Primary central nervous system lymphoma: Inter‐compartmental progression

Abstract There is limited understanding of the inter‐compartmental progression and treatment outcomes of primary central nervous system lymphoma (PCNSL). In this multicenter retrospective cohort study on 234 patients with PCNSL (median age: 62.5 years [18–92]; median follow‐up 35 months [0.1–237.0]) from 2000 till 2018 were divided into group 1 (ocular, 44 patients): 1A and 1B without and with CNS progression and group 2 (CNS, 190 patients): 2A and 2B without and with ocular progression, respectively. In group 1 (44 patients), 33 patients received local treatment, and 11 patients received systemic treatment. In group 2 (15 patients), six patients received combination treatment, while seven patients received only systemic treatment. A complete response was observed in 19 (43%) and 91 (48%) patients in groups 1 and 2, respectively. The 2‐year progression‐free survival (PFS) was 35% (95% CI: 0.23, 0.54) and 56% (95% CI: 0.49, 0.63) for groups 1 and 2, respectively (p < 0.0001). Age < 60 years was significantly associated with longer PFS (median PFS 48 vs. 24 months, p = 0.01). The overall survival (OS) at 2‐year was similar among groups 1 and 2 (83% and 67%), respectively (p = 0.06). Thus, Initial compartment of involvement does not influence local response rate or OS.


INTRODUCTION
Primary central nervous system lymphoma (PCNSL) is a rare, aggressive, non-Hodgkin lymphoma that is most frequently diffuse large B-cell lymphoma (DLBCL) in origin [1]. It affects the central nervous system (CNS) compartments including the brain, cranial nerves, spinal cord, and meninges [1,2]. PCNSL-ocular (O), an ocular subset of PCNSL, is known to predominantly affect the ocular structures such as the sub-retinal pigment epithelium or sub retinal space, retina, and vitreous. The ophthalmic manifestations can precede, occur simultaneously with, or follow extra-ocular CNS disease. About 15%-25% of patients with PCNSL may present with or eventually develop ocular involvement [3]. Conversely, 56%-90% of patients with PCNSL-O eventually involve the CNS compartment [1,4]. An extensive review of studies on PCNSL indicates relatively infrequent recurrences outside the CNS compartments thereby demonstrating a unique tropism for the CNS [4]. As compared to DLBCL of sytemic origin, the overall prognosis is poor, and current treatment is comprised predominantly of methotrexate (MTX)-based chemotherapy [5][6][7][8].
In this study, relapse refers to recurrence of disease after a period of improvement within the initially affected compartment and progression refers to involvement of a previously unaffected compartment such as eye, the CNS, or vice versa. Currently, the management of PCNSL-O is focused on local ocular control, given insufficient evidence that ocular treatment decreases progression to CNS. Despite achieving high rates of local ocular control with intravitreal agents including MTX [9] and rituximab [10], OS is poor as 65%-85% of patients eventually have disease progression into the CNS compartment followed by death within a median interval of 29 months [11,12].
In order to eliminate tumor-specific mortality due to CNS progression, outcome measures including progression-free survival (PFS) and overall survival (OS) should be reported. We report intercompartmental progression and treatment outcomes including local control, PFS, and OS from a multicenter retrospective series of patients with PCNSL treated in the modern era and reflecting a real-world scenario of patients treated by multidisciplinary teams [13].

Study groups
Patients were divided into two groups based upon the compartment involved at initial presentation.

Treatment regimens
Detailed treatment information and type of regimen were recorded for all patients. Patients who received ocular treatment were categorized by type of local ocular therapy: intravitreal therapy (MTX and/or rituximab) (Table S2)

Outcome measures
The primary outcome measure was assessment of local treatment response in the eye and CNS compartments. Local treatment response was divided into complete response (CR), partial response (PR), progressive disease, or relapsed disease to mirror terminology and criteria proposed by International PCNSL collaborative group for standardization of baseline evaluation and response criteria for primary CNS lymphoma [14]. The secondary outcome measure was inter-compartmental progression determined by PFS (defined as time from treatment initiation to disease progression into another compartment: ocular or CNS; brain parenchyma with or without CSF) or death, whichever occurred first. The tertiary outcome measure was OS defined as the time from initiation of treatment until death or last follow-up. Ocular and systemic risk factors predictive of PFS and OS were also analyzed.

Statistical analysis
All continuous variables were reported as median, interquartile range, and range whereas categorical variables were reported as frequency and percentage. Fisher's exact test was used to compare patient characteristics between disease groups. Kaplan-Meier analysis was used to estimate PFS and OS at 2 and 5 years, and log-rank tests were used to compare patient groups. Univariate and multivariate Cox proportional hazard models were performed to identify risk factors associated with PFS and OS. All tests were two-sided, and p-values < 0.05 were considered statistically significant. Statistical analysis was performed using SAS Studio 3.7 (SAS Institute, Cary, NC) and R version 3.6 (R Foundation, Vienna, Austria).

Primary outcome
3

DISCUSSION
Currently, the treatment goal in the management of PCNSL-O is focused on local control, given limited evidence that systemic therapy reduces CNS progression [12]. Despite achieving high rates of local ocular control with intravitreal agents like MTX [9] and rituximab [10], OS is poor as 65%-85% of patients ultimately progress to involve the

F I G U R E 1 Progression-free survival. Kaplan-Meier plot for entire cohort (A) and by groups defined by initial presentation (group 1 [PCNSL-O] and group 2 [PCNSL-CNS]) (B)
. Abbreviations: CNS, central nervous system; PCNSL, primary central nervous system lymphoma CNS followed by death within a median interval of 29 months [11]. We have reported a well-defined and informative set of outcome measures that can be used in future studies. Patients were treated by multidisciplinary teams, and we were unable to discern subtle differences in practice patterns between centers.

Local response
The initial ophthalmic CR rate of 80% and 50%, in group 1 and 2B patients, respectively was significantly different between the groups

Other outcome measures
Given that PCNSL-O is a subset of PCNSL [16] with a strong propensity to progress to the CNS (intercompartmental progression), treatment effectiveness should not be evaluated soley in terms of local ocular response [12]. Instead, evaluation of outcomes such as PFS and OS are critical. In this study, relapse refers to recurrence of disease after a period of improvement within the previously affected compartment, and progression refers to involvement of the previously unaffected compartment such as ocular to CNS involvement or vice versa. Based upon the natural history of PCNSL-O, intercompartmental progression is an important event for capturing of PFS, a critical outcome measure to assess overall the impact of ocular therapy. An intervention that prolongs PFS can be expected to improve OS, as death from PCNSL-O is due to CNS progression. Therefore calculation of OS in patients with only ocular disease at presentation (group 1) can be misleading unless time to CNS progression (PFS) is also reported to adjust for lead time bias (Figure 3). The interest in PFS also stems in part from the fact that some treatment strategies are aimed only toward stabilization of the disease thereby reducing the morbidity. Trials that report PFS may be conducted more quickly using fewer subjects and at lower costs than those incorporating OS [17]. It is important to emphasize that mere improvement of PFS does not always translate to an increase in OS, hence the great importance of analyzing OS outcomes.

Impact of PCNSL-O on PFS and OS
A few studies have demonstrated a lack of prognostic impact of ocular involvement [18,20,24], while others have shown that ocular involvement at diagnosis has a negative prognostic impact on PFS and OS [25,26]. The discrepancy in findings can be explained by the fact that these studies were retrospective and included heterogenous treatment regimens with HD-MTX-based regimens in only 55%-84% of patients, in contrast to studies showing poorer PFS and OS in the ocular involvement group, which were prospective and included HD-MTX-based chemotherapy regimens for all patients. In our series, we also show that there is no prognostic impact of ocular involvement (preceding, or following CNS involvement) on median OS when compared among the three groups (p = 0.1). Regression analyses showed that patients with older age (≥60 years) and the type of ocular involvement (vitreous seeding vs. sub-RPE infiltrates) were not a significant factor, similar to previous observations [27]. and or combined therapy [19][20][21]27]. In contrast, a large multicenter retrospective study by Castellino et al. [28] and a single-center study by Hashida et al. [29] have shown that PCNSL-O patients receiving combined treatment had improved PFS as compared to those who received local treatment alone. Similarly, in a prospective study, Akiyama et al. [23] showed that for a group receiving combined treatment (n = 18) compared to a group with local treatment (n = 8), 2-year PFS rates were 58.3% (95% CI: 23.0%-82.1%) and 37.5% (95% CI: 8.7%-67.4%), respectively. In addition, a study by Hormigo et al. [25] showed that patients who received systemic chemotherapy for local ocular disease before CNS progression had a significantly improved OS when compared to patients who received systemic treatment after CNS progression (39 months vs. 24 months, p < 0.03). Due to small number of cases in each study, sub-groups (≤20), and heterogeneous treatment regimens, all of the published studies to date appear to be underpowered to be able to demonstrate a statistically significant impact on PFS between those treated with local ocular versus systemic therapy.

Limitations
The major limitation in all large, retrospective, multicenter studies of PCNSL are the length of study duration and the possibility that treatment regimens may change over time. Because of retrospective nature of this study, data on toxicity and neurotoxicity are lacking in this study.
Another limitation is case selection bias depending on whether PCNSL-O or PCNSL disease is the presenting manifestation. Our study was over the past 18 years, a period during which treatments have consistently relied primarily on HD-MTX-based regimens. Lack of sufficient sample size in the sub-groups of patients treated with different treatment regimens.

Conclusions
Initial compartment involvement does not appear to influence the local response rate and OS (2 years