We enrolled patients with neurolymphomatosis who visited Severance Hospital, College of Medicine, Yonsei University and Mokdong Hospital, College of Medicine, Ewha Womans University from January 2004 to May 2017. Patients with the following conditions were identified from medical records: 1) pathological diagnosis of hematological malignancy, 2) clinical evidence of peripheral neuropathy, and 3) abnormal enlargement, enhancement, or FDG uptake of clinically affected neural structures on limb/plexus MRI and FDG-PET/CT scans. We then excluded patients with underlying peripheral neuropathies caused by diabetes mellitus, alcohol, autoimmune diseases, and heavy metals. Finally, 13 patients with neurolymphomatosis who met the above criteria were selected, of whom 11 and 2 were evaluated at Severance Hospital and Mokdong Hospital, respectively. The present study was initiated after obtaining approval from our Institutional Review Board (IRB number: 4-2017-0386). The need for written informed consent was waived by the review board because this study had a retrospective design.

Clinical data were obtained retrospectively through the medical records of the patients. The following clinical information was collected: sex, age at symptom onset, affected nerve structures, pain, muscle weakness, sensory deficit, malignant cell type, and mortality. We additionally analyzed the effectiveness of various diagnostic modalities such as limb/plexus MRI, FDG-PET/CT, cerebrospinal fluid (CSF) cytology, and nerve biopsy.

The medical records revealed that NCSs were performed in all but one patient (patient 9). These electrophysiological studies were performed using standardized techniques with two electromyography systems: Cadwell Sierra Wave (Cadwell Laboratories, Kennewick, WA, USA) and Medelec Synergy (Oxford Instruments, Surrey, UK). Skin temperature was controlled to within the range of 32℃ to 36°C. Motor and sensory NCSs were performed with surface electrodes using the conventional technique as outlined by Oh.13 Terminal latencies, compound muscle action potentials (CMAPs), and motor nerve conduction velocities (NCVs) had been recorded from the median, ulnar, peroneal, and tibial nerves by the belly-tendon method. The sensory nerve action potentials and sensory NCVs had been recorded from the median, ulnar, and superficial peroneal and sural nerves.

A partial conduction block is defined as more than 50% decline of the negative-peak amplitude of the CMAP between distal and proximal stimulations for median and ulnar nerves and more than 60% decline of the negative-peak amplitude of the CMAP for peroneal and posterior tibial nerves, based on an American Association of Electrodiagnostic Medicine (AAEM) consensus report.14 Additionally, affected nerves were defined as testable terminal branches of nerves with abnormal enlargement, enhancement, or FDG uptake on limb/plexus MRI and FDG-PET/CT scans. We then classified the findings of the motor NCSs into the conduction-block and simple-axon-degeneration groups. The conduction-block group was defined as motor NCS findings of partial conduction block. The simple-axon-degeneration group was defined as motor NCSs that showed CMAPs <80% of the lower limit of normal without marked slowing of the conduction velocity (<60% of the normal mean) and marked prolongation of the distal latency (>150% of the normal mean).13

The Mann-Whitney test was used to compare the median time intervals between symptom onset and NCS findings in the conduction-block and simple-axon-degeneration groups. Differences were considered statistically significant at p<0.05. Statistical analyses were performed with SPSS (version 18.0, SPSS Inc., Chicago, IL, USA).

The clinical presentations of the 13 patients with neurolymphomatosis are summarized in Table 1. Seven (54%) of the patients were men, and the median age at symptom onset was 60.0 years [interquartile range (IQR)=41.0–66.0 years]. The median time interval between symptom onset and diagnosis was 65 days (IQR=26.0–102.5 days). Neurolymphomatosis was the first manifestation of hematological malignancy in four (31%) patients and the relapse of a known malignancy in nine (69%). The underlying malignancy was diffuse large B-cell lymphoma in nine (69%) patients, natural-killer/T-cell lymphoma in two (15%), T-cell acute lymphoblastic leukemia in one (8%), and a not-classified lymphoma in one (8%). Six (46%) and six (46%) patients had brain and cranial nerve involvement, respectively. Twelve (92%) patients had painful asymmetric neuropathies.

All patients received systemic chemotherapy, and four patients (patients 4, 6, 7, and 13) received intrathecal injections of methotrexate. Radiation therapy was performed in four patients (patients 3, 4, 11, and 12). Intravenous immunoglobulin therapy was performed in only one patient (patient 5), and no patient received plasmapheresis. The median survival time after diagnosis was 7 months (IQR=3.0–15.0 months), and 12 (92%) patients died during the study period. Patient 12 was alive at the last follow-up (5 months after diagnosis), but lymphoma had spread to the right facial nerve at 4 months after the diagnosis of neurolymphomatosis.

Neurolymphomatosis was diagnosed in all 13 patients based on FDG-PET/CT scans demonstrating increased linear or focal FDG uptake of the affected nerves (Supplementary Fig. 1 in the online-only Data Supplement). Limb/plexus MRI was performed in 12 patients, which revealed abnormal enlargement or enhancement in 10 (83%) patients. All patients had negative findings for autoimmune antibodies including antinuclear antibodies, anti-double-stranded DNA antibodies, antineutrophil cytoplasmic antibodies, and anti-Ro and anti-La antibodies. CSF cytology was performed in eight patients, which revealed malignant cells in three (patients 1, 3, and 9). A nerve biopsy was performed in three patients: patients 5 and 10 exhibited axonal degeneration with malignant cell infiltrates, while patient 8 showed only axonal degeneration (Supplementary Fig. 2 in the online-only Data Supplement).

The NCS findings were compatible with multiple mononeuropathies in seven patients (patients 1, 2, 3, 5, 6, 8, and 10), with brachial plexopathies in two patients (patients 4 and 7), and with mononeuropathies in three patients (patients 11, 12, and 13). Thirty-eight motor NCSs were performed in the affected nerves showing abnormal uptake on FDG-PET/CT scans (Supplementary Table 1 in the online-only Data Supplement). The results of the motor NCSs are summarized in Table 2. Ten (26%) motor NCSs revealed motor conduction abnormalities that met the criteria for a partial conduction block, and these patients were therefore allocated to the conduction-block group. The other 28 (74%) motor NCSs were assigned to the simple-axon-degradation group. The median time interval between symptom onset and NCSs was significantly shorter in the conduction-block group (median=29 days, IQR=10–68 days) than in the simple-axon-degradation group (median=48 days, IQR=29–107 days, p=0.032) (Fig. 1A). However, no significant differences in the motor NCVs, terminal latencies, or distal CMAP amplitudes were identified between the conduction-block and simple-axon-degradation groups (Supplementary Table 2 in the online-only Data Supplement). The conduction-block group showed excessive temporal dispersion in five (50%) of the ten tests (Fig. 1B and C). Inching studies demonstrated partial conduction blocks without delayed latencies at nonentrapment sites in patients 2 and 12 (Supplementary Fig. 3 in the online-only Data Supplement).

Partial conduction blocks were found in 7 (58%) of 12 patients with neurolymphomatosis. Partial conduction blocks were present at the site of increasing FDG uptake on FDG-PET/CT scans in six of these patients. The exception was patient 12, who showed a partial conduction block between the wrist and elbow segment, but focal FDG uptake was present in the upper arm on FDG-PET/CT scans (Supplementary Fig. 1B in the online-only Data Supplement). However, the patients with and without partial conduction blocks did not differ with regard to other clinical characteristics such as sex, onset age, underlying malignancy, affected neural structures, pain, and survival time. Two patients (patients 2 and 5) were initially misdiagnosed as multifocal acquired demyelinating sensory and motor neuropathy due to the presence of partial conduction blocks. Follow-up NCSs demonstrated that the initial partial conduction blocks had converted to axonal degeneration (Table 3, Fig. 2). Patient 2 received systemic chemotherapy during the interval from the first to the follow-up NCS, whereas patient 5 received steroid and immunoglobulin therapies instead of chemotherapy because he was diagnosed as neurolymphomatosis after the follow-up NCS.