Biological factors associated with long COVID and comparative analysis of SARS-CoV-2 spike protein variants: a retrospective study in Thailand

Study design and participants

This cross-sectional study collected retrospective data from confirmed COVID-19 cases who acquired the SARS-CoV-2 infection during the Delta and Omicron epidemic periods, as reported by the Department of Medical Science of the Ministry of Public Health in Thailand (DMSc, 2023). Data were analyzed to determine the difference in prevalence and characteristics of long COVID symptoms between the two periods and to identify which biological factors were associated with long COVID. A minimum sample size of 206 positive COVD-19 cases was required, which was calculated using the n4Studies application (https://www.facebook.com/n4Studies/) based on an estimation infinite population proportion method by setting a prevalence difference proportion of 16% (ONS, 2022), and an alpha error of 0.05. Based on the medical records of the confirmed SARS-CoV-2 RT-qPCR positive cases at Maharat Nakhon Ratchasima Hospital from June to November 2021 (Delta period) and from December 2021 to the first week of August 2022 (Omicron period), all confirmed COVID-19 cases who were ≥ 18 years old, infected by SARS-CoV-2 for the first time within these periods, and had recovered from the infection for a minimum of three months were included as eligible participants. They were approached at random via telephone between March and April 2023 and requested to provide verbal consent for a single remote interview via telephone. To reduce the recall bias that influenced the validity assessment, we conducted a primary evaluation of the memorization participants by asking the time period of their infections. Fortunately, all participants were able to recall the disease period. Individuals who were pregnant during the COVID-19 infection, missed contact, had no recollection of long COVID experiences, passed away prior to interview, or lacked the ability to communicate in Thai were excluded from the study.

A peer-evaluated questionnaire was employed to record and analyze retrospective self-report data from each participant during the interview process. The self-reported questionnaire asked about basic demographic data and the experience of COVID-19 disease, including COVID-19 severity, the presence of long COVID symptoms, drug(s) for treatment, and vaccine status. According to the World Health Organization (WHO, 2022a), the definition of long COVID in this study is the persistence of existing infection symptoms or the emergence of new symptoms within three months of the initial SARS-CoV-2 infection. These symptoms must persist for a minimum of two months without any other explanation. In this investigation, the spectrum of long COVID disorders is categorized into seven systems, including cardiovascular: chest pain, fast-beating or heart palpitations; neurology: headaches, lightheadedness, pins-and-needles feelings, change in smell or taste, difficulty thinking or concentrating; dermatology: hair loss, rash; psychology: depression or anxiety, sleep problems; respiratory: difficulty breathing or shortness of breath, coughing; generalized symptoms of fatigue and myalgia; and others: diarrhea, stomach pain, constipation, etc.

The questionnaire for COVID-19 severity was evaluated by employing the specific criteria classification explanation, which classified the condition as mild, moderate, or severe according to the NIH categories (NIH, 2022). Participants were asked regarding the drug(s) they used for treatment in the five most used categories in Thailand, which include Andrographolide herbal medicine, Flavipiravir, Molnupiravir, symptomatic medicines (any medical therapy of a disease that only affects its symptoms, not the underlying cause, such as antipyretic medicine, cold medicine, cough medicine, etc.), and steroids and others.

This study was approved by the Ethical Committee of the Maharat Nakhon Ratchasima Hospital Institutional Review Board (MNRH IRB), Ministry of Public Health, Thailand (Certificate No. 139/2022). Participant identifiers were removed from the data and replaced with a participant code. The collected data will be destroyed immediately after publication.

Comparing genetic variations in the spike gene between the Delta and Omicron variants

The Global Initiative for Sharing Avian Influenza Data (GISAID) database (https://gisaid.org) is a free, open-access database that is available online. Its purpose is to facilitate the rapid exchange of data from key pathogens, such as influenza, SARS-CoV-2, respiratory syncytial virus (RSV), Mpox virus, and arboviruses such as chikungunya, dengue, and Zika. The database also offers genetic sequence and related clinical and epidemiological data associated with human viruses, as well as geographical and species-specific data associated with avian and other animal viruses, to further the informative data on the evolution and transmission of viruses during epidemics and pandemics.

In this study, GISAID database was accessed for datasets of Delta and Omicron SARS-CoV-2 variants collected in Thailand between July 1, 2021 and August 31, 2022. We collected only complete sequence datasets, which included 206 Delta variant sequences and 292 Omicron sequences. These genetic sequences are derived from the study objectives, who have different backgrounds in terms of gender, clinical status, and living areas in Thailand. Mutations causing amino acid changes within the spike protein of the two variants were examined to compare the percentage of accumulated mutation frequency. Subsequently, key mutations within each variant defined as having a difference in the percentage of accumulated mutation frequency over 50% were reviewed to determine whether their roles associated with pathogenesis, transmission, and immune evasion of the SARS-CoV-2 variant might help to explain a long COVID outcome.

Statistical analysis

All data were analyzed using the Statistical Package for the Social Sciences (Version 25.0; SPSS Inc., Chicago, IL, USA). Calculation of frequency, percentage, mean and standard deviation was used to characterize the baseline demographic data, to compare the data of long COVID between the Delta and Omicron epidemic periods, and to compare the genetic variations in the spike protein of the SARS-CoV-2 variants. A binary logistic regression test with enter method was used to determine the association between biological factors and long COVID outcome. The strength of association was determined using an odds ratio at 95% confidence interval and a P-value < 0.05 was considered as significant association. Furthermore, the model summary or goodness of fit derived from a binary logistic regression analysis was elucidated by the values of -2Log Likelihood (-2ll) and Nagelkerke R² (Pseudo R²).

Demographic characteristics

Between June 2021 and August 2022, during the Delta and Omicron epidemic periods in Thailand, a total of 490 confirmed COVID-19 cases at Maharat Nakhon Ratchasima Hospital were randomly asked for informed consent to participate in a telephone interview. A total of 251 subjects enrolled voluntarily; however, 247 were selected for analysis based on the defined inclusion and exclusion criteria. The number of participants who were infected by SARS-CoV-2 during the Delta and Omicron epidemic periods was 106 and 141, respectively. The flowchart of the subject enrollment process is shown in Fig. 1.

The baseline demographics and clinical characteristics of confirmed COVID-19 cases of the Delta and Omicron epidemic periods were compared are shown in Table 1. The mean age of the participants was 40.75 ± 14.20 years during Delta and 35.61 ± 11.70 years during Omicron. Female gender, age range of 18 to 29 years, obesity, and blood group O were the most prevalent characteristics of the participants in both epidemic periods. Nevertheless, during the Delta period, more than half of the cases (n = 54, or 50.9%) involved immunocompromised hosts who had at least one underlying disease. Hypertension, diabetes, hypercholesterolemia, asthma, and allergies were common underlying conditions. In contrast, infection with SARS-CoV-2 during the Omicron period primarily affected healthy hosts (n = 125, 88.7%). Almost one-third of the Delta cases (n = 38, 35.8%) exhibited symptoms classified as moderate to severe, whereas the Omicron cases predominantly presented mild symptoms. The antiviral drug Favipiravir and symptomatic drug usage was the key treatment utilized by the participants during both variant periods. The majority of the participants affected during the Delta period had not yet been immunized against COVID-19 (n = 63, 59.4%). Conversely, a significant numbers of participants affected during the Omicron period had received at least three vaccinations (n = 80, 56.7%). In respect to the vaccine type, the inactivated vaccine was predominant during the Delta period (n = 24, 55.8%), whereas the mRNA vaccine was predominant during the Omicron period (n = 75, 56.4%). Significantly, throughout both epidemic periods, the majority of participants reported having sustained long COVID, with a number of symptoms having comparable prevalence rates between the Delta and Omicron periods.

The prevalence of self-reported long COVID symptoms between the Delta and Omicron epidemic periods was compared (Fig. 2). The seven most frequently reported symptoms in long-term COVID patients were as follows: cardiovascular, neurology, dermatology, psychology, respiratory, generalized symptoms of fatigue and myalgia, and others. Both participant groups reported all symptoms but the rate of observation for each symptom varied between the groups. Participants who were infected during the Delta period showed a predominant psychological disorder, including insomnia, anxiety, and depression. Conversely, abnormalities in the respiratory and neuron systems were prevalent during the Omicron period.

A comparison of the baseline demographics of the participants between non-long COVID cases and long COVID cases is presented in Table 2. Non-long COVID cases had a mean age of 40.86 ± 13.74 years and long COVID cases were 36.56 ± 12.59 years old. Comparative frequencies were observed for almost all analyzed variables in the two groups. Significantly, the prevalence of long COVID experience was greater during the Omicron period (60%) compared to the Delta period (40%); additionally, a greater proportion of participants with moderate to severe COVID-19 exhibited a higher rate of long COVID.

A binary logistic regression analysis was used to determine biological factors associated with long COVID. Variant epidemic period, age, treatment with symptomatic medicines, and vaccination status were found to be significant predictors of long COVID outcomes. The odds of developing long COVID were significantly higher for SARS-CoV-2 infection during the Omicron period [OR = 2.976 (95% CI [1.202–7.365]); p = 0.018] than for infection during the Delta period. A 1-year increase in age was correlated with a decreased likelihood of developing long COVID [OR = 0.969 (95% CI [0.942–0.996]); p = 0.025]. In comparison to other antiviral and immunosuppressive drugs, treatment with symptomatic medications for COVID-19 was associated with a higher risk of developing long COVID [OR = 3.804 (95% CI [1.149–12.590]); p = 0.029]. It is worth mentioning that in comparison to vaccination with at least one dose, non-vaccination was a significant risk factor for long COVID [OR = 6.434 (95% CI [1.253–33.033]); p = 0.026]. All ten independent variables, as shown in Table 2, possessed the potential to account for 19.3% of the long COVID outcome explanation. The remaining 80.7% of the variance could be accounted for by variables that were not included in the binary logistic regression analysis. The dataset for analysis in this part was available in File S1.

Genetic variations in the spike gene between the Delta and Omicron variants isolated in Thailand

As it has been hypothesized that the spike protein of SARS-CoV-2 is accountable for long COVID, particularly neuropsychiatric symptoms (Theoharides, 2022), the prevalence and characteristics of long COVID outcomes may be influenced differently by the mutations in this protein found in the Delta and Omicron variants. This analysis compared the mutation frequencies observed in each region of the spike protein across the two variants as shown in Fig. 3. Most amino acid changes based on the wild-type spike protein sequence were observed in the S1 subunit compared to the S2 subunit. A higher number of amino acid changes were found in nearly all regions of the S1 subunit of the Omicron variant in comparison to the Delta variant, except for the signal peptide (SP) region. The amino acid substitutions observed in the fusion peptide (FP) and heptad repeat 1 (HR1) region of the S2 subunit were similar between the two variants. No amino acid changes were observed in the cytoplasmic domain (CD) and protease cleavage site of the S2 subunit (S2′) for the Omicron variant.

The frequency of each mutation in the S1 and S2 spike protein subunits was compared between the two variants and is presented as heatmap data in Figs. 4 and 5. The predominant high-frequency mutations were found in the S1 subunit of both variants, especially in the regions of the N-terminal domain (NTD), receptor-binding domain (RBD), receptor-binding motif (RBM), and protease cleavage site between the S1 and S2 subunits (S1S2′). For the S2 subunit, high-frequency mutations were observed in the FP region of the Omicron variant and the HR1 region of both variants.

Finally, the literature review of the pathogenesis evidence or suspected evidence regarding the role of the key mutations of each variant, defined here as percentage frequency differences between variants greater than 50%, is shown in Table 3. The Omicron variant sequences obtained from Thailand had a larger number of key mutations compared to the Delta variant sequences. Most of the significant mutations identified in the Delta variant sequences were associated with ACE2 affinity binding enhancement, viral infectivity, and immune evasion. The sole mutation, D950B, located in the HR1 region of the S2 subunit, lacked any evidence regarding its function. The greater part of the key mutations in the Omicron variant exhibited comparable functions to those observed in the Delta variant. However, a few mutations played addition roles in facilitating spike cleavage, virus fusion, and virus transmissibility; spike function impairment had also been published. Additionally, the function of five mutations in the NTD and RBD regions of the S1 subunit remained unknown: L24del, P25del, P26del, A27S, and R408S. The dataset for analysis in this part was available in tFile S2.

Conclusion

The SARS-CoV-2 variant, age, treatment with symptomatic medications, and non-immunization status were identified as biological factors associated with long COVID progression in our retrospective cross-sectional study. The identification of these factors should help to facilitate the development of a suitable health management strategy. Furthermore, the analysis of the frequency and biological implications of mutations observed in the spike protein gene of the Delta and Omicron variants could support the explanation for long COVID development and provide a scientific reference for monitoring, prevention, and vaccine development.