ABSTRACT
High prevalence of post‐acute sequelae of COVID‐19 (PASC/long COVID) has led to the determination of the pathophysiology of PASC. In a hospital‐based cross‐sectional survey, we assessed the clinical and immunological parameters in a set of 67 PASC and 81 recovered individuals from COVID‐19 (N‐PASC), to identify the immune biomarkers associated with the mechanistic cornerstone of this condition. PASC had higher chronic comorbidities, hospitalization, and ICU admissions during the COVID‐19 pandemic compared to the N‐PASC group. Though comparable SARS–CoV–2–specific antibodies were detected across the groups, their functionality (NAbs) was significantly higher in the N‐PASC. PASC patients exhibited features of immune dysregulation, characterized by altered coordination between cellular and humoral immune responses despite broadly comparable cytokine profiles and T‐cell responses. ROC analysis in hospitalized PASC and hospitalized N‐PASC subjects sampled at 1, 2, and 3 years post COVID‐19 supported the utility of IL‐6 as a biomarker for long‐term inflammatory activity. Higher FGF‐basic levels in the hospitalized PASC compared to non‐hospitalized PASC highlighted a potential association between elevated growth factor signaling and severity‐related immune dysregulation, tissue damage that may have utility as a biomarker candidate. Our analysis supports a dysregulated crosstalk between humoral and cellular immunity in PASC patients, which could be leading to inflammation and persistent clinical symptoms associated with this debilitating condition. In conclusion, comparable T‐cell and cytokine responses among long COVID and recovered individuals suggest that persistent symptoms are unlikely to be driven by impaired antiviral immunity, underscoring the potential role of immune dysregulation.
Keywords: IgG and NAbs, IL‐6 and FGF‐Basic, immune dysfunction, post‐acute sequelae of COVID‐19, T cell response
1. Introduction
The majority of the individuals infected with COVID‐19 recover completely from the acute phase of the disease. However, approximately 40%–45% of the COVID‐19 survivors suffer from a variety of persistent unresolved symptoms [1]. If the symptoms continue post 3 months after the initial SARS‐CoV‐2 infection with no other explanation, then the patients are referred to as long COVID [2]. The aftereffects of SARS‐CoV‐2 infection may impact a person's quality of life and ability to return to work. The clinical condition is commonly referred to as Long COVID or Post‐Acute Sequelae of COVID‐19 (PASC). The incidence of experiencing long COVID symptoms post‐infection is significantly higher in older adults and females compared to younger adults and men [3, 4]. The most common persistent symptoms are shortness of breath, exhaustion, loss of taste and smell, cognitive impairment, chest pain, and arthralgia [5].
Several lines of evidence indicate involvement of multi‐systemic humoral and cellular immune response during the acute phase of illness and post‐recovery from COVID‐19 infection [6]. Robust humoral (IgG and NT) and cellular (memory B and T cells) immune responses in the recovered individuals, indicative of protective immunity, are suggested to safeguard the recovered individuals from the circulating VoCs and VoIs, up to 8–9 months of recovery [7, 8, 9, 10].
Immune dysregulation is observed in individuals with residual post‐COVID syndrome (long COVID). Long COVID manifests as a diversity of symptoms affecting various organs, as seen in COVID‐19 patients [11]. T cells and persistent immune dysfunction are associated with Long COVID even after mild COVID‐19 [12]. Detailed clinical, virological, and immunological investigations would provide evidence to establish the drivers of immune dysfunction in long COVID and protective immune response post recovery/vaccination. Since persistent T cell activation and immune dysregulation are also associated with prolonged symptoms following viral infection, it is essential to study T cell responses in long‐term COVID‐19, which may indicate a potential link between immune response and long‐term COVID manifestations [12]. Certain immune cells, cytokines/chemokines, and host genetic factors have been identified to contribute to the host defense and recovery from COVID‐19 [13, 14]. Such evidence is crucial for identifying potential biomarkers and addressing the knowledge gaps in understanding the manifestation of long COVID.
With this background, the current study aimed to compare the clinical and immunological parameters of individuals with long COVID (PASC) with those who have recovered (N‐PASC) and identify the key factors/molecules associated with the possible manifestations of persistent post‐COVID syndrome.
2. Materials and Methods
2.1. Design and Setting
A hospital‐based cross‐sectional survey was conducted among the patients visiting the outpatient departments. The study was conducted in two government hospitals; BJ Medical College Hospital, Camp (a tertiary care centre) and Nimhan Hospital, Pashan (a primary care centre), both in Pune, Maharashtra, India.
2.2. Study Participants
Patients were recruited according to the WHO clinical case definition for “post‐COVID syndrome.” Individuals aged 18 years and above who were diagnosed with SARS‐CoV‐2 at least 3 months prior were included in the study. Assuming a prevalence of 43.4% of post‐COVID‐19 syndrome, we recruited 162 individuals for a level of significance of 5% (p < 0.05).
Data were collected through direct interviews with the eligible study participants. A pre‐tested questionnaire was used to collect data on social and demographic characteristics, co‐morbidity status, and severity of COVID‐19 infection, including the history of hospitalization and ICU admissions. Persistent or newly appeared symptoms lasting for more than 2 months without an alternate diagnosis were recorded along with their duration. Information on the current management strategy adopted for the post‐COVID‐19 symptoms was collected. Data regarding the COVID‐19 vaccination status were also collected. The participants were categorized into two groups: (a) PASC with symptoms of COVID‐19 and (b) Non‐PASC (N‐PASC), whose COVID‐19 symptoms had resolved/recovered from COVID‐19. Baseline characteristics and the immunological parameters were compared for these two groups.
For the analysis of the immunological parameters 148 study participants were considered, including 67 participants in the PASC group and 81participants in the N‐PASC group. The blood samples collected in K3/EDTA tubes were processed for PBMCs, and the plasma samples were stored at ‐80°C at the Indian Council of Medical Research‐National Institute of Virology (ICMR‐NIV), Pune, India. The experimental characterization of antibody and immune cell responses for all the study participants followed the same technique and used the same equipment. The various categories of the study subjects whose blood samples were analyzed are graphically represented in Figure 1.
Figure 1.
Graphical representation of characteristics of the study subjects.
The study was approved by the Institutional Ethical Committee (IEC) (IEC approval number NIV/IEC/Mar/2022/D‐7; Dated: April 18, 2022) for Research on Humans, based on the guidelines set by the Indian Council of Medical Research, New Delhi. Informed consent was obtained from all the participants.
2.3. Peripheral Blood Mononuclear Cell (PBMC) Isolation, Plasma Separation and Storage
Peripheral blood samples (approximately 3–4 mL) from the study participants were collected and processed for PBMC isolation by density gradient centrifugation [14]. Isolated PBMCs were used for T cell ELISPOT assay, while plasma samples were stored at ‐80°C and thawed for estimation of cytokines/chemokines and growth factors using a Bio‐plex Pro Human Cytokine 27‐plex assay kit, anti‐SARS‐CoV‐2 IgG antibody detection by ELISA, and 67 plasma samples from PASC and 92 plasma samples from N‐PASC were considered for neutralizing antibodies (NAbs) against SARS‐CoV‐2 using Surrogate Virus Neutralization (sVNT) assay (cPass, GenScript USA) [9, 10, 15]. To minimize potential bias, samples were selected without regard to clinical outcomes other than group classification (PASC vs. N‐PASC). The selected subset is representative of the broader cohort in terms of key demographic and clinical characteristics. A plasma sample was considered anti‐SARS‐CoV‐2‐S1 IgG antibody‐positive if the OD450nm was three times greater than that of the OD450nm of the negative control. Circulating NAbs against SARS‐CoV‐2 competitively inhibited the RBD‐ACE2 interactions. The percentage of inhibition was calculated by measuring the difference in the amount of labeled RBD between test and control samples. The cut‐off for percentage inhibition was set at 30%. Two‐fold dilutions from 1:10 to 1:320 were performed to determine the titer. Samples showing greater than 30% signal inhibition were considered positive.
2.4. Estimation of Plasma Cytokine, Chemokine, and Growth Factor Levels
Plasma concentrations of cytokines, chemokines, and growth factors were determined in PASC patients (n = 67) & N‐PASC (n = 81), on a Bio‐plex Multiplex Immunoassay System (Bio‐Rad, Hercules, CA, USA) using a Bio‐plex Pro Human Cytokine 27‐plex assay kit as reported previously [10] as per the manufacturer's instructions. Levels of 15 cytokines, including the pro‐inflammatory (IL‐1β, IL‐5, IL‐6, IL‐7, IL‐9, IL‐15, IL‐17, TNF‐α), anti‐inflammatory (IL1‐RA, IL‐4, IL‐10, IL‐13), and Th1 (IL‐2, IFN‐γ, IL‐12 p70) cytokines along with 7 chemokines (Eotaxin, CCL‐2, CCL‐3, CCL‐4, CCL‐5, IL‐8, CXCL‐10) and five growth factors (basic fibroblast growth factor [FGF], G‐CSF, GM‐CSF, vascular endothelial growth factor [VEGF], platelet‐derived growth factor‐BB [PDGF‐BB]) were estimated. The lowest value of the respective standards was used in case of undetected concentrations of the cytokines, chemokines, and growth factors in the tested samples [10].
2.5. SARS‐CoV‐2‐Specific T Cell ELISPOT Assay
SARS‐CoV‐2‐specific T cell response in terms of IFN‐γ release by ELISPOT assay was performed in PASC patients (n = 27) & N‐PASC individuals (n = 40) as previously described [9, 14]. To estimate the number of SARS‐CoV‐2 specific IFN‐γ secreting spot‐forming cells (SFCs), PBMCs were stimulated with recombinant S1 protein (Delta variant) [(SARS‐CoV‐2 Spike S1(E154K, L452R, E484Q, D614G, P681R His Recombinant protein Sino Biological, USA] and recombinant S1 protein (Omicron variant) [SARS‐CoV‐2 Spike S1 Protein, His Tag (B.1.1.529/Omicron), Acro Biosystems, USA]. Wells without any antigen served as negative controls, while those with 10 µg/mL of phytohemagglutinin (PHA) (Sigma Aldrich, USA) served as positive controls. All assays were carried out in triplicate. The IFN‐γ SFCs were counted on an ELISPOT reader, customized software (AID GmbH, Strassberg, Germany), and were expressed as the number per 105cells. The cut‐off level for SFCs was set as twice the average number of SFCs in the negative control wells. Results with high background readings or with no PHA responses were excluded. Due to limitations in sample availability following processing and storage, as well as the material requirements of functional assays, the ELISPOT assay was performed on subsets of the cohort, on 27 PASC and 40 N‐PASC individuals.
2.6. Statistical Analysis
Statistical analyses were performed using RStudio (R version 4.5.1) [15], while receiver operating characteristic (ROC) curve analyses were conducted using GraphPad Prism version 8.0 (GraphPad Software Inc. San Diego, CA, USA). Cytokine, chemokine, and growth factor concentrations were transformed by adding 1 to each observed value followed by logarithmic transformation prior to statistical analysis [16].
Differences in cytokine concentrations between study groups were assessed using the Mann–Whitney U test for comparisons between two groups and the Kruskal–Wallis test for comparisons involving more than two groups. For each comparison, an unadjusted p value was initially calculated, with statistical significance defined as p < 0.05. To account for multiple comparisons, p values were subsequently adjusted using the Benjamini–Hochberg (BH) false discovery rate (FDR) correction, and an adjusted p value < 0.05 was considered statistically significant. Both unadjusted and BH‐adjusted p values are reported.
In addition to p‐values, effect sizes were calculated to quantify the magnitude of differences between groups independently of sample size [17]. Effect sizes were interpreted according to established thresholds as small, moderate, or large, providing complementary information regarding the practical or biological relevance of observed differences. Effect sizes were estimated using epsilon‐squared (ε2) with the Kruskal_effsize() function from the rstatix package in R [18]. Effect sizes were interpreted as negligible (ε2 < 0.01), small (0.01–0.079), moderate (0.08–0.259), and large (≥ 0.26).
Also, to evaluate the independent association between significant cytokines and post‐acute sequelae of SARS‐CoV‐2 infection (PASC), multiple logistic regression analysis was performed with PASC status (PASC vs. N‐PASC) as the binary outcome. Cytokines that were significantly associated with PASC in the univariate analyses were included in the multivariable model, with adjustment for relevant demographic and clinical covariates to account for potential confounding. Adjusted odds ratios (aORs) and 95% confidence intervals (95% CIs) were estimated to quantify the effect of each cytokine on the odds of PASC.
3. Results
3.1. Characteristics of the Study Population
The median (IQR) age of the study participants was 40 years (34–55 years). Females constituted 38.3% (n = 62) of the study. Forty percent (n = 64) of them were well educated. With respect to occupation of the study group, it comprised skilled workers (32%, n = 52) and semi‐skilled workers (15.4%, n = 25). The rest of them belonged to the category of homemakers and retirees from various services, not actively engaged in other occupations at the time of the interview. At the time of the interview, 19.8% (n = 32) had taken the full course of COVID‐19 vaccination (including booster), 71% (n = 115) had taken two doses, 3.7% (n = 6) had taken one dose, and 5.6% (n = 9) were unvaccinated. At the time of the interview, 46.9% (n = 76) of the study participants satisfied the clinical definition of long COVID as stated by the WHO and were categorized as the PASC group. Table 1 shows the baseline characteristics of the study population comparing the PASC (n = 76) and N‐PASC (n = 86) study groups (Tables 1a and 1b).
Table 1.
Characteristics of the study population between the PASC and N‐PASC groups.
| Factor | Category | PASC (n = 76) | N‐PAS (n = 86) | p value |
|---|---|---|---|---|
| Age | Median (IQR) | 41.5 (35–55.75) | 40 (32–55) | 0.477 |
| Gender | Female (%) | 32 (42.1) | 30 (34.9) | 0.345 |
| Education | Primary or lesser (%) | 5 (6.6) | 17 (19.8) | 0.049 |
| Secondary (%) | 38 (50) | 38 (44.2) | ||
| Graduation and above (%) | 33 (43.4) | 31 (36) | ||
| Chronic comorbidities | Present (%) | 35 (46.1) | 25 (29.1) | 0.025 |
| Hospitalization for COVID‐19 | Yes (%) | 29 (38.2) | 12 (14) | 0.001 |
| ICU admission during COVID‐19 | Yes (%) | 9 (11.8) | 1 (1.2) | 0.005 |
| Oxygen supplementation during COVID‐19 | Yes (%) | 7 (9.2) | 5 (5.8) | 0.410 |
| BMI | Median (IQR) | 25.96 (23.31–9.34) | 26.53 (23.19–28.75) | 0.942 |
| Time since diagnosis | Up to 1 year (%) | 7 (9.2) | 5 (5.8) | 0.474 |
| Up to 2 years (%) | 36 (47.4) | 48 (55.8) | ||
| Up to 3 years and beyond (%) | 33 (43.4) | 33 (38.4) |
Abbreviations: BMI, body mass index; COVID‐19, coronavirus disease 2019; NA, not applicable; categorical and binary variables presented as n (%) and compared using the χ2 test.
Table 1a.
Comparison of Cytokine Levels Among Hospitalized PASC Patients Across SARS‐CoV‐2 Variant Groups Using Kruskal–Wallis Test with Benjamini–Hochberg Adjustment and effect size (ε2) (*indicates p‐value < 0.05).
| SARS Cov‐2 variant | |||||||
|---|---|---|---|---|---|---|---|
| Cytokine | Wild type (n = 13) Median | Delta (n = 6) Median | Omicron (n = 5) Median | Unadjusted p. value | Adjusted p value (FDR) | Effect size | Magnitude |
| Eotaxin | 1.9093 | 1.7124 | 1.8715 | 0.3578 | 0.4821 | 0.0027 | Negligible |
| FGF_basic | 0.8820 | 0.7657 | 0.8820 | 0.3750 | 0.4821 | 0.0000 | Negligible |
| G‐CSF | 1.3103 | 1.0473 | 1.4311 | 0.5134 | 0.5775 | 0.0000 | Negligible |
| GM‐CSF | 0.1367 | 0.1367 | 0.1139 | 0.8720 | 0.8720 | 0.0000 | Negligible |
| IFN‐g | 0.7694 | 0.4150 | 0.4800 | 0.3189 | 0.4783 | 0.0136 | Small |
| Hu IL‐1b (39) | 0.4786 | 0.2553 | 0.1139 | 0.0152 | 0.2569 | 0.3033 | Large |
| Hu IL‐1ra (25) | 2.1225 | 1.9779 | 2.0131 | 0.2835 | 0.4783 | 0.0248 | Small |
| Hu IL‐2 (38) | 0.4014 | 0.5563 | 0.5587 | 0.0430 | 0.3868 | 0.2045 | Moderate |
| Hu IL‐4 (52) | 0.4857 | 0.3304 | 0.4502 | 0.3994 | 0.4901 | 0.0000 | Negligible |
| Hu IL‐5 (33) | 0.9074 | 0.9074 | 1.0418 | 0.2619 | 0.4783 | 0.0324 | Small |
| Hu IL‐6 (19) | 0.2601 | 0.1523 | 0.3008 | 0.2407 | 0.4783 | 0.0404 | Small |
| Hu IL‐7 (74) | 0.6075 | 0.6075 | 1.3694 | 0.2619 | 0.4783 | 0.0324 | Small |
| Hu IL‐8 (54) | 0.5366 | 0.3856 | 0.5229 | 0.3169 | 0.4783 | 0.0142 | Small |
| Hu IL‐9 (77) | 1.7773 | 1.7018 | 1.8993 | 0.0876 | 0.4783 | 0.1367 | Moderate |
| Hu IL‐10 (56) | 0.8136 | 0.8136 | 0.8312 | 0.2619 | 0.4783 | 0.0324 | Small |
| Hu IL‐12(p70) (75) | 0.6964 | 0.3784 | 0.9518 | 0.1278 | 0.4783 | 0.1007 | Moderate |
| Hu IL‐13 (51) | 0.4456 | 0.3304 | 0.4065 | 0.5883 | 0.6354 | 0.0000 | Negligible |
| Hu IL‐15 (73) | 1.3341 | 1.3341 | 1.9956 | 0.2619 | 0.4783 | 0.0324 | Small |
| Hu IL‐17 (76) | 0.6243 | 0.5185 | 0.6166 | 0.4484 | 0.5263 | 0.0000 | Negligible |
| Hu IP‐10 (48) | 2.2788 | 1.8905 | 2.1265 | 0.2810 | 0.4783 | 0.0256 | Small |
| Hu MCP‐1(MCAF) (53) | 1.0770 | 0.8014 | 0.8159 | 0.3691 | 0.4821 | 0.0000 | Negligible |
| Hu MIP‐1a (55) | 0.2967 | 0.2304 | 0.2624 | 0.6877 | 0.7141 | 0.0000 | Negligible |
| Hu MIP‐1b (18) | 1.4853 | 1.4631 | 1.7550 | 0.1008 | 0.4783 | 0.1233 | Moderate |
| Hu PDGF‐bb (47) | 0.6972 | 0.6972 | 0.9545 | 0.1803 | 0.4783 | 0.0679 | Small |
| Hu RANTES (37) | 2.2529 | 2.4633 | 2.8294 | 0.0190 | 0.2569 | 0.2821 | Large |
| Hu TNF‐a (36) | 1.3860 | 1.4275 | 1.5153 | 0.2182 | 0.4783 | 0.0498 | Small |
| Hu VEGF (45) | 1.1389 | 1.0241 | 1.7190 | 0.2505 | 0.4783 | 0.0366 | Small |
Note: Pairwise comparisons were performed using Dunn's test with Benjamini–Hochberg false discovery rate adjustment. Results should be interpreted with caution because the small sample sizes reduce statistical power.
Table 1b.
Pairwise Comparisons of Cytokine Levels Among SARS‐CoV‐2 Variant Groups Using Dunn's Test with Benjamini–Hochberg Adjustment.
| Cytokine | Comparison | Unadjusted p value | Adjusted p value (BH) |
|---|---|---|---|
| Hu IL‐1 b (39) | Omicron ‐ Wild Type | 0.004303* | 0.012909 |
| Hu RANTES (37) | Omicron ‐ Wild Type | 0.005581* | 0.016742 |
| Hu IL‐2 (38) | Omicron ‐ Wild Type | 0.012174* | 0.036522 |
| Hu IL‐9 (77) | Delta ‐ Omicron | 0.040378 | 0.121133 |
| Hu IL‐12(p70) (75) | Delta ‐ Omicron | 0.044938 | 0.134813 |
| Hu MIP‐1b (18) | Delta ‐ Omicron | 0.047696 | 0.143088 |
| Hu IL‐1b (39) | Delta ‐ Omicron | 0.047783 | 0.071675 |
Indicates p‐value < 0.05.
The clinical symptoms as reported by the participants belonging to the PASC group were collected. More than half (52.6%) of the PASC study group (n = 40) experienced three or more persistent symptoms. The highest number of post‐COVID‐19 symptoms (8 symptoms) was reported by one person. The three most commonly experienced symptoms were fatigue, memory disturbances, and shortness of breath. The clinical symptoms, according to the reported frequencies, are shown in Figure 2. The PASC group of the current study had significantly higher chronic comorbidities, hospitalization, and ICU admissions during the COVID‐19 infection compared to the N‐PASC group. The schematic representation of the study design is shown in Figure 3.
Figure 2.
Frequency distribution of self‐reported clinical symptoms experienced by the study participants in the PASC group (n = 67).
Figure 3.
Schematic representation of study design.
3.2. Plasma Cytokines and Chemokines
3.2.1. PASC and N‐PASC Groups
Overall, the plasma concentrations of cytokines, chemokines, and growth factors were comparable between the PASC (n = 67) and N‐PASC (n = 81) groups (Table 2; Figure 4a). Plasma cytokine/chemokine analysis among the hospitalized PASC patients and N‐PASC individuals revealed significantly higher IL‐6 levels in PASC patients (IL‐6: PASC, 0.22 [0.13‐0.34] vs. N‐PASC, 0.13 [0.07‐ 0.15], p = 0.048). Cytokine and chemokine levels in the PASC group at 1 year, 2 year, and 3 year post‐SARS‐CoV‐2 infection were comparable, however, in the N‐PASC group, individuals at 1 year post‐infection had higher levels of FGF‐basic, IP‐10 and RANTES compared to that at 3 year post infection (FGF‐basic 1 year: 0.77 [0.61–0.98] vs. 3 years: 0.70 [0.66–0.833], p = 0.04), IP‐10 (1 year: 2.14 [2.02–2.21] vs. 3 years: 2 [1.67–2.22], p = 0.023), RANTES (1 year: 2.78 [2.51–2.81]; 3 years: 2.53 [2.25–2.72], p = 0.013). Cytokine comparisons across 1‐, 2‐, and 3‐year post‐infection groups represent cross‐sectional analyses of different individuals. Further, one‐way ANOVA analysis indicated that IP‐10, FGF‐basic and RANTES had a gradual decrease over a period of one, two and 3 years in the N‐PASC group. Mann–Whitney U test in the N‐PASC group showed the following results: IL‐12 was significantly higher in individuals at 1 year post COVID‐19 infection compared to 2 years post infection. RANTES levels were also significantly higher at 1 year post‐COVID‐19 infection compared to 3 years post‐infection (Table 3; Figure 4b).
Table 2.
Comparison of Cytokine Levels Among Non‐hospitalized PASC Patients Across SARS‐CoV‐2 Variant Groups Using Kruskal–Wallis Test with Benjamini–Hochberg Adjustment and effect size (ε2) (*indicates p‐value < 0.05).
| Cytokine | Wild Type (n = 14) Median | Delta (n = 24) Median | Omicron (n = 5) Median | Unadjusted p. value | Adjusted p value (FDR) | Effect size | Magnitude |
|---|---|---|---|---|---|---|---|
| Eotaxin | 1.6493 | 1.6994 | 1.8171 | 0.3428 | 0.9832 | 0.0035 | Negligible |
| FGF basic | 0.6391 | 0.6325 | 0.7042 | 0.8295 | 0.9832 | 0.0000 | Negligible |
| G‐CSF | 1.0276 | 0.8158 | 1.3118 | 0.5461 | 0.9832 | 0.0000 | Negligible |
| GM‐CSF | 0.1253 | 0.1367 | 0.1367 | 0.6229 | 0.9832 | 0.0000 | Negligible |
| IFN‐g | 0.3774 | 0.4048 | 0.3856 | 0.7360 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐1b (39) | 0.1780 | 0.2989 | 0.2989 | 0.5279 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐1ra (25) | 1.9384 | 1.9191 | 1.9709 | 0.9215 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐2 (38) | 0.5575 | 0.5563 | 0.1732 | 0.2824 | 0.9832 | 0.0132 | Small |
| Hu IL‐4 (52) | 0.3384 | 0.3464 | 0.4409 | 0.3147 | 0.9832 | 0.0078 | Negligible |
| Hu IL‐5 (33) | 0.9746 | 0.9074 | 1.0418 | 0.7243 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐6 (19) | 0.1335 | 0.1523 | 0.1523 | 0.3514 | 0.9832 | 0.0023 | Negligible |
| Hu IL‐7 (74) | 0.9884 | 0.6075 | 1.3694 | 0.7243 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐8 (54) | 0.3382 | 0.3589 | 0.3997 | 0.8822 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐9 (77) | 1.7750 | 1.7550 | 1.8589 | 0.9832 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐10 (56) | 0.8224 | 0.8136 | 0.8312 | 0.7736 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐12(p70) (75) | 0.7812 | 0.3784 | 0.2601 | 0.3184 | 0.9832 | 0.0072 | Negligible |
| Hu IL‐13 (51) | 0.2670 | 0.2317 | 0.3579 | 0.2980 | 0.9832 | 0.0105 | Small |
| Hu IL‐15 (73) | 1.6648 | 1.3341 | 1.9956 | 0.7243 | 0.9832 | 0.0000 | Negligible |
| Hu IL‐17 (76) | 0.3252 | 0.3944 | 0.5988 | 0.1466 | 0.9832 | 0.0460 | Small |
| Hu IP‐10 (48) | 2.0899 | 2.0156 | 2.1033 | 0.8746 | 0.9832 | 0.0000 | Negligible |
| Hu MCP‐1(MCAF) (53) | 0.7925 | 0.8114 | 0.8716 | 0.6962 | 0.9832 | 0.0000 | Negligible |
| Hu MIP‐1a (55) | 0.2157 | 0.2227 | 0.2480 | 0.7475 | 0.9832 | 0.0000 | Negligible |
| Hu MIP‐1b (18) | 1.5537 | 1.5057 | 1.7162 | 0.9634 | 0.9832 | 0.0000 | Negligible |
| Hu PDGF‐bb (47) | 0.7219 | 0.7633 | 0.6972 | 0.4909 | 0.9832 | 0.0000 | Negligible |
| Hu RANTES (37) | 2.6509 | 2.5596 | 2.5311 | 0.8462 | 0.9832 | 0.0000 | Negligible |
| Hu TNF‐a (36) | 1.3106 | 1.3193 | 1.4651 | 0.9597 | 0.9832 | 0.0000 | Negligible |
| Hu VEGF (45) | 1.3715 | 1.0241 | 1.5780 | 0.7762 | 0.9832 | 0.0000 | Negligible |
Figure 4.
(a) Heat map of levels of cytokines/chemokines in PASC and N‐PASC individuals. (b) Levels of cytokines (FGF basic, IP10, and CCL5/RANTES) in N‐PASC individuals post 1 year, 2 years, and 3 years of SARS‐CoV‐2 infection. (c) Graphical representation of plasma cytokine/chemokines in PASC and N‐PASC. (d) 1. ROC curves for IL‐6 in Hospitalized, PASC vs N‐PASC;p = 0.046. 2. ROC curves for FGF‐basic in PASC (hospitalized vs. non‐hospitalized in PASC); p = 0.005.
Table 3.
Comparison of Cytokine Levels Among Hospitalized NPASC Patients Across SARS‐CoV‐2 Variant Groups Using the Mann–Whitney U Test with Benjamini–Hochberg Adjustment and Effect Size (ε2)*indicates p‐value < 0.05).
| Cytokine | Wild Type (n = 3) Median | Delta (n = 8) Median | Unadjusted p. value | Adjusted p value (FDR) | Effect size | Magnitude |
|---|---|---|---|---|---|---|
| Eotaxin | 1.7010 | 1.7246 | 0.8383 | 0.8705 | 0.0000 | Negligible |
| FGF basic | 0.8325 | 0.7991 | 0.6761 | 0.8705 | 0.0000 | Negligible |
| G‐CSF | 1.1517 | 0.7948 | 0.6831 | 0.8705 | 0.0000 | Negligible |
| GM‐CSF | 0.1139 | 0.1367 | 0.2225 | 0.5838 | 0.0542 | Small |
| IFN‐g | 0.5855 | 0.3669 | 0.6082 | 0.8705 | 0.0000 | Negligible |
| Hu IL‐1b (39) | 0.3444 | 0.2753 | 0.4099 | 0.7906 | 0.0000 | Negligible |
| Hu IL‐1ra (25) | 2.0262 | 1.7780 | 0.3561 | 0.7906 | 0.0000 | Negligible |
| Hu IL‐2 (38) | 0.5587 | 0.5563 | 0.1630 | 0.5838 | 0.1051 | Moderate |
| Hu IL‐4 (52) | 0.4857 | 0.3139 | 0.2154 | 0.5838 | 0.0594 | Small |
| Hu IL‐5 (33) | 1.0418 | 0.9074 | 0.2225 | 0.5838 | 0.0542 | Small |
| Hu IL‐6 (19) | 0.1335 | 0.1319 | 0.7589 | 0.8705 | 0.0000 | Negligible |
| Hu IL‐7 (74) | 1.3694 | 0.6075 | 0.2225 | 0.5838 | 0.0542 | Small |
| Hu IL‐8 (54) | 0.4669 | 0.2900 | 0.2378 | 0.5838 | 0.0437 | Small |
| Hu IL‐9 (77) | 1.8246 | 1.6854 | 0.6831 | 0.8705 | 0.0000 | Negligible |
| Hu IL‐10 (56) | 0.8312 | 0.8136 | 0.2225 | 0.5838 | 0.0542 | Small |
| Hu IL‐12(p70) (75) | 0.3784 | 0.3784 | 0.8234 | 0.8705 | 0.0000 | Negligible |
| Hu IL‐13 (51) | 0.4065 | 0.2159 | 0.1836 | 0.5838 | 0.0854 | Moderate |
| Hu IL‐15 (73) | 1.9956 | 1.3341 | 0.2225 | 0.5838 | 0.0542 | Small |
| Hu IL‐17 (76) | 0.5988 | 0.3345 | 0.8375 | 0.8705 | 0.0000 | Negligible |
| Hu IP‐10 (48) | 1.9400 | 1.9031 | 0.6831 | 0.8705 | 0.0000 | Negligible |
| Hu MCP‐1(MCAF) (53) | 0.8573 | 0.8296 | 0.6090 | 0.8705 | 0.0000 | Negligible |
| Hu MIP‐1a (55) | 0.2480 | 0.2256 | 1.0000 | 1.0000 | 0.0000 | Negligible |
| Hu MIP‐1b (18) | 1.6871 | 1.4296 | 0.8383 | 0.8705 | 0.0000 | Negligible |
| Hu PDGF‐bb (47) | 0.6972 | 0.6972 | 0.3918 | 0.7906 | 0.0000 | Negligible |
| Hu RANTES (37) | 2.3644 | 2.6665 | 0.1530 | 0.5838 | 0.1157 | Moderate |
| Hu TNF‐a (36) | 1.4651 | 1.2821 | 0.8383 | 0.8705 | 0.0000 | Negligible |
| Hu VEGF (45) | 1.7190 | 1.0241 | 0.2225 | 0.5838 | 0.0542 | Small |
3.3. Subgroup Analysis Within the PASC Patients
PASC patients with persistence of one symptom (n = 16) had significantly higher IL‐9, MIP 1β, and IP‐10 compared to patients with persistence of 2 symptoms (n = 14) (IL‐9: 2 symptoms, 1.80 [1.71–1.91] 1.6 [0.13‐0.26] vs. 1 symptom, 1.66 [1.52–1.82], p = 0.02; MIP 1β: 1 symptom, 1.70 [1.46–1.78] vs. 2 symptoms, 1.38 [1.25–1.62], p = 0.02; IP‐10: 1 symptom, 2.09 [1.93–2.18] vs. 2 symptoms, 1.88 [1.77–2.02], p = 0.02). Hospitalized PASC patients had significantly higher IL‐6 and FGF‐basic levels compared to non‐hospitalized patients. (IL‐6: hospitalized, 0.26 [0.13–0.34], vs. non‐hospitalized, 0.13 [0.13–0.17]; p = 0.04, FGF‐basic: hospitalized, 0.86 [0.7–0.9] vs. non‐hospitalized, 0.6 [0.5–0.83], p = 0.005) (Table 4; Figure 4c) (Table 5).
Table 4.
Comparison of Cytokine Levels Among Nonhospitalized NPASC Patients Across SARS‐CoV‐2 Variant Groups Using the Mann–Whitney U Test with Benjamini–Hochberg Adjustment and Effect Size (ε2)*indicates p‐value < 0.05).
| Cytokine | Wild Type (n = 22) Median | Delta (n = 36) Median | Omicron (n = 10) Median | Unadjusted p value | Adjusted p value (FDR) | Effect size | Magnitude |
|---|---|---|---|---|---|---|---|
| Eotaxin | 1.7678 | 1.7420 | 1.7512 | 0.9607 | 0.9932 | 0.0000 | Negligible |
| FGF basic | 0.7042 | 0.7578 | 0.7701 | 0.8616 | 0.9932 | 0.0000 | Negligible |
| G‐CSF | 1.1604 | 1.1192 | 0.8675 | 0.8755 | 0.9932 | 0.0000 | Negligible |
| GM‐CSF | 0.1139 | 0.1367 | 0.1253 | 0.4094 | 0.9932 | 0.0000 | Negligible |
| IFN‐g | 0.4377 | 0.4800 | 0.4146 | 0.9045 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐1b (39) | 0.2122 | 0.2122 | 0.2174 | 0.9249 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐1ra (25) | 1.9086 | 1.9709 | 1.9367 | 0.8376 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐2 (38) | 0.5563 | 0.5575 | 0.5587 | 0.6452 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐4 (52) | 0.4065 | 0.3838 | 0.4065 | 0.9775 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐5 (33) | 1.0418 | 1.0418 | 1.0418 | 0.9584 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐6 (19) | 0.1523 | 0.1523 | 0.1523 | 0.8991 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐7 (74) | 1.3694 | 1.3694 | 1.3694 | 0.8940 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐8 (54) | 0.4090 | 0.3908 | 0.4090 | 0.9932 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐9 (77) | 1.8306 | 1.8164 | 1.8011 | 0.8085 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐10 (56) | 0.8312 | 0.8312 | 0.8312 | 0.9584 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐12(p70) (75) | 0.4431 | 0.3784 | 0.9518 | 0.4386 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐13 (51) | 0.3295 | 0.2922 | 0.3579 | 0.8934 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐15 (73) | 1.9956 | 1.9956 | 1.9956 | 0.9584 | 0.9932 | 0.0000 | Negligible |
| Hu IL‐17 (76) | 0.5611 | 0.4728 | 0.5398 | 0.7616 | 0.9932 | 0.0000 | Negligible |
| Hu IP‐10 (48) | 2.0289 | 2.0198 | 2.0905 | 0.5373 | 0.9932 | 0.0000 | Negligible |
| Hu MCP‐1(MCAF) (53) | 0.8577 | 0.8645 | 0.8431 | 0.9309 | 0.9932 | 0.0000 | Negligible |
| Hu MIP‐1a (55) | 0.2330 | 0.2414 | 0.2214 | 0.7698 | 0.9932 | 0.0000 | Negligible |
| Hu MIP‐1b (18) | 1.7043 | 1.6370 | 1.6769 | 0.8925 | 0.9932 | 0.0000 | Negligible |
| Hu PDGF‐bb (47) | 0.6972 | 0.6972 | 0.6972 | 0.7351 | 0.9932 | 0.0000 | Negligible |
| Hu RANTES (37) | 2.5577 | 2.5873 | 2.6970 | 0.4616 | 0.9932 | 0.0000 | Negligible |
| Hu TNF‐a (36) | 1.4426 | 1.3928 | 1.3835 | 0.8498 | 0.9932 | 0.0000 | Negligible |
| Hu VEGF (45) | 1.1364 | 1.2601 | 1.7190 | 0.8009 | 0.9932 | 0.0000 | Negligible |
Table 5.
Multiple Logistic Regression Model Showing Adjusted Odds Ratios and 95% Confidence Intervals for PASC Compared with Non‐PASC.
| Variable | Estimate | p value | Adjusted odds ratio | 95% confidence interval (CI) | |
|---|---|---|---|---|---|
| Lower CI | Upper CI | ||||
| Gender: Male (Ref: Female) | −0.3746 | 0.3108 | 0.6876 | 0.3307 | 1.4166 |
| Hospitalization: Yes (Ref: No) | 1.3884 | 0.0017* | 4.0084 | 1.7294 | 9.8759 |
| Vaccine One dose (Ref: No Vaccination) | 0.5345 | 0.6711 | 1.7066 | 0.1458 | 23.2030 |
| Vaccine Two doses (Ref: No Vaccination) | 0.4392 | 0.6029 | 1.5515 | 0.2973 | 9.0870 |
| VaccineTwo doses + booster (Ref: No Vaccination) | 0.0453 | 0.9604 | 1.0463 | 0.1754 | 6.8513 |
| Duration 2 years (Ref: Duration 1 year) | −0.4090 | 0.4468 | 0.6643 | 0.2266 | 1.9033 |
| Duration 3 years (Ref: Duration 1 year) | −0.4505 | 0.4352 | 0.6373 | 0.2016 | 1.9735 |
| FGF basic | −1.8287 | 0.0276* | 0.1606 | 0.0293 | 0.7803 |
Indicates p‐value < 0.05.
3.4. Variant Comparison
Twenty‐seven cytokines were compared among hospitalized PASC patients infected with the Wild Type, Delta, or Omicron SARS‐CoV‐2 variants. Owing to the small and unequal sample sizes across SARS‐CoV‐2 variant groups (Delta, n = 13; Omicron, n = 6; Wild Type, n = 5), cytokine concentrations were compared using the Kruskal–Wallis rank‐sum test. Nominally significant differences among the three variant groups were observed for IL‐1β (p = 0.015), RANTES (p = 0.019), and IL‐2 (p = 0.043). However, after adjustment for multiple comparisons using the Benjamini–Hochberg FDR, none of the cytokines remained statistically significant (all adjusted p > 0.05). Several cytokines, including IL‐9 (p = 0.088), MIP‐1β (p = 0.101), IL‐12(p70) (p = 0.128), and PDGF‐BB (p = 0.18), demonstrated non‐significant trends toward differential expression across the three variant groups. Effect size analysis indicated large effects for IL‐1β (effect size = 0.3033) and RANTES (effect size = 0.2821), moderate effects for IL‐2, IL‐9, IL‐12(p70), and MIP‐1β, and small or negligible effects for the remaining cytokines. Median cytokine concentrations suggested higher RANTES levels in the Omicron group and lower IL‐1β levels in Omicron compared with the Wild Type group, although these trends did not remain statistically significant after correction for multiple testing. Overall, these findings suggest that while IL‐1β, RANTES, and IL‐2 showed the strongest evidence of variation among Wild Type, Delta, and Omicron‐associated hospitalized PASC patients, no cytokine exhibited statistically significant differences after controlling for multiple testing. The observed trends for IL‐1β, RANTES, and IL‐2 should be interpreted cautiously because of the small sample sizes, which limit statistical power and may have reduced the ability to detect true differences.
3.5. ROC Analysis of FGF Basic and IL‐6
Based on the history of hospitalization, the PASC and N‐ PASC individuals were segregated, and their cytokine/chemokine values were analyzed by performing a ROC analysis. Sensitivity, specificity, and AUC (area under the ROC curve) were reported to see the prediction accuracy of the model. The ROC analysis determined cut‐offs for cytokines that could distinguish recovered individuals (N‐PASC in the current study) from Long COVID patients (PASC in the current study). These cut‐offs may have future diagnostic relevance for the PASCs. AUC for IL‐6 in the hospitalized group for PASC vs N‐PASC was 0.71 (95% CI 0.54–0.88; p = 0.046); the optimal cut‐off value was 0.37 pg/ml (sensitivity, 81%; specificity, 59%). Similarly, the FGF‐basic levels moderately discriminated between hospitalized and non‐hospitalized PASC patients, with an AUC of 0.70 (95% CI: 0.58–0.83; p = 0.005). The cut‐off values were 0.11 pg/mL (sensitivity, 60%; specificity, 50%) (Figure 4d).
3.6. SARS‐CoV‐2‐ Specific T Cell Response
In the PASC group, IFN‐γ responses in unstimulated, recombinant S1 protein (Omicron variant), recombinant S1 protein (Delta variant), and PHA‐stimulated cells were 2 (0.58–7.67), 5 (1–11.9), 7.5 (1.17–12.5), and 50 (18.8–167) SFCs/105 cells, respectively. In the N‐PASC group, IFN‐γ responses in unstimulated, recombinant S1 protein (Omicron variant), recombinant S1 protein (Delta variant), and PHA‐stimulated cells were 2 (0.67–4.5), 2.67 (1–7.67), 2.5 (1–5.63), and 60.8 (22.7–239) SFCs/105 cells, respectively. In the PASC group, 9/27 (33.3%) and 8/27 (29.6%) were responders to the S1 protein (Omicron variant) and the recombinant S1 protein (Delta variant), respectively. In the N‐PASC group, 10/40 (25%) and 13/40 (32.5%) were responders to the S1 protein (Omicron variant) and the recombinant S1 protein (Delta variant), respectively. The PASC group was further categorized as hospitalized (n = 8) and non‐hospitalized (n = 18) during COVID‐19 infection. Among the hospitalized PASC group, 3/8 (37.5%) and 2/8 (25%) were responders to the S1 protein (Omicron variant) and the recombinant S1 protein (Delta variant), respectively. In the non‐hospitalized group, 5/18 (27.7%) were responders to both the S1 protein (Omicron variant) and the recombinant S1 protein (Delta variant). Among the hospitalized N‐PASC group, 2/7 (28.6%) were responders to both the S1 protein (Omicron variant) and the recombinant S1 protein (Delta variant), while in the non‐hospitalized N‐PASC group, 8/33 (24.2%) were responders to the S1 protein (Omicron variant), and 11/33 (33%) were responders to the recombinant S1 protein (Delta variant). The T cell response was the lowest in the PASC and N‐PASC groups at 1 year post‐COVID‐19 infection, although the sample sizes assessed were very small in both groups. For patients in their 2nd and 3rd years post‐COVID‐19 infection, the T cell response was enhanced. ELISPOT analysis, conducted on a subset of participants with available viable PBMCs, showed comparable antigen‐specific T‐cell responses between PASC and N‐PASC groups.
3.7. Detection of anti‐SARS‐CoV‐2 spike/S1 IgG Antibodies, and SARS‐CoV‐2 Surrogate Virus Neutralization (sVNT) Antibodies
Plasma samples from 65/67 (97%) PASC group and 80/81 N‐PASC group (98.7%) were anti‐SARS‐CoV‐2 IgG antibody positive. SARS‐CoV‐2 surrogate virus neutralization (sVNT) antibodies were detected in 63/67 (94.0%) PASC participants and 85/92 (92.4%) N‐PASC participants. The median inhibition percentage was 90.90% (53.81–98.47) in the PASC group and 96.59% (61.66–98.47) in the N‐PASC group, with significantly higher inhibition observed in the N‐PASC group (p = 0.0028) (Figure 5).
Figure 5.
Neutralization assay comparing percentage inhibition between PASC vs N‐PASC.
4. Discussion
Symptoms of long COVID vary in severity, can be continuous, relapsing and remitting, or progressive, and are usually associated with economic and social functional impairment, emotional and physical distress [19]. Detection of viral RNA and spike proteins in blood and tissues months after acute COVID‐19 infection has suggested viral reservoirs or chronic antigen presence. Genetic and epigenetic factors have also been attributed to the immune reactivation and development of long COVID [20]. The severity of the initial COVID‐19 illness, hospitalization, and a high number of acute symptoms also strongly predict long COVID development. This multifactorial causation underscores the complexity of long COVID and highlights the need for (a) a biomarker as a measure of the disease to predict who is at risk and (b) preventive and therapeutic strategies to reduce the risk profile.
Profound fatigue, cognitive impairment, headache, sleep disturbances, shortness of breath, palpitations, cough, gastrointestinal disturbances, and loss of taste/smell are the globally reported clinical manifestations of PASC [21]. Complying with the above observations, our study also reports fatigue in more than half of the PASC cases (Figure 2). Those with chronic comorbidities, such as diabetes, cardiovascular disease, chronic pulmonary conditions, cancer, or immunosuppression are reported to face a significantly higher risk of developing long COVID or post‐acute sequelae of SARS‐CoV‐2 infection (PASC), especially if they were hospitalized or required ICU care during their acute illness [22]. Studies consistently demonstrate that hospitalization and ICU admission enhance the risk of long COVID, whereas the burden of underlying medical conditions increases the vulnerability to post‐COVID complications [23, 24, 25, 26]. In a similar line, the PASC group of the current study had higher chronic comorbidities, hospitalization, and ICU admissions during the COVID‐19 infection compared to the N‐PASC group.
Most studies have indicated similar cytokine and chemokine profiles among PASC and N‐PASC groups. However, long COVID patients also exhibit elevated levels of IL‐17 and IL‐2, while anti‐inflammatory cytokines such as IL‐4 and IL‐10 tend to be higher in N‐PASC individuals, suggesting some overlapping cytokine profiles between PASC and N‐PASC groups, with distinct pro‐inflammatory signatures characterizing long COVID, but not always true [27, 28]. In the current study, we are reporting comparable levels of cytokines, chemokines, and growth factors among PASC and N‐PASC groups.
Higher levels of IP‐10, MIF, MIG, and FGF‐basic are reported to differentiate COVID‐19 patients from healthy controls [29, 30], this finding partially aligns with higher levels of FGF‐basic, IP‐10 and RANTES at 1 year post infection compared to those at 3 year post infection in the N‐PASC group of the current study. This indicates that normalization of these mediators takes time. Overall, a reduction in FGF‐basic levels post‐infection signals a shift from active tissue repair and inflammation to restoration of normal tissue architecture and immune homeostasis [31]. Cytokine comparisons across 1‐, 2‐, and 3‐year post‐infection groups were based on cross‐sectional data from different individuals rather than longitudinal follow‐up.
Cytokine levels in the hospitalized (at the time of initial COVID‐19 infection) PASC group show a complex immune profile with elevated pro‐inflammatory cytokines such as IL‐6, IL‐1β, and TNF‐α persisting months after acute infection. Elevated IL‐6 in particular is a consistent finding associated with disease severity and prolonged symptoms in hospitalized PASC cohorts. These cytokine patterns reflect ongoing inflammation and immune dysregulation, supporting their role in the pathophysiology of long COVID [28]. This backs the proposed explanation that chronic immune dysregulation contributes to the maintenance of symptoms such as fatigue, brain fog, and musculoskeletal pain [29]. In a similar vein, our study also demonstrated elevated IL‐6 and FGF‐ basic in the hospitalized PASC compared to non‐hospitalized PASC patients. Higher levels of FGF‐basic in this subset of patients could be acting as a driver of cell regeneration, as reported elsewhere [29, 31, 32]. PASC patients of the current study population with three or more symptoms demonstrated elevated levels of IL‐6 compared to those with one symptom, associating the same with symptom heterogeneity and persistence, thus providing a rationale for targeted immunomodulatory therapies [33, 34]. Elevated serum IL‐6 levels in patients during initial hospital admission for SARS‐CoV‐2 infection are known to be at increased risk of developing PASC, which could be attenuated by an IL‐6 inhibitor [35]. Serum IL‐6 in distinguishing hospitalized PASC from hospitalized N‐PASC subjects of the current study population sampled at 1 year, 2 years, and 3 years post COVID‐19 supported the utility of IL‐6 as a biomarker for long‐term inflammatory activity. In addition, higher FGF‐basic levels in the hospitalized PASC compared to the non‐hospitalized PASC group highlighted a potential association between elevated growth factor signaling and severity‐related immune dysregulation.
The comparison of cytokine profiles among hospitalized PASC patients infected with the Wild type, Delta, and Omicron SARS‐CoV‐2 variants revealed largely comparable long‐term inflammatory profiles, suggesting that the infecting variant may have limited influence on persistent systemic immune dysregulation [19]. Although variation was observed in IL‐1β, RANTES, and IL‐2 across the three variant groups of our study, these differences were not retained after correction for multiple comparisons. IL‐1β is a key mediator of inflammasome activation and has been consistently associated with severe acute COVID‐19 and PASC. The trend toward a lower IL‐1β level in the Omicron‐associated PASC compared to the Wild type infection may reflect an attenuated inflammatory response and lower systemic cytokine production. In contrast RANTES, demonstrated relatively higher levels in the Omicron‐associated PASC, suggesting that persistent immune cell trafficking and tissue repair mechanisms may continue despite resolution of acute infection. Collectively, these findings suggest that although acute SARS‐CoV‐2 variants differ in transmissibility and disease severity, their long‐term immunological consequences may converge toward a common inflammatory phenotype as observed in the current PASC patients.
Detection of comparable SARS‐CoV‐2‐specific T‐cell response in both PASC and N‐PASC groups against recombinant S1 protein (Omicron variant) and recombinant S1 protein (Delta variant) as recall antigens infer sustained yet dysregulated T‐cell responses that may serve a dual role, providing ongoing protection and also possibly contributing to persistent symptoms. We have previously shown that SARS‐CoV‐2‐specific CD8 + T cell response can last up to 6–8 months post recovery from primary COVID‐19 infection, shifting toward effector memory phenotypes [7]. Considering that the current series of long COVID patients had primary COVID‐19 infection in the years 2020, 2021 and 2022, and the samples were collected in 2023, an enhanced T cell response against S1 in the second year, 2021 (Omicron = 37.5%; Delta = 31.25%) compared to the third year, 2020 (Omicron = 25%; Delta = 25%) could be attributed to the Omicron wave/vaccination. T‐cell response at 2 years and 3 years in the PASC patients suggests ongoing T‐cell activation, possibly driven by persistent antigenic stimulation or immune dysregulation, as reported [36]. The active role of T cells towards inflammation leading to symptom persistence cannot be ruled out, as reported elsewhere [31, 37]. The observed similarity in the T‐cell responses could be due to limited sample size and does not exclude the possibility of a type II error.
SARS‐CoV‐2‐specific IgG antibodies appeared to decline 8 − 9 months post‐recovery from COVID‐19 [7]. Chansaenroj et al have demonstrated a significant decline in anti‐nucleocapsid immunoglobulin G (anti‐N IgG) antibody seropositivity over time, dropping from 87.5% at 3 months to about 26.6% by 12 months after symptom onset. The study also highlighted that higher antibody persistence was associated with disease severity, influencing long‐term humoral immunity responses [38]. In the current study, 65/67 PASC and 80/81 N‐PASC groups tested positive for anti‐SARS‐CoV‐2 IgG antibodies. Thus, our findings, together with previous studies, suggest that although IgG responses are durable, their quality rather than quantity may play a decisive role in the pathogenesis of long COVID, highlighting the importance of immune analyses for a more complete understanding.
A study reported high neutralizing activity at 6 months among hospitalized individuals compared to mild symptomatic individuals, with a slow decline in neutralizing activity [38]. Jansen et al. in 2024 reported that individuals with PASC had significantly lower neutralizing antibody titers against both the wild‐type SARS‐CoV‐2 strain and the Omicron BA.1 variant compared to the recovered individuals. This indicates a reduced ability to neutralize the virus in PASC patients [39]. This goes in line with our findings of significantly higher neutralizing inhibition percentage in the N‐PASC group compared to the PASC group. Although a statistically significant difference in neutralization inhibition was observed between PASC and N‐PASC groups, both groups demonstrated robust neutralizing responses above the assay threshold.
A key limitation of this study is that functional assay, that is, T cell ELISPOT, was performed on subsets of participants due to the limited availability of viable PBMCs. Although samples were collected from the full cohort, only a proportion met the requirements for functional analyses. This reduced sample size may limit the statistical power and generalizability of the findings. Therefore, the results derived from these analyses should be interpreted as exploratory.
In a nutshell, our findings demonstrate that patients with ongoing long COVID maintain a robust humoral and T‐cell immune response months after SARS‐CoV‐2 infection. However, individuals who recovered without developing long COVID exhibited significantly higher neutralizing antibody activity than those with long COVID, despite high neutralizing responses in both groups. Most importantly, higher FGF‐basic levels in the hospitalized compared to non‐hospitalized (during initial COVID‐19 infection) long COVID patients highlighted a potential association between elevated growth factor signaling and disease severity that may have utility as a biomarker candidate.
Author Contributions
Conceptualization: P. R. Sreelakshmi. Anuradha S. Tripathy. Methodology: Priyanka Wagh, Prakash Sarje. Tanvi Shinde. Writing original draft preparation: Anuradha S. Tripathy, Priyanka Wagh, P. R. Sreelakshmi. Tanvi Shinde P. Anisha Sample collection/related contribution: Rahul Jagtap, Sachin Dhaigude, Babasaheb V Tandale. All authors have read and agreed to the final version of the manuscript.
Ethics Statement
The study was approved by the Institutional Ethical Committee (IEC approval number NIV/IEC/Mar/2022/D‐7; Dated 18 April 2022) for Research on Humans, based on the guidelines set by the Indian Council of Medical Research, New Delhi. Informed consent was obtained from all study participants.
Conflicts of Interest
The authors declare no conflicts of interest.
Acknowledgments
The authors would like to acknowledge ICMR‐National Institute of Virology, Pune for supporting the project having ID ETI 2201. The authors are grateful to the concerned local authorities for facilitating the collection of patients’ samples.
Sreelakshmi P. R., Wagh P., Sarje P., et al., “Distinct Cytokine Signatures and Antibody Neutralization Capacity Differentiate Post‐Acute Sequelae of COVID‐19 From Recovered Individuals Despite Comparable T Cell Responses,” Journal of Medical Virology 98 (2026): e71145, 10.1002/jmv.71145.
Prakash Sarje, Tanvi Shinde, and P. Anisha share equal third authorship.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Chen C., Haupert S. R., Zimmermann L., Shi X., Fritsche L. G., and Mukherjee B., “Global Prevalence of Post‐Coronavirus Disease 2019 (COVID‐19) Condition or Long COVID: A Meta‐Analysis and Systematic Review,” Journal of Infectious Diseases 226, no. 9 (2022): 1593–1607, 10.1093/infdis/jiac136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. https://www.who.int/europe/news-room/fact-sheets/item/post-covid-19-condition.
- 3. Mansell V., Hall Dykgraaf S., Kidd M., and Goodyear‐Smith F., “Long COVID and Older People,” Lancet Healthy Longevity 3, no. 12 (2022): e849–e854. [DOI] [PubMed] [Google Scholar]
- 4. Shah D. P., Thaweethai T., Karlson E. W., et al., “Sex Differences in Long COVID,” JAMA Network Open 8, no. 1 (2025): e2455430, 10.1001/jamanetworkopen.2024.55430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Nalbandian A., Sehgal K., Gupta A., et al., “Post‐Acute COVID‐19 Syndrome,” Nature Medicine 27, no. 4 (2021): 601–615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Pongkunakorn T., Manosan T., Surawit A., et al., “Immune Response After SARS‐Co V‐2 Infection With Residual Post‐COVID Symptoms,” Vaccines 11, no. 9 (2023): 1413, 10.3390/vaccines11091413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Trimbake D., Singh D., Y. G. K., Babar P., V. D. S., and Tripathy A. S., “Durability of Functional SARS‐CoV‐2‐Specific Immunological Memory and T Cell Response Up to 8–9 Months Postrecovery From COVID‐19,” Journal of Immunology Research 2025, no. 1 (2025): 9743866, 10.1155/jimr/9743866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Suryawanshi P., Patil‐Takbhate B., Athavale P., et al., “T‐Cell Responses in COVID‐19 Survivors 6− 8 Months After Infection: A Longitudinal Cohort Study in Pune,” Immunity, Inflammation and Disease 12, no. 6 (2024): e1238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Tripathy A. S., Singh D., Trimbake D., et al., “Humoral and Cellular Immune Response to AZD1222/Covishield and BV152/Covaxin COVID‐19 Vaccines Among Adults in India,” Human Vaccines and Immunotherapeutics 20, no. 1 (2024): 2410579, 10.1080/21645515.2024.2410579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Tripathy A. S., Vishwakarma S., Trimbake D., et al., “Pro‐Inflammatory CXCL‐10, TNF‐α, IL‐1β, and IL‐6: Biomarkers of SARS‐CoV‐2 Infection,” Archives of Virology 166, no. 12 (2021): 3301–3310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Davis H. E., McCorkell L., Vogel J. M., and Topol E. J., “Long COVID: Major Findings, Mechanisms and Recommendations,” Nature Reviews Microbiology 21, no. 3 (2023): 133–146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Yin K., Peluso M. J., Luo X., et al., “Long COVID Manifests With T Cell Dysregulation, Inflammation and an Uncoordinated Adaptive Immune Response to SARS‐CoV‐2,” Nature Immunology 25, no. 2 (2024): 218–225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Tripathy A. S., Wagh P., Vishwakarma S., et al., “Association of Human Leukocyte Antigen Class I and Class II Alleles and Haplotypes in COVID‐19 Infection in a Western Indian Population,” Infection, Genetics and Evolution 113 (2023): 105468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Yadav P. D., Sahay R. R., Salwe S., et al., “Broadly Reactive SARS‐CoV‐2‐specific T‐Cell Response and Participation of Memory B and T Cells in Patients With Omicron COVID‐19 Infection,” Journal of Immunology Research 2023 (2023): 8846953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. R Core Team , R Foundation for Statistical Computing. (R Foundation for Statistical Computing, 2024), https://www.r-project.org/. [Google Scholar]
- 16. Joerns E. K., Karp D., Zhang S., et al., “High Interleukin‐13 Level Is Associated With Disease Stability in Interstitial Lung Disease,” Heliyon 10, no. 11 (2024): e32118, 10.1016/j.heliyon.2024.e32118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Fiel Peres F., “Effect Sizes for Nonparametric Tests,” Biochem Med (Zagreb) 36, no. 1 (2026): 010101, 10.11613/BM.2026.010101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. https://cran.r-project.org/web/packages/rstatix/index.html.
- 19. Peluso M. J. and Deeks S. G., “Mechanisms of Long COVID and the Path Toward Therapeutics,” Cell 187, no. 20 (2024): 5500–5529, 10.1016/j.cell.2024.07.054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Korobova Z. R., Arsentieva N. A., Lyubimova N. E., and Totolian A. A., “Redefining Normal: Cytokine Dysregulation in Long COVID and the Post‐Pandemic Healthy Donors,” International Journal of Molecular Sciences 26, no. 17 (2025): 8432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Tsampasian V., Elghazaly H., Chattopadhyay R., et al., “Risk Factors Associated With Post− COVID‐19 Condition: A Systematic Review and Meta‐Analysis,” JAMA Internal Medicine 183, no. 6 (2023): 566–580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Makovski T. T., Steichen O., Rushyizekera M., van den Akker M., and Coste J., “Relationship Between Multimorbidity, SARS‐CoV‐2 Infection and Long COVID: A Cross‐Sectional Population‐Based French Survey,” BMC Medicine 23, no. 1 (2025): 222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Jacobs E. T., Catalfamo C. J., Colombo P. M., et al., “Pre‐Existing Conditions Associated With Post‐Acute Sequelae of COVID‐19,” Journal of Autoimmunity 135 (2023): 102991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Berends M. S., Homburg M., Kupers T., et al., “Impact of Pre‐Existing Comorbidities and Multimorbidities, Demography and Viral Variants on Post‐Acute Sequelae of COVID‐19 (‘Long COVID’) in Dutch Primary Care: A Retrospective Cohort Study,” International Journal of Infectious Diseases 156 (2025): 107912. [DOI] [PubMed] [Google Scholar]
- 25. Ashmawy R., Hammouda E. A., El‐Maradny Y. A., et al., “Interplay Between Comorbidities and Long COVID: Challenges and Multidisciplinary Approaches,” Biomolecules 14, no. 7 (2024): 835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Falsetti L., Zaccone V., Santoro L., et al., “The Relationship Between Post‐COVID Syndrome and the Burden of Comorbidities Assessed Using the Charlson Comorbidity Index,” Medicina 59, no. 9 (2023): 1583. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Queiroz M. A. F., Neves P. F. M., Lima S. S., et al., “Cytokine Profiles Associated With Acute COVID‐19 and Long COVID‐19 Syndrome,” Frontiers in Cellular and Infection Microbiology 12 (2022): 922422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Ganesh R., Yadav S., Hurt R. T., et al., “Pro Inflammatory Cytokines Profiles of Patients With Long COVID Differ Between Variant Epochs,” Journal of Primary Care and Community Health 15 (2024): 21501319241254751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Demir A. D., Cetin Z., and Sezgin F. M., “Serum FGF‐21 Levels During COVID‐19 Infection Recovery Period,” Medical Bulletin of Haseki 62 (2024): 168–174. [Google Scholar]
- 30. Wolszczak‐Biedrzycka B., Dorf J., Wojewódzka‐Żelezniakowicz M., et al., “Changes in Chemokine and Growth Factor Levels May Be Useful Biomarkers for Monitoring Disease Severity in COVID‐19 Patients: A Pilot Study,” Frontiers in Immunology 14 (2024): 1320362, 10.3389/fimmu.2023.1320362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Schultheib C., Willscher E., Paschold L., et al., “The IL‐1β, IL‐6, and TNF Cytokine Triad Is Associated With Post‐Acute Sequelae of COVID‐19,” Cell Reports Medicine 3, no. 6 (2022): 100663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Farooq M., Khan A. W., Kim M. S., and Choi S., “The Role of Fibroblast Growth Factor (FGF) Signaling in Tissue Repair and Regeneration,” Cells 10, no. 11 (2021): 3242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Low R. N., Low R. J., and Akrami A., “A Review of Cytokine‐Based Pathophysiology of Long COVID Symptoms,” Frontiers in Medicine 10 (2023): 1011936, 10.3389/fmed.2023.1011936. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Su Y., Yuan D., Chen D. G., et al., “Multiple Early Factors Anticipate Post‐Acute COVID‐19 Sequelae,” Cell 185, no. 5 (2022): 881–895.e20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Paranga T. G., Mitu I., Pavel‐Tanasa M., et al., “Cytokine Storm in COVID‐19: Exploring IL‐6 Signaling and Cytokine‐Microbiome Interactions as Emerging Therapeutic Approaches,” International Journal of Molecular Sciences 25, no. 21 (2024): 11411, 10.3390/ijms252111411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Zheng T., Gao R., Liu Y., et al., “T Cell‐Driven Sustained Inflammation and Immune Dysregulation Mimicking Immunosenescence for Up to Three Years Post‐COVID‐19,” Immunity and Inflammation 1, no. 1 (2025): 11, 10.1007/s44466-025-00012-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Marshall G. D., “The Pathophysiology of Postacute Sequelae of COVID‐19 (PASC): Possible Role for Persistent Inflammation,” Asia Pacific Allergy 13, no. 2 (2023): 77–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Chansaenroj J., Yorsaeng R., Posuwan N., et al., “Long‐Term Specific IgG Response to SARS‐CoV‐2 Nucleocapsid Protein in Recovered COVID‐19 Patients,” Scientific Reports 11, no. 1 (2021): 23216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Jansen E. B., Ostadgavahi A. T., Hewins B., et al., “PASC (Post Acute Sequelae of COVID‐19) is Associated With Decreased Neutralizing Antibody Titers in Both Biological Sexes and Increased ANG‐2 and GM‐CSF in Females,” Scientific Reports 14, no. 1 (2024): 9854. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.