All patients were recruited from the inpatient units of the Affiliated Brain Hospital of Guangzhou Medical University, and written informed consent was obtained from patients or their guardians, particularly in cases of severe cognitive impairment.

Comprehensive assessments were conducted, involving thorough history taking, physical and neurological examinations, neuropsychological assessments, appropriate laboratory tests, and magnetic resonance imaging (MRI) scans. Special attention was given to laboratory tests, specifically checking vitamin B12, folate, and thyroid hormone status to rule out dementia caused by anemia or thyroid hypofunction, thereby enhancing the accuracy of clinical diagnoses.

Cognitive function was evaluated using the Mini-Mental State Examination (MMSE), and diagnoses were reviewed by a senior neurologist with subspecialty training in neurodegenerative disorders.

Patients meeting the criteria for general paresis were carefully selected based on the following criteria: 1) meeting the DSM-IV criteria for dementia with a duration exceeding 1 year; 2) testing positive for the rapid plasma reagin (RPR) or the toluidine red unheated serum test (TRUST), and T. pallidum hemagglutination (TPHA) in both serum and cerebrospinal fluid (CSF); 3) having white cell counts in CSF of 10 × 10⁶/L and/or protein levels exceeding 500 mg/L. Exclusion criteria included patients with concurrent human immunodeficiency virus infection or cerebral infection caused by other microorganisms (Wang et al., 2011; Haishan et al., 2015).

The criteria for the normal control group included: 1) absence of neurological or psychiatric disorders; 2) no abnormal findings in conventional brain MR imaging; 3) no complaints of cognitive disorders; 4) an MMSE total score of 27 or higher.

¹H-MRS and structural MRI examinations were conducted using a Philips 3T Achieva MR scanner (Philips Medical Systems, Best, Netherlands). We selected the hippocampus as the region of interest. T1-weighted images facilitated tissue segmentation, and manual editing of the hippocampus was performed. A point-resolved spectroscopy (PRESS) sequence was employed to simultaneously acquire water-suppressed ¹H-MRS from the bilateral hippocampus regions. The timing parameters were set at TR/TE/NSA 2000ms/35ms/128, with an in-plane resolution of 25 × 10 mm² and a 10-mm-thick section, aligned approximately along the long axis of the hippocampus (Figure 1).

Structural MRI together with ¹H-MRS, including shimming and adjustments, required approximately 40 min of acquisition time. Peak areas of NAA, Cr, Cho and MI were estimated using fully automated spectral-fitting software. The ratios of the area under each peak were expressed relative to Cr in each spectrum. Voxel placements for spectroscopy and all data analysis were carried out by a trained radiologist who was blind to each subject’s diagnosis.

G*power software was used to calculate the power and effect size. A total of at least 52 subjects were included, with a minimum of 26 subjects in each group. Data analysis was conducted using the Statistical Package for Social Sciences (SPSS, version 29). Descriptive statistics including mean ± standard deviation (M±SD), median, and percentiles were calculated. Demographic and clinical variables between two groups were assessed using chi-square analysis and Independent-Sample t-Test. Two-group comparisons were performed with Independent-Sample t-Test or Mann-Whitney U Test. For comparisons among three groups, Oneway-ANOVA or Kruskal-Wallis H Test was employed, followed by post hoc Bonferroni’s test for multiple comparisons. This analysis included demographic and clinical characteristics data, energy metabolism data, and cognitive assessments.

Correlations between ¹H-MRS data and cognitive scale scores were examined using Spearmen correlations, and false discovery rate (FDR) correction was applied. A p-value less than 0.05 was considered statistically significant, and all tests were two-tailed.

We recruited 80 patients with GP and 57 NC subjects. There were no significant differences in age between the GP and the NC groups (p > 0.05). GP group had significant lower scores on MMSE test (p < 0.001) (Figure 2). There were significant differences between groups. Participants in each group were categorized based on their level of education. No significant differences were observed between the two groups in terms of education level (Table 1).

Metabolite concentrations in bilateral hippocampus of the GP and the NC groups were compared. GP patients showed significantly lower NAA/Cr and NAA/Cho ratios in bilateral hippocampus region (p < 0.05) compared to the NC subjects. It also exhibited a significantly higher Cho/NAA ratios in bilateral hippocampus region (p < 0.05) compared to the NC subjects. Differences in Cho/Cr and MI/Cr ratios between the GP and the NC groups were not significant in bilateral hippocampus region (p > 0.05) (Table 2; Figure 3).

According to the MMSE scores, participants were recruited and divided into three groups: 57 normal controls with MMSE score 27 to 30 (control group, NC group), 13 patients with MMSE score 20 to 26 (mild cognitive impairment group, GP 1 group) and 67 patients with MMSE score 0 to 19 (moderate-severe dementia group, GP 2 group).

There were no significant differences in age and education level between three groups (p > 0.05). Moderate-severe dementia group had significant lower scores on MMSE test compared to the others (p < 0.05). There were differences in sex between three groups (Table 3; Figure 2).

Metabolite concentrations in bilateral hippocampus of three groups were compared. GP patients showed significantly lower NAA/Cr and NAA/Cho ratios in the left hippocampus region (L-NAA/Cr and L-NAA/Cho ratios) compared to the NC subjects. So we did ANOVA test for multiple comparisons. Moderate-severe dementia groups showed significantly lower L-NAA/Cr and L-NAA/Cho ratios (p < 0.05), compared with the other two groups. But differences in L-NAA/Cr and L-NAA/Cho ratios between the mild cognitive impairment group and the NC groups were not significant (p > 0.05).

It also exhibited a significantly lower NAA/Cho and NAA/Cr ratios in right side of the hippocampus region (R-NAA/Cho and R-NAA/Cr ratios) in moderate-severe group (p < 0.05) compared to the NC subjects. Additionally, there were no difference between the GP 2 and GP 1 groups, nor between the GP 1 and NC groups (p > 0.05).

Conversely, Moderate-severe dementia group showed significantly higher L-Cho/NAA ratios (p < 0.001), compared with the other two groups. But differences in L-Cho/NAA ratios between the GP 1 and NC group were not significant (p > 0.05). It also showed a significantly higher Cho/NAA ratios in right side of the hippocampus region (R-Cho/NAA ratios) in GP 2 group (p < 0.05) compared to the NC subjects. We found no significant difference in R-Cho/NAA between the GP 1 and NC groups. Furthermore, there was no significant difference in R-Cho/NAA levels between the GP 1 and GP 2 groups (p > 0.05).

When we compared the differences of Cho/Cr and MI/Cr ratio in bilateral hippocampus region observed in all groups by means of Kruskal-Wallis Test analyze we did not find significant differences (p > 0.05) (Table 4; Figure 3).

To explore the relationships between the clinical characteristics and ¹H-MRS data in GP group, we analyzed the possible correlation of the metabolite concentrations with MMSE. There was no correlation between metabolite ratios and MMSE scores in hippocampus regions (Table 5).

¹H-MRS stands out among imaging modalities as it provides both qualitative and quantitative insights into the biochemical composition of brain tissue. This technique is valuable for assessing neuronal integrity, treatment effects, and exploring new therapies (Graff-Radford et al., 2014).

NAA is regarded as a marker of neuronal integrity (Tomiyasu and Harada, 2022). In our study, significantly lower NAA/Cr ratios were observed in the bilateral hippocampus region of subjects with GP, reflecting neuronal loss or compromised neuronal metabolism, particularly evident in patients with moderate to severe dementia.

Cr is linked to high-energy phosphate metabolism, and its levels tend to rise when energy metabolism declines. While certain research indicates that elevated Cr levels may signify a compensatory response to NAA loss, other studies utilize Cr as a reference due to its relatively stable concentration (Zeng et al., 2021; Tomiyasu and Harada, 2022).

Both hemispheres of the hippocampus in the GP group exhibited lower NAA/Cr ratios compared to the NC group, with a more pronounced trend noted in the left hippocampus region. Particularly noteworthy was the significant difference in NAA/Cr ratios observed among various severity groups within the GP cohort. Numerous studies support the role of NAA/Cr as a marker of neural functional integrity, suggesting that a decreased NAA/Cr ratio in the hippocampus indicates a greater loss of nerve cells. The observed pathophysiological mechanisms are attributed to the proliferation of T. pallidum, resulting in the occlusion of small blood vessels, hypoxic conditions, collapse of vessel walls, and reduced cerebral blood flow in the temporal lobe and limbic system (Czarnowska-Cubała et al., 2013; Xiang et al., 2013; Shang et al., 2020).

In our study, the more significant decrease in NAA/Cr ratios observed in patients with moderate to severe dementia suggests pronounced neuronal damage, which may be less apparent in the early stages of dementia. Results for individuals within the normal to mild general paresis range are inconclusive. This variability could be attributed to the differing extents of affected brain regions at various stages of dementia. Dementia associated with neurosyphilis is characterized by cortical hypointensity throughout the entire cortical thickness and brain atrophy, primarily impacting the frontal and temporal lobes. Patients with neurosyphilis-related dementia commonly exhibit clinically significant neuropsychiatric disturbances, including anger, aggression, irritability, and apathy, which are indicative of frontal lobe lesions (Zhong et al., 2017; Ha et al., 2024). As the disease progresses, spirochetal deposition may increase and gradually spread to other brain regions, causing extensive lesions in the temporal lobe or diffuse brain atrophy. And then, it shows significant abnormalities in brain metabolism and cognitive impairment. Thus, GP display a distinct metabolic pattern with a significant decrease in NAA and neuron count as the disease progresses. ¹H-MRS emerges as a valuable tool for monitoring clinical progression in patients.

Choline is a rate-limiting precursor in acetylcholine synthesis and a component of cell membrane phosphatidylcholine, often used as a marker for cellular density and membrane turnover (Maul et al., 2020; Tomiyasu and Harada, 2022). Moreover, the higher choline content in astrocytes and oligodendrocytes compared to neurons may suggest alterations in membrane phospholipids in GP, contributing to abnormal membrane repair processes, synaptic loss, and the formation of amyloid beta peptide. Higher choline contents have been detected in microglia-rich brain regions such as the thalamus (Urenjak et al., 1993; Plummer et al., 2023). The observed reduction in NAA/Cho and NAA/Cr in our patients reflects changes in membrane metabolism and neuronal loss.

Myoinositol is considered a glial marker, and elevated MI levels can result from increased glial content or activation (Tomiyasu and Harada, 2022). However, in our study, there was no significant difference in the MI/Cr ratios between the GP and NC groups in the bilateral hippocampus region. This suggests that gliosis in the bilateral hippocampus among GP patients may not be as severe as in AD patients, leading to nonsignificant differences in MI/Cr ratios among groups.

The lack of significant differences in Cho/Cr ratios among groups may be attributed to various factors. Cho/Cr ratios have shown inconsistent results in different studies. Different diseases, Brain regions, disease severity and course were influence factors (Wang et al., 2015; Joyce et al., 2022; Guan et al., 2022). In the future, other regions of the brain could be included to assess the significance of Cho/Cr ratios.

Our study revealed significantly lower neuropsychological assessment scores (MMSE) in the GP group compared to the NC group. However, within the GP group, no correlation was observed between metabolite ratios and MMSE scores in hippocampal regions. This does not imply that ¹H-MRS may not reflect the cognitive function of GP patients. One possible explanation is that a subset of patients with general paresis are more prone to experiencing diffuse cerebral atrophy. Moreover, syphilis can lead to cerebral vascular syndrome (Ha et al., 2024). It may have characteristics of vascular cognitive impairment such as white matter lesions, unlike AD, is not primarily characterised by medial temporal lobe or hippocampal atrophy. And some GP patients present with mental symptoms that are manifestations of frontal lobe syndrome. Another reason is that MMSE test is be preliminary screening scale for cognitive impairment, encompassing domains such as memory, attention, language ability, calculation and spatial ability. This test requires short time relatively, which is beneficial for patients to cooperate. However, MMSE test evaluates cognitive domains are limited which absences of testing for executive function. MMSE is lack of sensitivity for screening the early stage of dementia (Hammant et al., 2023). Thus, there was no correlation between MMSE scores and metabolite ratios.

Researchers have shown distinct patterns of CSF biomarkers (Aβ and tau) in GP and AD patients. The unique CSF Aβ pattern in general paresis suggests abnormal Aβ metabolism (Luo et al., 2015; Chiara et al., 2023). This raises questions about the potential association of GP pathogenesis with neurodegenerative factors. In clinical practice, a subset of GP patients does not experience improved cognitive function after penicillin antisyphilitic treatment. This prompts inquiries into whether these patients may have concurrent AD. Exploring the use of drugs employed for AD treatment, such as cholinesterase inhibitors, in GP patients warrants consideration. Our study identified a higher proportion of male patients in the GP group than females, prompting speculation about potential protective factors in females against syphilis. The exploration of metabolic characteristics in multiple brain regions in GP could enhance our understanding and guide treatment strategies. In conclusion, further studies are crucial to comprehensively understand the extent and persistence of brain injuries in GP.

There were several limitations that warrant consideration. The brain regions associated with cognitive functions include the frontal lobe, temporal lobe, parietal lobe, corpus callosum, among others. However, the clinical manifestations of patients with GP are diverse. In this study, we solely analyzed the levels of metabolites in the hippocampus. It is worth noting that different regions of brain damage may exhibit distinct metabolic characteristics. Future research could investigate the metabolic profiles of multiple brain regions and the concentrations of individual metabolites in patients with GP to obtain a more comprehensive understanding of the metabolic alterations in the GP brain.