The diagnostic value of serum Cystatin C in differentiating malignant from benign pulmonary nodules: a retrospective cohort study Authors: YuanLin He, LiJun He, YuTing Zhang, Cheng Shen Published: 2025 ABSTRACT AbstractBackgroundThe present study aims to evaluate the utility of Cystatin C testing for differentiating lung cancer from benign nodules of the lung in patient.MethodsConsecutive patients were enrolled in our research who had undergone a surgical resection or with fine needle biopsy between April 2015 and June 2024. All the patients were evaluated with their age, sex, smoking, drinking, and histology of lung cancer. Serum levels of Cystatin C in patients with lung nodules were evaluated using an automated analyzer.Results540 consecutive patients were enrolled in our research. There were 443 patients with primary lung cancer and 97 patients diagnosed as benign nodules. The serum level of Cys-C in patient with malignant nodule was significantly higher than in those with benign nodule (p<0.01), but they were comparable between patients with squamous cell carcinoma, adenocarcinoma, small cell lung cancer and large cell neuroendocrine carcinoma. Moreover, receiver operating characteristic curve analysis showed that a Cys-C level>0.92 mg/L was a good predictor for ruling out benign nodule, with a sensitivity of 64.0% and a specificity of 63%.ConclusionsOur study suggests that the serum Cys-C level test may be helpful for differentiating lung cancer from the benign nodule of the lung in patient.Keywords: Cystatin c, Lung cancer, Benign nodule, Receiver operating characteristic Background Morbidity and mortality of lung cancer are at the forefront of various types of tumors around the world. Influenced by factors such as smoking, occupational and environmental pollution, the incidence of lung cancer has increased significantly in many countries [1]. Patients with lung cancer often develop more obvious symptoms at an advanced stage of the disease. Therefore, the early diagnosis of lung cancer is a very critical step for improving the patient’s treatment status and quality of life. Despite advances in imaging techniques such as low-dose CT, differentiating benign from malignant pulmonary nodules remains a significant clinical challenge, often leading to unnecessary invasive procedures or delayed diagnosis. Cystatin C (Cys-C) is a 120 amino acid, low molecular weight protein without glycosylated [2]. As a housekeeping gene, serum Cys-C is produced from human nucleated cells [3]. Some previous studies have suggested that Cys-C results are not clearly correlated with age, muscle mass or body mass index (BMI) in healthy individuals [4]. Due to its stable chemical characteristic under room temperature in whole blood or serum, Cys-C is an ideal marker tool in diagnostic use [5–7]. Previous studies combined with relevant literature results have known that the expression level of serum cystatin C is a better detection method than serum creatinine in terms of glomerular filtration rate (GFR) marker [3, 8, 9]. Besides, serum creatinine and Cys-C-based indices showed moderate diagnostic accuracy for sarcopenia [10]. Beyond its role in renal function, Cystatin C is involved in regulating cysteine proteases such as cathepsins, which play a crucial role in tumor invasion, metastasis, and modulation of the tumor microenvironment. Altered expression of Cys-C has been reported in various cancers, suggesting a potential link to tumor biology. However, very few recent research reports suggest that the expression level of serum Cys-C may be correlated with lung cancer. Therefore, in order to further evaluate and select a marker tool for early detection and screening of lung cancer, our research content considers the detailed analysis of the correlation between serum Cys-C levels in lung cancer patients and lung cancer disease. This is, to our knowledge, the first report of confirming a correlation between serum Cys-C levels and lung cancer or benign nodules of the lung, and to evaluate Cys-C levels in differentiating lung cancer from benign nodules (BN) of the lung. Methods Patients and clinical data This was a retrospective study conducted on data collected from April 2015 to June 2024. The study included 540 patients (443 with primary lung cancer and 97 with benign nodules) who underwent diagnostic confirmation via surgical resection, CT-guided fine needle aspiration biopsy (FNAB), or bronchoscopic biopsy from the Department of Thoracic Surgery at West China Hospital between April 2015 and June 2024. The characteristics of the patients are shown in Table 1. All the patients were evaluated with their age, sex, smoking, alcohol consumption, and histology of lung cancer. The studies involving human participants were reviewed and approved by West China Hospital Ethics Committee (No. 202231) and affiliation of ethics committee. All patients in this article signed the text of the informed consent form and agreed to participate in our study. Table 1Clinicopathological features of lung nodule casesGroupCharacteristicNumber(%) Age >60287(53.1%)<=60253(46.9%) Sex Male393(72.8%)Female147(27.2%) Smoking Yes338(62.5%)No202(37.5%) Alcohol Consumption Yes120(22.2%)No420(77.8%) Pathological Diagnosis BN97(18.0%)MN443(82.0%) Histology BN97(18.0%)ADC148(27.4%)SCC147(27.2%)SCLC129(23.9%)LCNEC19(3.5%) Clinical Staging Ⅰ-Ⅱ192(43.3%)Ⅲ-Ⅳ198(44.7%)Unavailable53(12.0%)SCC - Squamous Cell Carcinoma, ADC - Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules Clinicopathological features of lung nodule cases SCC - Squamous Cell Carcinoma, ADC - Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules Inclusion criteria Patients were included in the study if they: Underwent a definitive diagnostic procedure (surgical resection, or biopsy) for a pulmonary nodule. Underwent a definitive diagnostic procedure (surgical resection, or biopsy) for a pulmonary nodule. Had a confirmed pathological diagnosis, which was used to classify the nodule as benign or malignant. For malignant cases, the specific histological subtype (e.g., Adenocarcinoma [ADC], Squamous Cell Carcinoma [SCC], etc.) was recorded. Had a confirmed pathological diagnosis, which was used to classify the nodule as benign or malignant. For malignant cases, the specific histological subtype (e.g., Adenocarcinoma [ADC], Squamous Cell Carcinoma [SCC], etc.) was recorded. Had complete clinical data available, including age, sex, smoking history, and alcohol consumption history. Had complete clinical data available, including age, sex, smoking history, and alcohol consumption history. Exclusion criteria Patients were excluded from the study if they: Lacked a definitive pathological confirmation of their nodule diagnosis. Lacked a definitive pathological confirmation of their nodule diagnosis. Had incomplete clinical or pathological data (e.g., missing smoking status or histology information). The “Unavailable” category for clinical staging in 53 malignant cases suggests that missing key data may not have been an absolute exclusion criterion for staging, but likely was for the primary classification. Had incomplete clinical or pathological data (e.g., missing smoking status or histology information). The “Unavailable” category for clinical staging in 53 malignant cases suggests that missing key data may not have been an absolute exclusion criterion for staging, but likely was for the primary classification. Serum Cystatin C analysis Serum cystatin-C levels were measured from blood samples obtained preoperatively, prior to any surgical intervention or biopsy. All serum samples were evaluated by an automated analyzer for serum Cys-C levels in patients with pulmonary nodules, maccura-160,920 (Maccura, Sichuan, China) using colloidal gold particles coated with anti-Cys-C antibodies by Immunolatex turbidimetry method (Maccura, Sichuan, China). Correlation between the serum level of Cystatin C and clinicopathologic factors of lung nodules of patients are shown in Table 2. Table 2Correlation between the serum level of Cystatin C and clinicopathologic factors of lung nodule patientsGroupsNumberLevel of Cystatin CMean ± SD(mg/L)StatisticP value Smoking 6.16< 0.01 Yes 3381.04 ± 0.24 No 2020.91 ± 0.17 Alcohol Consumption 3.11< 0.01 Yes 1201.04 ± 0.21 No 4200.97 ± 0.23 Histology 7.11< 0.01 BN 970.88 ± 0.16 ADC 1481.00 ± 0.22 SCC 1471.04 ± 0.26 SCLC 1291.00 ± 0.24 LCNEC 190.98 ± 0.12 Pathological Diagnosis 4.60< 0.01 Benign 970.89 ± 0.16 Malignancy 4431.01 ± 0.24 TNM stage 2.170.115 i-ii 1921.00 ± 0.23 iii-iv 1981.01 ± 0.23 Unavailable 531.07 ± 0.30SCC - Squamous Cell Carcinoma, ADC - Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules Correlation between the serum level of Cystatin C and clinicopathologic factors of lung nodule patients SCC - Squamous Cell Carcinoma, ADC - Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules Statistical analysis Measurement data was presented as mean ± standard deviation as they followed normal distribution. The Comparison was performed by One-Way ANOVA and the Post Hoc Multiple Comparisons was performed by Bonferroni Test. Categorical data was presented as frequencies and percentages. We used the area under the ROC (receiver operating characteristic) curve analysis for the study results. Multivariate logistic regression analysis was performed to adjust for potential confounders, including age, sex, smoking, alcohol consumption, and estimated glomerular filtration rate (eGFR). In this way, the accuracy of serum marker Cystatin C in predicting benign or malignant pulmonary nodules was evaluated. The area under the curve (AUC) was calculated. A P-value < 0.05 was defined to be of statistical significance. The statistical calculations were performed with SPSS 22.0 (SPSS Inc., Chicago, Illinois, USA). Results In our study, there were 393 males (72.8%) and 147 females (27.2%) and 287 patients (53.1%) were over 60-year-old. In all patients, 338 patients (62.5%) have smoking history and 22.2% patients of all (120 cases) have alcohol consumption history (Table 1). A total of 540 patients were enrolled with different histology, including 97 benign nodules (BN), 148 adenocarcinoma (ADC), 147 squamous cell carcinoma (SCC), 129 small-cell lung cancer (SCLC) and 19 large cell neuroendocrine carcinoma (LCNEC). The pathological results of all tumors were based on conventional optical microscopy of hematoxylin-eosin staining after the operation or the bronchoscopy biopsy or CT-guided fine needle aspiration biopsy. Staging was based on the 8th edition of the AJCC TNM classification. The serum level of Cys-C in BN was 0.89 ± 0.16 (mg/L) and in Malignant nodules (MN) was 1.01 ± 0.24 (mg/L) (Table 2; Fig. 1). From the Table 2, we can see that patients with smoking have higher level (1.04 ± 0.24 mg/L) of Cys-C than in patients without smoking (0.91 ± 0.17 mg/L). The same result tendency was seen in patients with alcohol consumption (1.04 ± 0.21 mg/L), and the serum level of Cys-C in patients without alcohol consumption was 0.97 ± 0.17 (mg/L). Comparison of serum level of Cys-C in benign nodule (BN) and in malignant nodule (MN) in patient As the consequence showed in our Table 2; Fig. 2, it was statistically significant in different pathological of all nodules. The serum level of Cys-C was elevated in patients with lung cancer including ADC, SCC, SCLC and LCNEC. Figure 1 shows the Cys-C level, which was (1.00 ± 0.22 mg/L), was meaningfully higher in patients with ADC than in patients with BN (0.88 ± 0.16 mg/L, p < 0.01). In patients with SCC and SCLC, the results also showed significantly higher Cys-C level expression than BN group respectively (SCC: 1.04 ± 0.26 mg/L; SCLC: 1.00 ± 0.24 mg/L p < 0.01), but it was comparable in patients with BN and patients with LCNEC (0.98 ± 0.12 mg/L, p = 0.065). Moreover, patients classified as early stage (TNM stage I and II) were comparable in patients classified as TNM stage III and IV (1.00 ± 0.23 mg/L vs. 1.01 ± 0.23 mg/L, p = 0.115) (Fig. 2). Comparison of serum level of Cys-C in benign nodule (BN) and in malignant nodule (MN) including adenocarcinoma (ADC), squamous cell carcinoma (SCC), small-cell lung cancer (SCLC) and large cell neuroendocrine carcinoma (LCNEC) Figure 3 shows the ROC for patients with lung cancer versus patients with benign nodules. The AUC value was 0.67 (95% CI: 0.62–0.73) for patients with BN vs. patients with MN and the best cut-off value of Cys-C level was 0.92 mg/L by ROC analysis. The AUC value of BN was 0.69 (95% CI: 0.62–0.75) vs. ADC, 0.70 (95% CI: 0.63–0.76) vs. SCC, 0.64 (95% CI: 0.57–0.71) vs. SCLC, and 0.71 (95% CI: 0.60–0.82) vs. patients with LCNEC diagnoses. Moreover, the best cut-off value of Cys-C level for predicting ADC, SCC, SCLC and LCNEC was 0.93 mg/L, 0.93 mg/L, 0.91 mg/L and 0.93 mg/L by ROC analysis, respectively. In our study, serum Cys-C demonstrated a high positive predictive value (88.8%), indicating its potential utility in identifying high-risk individuals for whom invasive diagnostic procedures may be recommended. However, its relatively low negative predictive value (27.7%) limits its clinical application as a standalone rule-out tool. Consequently, a normal Cys-C level should not be used to exclude the possibility of malignant tumors. ROC for the prediction by serum level of Cys-C in patients with malignant nodule versus benign nodule Discussion In clinical work, it is difficult for surgeons to differentiate between benign and cancer-like lung nodules preoperatively. Because percutaneous lung aspiration or bronchoscopy may also not be able to obtain pathological results of the focal tissue, the possibility that the pulmonary nodules are primary lung cancer cannot be ruled out in the end [11]. At present, although [18] F-FDG-PET/CT [18] F-fluorodeoxyglucose-positron emission tomography/computed tomography] has been widely used, however, this method is also insufficient to accurately determine the pulmonary nodules [12]. Therefore, choosing a less invasive and more effective screening method may be preferable to surgical removal of the nodule, although this approach may bring higher medical costs for patients with lung cancer [11, 13, 14]. By detecting the level of serum Cys-C in patients, the researchers found that there is a significant correlation between it and the malignant progression of colorectal cancer, melanoma and ovarian cancer. The level of serum Cys-C It can indirectly reflect the state of the patient’s cancer and has a certain evaluation value [15–19]. A few previous studies reported that Cys-C may serve as a replacement to be a marker of sarcopenia screening [20, 21]. However, there is little information on the determination of serum Cys-C between lung cancer and benign lung nodules. To determine whether the serum Cys-C level in patient can differentiate lung malignancy from lung benign diseases, we try to analyze the serum Cys-C level in patients with different pathological types of lung cancer or benign nodule. The observed elevation of serum Cys-C in patients with malignant nodules, regardless of histology, may be explained by its role in the tumor microenvironment. Tumors may upregulate cysteine proteases such as cathepsins for invasion and metastasis, leading to a compensatory increase in their endogenous inhibitor, Cystatin C. Alternatively, Cys-C could be released during tumor-associated inflammation or tissue remodeling processes common to various cancer types. With a modest sensitivity and specificity as a standalone test, serum Cys-C is unlikely to replace current diagnostic modalities. However, it may serve as a valuable adjunct within a multi-marker panel or in conjunction with clinical and radiological features to improve overall diagnostic accuracy. For example, in patients with indeterminate nodules on CT, an elevated Cys-C level might tip the balance towards recommending a more invasive biopsy. Our study underscoring the potential utility of Cys-C in the clinically critical population of patients with small, indeterminate nodules where diagnostic uncertainty is highest. We used the keywords including “cystatin C” and “lung cancer” without restriction to search and collect data in detail in PubMed, Web of Science and EMBASE databases. Then we found only 59 literatures are relevant to this research. Most of them show that the expression of serum Cys-C level can be used as a predictor of adverse reactions after chemotherapy in patients with lung cancer. Changzhi Huang et al.’s study showed that higher CysC concentrations are associated with increased mortality due to lung, blood, brain, esophageal, breast, and liver cancers [22]. Lower Cr/Cys-C ratios correlated with a higher incidence of chemotherapy related adverse effects in patients with Non-small cell lung cancer (NSCLC) [18, 23]. As showed in Naumnik et al.’s research, the expression of Cys-C plays an important role in the tumor invasion including lung cancer [24, 25]. In their study, there are 40 patients and 15 healthy volunteers. They collected peripheral blood samples from lung cancer patients before the start of chemotherapy and after four cycles of chemotherapy. The results showed that the average serum Cys-C level of advanced NSCLC patients was significantly higher than that of the healthy group. However, the serum Cys-C level of T4 patients was lower than that of T2 patients. (p = 0.033). Based on the results of the above studies, it is speculated that the expression level of serum Cys-C in lung cancer patients is higher than that in healthy people. It can also be used as an independent marker to assess the prognostic performance of patient with NSCLC [26–29]. As showed in our data (Table 2), patients with lung cancer including ADC, SCC and LCNEC, have a higher serum concentration of Cys-C compared to people with BN, which revealed a significant statistical difference (p < 0.01). But there was no statistically significant difference in outcomes between patients with early-stage and advanced-stage lung cancer, which is considered due to the small sample size of research. According to the results of several basic studies reported so far, it is found that the level of serum Cys-C in patients may be related to the migration, invasion and metastasis of cancer cells. One of the in vivo studies observed a significant correlation between the occurrence and progression of melanoma and colorectal cancer with elevated serum Cys-C concentrations in patients [30]. Secondly, as SCLC is a more aggressive malignancy than NSCLC, serum Cys-C from SCLC patients may be affected by the malignancy itself, thereby affecting the outcomes of these patients during treatment and their living state [31]. From the Table 3; Fig. 3, the best cut-off value of serum Cys-C level of SCLC group was 0.91 mg/L by ROC analysis with an AUC of 0.64 (95% CI: 0.57–0.71). The sensitivity and specificity were 77% and 45%, respectively. These results showed that serum Cys-C level expression in SCLC was statistically significant compared with BN group. But it was comparable in patients with SCLC and patients with ADC (p = 0.52), SCC (p = 0.12) and LCNEC (p = 0.64) respectively. Table 3Performance of Cystatin C in differential diagnosis of tumor nodulesGroup1Group2AUC(95% CI)Youden IndexSensitivitySpecificityCut offP ValuePPVNPV SCC ADC 0.54(0.47 to 0.60)0.120.180.930.990.23 SCLC ADC 0.52(0.45 to 0.59)0.120.940.180.970.52 LCNEC ADC 0.51(0.38 to 0.64)0.090.320.780.970.85 SCLC SCC 0.55(0.48 to 0.62)0.130.560.570.980.12 LCNEC SCC 0.52(0.41 to 0.68)0.200.250.951.000.65 LCNEC SCLC 0.53(0.42 to 0.64)0.181.000.180.960.64 BN ADC 0.69(0.62 to 0.75)0.310.680.630.93< 0.01 BN SCC 0.70 (0.63 to 0.76)0.320.540.770.93< 0.01 BN SCLC 0.64 (0.57 to 0.71)0.220.770.450.91< 0.01 BN LCNEC 0.71 (0.60 to 0.82)0.370.740.630.93< 0.01 MN BN 0.67 (0.62 to 0.73)0.270.640.630.92< 0.0188.8%27.7%SCC - Squamous Cell Carcinoma, ADC – Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules, AUC – Area Under Curve, CI - confidence interval, PPV and NPV were calculated and presented specifically for the comparison between malignant nodules and benign nodules Performance of Cystatin C in differential diagnosis of tumor nodules SCC - Squamous Cell Carcinoma, ADC – Adenocarcinoma, SCLC - Small Cell Lung Cancer, LCNEC - Large Cell Neuroendocrine Carcinoma, BN - Benign Nodules, MN - Malignant nodules, AUC – Area Under Curve, CI - confidence interval, PPV and NPV were calculated and presented specifically for the comparison between malignant nodules and benign nodules Only few researches mentioned the role of Cys-C in benign diseases of the lung [32–34]. In the study of benign lung diseases, Cys-C was found to be enriched in conditioned medium, and it could also inhibit the growth and activation of lung fibroblasts. Cys-C expression was markedly reduced obviously in mouse and human lungs. Such results suggest that Cys-C protects against the most dominant factor in fibroblast hyperproliferation and activation in damaged lungs [33]. Hmmier et al. [32] tried to find out that Proteomic signatures in bronchoalveolar fluid (BALF) and patient plasma can be used to differentiate benign lung nodules from NSCLC. The results presented in this study indicate that Cys-C was found to be significantly elevated in lung cancer especially in ADC and SCC compared to benign disease and BALF is the most straightforward way to discover and identify candidate lung cancer biomarkers. In our research, BN group had lower level of Cys-C than in MN group and it was statistically significant in two groups. Furthermore, the best cut-off value of Cys-C level for predicting MN was 0.92 mg/L by ROC analysis which means that patient with nodule in lung, whose serum Cys-C level over 0.92 mg/L, may be more possibilities considered as malignant nodule. However, it is important to note that this cut-off value was derived from a single-center cohort and requires external validation in larger, multi-center populations before clinical application, as demographics and comorbidities may influence the optimal threshold. Overall, there are still some deficiencies in this study that need to be brought to the attention of readers. First, the number of studies included in this study was relatively small, and the overall sample size was smaller than reported in other articles. Second, there is irreparable heterogeneity among the literatures included in the analysis. Including the clinical characteristics of patients (such as age, sex ratio, tumor stage, pathological type and ethnicity, etc.), data analysis methods, technical platforms of experimental research, cutoff valuesof data results, and differences in sensitivity and specificity of detection. Third, incidence, diagnostic criteria and medical resources vary in different regions, which might bring discrepancy in analyzing effect of Cys-C. Furthermore, there is publication bias among the included studies, which may affect the final results. The most likely reason for publication bias is that reports of positive research results will receive more attention from researchers, while negative results are rarely published and cited, so they are easily missed by researchers. It is important to note that the optimal cut-off value for cystatin-C identified in this single-center study may not be directly generalizable to other populations due to differences in demographics and comorbidities. External validation in larger, multi-center cohorts is necessary to confirm our findings and establish more universally applicable thresholds. In response to these problems, in our future research, we should describe the experimental design and the characteristics of the subjects in more detail, clarify the diagnostic criteria and treatment range, increase the sample size of the study, and control the deviation of each stage of the experiment. Therefore, large, prospective, multicenter studies are the next step to consider. Conclusion The present study confirmed a relationship between serum Cys-C levels in patient with lung cancer or benign nodule of the lung, and evaluated serum Cys-C levels in different pathological types of lung cancer and benign nodule of the lung. The display results of imaging can promote the accurate diagnosis of pulmonary nodules to a certain extent. However, these methods have limitations because they cannot be performed immediately at the bedside. Compared with simple blood test, there is no trend to reduce cost-effectiveness as a routine screening method for healthy people. Our study suggests that the serum Cys-C level test may be helpful for differentiating lung cancer from the benign nodule of the lung in patient. Acknowledgements We greatly appreciate the assistance of the staff of the Department of Respiratory and Critical Care Medicine and Department of Thoracic Surgery, West-China Hospital, Sichuan University for their efforts. Abbreviations cystatin C body mass index glomerular filtration rate receiver operating characteristic area under the curve non-small cell lung cancer squamous cell carcinoma small-cell lung cancer large cell neuroendocrine carcinoma benign nodules Malignant nodules (18) F-fluorodeoxyglucose-positron emission tomography/computed tomography bronchoalveolar fluid; TNM: Tumor, Node, Metastasis (cancer staging system) Author contributions YH was involved in drafting the manuscript. YH and YZ had undertaken the revision of the article and was involved in acquisition of data and preparing the figures. LH was in charge of collecting the primary data. CS designed and revised the manuscript. All authors have read and approved the final manuscript. Funding This paper was supported by the Natural Science Foundation of Sichuan Province (No.2024NSFSC1924) (to Cheng SHEN). Data availability All data for this study are publicly available and are ready for the public to download at no cost from the official websites of the PubMed, EMBASE, the Web of Science and China National Knowledge Infrastructure. There is no need to have the formal permission to use data for this study. The sources and data robustness have been described in the section of “Methods”. Declarations Ethics approval and consent to participate The studies involving human participants were reviewed and approved by West China Hospital Ethics Committee (No. 202231) and affiliation of ethics committee. The patients provided their written informed consent to participate in this study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. YuanLin He and LiJun He contributed equally to this work and should be considered co-first authors. References Associated Data Data Availability Statement All data for this study are publicly available and are ready for the public to download at no cost from the official websites of the PubMed, EMBASE, the Web of Science and China National Knowledge Infrastructure. There is no need to have the formal permission to use data for this study. The sources and data robustness have been described in the section of “Methods”.