This new serological study enhances and expands on our previous study of CSF serologic responses to 12 MS microbial candidates [12]. Candidates studied here include 20 different bacteria, previously identified to the genus level as enriched in brain samples from patients with MS (Table S3) [17]. Our method is improved by performing IgG concentration tests on each CSF and serum sample, improving the precision of CSF dilutions used for AI determinations. This study shows reactivity against multiple bacteria in CSF from patients with primary DD including MS, and with OND including GBS. Elevated AI values show intrathecal synthesis of antibacterial antibodies in some patients against some lysates, providing evidence for potential bacterial pathogenesis of disease.

The prior next-generation RNA sequencing analysis provided data to the genus but not the species level [17]. Thus, species selection within each genus was mainly driven by organism availability from ATCC or BEI (Table 2). At times, known human pathogens were selected over other species (e.g., Bacteroides fragilis over B. thetaiotaomicron). It is possible that the proper species were not selected for study, but it is likely that there is at least some serologic cross-reactivity among different species within the same genus.

The present study is an improvement over our previous one because we measured CSF and serum IgG concentrations experimentally in the lab using a commercially available kit, without relying on prior clinical laboratory determinations that may have been affected by collection issues or storage conditions. These techniques improve the confidence in the validity of the AI determinations, a key objective of our work. Any errors in the determination of CSF and serum total IgG concentrations would have been reflected in subsequent AI determination, especially if CSF IgG concentration was relatively low. For this reason, multiple replicates of CSF and serum IgG determinations were performed if possible. Unfortunately, the quantity of several CSF specimens was limited. Consequently, we could not confirm some values by retesting, nor could we test all bacterial lysates in some subjects where CSF was limited.

This study was not designed or powered to show statistical differences between the DD and OND groups. The groups were small, with only 24 patients total, reflecting the difficulties in obtaining paired CSF and serum samples from living subjects. We obtained most of these samples from inpatients at the University of Utah Hospital over the course of many years.

Another limitation of this study is likely cross-reactivity of the samples across different bacterial lysates. Figure 2, the ELISA Index (EI) cluster, addresses cross-reactivity to some degree. It shows mathematical relatedness among each of the rows, representing EI values against each of the bacterial lysates. Note that while some of the rows are closely related (e.g., Bacteroides and Lactobacillus), no row is identical to any other, showing that the antigens are not identical.

Another limitation concerns the use of disease-modifying therapies (DMT) in some subjects. DMT are immunosuppressive, so they would be expected to cause decreased intrathecal CSF immune responses. Some patients in both the DD and the OND groups received DMT, which may have affected CSF immune responses (Table 1 includes DMT use). DMT may also alter the gut microbial composition with an impact on the immune system [22]. Consequently, our results do not allow us to exclude the effects of DMT on immune responses that we measured in CSF. However, half the DD subjects did not receive specific treatments prior to CSF collection, and among the cluster of four patients with elevated AI (Fig. 3), three had prior treatment for DD. This argues against a strong DMT effect on the presence of elevated antibacterial AI.

Antibody index. Antibody index values were derived by diluting serum samples diluted in to match each sample’s total CSF IgG concentration. This allowed the direct assessment of intrathecal IgG synthesis using the EICSF:EIserum ratio or Antibody Index (AI). AI values > 1.5 were defined as positive, with those 1.0–1.49 equivocal, and < 1.0 negative. This cluster figure shows mathematical relatedness between the CSF samples (columns) and bacterial lysates (rows)

Furthermore, it was difficult to establish final neurologic diagnoses among several study participants. This necessarily led to phenotyping uncertainties. For example, OND-89 is a young woman who developed neurologic symptoms and acute brainstem disease, with typical demyelinating lesions on brain magnetic resonance imaging (MRI). She was initially diagnosed with MS; later, the MS diagnosis was rejected in favor of “progressive neurologic deterioration with inflammation,” which was then re-revised to “neurosarcoidosis.” In some cases (e.g., DD-21), different neurologists did not agree on the final diagnosis. To provide optimal accuracy and consistency, we report the latest neurologic diagnosis provided in the hospital or, when possible, at follow-up neurology clinic visits (Table 1).

The AI was developed to show intrathecal antibody production against a given lysate [17]. This study shows that multiple subjects from both the DD and OND groups did produce intrathecal antibodies against some of the tested lysates. This technique is similar to that used for the diagnosis of neuroborreliosis (CNS Lyme disease) and varicella-zoster encephalitis [23,24,25,26]. We believe that elevated AIs provide evidence for the pathogenicity of these microbes. To avoid further complexity, we did not correct for BBB function as others have done when analyzing AI levels [18]. Since AI is a reactivity ratio that compares CSF to serum diluted to equivalent total IgG concentrations, in the absence of intrathecal antibody production, even a severely impaired BBB should result in an AI no greater than 1.0. Furthermore, there were clearly elevated AI values from subjects with normal BBB function, as assessed in the albumin index, such as DD-85 and DD-87, women age 26 and 23 with new onset MS (Tables 1 and S2). However, not all subjects with confirmed DD had measurable increases in intrathecal antibody production (AI > 1.0). In contrast, each CSF sample we tested displayed antibacterial antibody activity above that of the negative control (i.e., EI > 1.0) for most tested lysates.

Clustering of AI values shows some intriguing patterns (Fig. 3). Notably, four DD specimens (85, 87, 92, 105: two females, two males) clustered together (adjacent columns) and were reactive against Bacteroides and Lactobacillus, which are also closely related (adjacent rows). These bacterial genera are part of the normal flora in the human intestinal and female reproductive tracts. Interestingly, subject 92 was a man who developed progressive DD starting a few days post colonoscopy. Subjects 85 and 87 are young females with normal CSF albumin indices, demonstrating intact BBB function despite MS development. Two OND subjects (89, 91) and one DD subject (92) demonstrated elevated AIs against many bacterial candidate pathogens, suggesting a relatively unfocused intrathecal antibody response.

After previously demonstrating Akkermansia sequences in the brains of MS patients, our current study again documents CSF Akkermansia reactivity, supporting the concept that Akkermansia is related to MS pathogenesis, given anti-Akk intrathecal antibody production in three subjects from the DD group (87, 102, 105) [27,28,29,30]. Another group has recently studied anti-Akk intrathecal responses and linked these to MS pathogenesis [31]. Despite great enthusiasm about a role for EBV in MS pathogenesis [6], our revised sequencing data did not show increased expression of EBV in brain samples we analyzed previously (Table S3), so we did not include EBV in our study. In addition, we did not study two virus families that were overexpressed in brain samples (tobamovirus and herpes simplex virus), due to incompatibility of viruses with our method for isolating and quantitating sonicated bacterial lysates.

To fully understand the meaning of these results, it is helpful to consider individual cases. Subject 87, a young woman presenting with severe fatigue and new sensory and ocular symptoms, had an MS diagnosis confirmed after OCB testing showing matched bands, with many T2 bright, flair-positive lesions on MRI. Our analysis showed AI > 1.0 against Akkermansia and Atopobium, and > 2.0 against Lactobacillus and Bacteroides. That is, measured IgG antibody levels (EI) against these bacterial lysates were higher than expected based on serum antibody levels (EI). Her normal albumin index value (4.3) indicated a relatively intact BBB. This suggests that Bacteroides and Lactobacillus exposure are responsible for at least part of the inflammatory response in her brain, leading to demyelination and MS. Subject 101 is a man who developed GBS, then chronic inflammatory demyelinating polyradiculoneuropathy (CIDP), soon after a documented episode of Campylobacter gastroenteritis. Campylobacter is believed to cause 30–40% of GBS cases [32,33,34]. Interestingly, his CSF AI against Campylobacter many months later was not elevated (0.77), while mild elevations were seen against Leptotrichia and Staphylococcus. This suggests decay in the antibody response to Campylobacter over the year since the onset of his neurologic symptoms, perhaps due to intensive treatment with multiple plasma exchanges. Subject 105, a man with longstanding MS, was relatively functional until he was admitted for debilitating transverse myelitis associated with zoster (shingles). Our study did not include viral antigens due to the lack of enrichment of zoster sequences in our prior study and due to technical issues as explained above. Comparing CSF and serum anti-varicella zoster virus IgG antibodies would have been interesting, but zoster is a known complication of his DMT and likely not the original microbial cause of his MS. AI against 9 of 20 bacterial lysates was elevated, supporting the hypothesis that his original underlying MS, but not subsequent transverse myelitis, was related to some or all of these bacterial genera.

The origin of OCB in MS has been elusive, despite years of investigation [3]. Research has shown that OCBs in CSF are antibodies that may be directed against a limited number of patient-specific self-antigens [35, 36]. We did not directly investigate OCB in this study, but we could compare clinical OCB determinations to our bacterial ELISA assays. Interestingly, the four DD subjects who had elevated AI against Bacteroides and Lactobacillus appeared to have relatively specific responses to these bacterial lysates and, in some cases, a few others (e.g., Alistipes). Our hypothesis is that the intrathecal antibody responses in MS are first directed against microbial pathogens, some of which cross-react with self-antigens, leading to tissue damage and subsequent demyelination.