Prognostic factors of cytomegalovirus retinitis after hematopoietic stem cell transplantation

Purpose To identify the visual prognostic factors in patients with cytomegalovirus (CMV) retinitis after hematopoietic stem cell transplantation (HSCT). Methods This retrospective cohort study included 4241 patients who underwent HSCT from April 1, 2010 to March 31, 2019 at Seoul St. Mary's Hospital. Of them, 1063 patients presented CMV viremia, and 67 patients (93 eyes) were diagnosed with CMV retinitis. We enrolled 66 patients (91 eyes). The main outcomes included the initial best-corrected visual acuity (BCVA), BCVA at the diagnosis of retinitis and last visit, involved retinal zone, peak CMV DNA levels in the peripheral blood and aqueous humor, time between HSCT and the diagnosis of retinitis, time between the diagnosis of viremia and retinitis, complications, recurrence, survival, and so on. Results The mean BCVA (logarithm of the minimum angle of resolution) values before HSCT, at the time of retinitis diagnosis, and at the last visit were 0.041 ± 0.076, 0.262 ± 0.529, and 0.309 ± 0.547, respectively. Multiple regression analysis revealed that the involved zone (P = 0.001), time between HSCT and retinitis diagnosis (P = 0.019), and survival status (P = 0.001) were associated with the final visual acuity. Conclusions The final visual prognosis was worse in patients with greater invasion of the central retinal zone, those with a longer interval between HSCT and the diagnosis of retinitis, and those who died. Prompt diagnosis of CMV retinitis through periodic fundus examinations of patients with CMV viremia can prevent severe vision loss. Once CMV viremia is confirmed, we recommend fundus examinations to be immediately performed and repeated every 2 weeks for at least 2 months, even if the CMV DNA titer in the peripheral blood becomes negative.


Patients
This retrospective cohort study included patients diagnosed with CMV retinitis among those presenting with CMV viremia after HSCT. From April 1, 2010 to March 31, 2019, 4241 patients underwent HSCT at Seoul Saint Mary's Hospital. From these, 1063 patients presented CMV viremia, which manifested as CMV retinitis in 67 patients (93 eyes). All patients diagnosed with CMV retinitis were of Korean descent. One patient with bilateral vision loss due to infectious endophthalmitis before the diagnosis of CMV retinitis was excluded. Eventually, we enrolled 66 patients (91 eyes) and retrospectively analyzed their medical records and laboratory reports. Institutional Review Board/Ethics Committee approval was obtained (from Seoul Saint Mary's Hospital), and the study was conducted in accordance with the Declaration of Helsinki. The need for informed consent was waived due to the retrospective nature of the study.

Diagnosis
All patients scheduled for HSCT underwent best-corrected visual acuity (BCVA) and intraocular pressure measurements, slit-lamp microscopy, and fundus examination before HSCT. We measured CMV DNA levels in peripheral venous blood samples using quantitative polymerase chain reaction (PCR) before and once a week for 3 months after HSCT. Fundus examinations were performed as soon as patients were diagnosed with CMV viremia, and they were performed such that the patients' general condition was not affected. If CMV viremia was not detected, without ophthalmic symptoms, no additional examination was performed. Patients with deteriorated systemic conditions occasionally received bedside examinations; however, fundus examinations were delayed for patients with an extremely poor general condition. If CMV viremia persisted without the onset of the clinical manifestations of CMV retinitis, we repeated the fundus examinations after 2 weeks. Even if CMV was no longer detected in the peripheral blood, fundus examinations were repeated every 2 weeks for 2 months. After 2 months, the interval was gradually increased, and the examination was discontinued. Fundus examination was then performed only when CMV viremia appeared again or ophthalmic symptoms occurred.
A retinal specialist diagnosed CMV retinitis when a pale necrotic retina with focal areas of hemorrhage or variable, small dot-like lesions, or retinal vasculitis with perivascular sheathing starting from the peripheral retina and centrifugally progressing toward the posterior pole, were observed [1,21]. In addition, we performed CMV DNA PCR using aqueous humor samples obtained by anterior chamber paracentesis in all patients showing clinical manifestations of CMV retinitis. Samples were collected under aseptic conditions in the ophthalmology clinic. This test was also performed when recurrence or aggravation was suspected or the treatment response was evaluated.
We divided the retina into zones 1, 2, and 3 in order to distinguish the extent of invasion by CMV. Zone 1 represents an area of up to 3000 μm (2 disc diameters) from the fovea or 1500 μm from the edge of the optic nerve head. Zone 2 represents an area extending forward from zone 1 to the clinical equator of the eye, and zone 3 represents an area extending forward from zone 2 to the ora serrata [22].
All patients with CMV retinitis initially received an intravitreal injection of ganciclovir. In cases of ganciclovir resistance, we administered intravitreal foscarnet [25,26]. Patients with reduced CMV viremia who required further systemic treatment for CMV retinitis received the necessary treatment.

Analysis
The following data were collected from the medical records of the included patients: the patient's age and sex, involved eye, initial BCVA (before HSCT), BCVA at the diagnosis of CMV retinitis, BCVA at the last visit, ophthalmic symptoms at diagnosis, involved retinal zone, hematological disease, type of HSCT (HLA-matched vs. mismatched donor, related vs. unrelated donor), peak CMV DNA levels detected in the peripheral blood and aqueous humor, systemic use of ganciclovir/foscarnet/valganciclovir, intravitreal injection of ganciclovir/foscarnet, recurrence of retinitis, survival status, and complications such as rhegmatogenous retinal detachment (RRD) and immune recovery uveitis (IRU).
BCVA values were converted into a logarithm of the minimal angle of resolution (logMAR) units. Patients who could only count fingers or detect hand motion were assigned logMAR values of 2.0 and 3.0, respectively. Cases of light perception and no light perception were excluded from the calculation for the mean logMAR BCVA [27].
All data were statistically analyzed using the Statistical Package for the Social Sciences software, version 24.0 (SPSS Inc, Chicago, IL, USA). Multiple regression analysis and Pearson's correlation analysis were performed for prognostic factors affecting the final visual acuity. The survival rate according to the follow-up period was evaluated by Kaplan-Meier survival analysis. A log-rank test was used to compare the survival rates of the 2 groups divided according to final visual acuity. A P-value of <0.05 was considered statistically significant.

Results
In total, 91 eyes of 66 patients diagnosed with CMV retinitis were included. CMV DNA was detected in the aqueous humor of all patients diagnosed with CMV retinitis. All participants had CMV viremia either at the time retinitis was diagnosed, or at some time previously. The patients' demographic characteristics are described in Table 1.
Only 44 eyes (48.4%) exhibited ophthalmic symptoms such as decreased visual acuity, blurred vision, floaters, and, rarely, visual field defects. Thirteen eyes (14.3%) of 10 patients (15.2%) were diagnosed with CMV retinitis after negative peripheral blood CMV testing. The average duration between the detection of CMV negativity and the diagnosis of retinitis was 24.64 ± 16.04 (7-54) days.
Fifteen eyes (16.5%) of 11 patients (16.7%) exhibited recurrence of retinitis after the retinal lesion had disappeared and the aqueous humor tested negative for CMV. From these, 5 patients had unilateral retinitis with unilateral recurrence, 2 had bilateral retinitis with unilateral recurrence, and 4 had bilateral retinitis with bilateral recurrence. In these cases, treatment with ganciclovir was continued if the initial treatment had been effective.
Twenty patients (30.3%) died because of a worsening systemic condition within 1 year after the diagnosis of retinitis. Ten patients died while receiving treatment for retinitis, while 10 did not receive ophthalmic treatment because they exhibited CMV negativity in the peripheral blood and amelioration of retinitis. The time of period from the last visit to patients' death was between 3 days and 3 months. The remaining 46 patients were followed-up for more than 1 year after the resolution of CMV retinitis (Table 2). Treatment outcomes for patients with CMV retinitis are described in Table 3.
According to Pearson's correlation analysis, the involved zone (P = 0.008), the time between HSCT and the diagnosis of CMV retinitis (P = 0.004), and survival status (P = 0.002) were significant factors influencing the final visual acuity. These factors were also significant in multiple regression analyses (P = 0.001, 0.019, and 0.001, respectively). Greater invasion of the central retina and a longer interval between HSCT and the diagnosis of CMV retinitis were associated with a worse final visual acuity. Moreover, the final visual acuity was worse in patients who died than in those who survived. In addition, Kaplan-Meier survival analysis showed a difference in survival rate between the 2 groups divided according to final visual acuity; those with poor final visual acuity had a lower overall survival rate (P = 0.013, log-rank test) (Fig 1). Other clinical factors, including the peak CMV DNA levels in the peripheral blood and aqueous humor, time between the diagnosis of CMV viremia and the diagnosis of CMV retinitis, treatment duration for CMV retinitis, number of intravitreal antiviral injections, recurrence of CMV retinitis, donor HLA matching, and donor relation showed no significant correlations with the final visual acuity ( Table 4). The clinical features and treatment outcomes of patients with severe complications of CMV retinitis are described in Table 5.

Discussion
In  the incidences of CMV retinitis after HSCT were 1.92%, 0.17%, and 2.3%, respectively. Several studies have shown that CMV seropositivity in HSCT recipients is an important risk factor for CMV viremia [10,21]. Koreans exhibit a high incidence of CMV retinitis because the majority exhibit CMV seropositivity [3]. CMV seroprevalence is about 83% worldwide [28], but 94% in Korea [29]. In the study by Crippa et al. [18] and Yan et al. [19], the median time between the diagnosis of CMV retinitis and HSCT was 251 (range, 106-365) and 167 (range, 60-445) days, respectively; this was considerably higher than the median interval of 75 (range, 28-309; average, 94.96 ± 67.21) days in the present study. The patients in this study (with the exception of some whose fundus examinations were delayed due to deteriorated general condition), received immediate and 2-weekly examinations after CMV viremia was diagnosed. This could have shortened the time until diagnosis of CMV retinitis.
Despite adequate systemic treatment for CMV viremia, 10 patients were diagnosed with CMV retinitis after an average of 24.64 ± 16.04 (range, 7-56) days, after CMV DNA titer in the  peripheral blood became negative, probably because CMV takes a long time to infect the retina through BRB. Yanping et al. [30] suggested that murine CMV retinitis requires more than 2 weeks to develop, and Gao et al. [31] showed that murine intraocular CMV replication is independent of systemic CMV replication and occurs after a certain time interval. Although periodic retinal screening after HSCT is expensive, time-consuming, and uncomfortable to the patient, CMV retinitis can be diagnosed in many patients despite the absence of ophthalmic symptoms. The patients with a shorter interval between HSCT and diagnosis of CMV retinitis exhibited better visual prognosis; therefore, once CMV viremia is confirmed, we recommend an immediate fundus examination. While CMV viremia is persistent, it is repeated every 2 weeks even if there is no CMV retinitis. We also recommend repeated fundus examinations every 2 weeks for at least 2 months, even after CMV DNA titers in the peripheral blood become negative. The minimum period of up to 2 months is because of the 10 patients (13 eyes) that were diagnosed with CMV retinitis even after 24.64 ± 16.04 (7-56) days after CMV DNA became negative in the peripheral blood. Larochelle et al. [32] designed a CMV retinitis screening protocol for pediatric HSCT recipients. They suggested that all HSCT recipients that are confirmed with CMV viremia or any with CMV organ disease undergo a fundus examination within 2 weeks of viremia being diagnosed, and every 6-8 weeks until no viremia. However, based on our experience, it can be risky to discontinue fundus examinations once blood samples' CMV negativity is achieved. Further research is needed to determine the usefulness of the screening protocol.
In the present study, the patients who died, presented a poor final visual acuity, similar to findings in previous studies of patients with CMV retinitis and AIDS. In 2 previous studies conducted by Jabs et al., patients who experienced immune recovery exhibited a high survival rate, whereas those without immune recovery exhibited a low survival rate. In their first study [14], the rates of retinitis progression, second eye involvement, and decreased visual acuity were significantly higher in the no immune recovery group than in the immune recovery group. In the second study [13], however, visual impairment was independent of the immune status. According to the study of Holland et al. [32], in AIDS patients, CMV activity increases in the presence of immunodeficiency. In addition, in the study of Yan et al. [19], patients with CMV retinitis had significantly fewer median absolute counts of lymphocytes, CD3+ cells, and CD8+ T cells than those without CMV retinitis after haploidentical HSCT. The most common cause of death in HSCT recipients was a relapse, infection, and GVHD [33]. In the present study also; the most common causes of death in 20 patients were GVHD and infection. Four patients died of CMV infection in other organs, and 10 died while receiving treatment for CMV retinitis. Control of life-threatening infections, including CMV infection, may be difficult in patients with immunodeficiency; therefore, the final visual acuity of CMV retinitis is likely to be poor.
In the present study, the CMV DNA level in the aqueous humor was used for the diagnosis of CMV retinitis in addition to the characteristic clinical features. In the study by Smith et al., CMV DNA levels in the aqueous humor correlated with the activity and area of retinitis [34]. However, in this study, CMV DNA levels in the peripheral blood, as well as aqueous humor, were not related to the final visual acuity. CMV retinitis usually extends from the periphery to the center. Although the CMV DNA level in the aqueous humor reflects CMV activity and the area of retinitis at the time of measurement, the visual prognosis is determined by the extent of central involvement. If the lesion is confined to the periphery, there may be no correlation between the area affected by retinitis and the visual prognosis. In patients with CMV retinitis, elevated CMV DNA levels in the peripheral blood may indicate extraocular CMV infection or reactivation of CMV retinitis [35], while elevated CMV DNA levels in the aqueous humor may indicate CMV activity in the eye [34]. Therefore, CMV DNA levels can be used to determine recurrence even if it is not related to visual acuity.
This study has some limitations. First, unlike the CMV DNA level in the peripheral blood, which was measured weekly, the CMV DNA level in the aqueous humor was measured only at the time of diagnosis, aggravation, recurrence, and at the end of treatment; this was because aqueous humor sampling is an invasive procedure. Second, it was difficult to determine the precise timing for the onset of immunosuppression, considering HSCT recipients are already immunocompromised because of the underlying hematological disease and the chemotherapy administered before HSCT. Third, the study included many patients in poor general condition, and in some cases, there was no choice but to limit examination or treatment. Some patients had a delayed fundus examination due to worsening of overall condition, and some patients were moved to an intensive care unit during retinitis treatment, so the intravitreal injection was postponed. All patients could undergo visual acuity testing, even by finger counting or hand motion, when CMV retinitis was first diagnosed. However, visual acuity testing was difficult in patients transferred to the intensive care unit during treatment.
In conclusion, our findings suggest that the visual prognosis was worse in patients with greater invasion of the central retinal zone, those with a longer interval between HSCT and the diagnosis of CMV retinitis, and those who died. CMV retinitis can lead to various complications despite active and aggressive treatment, and it can result in severe vision loss. Therefore, once CMV viremia is confirmed, we recommend an immediate fundus examination. While CMV viremia is persistent, it is repeated every 2 weeks even if there is no CMV retinitis. In addition, repeated examinations are recommeded every 2 weeks for at least 2 months, even after CMV DNA titers in peripheral blood become negative.