Long‐term outcomes of combined intravitreal methotrexate and systemic high‐dose methotrexate therapy in vitreoretinal lymphoma

The optimal treatment for vitreoretinal lymphoma (VRL) remains a challenge, as central nervous system (CNS) relapse occurs frequently, leading to the worst impact on survival. We previously proposed combined intravitreal methotrexate and systemic high‐dose methotrexate therapy for this disease. This study aimed to report the long‐term outcomes of patients with VRL using this combination treatment.


| INTRODUCTION
Vitreoretinal lymphoma (VRL) is the most common intraocular lymphoma worldwide. 1 At initial diagnosis, VRL may be restricted to the intraocular compartment without brain or cerebrospinal fluid (CSF) involvement, called primary VRL (PVRL), or have concurrent central nervous system (CNS) diseases. Like primary CNS lymphoma, the majority of VRL cases are high-grade non-Hodgkin lymphoma, belonging to the activated B-cell subtype of diffuse large B-cell lymphoma. 2 However, few cases of T cell or NK cell origin had been reported. [3][4][5] The median age at VRL diagnosis is in the fifth-sixth decades of life. 6,7 It seems that VRL has no racial predilection. 1 Additionally, growing evidence shows that women are more commonly affected than men. 5,[8][9][10] Patients with VRL commonly have bilateral ocular involvements. 1,11,12 Blurred vision and floaters are the most common ophthalmologic symptoms. Notably, these nonspecific clinical manifestations mimic posterior uveitis, letting VRL been termed as a masquerade syndrome. 7 The golden standard for the diagnosis of VRL is the cytologic identification of lymphoma cells in the vitrectomy specimens. 13 Flow cytometry, measurements of interleukin levels, 14 detection of immunoglobulin heavy chain or T-cell receptor gene rearrangements, 15,16 more recently novel molecular biomarker tests of vitreous samples, [17][18][19] as well as the multimodal retinal imaging in the ophthalmic field 6,20 all contribute to the improvement of diagnostic yield.
VRL usually has a good response to initial treatment. Nevertheless, subsequent local or CNS relapse frequently occurs. It has been reported that over half of patients with VRL will eventually develop CNS relapse or progression, which accounts for the most frequent cause of death. 7,21 Nowadays, several therapeutic strategies are available for VRL: local intraocular therapy including ocular radiotherapy or intravitreal (IVT) injection of methotrexate or rituximab; systemic high-dose methotrexate (HDMTX)based chemotherapy with or without intrathecal (IT) therapy; or a combination of both. However, due to the rarity of the disease, the best frontline therapeutic strategy for VRL remains a matter of debate. 13 Despite no standard therapy for VRL, the goal of treatment is to achieve complete response (CR) of intraocular disease for restoring the visual acuity and reduce the risk of subsequent CNS or systemic relapse. In 2003, we had adopted the therapy combining IVT methotrexate and systemic HDMTX with or without IT therapy in the management of all patients with newly diagnosed VRL at the National Taiwan University Hospital (NTUH). 22 IVT methotrexate is used to quickly reach effective tumoricidal level to eliminate the intraocular disease, while systemic HDMTX with or without IT therapy is intended to eradicate the concurrent CNS disease or potentially microscopic CNS seeding to prevent the subsequent relapses. In this study, we expanded the patient number from 19 in the previous report to 32 and presented an update on the long-term follow-up outcomes of patients with VRL treated with this combination approach.

| Patient selection and clinical data collection
From April 2003 to December 2018, patients with newly diagnosed VRL without a concurrent human immunodeficiency virus infection at NTUH were considered for enrolment. For the diagnosis of VRL, all enrolled patients underwent diagnostic vitrectomy to obtain adequate vitreous samples for cytologic and flow cytometric analyses. VRL was diagnosed when large, abnormal lymphoid cells with high nuclear/cytoplasm ratio, pleomorphic nuclei, and basophilic cytoplasm were identified by a cytologic examination of the vitreous samples. The immunophenotype and monoclonality of the abnormal cells were determined by flow cytometry. Additionally, all patients underwent CSF analysis, brain magnetic resonance imaging (MRI), bone marrow biopsy, and whole-body computed tomography (CT) with contrast or positron emission tomography/CT to clarify the disease extent. VRL with disease limiting to the intraocular compartment was defined as PVRL, whereas VRL with concomitant brain or CSF involvement was defined as concurrent VRL.
We performed a retrospective chart review to collect data on clinical features, treatment responses, relapse events, and outcomes. Treatment-related toxicities were evaluated in accordance with the National Cancer Institute Common Terminology Criteria for Adverse Events Version 5.

| Combination therapy
All enrolled patients had received combined IVT methotrexate and systemic HDMTX therapy as the frontline treatment. IVT injection of methotrexate was performed at a dose of 400 μg. For the dosage of HDMTX, it was administered at a target dose of 8 g/m 2 over 6 h per cycle. Dose reduction of methotrexate to 6 g/m 2 was acceptable, especially for older patients. The detailed treatment protocol of the combination approach has been described in our previous study. 22 Patients with concomitant CSF involvement also received concurrent IT chemotherapy consisting of methotrexate, cytarabine, and hydrocortisone.

| Treatment response evaluation
The assessment of therapeutic response was scheduled to perform after 1 month of treatment and was followed up at intervals of 3 months thereafter. CR is defined as the disappearance of lymphomatous infiltrates within the eye or CNS as determined by ophthalmologic evaluation, brain MRI scan, and CSF examination. Partial response (PR) is defined as more than 50% reduction of lymphomatous infiltrates. Progressive disease (PD) is defined as deterioration of previous ocular or CNS lesions, and stable disease is defined as less than a PR but is not PD. Relapsed disease means reappearance of any new lesion after achieving CR.

| Statistical analysis
A chi-square test or Fisher's exact test was used to compare categorical data. A Mann-Whitney U test was used to compare the medians of continuous variables. Overall survival (OS) was measured from the date of diagnosis to death from any cause or the date of the last follow-up. Progression-free survival (PFS) was measured from the date of diagnosis until the earliest occurring time point among the following: the end of the follow-up period, the date of relapse (regardless of location), PD, or death from any cause. CNS/systemic PFS was defined as the time without CNS and/or systemic relapse. Cumulative incidence curves were determined for relapse events, and Gray's test was used to examine the significance of between-group differences. The binary logistic regression analysis was performed to identify the independent risk factors associated with CNS/systemic relapse. The Kaplan-Meier method was used to calculate the curves of OS and PFS, and a log-rank test was used to identify significant between-group differences. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using univariate and multivariate Cox proportional hazard regression models to determine the independent risk factors associated with survival. A two-sided p value <0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 21 software (IBM).

| Patient characteristics
A total of 32 patients were evaluated, with a median age at diagnosis being 63 years (range: 39-78 years) and 22 (68.8%) being women. Among them, 23 had PVRL and nine had concurrent VRL, and 18 had unilateral and 14 had bilateral ocular involvements. The clinical characteristics of patients with VRL are shown in Table 1. Furthermore, the data of cytologic and flow cytometric analyses of the vitreous fluid in three representative patients are presented in Figure S1. The median duration of symptoms before diagnosis was 6 months (range: 2-32 months). All patients presented with ocular complaints without associated neurologic symptoms or signs at diagnosis. The most common presenting symptoms were blurred vision in 26 (81.3%) and floaters in 15 (46.9%) patients. The disease laterality was not associated with concomitant CNS involvement.

| Treatment response
The treatment responses to the combination therapy are presented in Table 1. Twenty-six (81.3%) patients achieved CR, 87% in the PVRL subgroup, and 66.7% in the concurrent VRL subgroup (p = 0.314). Intriguingly, the six nonresponder patients all achieved CR after the salvage therapies.

| Relapse events
During the median follow-up period of 103.5 months (range: 4.4-195.2 months), 24 (75%) patients experienced relapse events. The median time to first relapse was 20.3 months (95% CI, 12.9-27.7 months). Of these 24 patients, 11 had isolated intraocular relapse, 12 had CNS/ systemic relapse, and one had simultaneous intraocular and CNS involvement at the first relapse.
The clinical characteristics and treatment outcomes of the 11 patients with isolated intraocular involvement at first relapse are displayed in Table S1. All 11 patients received more frequent IVT chemotherapy as the salvage therapy, and one of them also received consolidated whole-brain radiotherapy (WBRT). They all achieved a second CR. Afterward, four patients had continuous CR, three had subsequent intraocular relapse alone, and four developed subsequent CNS/systemic relapse. The three patients with serial intraocular relapse alone were all alive without disease after further treatments. Nevertheless, two of the four patients with subsequent CNS/systemic relapse died of disease (Cases 14 and 19, Table S1).
The only one patient with simultaneous intraocular and CNS involvement at first relapse received carmustine-based systemic chemotherapy in combination with IVT chemotherapy as the salvage therapy (Case 6, Table 2). Nevertheless, the disease rapidly progressed, and the patient died of disease in 2 months.
Overall, intraocular relapse was observed in 12 (37.5%) patients. The median time to intraocular relapse was 20.3 months (95% CI, 12.2 to 28.4 months). The cumulative incidence rate (CIR) of intraocular relapse at 2 and 5 years were respectively 20.6% and 42%, reaching a plateau at 33.6 months ( Figure S2). The incidence of intraocular relapse did not differ between PVRL and concurrent VRL subgroups (p = 0.422).
The clinical features and treatment outcomes of the 17 patients with CNS/systemic relapse are summarized in Table 2. The therapeutic strategies for CNS/systemic relapse were determined as the discretion of the primary -care hematologists. Eleven of 17 patients with CNS/systemic relapse eventually died of disease.

| Survival
After a median follow-up time of 103.5 months, the median OS had not been reached. OS rate at 5 and 10 years was 73.3% and 58.9%, respectively ( Figure S4A). The median PFS was 20.3 months (95% CI, 11.8 to 28.8 months), and the 2 and 5-year cumulative PFS rates were 50% and 29.9%, respectively ( Figure S4B). Additionally, the median CNS/systemic PFS was 69.5 months (95% CI, 31.5 to 107.5 months, Figure S5). Twenty patients remained alive at the time of last follow-up, while 12 patients died-11 from recurrent disease and the remaining one from infection.
Age, sex, and disease laterality had no prognostic implication in our cohort. Compared with the patients with PVRL, those with concurrent VRL had a trend of poorer OS (median OS, 59.1 months vs. not reached, log-rank p = 0.167, Figure 2A) and a significantly shorter PFS (median PFS, 7.3 vs. 33.5 months, log-rank p = 0.034, Figure 2B). Additionally, a shorter CNS/systemic PFS was observed   Figure S6). Intraocular relapse had no impact on OS. In contrast, multivariate analysis identified CNS/systemic relapse as an independent unfavorable factor for OS (HR: 11.315; 95% CI: 1.335-95.892; p = 0.026, Table 3). The 2 and 5-years OS rates post-CNS/systemic relapse were 58.2% and 34.5%, respectively.

| Risk factors associated with CNS/ systemic relapse
Several clinical features, including age, sex, disease laterality, and concomitant CNS involvement at diagnosis, were examined to explore the factors predictive of CNS or systemic relapse. By multivariate binary logistic regression analysis, we identified that concomitant CNS involvement at diagnosis is the only risk factor significantly associated with the development of CNS and/or systemic relapse (Table 4).

| Treatment-related adverse effects
The most common adverse effect of IVT methotrexate was superficial punctate keratitis, which occurred in 21 (65.6%) patients. Dry eye and sterile endophthalmitis were occurred in five (15.6%) and two (6.3%) patients, respectively. Only one (3.1%) patient developed maculopathy, and none had a vitreous hemorrhage. Concerning the hematologic toxicities, grade 3/4 neutropenia was observed in four (12.5%) and grade 3/4 thrombocytopenia in three (9.4%) patients. No grade of anemia was recorded. Two out of 32 patients also developed grade 4 febrile neutropenia. Intriguingly, all but one (31/32, 96.9%) patients developed hepatotoxicity, of which 15 was grade 1, six was grade 2, and 10 was grade 3 toxicity. Nonetheless, the hepatotoxicity was reversible, and no late sequelae were reported. Seven (21.9%) patients developed HDMTX-related nephrotoxicity, among whom one had grade 1 and six had grade 2 toxicity. Notably, the renal function was completely recovered in six of the seven patients during the posttreatment follow-up period. Overall, there was no toxic death or premature discontinuation of HDMTX due to toxicity.

| DISCUSSION
VRL is a rare intraocular malignancy with a slight female predominance. Our study also showed that approximately 70% of patients with VRL were female. In contrast to previous studies that VRL are mainly bilateral, more than half of patients in our study had unilateral involvement at diagnosis. Our finding might underestimate the true value because VRL is often asymmetrical and initially may present as being unilateral. 23 VRL commonly masquerades as posterior uveitis, which is responsible for its diagnostic delay. In our study, the median time from the onset of clinical symptoms to the definite diagnosis of VRL was 6 months, comparably with other reported data. 12,24,25 We had previously reported the efficacy of combined IVT methotrexate and systemic HDMTX treatment in 19 patients with VRL. 22 During a median follow-up period of 40 months, the 5-year survival rate was 55.8%. CNS/ F I G U R E 1 Cumulative incidence of central nervous system (CNS)/systemic relapse in patients with vitreoretinal lymphoma (VRL), stratified by the presence or absence of concurrent CNS disease at diagnosis. A higher cumulative incidence rate (CIR) of CNS/systemic relapse was observed in patients with concurrent intraocular and CNS disease (concurrent VRL) than in patients with primary VRL (PVRL). The CIR of CNS/ systemic relapse at 2, 5, and 10 years were respectively 44.4%, 77.8%, and 88.9% in patients with concurrent VRL and 13.6%, 29.4%, and 50.5% in patients with PVRL. systemic relapses arose in six (31.6%) patients. In this extended and updated study with more patients enrolled, the overall survival remained high (5-year survival rate of 73.3%), supporting the survival benefit regarding this combination approach. Nevertheless, a higher proportion of patients (17/32, 53.1%) experienced CNS/systemic relapse during a longer follow-up period (9/23 [39.1%] among PVRL, 8/9 [88.9%] among concurrent VRL). Furthermore, the CIR of CNS/systemic relapse increased steadily over time, especially for the concurrent VRL subgroup. Intriguingly, late CNS relapse occurred in one patient beyond 10 years from diagnosis, underlying the inherent risk of CNS relapse in VRL. All these results suggest that the present combination approach could not effectively reduce the occurrence of subsequent CNS/systemic relapse, even if systemic HDMTX was given for a relatively high number of cycles (median 10 cycles).
The effectiveness of combined IVT methotrexate and systemic HDMTX therapy in VRL had been addressed. Akiyma et al. prospectively investigated 10 patients with PVRL treated with IVT methotrexate till CR of ocular lesions followed by five cycles of HDMTX (3.5 g/m 2 ). 26 Even though all patients achieved CR, a significant portion (40%) of them experienced CNS relapse. Hashida et al. retrospectively explored the association between CNS prophylactic treatments and CNS relapse in 26 patients with PVRL and reported that the time to CNS progression was significantly prolonged with the combined approach versus IVT therapy alone. 27 However, four of six (66.7%) patients using systemic HDMTX as the preventive therapy eventually developed CNS relapse. More recently, a retrospective study of 59 patients with PVRL showed a high CR rate and a long median brain-free survival after the HDMTX-based polychemotherapies. 10 Nevertheless, a total of 25 (42.4%) patients ultimately developed brain/systemic relapses. These studies further support our perspective that systemic HDMTX therapy can't effectively prevent the onset of CNS/ systemic relapse, even if it can considerably prolong the time to subsequent CNS/systemic involvement.
Previous studies had demonstrated heterogeneous outcomes while using different forms of combination therapy in VRL treatment. From a small prospective study enrolling 17 patients, Kaburaki et al. reported a relatively low 4year CIR of CNS progression (14.3%) after a conventional PCNSL-like treatment consisting of rituximab, HDMTX, procarbazine, and vincristine (R-MPV) with IVT methotrexate as the induction therapy followed by consolidative reduced-dose WBRT and high-dose cytarabine (HDAC). 28 In addition, de la Fuente et al. reported the outcomes of combined modality comprised of bilateral ocular radiotherapy followed by systemic HDMTX-based chemotherapy in 12 patients with PVRL. 25 With a median follow-up time of 68 months, a low 5-year CIR of CNS relapse (37.5%) was observed but at the expense of radiation retinopathy, cataract, and optic atrophy. By contrast, Cheah et al demonstrated a high 4-year CIR of CNS progression (58%) in a retrospective single-center study consisting of 11 patients with PVRL treated with R-MPV followed by binocular radiotherapy and HDAC. 29 Whether the incorporation of ocular radiotherapy, WBRT, or intensive chemotherapy into the combination strategy results in lower incidence of subsequent CNS/systemic relapse needs to be further investigated.
Despite similar CR rates after the combination approach, concurrent VRL was associated with a significantly higher CIR of CNS/systemic relapse than PVRL. Moreover, all but one patient with concurrent VRL developed CNS/ systemic relapse, indicating that systemic HDMTX monotherapy adopted in this study is not enough to eliminate the occult CNS or systemic diseases in this subgroup of patients. Thus, the PCNSL-like treatment 30 that were resistant to the frontline HDMTX monotherapy. Whether the addition of other chemotherapeutic regimens into HDMTX is more effective than HDMTX monotherapy to eradicate the occult systemic lesions in VRL requires further explorations. More recently, increased insight into the pathophysiology of VRL has contributed to the development of several novel agents for this rare malignancy, especially for the Bruton's tyrosine kinase inhibitors (BTKi). 32,33 Guan et al. also demonstrated a high disease control rate (90%) after 1 month of BTKi monotherapy in a prospective, singlecenter, phase 2 study enrolling 10 patients with relapsed or newly diagnosed VRL, paving the way for a paradigm change in the choices of VRL treatment. 34 The main limitation of this study is the inherent biases due to the retrospective nature. Additionally, only one type of systemic treatment was performed, so the difference in the comparison between HDMTX-based polychemotherapies and HDMTX monotherapy as the frontline regimen in VRL treatment couldn't be known. Furthermore, we did not routinely examine the levels of interleukins or novel biomarkers in vitreous or aqueous samples at diagnosis and at the end of treatment, which might influence the ocular assessments in our study. Nevertheless, the treatment adopted in our study is relatively homogeneous and the length of follow-up was sufficient to draw conclusions. Our study truly showed the real-world data regarding the use of combined IVT methotrexate and systemic HDMTX monotherapy in a relatively large cohort of Asian patients with VRL.
To sum up, we demonstrated the long-term outcomes regarding the combination approach of IVT methotrexate and systemic HDMTX monotherapy as the frontline therapy for patients with VRL. This combination approach was well tolerated with acceptable toxic profiles. Moreover, it induced a high CR and survival rates as well as a long median CNS/systemic PFS. During an extended follow-up period of nearly 9 years, nonetheless, more than half of the patients experienced CNS/systemic relapse, which represented the main cause of death in these patients. Intriguingly, constant pattern of CNS/systemic relapse was observed, indicating the suboptimal efficacy of this combination approach to prevent subsequent CNS/systemic progression. Incorporation of more intensive consolidation strategies into the treatment of VRL deserves to be considered, especially for patients with concurrent CNS disease at diagnosis.