MRI Patterns of Non-Enhancing T2-FLAIR Hyperintensity Lesions in Primary CNS Lymphoma

Purpose: The non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor site at baseline have been observed of neoplastic nature in primary central nervous system lymphomas (PCNSL). Our aim was to explore the incidence, location, and morphology of the non-enhancing T2-FLAIR hyperintensity lesions in PCNSL. Methods: We retrospectively reviewed patients diagnosed with immunocompetent PCNSL at our institution. We identied and evaluated the T2-FLAIR hyperintensity lesions without enhancement that markedly decrease or disappear in the MRI after treatment. MRI characteristics of PCNSL at initial presentation were analyzed and compared between patients with non-enhancing T2-FLAIR hyperintensity lesions and patients without these lesions. Results: Among 89 patients, 10 patients (11.2%) were found to have non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor site at baseline, that showed a markedly decrease or disappearance after treatment. The locations of these lesions were as follows: the juxtacortical and deep white matter in 7 lesions, periventricular white matter in 2 lesions, basal ganglia in 1 lesion, and infratentorial aera in 1 lesion. Baseline MRI characteristics in patients with non-enhancing T2-FLAIR hyperintensity lesions exhibited a higher rate of multiple enhancing lesions (P = 0.027), and bilateral enhancing lesions (P = 0.001), compared to patients without non-enhancing T2-FLAIR hyperintensity lesions. Conclusion: Non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor lesions in patients with PCNSL, which indicated of neoplastic nature, was not rare. Those non-enhancing lesions should be incorporated in the initial evaluation of tumor burden and response in the follow up.

1 lesion, and infratentorial aera in 1 lesion. Baseline MRI characteristics in patients with non-enhancing T2-FLAIR hyperintensity lesions exhibited a higher rate of multiple enhancing lesions (P = 0.027), and bilateral enhancing lesions (P = 0.001), compared to patients without non-enhancing T2-FLAIR hyperintensity lesions.
Conclusion: Non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor lesions in patients with PCNSL, which indicated of neoplastic nature, was not rare. Those non-enhancing lesions should be incorporated in the initial evaluation of tumor burden and response in the follow up.
Introductions Primary central nervous system lymphomas (PCNSL) are mainly diffuse large B-cell lymphomas (DLBCL) that restricted to the brain, meninges, and eye at the time of diagnosis [1,2]. It accounts for 6% of all intracranial malignant tumors and for 1-2% of all lymphomas [3]. Treatment for PCNSL has been advanced signi cantly with improved survival for this rare and aggressive lymphoma of the CNS after high-dose methotrexate-based chemotherapy. This is particularly true for younger patients and those with a good performance status. However, the 5-year survival proportion remains low [4].
Magnetic resonance imaging (MRI) with contrast enhancement is the most sensitive imaging modality when PSCNL is suspected. Typical characters in immunocompetent patients with PCNSL are intense and homogeneous enhancement with well-de ned borders, usually localized in the periventricular areas [5][6][7]. Non-enhancing lesions in PSCNL are considered rare [8][9][10][11]. The most commonly used response criteria of treatment are those published by the International PCNSL Collaborative Group (IPCG), which include MRI ndings [12]. MRI response criteria rely on the changes of contrast enhancement measures.
It has been observed that the non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor site at baseline have the neoplastic nature [13]. However, the incidence, location, and shape of these non-enhancing T2-FLAIR hyperintensity lesions in PCNSL are still poorly characterized.

Patients
We retrospectively reviewed consecutive patients with all newly diagnosed immunocompetent PCNSL patients between 02/2009 and 09/2018 at our institution.
Eligibility criteria of this study included: (1) complete and available patient chart; (2) histologically con rmed diffuse large B cell lymphoma con ned to the brain at initial diagnosis, without evidence of non-CNS involvement; (3) had MRI images including at least T2-FLAIR, T2WI, T1WI, and T1WI with contrast-enhancement at baseline and follow up; (4) age over 18 years old; (5) patients didn't receive corticosteroid treatment before baseline MRI. Patients with poor images quality were excluded. Evaluation of MRI images were performed at baseline, before histological diagnosis and initiation of corticosteroid treatment, and at the time of following up after treatment. The MRI parameters included at least T1WI, T1WI with gadolinium injection, T2WI, and T2-FLAIR sequences. All MRI images were reviewed by two radiologists blinded to individual patient outcomes. Disagreements were resolved with consensus.
MRI analysis of non-enhancing T2-FLAIR hyperintensity lesions of PCNSL Initially, white matter hyperintensities in axial T2-FLAIR sequences of cerebral MRI were evaluated in all patients at the baseline and following up. Then, we identi ed the white matter hyperintensity lesions without enhancement that markedly decrease or disappear in the following-up MRI. The information including number, anatomical site, shape of those lesions was recorded. The location of lesions was divided into periventricular white matter, deep white matter, juxtacortical white matter, internal capsule, and infratentorial white matter. Infratentorial localization was de ned as lesions in the brainstem and/or the cerebellum.

MRI analysis of enhancing lesions of PCNSL
The following characteristics including number, lateralization, anatomical location, and contrast enhancement pattern of the PCNSL at initial presentation on MRI were analyzed and documented. Lateralization was based on the Talairach atlas; deep lesions were de ned as lesions in periventricular regions, basal ganglia, brainstem, and/or cerebellum according to previous study [14]. Infratentorial localization was de ned as lesions in the brainstem and/or the cerebellum. Enhancement pattern included homogeneous and heterogeneous enhancement. MRI characteristics of PCNSL at initial presentation were compared between patients with non-enhancing T2-FLAIR hyperintensity lesions and patients without these lesions.

Statistical Analysis
Clinical characteristics and imaging pro les are reported as means ±standard deviations for continuous variables. Categorical variables are reported as proportions. The independent samples Mann-Whitney Utest was used to compare continuous variables. Fisher's exact test was used to compare dichotomous variables. Signi cance was set at P < 0.05. All statistical analyses were performed using SPSS, Version 25.0 (IBM Corp., Armonk, NY, USA)

Patient Characteristics
At the time of analysis, a total of 89 patients were included in this neuroimaging study. The mean age of patients was 54.7 ± 12.3 years, and 33.7% were male. 20 (22.5%) received stereotactic biopsy and 69 (77.5%) patients underwent surgical resection. Then all patients were given methotrexate-based chemotherapy. Demographics and baseline MRI characteristics of patients were detailed in Table 1.
Characteristics of non-enhancing T2-FLAIR hyperintensity lesions Ten patients (11.2%) were found to have non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing tumor site at baseline, and these lesions showed a markedly decrease or disappearance after treatment. Total 11 lesions were con rmed in these 10 patients. One lesion markedly decreased, and ten lesions disappeared after treatment. The mean age of those patients was 53.9±11.2 years, and 4 (40%) were male.
Among these lesions, seven (63.7%) lesions located in the juxtacortical and deep white matter. Two (18.2%) lesions involved periventricular white matter. One (9.1%) lesion located in the basal ganglia, and one (9.1%) lesion located in the infratentorial area. Those lesions showed large volume, and irregular shape (Fig1, 2). Four lesions located in the right hemisphere, and the other seven lesions located in the left hemisphere.
Comparison of baseline demographics and MRI characteristics between patients with non-enhancing T2-FLAIR hyperintensity lesions and patients without non-enhancing T2-FLAIR hyperintensity lesions Demographic information and MRI characteristics in patients with or without non non-enhancing T2-FLAIR hyperintensity lesions were detailed in Table 2. Age and gender showed no signi cant difference between two groups. Patients with non-enhancing T2-FLAIR hyperintensity lesions (n = 10) exhibited a higher rate of multiple enhancing lesions (60.0% vs 24.1%, P = 0.027), and a higher rate of multiple enhancing lesions (70.0% vs 19.0%, P = 0.001), compared to patients without non-enhancing T2-FLAIR hyperintensity lesions (n = 79). There were no signi cant differences in other radiological parameters including location, enhancement pattern between these two groups.

Discussion
The aim of this study was to analyze the prevalence and characters of non-enhancing T2-FLAIR hyperintensity lesions of neoplasia nature in PCNSL patients. Better understanding of the pattern and mechanism of non-enhancing T2-FLAIR hyperintensity lesions is vital for early detection and accurate evaluation of the true burden of disease, potentially guiding the management.
Of 89 PCNSL patients, we observed that 10 (11.2%) patients had non-enhancing T2-FLAIR hyperintensity lesions at a distance from the enhancing lesions that markedly decrease or disappear. This dynamic change indicated that these white matter non-enhancing T2-FLAIR hyperintensity lesions had neoplastic nature. It has been reported that the prevalence of this phenomenon was about 23% [13]. In our research, the most frequently involved locations of non-enhancing T2-FLAIR hyperintensity lesions were deep and juxtacortical white matter. We found that patients with non-enhancing T2-FLAIR hyperintensity lesions exhibited signi cantly more multiple enhancing tumor lesions (p =0.027) and bilateral enhancing lesions (p =0.001) compared to the patients without non-enhancing T2-FLAIR hyperintensity lesions. The underlying mechanism of this phenomenon is still unknown. The consideration of PCNSL as a whole brain disease might contribute to this [15].
Pathology reveals that PCNSL is highly proliferative tumor cells in an angiocentric growth pattern, in which lymphoma cells accumulate around small and medium-sized blood vessels [16]. During the angiotropic invasive growth of tumor, tumor cells arrange themselves centripetally around the Virchow-Robin space and show a sleeve-like in ltration pattern. When those lesions grow larger, this will lead to disruption of the blood-brain barrier and enable their detection by virtue of pathological contrast enhancement. We thought the non-enhancing lesions might be the early phase of lymphoma, while they have not yet disrupted the blood-brain barrier. Routine contrast enhanced MRI only re ects the damage of the blood brain barrier, but not the degree of tumor angiogenesis. It is important to keep in mind that contrast enhancement is a surrogate for disruption of blood-brain barrier, which does not exhibit the true extent of tumor in the central nerve system. Very few PCNSL cases have been reported that do not have contrast enhancement [9,11]. Most of those cases exhibited a diffuse brain infiltration with lymphoma cells. Those diseases include lymphomatosis cerebri and intravascular lymphoma (IVL), a rare variant of aggressive (usually B cell) non-Hodgkin's lymphoma with selective growth of neoplastic cells within blood vessel lumina.
Recognizing the features of non-enhancing T2-FLAIR hyperintensity lesions that of neoplastic nature is key for early detection. However, it is challenging to distinguish the non-enhancing lymphomatous in ltration from non-speci c white matter hyperintensities lesion, which can be caused by diverse etiologies, including migraine, stroke, dementia, and healthy aging [17][18][19]. We observed that most of non-enhancing T2-FLAIR hyperintensity lesions of tumor nature mostly located in deep and juxtacortical white matter, with relatively large volume and irregular shape. Those characters might be helpful to differentiate non-enhancing white matter high signal lesions of tumor nature from other diseases. MRI spectroscopy and PET scan should be investigated as complementary tools. However, there is no speci c diagnostic method, and the most important is the follow up.
More importantly, knowing the features of non-enhancing T2-FLAIR hyperintensity lesions that of neoplastic nature is key for accurate evaluation of tumor burden. This might provide potential insight into the mechanisms of PCNSL pathogenesis, as well as help develop appropriate strategies of treatment. It is generally appreciated that the radiographic appearance of the tumor underestimates the disease extent [15]. Our result was consistent with this. However, non-enhancing lesions have not been taken into consideration in the current IPCG response criteria [12]. According to IPCG criteria, lack of contrastenhancing lesions in the absence of other lymphoma manifestations is regarded as complete response in PCNSL. Non-enhancing T2-FLAIR hyperintensity lesions that of neoplastic nature should be incorporated in re ned criteria de ning response and progression in PCNSL. Previous research has reported that the non-enhancing high signal could become the relapsing site of tumor during the follow up [13], though we did not nd this. Perhaps it is because of the small sample size in our study.
Our study has several limitations. First, it is an observational, retrospective cohort study, which may generate some biases due to confounding factors. Second, the sample size is relatively small in our study. Third, MRI scans were obtained in different type of scanner and imaging parameters, but all 3 T scanner. There was no systematic difference in diagnostic imaging qualities which in uence the assessment.

Conclusions
The present study found that 11.2% of patients with PCNSL display non-contrast enhancing lesions at a distance from the enhancing tumor lesions. The remarkedly decrease or disappearance of these white matter lesions after treatment indicated their neoplastic nature. Those lesions should be incorporated in initial diagnosis, and evaluation of response and progression in the follow up. However, further prospective studies are needed to investigate the underlying mechanism and clinical consequences of those lesions in PCNSL patients.