Fatal Case of Chronic Jamestown Canyon Virus Encephalitis Diagnosed by Metagenomic Sequencing in Patient Receiving Rituximab

A 56-year-old man receiving rituximab who had months of neurologic symptoms was found to have Jamestown Canyon virus in cerebrospinal fluid by clinical metagenomic sequencing. The patient died, and postmortem examination revealed extensive neuropathologic abnormalities. Deep sequencing enabled detailed characterization of viral genomes from the cerebrospinal fluid, cerebellum, and cerebral cortex.

A CSF sample collected in April 2018 underwent clinical metagenomic next-generation sequencing (mNGS) testing at the University of California-San Francisco (8) and was positive for California encephalitis virus most closely matching JCV, with reads mapping to 2 of the 3 viral genome segments (Appendix Figure 1). Another CSF sample, obtained approximately 3 weeks later in May, was negative for JCV by RT-PCR performed by CDC's Arboviral Diseases Branch (Division of Vector-Borne Diseases, National Center for Emerging and Zoonotic Infectious Diseases; Fort Collins, CO, USA); however, concurrent serum JCV RT-PCR was positive. Results of JCV IgM and neutralizing antibody testing were negative for CSF and blood from the samples obtained in May. Concurrent samples had 0% CD20+ circulating lymphocytes (reference 3%-20% lymphocytes), attributed to rituximab treatment, last administered in December 2017.
The patient was treated with intravenous immunoglobulin (total 2 g/kg), followed by a 2-week course of favipiravir, an experimental inhibitor of viral RNA polymerase, without improvement. His mental status deteriorated to a comatose state. He was transitioned to comfort care and died in June 2018, ≈1 year after suspected symptom onset.
At autopsy, the unfixed brain weighed 1,240 g and appeared grossly normal, with no masses, hemorrhage, infarctions, or herniation. Histologic abnormalities were most prominent in the cerebral cortex (particularly frontal and temporal lobes), cerebellum, and hippocampus; milder changes in basal ganglia, thalamus, and brainstem were observed, including severe loss of neurons, diffuse microgliosis with microglial nodules and neuronophagia, and perivascular and Complete or near-complete JCV genomes were recovered from premortem CSF and postmortem cerebellum and cortex tissue (both frozen and formalin-fixed paraffin-embedded) (Table; Appendix Supplementary Methods, Figure 2). Phylogenetic analysis of the small (nucleocapsid) segment showed that sequences from this patient were most closely related to JCV from mosquitoes in Connecticut (Figure 2, panel A) (9). Comparison of JCV genomes between this patient's CSF, cerebellum, and cortex revealed 27 high-confidence within-patient single-nucleotide polymorphisms (SNPs) (Figure 2, panel B; Appendix Tables 3, 4). For 13 SNPs, the variant present in CSF was different from that in cerebellum and cortex, suggesting evolution over time. For another 4 SNPs, the variant present in cerebellum was different from that in cortex, suggesting compartmentalization. The remaining 10 SNPs could represent either compartmentalization or evolution over time, because only 1 brain tissue (cerebellum or cortex) was sequenced to sufficient depth. Variability was greater in the small segment (nucleocapsid) and medium segment (glycoprotein) than the large segment (polymerase).

Conclusions
We describe an unusual fatal case of chronic JCV encephalitis in a patient who was being treated with rituximab. In contrast to this case, previously described patients with JCV have had acute illness, and JCV infection is rarely fatal (7,10). The neuropathologic findings in this patient, although nonspecific, are similar to those of a cerebellar biopsy from a patient with JCV encephalitis that showed severe loss of Purkinje and granule cells, diffuse microgliosis of the molecular layer, and leptomeningeal inflammation (5).
The lack of distinguishing clinical, radiographic, and pathologic features of JCV underscores the diagnostic utility of clinical mNGS (8). Attributable in part to low incidence and lack of commercially available targeted testing, JCV is often not considered a priori, especially in the setting of chronic progressive neurologic illness. As a further complication, standard clinical testing by serology can be negative in the setting of B-cell-depleting therapy; our patient had negative JCV serologic tests and lack of B lymphocytes by immunohistochemical staining. Similar phenomena have been reported in rituximab-treated patients with other arboviral infections (e.g., Cache Valley orthobunyavirus, Powassan virus, and West Nile virus) who lack detectable antibodies but remain viremic longer than immunocompetent patients, highlighting the importance of nucleic acidbased testing methods (11)(12)(13).
In addition to diagnosis, mNGS also provides valuable information about pathogen genomics. We report the unique assembly of a JCV genome from human clinical samples, an important advance in the study of JVC pathogenesis, virus evolution, and differences between the enzootic transmission cycle and human infection (14). The functional importance of the identified SNPs could not be evaluated from the genomic data alone; however, none  were associated with alterations of potential Nlinked glycosylation sites, cysteine bonds, or the conserved fusion domain (15). One SNP that arose between CSF and brain (small segment gA397G; aT109A) also varied between JCV strains with different neurovirulence in mice, although the functional importance is unknown (4). Although treatment options for JCV infection are largely unexplored, response to antiviral drugs probably depends on initiating treatment early in the disease course and reaching therapeutic levels in the CSF before extensive neuronal loss. Thus, broad-spectrum molecular assays such as mNGS could potentially lead to earlier treatment with improved outcomes (8).