This study was a prospective observational study. The study aimed to validate the performance of an in-house 2-steps PCR-SSP against the sequencing method for the detection of HLA-B*57:01 and HLA-B*58:01 alleles.

DNA samples were selected from residual specimens used in routine HLA testing at Molecular Biology Laboratory, Division of Transfusion Science, Department of Medical Technology, Faculty of Associated Medical Sciences, Chiang Mai University. Purposive sampling was employed to select DNA samples with known presence of HLA-B*57:01 and/or HLA-B*58:01 alleles to ensure the inclusion of positive cases for concordance analysis. The sample size calculation was based on an expected prevalence p = 50% to reflect the intention of sample selection rather than estimate prevalence using the actual allele frequencies in the population. For reference, allele frequencies in Thai population reported in Allele Frequency Net database (http://www.allelefrequencies.net/hla.asp) are approximately 1.8% for HLA-B*57:01% and 7.7% for HLA-B*58:01 (Gonzalez-Galarza et al., 2020). These frequency values are provided for further random sampling. The sample size formula applied was; n = Z ² × p × (1−p)/d ², where Z is 1.96 for 95% confidence, p is the expected prevalence (here set to 0.5) and d is precision (0.05) (Pourhoseingholi et al., 2013). The calculated sample size based on this formula was approximately 385 samples. However, in this preliminary study, a smaller sample size was chosen due to the limited availability of positive samples with a total of 30 samples purposively selected to enable direct concordance analysis.

DNA was extracted using QIAamp^® Blood Mini Kit (QIAGEN, Germany). DNA purity and concentration was then measured using the EON™microplate spectrophotometer (RI technologies ltd., Singapore) based on the principle of light absorption.

The original primer designs, as published by Martin et al. (2005) and Virakul et al. (2010) for HLA-B*57:01 and HLA-B*58:01, respectively, were modified to facilitate a 2-step PCR-SSP approach binding across exon 2–3 boundaries (patent pending). The reference sequences for HLA-B*57:01:01:01 (IMGT/HLA Acc No: HLA00381) and HLA-B*58:01:01:01 (IMGT/HLA Acc No: HLA00386) was retrieved from IMGT/HLA database. To guarantee effective and targeted amplification within a two-step cycling, the modifications focused on enhancing primer annealing and extension properties. All primers have T_m of 73 °C–75 °C and GC content 60%–68%. The in silico validation of primers was performed using multiple bioinformatics tools to confirm their specificity and efficiency. For specificity, the primers designed to target HLA-B*57:01 and HLA-B*58:01 alleles were validated using BLAST against the nucleotide BLAST in NCBI database with the Human RefSeq Gene database and megablast selection to confirm allele specificity and exclude cross-reactivity with non-target alleles or related sequences. For cross-reactivity within the HLA region, Probe and Primer Search Tool in IPD-IMGT/HLA database was used. Moreover, the primers were also assessed for melting temperature (Tm), GC content, self-complementarity, and the absence of potential primer-dimer formations using Primer3 software. These metrics ensured high specificity, minimized nonspecific amplifications, and optimized annealing conditions. Additionally, an internal control primer targeted the human growth hormone (HGH) gene. The reference sequence for the HGH gene (accession no. NG_042788) was obtained from RefSeq database. Both forward and reverse primer of internal control were also analyzed for specificity and cross-reactivity using the nucleotide BLAST in NCBI database.

For the detection of HLA-B*57:01, a 12.5 µL reaction mix was prepared containing 6.25 µL of 2x Quick Tag HS Dye mix (Toyabo, Japan), 2.0 µL of 10 µM specific forward and reverse primers, 1.0 µL of 10 µM internal control forward and reverse primers, and 1.25 µL of sterilized distilled water. Input DNA of 100 ng yielded optimal amplification.

For detection of HLA-B*58:01, a 12.5 µL reaction mix was prepared containing 6.25 µL of 2x Quick Tag HS Dye mix (Toyabo, Japan), 1.0 µL of 10 µM specific forward and reverse primers, 1.0 µL of 10 µM internal control forward and reverse primers and 3.25 µL of sterilized distilled water. Input DNA of 25 ng yielded optimal amplification.

The PCR amplification for both HLA-B*57:01 and HLA-B*58:01 followed this steps: 1 cycle of an initial denaturation at 94 °C for 2 min, followed by 30 cycles of 94 °C for 30 s and extension at 68 °C for 1 min. The reaction was then held at 4 °C. A summary of PCR protocol was presented in Table 1. Following the PCR amplification, the quality of the PCR product was assessed using 2% agarose gel electrophoresis at 100 V for 40 min. The key characteristic of the DNA polymerase in the 2x Quick Tag HS Dye mix that allows a 2-step PCR protocol is its robust activity and fidelity at relatively high annealing/extension temperatures, enabling the annealing and extension steps to be combined as shown in PCR protocol. Also the enzyme’s optimal activity is at 68 °C which allows the PCR cycling to be simplified into two steps with not very high extension temperature. These reduces total PCR run time and reduces handling steps without loss of sensitivity or specificity (Toyobo Co.Ltd, 2025).

In this study, the full-length HLA-B was amplified and barcoded using the primers from Matern et al. (2020). The barcoded amplicons were pooled and library preparation was performed using ligation sequencing V14 (SQK-LSK114, Oxford Nanopore Technology, United Kingdom) before loading onto the flow cell, which was the MinION R10.4.1. (FLO-MIN114, Oxford Nanopore Technology, United Kingdom) The sequencer used was the GridION and the sequencing steps followed the recommended protocols provided by Oxford Nanopore Technologies (ONT). Subsequently, the analysis step was performed using an in-house bioinformatics tool. Nanopore sequencing data were processed using an in-house bioinformatics pipeline adapted from Anukul et al. (2023). Nanopore sequencing data were acquired with MinKNOW (Oxford Nanopore Technologies, ONT), which generated POD5 files. Basecalling, demultiplexing, and simultaneous adapter and barcode trimming were performed using Dorado with SUP model (ONT), yielding per-sample FASTQ reads. The reads were aligned to HLA reference sequences for HLA-B downloaded from the IPD-IMGT/HLA database (version 3.48; https://www.ebi.ac.uk/ipd/imgt/hla/) using Minimap2 v2.28 (Li, 2018). Small-variant calling was performed with Clair3, and all variants were phased using WhatsHap v1.6 (Patterson et al., 2015) to produce phased VCF and BAM files. For each haplotype block, supporting reads were extracted from the phased BAM and polished with Racon v1.5.0 (Vaser et al., 2017) to generate allele-specific consensus sequences. Each consensus allele was aligned against the IPD-IMGT/HLA database using BLASTN v2.13 (Altschul et al., 1990), and the best match with the highest bit-score was selected as the final assignment for each HLA allele. This pipeline achieves high-resolution typing (fields 3–4) with low error rates and has demonstrated robust performance across various HLA class I loci beyond HLA-B*57:01 and HLA-B*58:01.

PCR-SSP results were compared to sequencing using agreement measured by Cohen’s kappa with 95% CI. With the observed perfect agreement (kappa = 0.81-1.00), it indicates the strange of agreement as almost perfected at a significance level of 0.05 (Wang et al., 2019).

The presence of the HLA-B*57:01 allele is indicated by the appearance of two specific bands for HLA-B*57 at 226 bp and HLA-B*57:01 at 108 bp. Both amplicons were located within exon 3 of HLA-B allele based on sequence alignments from the IMGT/HLA database. This dual-band approach reduces the risk of false positives and allows confident identification of the HLA-B*57:01 allele. A schematic figure showing primer positions and expected band sizes of HLA-B*57:01 allele detection is shown in Figure 1. However, in HLA-B*57:01 negative sample, only one band of the internal control at 429 bp will show up (Figure 2). Among the thirty samples tested, nine samples were positive for the HLA-B*57:01, while twenty-one samples were negative (Table 2).

According to the test results and interpretation of the HLA-B*58:01 test using in-house 2-steps PCR-SSP method, it was found that using a DNA concentration of at least 100 ng/μL which similar to the HLA-B*57:01 test, was not suitable for detecting HLA-B*58:01. Therefore, the appropriate concentration for this test was determined to be 25 ng/μL. Notably, the optimal DNA input for HLA-B*58:01 detection was lower (25 ng/μL) compared to HLA-B*57:01 (100 ng/μL). Higher DNA concentrations may result in reduced amplification efficiency for HLA-B*58:01, likely due to PCR inhibition or competition effects between genomic DNA carried non-target alleles, specific primers and internal control primers (Chou et al., 1992; Schrader et al., 2012). Reducing the template concentration can improve primer specificity and amplification efficiency for HLA-B*58:01, yielding clearer and more specific PCR bands.

The presence of the HLA-B*58:01 allele is indicated by the appearance of a specific band at 358 bp spans within exon 2-3 of HLA-B allele (Figure 1). Sample lacking the HLA-B*58:01 showed only one band of the internal control at 141 bp (Figure 3). Among the thirty samples tested, eight samples were positive for HLA-B*58:01 and twenty-two samples were negative (Table 2).

A total of nine samples were positive for HLA-B*57:01 and eight samples were positive for HLA-B*58:01, while the remaining samples were negative for the HLA-B*57:01 and HLA-B*58:01 alleles (Table 2).

The reliability of the in-house 2-step PCR-SSP method was assessed by comparing its results with those from nanopore sequencing (Table 2). The two methods provide different levels of detection resolution. The 2-step PCR-SSP method offers a maximum detection resolution of 2 fields, while the nanopore sequencing method provides a minimum detection resolution of 3 fields. This indicates that the test results obtained from the nanopore sequencing are more detailed and precise, enabling better data differentiation compared to the 2-step PCR-SSP method, which has lower resolution. However, when the results from both methods were compared and statistically examined for consistency using Cohen’s kappa statistic (kappa score), both methods produced a kappa value of 1.00 indicating highly consistent at 2-field resolution. In terms of time consuming and cost per sample of the assay, in-house 2-steps PCR-SSP assay required 1 h PCR hand-on time excluding DNA extraction and agarose gel analysis and reagents costing USD 7 per sample, versus 3–5 h and USD 190/sample for Oxford Nanopore sequencing (Anukul et al., 2023).