Deep phenotyping of post-infectious myalgic encephalomyelitis/chronic fatigue syndrome

Abstract

Post-infectious myalgic encephalomyelitis/chronic fatigue syndrome (PI-ME/CFS) is a disabling disorder, yet the clinical phenotype is poorly defined, the pathophysiology is unknown, and no disease-modifying treatments are available. We used rigorous criteria to recruit PI-ME/CFS participants with matched controls to conduct deep phenotyping. Among the many physical and cognitive complaints, one defining feature of PI-ME/CFS was an alteration of effort preference, rather than physical or central fatigue, due to dysfunction of integrative brain regions potentially associated with central catechol pathway dysregulation, with consequences on autonomic functioning and physical conditioning. Immune profiling suggested chronic antigenic stimulation with increase in naïve and decrease in switched memory B-cells. Alterations in gene expression profiles of peripheral blood mononuclear cells and metabolic pathways were consistent with cellular phenotypic studies and demonstrated differences according to sex. Together these clinical abnormalities and biomarker differences provide unique insight into the underlying pathophysiology of PI-ME/CFS, which may guide future intervention.

Fig. 1. Patient recruitment and characteristics.

a Diagram showing the procedure followed to recruit adjudicated Post-infectious Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (PI-ME/CFS) and matched healthy volunteers (HV) and measurements undertaken for deep phenotyping of the cohorts. The number of participants at each stage of recruitment are noted. b Comparisons of age, sex, and BMI distribution in HV (blue; n = 21 independent participants) and PI-ME/CFS (red; n = 17 independent participants) using unadjusted two-sided t-tests for independent samples. c Distribution of the response of HV (blue; n = 21 independent participants) and PI-ME/CFS (red; n = 17 independent participants) to the indicated patient reported outcome questionnaires. Group comparisons performed using unadjusted two-sided Mann–Whitney-U tests. CSF cerebrospinal fluid, PBMC peripheral blood mononuclear cell, BMI body mass index, PROMIS Patient-Reported Outcomes Measurement Information System, PHQ-15 Patient Health Questionnaire – 15, MASQ Multiple Ability Self-Report Questionnaire. Figure 1A created with Biorender.com. Source data are provided as a Source Data file.

Fig. 2. Diminished heart rate variability measures are consistent with decreased parasympathetic activity in the PI-ME/CFS cohort compared to HV.

a Table of time and frequency domain heart rate variability measurements. Group comparisons for panel a were performed with unadjusted two-sided Mann–Whitney U tests. Box plots comparing HV (blue; n = 19 independent participants) and PI-ME/CFS (red; n = 14 independent participants) for (b) SDNNI (msec) (c) rMSSD (msec, p = 0.019, unadjusted two-sided t-test for independent samples with equal variance) (d) pNN50 (%, p = 0.017, unadjusted two-sided Mann–Whitney U test) (e) lnHF(ms²) (f) lnLF (ms²). Box plots depict the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values g: Mean heart rate of HV (blue; n = 20 independent participants) and PI-ME/CFS (red; n = 13 independent participants) of 5-min segmented intervals over a 24-h period graphed over 24-h period. Error bars represent ±SE for each 5-min time block for each group. Note HV graph (blue) demonstrates fluctuations throughout the day with subject heart rates displaced slightly higher, suggesting increased sympathetic activity. Similarly, the typical sinusoidal drop in heart rate over sleeping hours is diminished in subjects also suggesting diminished parasympathetic and/or increased sympathetic activity. h Box plot of baroreflex-cardiovagal gain as measured by mean baroslope (ms/mmHg). HVs (blue; n = 19 independent participants) and PI-ME/CFS (red; n = 16 independent participants) are compared using an unadjusted two-sided t-test for independent samples with equal variance (p = 0.015). Box plot H depicts the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values. SDNNi standard deviation of the average NN intervals for each 5 min segment of a 24 h HRV recording, rMSSD root mean square of successive differences between normal heartbeats, pNN50 proportion of NN50 divided by the total number of NN (R-R) intervals, HF high frequency, LF low frequency, SD1 standard deviation of Poincaré plot of RR intervals perpendicular to the line-of-identity, SD2 standard deviation of the Poincaré plot of RR intervals along the line-of-identity. Source data are provided as a Source Data file.

Fig. 3. Impaired effort measures and motor performance were observed in PI-ME/CFS cohort compared to HV.

a, b Effort-Expenditure for Rewards Task: a Probability of choosing the hard task is significantly more in HV (blue, n = 16 independent participants) compared with PI-ME/CFS (red; n = 15 independent participants) at the start of and throughout the task. The Odds Ratio for the probability of choosing the hard task at the start of the task is 1.65 [1.03, 2.65], p = 0.04 using Fisher’s Exact test. The lines are the curvilinear fits and the error bands are the confidence intervals. Decline rates (i.e. response to fatigue) between the groups is similar as the trial progresses. b Button press rates for easy (right) and hard (left) tasks as the trial progresses is shown for HV (blue) and PI-ME/CFS (red) participants. The lines are the linear fits and the shaded error bands are the confidence intervals. The decline in button press rate over time during the easy tasks in PI-ME/CFS did not impact easy task performance, which is supportive of pacing in PI-ME/CFS participants. c–e Grip Strength test: Box plots of (c) maximum grip force of HV (blue; n = 20 independent participants) and PI-ME/CFS (red; n = 16 independent participants) and (d) time to failure of HV (blue; n = 18 independent participants) and PI-ME/CFS (red; n = 16 independent participants), unadjusted two-sided t-test for independent samples with equal variance, p = 0.0002. Correlation between time to failure and (e) proportion of hard task choices in HV (blue; n = 15 independent participants) and PI-ME/CFS (red; n = 14 independent participants). For figure e, the relationship between indicated variables in x and y axis were fitted by linear regression in each group. Linear regression t-tests were used to determine non-zero slope. Exact p values of the correlations are presented on the graph. For box plots c and d, boxes depict the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values. Source data are provided as a Source Data file.

Fig. 4. Impaired sustained effort and motor performance was observed in PI-ME/CFS cohort compared to HV.

a–e Repetitive Grip Strength test: a Grip force normalized to maximum voluntary contraction (MVC) in the first block, the last block prior to fatigue onset, and the first three blocks after fatigue onset in HV (blue) and PI-ME/CFS cohorts (red). A significant grip force difference was noted between the groups (−1.2 ± 4 versus −6.4 ± 4 kilogram-force units, t(12) = 2.46, p = 0.03). The number of non-fatigued grip test blocks of HV (blue) and PI-ME/CFS (red) participants is also displayed. b Slope of the Dimitrov index across the first block (b₁), the last block prior to fatigue onset (b_n), and the first three blocks after fatigue onset (f₁, f₂, and f₃) in HV (blue; n = 6 independent participants) and PI-ME/CFS (red; n = 8 independent participants) patients. A significant difference was noted between the groups (0.2 ± 0.5 versus −0.43 ± 0.3, t(12) = 3.2, p = 0.008). c Mean and standard error of the motor evoked potential of HV (blue; n = 6 independent participants) and PI-ME/CFS (red; n = 8 independent participants) participants spanning the last five grip test blocks prior to fatigue onset. The amplitudes of the MEPs of HVs significantly decreased over the course of the task while the amplitudes of the MEPs of PI-ME/CFS participants significantly increased (−0.13 ± 0.2 versus 0.13 ± 0.2 MEP units; t(12) = 2.4, p = 0.03 D. Brain regions where Blood Oxygen Dependent (BOLD) signal decreased over grip strength blocks in PI-ME/CFS patients and increased over grip strength blocks in HVs. e Brain activation of the regions depicted in d measured in the blocks of four over the course of the experiment in HV (blue; n = 10 independent participants) and PI-ME/CFS (red; n = 8 independent participants) cohorts. For e, a two-way ANOVA was run where F(3,45) = 5.4 with a voxel threshold of p ≤ 0.01, corrected for multiple comparisons p ≤ 0.05, k > 65 (p-values were 0.976, 0.43, 0.02 (*), and 0.02 (*) for blocks 1 to 4 respectively). Source data are provided as a Source Data file.

Fig. 5. Impaired cardiopulmonary performance was observed in PI-ME/CFS cohort compared to HV.

Cardiopulmonary Exercise Test (CPET) a–h: Box plots of (a) peak power attained during CPET for HV (blue; n = 9 independent participants) and PI-ME/CFS (red; n = 8 independent participants) participants, (b) peak respiratory rate with the error bars representing ±SD at each work rate, (c) peak heart rate (unadjusted two-sided t-test for independent samples with unequal variance, p = 0.072), (d) Peak VO2 (unadjusted two-sided t-test with equal variance, p = 0.002), (e) Percent of predicted VO2 achieved of HV (blue; n = 9 independent participants) and PI-ME/CFS (red; n = 8 independent participants) participants (unadjusted two-sided t-test for independent samples with equal variance, p = 0.004), and (f) heart rate reserve for HV (blue; n = 9 independent participants) and PI-ME/CFS (red; n = 8 independent participants) participants (unadjusted two-sided t-test for independent samples with equal variance, p = 0.011). g Heart rate as a function of % total CPET time depicted as expected for age and gender of HV (n = 9 independent participants) and PI-ME/CFS (n = 8 independent participants) (gray solid and dashed lines, respectively, unadjusted two-sided Mann–Whitney U test for independent samples). Mean heart rate responses from the CPET are depicted for HV (blue line) and PI-ME/CFS (red line). A significant difference was observed for the heart rate slope between PI-ME/CFS and expected (0.70 ± 0.27 versus 1.05 ± 0.12, p = 0.014), but not for HV and expected (1.03 ± 0.16 versus 1.08 ± 0.13; p = 0.479). The deviation from expected relation reflects chronotropic incompetence in the PI-ME/CFS group. h Box plot of VO2 at the anaerobic threshold (AT) in HV (blue; n = 9 independent participants) and PI-ME/CFS (red; n = 8 independent participants) using an unadjusted two-sided t-test for independent samples with equal variance (p = 0.024). For box plots a, c–f, and h boxes depict the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values. Source data are provided as a Source Data file.

Fig. 6. Differential catecholamines and tryptophan pathway metabolites levels in PI-ME/CFS patients cerebrospinal fluid.

a–c Box plot of the indicated neurotransmitters on y axis in HV (blue; n = 21 independent participants) and PI-ME/CFS (red; n = 16 independent participants) in (a) unadjusted two-sided Mann–Whitney U test (p = 0.021), (b), unadjusted two-sided Mann–Whitney U test (p = 0.025), and (c) unadjusted two-sided t-test for independent samples with equal variance (p = 0.006). For box plots a–c boxes depict the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values. Correlation between cerebrospinal fluid norepinephrine (NE) and (d) time to failure on grip strength task in HV (blue; n = 18 independent participants) and PI-ME/CFS (red; n = 15 independent participants) or (e) proportion of hard task choices (i.e., effort preference) in HVs (blue; n = 14 independent participants) and PI-ME/CFS (red; n = 14 independent participants). f Correlation between cerebrospinal fluid dopamine and time to failure in HVs (blue; n = 17 independent participants) and PI-ME/CFS (red; n = 14 independent participants). g Correlation between cerebrospinal fluid DHPG and proportion of hard task choices (i.e., effort preference) in HVs (blue; n = 14 independent participants) and PI-ME/CFS (red; n = 14 independent participants). For figures d–g, the relationship between indicated variables in x and y axis were fitted by linear regression in each group. The exact p value of each regression is presented on the graph, linear regression t-test for nonzero slope. h PCA computed from all metabolites measured from the cerebrospinal fluid samples in the indicated groups. i Heatmap of statistically significant (false discovery rate adjusted p-value < 0.05) differentially expressed metabolites in the indicated groups on x axis and the metabolites labeled on y axis. Red: upregulated; Blue: downregulated. Supervised clustering of metabolites measured from the cerebrospinal fluid samples in (j) male cohorts and (k) female cohorts from PLSDA analysis. DHPG (S)−3,5-Dihydroxyphenylglycine, PLSDA Partial least square discriminant analysis, PC principal component. Source data are provided as a Source Data file.

Fig. 7. Flow cytometry demonstrates distinct perturbation in immune cell subpopulation in PBMCs.

a Boxplot of the B cell subset naïve (%) in HV (blue; n = 20 independent participants) and PI-ME/CFS (red; n = 17 independent participants) groups, unadjusted two-sided t-test for independent samples with equal variance (p = 0.037). b Boxplot of B cell switched memory in HV (blue; n = 20 independent participants) and PI-ME/CFS (red; n = 16 independent participants) groups, unadjusted two-sided t-test for independent samples with equal variance (p = 0.008). c Boxplot of the CD8 + T cell subset PD-1 (%), Mann–Whitney U test, exact p-value = 0.033. d Boxplot of the CD8 + T cell subset CD226 (%), unadjusted two-sided t-test for independent samples with equal variance (p = 0.055). The samples used for boxplot (d) were collected at a separate time point than the others boxplots; HV (blue; n = 7 independent participants) and PI-ME/CFS (red; n = 8 independent participants) groups. e Boxplot of CD8 + T cell CXCR5 (%), unadjusted two-sided t-test for independent samples with equal variance (p = 0.014). f Boxplot of CD8 + T cell naïve (%), unadjusted two-sided t-test for independent samples with equal variance (p = 0.016). Where indicated the plots shows the measurements from female and male cohorts. Measurements in PBMC samples are shown in shaded box and cerebrospinal fluid samples in open box. For box plots a–f boxes depict the median (horizontal line) within quartiles 1–3 (bounds of box). Whiskers extend to minimum and maximum values. Source data are provided as a Source Data file.

Fig. 8. Male and female cohorts have distinct perturbation in immune cell subpopulation and biological processes in PBMCs.

a, b PCA computed from all gene expression values with indicated groups: HV (blue), PI-ME/CFS (red) clusters or males (turquoise) and females (orange) highlighted for the indicated PCs. c Venn diagram showing common DE genes identified from male and female cohorts (p value < 0.05 as filter for DE genes using an unadjusted two-sided moderated t-test). DE analysis on the (d, e, h, i) male cohorts and (f, g, j, k) female cohorts. d, f PCA plots computed from DE genes shows robust clustering of samples based on the PI-ME/CFS status. e, g Volcano plots shows log transformed statistically significant (unadjusted p-values < 0.05 using an unadjusted two-sided moderated t-test) up (red) and down regulated (blue) genes in PI-ME/CFS male and female cohorts, respectively. h, j Heatmaps of a subset of T cell process genes in males and B cell related processes in females. i, k Pathway enrichment plots showing top 15 pathways for which the DE genes from male and female cohorts selected for. The top 15 pathways are labeled on the y axis and the color of the circle is scaled with –log-10 p-value and the size of the pathway circles inside the plot are proportional to the number of genes that overlapped with the indicated pathway. Fisher’s exact test was used in the ‘clusterProfiler’ R package to obtain the log-transformed p-values. Red color nodes: upregulated in PI-ME/CFS and blue color nodes: downregulated in PI-ME/CFS. DE differentially expressed. Source data are provided as a Source Data file.

Fig. 9. Male and female cohorts have distinct differential gene expression profiles in the muscle.

a, b PCA computed from all gene expression values in samples highlighted from the indicated groups: HV (blue) and PI-ME/CFS (red) or males (turquoise) and females (orange) for the indicated PCs. c Venn diagram showing common DE genes identified from male and female cohorts (DE genes are genes with p value < 0.05 using an unadjusted two-sided moderated t-test). DE analysis using an unadjusted two-sided moderated t-test on the (d, e, h, i) male cohorts and (f, g, j, k) female cohorts. d, f PCA plot computed from DE genes shows robust clustering of samples based on the PI-ME/CFS status. e, g Volcano plots shows log transformed statistically significant (unadjusted p-values < 0.05 using an unadjusted two-sided moderated t-test) up (red) and down regulated (blue) genes in PI-ME/CFS male and female cohorts, respectively. Pathway enrichment plot of DE genes (h) upregulated and (i) downregulated in PI-ME/CFS male cohort. Pathway enrichment plot of DE genes (j) upregulated and (k) downregulated in PI-ME/CFS female cohorts. The top 15 pathways are labeled on the y axis and the color of the circle is scaled with –log-10 p-value and the size of the pathway circles inside the plot are proportional to the number of genes that overlapped with the indicated pathway. Fisher’s exact test was used in the ‘clusterProfiler’ R package to obtain the log-transformed p-values. Red color nodes: upregulated in PI-ME/CFS and blue color nodes: downregulated in PI-ME/CFS. DE differentially expressed, PC principal component. Source data are provided as a Source Data file.

Fig. 10. Pathophysiology of PI-ME/CFS.

Diagram illustrates potential mechanisms and a cascade of events that lead to the development of ME/CFS after an infection. Exposure to an infection leads to concomitant and persistent immune dysfunction and changes in gut microbiome. Immune dysfunction affects both innate and adaptive immune systems that are sex dependent. We hypothesize that these changes are driven by antigen persistence of the infectious pathogen. These immune and microbial alterations impact the brain, leading to decreased concentrations of metabolites which impacts brain function. The catecholamine nuclei release lower levels of catechols, which impacts the autonomic nervous system and manifests with decreased heart rate variability and decreased baroreflex cardiovascular function, with downstream effects on cardiopulmonary capacity. Altered hypothalamic function leads to decreased activation of the temporoparietal junction during motor tasks, suggesting a failure of the integrative brain regions necessary to drive the motor cortex. This decreased brain activity is experienced as physical and psychological symptoms and impacts effort preferences, leading to decreased engagement of the motor system and decreases in maintaining force output during motor tasks. Both the autonomic and central motor dysfunction result in a reduction in physical activity. With time, the reduction in physical activity leads to muscular and cardiovascular deconditioning, and functional disability. All these features make up the PI-ME/CFS phenotype.