times to retinitis progression than systemic ganciclovir,29, 30 the analytic model included ganciclovir implant use and highly active antiretroviral therapy use, as well as resistance status.

Characteristics of patients who developed phenotypic resistance were compared with those of patients who did not develop resistance, using the Kruskal–Wallis test for continuous variables and the χ2 test or the Fisher exact test for categorical variables. The association of resistance with progression was analyzed using generalized estimating equations logistic regression, and the associations of resistance with change in visual acuity and area of retinitis were analyzed using generalized estimating equations linear regression of the ranks of the changes. Ranks were used in the latter analyses to protect against the influence of outliers with large changes. Time to the development of contralateral ocular disease among patients with and without resistance was analyzed with a time-dependent Cox proportional hazards model.31, 32

Results 

One hundred ninety-seven patients were enrolled in the study. Of these, 19 developed phenotypic resistance in the blood or urine, 18 genotypic resistance in the blood or urine, 19 phenotypic resistance in the blood, and 16 genotypic resistance in the blood. Fifteen patients developed both phenotypic and genotypic resistance to ganciclovir, 4 had phenotypic resistance only, and 3 had genotypic resistance only. All 18 patients classified as having genotypic resistance had UL97 gene mutations.13 Six patients had sequence changes in the UL54 gene, all of whom also had UL97 gene mutations.13 Resistance data on these patients, correlating phenotypic and genotypic resistance and giving the sequence changes over time, have been published previously.13, 23 Characteristics of the study population are shown as Table 1. Patients who never developed resistance and those who developed phenotypic resistance were similar at enrollment, except that patients who never developed resistance were more likely to be on highly active antiretroviral therapy at the time of diagnosis of the retinitis than were patients who subsequently developed resistance. However, there was no significant difference between the two groups in the percentage of patients who received highly active antiretroviral therapy at some time during follow-up. Although initial treatment was similar between the two groups, patients who developed resistance were more likely to receive foscarnet or cidofovir at some time during follow-up. Patients who developed resistance appeared to have longer follow-up than those who did not (median 15.1 months vs 8.6 months, P = .0128).

Table 1. Characteristics of the Study Population

	Characteristic	No Resistance	Phenotypic Resistance	Total	P Value
	Number of patients	178	19	197
	Age (years)
	Mean	38.4	37.2	38.3	.436
	Range	22–68	26–55	22–68
	Gender (%)
	Men	71.3	78.9	72.1	.861
	Women	28.7	21.1	27.9
	Risk factor for HIV infection (%)				.319
	MSM	49.7	66.7	51.3
	IDU	20.7	11.1	19.8
	Both MSM and IDU	1.2	5.6	1.6
	Heterosexual	24.0	16.7	24.1
	Other	3.5	0.0	3.2
	Median time since AIDS Dx (mo)	22.7	17.6	22.5	.339
	Highly active antiretroviral therapy (%)
	At Dx CMV retinitis	30.9	5.3	28.4	.019
	At any time	40.4	26.3	39.1	.230
	Anti-CMV regimen (%)
	Initial therapy (first 30 days)				.376
	IV ganciclovir	48.6	63.2	50.3
	Oral ganciclovir*	6.8	0.0	6.1
	Ganciclovir implant†	30.0	10.5	27.9
	Other‡	14.6	26.3	15.7
	Ever received
	IV ganciclovir	74.7	89.5	76.1	.255
	Oral ganciclovir	48.3	57.9	49.2	.427
	Ganciclovir implant	52.2	47.4	51.8	.686
	Foscarnet	19.1	42.1	21.3	.035
	Cidofovir	11.8	31.6	13.7	.029
	Bilateral CMV retinitis at Dx (%)	37.5	42.1	37.9	.694
	Zone 1 disease in either eye at Dx (%)	58.5	42.1	56.9	.170
	Retinal area involved by CMV ≥25% at Dx retinitis
	Either eye (%)	27.2	15.8	26.0	.411
	Worse eye (%)	13.0	7.4	12.5	.117
	Visual acuity at Dx CMV retinitis (median letters)
	Better eye	85.1	84.1	85.1	.710
	Worse eye	74.0	80.1	75.1	.299
	Follow-up (months)
	Median	8.6	15.1	9.2	.0128
	Range	0.7–76.7	2.7–52.5	0.7–76.7

CMV = cytomegalovirus; Dx = diagnosis; IDU = injection drug use; IV = intravenous; MSM = men having sex with men.

* Oral ganciclovir is the regimen of intravenous ganciclovir induction and oral ganciclovir maintenance therapy. † Any regimen containing a ganciclovir implant. ‡ Foscarnet, cidofovir, deferred therapy for first 30 days.

At baseline, 67.9% of patients had either a blood or urine culture that grew cytomegalovirus (blood 47.2%; urine 51.1%), and during follow-up, 45.1% of patients had at least one blood or urine culture positive for cytomegalovirus (blood 33.9%; urine 27.3%).

The median time to first progression of retinitis for the entire study group was 127 days, and the observed progression rate for the entire study group was 1.83 progressions/person-year. Relative odds for retinitis progression are shown in Table 2. The results of the analyses are shown both by patient, in which retinitis progression can occur in either eye, and by eye involved by cytomegalovirus retinitis, in which progression is defined as occurring only in that eye. For each analysis, results are shown using the four definitions of resistance outlined above. In the by-patient analysis, ganciclovir resistance was associated with a 4.17-fold to 5.61-fold increase in the odds of retinitis progression, depending upon the definition of ganciclovir resistance and the compartments (blood only vs blood and urine) sampled (P values all ≤ .0002). Highly active antiretroviral therapy consistently was associated with an approximate 50% reduction in the odds of retinitis progression, and the ganciclovir implant consistently was associated with an approximate 60% reduction in the odds of retinitis progression, both of which were significant associations. In the by-eye analysis, ganciclovir resistance was associated with a 2.05-fold to 2.60-fold increase in the odds of retinitis progression, depending upon the definition of resistance and the compartments sampled (P values < .0001 to .0327). Highly active antiretroviral therapy and the ganciclovir implant were associated with decreases in the odds of retinitis progression of a magnitude similar to those seen in the by-patient analysis. These data suggest that the detection of ganciclovir resistance in a patient on ganciclovir therapy was associated with an increased risk of retinitis progression.

Table 2. Relative Odds for Retinitis Progression

		Resistance	HAART	Ganciclovir Implant
	Retinitis progression in a patient
	Phenotypic resistance, blood or urine
	Relative odds	4.17	0.52	0.40
	P value	<.0001	.0078	.0002
	Phenotypic resistance, blood
	Relative odds	4.69	0.55	0.37
	P value	<.0001	.0158	<.0001
	Genotypic resistance, blood or urine
	Relative odds	5.61	0.52	0.40
	P value	<.0001	.0070	.0002
	Genotypic resistance, blood
	Relative odds	4.88	0.55	0.38
	P value	<.0001	.015	<.0001
	Retinitis progression in eyes involved by cytomegalovirus retinitis
	Phenotypic resistance, blood or urine
	Relative odds	2.22	0.53	0.37
	P value	.0111	.0151	<.0001
	Phenotypic resistance, blood
	Relative odds	2.37	0.55	0.34
	P value	.0069	.0224	<.0001
	Genotypic resistance, blood or urine
	Relative odds	2.60	0.53	0.37
	P value	.0079	.0151	<.0001
	Genotypic resistance, blood
	Relative odds	2.05	0.55	0.35
	P value	.0327	.0216	<.0001

HAART = highly active antiretroviral therapy.

The associations between resistant cytomegalovirus and other ocular outcomes are shown in Table 3. Results are expressed as the change by 3-month interval. For retinal area, the result represents the change in the percent of the total retinal area in zones 1 and 2 (the area accessible to conventional fundus photography) affected by retinitis lesions and indicates an increase in the size of the cytomegalovirus lesion. For visual acuity, the result represents the change in the number of letters read, and a negative number indicates a loss of visual acuity.

Table 3. Resistant Cytomegalovirus and Ocular Outcomes

	Resistance Definition/Outcome	Resistant	Susceptible	P Value
	Median change in retinal area involved by CMV retinitis (% retinal area by 3-month interval)
	Phenotypic, blood or urine	1.10	0.10	.021
	Phenotypic, blood	1.10	0.05	.016
	Genotypic, blood or urine	1.10	0.05	.012
	Genotypic, blood	1.10	0.05	.098
	25th percentile for change in visual acuity (worse eye; letters by 3-month interval)
	Phenotypic, blood or urine	−13.5	−5.0	.066
	Phenotypic, blood	−14.0	−5.0	.058
	Genotypic, blood or urine	−14.0	−5.0	.009
	Genotypic, blood	−14.0	−5.0	.096

CMV = cytomegalovirus.

Resistance, regardless of the definition, was associated with a greater increase in the amount of retinal area involved by cytomegalovirus during a 3-month interval (1.10% for patients with resistant virus vs 0.05% to 0.10% for patients with susceptible virus). For phenotypic definitions of resistance and genotypic resistance in the blood or urine the change was statistically significant (P = .012 to .021), whereas for genotypic resistance in the blood, the change was of borderline statistical significance (P = .098). The median change in visual acuity for 3-month interval, regardless of whether the patient harbored a resistant or susceptible virus during that interval, was zero letters. However, there was a difference in the distributions of loss of visual acuity between the two groups, which was of borderline statistical significance, in that those patients who did poorly tended to do more poorly if they harbored a resistant virus. Hence, the 25th percentile for change in visual acuity is shown in Table 3; there was a suggestion that patients who harbored resistant virus lost vision at a greater rate than did those who harbored a susceptible virus. Rates of visual loss at the 25th percentile for patients with resistant virus were 13.5 to 14 letters for 3-month interval, depending on the definition of resistance and the sites sampled, compared with a loss of 5 letters among those with susceptible virus (P = .009–.096).

Among patients with unilateral cytomegalovirus retinitis at enrollment, resistant cytomegalovirus (either phenotypic or genotypic, in either blood or urine) was associated with a 9.10-fold increase in the odds of the occurrence of contralateral retinitis (P = .0004); highly active antiretroviral therapy was associated with a relative odds for contralateral eye disease of 0.50 (P = .165); and the ganciclovir implant was associated with a 4.10-fold increase in the odds of development of contralateral eye disease (P = .0041).

Discussion 

Anticytomegalovirus agents suppress viral replication, but do not eliminate the virus. Hence, among patients with AIDS, unless there is immune reconstitution as a consequence of highly active antiretroviral therapy, chronic suppressive “maintenance” therapy is required.9 However, cytomegalovirus resistant to anticytomegalovirus agents will occur with prolonged therapy.10, 11 Previously, our group has reported on the epidemiology of resistant cytomegalovirus, the phenotype-genotype correlation for ganciclovir resistance among cytomegalovirus isolates from patients with cytomegalovirus retinitis, and the correlation of cytomegalovirus UL97 gene sequences between vitreous specimens and blood specimens.10, 23, 33 These results suggested that the identification of resistant virus in the blood or urine might correlate with adverse clinical outcomes of the retinitis.13 Furthermore, in these studies, patients who harbored resistant virus appeared to do poorly, as evidenced by the frequent occurrence of retinitis progression among these patients.13 However, because of the nearly universal occurrence of relapse of the retinitis, despite initial successful control, among patients treated with systemic anticytomegalovirus therapy,14 it was unclear how well resistance identified in the blood or urine would correlate with adverse outcomes of the retinitis. Therefore, we evaluated the ocular outcomes of cytomegalovirus retinitis, including retinitis progression, amount of retinal area involved by cytomegalovirus, and visual loss among patients with cytomegalovirus retinitis.

The strengths of our study included the prospective data collection and the masked evaluation of both retinitis progression and the amount of retinal area involved by cytomegalovirus. Because cytomegalovirus resistance results were not available until several weeks after the patient was seen, the evaluation of visual acuity also was performed by individuals typically unaware of the “resistance status” of the patient at the time of the patient visit. Furthermore, visual acuity measurements were performed using standardized logarithmic visual acuity charts and techniques.17 However, there are several caveats regarding the interpretation of these data. The proportion of patients who developed ganciclovir resistance was small. Therefore, the power to detect moderate relative differences may have been limited. This problem is apparent for the group of patients with genotypic resistance detected in the blood, which constituted the smallest of the four exposure groups, and presumably explains the borderline P values for the correlation of loss of retinal area with the detection of genotypic resistance in the blood. Retinal area and visual acuity results were of similar magnitude to those seen with other definitions of resistance, but the P values were borderline.

In addition, clinicians were allowed to change therapy based on their clinical observations. If a patient was relapsing frequently (multiple progressions), clinicians were likely to change therapy in an effort to control the retinitis and preserve visual acuity. Evidence for the success of this approach is that the median loss of visual acuity among all patients was zero letters during a 3-month interval. The attempt of clinicians to control the retinitis by switching to alternate therapies presumably explains the greater use of foscarnet and cidofovir among patients with resistant virus on follow-up, despite the use of similar treatment regimens as initial therapy between the two groups.

Nevertheless, our results show a substantial increase in the odds of retinitis progression among patients who harbor resistant virus in either the blood or urine. The odds of retinitis progression in involved eyes of patients with resistant virus identified in the blood or urine also was increased. The magnitude of the “by-eye” increase was less than that seen in the “by-patient” analysis, as the patient could have retinitis progression in either eye and, therefore, had a greater chance for retinitis progression being identified by the Fundus Photograph Reading Center. Because retinitis progression in either eye typically results in treatment modification, the by-patient analysis may be more relevant to subsequent therapeutic maneuvers. However, regardless of the definition of resistance used, the increase in the odds of retinitis progression was significant.

In addition to the increased risk of retinitis progression among patients with resistant cytomegalovirus, there was an increase in the amount of retinal area involved by cytomegalovirus, which was consistent regardless of the definition of resistance used, and statistically significant for all but the genotypic, blood only, definition of resistance (which was of borderline significance). The amount of retinal area involved by cytomegalovirus retinitis on fundus photography correlates well with the loss of visual field, which is an element of visual function.34 Furthermore, there was a suggestion that resistance was associated with a greater loss of visual acuity.

Our study also confirms previous work on the efficacy of the ganciclovir implant for controlling retinitis progression,29, 30 as the ganciclovir implant was associated with an approximate 60% reduction in the odds of retinitis progression. Several case series26, 27, 28 have reported that highly active antiretroviral therapy resulting in immune reconstitution can control cytomegalovirus retinitis without specific anticytomegalovirus therapy. In the current study, highly active antiretroviral therapy was associated with an approximate 50% reduction in the odds of retinitis progression. Other reported benefits of highly active antiretroviral therapy among patients with cytomegalovirus include a reduction in the incidence of retinal detachment of similar magnitude35 and an apparent reduction in the rate of retinitis progression among patients with cytomegalovirus retinitis without immune reconstitution.36

These data also confirm our previously reported ninefold increase in the rate of contralateral retinitis among those presenting with unilateral retinitis who develop resistance in the blood or urine, despite the fact that a more simplistic analysis was used previously.10 The ganciclovir implant was associated with a 4.10-fold increase in the risk of contralateral eye retinitis. Although patients typically were given concomitant oral ganciclovir to decrease the probability of the occurrence of contralateral ocular and visceral disease, approximately one third of the patients receiving the ganciclovir implant did not receive such therapy, for various reasons, including neutropenia, poor compliance, and others. Similar results have been reported from clinical trials of the ganciclovir implant, where patients treated with the implant alone have a higher rate of contralateral ocular or visceral disease.29, 30

In conclusion, ganciclovir resistance, detected in the blood or urine, is associated with an increased risk of adverse ocular outcomes. Because patients who harbor a resistant virus can be treated with alternative therapy, to which the virus is susceptible, detection of resistant cytomegalovirus may enable the treating clinician to alter therapy to a more appropriate drug. Although culture of the blood or urine for cytomegalovirus and testing of culture isolates for phenotypic resistance correlates well with clinical outcomes, such testing may take several weeks, thereby limiting its clinical utility. Efforts to rapidly identify patients who harbor resistant virus using technology such as the measurement of the cytomegalovirus viral load in the blood37 or direct sequencing of blood specimens for cytomegalovirus UL97 gene mutations, appear to be warranted, as such testing procedures have the potential for rapid (less than 48 hours) turnaround and increased clinical utility.