Clinical and Genomic Profile of Primary Cranial Neurolymphomatosis

Introduction

Neurolymphomatosis (NL) refers to infiltrative lymphomatous involvement of the cranial, spinal, and peripheral nerves and nerve plexuses and is the least common neurological presentation of lymphoma.^1,2 Neuropathy resulting from compression or entrapment by adjacent large lymphomatous tumors is not considered NL. It is a rare lymphoma entity, as revealed by the largest retrospective studies from large institutions (Mayo Clinic and Massachusetts General Hospital) and the International Primary CNS Lymphoma Collaborative Group (IPCG) in which 40, 25, and 50 cases were gathered over 16, 28, and 15 years respectively.^1–3

Based on these large studies, the median age at diagnosis of NL is approximately 55.5–63 years with a slight male preponderance of 54–60%. Most cases were B-cell non-Hodgkin lymphoma (NHL) (82–97%). Diffuse large B-cell lymphoma (DLBCL) is the most common type of lymphoma associated with NL (68–75.5%).^1–3 NL is divided into primary and secondary NL depending on the absence or presence of systemic lymphoma. Clinical manifestations depend on the type of nerve involved: cranial nerves, spinal nerves, cauda equina, peripheral nerves, or nerve plexuses. The neurological symptoms can be sensory, motor, or mixed.

Cranial nerve involvement in DLBCL has been regarded as central nervous system (CNS) involvement.⁴ Thus, primary cranial nerve NL is a subtype of PCNSL in which lymphomatous involvement is confined to cranial nerves. However, this condition is extremely rare, and the biology, clinical course, and natural history have not been well studied.

Here, we describe the clinical course of a 57-year-old patient with PCNL, the genomic profile of the tumor, and a literature review.

Case Presentation

A 57-year-old male patient presented to our institution for the evaluation of multiple cranial neuropathies, including bilateral cranial nerve (CN) VII and right CNs III, V, and VI, spanning the previous 30 months. The outside magnetic resonance imaging (MRI) brain showed some enhancement of the cranial nerves, but cerebrospinal fluid (CSF) analysis was negative. He was initially managed with steroids and azathioprine was subsequently administered. His neurological symptoms improved transiently with these treatments but recurred with a reduction in the steroid dose.

At the time of evaluation, he experienced right-sided facial weakness, diplopia, and pain. Positive findings on neurologic examination included ophthalmoplegia of the right eye with limited abduction and adduction, decreased sensation of soft touch in the right V1-V2 distribution, and right-sided facial droop with weakness of the right eye closure and inability to puff the right cheek. The neurological findings were consistent with palsies of the right CN V, VI, and VII. The patient underwent an extensive diagnostic evaluation.

Magnetic resonance imaging (MRI) of the brain revealed asymmetric prominent enhancement of the right seventh cranial nerve at the mastoid segment, extending into the right parotid space, and asymmetric thickening and enhancement of the right cavernous sinus and right foramen rotundum (Figure 1). No significant abnormalities were observed in the brain parenchyma or spinal cord. Positron emission tomography (PET) scan utilizing fludeoxyglucose (FDG) radiotracer showed a pancreatic lesion, and biopsy revealed a low-grade neuroendocrine tumor. The abnormalities found on MRI involving the brain were not seen on PET scan. The CSF analysis results were negative. He underwent a right craniotomy and biopsy of the enhancing right cavernous sinus lesion, which was consistent with a large B-cell lymphoma involving nerve fibers (Figure 2). Pathology revealed multiple fragments of nerves diffusely infiltrated by CD20+ and PAX5+ large atypical cells, with a Ki-67 proliferation index of 80%. A pathological diagnosis of cranial large B-cell neurolymphomatosis was made.

Figure 1 57-year-old man with multiple cranial neuropathies. Coronal (A) and axial (B and C) contrast-enhanced T1-weighted MR images showing a nodular enhancing lesion within the anterior right cavernous sinus (arrows) (B-cell lymphoma on biopsy).

Figure 2 Biopsy of the right cavernous sinus lesion. The biopsy of the right cavernous sinus lesion comprises multiple fragments of nerve with the diffuse interstitial infiltrate of large, atypical cells (A and B, x 20 and x 40; respectively). The large, atypical cells are diffusely positive for PAX-5 (C, x 40) and CD20. The proliferative rate by Ki-67 is high (80%) (D, x 40). The findings confirmed a high-grade B cell lymphoma involving the cranial nerve.

Formalin-fixed paraffin-embedded lymphoma tissue was sent for comprehensive genetic profiling using Hematology Profile Plus by the Genomic Testing Cooperative. Hematology Profile Plus uses next-generation sequencing, Sanger Sequencing, and fragment length analysis to identify molecular abnormalities in the DNA of 179 genes and RNA of 1408 genes associated with hematologic malignancies. The results are presented in Table 1. These findings are consistent with the activated B-cell (non-germinal center phenotype) signature with mutations in MYD88, CD79b, PIM1, CXCR4, and IKZF3.

Table 1 Genomic Profiling of the Tumor

The initial induction therapy used was the RMA regimen: rituximab, high-dose methotrexate (HD-MTX), and high-dose ara-c (HiDAC). The patient had an excellent clinical response, with resolution of the neurological symptoms after the first cycle. Unfortunately, he developed acute renal failure related to HD-MTX following the second cycle, which required hemodialysis. The patient’s renal function eventually completely recovered. Repeat MRI after 2 cycles of chemotherapy showed persistent enhancement around the right trigeminal and facial nerves. He was then treated with skull-base radiation therapy, followed by the initiation of ibrutinib (560 mg daily). After radiation and four months of ibrutinib treatment, MRI showed improved enhancement around the cranial nerves, with no new lesions. He underwent high-dose chemotherapy with carmustine (BCNU) and thiotepa, followed by autologous stem cell transplantation (ASCT).

He was found to have relapse/progression four months after ASCT, with MRI showing new abnormal enhancement involving the cisternal segments of bilateral CN III and V with new enhancements within the left internal auditory canal along CN VII and VIII. Within a month, he experienced rapid neurological progression with severe bilateral shoulder pain, new left facial palsy, and difficulty with mastication. MRI of the cervical spine also showed multilevel cervical nerve root enhancement (Figure 3), whereas MRI of the lumbar spine showed enhancement of the cauda equina. The CSF cytology was positive for lymphoma. The patient underwent cervical spinal radiation with symptomatic relief. He received one cycle of rituximab, ibrutinib, and lenalidomide. Owing to rapid clinical deterioration, he was switched to two cycles of R-EPOCH (rituximab, etoposide, vincristine, cyclophosphamide, and doxorubicin) in combination with intrathecal triple chemotherapy with methotrexate, ara-c, and hydrocortisone. Due to relapsed and chemorefractory disease, the patient was referred for chimeric antigen receptor (CAR) T-cell therapy (CAR-T). He received low-dose fludarabine and cyclophosphamide lymphodepleting therapy followed by tisagenlecleucel infusion. CAR-T cell therapy was complicated by grade 1 cytokine release syndrome (CRS). MRIs performed 30 days post-CAR-T cell therapy showed interval improvement in the cranial nerve and cervical spine enhancement. However, MRI three months post-CAR-T showed disease progression with leptomeningeal and ependymal involvement (Figure 3). The patient returned home and underwent modified whole-brain radiation, excluding the skull base and upper cervical spine, based on prior treatments. Three months later, his disease progressed rapidly, with an outside MRI report indicating disseminated meningeal, ependymal, and brain parenchymal involvement, leading to his death 22 months after his initial diagnosis and 52 months after the initial episode of cranial neuropathy. The clinical timeline is shown in Figure 4. Throughout the disease course, he had periodic CT imaging that did not identify any lymphomatous involvement outside of the CNS. Additionally, monitoring of his pancreatic low-grade neuroendocrine tumor showed stability, and he never required treatment for this lesion.

Figure 3 Follow up MRI of the cervical spine (A and B) and brain (C and D) after 1-year of initial presentation with progression of neurolymphomatosis. Axial (A and B) contrast-enhanced T1-weighted MR images of the cervical spine with thickening and enhancement of multiple dorsal and ventral nerve roots. Contrast-enhanced T1-weighted MRI of the brain reveals nodular enhancing lesion with ependymal involvement in the right peritrigonal region (C, arrow). Lesion shows restricted diffusion (D, arrow) suggesting high cellularity.

Figure 4 Diagram depicting the clinical timeline of our patient.

Discussion

Our case illustrates many unique features that can be seen in PCNL. The patient had a rather long pre-diagnostic phase, spanning approximately 30 months, during which he had multiple relapsing episodes of cranial neuropathies. The patient was managed with steroids, which alleviated the symptoms; however, he was never in remission. Although there was no definitive evidence of PCNL during these months, it is quite likely that the patient already had PCNL. An analogous clinical scenario has been observed in PCNSL, in which some patients have control or even remission of the disease with steroid treatment only.⁵ The PCNL in our patient shares this feature of steroid responsiveness, as described in PCNSL. As such, PCNL should be considered as a diagnostic possibility in patients with steroid-responsive but unexplained cranial neuropathies.

PCNL initially manifested in our patient with cranial neuropathies involving CN V, VI, VII, and VIII. He developed cervical spinal NL with CSF and brain involvement. Brain progression starts with ependymal tumors, followed by cerebral parenchymal tumors. A similar pattern of brain progression has been previously described.¹ In both cases, PCNL was initially observed in the brain as an ependymal tumor, followed by further dissemination in the brain parenchyma. This pattern of spread suggests that cerebral involvement occurs through the CSF pathway. Neither case showed evidence of PCNL via retrograde spread through the cranial nerves.

NL is difficult to diagnose in such cases. In the Massachusetts General Hospital (MGH) study covering the period 1972–2000, diagnosis was made by autopsy in 46% of the cases (33/72).¹ However, the IPCG study (1993–2008) showed a much lower incidence of diagnosis at autopsy at 8% (4/50), likely due to increased awareness and technology advancement.² In terms of diagnostic workup, MRI scans, PET scans, CSF analysis, and adequate biopsy of lesions on imaging are essential. PCNL appears as an enhancement along the tracts of cranial nerves. CSF cytology can be positive in approximately 40% of the cases.^1,2 PET can be used to rule out systemic lymphoma and peripheral nerve involvement. Collaboration with neurosurgical and radiology colleagues is essential to obtain adequate biopsies.

To the best of our knowledge, this is the first report of PCNL with genomic profiling. These findings include an activated B cell (non-germinal center phenotype) signature with MYD88 L265P, CD79b, and PIM1 mutations, which are similar to the typical genomic signature in PCNSL.⁶ In terms of the mutation profile, this case can be classified as the MCD/C5 subtype.⁷ Our patient had 22 mutations affecting PIM1, indicating highly active aberrant somatic hypermutation (Figure 5). PIM1 mutations have been associated with resistance to Bruton’s tyrosine kinase inhibitors in ABC (non-germinal center phenotype)-DLBCL.⁸ We identified four deleterious missense mutations using SIFT and Polyphen prediction models (R136M, L182R, G45D, and G78R), two truncation mutations upstream of the functional kinase domain (W109* and Q127*), and four 5’UTR mutations.⁹ 5’UTR mutations have been previously reported to result in overexpression of PIM1 and confer protection against autophagy.¹⁰ Interestingly, the 5’UTR mutations were seen at lower allele frequencies than the non 5’UTR mutations (median 5’UTR Allele Frequency 7.11 (6.42–9.25) vs median non 5’UTR Allele Frequency 16.02 (7.86–36.48)). However, the significance of these findings remains unclear. Interestingly, our patient had two missense CXCR4 mutations that affected the carboxyl-terminus. This appears to be a unique feature of our case, as CXCR4 mutations have not been previously described in PCNSL.¹¹ The combination of MYD88, CD79, and CXCR4 mutations has been well-described in Waldenstrom macroglobulinemia (WM). Missense CXCR4 mutations in WM are associated with a more aggressive clinical course and resistance to ibrutinib.^12,13 Another unusual mutation found in our patient was IKZF3, which has not been previously described in PCNSL. Mutant IKZF3 in chronic lymphocytic leukemia has been associated with the hyperactive B-cell receptor signaling pathway and ibrutinib resistance.¹⁴ Another novel finding was the TBL1XR1-ITPR1 fusion, which has not been described in the literature. TBL1XR1 mutations have been reported in 14–19% of PCNSL.^15,16 The profiling also revealed inversion 3, which has been previously described in myeloid leukemia and T-cell lymphoblastic leukemia/lymphoma.^17,18 The genomic profile in our case indicated highly upregulated signaling in B-cell receptors, Toll-like receptors, and NFKB signaling.

Figure 5 Lollipop plot of PIM1.

The patient did not respond to multiple lines of therapy in a sustained manner. Although there is no standardized approach for treating NL, it is reasonable to use the therapeutic approach adopted for PCNSL. HD-MTX based therapy appears to be associated with long-term survival in a subset of patients.^19–22 Survival has improved with the introduction of rituximab.³ Radiation therapy was effective in terms of pain control, as observed in our patient. BTK inhibition was reasonable to consider for PCNSL, based on the MCD/C5 mutation profile. However, the ibrutinib therapy was ineffective. Thus, ASCT appears reasonable for eligible patients.³ The blood-nerve barrier (BNB) appears to be structurally similar to the blood–brain barrier (BBB) and has been reported to be leakier.^23–25 BNB is also likely to be disrupted in the NL. In our case, R-EPOCH treatment had a disease-stabilizing effect, indicating its ability to permeate the BNB.

The findings of the literature review of the three largest NL studies are presented in Table 2. Cranial nerve involvement ranges 20–51% with CNS involvement 10–26%. 26–52% of the cases are primary NL. DLBCL comprised 68–75.5% of the cases.^1–3 HD-MTX-based therapy was most frequently used in two recent studies.^1,3 The median overall survival for primary NL ranged from 20 (IPCG) to 138 months (Mayo). The Mayo series included patients with indolent B-cell lymphoma and peripheral nerve involvement. NL involvement in aggressive lymphoma was associated with a median OS of 46.9 months.³

Table 2 Summary of the Three Largest NL Studies

We identified 12 cases of PCNL in the literature, with eight males and ages ranging from 33–74 (Table 3). All patients had B-cell lymphoma, with seven cases of DLBCL. The most common clinical manifestation was neurological facial symptoms seen in 8/12 patients with facial pain, numbness, or paralysis. The diagnosis was established by biopsy in 11/12 cases. CSF cytology was positive in 2/5 cases. HD-MTX-based therapy was used 7/12 cases. Complete response (CR) was reported in 8/12 patients. Long-term follow-up data was not available for all patients. Three patients died at 3, 14, and 18 months of follow-up. Two patients achieved CR at one-year follow up. One patient was in CR at 3-year follow up. One patient relapsed at eight months and was alive.

Table 3 Summary of Reported Cases of PCNL

Conclusion

Primary cranial NL is a rare subtype of PCNSL. Our patient shares a similar cell-of-origin signature (DLBCL-non-germinal center phenotype) and mutation profile with those of PCNSL (MYD88, CD79b, and PIM1 mutations). It is also associated with novel genomic alterations involving multiple mutations, such as PIM1, missense CXCR4 mutations, IKZF3 mutations, inversion 3, and TXL1XR1-ITPR1 fusion. The most common clinical symptoms are facial pain, numbness, and palsy, followed by diplopia. Therefore, PCNL should be considered in patients with unexplained steroid-responsive cranial neuropathies. Diagnostic evaluations should include MRIs, PET, CSF analysis, and biopsy. HD-MTX-based therapy has been frequently used in the management of NL in a subset of patients achieving long-term remission. Further research involving multiple institutions is needed to elucidate the biology and best therapeutic approach to PCNL.