Introduction

Warfarin is an oral anticoagulant of the bicoumarin derivative class that works by inhibiting the synthesis of coagulation factors through vitamin K in hepatic cells. It is widely used in the prevention and treatment of thrombotic diseases1. Owing to its efficacy and low price, warfarin is the most widely used oral anticoagulant in clinical practice2. Despite its widespread use, determining the optimal and safest dosage remains challenging owing to its narrow therapeutic window3. Although the introduction of the International Normalized Ratio in the early 1990s allowed for a more consistent and universal management of warfarin therapy, this drug continues to be a key cause of drug-related adverse events, including thromboembolism, ecchymosis, and severe gastrointestinal or intracranial hemorrhage4. Hence, it is essential to personalize the warfarin dosage. In recent years, several studies have shown that genetic factors, including polymorphisms in CYP2C9, VKORC1, and CYP4F2, along with 30 other key factors5, are major contributors to inter-individual differences in warfarin maintenance doses.

The 2017 Clinical Pharmacogenomics Implementation Consortium Guidelines for Genetic Pharmacology to Guide Dosing of Warfarin recommend the three genes with the strongest current scientific evidence for use in guiding warfarin dosing, namely CYP2C9, VKORC1, and CYP4F26. The CYP2C9 gene encodes an enzyme essential for warfarin metabolism, converting it into inactive components. Genetic polymorphisms in the human CYP2C9 gene, such as CYP2C9*3 (c.1075 A > C, rs1057910) and CYP2C9*2 (c.430 C > T, rs1799853), which are common in the Chinese population, reduce the activity of the CYP2C9 enzyme. This results in slower metabolism and clearance of warfarin, making patients more sensitive to the drug and requiring a reduced dose to minimize adverse reactions7,8. Genetic polymorphisms in the VKORC1 gene, which encodes a subunit of the vitamin K epoxide reductase complex, also cause changes in the enzymatic activity of VKORC, which in turn affects the anticoagulant effect of warfarin. In particular, VKORC1 (c.-1639G > A, rs9923231) leads to differences in gene promoter activity, with patients carrying allele A requiring a lower dose of warfarin than GG-pure patients9,10,11. CYP4F2, a member of the CYP superfamily mainly found in the liver and kidney, acts as a monooxygenase of vitamin K. The CYP4F2*3 (c.1297G > A, rs2108622) polymorphism contributes to individual variations in warfarin metabolism in 1-2% of cases. Patients with the AA genotype require a higher warfarin dosage to achieve the same anticoagulant effect12,13,14.

Current methods for detecting individualized dose-related polymorphisms in warfarin include restriction endonuclease fragment polymorphism analysis, fluorescence quantitative polymerase chain reaction (PCR), high-resolution solving curves, allele-specific PCR, denaturing high-performance liquid chromatography, gene chips, and gene sequencing15,16,17,18. However, these methods require purified nucleic acids as templates; this key step increases the testing time and cost. In addition, disadvantages such as expensive instruments and the inability to simultaneously detect multiple gene loci make them unsuitable for clinical testing. In this study, PCR was performed using transgenic DNA polymerase, enabling direct PCR amplification from whole blood samples with just 1 µl of blood, eliminating the need for DNA extraction and minimizing contamination. The use of nested PCR technology enables simultaneous detection of multiple specimens, greatly reducing detection time and meeting the batch processing and speed requirements of clinical testing. In summary, this study aimed to establish the simultaneous detection and genotyping of four polymorphic loci (rs1799853, rs1057910, rs9923231, and rs2108622) in three warfarin-individualized dosing-related genes (CYP2C9, VKORC1, and CYP4F2) in a single reaction system using combined nested PCR and fluorescent probe lysis curve technology.

Materials and methods

Blood sample collection

Patients who received warfarin treatment after cardiac stenting from May to September 2023 at the First Affiliated Hospital of Gannan Medical University were enrolled. In total, 181 whole blood specimens were collected using EDTA anticoagulant or citrate anticoagulant. Informed consent was obtained from all the participants. This study was approved by the Ethics Committee of First Affiliated Hospital of Gannan Medical University. All the studies were conducted in accordance with the principles of the Declaration of Helsinki. Whole blood specimens were collected and stored at 4–6 °C for backup and at -80 °C for long-term storage.

Synthesis of plasmids and design of primer probes

Whole-genome sequencing data for CYP2C9, VKORC1, and CYP4F2 were obtained from the NCBI database (https://www.ncbi.nlm.nih.gov/). Primer premier 5.0 and Oligo 6.0 software were leveraged to design four self-quenching probes based on the rs1799853, rs1057910, rs9923231, and rs2108622 polymorphic sites, and the amplification primer pairs were designed according to the corresponding probes. The primers and probes were compared using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to ensure the specificity of amplification and detection. Similarly, the gene sequences of these four polymorphic sites and gene variants were selected for the synthesis of plasmids and homologous sequences. All primers and probes were synthesized by Shanghai Bioengineering Company Limited. The corresponding sequences are listed in Table 1.

Table 1 Gene polymorphism site primers, probe sequences.
Full size table

Polymerase chain reaction

The PCR system was optimized and tailored to its specific components (final concentrations): 15 µL of 2×SuperEasyTM Mix (UNG)-EDTA, 0.08 µM each of upstream primers for CYP2C9*2, CYP2C9*3, VKORC1, and CYP4F2*3, 0.33 µM of downstream primers, 0.08 µM of probes, and 1 µL of whole blood (DNA template), bringing the total volume to 30 µL.

The PCR reaction experiment used SLAN® 96P fluorescent quantitative PCR (Hongshitech, Shanghai, China), with the following reaction conditions: UNG enzyme digestion was set at 37 °C for 5 min. PCR enzyme activation and template degeneration was set in the following order: 94 °C for 5 min, 94 °C for 30 s, 55 °C for 60 s; 50 cycles to obtain amplification products. The fluorescence signal of the corresponding detection channel was collected during the annealing stage, and the melting curve was analyzed after PCR. Melt curve analysis procedure was as follows: denaturation at 94 °C for 2 min and 45 °C constant temperature for 2 min. The melting curve was analyzed utilizing SLAN® 96P fluorescence PCR control software (Hongshitech, Shanghai, China), and 181 PCR products were sent to Anhui General Biology Co., Ltd. for bidirectional sequencing. The Applied Biosystems 3500 genetic analyzer (Carlsbad, California, USA) was used for sequencing.

Statistical analyses

SPSS 18.0 software was used for data processing, with the χ2 test employed to assess whether the allele and genotype frequency distributions of the rs1057910, rs9332127, rs9923231, and rs2108622 polymorphic loci conformed to Hardy–Weinberg equilibrium. The Stata 12.0 software was used to calculate the correlation between genotype and warfarin dose, along with the 95% confidence interval.

Result

Evaluating the effectiveness of the PCR melting curve method

Four pairs of primers and four fluorescent probes were designed to target and detect the four gene polymorphism sites, CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), VKORC1 (rs9923231), and CYP4F2*3 (rs2108622). Each fluorescent probe was labeled with different fluorescent motifs corresponding to different detection channels (FAM, HEX, CY5, and ROX). Plasmid DNA standards for different genotypes were used to establish standard melting curves for the four gene polymorphism sites (Fig. 1a-d). The melting curves obtained from the wild-type and pure mutant plasmid DNA standards of the four polymorphic loci showed a single melting peak (with different Tm values). Two melting peaks appeared when the specimens tested were heterozygous mutants. Therefore, the number of melting peaks and Tm values could be used to genotype the four polymorphic loci (the Tm values of the melting peaks corresponding to different genotypes are listed in Table 2.)

Fig. 1
figure 1

Plasmid DNA standards to establish standard melting curves for polymorphic sites in four genes. (a–d) The polymorphic site melting curves graph of CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), VKORC1 (rs9923231) and CYP4F2*3 (rs2108622), respectively. As CYP2C9*2 gene for example, a single melting peak appeared and the Tm value was 49.99 ± 1.5 °C, the gene was judged to be wild-type. a single melting peak appeared and the Tm value was 59.08 ± 1.5 °C, the gene was judged to be homozygous variant. two melting peaks appeared, and the Tm values were 49.99 ± 1.5 °C and 59.08 ± 1.5 °C, respectively, the gene was judged to be a heterozygous variant. Genotyping of four polymorphic sites was achieved by the number of melting peaks and Tm values.

Full size image
Table 2 Tm values of wild-type and mutant melting peak in each gene polymorphism site.
Full size table

Next, we tested seven clinical whole blood samples. The assay genotyping results of the seven clinical samples (Fig. 2a-g) showed 100% concordance with the sequencing results (Table 3), and there were no melting peaks in the homologous sequence plasmid standards for polymorphic loci (Fig. 2h). Two clinical whole blood samples were randomly selected for duplicate testing (20 replicate wells per sample) (Fig. 2i-j). These results indicated that the assay had good accuracy, specificity, and reproducibility.

Fig. 2
figure 2

Evaluating the effectiveness of the PCR melting curve method. (a–g) The melting curves graph of 4 gene polymorphisms in 7 clinical whole blood samples, and the number of white blood cells in whole blood specimens is 2.59×109/L, 4.08×109/L, 7.28×109/L, 10.24×109/L, 15.05109/L, 20.99×109/L, 40.07×109/L respectively; (h) The homologous sequence melting curve graph of the CYP2C9 and VKORC1 genes, including homologous sequences CYP2C19, CYP2E1, and CYP2A13-TYW1; (i,j) The melting curves obtained from 20 repeat trials of two randomly selected clinical specimens.

Full size image
Table 3 Comparison of gene polymorphisms detected by multiplex fluorescence.
Full size table

Lower limit of white blood cell number of PCR melting curve method

For clinical testing convenience, we also determined the lower limit of the minimum number of white blood cells required for our method to estimate the minimum blood volume needed for testing. We tested seven clinical whole blood samples corresponding to white blood cell counts of 2,590, 4,080, 7,280, 10,240, 15,050, 20,990, and 40,070. The melting curve peaks of all the assay results distinguished the different genotypes well (Fig. 2a-g). Approximately 1 µL of whole blood of a healthy individual contains 4000–10,000 white blood cells. Thus, we conclude that 1 µL of whole blood was sufficient for our method.

Patients with clinical warfarin anticoagulation usually require monitoring of prothrombin time and international normalized ratio (with a sodium citrate anticoagulant). We tested EDTA-anticoagulated whole blood samples and sodium citrate-anticoagulated whole blood samples, and the results showed that neither anticoagulant affected the test results (Fig. 3a-e). This indicated that the assay was easy to perform in a clinical setting.

Fig. 3
figure 3

Analysis of detection results of gene polymorphism by different anticoagulants. (a–e) The melting curves obtained by multiplex fluorescence melting curve assay of whole blood direct amplification of 5 patients whole blood samples anticoagulated by EDTA and sodium citrate, respectively. The melting curves result of four polymorphisms loci from two samples of whole blood from the same patient were consistent.

Full size image

Detection result analysis of patient specimens treated with warfarin

Whole blood specimens from 181 patients treated with warfarin were tested for four polymorphic loci, CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), VKORC1 (rs9923231), and CYP4F2*3 (rs2108622), using whole-blood direct-amplification multiplexed fluorescence lysis curve technology. The results showed the following (Fig. 4a-b) (Table 4): all genotypes at the CYP2C9*2 (rs1799853) locus were wild-type (CC, genotype frequency 100%); 89.0% of the genotypes at the CYP2C9*3 (rs1057910) locus were wild-type (AA), 11% were heterozygous mutant (AC), and no purist mutant (CC) was detected; most genotypes at the VKORC1 (rs9923231) locus genotypes were primarily pure heterozygous mutant (AA), accounting for 85.6%, and wild-type (GG) accounted for only 0.6%; CYP4F2*3 (rs2108622) locus genotypes GG, GA, and AA were 58.0%, 38.1% and 3.9%, respectively. All samples were sequenced (Supplementary Table 1), and the sequencing results were in complete agreement with the results of the method used, indicating that the reaction system of the whole blood direct amplification multiplex fluorescence melting curve technology had high specificity and accuracy. In addition, we analyzed the relationship between the mutation results of the relevant genes in patients and the warfarin dose. The results revealed that patients with mutations in the VKORC1 gene had a lower warfarin dose than that of the wild type and a significantly longer warfarin dose adjustment cycle. The intermediate-metabolizing population (CYP2C9*1*3) had a lower warfarin dose than the normal metabolizing population (CYP2C9*1*1) and a high risk of hemorrhage. Carriers of the A allele in the CYP4F2*3 gene had a significantly higher warfarin dose than that of the wild type (Fig. 5a-b).

Fig. 4
figure 4

Gene polymorphism detection melting curve of 181 clinical samples. (a) Melting diagram of sample No. 1–95; (b) Melting diagram of sample No. 96 − 18.

Full size image
Table 4 Genotyping results of 181 samples and frequency of genetic polymorphisms.
Full size table
Fig. 5
figure 5

Correlations between genotyping of 4 polymorphisms and warfarin dose. Divided into 8 groups, CYP2C9*1*1, CYP2C9*1*3, VKORC1 (GG), VKORC1 (GA), VKORC1 (AA), CYP4F2*3 (GG), CYP4F2*3 (GA), CYP4F2*3 (AA) according to the polymorphism site gene detection results. Among them, CYP2C9*1*1 represents CYP2C9*2 and CYP2C9*3 are both wild types, and CYP2C9*1*3 represents CYP2C9*2 wild type and CYP2C9*3 heterozygous variant type. Comparison of genotyping with clinical warfarin dose (a) and corresponding PT-INR values (b). The difference in p < 0.05 was statistically significant.

Full size image

Discussion

Warfarin dose is significantly influenced by the CYP2C9*2, CYP2C9*3, VKORC1, and CYP4F2 alleles, which is why these genetic variants have been incorporated into algorithms for determining the initial warfarin dose in clinical settings19. The frequencies of these alleles and their impact on warfarin dosing have been studied in patients and healthy donors of different ancestries, including African Americans, Caucasians, Japanese, Han Chinese, Indians, and Hispanics20,21,22. The results of these previous studies highlight the differences in the frequency of these alleles and, therefore, in the general warfarin dose requirement among individuals from different geographical regions. Various methods have been developed to detect polymorphisms in individual warfarin dosing-related genes. Although these methods are available and used in clinical practice, their cumbersome and time-consuming procedures and limited number of detectable sites have restricted their broader clinical application. Liu et al. developed a fully integrated and automated microsystem consisting of disposable plastic chips for DNA extraction and PCR amplification combined with a reusable glass capillary array electrophoresis chip in a modular-based format was successfully developed for warfarin pharmacogenetic testing23. Although this method is greatly improved compared to some previous methods, it still requires DNA extraction, which is more costly and cumbersome. The whole blood direct amplification multiplex fluorescence melting curve technique established in this study enables genotyping of the warfarin loci at CYP2C9*2 (rs1799853), CYP2C9*3 (rs1057910), VKORC1 (rs9923231), as well as CYP4F2*3 (rs2108622) genotyping of the warfarin locus. This method accurately detected four nucleotide polymorphism sites with good reproducibility. Fluorescence intensities of the four channels were determined using fluorescent probes. The melting curves of each polymorphic site had a △Tm value greater than 5℃, which prevented misinterpretation of genotypic results caused by the close proximity of the melting peaks between single nucleotide mutation sites and unmutated gene sites. The method requires only 1 µL of whole blood, no DNA extraction, takes less than 2 h, costs less than $1, is highly sensitive, can accurately distinguish between different single-nucleotide polymorphism (SNP) sites, and can meet the needs of most clinical whole blood samples for genetic polymorphism detection.

Genotyping of four SNP loci in the CYP2C9, VKORC1, and CYP4F2 genes was performed on whole blood samples from 181 patients in the Ganzhou region following warfarin treatment. The aim was to assess the mutation frequency of genes related to individualized warfarin administration and their relationship with warfarin dosing. As shown in Table 4, VKORC1 (c.-1639G > A, rs9923231) had the highest mutation frequency among the genes related to warfarin individualized dosing, with a frequency of allele A as high as 92.5%; the frequencies of alleles G and A at CYP4F2*3 (c.1297G > A, rs2108622) were 77.1% and 22.9%, respectively; CYP2C9*3 (c.1075 A > C, rs1057910) had allele A and C frequencies of 94.5% and 5.5%, respectively; CYP2C9*2 (c.430 C > T, rs1799853) was detected only with allele C (allele frequency of 100%), and no mutant genes were detected, which is in line with the distribution pattern of the gene in Asian populations. Analysis of the relationship between gene mutations related to individualized warfarin dosing and warfarin dosage in 181 patients revealed that those with VKORC1 gene mutations required lower doses of warfarin than the wild type group and had significantly longer dosage adjustment cycles. Additionally, patients with intermediate metabolism (CYP2C9*1*3) required warfarin doses than those with normal metabolism population (CYP2C9*1*1) and had a higher risk of hemorrhage. Carriers of the A allele in the CYP4F2*3 gene had significantly higher warfarin doses than the wild type. In summary, the CYP2C9*2, CYP2C9*3, VKORC1, and CYP4F2*3 genes are the primary factors affecting differences in individual warfarin doses in the Chinese population. Genetic polymorphism testing in patients undergoing warfarin therapy is crucial for shortening the dosage adjustment cycle and reducing the risk of bleeding.

However, this study had certain limitations. Our assay was performed in a controlled environment, and measurements were taken for only a subset of the population. In addition, we were unable to assess certain clinical factors of the patients, such as physical activity and dietary structure, which may affect the daily stable dose of warfarin24,25. Furthermore, owing to the limited sample size in this study, expanding the sample size in future phases is essential. Additionally, timely recording of initial doses and dosage adjustment cycles is necessary to provide a theoretical basis for individualized warfarin dosing in clinical practice.

In summary, we established a reliable direct blood PCR method to detect CYP2C9*2, CYP2C9*3, VKORC1, and CYP4F2*3 SNP loci. This study also demonstrated that the frequencies of CYP2C9*2, CYP2C9*3, VKORC1, and CYP4F2*3—the key genetic factors influencing warfarin dose requirements—differ in a sample from Ganzhou (China) compared to other populations of diverse ancestries. This highlights the need to evaluate various Chinese populations to accurately determine the true frequency of these genetic variations.