With the help of the hospital information system, we screened the cases of hospitalized tumor patients using tigecycline in Sun Yat-sen University Cancer Center from November 2015 to October 2019 using a retrospective analysis method. The exclusion criteria: 1) patients with hospitalization periods <4 days; 2) patients with severe hepatic insufficiency (Child-Pugh class C); 3) pregnant or lactating women; 4) patients with abnormal coagulation or congenital coagulation disorders such as hemophilia before hospitalization; 5) patients using other drugs that affect coagulation indexes 2 weeks before or during the medication; 6) patients without the detection indexes of coagulation function before and after the medication; 7) patients with incomplete medical records. All participants provided written informed consent according to the regulations of the Institutional Review Boards of the Hospitals in agreement with the Declaration of Helsinki with the number SL-B2021-340-01.

The main safety observation indexes: the coagulation function-related indexes during hospitalization, including prothrombin time (PT) before and after treatment, prothrombin time activity (PT%), international standardization ratio (INR), partially activated prothrombin time (APTT), fibrinogen (FBG), thrombin time (TT), D-dime and fibrin degradation product (FDP). Minor safety observation indexes: the levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatinine (CRE), and C-reactive protein (CRP) before and during the treatment with tigecycline.

We defined the abnormal coagulation index as the APTT higher than the normal value of 10 s (i.e., APTT >50 s), the PT and TT higher than the normal values for 3 s (i.e., PT > 17 s and TT > 24 s), or the FBG lower than the normal value (i.e., FBG < 2 g/L) and paid attention to the platelet count (PLT) lower than normal value (PLT <100×10⁹/L). In logistic regression analysis, patients with the ages of < years old, 19–60 years old, and ≥61 years old were divided into the adolescent group, the normal age group, and the elderly group, respectively; patients with the tigecycline use times of 7 days, 8–14 days, and >14 days were divided into the medium-course treatment group, the medium-long-course treatment group, and the long-course treatment group, respectively; patients with both ALT and AST ≤50 IU/L were considered to have a normal liver function, and patients with ALT or AST >50 IU/L were considered to have an abnormal liver function; patients with CRE ≤106 μmol/L were considered to have a normal renal function, patients with CRE >107 μmol/L were considered to have an abnormal renal function.

The cell adhesion assay was performed as previous described (Huang et al., 2015). Chinese hamster ovary cells were obtained from Da-Wei Wang (Ruijin hospital) and cultured in F12 (Gibco, NY, United States) supplemented with 10% fetal bovine serum (Gibco, 10270-106) and 2 mM Glutamine (Gibco, NY, United States). The cells were seeded into 96-well plates at a density of 10⁵ cells/well. To determine the biological effect of the tigecycline (Hansoh Pharma, Jiangsu, China), cells were separately incubated with appropriate concentrations (0 mg/ml, 0.2 mg/ml, and 0.5 mg/ml) of tigecycline. After 48 h, 10 μL of reagent from a Cell Counting Kit-8 (CCK-8, Dojindo Laboratories, Kumamoto, Japan) was added to each well. The samples were then incubated at 37°C for an additional 4–6 h and measured at the absorbance of 450 nm using a spectrophotometer.

The cell spreading assay was performed as previous described (Huang et al., 2015). The tigecycline was diluted into three groups of 0 mg/ml, 0.2 mg/ml, and 0.5 mg/ml and treated in CHO cells for 48 h. Then, 10 mg/ml BSA was added to a 96-well plate at 50 μL/well, coated at 4 °C overnight, washed twice with DPBS the next day, and blocked with 20 mg/ml BSA at room temperature for 2 h. The CHO cells treated with tigecycline for 48 h were digested and made into a single-cell suspension, washed twice with 1 ml of F12 medium, and then washed with 1 × 10⁶/ml density suspended in the medium. Next, 100 μL of cell suspension was added to the coated 96-well plate. Each group of cells was connected to 3 replicate wells, placed in a 37°C, 5% CO₂ cell incubator for 1–2 h, and then observed under the microscope when the CHO cells were well stretched. Incubation was terminated, and non-adherent cells were removed by being washed three times with DPBS. After being washed with DPBS, pictures were taken under a microscope, and ImageJ software was used to calculate the area of single cells one by one. We defined cells with an area of ≥30 μm² as stretching, and cells with an area of <30 μm² as adherent only. The total number of cells in each group was about 300, and the percentage of cells of each shape in all groups were calculated. Fluorescence microscopy was used to reflect cell viability, and we used cell staining to make cells appear green under fluorescence, while dead cells were not.

The CHO cells treated with tigecycline (0 mg/ml and 0.2 mg/ml) for 48 h were used for RNA-sequencing. KOBAS (http://kobas.cbi.pku.edu.cn/home.do) was used for the KEGG pathway enrichment analysis, in which the calculation principle is the same as that in GO functional enrichment analysis, and the calculation is performed using Fisher’s exact test (Wu et al., 2006). In order to control the calculated false positive rate, the BH (FDR) method was used for multiple testing, and the calculation formula was not the same as that in the previous section. The corrected P-value threshold was 0.05.

SPSS 22.0 statistical software was used for analysis. Measured data were expressed as mean ± standard deviation and were tested for homogeneity of variance. When the variance was homogeneous, a t-test was used, and when the variance was unequal, a corrected t-test was used. A rank-sum test or Ridit test was used for grade data. The categorical data were analyzed using the χ2 test. According to α = 0.05 level, p < 0.05 implied a statistical difference, and p > 0.05 implied no statistical difference.

The function of platelet was detected by thromboelastography as previous studies (Wang et al., 2014; Guo et al., 2021; Miles et al., 2022; Sekar et al., 2022). Assays were performed according to the manufacturer’s instructions (Hemostasis System Kaolin Testing Kit, Glodmag, China). Briefly, samples were fresh whole blood, 2 ml was collected with citric acid anticoagulation, and kept at room temperature for at least 15 min. Then, draw 340 ul blood sample for on-machine testing (Thrombelastography TEG5000, HAS). All samples were detected and analyzed within 2 h after sampling. The main indicators of thromboelastometry were: (1) Reaction time (R) indicated that there was no fibrin formation in the tested sample; (2) Coagulation time (K) indicated that fibrin began to form in the tested sample, which had a certain firmness; (3) The widest distance (MA) on both sides of the curve represented the maximum amplitude of thrombus formation; (4) Thromboelastometry (ε), which represented the size of the elasticity of the thrombus. (5) Maximum coagulation time (m) indicated the time from coagulation time to the maximum amplitude.

A total of 205 cases of tigecycline used in tumor patients were recruited from our hospital, and 129 valid cases were finally included. Basic characteristics of this cases were shown in Figure 1. Among them, the most common patients were hematological tumors, accounting for about 62.8% (81 cases), followed by digestive system tumors, accounting for about 23.2% (Figure 1A). There were 71 males (55%) and 58 females (45%). The average age was (49.5 ± 15.9) years. There were 2 cases (1.5%) in the adolescent group (age <18 years), 47 cases (72.1%) in the normal age group (19–60 age), and 22 cases (26.4%) in the elderly group (≥61 years old) (Figure 1B). In addition, the days of tigecycline administration were 10.58 ± 9.12days and we found 113 cases (87.6%) had abnormal coagulation function.

Then, we further examined which coagulation indexes were affected by tigecycline. After medication, the coagulation indexes, mainly including PT, INR, APTT, TT, D-Dimer, and FDG, showed an upward trend (Figure 1C). As shown in Table 1, the coagulation function of the tumor patients before using tigecycline was near the normal baseline. However, after 3 days of tigecycline administration, the PT of the patients was significantly prolonged (17.56 ± 5.72 s) and aggravated with the prolongation of the administration time. After continuous administration for more than 14 days, the PT of the tumor patients was abnormally prolonged (19.91 ± 12.75 s), and even the lives of the patients might be threatened. The FBG of the patients before tigecycline treatment was 3.98 ± 3.45 g/L and showed a downward trend at different time points after treatment. The FBG decreased significantly after 1–3 days, 4–7 days, 8–14 days, and more than 14 days of medication (p < 0.05). The PT% and FBG decreased in cancer patients with tigecycline treatment, while the TT, APTT, and D-Dimer significantly increased after treatment, and this effect did not increase with the prolongation of the administration time. In addition, the INR also increased after treatment. However, unlike others, the INR peaked within 3 days after administration, then decreased slowly, and remained stable at a relatively high level. Further, the FDP peaked on d 4–14 of tigecycline administration and then gradually declined (Table 1).

In order to further explore how tigecycline causes abnormal coagulation function in tumor patients, we retrospectively analyzed the liver function and platelet function of these tumor patients according to our screening conditions. Our results showed that tigecycline did not affect the liver and kidney function of patients (Table 2). Both the ALT and AST tended to decrease after administration, which might be related to the infection status of patients before treatment. The ALT/AST ratios and serum creatinine did not change significantly before and after treatment (Table 2). We further detected and compared the changes of coagulation function indexes in tumor patients before and after the use of tigecycline by thromboelastography. The R value remained within the normal range (5-10) before and after the use of tigecycline with 5.1 (Before, before use of tigecycline), 8.3 (Day 3, 3 days after use of tigecycline), and 5.6 (Day 14, 14 days after use of tigecycline), respectively. However, the K value did not change significantly after using tigecycline for 3 days, but increased significantly after 14 days of using tigecycline. The prolongation of the K value indicated that the patient was in a hypocoagulable status. In addition, the MA value began to decrease after the third day of use of tigecycline, and further decreased with the prolongation of the medication time with 62.9 (Before), 50.6 (Day 3), and 44.6 (Day14), respectively. The decreased MA value reflected that tigecycline induce platelet function insufficiency (Figure 2). Next, we used CHO cells to explore the mechanism of tigecycline on coagulation function in vitro. We defined cells with an area of ≥30 μm² as stretching, and cells with an area of <30 μm² as adherent only. The results of the cell spreading experiment showed that the average numbers of CHO cells in the three groups of tigecycline with different concentrations (0 mg/ml, 0.2 mg/ml, and 0.5 mg/ml) were 226, 266, and 4, respectively. The percentages of cells with stretched morphology in the three groups were 73.14, 75.78, and 40%, respectively (Figure 3). In addition, the results of the cell adhesion experiment showed that the numbers of the adherent cells also decreased with the increase in the concentration of tigecycline. However, the percentages of cells with adherent morphology increased significantly, which were 26.86% (0 mg/ml), 24.22% (0.2 mg/ml), and 60% (0.5 mg/ml), respectively (Table 3).

To further study the mechanism of tigecycline regulating coagulation abnormalities, we harvested CHO cells treated with tigecycline with the concentrations of 0 mg/ml and 0.2 mg/ml for 48 h for RNA-sequencing. We used the software edgeR to analyze the differentially expressed genes. Based on the gene read-count data, the differential expression was calculated, and the significantly differentially expressed genes were screened (FDR <0.05 and |log2FC| ≥ 1) (Figure 4A). According to the criteria, we screened out a total of 1,269 genes that meet the requirements (Supplementary Table S1). The GO functional enrichment analysis of these differentially expressed genes indicated that membrane-bounded organelle, cellular response to chemical stimulus, and binding pathways were more enriched in the group with tigecycline treatment (Figure 4B and Supplementary Table S2). Then, KEGG pathway enrichment analysis was performed using KOBAS, and calculation was performed using Fisher’s exact test. We found that the expression of genes involved in the TNF signaling pathway, IL-17 signaling pathway, and Focal adhesion was significantly increased after tigecycline treatment (Figure 4C and Supplementary Table S3). Additionally, bulk RNA-sequencing assay showed significant abnormal expression in gene sets associated with platelet-activation and coagulation, such as PKFP, PLA2g7, PDGFRA, PDGFRB, PAFAH1b3, PDGFA, PECAM1, PDGFC, and MPIG6b. Furthermore, adhesion regulating molecules RPN13, CHL1, ICAM5, and TLCN were also down-regulated by tigecycline in CHO cells. These key genes shown above need to be further validated in blood samples of acute myelocytic leukemia patients treated with tigecycline.

The SNPs and AS events can influence gene expression and the corresponding downstream functions (Zimdahl et al., 2004; Chen et al., 2020). Here, we further explored the SNPs and AS events in CHO cells with tigecycline treatment. We used the assembled transcript as a template sequence and aligned the template sequence with the original sequence using Samtools (http://samtools.sourceforge.net/) and VarScan v.2.2.7 (http://varscan.sourceforge.net/) software to find candidate SNPs. We found no significant differences in SNP types and distribution in CHO cells before and after treatment with tigecycline (Figures 5A,B, Table 4 and Supplementary Table S4). In addition, the splicing analysis indicated that the number of AS events was near 30,000, and skipped exon (SE) was the most common type in the two groups (Figures 5B,C). The pie chart of differentially expressed AS events in these two groups showed that SE (69.17%) was also the most occupied subtype, followed by alternative 3’splice site (A3SS,10.74%) and alternative 5’splice site (A5SS, 10.29%) (Figure 5C).