Study sample characteristics and symptom burden are shown in Table 1 and Fig. 2. As depicted in Table 1, most of the subjects were females. The mean age was comparable between the groups and did not differ significantly. The PACS patients were free from comorbidities except for a subset of PACS-POTS (40%) who declared depression/anxiety. For this reason, a larger proportion of patients in this group were on sedative and hypnotics. The majority (60%) of patients in the PACS + POTS group were on ivabradine. As expected, the increase in heart rate following head-up tilt test was higher in the PACS + POTS group. However, the mean heart rate over 24 h did not differ between the two PACS groups. The symptom burden between the two PACS groups did not differ in major aspects as MAPS score was similar (Fig. 2A) and the proportion of patients reporting common symptoms (13 of 26 most common are shown) in PACS at the day of sampling was, overall, comparable with some exceptions (Fig. 2B).
Plasma proteomic profiling
In the differential expression analysis, we found 204 proteins upregulated in the PACS + POTS group when compared to healthy controls (Fig. 3A), and 201 proteins upregulated in the PACS-POTS group (Fig. 3B). As POTS has been described as a distinct phenotype of PACS, we assessed differences in the proteome between the two PACS groups. Interestingly, there were no significantly altered proteins between the groups, and there were also few proteins with a log2-fold change > 1 (Fig. 3C). Next, we performed principal component analysis (PCA) of our proteomics dataset, which showed clear distinction between healthy controls and both PACS groups, but no major distinction between the two PACS groups (Fig. 4A). Among the upregulated proteins in both PACS groups vs. healthy controls, we observed more than 90% overlap between the PACS groups, with only a few proteins uniquely upregulated in each of the groups (Fig. 4B). Most investigated proteins were upregulated, while only 9 proteins in the PACS + POTS group and 6 in the PACS-POTS group were downregulated (Fig. 4C). All proteins with associated log2-fold changes, raw p-values and adjusted p-values are listed in Supplementary File 1.
Finally, an interactive heatmap with unsupervised hierarchical clustering for the significantly different proteins with log2-fold change > ± 2 was used to further depict the differentially expressed proteins in healthy controls and PACS patients. This resulted in a clear cluster of healthy controls, but no major clusters comparing PACS + POTS with PACS-POTS, further supporting the similarities in the plasma proteome of PACS patients regardless of concomitant POTS (Fig. 4D).
To further dissect the dysregulated outcomes across the groups and to find possible small differences that were undetected due to lack of power, we chose to look whether there were any uniquely up- or downregulated proteins when comparing PACS + POTS and PACS-POTS groups, respectively, with healthy controls. This analysis showed only minor differences between the PACS + POTS and PACS-POTS groups, as there were 11 and 8 uniquely upregulated proteins for the PACS + POTS and PACS-POTS groups, respectively, as well as 6 and 3 uniquely downregulated proteins for the PACS + POTS and PACS-POTS groups, respectively (Fig. 4B,C). These proteins are listed in Supplementary File 1. Taken together, PACS patients had a clearly dysregulated plasma proteome with a substantial proportion (almost 30%) of dysregulated proteins compared to healthy controls, but there were no major differences between PACS patients with or without POTS.
Protein functionality analysis
We next sought to gain knowledge about the potential function of the plasma protein dysregulation seen in PACS. As there was no difference between the PACS + POTS and PACS-POTS groups, we combined the data from these groups and analyzed them vs. healthy controls. Gene ontology (GO) pathway enrichment analysis led to a large amount of significantly altered pathways, including processes involved in hemostasis, inflammation, amino acid metabolism and apoptosis (Fig. 5A–D). Network plots were then used to depict which specific proteins were altered in each pathway. Interestingly, for hemostasis and coagulation, we found that key proteins such as SERPINE1 (also known as plasminogen activator inhibitor 1) were strongly upregulated in PACS patients, implying increases in clotting susceptibility (Fig. 5A). Moreover, CCL5, a chemokine often expressed by T-cells involved in inflammation, was upregulated in PACS patients (Fig. 5B), implying a key role of T-cells in the hyperinflammatory response often seen in PACS patients. Finally, amino acid metabolism seemed to be increased with expansion of key amino acid metabolic enzymes (Fig. 5C), whilst key apoptotic markers such as Caspase 3 and DIABLO were upregulated in PACS patients (Fig. 5D), indicating activation of apoptotic processes.
Plasma cytokine profiling
We next analyzed cytokine profiles since both acute COVID-19 and PACS are associated with hyperinflammation, and our proteomics data suggested dysregulation in T-cell responses (Fig. 5B). Differential expression analysis showed similar patterns in cytokine dysregulation in both PACS groups compared to healthy controls, but no significant difference between PACS + POTS and PACS-POTS (Fig. 6A–C). Like the proteomics data, initial PCA analysis displayed clear difference between healthy controls and both PACS groups, but no major difference between PACS + POTS and PACS-POTS (Fig. 7A). There was a large overlap in the upregulated cytokines in both PACS groups vs healthy controls (Fig. 7B), with only one uniquely upregulated cytokine in each of the groups (Supplementary File 2). However, there were no downregulated cytokines in any of the groups (Fig. 7C), further strengthening the distinct pro-inflammatory signature of the PACS population. Interestingly, several of the commonly upregulated cytokines such as vascular endothelial growth factor-A (VEGF-A) and epidermal growth factor (EGF) play a role in angiogenesis. Unsupervised hierarchical clustering reaffirmed the results of the differential expression analysis by showing clear clustering when healthy controls were compared with PACS, but no pronounced clustering when PACS + POTS were compared with PACS-POTS (Fig. 7D). All cytokines with associated log2 fold changes, raw p-values and adjusted p-values are listed in Supplementary File 2.
Plasma sphingolipid profiling
It has previously been shown that sphingolipids, in particular ceramides, represent potential drivers of cardiovascular disease through various actions, including accumulation in tissues relevant for cardiovascular disease such as the vasculature and the heart with adverse metabolic regulation17. A proposed key-mechanism by which sphingolipids act as mediators of cardiovascular disease is through their pleiotropic effect as inhibitors of nitric oxide formation and increased production of reactive oxygen species17,18. Since our previous work showed that PACS + POTS patients frequently presented with microvascular dysfunction (MVD), we hypothesized that sphingolipid-related dysregulation could accompany POTS and/or MVD in PACS patients. Using a targeted LC–MS based lipidomic approach specifically for sphingolipid detection, we analyzed a total of 88 sphingolipids in our cohort.
Again, we analyzed the data to identify differentially expressed lipids between the groups. Compared to healthy controls, we found 16 and 19 dysregulated lipids out of 88 in the PACS + POTS and PACS-POTS groups, respectively (Fig. 8A,B). Moreover, no significant differences were detected between PACS-POTS and PACS + POTS groups (Fig. 8C). A few of the most upregulated sphingolipids in both PACS groups are sphingosine (d18:1) and sphingosine (d16:1) which are the basis of sphingolipids, and sphingosine 1-phosphate (S1P) with diverse bioactive actions including critical roles in the immune system, blood pressure and endothelial function19,20. Interestingly, PCA analysis once again showed a clear cluster between healthy controls and PACS patients but failed to differentiate PACS-POTS from PACS + POTS group, further highlighting the similarities in plasma composition in both PACS-POTS and PACS + POTS groups (Fig. 9A). There were only two unique significantly upregulated lipids in both groups, as well as 2 and 5 unique significantly downregulated lipids in PACS + POTS and PACS-POTS groups, respectively (Fig. 9B,C). Similar to the plasma proteins and cytokines, unsupervised hierarchical clustering resulted in a clear cluster of healthy controls, but no major clusters among PACS-POTS and PACS + POTS patients (Fig. 9D). All sphingolipids with associated log2-fold changes, raw p-values and adjusted p-values and the uniquely dysregulated sphingolipids are listed in Supplementary File 3. None of the sphingolipids fell below the LOD.
Collectively, our data show clear dysregulation of the molecular plasma profile in PACS patients. This is most pronounced in cardiometabolic proteins but also in cytokines and sphingolipids compared to healthy controls (Fig. 1). Interestingly, all analyses showed a similar picture with absence of major differences between PACS patients with and without POTS. The common denominator for the outcomes of all molecular analysis is the up-regulation of pro-inflammatory proteins/molecules of importance for maintaining hemostasis, metabolism, clot formation and vascular function.