Prefrontal Planning Latency Associated With Age and Serological Evidence of Microbial Exposure

ABSTRACT

Background

It has been hypothesized that certain cognitive disorders characterized by frontal executive dysfunction, such as Alzheimer's disease (a neurodegenerative conformational proteinopathy involving amyloid‐β and tau), may be associated with prior exposure to infectious agents. This study aimed to test the hypothesis that serological markers of microbial exposure are associated with frontal lobe activity in humans.

Methods

Executive planning latency was measured in 53 adult participants in this cross‐sectional study. A venous blood sample was collected from each participant. Assays for the serological markers of bacterial and viral exposure were performed in a blinded manner.

Results

Generalized linear modeling (likelihood ratio χ ² = 27.76, df = 5, p = 0.0001) included five statistically significant independent variables. Among these, age (|t| = 4.360, p = 0.0001), Borrelia burgdorferi lymphocyte function‐associated antigen‐1 (| t | = 3.260, p = 0.002), and Ehrlichia chaffeensis and Anaplasma phagocytophilum assays (| t | = 3.138, p = 0.003) were highly significant.

Conclusion

This study supports the hypothesis that frontal lobe executive function, as indexed by prefrontal cortical planning latency, is associated with serological evidence of exposure to several bacterial genera, particularly Borrelia and Ehrlichia/Anaplasma, as well as with age.

Planning latency versus age by group. Scatterplot showing individual observations (n = 53) stratified by fibromyalgia status. LOESS smoothing curves with 95% confidence intervals (shaded regions) illustrate the relationship within each group. Overall, rₛ = 0.394 (p = 0.003).

Summary

• Frontal lobe executive function, which is indexed by the latency of prefrontal planning, is associated with serological evidence of bacterial exposure.

• These findings suggest a potential link between serological evidence of exposure to specific pathogens and executive dysfunction, warranting further investigation into the role of microbial exposure in cognitive aging and neurodegenerative risk.

1. Introduction

Norman and Shallice (1980) published a detailed model of high‐level cerebral information processing [1, 2]. This involves information being transferred from a perceptual system, via a trigger database, to schema control units and thence to an effector system. In this model, schemas refer to highly specialized routine programs. At the schema control unit level, nonroutine selection (but not routine selection) requires the action of a mechanism called the supervisory attentional system [1, 3]. This planning function of the supervisory attentional system can be assessed using the problem‐solving format of look‐ahead puzzles, in which target arrangements of colored and/or differently sized beads, disks, or blocks must be created from initial arrangements [4]. Physical examples include the Tower of Hanoi, Tower of London, and Tower of Toronto tasks, while computerized versions include the groundbreaking HACKER program and the current Cambridge Automated Neuropsychological Test Battery (CANTAB) Stockings of Cambridge task [4, 5]. These look‐ahead puzzles have two main components: a planning component and a spatial processing component; the former is indexed by the planning latency, that is, the time interval from the presentation of the target arrangement to the first move [4, 5].

The ability to organize a sequence of actions toward a specific target or goal is known as an executive function. Thus, the above look‐ahead puzzles assess this function [6]. Executive planning latency is related to frontal lobe activity, with patients with frontal lobe lesions exhibiting longer latencies [7]. A trend toward longer latency was also reported in frontal lobe dementia (one‐touch Tower of London latency, p = 0.058) [8].

It has been hypothesized that infectious diseases may play an important etiological role in some neuropsychiatric disorders [9]. For example, during the coronavirus disease 2019 (COVID‐19) pandemic, acute delirium and other neuropsychiatric symptoms have been frequently reported in affected patients [10, 11]. Indeed, severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection has been associated with neuroinflammatory complications that may involve the frontal lobes [12]. Furthermore, it is well established that Treponema pallidum can cross the blood‐brain barrier and cause neurosyphilis [13]. Bransfield et al. further documented that the microbial effect on mental functioning is much greater than that classically recognized [9]. They posit that a wide variety of infectious agents have potential psychiatric manifestations, including members of the genus Borrelia, which, like members of the genus Treponema, are spirochetes; non‐spirochetal bacteria such as Chlamydia spp., Ehrlichia, and Mycoplasma pneumoniae; and certain viruses [9].

This study aimed to test the hypothesis that serological markers of exposure to these microbial agents are associated with frontal lobe activity in humans.

2. Materials and Methods

2.1. Study Design and Subjects

This was a cross‐sectional study of 53 adult subjects (age ≥ 18 years), comprising 27 patients who met the revised American College of Rheumatology criteria for fibromyalgia and 26 age‐ and sex‐matched healthy volunteers [14]. None of the participants were taking any medication or nutritional supplements known to be associated with changes in executive planning latency. This study was approved by the CPUREC Research Ethics Committee (approval no. 001/18). Further details are provided in the Ethics Statement. Each participant provided written informed consent.

2.2. Executive Planning Latency

Executive planning latency was measured as the time interval between the presentation of the target arrangement and the first move in the one‐touch Stockings of Cambridge task in CANTAB (Cambridge Cognition, Cambridge, UK). This task was performed using a touchscreen‐enabled iPad (Apple Inc., Cupertino, CA, USA). Two displays, one above the other, were shown on the touchscreen. Each display contained suspended stockings, some of which contained one or more colored balls. The upper display showed the target arrangement. The subjects manipulated the balls in the lower display (to reach the target arrangement with the minimum number of moves) by touching a given ball, then the desired position. Only one ball could be moved at a time. As would occur with physical stockings containing physical balls, only the uppermost ball in the stack could be moved.

The executive planning latency for each participant was measured in milliseconds. This was defined as the median latency across all assessed trials in which the first response was correct. In each trial, latency was defined as the time interval between the first appearance of the balls in the stockings and the first choice.

2.3. Bacterial and Viral Infection Assays

Within 1 h of the executive planning latency measurement, venous blood was collected directly into Vacutainer® serum separator tubes (Beckman Dickinson, Franklin Lakes, NJ, USA) and VACUETTE citric acid, sodium phosphate, dextrose, and adenine tubes (Greiner Bio‐One; Kremsmünster, Austria). These tubes were transported by courier to the laboratory, where bacterial and viral infection assays were performed by personnel blinded to the participants' names, ages, and group assignments. The bacterial assays were Borrelia burgdorferi B31 reference strain full antigen (enzyme‐linked immunosorbent spot assay; ELISpot); B. burgdorferi outer surface protein (Osp) peptide mix (OspA from B. burgdorferi sensu stricto, Borrelia afzelii, Borrelia garinii + OspC (ELISpot)); B. burgdorferi lymphocyte function‐associated antigen‐1 (LFA‐1; ELISpot); Chlamydia pneumoniae immunoglobulin A (IgA) and immunoglobulin G (IgG) (enzyme‐linked immunosorbent assay; ELISA); Chlamydia trachomatis IgA and IgG (ELISA); M. pneumoniae IgA and IgG (ELISA); Yersinia spp. IgA and IgG (ELISA); and Ehrlichia chaffeensis and Anaplasma phagocytophilum (ELISpot). The viral assays used were coxsackievirus A7 IgA and IgG (immunofluorescence); coxsackievirus B1 IgA and IgG (immunofluorescence); and echovirus IgA and IgG (immunofluorescence). The commercial assay kits used for the bacterial and viral assays are listed in Table 1. It should be noted that serum IgA detected in this study primarily reflects systemic immune exposure rather than mucosal immunity alone. While secretory IgA (dimeric form) predominates on mucosal surfaces, monomeric IgA circulates in the serum and can serve as a marker of systemic immune activation following infections originating at the epithelial barriers [15, 16, 17]. Therefore, elevated serum IgA levels against respiratory pathogens, such as C. pneumoniae and M. pneumoniae, may indicate prior infection, although they do not distinguish acute from chronic exposure. Similarly, IgG antibodies reflect cumulative exposure and typically persist for months to years after the resolution of an active infection. These caveats apply equally to viral serology: antibody detection indicates immune exposure rather than active infection, with IgG antibodies potentially persisting for months to years post‐infection.

Table 1.

Commercial assay kits used for bacterial and viral testing.

Pathogen assay	Assay type	Manufacturer	Catalog number
Borrelia burgdorferi strain B31	ELISpot	Autoimmun Diagnostika GmbH (AID), Strassberg, Germany	ELSP5905X
B. burgdorferi Osp	ELISpot	Autoimmun Diagnostika GmbH (AID), Strassberg, Germany	ELSP5946X
B. burgdorferi LFA‐1	ELISpot	Autoimmun Diagnostika GmbH (AID), Strassberg, Germany	ELSP5941X
Chlamydia pneumoniae IgA	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2192‐9601A
C. pneumoniae IgG	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2192‐9601G
Chlamydia trachomatis IgA	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2191‐9601A
C. trachomatis IgG	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2191‐9601G
Mycoplasma pneumoniae IgA	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2202‐9601A
M. pneumoniae IgG	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2202‐9601G
Yersinia spp. IgA	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2173‐9601A
Yersinia spp. IgG	ELISA	Euroimmun, Revvity Inc., Waltham, MA	EI 2173‐9601G
Ehrlichia chaffeensis & Anaplasma phagocytophilum	ELISpot	Autoimmun Diagnostika GmbH (AID), Strassberg, Germany	ELSP5981X
coxsackievirus A7 IgA	Immunofluorescence	Euroimmun, Revvity Inc., Waltham, MA	FI 2730‐1010‐4A
coxsackievirus A7 IgG	Immunofluorescence	Euroimmun, Revvity Inc., Waltham, MA	FI 2730‐1010‐4G
coxsackievirus B1 IgA	Immunofluorescence	Euroimmun, Revvity Inc., Waltham, MA	FI 2730‐1010‐4A
coxsackievirus B1 IgG	Immunofluorescence	Euroimmun, Revvity Inc., Waltham, MA	FI 2730‐1010‐4G

2.4. Statistical Analyses

Generalized linear modeling was carried out, with executive planning latency as the dependent variable, using backward selection with α = 0.05. Given that the numerical codomain was, in practice, (0, ∞), the natural logarithm was used as the link function for a gamma family. Sex and group were entered into the model as independent variables. Age was entered directly as a covariate, as were the ELISpot and ELISA results. The immunofluorescence assay results were entered as reciprocally transformed data and were positively monotonic with the corresponding viral IgA and IgG assay results. The chi‐square test was used to compare the final model (H₁) to the null model (H₀). Case‐wise diagnostic testing was performed using a standardized residual > 3, and deviance and Pearson goodness‐of‐fit tests were used to assess the adequacy of fit. These statistical analyses were performed using R version 4.5.1, JASP 0.19.3, and the R packages stats, statmod (Statistical Modeling), and emmeans (Estimated Marginal Means, also known as Least‐Squares Means) [18, 19, 20, 21, 22]. Graphical plotting was performed using ggplot2 software [23].

3. Results

The cohort consisted of 53 subjects, two of whom were male, with a mean (standard error) age of 48.9 (1.5) years. All subjects were of the same ethnicity, namely white Caucasians. The mean executive planning latency was 11,328 (562) ms.

The mean planning latency in patients with fibromyalgia (11,985 (915) ms) was not significantly different from that in the control group (10,424 (671) ms; p = 0.17). This difference was not significant after adjusting for age and infectious markers in a multivariate generalized linear model (GLM), suggesting that the observed prolongation of latency was better explained by age and markers of microbial exposure than by fibromyalgia diagnosis.

The GLM was highly significant (χ ² = 27.76, df = 5, p = 0.0001). The goodness of fit of the final model relative to the saturated model was excellent (deviance D = 3.781, df = 47, p = 1; Pearson goodness‐of‐fit statistic based on Pearson residuals = 3.734, df = 47, p = 1). No influential cases were found (standardized residuals > 3). The quantile–quantile (Q–Q) plot of the standardized deviance residuals is shown in Figure 1.

Figure 1.

Q–Q plot of standardized deviance residuals.

A significant positive correlation was observed between age and planning latency across the entire cohort (Spearman's r _s = 0.394, p = 0.003). When stratified by group, the correlations remained positive in both controls (r _s = 0.368, p = 0.059) and patients with fibromyalgia (r _s = 0.410, p = 0.034), as illustrated in Figure 2.

Figure 2.

Planning latency versus age by group. Scatterplot showing individual observations (n = 53) stratified by fibromyalgia status. LOESS smoothing curves with 95% confidence intervals (shaded regions) illustrate the relationship within each group. Overall, r _s = 0.394 (p = 0.003); control group r _s = 0.368 (p = 0.059); fibromyalgia group r _s = 0.410 (p = 0.034).

Exploratory post hoc analysis revealed a significant positive correlation between assay results for E. chaffeensis/A. phagocytophilum and those for B. burgdorferi (r _s = 0.433, p = 0.001). Figure 3 illustrates the observed and predicted median latency values from the final gamma GLM with 95% confidence intervals.

Figure 3.

Observed versus predicted median planning latency from the final gamma generalized linear model. Each point represents one subject (n = 53). The dashed gray line shows the line of identity, representing perfect prediction. The solid diagonal line shows the linear regression fit through the observed data, with shaded 95% confidence band. Points cluster closely around the line of identity, indicating good model fit (pseudo R ² = 0.40; see text for statistics).

Five independent variables, together with the intercept, were statistically significant. Of these, age, B. burgdorferi LFA‐1, and the assays for E. chaffeensis and A. phagocytophilum were the most significant. The details are presented in Table 2.

Table 2.

Significant model coefficients (omitting the intercept), in order of statistical significance, with their 95% confidence intervals.

Independent variable	Regression coefficient (β)/ln(ms)	Standard error/ln(ms)	|t|	p	95% confidence interval of coefficient
					Lower bound/ln(ms)	Upper bound/ln(ms)
Age	0.0155850	0.0035747	4.3598746	0.0000704	0.0086395	0.0225268
Borrelia burgdorferi LFA‐1	0.1237321	0.0379555	3.2599250	0.0020760	0.0509069	0.1980777
Ehrlichia chaffeensis and Anaplasma phagocytophilum	−0.1201517	0.0382948	3.1375477	0.0029393	−0.1915913	−0.0467751
Mycoplasma pneumoniae IgA	0.1815919	0.0817634	2.2209434	0.0312108	0.0300410	0.3405268
Chlamydia pneumoniae IgG	−0.1130832	0.0543394	2.0810521	0.0429033	−0.2236863	−0.0022612

Note: β‐coefficients represent change in ln(latency in ms) per unit increase in predictor. |t| denotes the absolute value of the Wald test statistic from the gamma generalized linear model. Standard errors and 95% confidence intervals are expressed on the same logarithmic scale. Exponentiating coefficients yields incidence rate ratios.

4. Discussion

This study supports the hypothesis that frontal lobe executive function, as indexed by prefrontal cortical planning latency, is associated with serological evidence of exposure to the bacterial genera Borrelia, Ehrlichia/Anaplasma, Mycoplasma, and Chlamydia, as well as with age. Associations with age are first considered.

While there have been few adult human studies on prefrontal cortical planning latency and aging, the present findings are consistent with reports on synaptic aging in the prefrontal cortex of non‐human primates [24]. For example, synaptic loss occurs in area 46 of the rhesus monkey brain [25, 26]. Indeed, this region corresponds to the dorsolateral prefrontal cortex, which is implicated in planning performance in look‐ahead puzzles such as the Tower of London [27]. The interaction between prefrontal cortical planning latency and aging may explain why studies on Alzheimer's disease (AD) tend not to show a significant increase in latency. However, when a young control group was included, the increased latency could be unmasked. For example, in their study of normal aging and AD, Coubard and colleagues reported that the latency in the Tower of London task was 1.23 s (standard deviation 0.69 s) in their AD cohort (mean age 78 years) and 1.11 (0.84) s in a control group matched for age [28]. However, the corresponding latency in the younger control cohort (mean age: 25 years) was significantly lower, at 0.68 (0.39) s [28].

It is important to emphasize that the present findings demonstrate an association rather than causation. The hypothesis that chronic bacterial exposure contributes to cognitive aging and dementia risk remains speculative and requires longitudinal validation. Several mechanisms have been proposed to link peripheral infections to frontal lobe dysfunction. Systemic inflammation can trigger changes in blood–brain barrier (BBB) permeability, allowing cytokines and potential microbial products to enter the central nervous system parenchyma [29, 30, 31]. Peripheral immune activation may induce microglial priming, leading to an exaggerated neuroinflammatory response to subsequent challenges [32, 33, 34, 35, 36]. Furthermore, certain pathogens may directly invade neural tissues or produce molecular mimics that trigger autoimmune responses against neuronal proteins [9, 37]. Nevertheless, alternative explanations—including reverse causation (e.g., age‐related immune senescence increasing susceptibility to infection), unmeasured confounders, or artifactual associations arising from the small sample size—cannot be ruled out. Therefore, the present results should be interpreted as preliminary evidence supporting further investigation, rather than as definitive proof of an infectious etiology for cognitive decline.

The strong relationship between prefrontal cortical planning latency and serological evidence of Borrelia exposure reported in this study is the first such association to be documented. This is consistent with the findings of several studies that associate frontal lobe disorders with Lyme disease. For example, in 1995, Waniek et al. reported a case of fatal Lyme disease associated with progressive frontal‐type dementia [38]. Nineteen years later, Blanc et al. described a cohort of 20 patients with dementia and positive intrathecal anti‐Borrelia antibodies [39]. Four (20%) patients in this cohort showed progressive worsening of dementia with a final diagnosis of AD [39]. While increased prefrontal cortical planning latency is unlikely to occur in patients with AD compared with age‐matched controls, for the reasons given above, it seems plausible that our results may be consistent with an association between prior Borrelia exposure and AD. This is consistent with the autopsy results reported by MacDonald et al. and Miklossy et al. In 1987, MacDonald and Miranda reported the direct visualization of Borrelia spirochetes in autopsy cerebral tissue derived from a patient who died of AD [40]. In due course, MacDonald went on to hypothesize that the characteristic plaques of AD may be derived from cystic forms of Borrelia [41, 42]. Similarly, in 1993, Miklossy reported that in 14 cases of AD, spirochetes were found in autopsy cerebral tissue as well as in blood and cerebrospinal fluid, but in none of 13 age‐matched controls [43]. These spirochetes were identified as B. burgdorferi sensu stricto [44]. Miklossy's group went on to show that in vitro exposure of mammalian neuronal, astrocytic, and microglial cells to B. burgdorferi for 2 to 8 weeks was associated with the appearance of morphological structures similar to AD plaques with similar β‐amyloid immunoreactivity [45].

The BBB restricts the direct passage of many pathogens into the central nervous system; however, several mechanisms allow for immune‐mediated effects on frontal lobe function. As mentioned above, Borrelia invades the cerebral endothelium and produces spirochetes in the cerebrospinal fluid [40, 43]. Some viruses may also enter via infected immune cells (the “Trojan horse” mechanism) or by disrupting tight junctions through inflammatory signaling [46, 47, 48]. Other bacteria, including C. pneumoniae and M. pneumoniae, may exert indirect effects via systemic inflammatory cytokine cascades that activate microglia without requiring direct microbial entry into the parenchyma and may also enter the brain via infection of the olfactory neuroepithelium [9, 49, 50]. Serum IgA and IgG antibodies reflect cumulative exposure and immune activation rather than necessarily indicating active central nervous system infection. These considerations temper the interpretation of the present findings as suggestive of peripheral immune dysregulation affecting frontal‐lobe executive function rather than of active neuroinfection.

The negative coefficient for the variable indicating serological reactivity to E. chaffeensis and A. phagocytophilum was likely attributable to the relationship between this reactivity and exposure to B. burgdorferi. A post hoc analysis showed a highly significant positive correlation between reactivity to E. chaffeensis and A. phagocytophilum and exposure to B. burgdorferi (r _s = 0.433, p = 0.001).

The two final statistically significant predictive variables in the model were M. pneumoniae IgA and C. pneumoniae IgG levels. Although exposure to M. pneumoniae has not been specifically linked to frontal lobe dysfunction, such infections have been noted in patients with long COVID [51]. Further research is required to determine whether M. pneumoniae contributes to the cognitive disorders observed in long COVID in such patients. The coefficient of C. pneumoniae IgG in the model was negative. At first glance, this may appear to contradict earlier findings from the Northern Manhattan Study that the composite serologic index, known as the infectious burden index, which includes C. pneumoniae IgG, is associated with decreased cognitive performance (measured using the Mini‐Mental State Examination and the modified Telephone Interview for Cognitive Status), diminished executive function at baseline (Color Trails and Odd‐Man‐Out), and memory decline over several years [52, 53]. However, the negative coefficient in the present study is consistent with the more recent study by Roberts et al., based on a subset of the Northern Manhattan Study subjects, which showed that C. pneumoniae exposure was not associated with impaired executive function but rather with impaired performance on language tests [54].

It is interesting briefly to speculate about the two implications of these findings for B. burgdorferi. First, the advantage of the relationship between the prefrontal cortical planning latency and serological evidence of Borrelia exposure is that, as part of its enzootic cycle, this zoonotic pathogen is maintained in mammalian hosts [55]. Perhaps an increase in prefrontal cortical planning latency in the host with prior exposure is associated with a greater likelihood of arthropod vectors (including their newly hatched larvae) attaching to the mammal and acquiring B. burgdorferi via blood meals. Second, if frontal‐type dementia and AD are related to prior Borrelia exposure or persistent infection, antibiotic treatment may have therapeutic value; there is some anecdotal evidence that this might indeed be the case. In the aforementioned case report by Waniek et al., antibiotic treatment was associated with a transient improvement in progressive frontal lobe dementia associated with Lyme disease; discontinuation of this treatment was followed by relapse [38]. Furthermore, in the above study by Blanc et al., involving 20 patients with both dementia and positive intrathecal anti‐Borrelia antibodies, dementia in seven of the patients who were treated with the antibiotic ceftriaxone either remained stable or showed mild improvement [39]. While these historical cases suggest a potential therapeutic avenue, it should be noted that the present study did not provide evidence for treatment efficacy in the studied cohort, and these findings should be viewed strictly as hypothesis‐generating.

Despite these mechanistic insights, this study has some limitations. The first concerns the sample size. Although the sample size was relatively small, the final GLM demonstrated robust stability. Multicollinearity diagnostics revealed negligible variance inflation factors (range: 1.06–1.61), and no influential outliers were detected. The retained predictors were statistically significant (p < 0.05), and the deterministic nature of the backward elimination procedure ensured that the model structure was fully reproducible. Another limitation of the present study is the relatively restricted panel of viruses tested. While coxsackievirus A7, coxsackievirus B1, and echovirus were included, other prevalent viruses, such as herpes simplex viruses 1 and 2, cytomegalovirus, and Epstein‐Barr virus, were not assessed. Future studies should include an expanded serological panel that includes these viruses. Another limitation is the cross‐sectional nature of this study. Because exposure and outcome were assessed concurrently, it is not possible to rule out the admittedly biologically implausible possibility that latency changes precede infection. Finally, the fact that the cohort was drawn from a single clinical site and that all subjects were of a single ethnicity, namely white Caucasians, limits the generalizability of the findings to other settings or population groups.

Author Contributions

Basant K. Puri: conceptualization, methodology, data analysis, writing – original draft preparation, writing – review and editing, supervision. Rosemarie Preyer: conceptualization, data collection, data curation, writing – review and editing. Gary S. Lee: conceptualization, data collection, data curation, writing – review and editing. Lucy Monaghan: conceptualization, data collection, data curation, writing – review and editing. Armin Schwarzbach: conceptualization, data collection, data curation, writing – review and editing, supervision.

Ethics Statement

This study received approval from the CPUREC Research Ethics Committee (country: UK; approval no. 001/18). Each participant gave informed written consent to participate. The study was performed in accordance with the Declaration of Helsinki.

Conflicts of Interest

The authors declare no conflicts of interest.

Use of Large Language Models, AI and Machine Learning Tools

Not used.

Data Availability Statement

The raw dataset used in this study is available in an anonymized form from the corresponding author upon reasonable request.