Multiomic characterisation of the long-term sequelae of SARS survivors: a clinical observational study

Summary

Background

We aimed to characterise the long-term health outcomes of survivors of severe acute respiratory syndrome (SARS) and determine their recovery status and possible immunological basis.

Methods

We performed a clinical observational study on 14 health workers who survived SARS coronavirus infection between Apr 20, 2003 and Jun 6, 2003 in Haihe Hospital (Tianjin, China). Eighteen years after discharge, SARS survivors were interviewed using questionnaires on symptoms and quality of life, and received physical examination, laboratory tests, pulmonary function tests, arterial blood gas analysis, and chest imaging. Plasma samples were collected for metabolomic, proteomic, and single-cell transcriptomic analyses. The health outcomes were compared 18 and 12 years after discharge. Control individuals were also health workers from the same hospital but did not infect with SARS coronavirus.

Findings

Fatigue was the most common symptom in SARS survivors 18 years after discharge, with osteoporosis and necrosis of the femoral head being the main sequelae. The respiratory function and hip function scores of the SARS survivors were significantly lower than those of the controls. Physical and social functioning at 18 years was improved compared to that after 12 years but still worse than the controls. Emotional and mental health were fully recovered. Lung lesions on CT scans remained consistent at 18 years, especially in the right upper lobe and left lower lobe lesions. Plasma multiomics analysis indicated an abnormal metabolism of amino acids and lipids, promoted host defense immune responses to bacteria and external stimuli, B-cell activation, and enhanced cytotoxicity of CD8⁺ T cells but impaired antigen presentation capacity of CD4⁺ T cells.

Interpretation

Although health outcomes continued to improve, our study suggested that SARS survivors still suffered from physical fatigue, osteoporosis, and necrosis of the femoral head 18 years after discharge, possibly related to plasma metabolic disorders and immunological alterations.

Funding

This study was funded by the Tianjin Haihe Hospital Science and Technology Fund (HHYY-202012) and Tianjin Key Medical Discipline (Specialty) Construction Project (TJYXZDXK-063B, TJYXZDXK-067C).

Research in context.

Evidence before this study

We searched PubMed for follow-up studies on the long-term consequences of SARS published between January 1, 2003, and June 13, 2022, without applying any language restrictions. We used search terms (“SARS”) AND (“survivor∗” OR “recover∗” OR “persistent” OR “follow up” OR “discharge∗” OR “long term” OR “sequelae”). To our knowledge, most SARS follow-up studies are cross-sectional surveys and only a few longitudinal cohort studies have described the dynamic recovery of health outcomes in patients who survived hospitalisation with SARS. Furthermore, the molecular and cellular basis for the sequelae observed in survivors of SARS was lacking in all existing studies.

Added value of this study

To our knowledge, this is the most extensive and longest cohort study of SARS survivors, including an age-matched, sex-matched, and occupation-matched non-SARS control group. Fatigue was the most common symptom during follow-up. The other two common sequelae were osteoporosis and necrosis of the femoral head, with a poor score of hip function. The SF-36 score of SARS survivors 18 years after discharge was improved compared to 12 years after discharge. However, the score was still worse than the control group, mainly in physical and social function. We used metabolomic, proteomic, and single-cell RNA sequencing technologies to reveal several possible pathological mechanisms that underlie the clinical manifestations of SARS survivors 18 years after discharge.

Implications of all the available evidence

Social functioning continued to improve, and mental and emotional health recovered fully to normal 18 years after discharge. However, some sequelae, especially physical aspects (e.g., fatigue), of SARS can last for a lifetime, although a longer follow-up is needed in the future. Chest imaging tests did not indicate an improvement in residual lesions of pulmonary abnormalities between 12- and 18-year follow-ups. Osteoporosis and necrosis of the femoral head may result from host metabolic and immunological alterations, as well as high pulse administration of glucocorticoids. Future interventions targeting these alterations may improve the health outcomes of SARS survivors.

Introduction

Coronaviruses (CoVs) have caused several pandemics over the past two decades, including severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and the ongoing coronavirus disease 2019 (COVID-19). These have had a devastating impact on human health and the economy worldwide. The high fatality rates of SARS (13%) and MERS (35%) limited their transmissibility and were quickly contained.¹ The lack of deep understanding of the pathogenesis of coronavirus infections and insufficient research on optimizing treatment options for infected individuals are important reasons for our inability to cope with the current pandemic. The long-term effects of coronavirus infections have not been fully studied.

The first known case of SARS infection occurred in Foshan, China, in November 2002, and spread to 26 other countries.² Of 5663 individuals, including healthcare workers, confirmed to have SARS infection in China, 372 died.^3,4 The most common symptoms included persistent fever, rigor, dry cough, malaise, fatigue, headache, dyspnea, hypoxemia, and progressive chest radiographic changes.^5,6 One year after hospital discharge, the health status of SARS survivors was worse than that of the healthy group; about one-third still presented pulmonary function impairment.⁷ Lung thin-section computed tomography (CT) showed predominantly intralobular and interlobular septal thickening with ground-glass opacity in some SARS survivors seven years after recovery.⁸ Pulmonary lesions on CT scans and necrosis of the femoral head continued to diminish but remained in approximately 5% and 25% of SARS survivors, respectively, in 2018.⁹ These indicate that SARS infection causes long-term sequelae. However, information on how long these sequelae will last and on host immunity in SARS survivors is scarce.

We previously finished a cross-sectional study of SARS survivors 12 years after hospital discharge.¹⁰ In this study, we conducted a comprehensive 18-year follow-up of some of these survivors with respect to persistent symptoms, hip function, and pulmonary function. Plasma samples were assayed for metabolomic, proteomic, and single-cell RNA sequencing profiling to assess host immunity status in these survivors. The findings of this study may have implications in predicting the prognosis of coronavirus-induced pathologies, including post-acute sequelae of SARS-CoV-2 infection.

Methods

Study design

Among the 31 health workers who survived SARS-CoV infection between Apr 20, 2003 and Jun 6, 2003 in Haihe Hospital (Tianjin, China), we performed a clinical observational study of 14 survivors in 2021 to determine whether they fully recovered after 18 years. Control individuals were also health workers from our hospital who were not infected with SARS-CoV. Controls and SARS survivors were individually matched 1:1 on age, sex, and occupation (health workers from Haihe Hospital, Tianjin, China) manually, and the occurrence of SARS infection was taken as the grouping basis. Our previously published dataset on the health outcomes of the same 14 patients at the 12-year follow-up was adopted to show healthy improvements of SARS survivors between two follow-up visits.¹⁰ The Research Ethics Commission of Haihe Hospital approved this study (2021HHKT-006). Written informed consent was collected from SARS survivors and control participants.

Health assessment

SARS survivors and the controls participated in face-to-face interviews in the outpatient clinic of Haihe Hospital. Both groups underwent physical examination and routine laboratory tests, and completed questionnaires, including a self-reported symptom questionnaire, the modified British Medical Research Council (mMRC) dyspnea scale (scores range from 0 to 4; a higher score indicates worse dyspnea),¹¹ the Borg scale (scores range from 0 to 10; a higher score indicates worse dyspnea or fatigue),¹² manual muscle test (MMT),¹³ Harris hip joint function assessment, and the Short Form 36-item Health Survey (SF-36). To limit recall bias, we explained each item of questionnaire clearly to each participant and each of them had enough time to finish the questionnaire with a family member. Pulmonary function tests were performed using an automatic pulmonary function testing system (Master Screen Body [Jaeger ms-pft analysis unit, Wurzburg, Germany]) according to the standards of the American Thoracic Society.¹⁴ Chest CT images were reviewed by three radiologists. Statistical frequency of CT manifestations according to the Fleischner Association chest CT manifestation classification criteria.¹⁵ In the event of inconsistency in interpretation among the three radiologists, clinical data and examination results were discussed in a group discussion until a consensus was reached and a conclusion was made. Using semi-quantitative visual scoring method, CT images of single lung lobe were scored according to the area percentage of single lobe lesions.^16,17 Lobes without lesions were scored as 0, and lobes with percentages of lesion area of <25%, 25% to <50%, 50% to <75%, and ≥75% were scored 1, 2, 3, and 4, respectively. The total score for the five lobe categories ranges from 0 to 20. Each CT image was independently reviewed and scored by three radiologists, and the average of the scores represented the final CT image score.^18,19

Multiomics analysis

Venous blood samples (4 mL) were collected from each participant into a BD Vacutainer EDTA tube (BD, Cat#:367861). Half the sample was processed for plasma metabolomics and proteomic profiling, and peripheral blood mononuclear cells (PBMCs) were isolated from the other half for single-cell RNA sequencing (scRNA-seq).20, 21, 22 Due to technical issue, we did not get enough serum samples from two patients, so metabolomics analysis was performed in only 12 patients.

Statistical analysis

Data following a normal distribution are expressed as mean ± standard deviation (x ± s), and comparisons between groups were performed using an independent-sample t-test and paired t-test. Data not following a normal distribution are presented as medians (IQR, 25th percentile, 75th percentile), and the Mann–Whitney U-test was used for comparison between groups. Count data are expressed as percentages (%) and compared using the χ² test. Data normality assessment and statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA). P-Values <0.05 were considered statistically significant.

For metabolomic analysis, a heatmap of the hierarchical clustering analysis was generated using pheatmap (version 1.0.12). Volcano plots were obtained using ggplot2 (version 3.3.5). Boxplots were constructed using GraphPad Prism (version 9.4.0). Violin plots were constructed using the Seurat package. Differential abundance scoring based on KEGG was performed using ggplot2 (version 3.3.5) and KEGG (http://www.genome.jp/kegg/). For proteomic analysis, Gene Ontology (GO) analysis revealed biological processes using ggplot2 (version 3.3.2) based on the GO database (http://geneontology.org/), and the q value was calculated using fdrtool (version 1.2.15, q value < 0.05). For scRNA-seq, differentially expressed genes of each cell type were identified using Seurat package function of ‘FindMarkers’. P values were performed using the wilcoxon rank sum test, then adjusted by bonferroni correction based on the total number of genes in dataset. GO analysis revealed BPs using DAVID (version 6.8, https://david.ncifcrf.gov/), and p-values were calculated using the chi-square test. The p values of each cell type based on single-cell RNA-seq were calculated using Student's t-test (p < 0.05). Principal component analysis (PCA) was performed over metabolomic, proteomic, and single cell RNA transcriptional data using the R statistical language (version 4.2.0) to analyse the differences between SARS survivors and control individuals.

Role of the funding source

Funders had no role in the design and conduct of the study, and preparation and submission of the manuscript for publication. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Results

Health outcomes

Among 31 SARS survivors, ten refused follow-up visits and seven were unable to come to finish follow-up visits because of COVID-19 restrictions; 14 SARS survivors completed the assessments and were included in the final analysis to determine whether they fully recovered after 18 years (Fig. S1). These survivors, during the acute SARS episode, received an average dose of 5167 mg methylprednisolone for an average of 35 days. The median age was 61.0 (54, 69) years, and ten patients (71.4%) were women (Table 1). Two (15.4%) had a history of smoking. Eight (57.1%) survivors had hypertension and six (42.9%) survivors had type 2 diabetes. Fatigue was the most common symptom in SARS survivors (50%) and was significantly higher than that in the control group (7.1%) (p = 0.04). Regarding respiratory function of SARS survivors at 18 years after discharge, the mMRC and Borg scores were worse than those of controls. The proportion of dyspnea, defined by an mMRC score of 1 or more, in SARS survivors at 18 years after discharge was 10 (71.4%) compared to 1 (7.1%) in controls (p = 0.001). The proportion of those with a Borg score of 1 or more among SARS survivors at 18 years after discharge was 10 (71.4%), compared to 3 (21.4%) in controls (p = 0.04). Scores in most SF-36 domains were lower in SARS survivors at 18 years after discharge compared to those in controls (Table S1). Chest CT showed that the number of affected lobes among SARS survivors at 18 years after discharge was higher than that in controls (p = 0.02), especially the right upper (p = 0.001) and the left lower lobes (p = 0.02) (Table 1). The semi-quantitative visual score showed that the total score of the SARS survivors at 18 years after discharge was higher than that of controls (p = 0.02). Pulmonary function test results showed no difference in diffusion or ventilation between SARS survivors at 18 years after discharge and controls. Blood gas analysis showed that the alveolar-arterial oxygen gradient (Aa-DO2) was comparable between the SARS survivors at 18 years after discharge and controls (Table S2). The most common complications in SARS survivors at 18 years after discharge were osteoporosis (p < 0.001) and necrosis of the femoral head (p = 0.048) (71.4% and 35.7%, respectively) (Table 1). The Harris hip function assessment of SARS survivors at 18 years after discharge showed a lower score than that in controls (p < 0.001) (Table 1). Blood lipid indicators, liver function, renal function, tumor markers, thyroid function, blood routine and urine routine indicators of SARS survivors were all within the normal range 18 years after discharge. In addition, there were no significant differences in blood glucose levels between SARS survivors at 18 years after discharge and controls (Tables S3–S8).

Table 1.

Demographic and clinical characteristics of SARS survivors 18 years after infection and control group.

Characteristics	SARS survivors at 18 years after discharge (n = 14)	Control (n = 14)	P value
Age, Median, years	61 (54,69)	62 (54,68)	0.98
Sex (%)			1.00
Male	4 (28.6)	4 (28.6)
Female	10 (71.4)	10 (71.4)
Smoking history	2 (15.4)	4 (28.6)	0.72
Comorbidity (%)
Hypertension	8 (57.1)	4 (28.6)	0.13
Type 2 diabetes	6 (42.9)	3 (21.4)	0.42
Coronary heart disease	5 (35.7)	2 (14.3)	0.38
Femoral head necrosis	5 (35.7)	0 (0)	0.048
Osteoporosis	10 (71.4)	1 (7.1)	<0.001
Symptoms (%)
Myalgia or fatigue	7 (50.0)	1 (7.1)	0.04
Exertional dyspnea	3 (21.4)	0 (0)	0.22
Cough	3 (21.4)	5 (35.7)	0.68
Expectoration	3 (21.4)	4 (28.6)	1.000
Diarrhea	2 (14.3)	0 (0)	0.46
Poor appetite	2 (14.3)	0 (0)	0.46
Function assessment
Modified Medical Research Council score			0.001
0	4 (28.6)	13 (92.9)
≥1	10 (71.4)	1 (7.1)
Borg dyspnea scale			0.04
0	4 (28.6)	11 (78.6)
≥1	10 (71.4)	3 (21.4)
Manual muscle test score
5	9 (64.3)	12 (85.7)	0.43
≤4	5 (35.7)	2 (12.3)
Harris score	72 (49.5,93.3)	100 (95.5,100)	<0.001
SF-36 Quality of Life assessment
Total score	110 (78,124)	135 (127,145)	0.001
Pulmonary function
TLC (%)	93.02 ± 27.14	90.72 ± 9.89	0.78
FVC of predicted/%	90.45 ± 17.15	96.84 ± 15.49	0.33
FEV1 of predicted/%	92.28 ± 15.20	92.79 ± 13.68	0.93
(FEV1/FVC)/%	84.65 ± 8.18	85.67 ± 17.64	0.85
FEV1/VCMAX/%	105.71 ± 12.55	99.54 ± 9.89	0.18
FEF50 (%)	94.04 ± 31.63	85.10 ± 27.79	0.45
FEF75 (%)	103.27 ± 24.38	100.74 ± 22.68	0.79
MMEF (%)	81.85 ± 34.49	76.05 ± 21.59	0.61
DLCO-SB (%)	81.38 ± 26.19	82.86 ± 15.12	0.87
Chest CT
Affected lung lobes	2.71 ± 1.44	1 (0,2)	0.02
The right upper lobe	10 (71.4%)	1 (7.1%)	0.001
The right middle lobe	4 (28.6%)	5 (35.7%)	1.00
The right lower lobe	9 (64.3%)	4 (28.6%)	0.13
The left upper lobe	5 (35.7%)	6 (42.9%)	1.00
The left lower lobe	10 (71.4%)	3 (21.4%)	0.02
Consolidation opacities	2 (14.3%)	0 (0)	0.48
Ground-glass opacities	7 (50%)	4 (28.6%)	0.44
Ground-glass opacities as the main performance	7 (50%)	4 (28.6%)	0.44
Consolidation opacities as the main performance	1 (7.1%)	0 (0)	1.00
Subpleural relatively unaffected	10 (71.4%)	6 (42.9%)	0.25
Lobular core nodules	8 (57.1%)	1 (7.1%)	0.01
Semiquantitative visual score
The right upper lobe lesions	1 (0.25,1)	0 (0,0)	0.003
The right middle lobe lesions	0 (0,0.75)	0 (0,1)	0.77
The right lower lobe lesions	1 (0,1)	0 (0,0.75)	0.09
The left upper lobe lesions	0 (0,1)	0 (0,1)	0.77
The left lower lobe lesions	1 (0.25,1)	0 (0,0)	0.02
Total score	3 (1.25,3.75)	1(0,2)	0.02

Questionnaire indicated that four of 14 enrolled SARS survivors claimed one hospital visit because of pneumonia between 12 and 18 years after discharge. While in the control group, only one had pneumonia during the same period. The total SF-36 score increased at 18 years compared to that at 12 years after discharge (p < 0.001) (Table 2). Scores of physical roles (p < 0.001), vitality (p = 0.01), social functioning (p < 0.001), mental health (p < 0.001), and role emotional (p < 0.001) were improved at 18 years, as compared to those at 12 years (Table S9). Pulmonary function tests showed no significant difference in the proportion of lung diffusion impairment over time in SARS survivors 18 and 12 years after discharge (Table 2). Blood gas analysis showed similar Aa-DO2 in SARS survivors at 18 years and at 12 years (Table S10). Chest CT showed that seven SARS survivors (50%) had ground-glass opacities, the most common remaining imaging abnormality (Table 2). The Harris hip function score exhibited no significant differences compared to that at 12-year visit. The blood glucose level of SARS survivors 18 years after discharge was higher than that at 12 years (p = 0.041) (Table S12). Blood lipids, liver function, renal function, and routine blood indicators of SARS patients at 18 and 12 years after discharge were all within the normal physiological range (Tables S11–S13).

Table 2.

Clinical characteristics of SARS survivors in 12 years and 18 years after infection.

Characteristics	SARS survivors at 12 years after discharge (n = 14)	SARS survivors at 18 years after discharge (n = 14)	P value
Harris score	62.14 ± 15.39	70.07 ± 21.57	0.23
SF-36 Quality of Life score	33.98 ± 19.13	101.7 ± 29.37	<0.001
Pulmonary function
TLC (%)	80.01 ± 10.35	94.40 ± 27.86	0.09
FVC of predicted/%	95.74 ± 13.30	91.61 ± 17.37	0.31
FEV1 of predicted/%	94.91 ± 11.24	93.33 ± 15.38	0.52
(FEV1/FVC)/%	83.09 ± 8.09	84.49 ± 8.52	0.48
FEV1/VCMAX/%	104.68 ± 10.62	105.24 ± 12.99	0.84
FEF50 (%)	93.12 ± 24.86	94.98 ± 32.84	0.80
FEF75 (%)	75.79 ± 33.00	103.75 ± 26.40	0.01
MMEF (%)	87.25 ± 23.06	81.73 ± 36.02	0.59
VC (%)	95.81 ± 14.51	85.16 ± 29.37	0.16
Chest CT
Affected lung lobes	2.71 ± 1.44	2.29 ± 1.38	0.11
The right upper lobe	10 (71.4%)	9 (64.3%)	1.00
The right middle lobe	4 (28.6%)	3 (21.4%)	1.00
The right lower lobe	9 (64.3%)	8 (57.1%)	1.00
The left upper lobe	5 (35.7%)	3 (21.4%)	0.68
The left lower lobe	10 (71.4%)	9 (64.3%)	1.00
Consolidation opacities	2 (14.3%)	3 (21.4%)	1.00
Ground-glass opacities	7 (50%)	7 (50%)	1.00
Ground-glass opacities as the main performance	7 (50%)	7 (50%)	1.00
Consolidation opacities as the main performance	1 (7.1%)	1 (7.1%)	1.00
Subpleural relatively unaffected	10 (71.4%)	12 (85.7%)	0.65
Lobular core nodules	8 (57.1%)	6 (42.9%)	0.71
Semi-quantitative visual score of lung lesions
The right upper lobe lesions	1 (0.25,1)	1 (0,1)	0.77
The right middle lobe lesions	0 (0,0.75)	0 (0,0)	0.77
The right lower lobe lesions	1 (0,1)	1 (0,1)	0.64
The left upper lobe lesions	0 (0,1)	0 (0,0)	0.54
The left lower lobe lesions	1 (0.25,1)	1 (0,1)	0.60
Total score	2.93 ± 1.94	2.29 ± 1.38	0.06

Metabolism

An untargeted plasma metabolomics approach was used for the plasma metabolism analysis of the 12 SARS survivors at 18 years after discharge and 12 control participants. Principal component analysis (PCA) from SARS survivors at 18 years after discharge and controls were dispersed similarly (Fig. S2A). A total of 19 differentially expressed metabolites (DEMs) were identified (Fig. S2B), including six upregulated metabolites and 13 downregulated metabolites (Fig. 1A). SARS survivors had fewer DEMs in plasma at 18 years than at 12 years after infection.¹⁰ Most amino acid metabolic pathways, including tyrosine metabolism, arginine biosynthesis, lysine degradation, and arginine and proline metabolism, were upregulated in SARS survivors at 18 years after discharge compared to those in controls (Fig. 1B). However, the lipid metabolic pathways, including metabolism of glycerophospholipid, sphingolipid, alpha-linolenic acid, linoleic acid, and arachidonic acid, were downregulated (Fig. 1B). In particular, an increase in ornithine and a decrease in its product N-methylputrescine, an important component of the urea cycle, were observed in the plasma of SARS survivors at 18 years after discharge compared to those in controls (Fig. 1A and B). As ornithine is a metabolic biomarker of osteoporosis,²³ these data provide a possible explanation for osteoporosis as the most common complication of SARS survivors 18 years after infection. Because ornithine increases growth hormone levels and accelerates muscle tissue production, elevated plasma ornithine levels may be a restorative response to fatigue in these survivors.²⁴

Fig. 1.

Plasma metabolomic profiling of SARS survivors 18 years after discharge. Panel A shows the heatmap of 19 differentially expressed metabolites screened in plasma samples from SARS survivors 18 years after discharge versus controls. Red indicates that the substance was highly expressed in the corresponding group, while blue indicates that it was low. Panel B shows differential abundance score based on KEGG. Different colors indicate different metabolic classifications of pathways. The positive line segments indicate the overall up-regulation of this pathway; otherwise, it indicates the down-regulation. The size of the end point of the line segment indicates the amount of substance annotated in the pathway.

Humoral immune defense response

A data-independent acquisition (DIA) proteomics strategy was used for plasma proteomics analysis, and 611 proteins were quantified. PCA score plots showed similar cloudy separations between SARS survivors at 18 years after discharge and controls (Fig. S3A). Thirty-two differentially expressed proteins (DEPs) were identified between SARS survivors at 18 years after discharge and controls, including 22 up-regulated and 10 down-regulated proteins (Fig. 2A and Fig. S3B). There were no significant enrichments in GO analysis based on the downregulated DEPs. However, the GO terms of upregulated DEP enrichment were classified into eight types of biological processes (BPs) (Fig. 2B). B cell activation and humoral immune response-related BP types, including “Phagocytosis”, “Complement activation”, “B cell activation and immunoglobin-mediated humoral immune response”, “Immune effector process” and “Lymphocyte activation”, were annotated in the functional classification of GO enrichments of upregulated DEPs (Fig. 2B). Similarly, the levels of immunoglobulin heavy variable (IGH) IGHV1-2 (p = 0.037), IGHV2-26 (p = 0.037), IGHV3-64D (p = 0.0046), IGHV4OR15-8 (p = 0.0029), and IGHV5-10-1 (p = 0.037) were elevated in SARS survivors at 18 years after discharge compared to that in controls (Fig. 2C), suggesting that B cell activation and immunoglobin-mediated humoral immune responses remain in SARS survivors at 18 years after discharge.

Fig. 2.

Plasma proteomic profiling of SARS survivors 18 years after discharge. Panel A shows the heatmap of 32 differentially expressed proteins (DEPs) screened out in plasma samples from SARS survivors at 18 years after discharge versus controls. Red indicates that the protein was highly expressed in the corresponding group, while blue indicates that it was low. Panel B shows Gene ontology (GO) analysis of DEPs (q value < 0.05). The color bars are based on different functional classifications. Panel C shows Boxplots of DEPs related to “B cell activation and immunoglobin-mediated humoral immune response,” Panel D shows Boxplots of DEPs related to defense response to “bacteria and external stimulus”.

Furthermore, the types of “Defense response to bacterium” and “Defense response to external stimulus” were enriched according to upregulated proteins. Plasma levels of DEFA1 (p = 0.00054), a member of the DEFA family of antimicrobial and cytotoxic peptides involved in host defense,²⁵ as well as RNASE1 (p = 0.02) and HSP90AB1 (p = 0.04) levels (Fig. 2D), included in the GO annotations of “defense response to bacterium” and “positive regulation of response to stimulus,” respectively, were elevated in SARS survivors at 18 years after discharge compared to those in controls. These data suggested that the immune defense response remain activated in SARS survivors at 18 years after discharge.

Lymphocyte and monocyte functions

To investigate cellular immunity function in SARS survivors 18 years after infection, PBMCs obtained from fresh peripheral blood samples of six SARS survivors at 18 years after discharge and six age- and sex-matched controls were subjected to scRNA-seq analysis. In total, 99,383 cells with high-quality transcriptomes were retained after cell filtration. These clusters were readily assigned to several cell lineages and annotated in the t-SNE plots of total PBMCs (Figs. S4–S9, Fig. 3A and B). Of all the identified cell subtypes, the only change was the proportion of effector memory CD8⁺ cytotoxic T cells, which decreased significantly (p = 0.039) in SARS survivors at 18 years after discharge compared to that in controls (Fig. 3C and D, Fig. S10A and B).

Fig. 3.

Single-cell RNA-seq profiling of peripheral blood mononuclear cells of SARS survivors 18 years after discharge. Panel A shows tSNE plots of 99,383 cells of peripheral blood mononuclear cells from SARS survivors at 18 years after discharge (n = 6) and controls (n = 6). Each color represents a cell type. Panel B shows sample origin. Red: Control group (50,630 cells). Blue: group of SARS survivors at 18 years after discharge (48,753 cells). Panel C shows the fraction of cells from each population. Each color represents a population. Panel D shows the percentage of effector memory CD8⁺ T cells. Panel E shows DEGs of effector memory CD8⁺ T cells related to “T cell activation, survival and generation” and “Cell killing.” Panel F shows DEGs of effector memory CD4⁺ T cells related to “APP and viral entry into host cell.” Panel G shows DEGs of central memory CD4⁺ T cells related to “Regulation of cell proliferation.” Panel H shows DEGs of memory B cells related to “B cell activation and immunoglobin mediated humoral immune response.” Panel I shows Gene ontology (GO) analysis of DEGs from classical monocytes (chi-square test, p value < 0.05). The color bars are based on different functional classifications. Panel J shows DEGs of classical monocytes related to “Antimicrobial humoral immune response and defense response to external stimuli.” Panel K shows GO analysis of DEGs from intermediate monocytes (p value < 0.05). Panel L shows DEGs of intermediate monocytes related to “osteoclast differentiation” and “cellular response to hormone stimuli.” Panel M shows GO analysis of DEGs from nonclassical monocytes (p value < 0.05). Panel N shows DEGs of nonclassical monocytes related to “osteoclast differentiation” and “cellular response to hormone stimuli.” For genes in E, G, I, and K, red pointing arrows indicate upregulated genes or activated function, blue pointing arrows indicate downregulated genes or inhibited function. P-values were calculated using the Wilcoxon rank-sum test. The p value was adjusted by Bonferroni correction based on the total number of genes in the dataset and differentially expressed genes with an adjusted p value < 0.05.

The expression of the ribosomal protein S26 (RPS26) (adjusted p = 0) was significantly down-regulated in SARS survivors at 18 years after discharge compared to that in controls, while the expressions of the JunB proto-oncogene (JUNB) (adjusted p = 2.37 × 10⁻¹¹⁸) and the Jun proto-oncogene (JUN) (adjusted p = 9.44 × 10⁻⁸⁸) were significantly up-regulated, suggesting that effector memory CD8⁺ T cell survival and generation were restrained in SARS survivors at 18 years after discharge, consistent with the decreased percentage of effector memory CD8⁺ T cells (Fig. 3D and E). Effector memory CD8⁺ T cells exhibited elevated expressions of the Fc gamma receptor IIIa (FCGR3A) (adjusted p = 8.11 × 10⁻¹⁵), granzyme B (GZMB) (adjusted p = 1.69 × 10⁻²⁶), and granulysin (GNLY) (adjusted p = 3.58 × 10⁻²³) in SARS survivors at 18 years after discharge compared to those in controls (Fig. 3E). Additionally, the expression of granzyme K (GZMK) (adjusted p = 1.23 × 10⁻³¹) was downregulated. These results suggest that the activation and cell-killing ability of effector memory CD8 + T cells were enhanced in SARS survivors 18 years after infection, although degressive survival and proportion were observed. Furthermore, effector memory CD4⁺ T cells showed lower expression of HLA class II DR beta 1 (HLA-DRB1) (adjusted p = 2.74 × 10⁻³⁹) and CD74 (adjusted p = 5.01 × 10⁻²⁴) in SARS survivors at 18 years after discharge than in controls (Fig. 3F). For central memory CD4⁺ T cells, JUNB (adjusted p = 1.87 × 10⁻⁴⁶) and JUN (adjusted p = 3.03 × 10⁻⁵¹) expression was up-regulated in SARS survivors at 18 years after discharge (Fig. 3G). These results indicate the reduced immune function of effector memory CD4⁺ T cells and central memory CD4⁺ T cells in SARS survivors at 18 years.

For memory B cells, we observed up-regulated expression of HLA class II DQ alpha 1 (HLA-DQA1) (adjusted p = 1.97 × 10⁻¹³) and CD1C (adjusted p = 1.24 × 10⁻⁷) in SARS survivors at 18 years after discharge, suggesting that B cell activation was enhanced in these survivors, consistent with the findings observed in plasma proteomic analysis (Fig. 2, Fig. 3H). In contrast to the elevated levels of IGHV gene-coded proteins in plasma proteomics analysis (Fig. 2C), the transcript expression of immunoglobulin heavy constant alpha 1 (IGHA1) (adjusted p = 1.27 × 10⁻²) and immunoglobulin heavy constant gamma 3 (IGHG3) (adjusted p = 2.92 × 10⁻⁴) in memory B cells was downregulated in SARS survivors at 18 years after discharge (Fig. 3H), indicating a negative feedback between the transcription and translation of these genes in memory B cells.

For classical monocytes, upregulated DEGs were enriched in the GO terms related to “Antimicrobial humoral immune response” and “Defense response to external stimulus” (Fig. 3I). The expressions of pro-platelet basic protein (PPBP) (adjusted p = 1.40 × 10⁻⁶⁷), platelet factor 4 (PF4) (adjusted p = 7.79 × 10⁻³⁹), JUN(adjusted p = 3.23 × 10⁻⁵²), FosB proto-oncogene (FOSB) (adjusted p = 2.71 × 10⁻³⁷), and resistin (RETN) (adjusted p = 1.67 × 10⁻¹¹⁹) were up-regulated in SARS survivors at 18 years after discharge compared to those in controls (Fig. 3J), suggesting the antimicrobial humoral immune response was enhanced and the defense response to external stimuli was promoted, consistent with the results of the proteomic analysis (Fig. 2C and D). DEGs of intermediate monocytes were enriched in GO terms related to “osteoclast differentiation” and “cellular response to hormone stimuli”, as well as non-classical monocytes (Fig. 3K, L, and M). The transcript expressions of JUNB (intermediate monocyte adjusted p = 0.015, non-classical monocytes adjusted p = 2.48 × 10⁻²⁴), Fos proto-oncogene (FOS) (intermediate monocyte adjusted p = 0.016, non-classical monocytes adjusted p = 1.55 × 10⁻¹⁴), and dual specificity phosphatase 1 (DUSP1) (intermediate monocyte adjusted p = 0.005, non-classical monocytes adjusted p = 5.63 × 10⁻¹⁹) in intermediate and nonclassical monocytes were up-regulated in SARS survivors at 18 years after discharge compared to those in controls (Fig. 3N). These results provide a molecular basis for increased osteoclast differentiation and femoral head necrosis in SARS survivors 18 years after discharge.

Discussion

To the best of our knowledge, this is the longest follow-up study that systematically and comprehensively describes the longitudinal evolution of health and functional outcomes of SARS survivors for up to 18 years. More than half the SARS survivors still had symptomatic sequelae at 18 years post-discharge, and the most common symptom was fatigue. Fatigue is a common symptom in patients with SARS-CoV infection, and a high prevalence of fatigue has also been identified in SARS survivors during the early recovery phase after discharge.^26,27 Fatigue may be caused by a variety of factors such as limited lung function, gas exchange disorders, muscle disorders, psychosocial factors, and host immunological alterations.²⁸

SARS survivors continued to improve at 18 years compared to 12 years after discharge. Emotional and mental health restored to the normal. However, SARS survivors still had lower quality of life 18 years after discharge than controls, mainly in physiological functioning, physical role, bodily pain, general health, vitality, and social functioning. Residual lung lesions mainly involved the upper lobe of the right lung and the lower lobe of the left lung. In addition, the lesions were mainly central-lobular nodules. Survivors of SARS and MERS also showed fibrotic abnormalities during convalescence.^29,30 A follow-up study suggested that the pulmonary lesions caused by SARS-CoV infection improved mainly within 1 year after discharge but remained unchanged up to 15 years.⁹ Our study confirmed that such lesions in the lungs last up to 18 years after discharge. However, the pulmonary ventilation and gas diffusion capacity of these survivors returned to normal levels.

The most common complications of SARS survivors 18 years after discharge were osteoporosis and necrosis of the femoral head. This is partially due to using high-dose glucocorticoids in the initial treatment for these patients because glucocorticoids inhibit osteoblast production and promote apoptosis of osteoblasts and osteocytes. In addition, indirect effects of glucocorticoids on bone metabolism also involve in the development of glucocorticoid-induced osteoporosis.³¹ The goal of osteoporosis management is to prevent fractures. Bisphosphonates are available to lower fracture risk by reducing bone resorption.³² Preventive measures could be undertaken for other SARS survivors, especially for osteoporosis.

Harris hip score of these SARS survivors was significantly lower than that of the normal population. There was no improvement in this aspect from 12 to 18 years post-discharge. Femoral head necrosis and osteoporosis are direct consequences hindering osteoblast proliferation and/or excessive osteoclast differentiation.^33,34 The gene expressions of JUNB and FOS, known to promote osteoclast differentiation,35, 36, 37, 38 were higher in the intermediate and non-classical monocytes of SARS survivors 18 years after infection than those in controls. Moreover, the response to hormone stimulus in these monocytes was also enhanced, in addition to the ornithine-induced hormone activation. This suggests that fatigue elicits a series of negative feedback host responses during recovery from SARS-CoV infection.

Although abundance decreased compared to that in control individuals, CD8⁺ T cells exhibited stronger cytotoxicity in SARS survivors. However, antigen processing and presentation, as well as viral entry into host cells, were suppressed in the effector memory CD4⁺ T cells of these survivors. In addition, the proliferation of central memory CD4⁺ T cells was restrained. The severity of SARS is also associated with the intense IgG response to a greater extent.³⁹ Some SARS survivors showed a decreasing level of IgG antibodies against SARS-CoV three years after discharge, as well as memory B cell-based responses.⁴⁰ Individual differences exist in the protective humoral response to SARS-CoV infection as anti-SARS-specific IgG antibodies were not detected in some SARS survivors six years after discharge but persisted in other SARS survivors for up to 15 years.^41,42 Furthermore, both plasma proteomic and scRNA-seq analyses provided evidence suggesting activation of memory B cells and enhancement of the host defense response to bacteria and external stimuli in SARS survivors 18 years after discharge. These survivors had fewer plasma differential metabolites and enhanced amino acid metabolism as compared with upregulated lipid metabolism at 12 years, reflecting changes in the body's response with time.

The strengths of our study are its longitudinal design to evaluate a long-term recovery status of SARS survivors. Furthermore, we integrated plasma metabolomic, proteomic, and single-cell transcriptomic analysis to reveal possible mechanisms underlying the long-term sequelae observed in SARS survivors 18 years after discharge.

Our study had several limitations. First, the number of individuals enrolled in this study was still small despite our efforts. We compared the demographic information and clinical data (including sex, age, comorbidities, and steroid use) of 14 SARS patients enrolled in this study with the corresponding data of all 31 SARS survivors at our hospital. The results showed that there were no statistical differences in the demography and clinical status between the 14 enrolled SARS survivors and 31 SARS survivor (Table S14). Additionally, none of the 17 non-enrolled SARS survivors did not participate in the follow-up study because of serious illness problems or death. Second, this was a single-centre study. Third, it was difficult to determine whether the observed abnormalities are specific to SARS without a control group with other lung infections. Control individuals were not matched by comorbidity because these survivors suffered from various SARS infection-associated sequelae including hyperlipidemia and cardiovascular abnormality. And the wellbeing of these survivors before SARS infection may also have impacts. Fourth, the clinical details and the severity of the SARS patients could not be retrieved. Fifth, Since the 4 SARS patients had pneumonia 12–18 years after the SARS episode, the abnormalities in amino acid, lipid metabolism and immunological parameters could be affected by these subsequent episodes of pneumonia, and therefore may not be entirely due to the SARS episode in 2003. Sixth, multivariate analysis cannot be conducted to adjust for comorbidities because of the small number of enrolled cases. Lastly, it was also not possible to distinguish whether the reported innate/adaptive immune function differences reflect sequelae of the infection versus basal biological distinctions that predisposed that to SARS in the first place.

To conclude, SARS survivors continued to recover both physically and mentally. Mental and emotional health recovered fully to normal 18 years after discharge, but some physical aspects were still worse compared with other health-care workers. Therefore, future follow-up is required to characterise the prolonged natural history of SARS.

Contributors

Dr HC and Dr LL have accessed and verified all the data in this study. Dr HC is the lead contact and takes responsibility for the integrity of the data. HC, HS, ZX, and LL were responsible for study concept and design. KL, QW, and HL were responsible for acquisition of data. HC, KL, and QW were responsible for analysis of data. HC, KL, QW, HL, HS, ZX, and LL were responsible for interpretation of data. HC, KL, and QW were responsible for drafting of the manuscript. HC, KL, QW, HS, ZX, and LL critically revised important knowledge contents. Statistical analysis was performed by HC, KL, and QW. HC and LL funded the project. HC, HS, ZX, and LL provided administrative, technical or material support. HC, HS, ZX, and LL supervised the study. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Data sharing statement

Multiomic data are openly available in GEO bank (GSE216357) at https://www.ncbi.nlm.nih.gov/geo/. Clinical data are available from the corresponding authors upon reasonable request and with the permission of the institution.

Declaration of interests

No disclosures were reported.