Thymosin alpha1 use in adult COVID-19 patients: A systematic review and meta-analysis on clinical outcomes

Abstract

Objective

Thymosin alpha1 (Ta1) is widely used to treat patients with coronavirus disease 2019 (COVID-19), however, its effect remains unclear. This systematic review and meta-analysis aimed to evaluate the effect of Ta1 as a COVID-19 therapy.

Methods

PubMed, EMBASE, the Cochrane library, Web of Science, and the reference lists of relevant articles were searched to identify eligible studies. Assessment of heterogeneity was done using the I-squared (I²) test and random/fixed effect analysis was done to determine the risk ratio (RR). We polled the data related to mortality mainly by using Review Manager 5.4. Predefined subgroup analyses and sensitivity analyses were also performed.

Results

A total of 9 studies were included, on a total of 5352 (Ta1 = 1152, control = 4200) patient outcomes. Meta-analysis results indicated that Ta1 therapy had no statistically significant effect on mortality [RR 1.03 (0.60, 1.75), p = 0.92, I² = 90 %]. Subgroup analyses demonstrated that the beneficial effect in mortality was associated with mean age＞60 years in the Tα1 group [RR 0.68 (0.58, 0.78), p < 0.0000.1, I² = 0 %], the proportion of female ≤ 40 % in the Tα1 group [RR 0.67 (0.58, 0.77), p < 0.0000.1, I² = 0 %], and severe/critical COVID-19 patients [RR 0.66 (0.57, 0.76), p < 0.0000.1, I² = 0 %]. Sensitivity analysis further demonstrated the results to be robust.

Conclusions

The results of this meta-analysis do not support the use of Ta1 in hospitalized adult COVID-19 patients.

1. Introduction

The coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) poses a current worldwide public health threat. From the start of the COVID-19 pandemic to January 31, 2022, a total of 349,641,119 of confirmed COVID-19 cases were reported globally, including about 5.6 million deaths according to the World Health Organization [1]. After the outbreak of COVID-19, various mutations occur in the virus resulting in various mutant strains: Alpha, Beta, Gamma, Delta, and now Omicron. Currently, the Omicron variant is rapidly becoming dominant worldwide due to enhanced immune escape and increased transmissibility [2]. Although the most prevailing type of Omicron infection is asymptomatic, older adults, people who have not been immunized, and those with underlying medical conditions are at higher risk for severe COVID-19 [3]. Delayed immune reconstitution and cytokine storm remain serious obstacles to COVID-19 cure. T-cell immune responses are essential for at least partial protection from many coronavirus infections, including COVID-19, and participate in abating the strong innate immune responses involved in cytokine syndrome [4]. Thymosin alpha1 (Tα1) is a polypeptide hormone secreted by thymic epithelial cells, which can effectively increase the number of T cells, promote the differentiation and maturation of T cells, and reduce apoptosis [5]. Tα1 has been used in immunodeficiency-related diseases and has been proven efficient, such as chronic B and C hepatitis as well as some types of cancers [6], [7]. Therefore, Tα1 has the potential to be a treatment for COVID-19 patients.

Clinical studies have examined the effects of Ta1 on COVID-19, with conflicting results [8], [9], [10], [11], [12], [13], [14], [15], [16]. Some studies have found that Ta1 use was associated with lower mortality [8], [9], [15], [16], while others suggested that Ta1 use did not affect COVID-19 mortality [11], [13], [14]. However, some findings were even contradictory [10], [12]. For example, Wang et al. conducted a single-center retrospective cohort study including 317 COVID-19 patients, treatments with immunomodulatory therapies, including immunoglobulin, glucocorticoids, and thymosin, were significantly associated with increased mortality in COVID-19 patients [10]. Consequently, we performed a meta-analysis to systematically analyze the effects of Ta1 on COVID-19 to establish current evidence for the role of this intervention.

2. Data and methods

Our study was conducted following the Preferred Reporting Items for Systematic Reviews and meta-Analyses (PRISMA) statement checklist [17] and according to the protocol registered in the PROSPERO database (CRD42022370182).

2.1. Search strategy

PubMed, EMBASE, and Cochrane Library, Web of Science databases were searched for observational cohort studies and randomized controlled trials (RCTs) to October 25, 2022. The used search terms were as follows: ‘‘thymosin alpha1′’ or ‘‘thymosin’’ or “thymus” or “maipuxin” or “thymalfasin” or “zadaxin” or “thymosin α1 (Tα1)’’ or “thymosin-alpha (1)’’ or ‘‘COVID-19’’ or ‘‘SARS-CoV-2’’ or ‘‘SARS-COV-2’’ or ‘‘coronavirus Disease 2019’’. Furthermore, we searched reference lists of all included studies for additional eligible studies. Two of the authors (YL and BZ) independently screened titles and abstracts, analyzed full-text articles, and ascertained the final eligible records. Conflicting results were resolved by discussion. We merged retrieved citations using EndNote X9.

2.2. Inclusion and exclusion criteria

The inclusion criteria were: (1) the study design was a cohort study or RCT; (2) studies had to provide clinical outcomes of Tα1 in adults who had COVID-19 or sufficient data to calculate these outcomes; (3) the intervention included Tα1, which was not limited by type, dose, and duration, compared with placebo or standard of care alone. Reviews, letters, abstracts, case reports, and animal studies were excluded.

2.3. Data extraction

Study characteristics were extracted by two authors (WW and WS) separately as follows: first author’s name, publication year, study design, country origin, sample size, patient severity, mean age in the Tα1 group, the proportion of female in the Tα1 group (%), Tα1 dosages, and clinical outcomes. When needed, we contacted the original author for clarification. The primary endpoint was mortality, and the second endpoints were the length of hospitalization, length of intensive care unit (ICU) stay, invasive mechanical ventilation (IMV) rate, and adverse events. If mortality could not be acquired, we used 28-day mortality or in-hospital mortality instead.

2.4. Quality assessment

The Newcastle Ottawa Scale (NOS) was used for the validity assessment of observational studies [18]. The NOS score ranges from 0 (low quality) to 9 (high quality) points. For RCTs, the risk of bias was assessed on seven domains (random sequence generation, concealment of allocation, performance bias, detection bias, incomplete outcome data, selective outcome reporting, and others) [19]. Two authors (WW and YR) performed the quality assessment independently. Disagreements were resolved by discussion.

2.5. Statistical analyses

Eligible meta-analysis studies were studies with identical results. A meta-analysis was performed only for sets of three or more studies that met the above-mentioned criteria. Pooled relative risks (RRs), mean differences (MDs), and the corresponding 95 % confidence intervals (CIs) were calculated. If the study did not report the mean and standard deviation, the median, sample size, and interquartile range were used to calculate the MD for ICU and hospital stay between the two groups [20]. When available, the adjusted hazard ratio (HR) and in-hospital mortality after propensity score matching (PSM) were used for pooling to reduce confounding. Statistical heterogeneity was quantified using the I² statistic and Chi² test, and if significant heterogeneity (P ≤ 0.05 or I² ≥ 50 %) was obtained we used a random-effects model, and otherwise a fixed-effects model was used [21]. Subgroup analyses were further conducted to explore the potential source of heterogeneity. Stratified analyses were conducted based on study design (cohort or RCT), sample size (＞600 or ≤ 600), mean age in the Tα1 group (age＞60 or age ≤ 60), the proportion of female in the Tα1 group(%) (female＞40 % or female ≤ 40 %), and COVID-19 patient severity (severe/critical or COVID-19). We conducted sensitivity analyses to estimate the influence of a study on the pooled prevalence by removing one study in each turn. Publication bias was evaluated using a funnel plot [22]. P < 0.05 in the 2-tailed test was considered to be statistically significant. Review Manager 5.4 and STATA 16.0 were used to perform data analysis. A trial sequential analysis was performed to verify that cumulative data were sufficient to estimate outcomes. This analysis was performed using trial sequential analysis software, version 0.9.5.9 (Centre for Clinical Intervention Research) [23]. The TSA was performed to maintain an overall 5 % risk of a type I error and 20 % of the type II error (a power of 80 %), a relative risk reduction of 20 %, and the pooled control group event rate across the included studies. This analysis was conducted according to the previous meta-analysis [24].

3. Results

3.1. Study selection, characteristics, and quality

As shown in Fig. 1 , our literature search returned 229 results for relevant articles, and the full text was retrieved for 23 articles. Finally, 9 full texts were included in the meta-analysis [8], [9], [10], [11], [12], [13], [14], [15], [16]. The main characteristic of the studies included is presented in Table 1 . These articles included 7 retrospective cohort studies [8], [9], [10], [11], [12], [13], [16], and 2 RCTs [14], [15]. Sample sizes of the included studies ranged from 40 to 2,282 patients. For the observational studies [8], [9], [10], [11], [12], [13], [16], the mean NOS score was 8.6 points. For the RCT [14], [15], the risk of bias table is depicted in Table S1.

Fig. 1.

PRISMA flow diagram.

Table 1.

Characteristics of the included studies.

Study	Design	Country	Population	Mean age in the Tα1 group	Female in the Tα1 group (%)	Tα1 dosages	Clinical outcomes
Wu M 2020	multicenter retrospective cohort study	China	severe/critical (3 3 4) Tα1 (1 0 2) Control (2 3 2)	64 (56,69)	34 (33.3 %)	1.6 mg, qd or q12h	28-day and 60-day mortality; hospital length of stay; total duration of the disease
Liu Yp 2020	multicenter retrospective cohort study	China	severe/critical (76) Tα1 (36) Control (40)	57.5 (53.50, 73.75)	11(30.6 %)	10 mg qd (at least seven days), then recommended to continue to use till the endpoints	mortality; ICU length of stay of survivors; Hospital stay of survivors; NIMV and IMV (%)
Wang Y 2020	retrospective cohort study	China	COVID-19 (3 1 7) Tα1 (68) Control (2 4 9)	NA	NA	10 mg-30 mg qd	in-hospital mortality
Huang Cl 2021	retrospective cohort study	China	non-severe(1388) Tα1 (2 3 2) Control (1156)	38 (28,53)	95 (40.9 %)	1.6 mg:three times a week, for at least 1 week; qod, for at least 6 days; qd, for at least 3 days	mortality; Progression to severe cases
Liu J 2021	multicenter retrospective cohort study	China	COVID-19 (2282) Tα1 (3 0 6) Control (1976)	57.9 ± 14.5	141 (46.1 %)	NA	non-recovery; In-hospital mortality; Duration of mechanical ventilation; length of ICU stay; length of hospital stay
Sun Q 2021	multicenter retrospective cohort study	China	Critical (7 7 1) Tα1 (3 2 7) Control (4 4 4)	64 (55, 71)	125 (38.2 %)	NA	28-day mortality; IMV; NIMV
Wang T 2022	retrospective cohort study	China	COVID-19 (95) Tα1 (31) Control (64)	53.55 ± 12.13	14 (45.16 %)	1.6 mg qod	28-day mortality
Shetty 2022	RCT	India	Severe (40) Tα1 (27) Control (13)	48 (35, 58)	11 (40.7 %)	1.6 mg	all-cause mortality, clinical progression, duration of hospital/ ICU stay
Shehadeh 2022	RCT	USA	Severe (49) Tα1 (23) Control (26)	64 (49, 80)	9 (39 %)	1.6 mg qd	mortality; clinical recovery IMV

Abbreviations: NA, not available; RCT, Randomized controlled clinical trial; Tα1, Thymosin α1 group; ICU, intensive care unit; qd, once a day; qod, once every other day; NIMV, Noninvasive mechanical ventilation; IMV, Invasive Mechanical Ventilation.

3.2. Mortality

As shown in Fig. 2 , nine trials with 5352 patients reported all-cause mortality, 28-day mortality, or in-hospital mortality [8], [9], [10], [11], [12], [13], [14], [15], [16]. Overall, mortality was 22 % in the patients taking Tα1 and 15 % in the patients not Tα1. No statistically significant impact of Tα1 treatment on mortality was observed [RR 1.03 (0.60, 1.75), p = 0.92], with high inter-study heterogeneity (I² = 90 %). In addition, three trials with 1916 patients reported 28-day mortality or in-hospital mortality after PSM [11], [12], [13]. No statistically significant impact of Tα1 treatment on mortality was observed [RR 1.11 (0.83, 1.47), p = 0.48], with high inter-study heterogeneity (I² = 52 %). (Fig. S1). Further, three trials reported adjusted hazards ratios (HRs) and 95 % confidence intervals (95 % CIs) of 28-day mortality in COVID-19 patients [9], [13], [16]. No statistically significant impact of Tα1 treatment on 28-day mortality was observed [HR 0.51 (0.26, 1.00), p = 0.065], with high inter-study heterogeneity (I² = 58.5 %) (Fig. 3 ).

Fig. 2.

Forest plot of the effect of thymosin alpha1 treatment on hospitalized COVID-19 patient mortality.

Fig. 3.

Forest plot of the effect of thymosin alpha1 treatment on hospitalized COVID-19 patient mortality adjusted hazards ratios (HRs).

3.2.1. Subgroup analyses

In most cases, high heterogeneity was still present in stratified analyses unless RCT, severe/critical group, age＞60 years in the Tα1 group, and the proportion of female ≤ 40 % in the Tα1 group. In subgroup analyses, no significant difference (p = 0.18) was observed between RCTs [RR 0.48(0.17, 1.37), I² = 21 %] and retrospective cohort studies’ results [RR 1.19(0.66, 2.14), I² = 92 %](Fig. S2a). The observed association was also similar among studies with sample size＞600 [ RR 1.25(0.56, 2.80), I² = 94 %] and those with sample size ≤ 600 [RR 0.85(0.34, 2.10), I² = 87 %] (Fig. S2b). Stratification based on mean age in the Tα1 resulted in a pooled RR of 0.68(0.58,0.78) for studies with mean age＞60 years in the Tα1 group and a pooled RR of 0.97(0.46,2.05) for studies with mean age ≤ 60 years in the Tα1 group, the differences between the subgroups were not statistically significant (p = 0.36) (Fig. 4 a). Subgroup analysis according to the proportion of female in the Tα1 group showed that Tα1 decreased the mortality in the Tα1 group (proportion of female ≤ 40 %)[RR 0.67(0.58, 0.77), I² = 0 %], but had no significant effect on the mortality in Tα1 group (proportion of female＞40 %) [RR 1.22(0.57, 2.61), I² = 72 %], the differences between the subgroups were not statistically significant(p = 0.13) (Fig. 4b). Stratification based on COVID-19 patient severity, resulted in a pooled RR of 0.66 (0.57, 0.76) for studies with severe/critical COVID-19 group and a pooled RR of 2.24 (1.11, 4.54) for studies with COVID-19/non-severe group, the differences between the subgroups were statistically significant (p = 0.0009) (Fig. 4c).

Fig. 4.

Subgroup analysis forest plots of the effect of thymosin alpha1(Tα1) treatment on hospitalized COVID-19 patient mortality and funnel plot of publication bias. Fig. 4(a) subgroup analysis according to mean age in the Tα1 group, Fig. 4(b) subgroup analysis according to the proportion of female in the Tα1 group, Fig. 4(c) subgroup analysis according to patient severity.

3.2.2. Sensitivity analyses and reporting bias

Sensitivity analyses were performed by excluding one study at a time. And they indicated that the omission of any of the studies led to an estimated RR change of between 0.82(95 % CI: 0.51–1.32) and 1.16 (95 % CI: 0.66–2.01) (Table 2 ). The changes were not significant. The funnel plot was symmetric, indicating no publication bias (Fig. S3).

Table 2.

Sensitivity analysis for mortality.

Study omitted	RR	95 %CI		I2 (%)	P
Wu M 2020	1.13	0.63	2.02	91	0.69
Liu Yp 2020	1.15	0.66	2.02	91	0.62
Wang Y 2020	0.82	0.51	1.32	83	0.42
Huang Cl 2021	0.94	0.55	1.58	90	0.81
Liu J 2021	0.97	0.49	1.93	88	0.93
Sun Q 2021	1.11	0.59	2.06	83	0.75
Wang T 2022	1.00	0.55	1.80	91	0.99
Shetty 2022	1.16	0.66	2.01	91	0.61
Shehadeh 2022	1.04	0.60	1.83	91	0.88

Abbreviations: RR, relative risk; CI, confidence interval.

3.2.3. Trial sequential analysis

Trial sequential analysis suggested that there may be no statistical difference in mortality at 28 days between the intervention group and the control group, and more trials are needed to prove it (Fig. 5 ).

Fig. 5.

Trial sequential analysis for mortality.

3.3. Length of hospitalization

Only four out of nine studies reported on the length of hospitalization [8], [9], [12], [15]. Consequently, a total of 2716 (Tα1 = 467, control = 2249) patient outcomes were included in the meta-analysis regarding the length of hospital stay. No statistically significant impact of Tα1 treatment on length of hospitalization was observed (MD 1.90[−4.31, 8.11], P = 0.55), with high inter-study heterogeneity (I² = 95 %), Fig. S4. Sensitivity analysis indicated that no literature significantly interfered with the results of this meta-analysis.

3.4. Length of ICU stay

Only three out of nine studies reported on the length of hospitalization [8], [12], [15]. Consequently, a total of 2382 (Tα1 = 365, control = 2017) patient outcomes were included in the meta-analysis regarding the length of hospital stay. No statistically significant impact of Tα1 treatment on length of hospitalization was observed (MD 0.79[−6.35, 7.94], P = 0.83), with high inter-study heterogeneity (I² = 94 %), Fig. S5. Sensitivity analysis showed that no studies had a significant impact on this outcome.

3.5. IMV rate

Three trials presented information analyzing IMV rate [8], [13], [14]. Pooling the outcomes of the 3 studies showed that there was no significant statistical difference in the IMV rate between the Tα1 and non-Tα1-treated groups [RR 0.51 95 % CI (0.11, 2.45), I² = 57 %; studies = 3; participants = 896] (Fig. S6). Sensitivity analysis showed that the result was stable.

4. Discussion

In the meta-analysis, we explored the potential benefits that Tα1 can bring in patients with COVID-19 by systematically analyzing all nine relevant studies. Overall, our results suggested that Tα1 therapy provided no significant benefit regarding COVID-19 patients' mortality. In addition, this meta-analysis also showed no association between length of hospitalization, length of ICU stay, IMV rate, and tymosin-a1 treatment.

The results of our meta-analysis were partially in accord with a previously conducted meta-analysis on the topic [25], but certain potential differences should be highlighted. The meta-analysis conducted by Liu et al. analyzed 4 studies (all of which were included in our analysis as well) and failed to find a beneficial effect of Tα1 regarding the patient mortality outcome [RR 1.24 (0.33, 4.68), p = 0.76] [25]. However, statistical bias should be considered due to the small number of studies included, all included studies were retrospective, and studies with no significant association between Tα1 treatment and mortality were heavily weighted. In addition, their study had a high level of heterogeneity (I² = 94 %) and no subgroup analysis, sensitivity analysis, funnel plot, etc. were conducted to explore the source of heterogeneity. Our literature search identified five additional studies [12], [13], [14], [15], [16], two of which were RCTs [14], [15] and found no significant effects. This finding was proved via PSM, in the cox proportional hazards model, and trial sequential analysis. However, the insignificant association between Tα1 therapy and mortality cannot exclude the beneficial effect of Tα1 therapy among specific patients with COVID-19.

Subgroup analyses provided that the beneficial effect in mortality was associated with mean age＞60 years in the Tα1 group, the percentage of female ≤ 40 % in the Tα1 group, and severe/critical COVID-19 population group. Liu et al. reported Tα1 treatment reverses T cell exhaustion, recovers immune reconstitution through promoting thymus output and significantly reduces mortality of severe COVID-19 patients compared with the untreated group (11.11 % vs 30.00 %, p = 0.044). In an RCT, Shetty et al. found the death rate was statistically lower in the Tα1 group (11.1 %) than in the placebo group (38.5 %) in patients with severe COVID-19. Wu et al. also found Tα1 therapy significantly reduced 28-day mortality [HR 0.11,(0.02, 0.63); p = 0.013] in critical type patients, especially those aged over 64 years. Nevertheless, inconsistent results were also reported in other studies. Sun et al. found that Tα1 therapy was not associated with a difference in 28-day mortality in critical COVID-19 patients after adjustment for baseline confounders [13]. This may be due to the critical and complex pathophysiological changes in the patients enrolled in their study. Nearly half of the patients required IMV, and the overall mortality rate was 52.4 % in their study. Similarly, Shehadeh et al. found that Tα1 was not associated with the likelihood of death, possibly due to a significant difference in baseline high-flow oxygen support between treatment groups [14]. Interestingly, our subgroup analyses also found Tα1 treatment reduces the mortality of COVID-19 patients compared with the untreated group in the Tα1 group(female ≤ 40 %). The same findings were also found in Sun et al.’s study [13]. However, the underlying mechanism remains unclear. There is growing evidence of gender differences in clinical outcomes for COVID-19 [26], [27]. Further, a recent meta-analysis collating data from 8 countries demonstrated that once infected males had more severe adverse clinical outcomes, leading to higher death rates [28]. The reason may be differences in molecular level, immune response, and endocrine function between men and women [27], [29], [30], [31]. For example, it has been observed that women develop mount stronger humoral and adaptive immune responses than men after a virus infection. The intensity and prevalence of the virus tend to be lower in women than in men due to increased immunity to the virus [32]. Based on this information, male patients may receive more aggressive inpatient care interventions, such as early administration of Tα1, especially those with other risk factors for COVID-19. In addition, Liu et al. found that Ta1 use at a later stage may exacerbate the inflammatory response and lead to a poor prognosis compared to early use [12]. In our study, the proportion of men in the Tα1 group was higher than that in the control group, we inferred that the low mortality rate in male patients in the Ta1 group may be attributed to the use of Ta1 in the early stage of the disease. More research is needed to confirm our hypothesis.

Adverse events in the Ta1 group, such as respiratory failure, epistaxis, thromboembolism, etc., have been reported in three studies, but none of them were identified as being associated with the Ta1 [11], [14], [15]. Moreover, there were no allergic reactions or drug eruptions in the Ta1 group. Therefore, the administration of Ta1 to moderate-to-severe COVID-19 patients appears to have acceptable safety.

Several limitations of this meta-analysis must be considered. First, significant heterogeneity was detected in mortality and Tα1 use among COVID-19 patients, the differences in mean age in the Tα1 group, the proportion of female in the Tα1 group, and COVID-19 patient severity may be responsible for this between-study variation. Second, most of the included studies were conducted in China, which may limit generalizability. Third, Liu et al. reported Ta1 use at a later stage was significantly associated with a poor prognosis than Ta1 use at an earlier stage and the duration of Ta1 use had a less significant effect on the recovery rate [12]. However, we had no access to the information on the type, dose, and duration of Tα1 in some of the studies. Therefore, we could not evaluate the association between mortality and different types, doses, and durations of Tα1 in patients with COVID-19. Fourth, the sample sizes of most of the studies included were relatively small, only three trials had more than 600 participants. Fortunately, the sensitive analyses proved that the results were relatively stable. Finally, most of the included studies were observational studies in nature, residual confounding cannot be excluded.

5. Conclusions

Our study found that Tα1 therapy provided no significant benefit regarding COVID-19-related mortality rate. More well-designed worldwide multicenter clinical trials are needed to associate personalized medicine with the Tα1 treatment to select suitable patients who are more likely to show a benefit.

Funding

Not applicable.

Patient consent

Not required.

Ethics approval

Not required.

Data Availability Statement

All data generated or analyzed during this study are included in this published article.

CRediT authorship contribution statement

Weifeng Shang: Formal analysis, Data curation, Original draft preparation. Bo Zhang: Original draft preparation. Yali Ren: Methodology. Weina Wang: Validation, Investigation. Dengfeng Zhou: Conceptualization, Project administration, Supervision, Writing – review & editing. Yuanyuan Li: Conceptualization, Project administration, Supervision, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

The following are the Supplementary data to this article:

Supplementary Fig. S1.

Forest plot of the effect of thymosin alpha1 treatment on hospitalized COVID-19 patient mortality after propensity score matched.

Supplementary Fig. S2.

Subgroup analysis forest plots of the effect of thymosin alpha1 treatment on hospitalized COVID-19 patient mortality. Figure S2(a) subgroup analysis according to study type, Figure S2(b) subgroup analysis according to sample size.

Supplementary Fig. S3.

Funnel plot of publication bias.

Supplementary Fig. S4.

Forest plot of length of hospitalization.

Supplementary Fig. S5.

Forest plot of length of ICU stay.

Supplementary Fig. S6.

Forest plot of IMV rate.

Supplementary Table S1

Quality assessment of included studies.

Data availability

The data is shown in the article