The mean age of the cohort was 59.83 years, and 54.55% were male. Non-severe patients represented 45.45% of the total, and the rest (54.55%) were severe, in accordance with the Diagnosis and Treatment Program of COVID-19 (7th edition). Hypertension was the leading pre-existing disease (45.45%), followed by diabetes (40.91%), chronic respiratory diseases (13.64%), chronic renal disease (13.64%), coronary heart disease (9.09%), cancer (9.09%) and chronic hepatitis (4.55%). All patients (100%) were clinically cured, presenting no symptoms and significantly improved chest computed tomography scan. Only one patient exhibited mild somnolence (Table 1).
Characteristics Age-years, mean ± SD 59.83 ± 12.94 Gender—n/N (%) Male 12/22 (54.55) Female 10/22 (45.45) Underlying diseases—n/N (%) Hypertension 10/22 (45.45) Diabetes 9/22 (40.91) Chronic respiratory diseases 3/22 (13.64) Chronic renal disease 3/22 (13.64) Coronary heart disease 2/22 (9.09) Cancer 2/22 (9.09) Chronic hepatitis 1/22 (4.55) Disease Severity of COVID-19—n/N (%) Non-severe 12 (54.55) Severe 10 (45.45) Initial symptoms—n/N (%) Cough 20/22 (90.90) Fever 18/22 (81.82) Fatigue 12/22 (54.55) Dyspnea 8/22 (36.36) Chest tightness 7/22 (31.82) Myalgia 4/22 (18.18) Diarrhoea 2/22 (9.09) Nausea or vomit 1/22 (4.55) Clinical outcomes†—n/N (%) Symptoms alleviated 22 (100) Presented symptoms 1 (4.55) Chest computed tomography improved or almost normal 22/22 (100.00) Freedom of oxygen treatment 22/22 (100.00) Highest body temperature within 72 h prior to sampling (℃) 36.64 ± 0.38 Data presented as n(n/N %) or otherwise indicated, where N is the total patients.
†as of studied time.
SD, standard deviation.
Table 1. Demographics and clinical characteristics of studied patients.
All 22 of the studied patients were visibly positive for SARS-CoV-2 nucleic acids in nasopharyngeal swab specimens 50 days after initial symptoms. The final positive result was observed and demonstrated in a timeframe of 50 to 120 days post initial symptoms (Fig. 1A). The positive rate gradually declined to 86.36% at 70 days, 36.36% at 80 days, and 4.55% at 100 days post initial symptoms. Such decline became significant after 75 days post initial symptoms (positive rate 59.09%), and dropped to 0% after 110 days. In turn, patients were consistently converted to negative for SARS-CoV-2 nucleic acid, and the conversions were mainly centralized between 70 and 90 days, with the latest negative result pronounced 115 days post initial symptoms (Fig. 1B).
Figure 1. Evolutionary course, conversion, and viral load of long-term SARS-CoV-2 patients. Positive rate of latest SARS-CoV-2 (A), negative conversion (B), and Ct value of ORF1ab and N for latest positive SARS-CoV-2 (C) in studied 22 patients from 50 to 120 days post initial symptoms.
Meanwhile, we analysed SARS-CoV-2 load in nasopharyngeal swab specimens by RT-PCR. Ct values were used as a relative standard for SARS-CoV-2 RNA expression, with lower Ct values corresponding to higher viral copy numbers. For each specimen, SARS-CoV-2 was considered positive if the Ct value of either the ORF1ab gene or the N gene was less than 40. The Ct values of the latest positive SARS-CoV-2, categorized to ORF1ab (Ct value 36.40 ± 3.36) and N (Ct value 36.06 ± 3.23) were demonstrated in Fig. 1C, in an observational period of 50–120 days after symptom onset. Eighteen patients (81.8%) were double positive for both genes, while single positive either for the ORF1ab gene or the N gene were 3 (13.6%) or 1 (4.60%), respectively.
To explore the immune response in COVID-19 patients with long-term SARS-CoV-2 infection, the peripheral blood subgroup of lymphocytes and IFN-γ-secreting T cells were assessed at the time of latest positive SARS-CoV-2 tests. IFN-γ producing capability can be used as a marker of lymphocyte function. The counts of total T lymphocytes, total B lymphocytes (tBL), CD4+ T cells, CD8+ T cells, and NK cells were 1184 ± 297.2/μL, 170.6 ± 75.29/μL, 709.3 ± 237.7/μL, 413.9 ± 192.8/μL, and 402.7 ± 537.7/μL, respectively (Fig. 2A). Functionality assessment disclosed that the ratios of IFN-γ-secreting cells to total CD4+ and CD8+ T cells were 24.68% ± 9.60% and 66.41% ± 14.87%, respectively, which were generally in the normal range (14.54%–36.96%; 34.93%–87.95%). IFN-γ-secreting NK cells were 58.03% ± 11.78%, lower than the given range (61.2%–92.65%) (Fig. 2B).
Figure 2. Long-standing SARS-CoV-2-induced host immune response. Peripheral blood lymphocyte counts (A). Lymphocyte functionality assessed by IFN-γ produced CD4+/CD8+ T cells or NK cells (B). Detection of SARS-CoV-2-specific IgM, IgG, and neutralizing antibody (C). Correlation of SARS-CoV-2-neutralizing activity with IgG (D), B lymphocytes (E), and CD4+ T cell (F), respectively. Dashed lines represent the normal range of giving items.
Next, we detected SARS-CoV-2-specific antibodies and their SARS-CoV-2 neutralizing activity, which was evaluated by neutralization antibody titres. Plasma samples were collected upon the time of latest positive SARS-CoV-2 test. All patients generated moderate IgG titres (94.41, 83.46–114.40 mg/mL) against SARS-CoV-2. Also, SARS-CoV-2-specific IgM (5.57, 1.75–17.40 mg/mL) was detectable but deemed negative according to the user guide of the test (< 10 mg/mL) in the same time point (Fig. 2C). The mean value of SARS-CoV-2 neutralizing activity of the 22 studied patients was 157.2, of which 5 exhibited no neutralization antibody (NAb) (Fig. 2C). The production of SARS-CoV-2-specific NAb was positively correlated to the titres of IgG (P = 0.054, R2 = 0.1738, Fig. 2D), but had no correlation to the B cell (P = 0.5932) and CD4+ T cell (P = 0.6184) counts, as shown in Fig. 2E and 2F.
We also analysed whether the SARS-CoV-2 load was correlated to the immune response (Fig. 3). The expression of the N gene was found to be positively correlated to the NK cells (P = 0.0023, R2 = 0.4105, Fig. 3E), but not for the other lymphocyte subsets. Furthermore, the OFR1ab gene was not correlated to any lymphocyte subsets, neither for cell count, nor for functionality (Fig. 3A–3H).
To investigate whether SARS-CoV-2 virulence had altered, virus isolation was performed with Vero E6 cells from nasopharyngeal swab specimens 50 days after initial symptoms in COVID-19 patients. We were not able to isolate SARS-CoV-2 and did not observe a cytopathic effect (Fig. 4A). However, only one sample from one COVID-19 patient (Patient #4) was positive for SARS-CoV-2 nucleic acid in cell supernatants from two passages, with Ct values gradually increasing during cell passages (Fig. 4B). However, further passages of the virus in Vero E6 cells were unsuccessful.
Figure 4. The result of virus isolation and sequence. A Cytopathic effect of SARS-CoV-2-infected cells. Vero E6 cells were inoculated with sample, and the cytopathic effect was not observed 72 h post-infection. B The Ct values of the sample from patient #4. Vero E6 cells were inoculated with the sample for six days, and the supernatant was subjected to real-time RT-PCR detection with CFDA approved testing kits. The replication of the virus was indicated as Ct values. C NGS raw reads of patient #4 mapped to the SARS-CoV-2 sequence. Reads coverage and sequencing depth of the metatranscriptome sequencing.
Metatranscriptome sequencing was performed on the abovementioned positive sample to analyse the virus genome. Amplicon sequencing of the multiplex PCR products generated extra 65, 847, 006 reads, 69, 506 of which were SARS-CoV-2 reads. Those SARS-CoV-2 reads together covered 11, 160 nucleotides (37.5% of the genome) of the virus genome (Fig. 4C). Eleven consensus sequences longer than 200 bp were obtained, with a total length of 7032 nt (23.6% of the genome) and length ranging from 225 nt to 1500 nt. No variant in the covered virus genome region was observed. One of the consensus sequences had an overlap of 354 nucleotides with the receptor-binging domain of the spike protein coding sequence, showing 100% identity with the reference of Wuhan-Hu-1. Next, nested-PCR assays targeting the spike gene of SARS-CoV-2 were also performed on this sample, and the amplified fragments were sequenced. Full-genome sequences of the S gene, named hCov-19/Wuhan/Tongji-04-2/2020, were submitted to GISAID and are available under accession numbers EPI_ISL_4470802|2020-04-20. To better understand the evolutionary relationships of hCov-19/Wuhan/Tongji-04-2/2020 in detail, phylogenetic analyses were conducted (Supplementary Figure S1). The S gene of hCoV-19/Wuhan/Tongji-04-2/2020 were more closely related to EPI_ISL_480345 and EPI_ISL_478457, while EPI_ISL_434534 was more distant (Supplementary Figure S1). Compared to the first released genome (Wuhan-Hu-1), a total of only 3 synonymous variants were identified in spike protein-coding regions.