Introduction

Primary central nervous system lymphomas (PCNSLs) of T cell origin are rarely seen in pediatric patients, and very few detailed cases are reported in the literature [1–15]. The natural history, optimal treatment regimen, and expected outcomes have not been determined. Further, establishing a definitive diagnosis remains a challenge. We report a rare case of T cell PCNSL in a 12-year-old boy and review the literature to highlight the importance of brain biopsy to definitively establish the diagnosis when PCNSL is suspected.

Case report

Presentation and examination

A previously healthy 12-year-old boy presented with worsening left-sided weakness, nausea, vomiting, headache, blurred vision, and diplopia. There was no history of fever, weight loss, or recent infection. Neurological examination revealed bilateral cranial nerve 6 palsies, weakness of the left triceps and deltoid, diminished sensation to light touch over the lateral aspect of his left forearm, and dysmetria on left-sided finger-to-nose testing. Hepatosplenomegaly and peripheral lymphadenopathy were not detected.

Diagnostic testing

Magnetic resonance imaging (MRI) of the brain revealed right parietal gyral thickening with faint meningeal contrast enhancement on T1-weighted imaging, high-intensity signal abnormalities in the right parietal and occipital subcortical and periventricular white matter and splenium of the corpus callosum on T2-weighted and fluid-attenuated inversion recovery imaging (FLAIR), and patchy cortical areas of restricted diffusion on diffusion-weighted imaging (DWI) (Fig. 1). The differential diagnosis included lymphoma, gliomatosis cerebri, vasculitis, encephalitis, and demyelinating disease.

Fig. 1
figure 1

Abnormal hyperintensity in the right parietal cortical, subcortical, and periventricular regions on FLAIR MRI (a, b). Patchy regions of cortical high intensity appeared on DWI (c) with associated low intensity on apparent diffusion coefficient (ADC) mapping (d). Faint heterogeneous contrast enhancement of the leptomeninges was also observed in the right parieto-occipital region (not shown). These findings were suggestive, but not specific, for lymphoma

Full size image

The levels of serum inflammatory markers (erythrocyte sedimentation rate and C-reactive protein) were within normal limits, and cerebral angiography did not reveal evidence of vasculitis. There were no risk factors for or evidence of an underlying immunodeficiency. Cerebrospinal fluid (CSF) was obtained via lumbar puncture with an initial CSF white blood cell count of 1790/μL, and flow cytometry and immunophenotyping showed an atypical T cell population comprising 92 % of the total leukocytes (Fig. 2a). The atypical T cell population expressed CD3, CD5, CD7, CD8, CD38, CD56, CD57, TdT, and T cell receptor (TCR) γδ and lacked reactivity for CD1a, CD2, CD4, CD16, CD20, CD34, CD99, and TCR αβ. The CD4/CD8 ratio was 1.67. Fluorescent in situ hybridization (FISH) probes for lymphoma/leukemia markers revealed no evidence of 9p deletion, but ETV6 deletion and RUNX1 expression were present. These results were suspicious for a lymphoproliferative disorder but could not differentiate a mature T cell lymphoma (CD3 and TCR γδ positivity with lack of CD1a and CD99) from an immature T cell lymphoblastic lymphoma (TdT positivity). Such histological distinction may have survival implications because the lymphoblastic subtype has a more aggressive clinical course but may be curable [16].

Fig. 2
figure 2

Cerebrospinal fluid cytology revealed small- to intermediate-size atypical lymphoid cells with irregular nuclear contours and scant cytoplasm (a). Brain biopsy of parietal cortex showed a predominantly perivascular cuffing distribution with sheets of lymphoid cells within the leptomeninges on hematoxylin and eosin staining (b). Immunohistochemistry demonstrated positivity for CD3 (c) and TdT (d) and >90 % proliferation rate on Ki-67 staining (e). These findings were consistent with T cell lymphoblastic lymphoma

Full size image

Further evaluation included a positron emission tomography scan; computed tomography (CT) imaging of the neck, chest, abdomen, and pelvis; and a total spine MRI. These studies revealed no other lesions. A peripheral blood smear found no atypical lymphocytes, and bilateral bone marrow aspirates were considered normal. Direct funduscopic examination revealed no evidence of intraocular lymphoma or papilledema. In the context of symptom progression and data most consistent with lymphoma, empiric corticosteroid therapy was initiated and resulted in clinical improvement.

To definitively establish the diagnosis, we performed a right craniotomy and open, frameless stereotactic biopsy of MRI-defined abnormal parietal cortex and subcortical white matter, as well as overlying dura and calvaria. Intraoperatively, phase reversal was conducted to verify the location of the central sulcus and primary sensorimotor cortex. The open biopsy was obtained using an 11-blade scalpel, and the specimen was sent for pathological evaluation and divided for analyses by flow cytometry, karyotype, FISH, histology, and immunohistochemistry.

Histopathological evaluation of the cortical and subcortical white matter biopsy specimen demonstrated atypical lymphoid proliferation in a predominantly cuffed perivascular distribution with sheets of lymphoid cells within the leptomeninges. The lymphoid cells were small to intermediate in size with irregular nuclear contours, immature-appearing chromatin, occasional mitotic figures, and scant amount of cytoplasm (Fig. 2b). Immunohistochemical staining revealed that the lymphoid cells were positive for CD3 and TdT and negative for CD1a, CD2, CD4, CD8, CD20, and CD33 (Fig. 2c). The proliferation exceeded 90 % by Ki-67 staining. There was no evidence of Epstein-Barr virus by in situ hybridization. The diagnosis of lymphoblastic T cell lymphoma was made on the basis of the strong TdT expression and the high proliferation index. The gene rearrangement phenotype favored the late cortical thymocyte stage of T cell lymphoblast differentiation.

Treatment

Because of progressive left-sided weakness, the patient was given dexamethasone while the non-invasive diagnostic testing was performed, but this was discontinued 3 days prior to the brain biopsy. After the diagnosis of lymphoblastic T cell PCNSL was made, the patient received a total of four cycles of systemic chemotherapy: two cycles of high-dose cytarabine and etoposide, a regimen commonly used in children with mature B cell non-Hodgkin’s lymphoma with CNS involvement, and two cycles of high-dose methotrexate (5 g/m2), vincristine, and 6-mercaptopurine, a regimen commonly used in acute lymphoblastic leukemia with CNS involvement. He also received four cycles of intrathecal methotrexate and systemic hydrocortisone and cytarabine. Shortly after beginning chemotherapy, he had complete resolution of prior neurological symptoms, including diplopia. He experienced a steady reduction in CSF pleocytosis and ultimately had no detectable disease in his CSF by morphology or flow cytometry after the completion of four cycles of systemic and intrathecal chemotherapy.

Following chemotherapy, the patient underwent craniospinal radiation therapy to a total dose of 22.5 Gy, given in 1.5-Gy daily fractions. This was followed 3 weeks later by 13.2 Gy of total body irradiation, given in 1.65-Gy, twice-daily fractions for 4 days, prior to allogeneic stem cell transplant. Serial MRIs throughout treatment have revealed steady improvement with resolution of meningeal enhancement on T1-weighted contrast-enhanced imaging and decrease in signal abnormality involving the corpus callosum and periventricular and subcortical white matter. The patient remains alive 12 months after the initial presentation with no evidence of active disease.

Discussion

Epidemiology

PCNSL is distinct from systemic lymphoma of the same histological subtype, with a significantly lower incidence and worse prognosis [17]. In immunocompetent patients, PCNSL accounts for only 5 % of primary intracranial neoplasms and 1–2 % of non-Hodgkin’s lymphoma [18, 19]. Only 1 % of all reported PCNSL cases occur in children, giving an estimate of 15–20 pediatric cases per year in North America [20, 21], but the overwhelming majority of PCNSLs are of B cell origin. T cell lymphomas are rare among both adults and children, comprising only 2–8 % of PCNSL cases [22–26]. In contrast, systemic T cell lymphomas account for 15–20 % of cases of non-Hodgkin’s lymphoma [25]. The reason for the difference in proportion of T cell versus B cell lymphomas in systemic versus PCNSL is largely unknown [22–24, 26]. Most published clinical series and literature reviews of T cell PCNSL are composed of a heterogenous group of patients with a wide age range. Few studies have focused on the pediatric age group.

After review of the literature, we identified only 18 cases (13 male, 5 female) of T cell PCNSL reported in patients under 18 years (range 2–17 years), including our patient (Table 1). Three patients were immunocompromised, 11 were immunocompetent, and the immune status of 4 was unknown. Eleven patients had more than 18 months of follow-up and were still alive at the time that the cases were reported; six of these survived at least 5 years after diagnosis. Five patients died within 6 months of diagnosis. The three immunocompromised patients were still alive after a follow-up of 20 months, 6 years, and 10 years, respectively. There is considerable variation in presentation, diagnostic workup, treatment modalities, and clinical outcomes among these cases, which limited the opportunity to draw conclusions about T cell PCNSL as a disease entity. Given the paucity of reported cases in children, it is likely that T cell PCNSL is under-recognized [27]. This may be explained in part because it presents a diagnostic challenge, both radiographically and pathologically.

Table 1 Literature review of pediatric T cell primary CNS lymphoma cases
Full size table

Diagnosis and imaging

T cell PCNSL can present in many different forms. Tumors can involve the cortical and subcortical cerebrum, cerebellum, basal ganglia, leptomeninges, and dura. Multifocal disease at the time of presentation is common, even in immunocompetent patients. A space-occupying parenchymal mass is the most common presentation of PCNSL; solitary leptomeningeal disease is a rare variant [28, 29]. Leptomeningeal dissemination, however, is a common manifestation for PCNSL of either B cell or T cell origin, occurring in 7–42 % of patients [30, 31]. The presence of leptomeningeal involvement does not appear to influence outcome [30–34].

Because of the variability in appearance on radiographic imaging, PCNSL cannot be distinguished definitively from reactive processes, including demyelinating lesions, infection, abscess, inflammation, and other neoplasms, such as high-grade glial tumors or metastases. Homogeneous enhancement is common in PCNSL of B cell origin [35, 36], but T cell PCNSL is often characterized by heterogeneously enhancing lesions [37, 38]. The presence of ring enhancement or significant peritumoral edema is variable [27, 37]. Cystic degeneration with or without hemorrhage was thought to be more common in immunocompromised patients but has also been seen in immunocompetent patients and may be more common in PCNSL of T cell origin than cases of B cell origin [37]. Regions of high and low signal intensities on DWI resembling an abscess or necrosis have been observed [37]. MRI may underestimate the extent of disease, because microscopic infiltration of parenchyma may not be associated with signal abnormality [39]. Because of this significant variability, low- and high-grade lymphomas cannot be distinguished by radiographic characteristics alone.

Establishing a pathological diagnosis requires obtaining a sufficient amount of abnormal tissue from a CSF sample or brain biopsy; however, a definitive diagnosis often cannot be established even if a relatively large volume of tissue is obtained. CSF analysis fails to establish a diagnosis in approximately one third of patients with PCNSL and leptomeningeal disease [40]. Flow cytometry may enhance detection in patients with a low concentration of malignant cells in CSF [41] but is non-specific given the wide variety of conditions that exhibit monoclonal populations of lymphocytes [27, 42–45].

In this case, we could not differentiate between an immature and a mature lymphoproliferative process based on flow cytometry results alone. The specimen expressed both mature (CD3 positive, TCR γδ positive, CD1a negative, and CD99 negative) and immature features (TdT positive). When PCNSL is suspected, brain biopsy corresponding to an area of abnormal enhancement or signal intensity on T2-weighted MRI may maximize the likelihood of establishing a definitive diagnosis. Identification of histological features may have implications regarding treatment and survival. Systemic lymphomas with immature cellular features, such as the lymphoblastic subtype seen in our patient, are more aggressive but frequently responsive to treatment and can be curable [16].

Potential pitfalls in diagnosis

Even after obtaining a brain biopsy, there are several potential pitfalls when making a pathological diagnosis of T cell PCNSL. In contrast to B cell lymphomas, most PCNSLs of T cell origin do not show high-grade morphology or atypical features [27]. Further complicating the diagnosis is the presence of a mixed infiltrate, containing reactive T cells, B cells, and macrophages, that is a common feature in any lymphoproliferative process. Concentric perivascular lymphoid aggregates can be present in both neoplastic and reactive processes. In some cases of T cell PCNSL, reactive B cells outnumber neoplastic T cells [27]. High proliferation rates and Ki-67 staining are not specific for differentiating reactive from neoplastic infiltrates [27, 45–47].

If corticosteroids were administered prior to the biopsy, an abundance of macrophages may also obscure the identification of neoplastic T cells. Corticosteroids can induce lymphoid cells to undergo apoptosis, depleting neoplastic cells within 24–48 h, rendering the sample non-diagnostic. After a few days, PCNSL cells may develop resistance to corticosteroids and repopulate the brain parenchyma. At least one report has described obtaining a successful PCNSL diagnosis, via molecular testing, after corticosteroids were started [7], but obtaining a biopsy before initiating corticosteroid treatment is ideal. In the case presented, our patient improved clinically after initiation of empiric corticosteroids, but these were rapidly tapered and discontinued 72 h prior to surgery to optimize diagnostic yield.

Normal markers of T cells include CD3, CD4, and CD8. Alterations of T cell surface markers can occur during both neoplastic transformation and reactive processes. Such change in antigen expression can include loss of CD7 and/or CD5 [43]. Cases have been reported of T cell lymphomas that are CD4 positive, CD8 positive, both CD4 and CD8 positive, and both CD4 and CD8 negative [27, 37]. CD5 expression is present in normal T cells but can also be seen in certain B cell lymphomas [48]. Aberrant expression of CD20, a marker traditionally thought to be B cell specific, may be present in up to 12 % of normal T cells as well as in lymphomas [49, 50]. A shift in the CD4/CD8 ratio has been investigated as a means to distinguish among T cell proliferative disorders; however, this ratio may fluctuate over the course of a pathologic process and CD4 or CD8 predominance can be seen in a variety of conditions [27, 42–45]. Monoclonality is usually assessed by molecular testing for TCR gene rearrangements [42, 44]; however, given the variety of conditions in which monoclonal populations of T cells have been observed, results of TCR gene rearrangements need to be interpreted within the appropriate clinical context [27, 42–45].

Treatment

In the few reported cases of pediatric T cell PCNSL (Table 1), a wide variety of treatment modalities have been used, including surgical resection without adjuvant treatment, radiation or chemotherapy alone, allogeneic stem cell transplant, and multimodality regimens with various combinations. Because of the paucity of confirmed cases, little information can be extrapolated from the literature to help guide treatment for T cell PCNSL in children. Current management is based on analysis of patients with PCNSL of B cell origin.

The IPCG review included 29 child and adolescent patients (mean age 14 years) with PCNSL, including only 5 with T cell PCNSL, who underwent a variety of treatment regimens with 5.7 years of mean follow-up [20]. The authors found that prognosis depended on the intensity and type of chemotherapy. Good outcomes were observed in children treated with chemotherapy alone consisting mainly of regimens utilizing high-dose methotrexate. The authors also concluded that radiation can be deferred for refractory or relapsed cases, a finding that has been corroborated by others [2, 14]. In the case presented, craniospinal radiation therapy was used because of a slow clinical response to intensive chemotherapy reflected by persistent CSF pleocytosis, but evidence to guide the use of radiation therapy in patients with T cell PCNSL is lacking.

Differentiating T cell lymphoma from a B cell malignancy may be important for some treatment decisions, such as whether to include rituximab into the treatment protocol; however, prognosis and treatment responses do not appear to differ between T cell and B cell PCNSL, at least in adults [26]. In addition to cellular lineage, the role of tumor cell maturity and molecular phenotyping also remains unclear. While a variety of subtypes of T cell PCNSL have been described, there appears to be a distinct anaplastic large cell variant, which is typically CD30-positive [8], as well as a small cell variant. This case was characterized by lymphoblastic T cells of small to intermediate size that favored the late thymocyte stage of T cell differentiation. The significance of the T cell PCNSL subtypes has not been determined; Shenkier and colleagues [26] reported evidence that cytologic type does not affect survival. Larger studies are needed to determine whether there is a difference in treatment or survival among the various subtypes.

Conclusion

Although PCNSL is rarely seen in pediatric patients, this case demonstrates the importance of brain biopsy in cases in which it is suspected. PCNSL of T cell origin in children remains poorly studied, with only 18 detailed cases reported over the last three decades, including this report. Increasing awareness and identification of children diagnosed with T cell PCNSL is needed to better understand the molecular biology of this disease and develop more standardized treatment regimens.