Assessment of a High-Resolution HPLC System to Quantify HbE for α⁰-Thalassemia Diagnosis in HbE Heterozygotes

Abstract

Background

HbE and α⁰-thalassemia are highly prevalent in Southeast Asia and co-inherited in asymptomatic carriers. α⁰-Thalassemia screening in HbE heterozygotes relies on HbE quantification. We evaluated the performance of a high-resolution HPLC system in quantifying HbE and HbA₂ to distinguish HbE heterozygotes with and without α⁰-thalassemia.

Methods

HbE heterozygotes (N=650) presenting EA patterns on capillary electrophoresis (CE) were analyzed using the Premier Resolution (PR) HPLC system (Trinity Biotech). HbA₂ and HbE levels were quantified, Pearson correlations with CE assessed, and diagnostic cutoffs determined. A prospective cohort of HbE heterozygotes (N=884) was used for validation. α⁰-Thalassemia genotyping was performed using gap-PCR.

Results

PR-HPLC confirmed HbE heterozygosity and successfully separated HbA₂ from HbE. The HbA₂, HbE, and combined HbA₂+HbE levels were consistently higher than but correlated with CE values (HbA₂ bias=1.213, r=0.686; HbE bias=0.286, r=0.960; combined bias=1.502, r=0.952). HbE and combined levels were substantially lower in α⁰-thalassemia carriers (14.65%±2.2% vs. 19.38%±2.5%); HbA₂ levels did not significantly differ. HbE cutoff <19% resulted in an area under the curve (AUC) of 0.975, with 100% sensitivity and 91.9% specificity. HbE+HbA₂ <24.0% yielded an AUC of 0.970, with 100% sensitivity and 89.7% specificity. Validation experiments supported the robust performance for HbE detection alone (100% sensitivity, 93.6% specificity, and 94.5% accuracy).

Conclusions

PR-HPLC can quantify HbE precisely, potentially aiding in robust α⁰-thalassemia diagnosis and screening without HbA₂ assessment.

INTRODUCTION

HbE is the most common structural Hb variant in Southeast Asia. It arises from a point mutation in the β-globin gene (HBB: c.79G>A), producing a spectrum of clinical phenotypes ranging from asymptomatic carrier states to transfusion-dependent thalassemia when co-inherited with β-thalassemia mutations [1]. In Thailand, Laos, and Cambodia, HbE prevalence is high (20–40%) [2–4] and α⁰-thalassemia is also common, with frequencies ranging from 20% to 50% [5–8]. HbE and α⁰-thalassemia are often co-inherited, causing severe thalassemia syndromes, with co-inheritance rates of 15–35% reported in HbE heterozygotes in Southeast Asian cohorts [2, 9–11].

In double heterozygotes, reduced α-globin availability in α⁰-thalassemia, along with less efficient αβ-heterodimer assembly by β^E-globin, results in markedly reduced HbE levels [12], suggesting that HbE quantification can be used for α⁰-thalassemia diagnosis and screening in HbE heterozygotes. Accurate HbE measurement is critical, as minor deviations affect α⁰-thalassemia diagnosis. Capillary electrophoresis (CE) and HPLC are widely used for Hb variant identification and precise HbA₂ and HbF quantification [13, 14]. However, HbE quantification via conventional cation-exchange HPLC is limited by the frequent HbA₂ co-elution [15]. Therefore, α⁰-thalassemia screening is based on thresholds for combined HbE and HbA₂ (HbE+HbA₂) rather than HbE alone [9, 16–19]. Although CE provides improved resolution in separating HbE from HbA₂, many laboratories apply combined thresholds, because original cutoffs were derived from HPLC data [10, 20, 21].

The Premier Resolution (PR) HPLC system, approved by the US Food and Drug Administration in 2023 for evaluating hemoglobinopathies, has enhanced separation capability for structurally similar Hbs. We characterized the chromatographic profile of HbE using PR-HPLC, determined the optimal parameters and cutoffs for detecting α⁰-thalassemia in HbE heterozygotes, and validated the selected parameters and cutoffs in a prospective cohort.

MATERIALS AND METHODS

Study design

The study was conducted in two phases: in phase I (June through November 2024), we assessed the correlation between HbE and HbA₂ levels obtained using PR-HPLC and CE and determined optimal cutoff values for HbE, HbA₂, and HbA₂+HbE as screening indicators for detecting double heterozygosity of HbE and α⁰-thalassemia. In phase II (December 2024 through June 2025), we validated the diagnostic performance and reliability of the selected cutoffs.

Sample collection and hematological analysis

In phase I, 650 EDTA-anticoagulated samples were collected from individuals undergoing Hb typing using CE, showing HbE heterozygosity as demonstrated by an EA pattern on the electropherogram—indicating the presence of both HbE and HbA fractions characteristic of HbE heterozygotes—and having a sufficient blood sample volume for complete analysis. The sample size of 650 represents all eligible consecutive cases collected during the study period (June through November 2024). CE was performed using the MINICAP Flex Piercing system (Sebia, Lisses, France) with the following conditions: sample volume, 18 μL, diluted 1:6 with hemolysate buffer; separation voltage, 10,000 volts; run time, 10 mins; and detection at 415 nm. Hb fractions were identified and quantified using the manufacturer’s software, with automatic peak integration and manual verification. Hematological parameters were measured using a Mindray CAL 6000 automated blood cell counter (Shenzhen Mindray Bio-Medical Electronics, Shenzhen, China). Hb profiling and HbE/HbA₂ quantification were performed using a PR-HPLC system (Trinity Biotech, County Wicklow, Ireland) in high-resolution mode within 24 hrs after CE testing.

In phase II, 5,082 prospective EDTA-anticoagulated samples were collected from 1,903 individuals evaluated for anemia or during antenatal screening and 3,179 individuals attending routine annual health checkups, for validation. Immediately after erythrocyte analysis, Hb profiling and HbE/HbA₂ quantification were performed using PR-HPLC.

Hb analysis and measurement of HbE and HbA₂ using PR-HPLC

The PR-HPLC system is an automated ion-exchange HPLC platform designed for the separation and quantification of HbA, HbA₂, and HbF and presumptive identification of common Hb variants, including HbE. It provides superior separation of HbA₂ and HbE peaks compared with conventional HPLC systems. Technical specifications and operating conditions are provided in Supplemental Data Table S1. QC was performed per the manufacturer’s protocols using certified reference materials.

Molecular analysis of α⁰-thalassemia deletions

Genomic DNA was extracted using the GF-1 Blood DNA Extraction Kit (GF-BD-100; Vivantis Technologies, Selangor Darul Ehsan, Malaysia) per the manufacturer’s instructions. Common α⁰-thalassemia deletions (Southeast Asian [SEA; --^SEA] and Thai [THAI; --^THAI]) were identified using multiplex gap-PCR [22].

To identify the South Africa (SA) and Chiang Rai (CR) deletions, two rare α⁰-thalassemia deletions occasionally detected in the Thai population [8], we developed gap-PCR assays. To identify the SA deletion, a 1,129-bp deletion-specific fragment was amplified using primers αG57 (5′-CTCATCCAGACTCTCCAGCT-3′) and α53 (5′-ACTCGAGCTACCCCAAGGAT-3′) [23]. An internal control fragment (314 bp) was amplified with primers A1B (5′-GGTTCCC TGAGCCCCGACACG-3′) and A7 (5′-CTCTGTGTTCTCAGTATTGGAG-3′) [22]. To identify the CR deletion, a 335-bp deletion-specific fragment was amplified using primers NVF (5′-TAGCCTGGAGGA CAGAGTAAG-3′) and NVR (5′-CGGGTCATGTTGAGTAGGAATAA-3′), and an internal control fragment (198 bp) was amplified with primers SAF (5′-GAGAGGGGTATGCAAATATCCGTA-3′) and SA1 (5′-TGGAGTAAATGGCTGGTATCATTG-3′).

PCR mixtures (50 µL) contained 200 ng genomic DNA, 1.5 mM MgCl₂, 200 µM dNTPs, 1.5 M betaine, 5% DMSO, and 1.5 U Taq DNA polymerase (Vivantis Technologies) in a buffer comprising 10 mM Tris-HCl (pH 9.1), 50 mM KCl, and 0.1% Triton X-100. Primer concentrations were as follows: SA: 15 pmol αG57/α53, 1 pmol A1B, 30 pmol A7; CR: 15 pmol NVF, 7.5 pmol NVR, 120 pmol SAF, 30 pmol SA1. Thermal cycles comprised initial denaturation at 94°C for 3 min, followed by 30 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min, and final extension at 72°C for 10 min. PCR products were resolved on 1.5% agarose gels, stained with ethidium bromide (0.5 μg/mL), and visualized using a Gel Doc EZ Imager (Bio-Rad Laboratories, Hercules, CA, USA) equipped with UV transillumination and digital imaging capabilities.

Ethics

This study adhered to the principles of the Declaration of Helsinki and was approved by the Ethics Committee and Institutional Review Board of the University of Phayao, Phayao, Thailand (approval No. 1.2/104/67). Residual samples from routine clinical testing were used. Written informed consent was obtained from all participants.

Statistical analyses

Hematological parameters and HbA₂/HbE values were summarized using descriptive statistics (mean±SD and 95% confidence interval [CI]). Group differences were evaluated using the Mann–Whitney U test. For method comparison, paired differences were assessed using the Wilcoxon signed-rank test. Pearson’s correlation coefficient was used to determine linear associations between HbA₂, HbE, and HbA₂+HbE values obtained using the two methods. Method agreement was examined using Bland–Altman plots.

Optimal cutoffs for HbE and HbA₂+HbE for α⁰-thalassemia screening in HbE heterozygotes were determined using ROC curve analysis. The area under the curve (AUC) was calculated for each parameter. Using the established cutoffs, diagnostic performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and Youden’s index, were computed for CE and PR-HPLC. Diagnostic performance was validated by comparing sensitivity, specificity, PPV, NPV, and overall accuracy with those of molecular genotyping results.

Statistical analyses were conducted using SPSS Statistics v.26.0 (IBM Corp., Armonk, NY, USA). Statistical significance was set at a two-tailed P<0.05.

RESULTS

Hb separation using CE and PR-HPLC

CE yielded a distinct electropherogram characterized by major fractions of HbA and HbE of variable levels, with a minor HbA₂ fraction, for the 650 individuals with heterozygous HbE (Fig. 1A). HbA₂, HbE, and HbA₂+HbE value distributions are summarized in Table 1. PR-HPLC yielded a uniform chromatographic pattern with three major peaks corresponding to HbA, HbE, and HbA₂. HbE eluted at 5.389±0.14 mins (range: 5.042–5.658), while HbA₂ eluted at 5.586±0.13 mins (range: 5.263–5.939). The HbA₂ peak was partially resolved from HbE, with minor shoulder overlap in all samples (Fig. 1B). Nevertheless, HbE and HbA₂ quantification was feasible using integrated peak area analysis.

Fig. 1. Hb fraction profiles and distribution of HbA₂, HbE, and HbA₂+HbE in heterozygous HbE individuals. (A) Representative capillary electropherogram showing distinct separation of the HbE, HbA, and HbA₂ fractions. (B) Representative PR-HPLC chromatogram. (C) Scatter plots illustrating the distributions of HbA₂, HbE, and HbA₂+HbE levels obtained using CE and PR-HPLC in 650 samples overall, 108 samples with α⁰-thalassemia, and 542 samples without α⁰-thalassemia. Horizontal lines represent mean values, with upper and lower limits indicating±1 SD.

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC.

Table 1. Comparison of HbA₂, HbE, and HbA₂+HbE levels measured using CE and PR-HPLC in heterozygous HbE individuals with or without α⁰-thalassemia (training cohort).

Group	Parameter	CE				PR-HPLC			P *
		Mean±SD	Range	95% CI		Mean±SD	Range	95% CI
Overall (N=650)	HbA2	3.50±0.4	1.80–5.20	3.46–3.53		4.71±0.6	2.90–7.50	4.66–4.75	<0.0001
	HbE	21.77±4.4	6.80–28.50	21.43–22.11		22.06±4.5	7.50–29.20	21.71–22.40	<0.0001
	HbA2+HbE	25.27±4.4	9.30–31.70	24.93–25.60		26.77±4.6	10.60–33.90	26.41–27.12	<0.0001
With co-inherited α0-thalassemia (N=108)	HbA2	3.59±0.4	2.50–4.60	3.51–3.67		4.73±0.5	3.70–6.00	4.64–4.82	<0.0001
	HbE	14.21±2.3	10.00–17.90	13.77–14.65		14.65±2.2	10.30–19.00	14.22–15.08	<0.0001
	HbA2+HbE	17.80±2.5	12.50–22.10	17.32–18.27		19.38±2.5	14.70–24.00	18.90–19.85	<0.0001
Without co-inherited α0-thalassemia (N=542)	HbA2	3.47±0.4	1.80–5.20	3.44–3.52		4.71±0.6	2.90–7.50	4.65–4.76	<0.0001
	HbE	23.28±2.9	6.80–28.50	23.04–23.52		23.53±3.1	7.50–29.20	23.27–23.80	<0.0001
	HbA2+HbE	26.75±3.0	9.30–31.70	26.50–27.00		28.24±3.3	10.60–33.90	27.96–28.52	<0.0001

Data are presented as mean±SD, range, and 95% CI.

*P-values were calculated using the Wilcoxon signed-rank test to assess differences between CE and PR-HPLC.

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC; CI, confidence interval.

Genotypic classification of HbE heterozygotes in the training cohort

Among the 650 individuals with heterozygous HbE, 108 (16.6%) were identified as double heterozygotes for HbE and α⁰-thalassemia through molecular analysis. Of these, 107 (99.1%) harbored the SEA deletion and one (0.9%) harbored the CR deletion. The remaining 542 individuals (83.4%) did not have α⁰-thalassemia. Hematological differences between the groups with and without α⁰-thalassemia are summarized in Supplemental Data Table S2.

Correlation of HbE, HbA₂, and HbE+HbA₂ values between CE and PR-HPLC

We compared HbE, HbA₂, and HbA₂+HbE levels obtained through CE and PR-HPLC in the 650 training samples (Table 1). PR-HPLC yielded significantly higher mean HbA₂ (4.71%±0.6%), HbE (22.06%±4.5%), and HbA₂+HbE (26.77%±4.6%) levels than CE (3.50%±0.4%, 21.77%±4.4%, and 25.27%±4.4%, respectively, P<0.0001) (Fig. 1C). In the 108 HbE heterozygotes with α⁰-thalassemia, PR-HPLC consistently yielded higher values than CE (4.73±0.5% vs. 3.59%±0.4% for HbA₂, 14.65%±2.2% vs. 14.21%±2.3% for HbE, and 19.38±2.5% vs. 17.80%±2.5% for HbA₂+HbE, respectively, all P<0.0001) (Fig. 1C). This trend was comparable in those without α⁰-thalassemia (N=542) (4.71%±0.6% vs. 3.47%±0.4% for HbA₂, 23.53%±3.1% vs. 23.28%±2.9% for HbE, and 28.24%±3.3% vs. 26.75±3.0% for HbA₂+HbE, respectively, all P<0.0001) (Fig. 1C).

Bland–Altman analysis of agreement between the CE and PR-HPLC results revealed that the mean difference (PR-HPLC – CE) in HbA₂ values was 1.213, with a limit of agreement (LoA) ranging from 0.364 to 2.062 (Fig. 2A). For HbE, the mean difference was 0.286, with an LoA ranging from –2.172 to 2.749. For HbA₂+HbE, the mean difference was 1.502, with an LoA ranging from –1.234 to 4.237. Subgroup analysis revealed similar patterns in those with and without α⁰-thalassemia. In the α⁰-thalassemia group (Fig. 2B), the mean differences were 1.442, –0.182, and 1.259 for HbA₂, HbE, and HbA₂+HbE values, respectively. In the non-α⁰-thalassemia group, the mean differences were 1.168, 1.259, and 1.550 for HbA₂, HbE, and HbE+HbA₂ values, respectively (Fig. 2C).

Fig. 2. Bland–Altman comparison of HbA₂, HbE, and HbA₂+HbE levels measured using CE and PR-HPLC for (A) all heterozygous HbE samples (N=650), (B) HbE heterozygotes with co-inherited α⁰-thalassemia (N=108), and (C) HbE heterozygotes without α⁰-thalassemia (N=542). Blue dotted lines indicate the mean difference; red dotted lines denote the upper and lower limits of agreement.

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC.

Correlation between CE and PR-HPLC measurements

Pearson correlation analysis revealed the agreement between CE and PR-HPLC for the quantification of HbA₂, HbE, and HbA₂+HbE (Fig. 3A–3C). Significant correlations were observed across all sample groups (P<0.0001). For HbA₂, the overall correlation coefficient was r=0.686, with a linear regression equation of y (PR-HPLC)=0.9270x (CE)+1.468 (R²=0.471). The subgroup regression equations were y=0.9660x+1.557 (R²=0.688) for individuals with α⁰-thalassemia and y=0.9456x+1.359 (R²=0.471) for those without. For HbE, a strong correlation was observed (r=0.9597, R²=0.921), with a regression equation of y=0.9799x+0.7255. The subgroup equations were y= 0.9312x+1.353 (R²=0.9033) in the α⁰-thalassemia group and y=0.9903x+0.5934 (R²=0.9262) in the non-α⁰-thalassemia group. For HbA₂+HbE, the overall correlation coefficient was r=0.9524, with a regression equation of y=0.9888x+1.785 (R²=0.9071). The subgroup equations were y=0.9576x+2.349 (R²=0.8932) in the α⁰-thalassemia group and y=0.9948x+ 1.681 (R²=0.9098) in the non-α⁰-thalassemia group.

Fig. 3. Pearson correlation results for HbA₂, HbE, and HbA₂+HbE levels obtained using CE and PR-HPLC for (A) all heterozygous HbE samples (N=650), (B) HbE heterozygotes with co-inherited α⁰-thalassemia (N=108), and (C) HbE heterozygotes without α⁰-thalassemia (N=542). Regression lines, 95% CIs, and equations are shown for each comparison.

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC; CI, confidence interval.

Comparison of Hb fractions between HbE heterozygotes with and without α⁰-thalassemia

We evaluated the effect of α⁰-thalassemia co-inheritance on Hb fractions by quantifying HbA₂, HbE, and HbA₂+HbE levels using CE and PR-HPLC (Table 1). For HbA₂, individuals with α⁰-thalassemia showed slightly higher mean values than those without. With PR-HPLC, the mean HbA₂ level was 4.73%±0.5% in the α⁰-thalassemia group vs. 4.71%±0.6% in the non α⁰-thalassemia group (P=0.0812). In contrast, CE detected a small but significant difference (3.59%±0.4% vs. 3.47%±0.4%, P=0.0047) (Fig. 4A). HbE levels were markedly lower in individuals with co-inherited α⁰-thalassemia than in those without. PR-HPLC measured a mean HbE of 14.65%±2.2% in the α⁰-thalassemia group vs. 23.53%±3.1% in the non-α⁰-thalassemia group (P<0.0001). A similar difference was observed with CE, confirming consistency between the methods (Fig. 4A). Moreover, the HbA₂+HbE fraction showed an evident separation between the two groups, with mean values of 19.38%±2.5% in those with α⁰-thalassemia and 28.24%±3.3% in those without (P<0.0001).

Fig. 4. Quantitative comparison and diagnostic performance of HbA₂, HbE, and HbE+HbA₂ levels determined using PR-HPLC and CE. (A) Bar graphs showing HbA₂, HbE, and HbE+HbA₂ levels in individuals with and without α⁰-thalassemia, measured using PR-HPLC and CE. Data are presented for the overall samples and stratified by α⁰-thalassemia status. *P<0.05 vs. HbE heterozygotes without α⁰-thalassemia (Mann–Whitney U test). (B) ROC curves indicating the diagnostic performance of isolated HbE and HbE+HbA₂ values in identifying co-inherited α⁰-thalassemia in HbE heterozygotes.

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC.

Optimization of diagnostic HbE and HbA₂+HbE cutoffs for PR-HPLC

As HbA₂ alone showed limited discriminatory power, optimal diagnostic cutoffs for PR-HPLC were established for HbE and HbA₂+HbE using ROC curve analysis. ROC curves are shown in Fig. 4B, and related metrics are detailed in Table 2. For HbE quantification, CE and PR-HPLC yielded comparable AUCs of 0.978 and 0.975, respectively. The optimal PR-HPLC cutoff (<19.0%) demonstrated 100% sensitivity and 91.9% specificity. For HbA₂+HbE values, CE and PR-HPLC yielded AUCs of 0.974 and 0.970, respectively, indicating robust diagnostic performance. The optimal PR-HPLC cutoff (<24.0%) achieved 100% sensitivity and 89.7% specificity. In contrast, application of the conventional <25% cutoff (based on existing CE-based thresholds) yielded 100% sensitivity but lower specificity: 85.6% and 82.5% for PR-HPLC and CE, respectively (Table 2).

Table 2. Diagnostic performance of different HbE and HbA₂+HbE cutoff values in detecting α⁰-thalassemia among HbE heterozygotes (training cohort).

Method	Cutoff (%)	Sensitivity (%)	Specificity (%)	FP (%)	FN (%)	Youden’s index (%)
CE	HbE (%) AUC=0.978; 95% CI=0.968–0.988
	HbA2+HbE (%) AUC=0.974; 95% CI=0.963–0.986
	<25.0*	100	82.5	17.53	0	82.5
PR- HPLC	HbE (%) AUC=0.975; 95% CI=0.964–0.986
	<18.5	99.1	92.6	7.4	0.9	91.7
	<19.0	100	91.9	8.1	0	91.9
	<19.5	100	90.2	9.8	0	90.2
	<20.0	100	88.2	11.8	0	88.2
	HbA2+HbE (%) AUC=0.970; 95% CI=0.958–0.982
	<23.5	98.1	91.5	8.5	1.9	89.6
	<24.0	100	89.7	10.3	0	89.7
	<24.5	100	87.1	12.9	0	87.1
	<25.0	100	85.6	14.4	0	85.6

Values represent diagnostic performance metrics calculated through ROC analysis in the training cohort.

*Original cutoff based on conventional HPLC and CE [16, 17].

Abbreviations: CE, capillary electrophoresis; PR-HPLC, Premier Resolution HPLC; FP, false-positive rate; FN, false-negative rate; AUC, area under the ROC curve; CI, confidence interval.

Performance evaluation of the selected optimal cutoffs for HbE heterozygotes with co-inherited α⁰-thalassemia

The cutoffs established in the training cohort were applied to a prospective cohort of 5,082 samples to validate the diagnostic utility of PR-HPLC in detecting α⁰-thalassemia in HbE heterozygotes. Based on characteristic Hb fractions with retention times corresponding to HbA, HbE, and HbA₂, 884 individuals (17.39%) were identified as HbE heterozygotes. As in the training cohort, the HbE peak eluted consistently at 5.11–5.67 mins (5.35± 0.14 mins). Distributions for HbA₂, HbE, and HbA₂+HbE were 2.4%–6.9% (4.76%±0.72%), 8.3%–29.0% (23.08%±4.04%), and 12.8%–34.4% (27.84%±4.18%), respectively. Using the HbE cutoff of <19.0%, 118 samples were flagged as suspected co-inherited α⁰-thalassemia, whereas 766 samples with HbE ≥19.0% were classified as non α⁰-thalassemia. The same 117 samples were identified when applying the HbA₂+HbE cutoff of <24.0%.

Molecular analysis helped confirm the diagnosis of 118 individuals (13.35%) as having α⁰-thalassemia with SEA deletion. Of these, 117 were correctly classified based on the combined cutoff (12.8%–23.7%), whereas one sample slightly exceeded the threshold (24.1%). Notably, all 118 individuals were correctly identified based on the HbE cutoff (8.6%–18.4%). Diagnostic performance metrics are summarized in Supplemental Data Table S3. With HbE <19.0%, PR-HPLC achieved 100% sensitivity, 93.6% specificity, 70.7% PPV, 100% NPV, and 94.5% overall accuracy. Using the HbA₂+HbE cutoff of <24.0%, sensitivity was 99.2%, specificity 93.73%, PPV 70.91%, NPV 99.9%, and accuracy 94.46%. Subgroup analysis showed consistent results: in routine samples (N=430), the combined cutoff provided 100% sensitivity and 88.42% specificity, whereas in health check-up samples (N=454), specificity improved to 98.30%, with a modest reduction in sensitivity (97.62%).

Development and validation of gap-PCR for rare α⁰-thalassemia genotyping

Gap-PCR successfully generated distinct fragments of 1,129 bp and 335 bp for SA and CR deletions, respectively (Supplemental Data Fig. S1A and S1B). Normal alleles were consistently amplified as internal controls, ensuring sample integrity and assay reliability. No cross-reactivity was observed between the deletion types. Screening of the HbE heterozygotes in the training (N=650) and validation (N=884) cohorts identified 225 individuals as SEA-deletion carriers, one individual with the CR deletion, and no SA deletions.

DISCUSSION

The PR-HPLC system enables precise and reproducible quantification of HbE and HbA₂ in HbE heterozygotes, effectively overcoming the co-elution limitations of conventional HPLC. PR-HPLC demonstrated high analytical sensitivity, detecting HbE fractions even in samples with very low HbE levels (<10% of total Hb) as derived using CE. In phase I, all HbE heterozygotes identified via CE were consistently detected using PR-HPLC, underscoring its qualitative concordance with CE. Notably, chromatographic profiles of individuals from both (training and validation) cohorts showed partially overlapping HbE and HbA₂ peaks, whereas CE achieved complete separation. Despite this overlap, reliable quantification of HbE and HbA₂ was possible, facilitating direct method comparison.

To our knowledge, this is the first study to directly compare HbE and HbA₂ quantification in HbE heterozygotes with and without co-inherited α⁰-thalassemia using both CE and HPLC. While HbA₂ and HbE levels measured with PR-HPLC were significantly higher than those obtained via CE, Bland–Altman analysis demonstrated good agreement between the methods, particularly for HbE. Pearson correlation results supported these findings, revealing strong correlations for HbE (r>0.95) across all subgroups, whereas HbA₂ correlations were moderate (r=0.69–0.82). In line with the consistently higher HbA₂+HbE values obtained using PR-HPLC than using CE, Bio-Rad HPLC has been shown to yield higher HbE measurements (30.1%) than Capillarys 2 (23.9%) in HbE carriers [24].

Separate quantification of HbE and HbA₂ using PR-HPLC and CE enabled a precise assessment of the effect of α⁰-thalassemia co-inheritance on Hb expression. In HbE heterozygotes, α⁰-thalassemia co-inheritance was consistently associated with substantially reduced HbE levels, whereas HbA₂ levels were unaffected, suggesting that α⁰-thalassemia selectively suppresses HbE (α₂β^E₂) synthesis without altering HbA₂ (α₂δ₂) expression and corroborating HbE quantification as a surrogate marker for α⁰-thalassemia screening in HbE carriers. To our knowledge, this study is the first to demonstrate this selective HbE suppression through separate HbE measurement. This effect has not been directly addressed previously, as conventional HPLC cannot resolve HbE from HbA₂, and CE-based studies have typically relied on combined values [10, 20, 21, 24, 25]. Notably, in line with our observations, several studies have reported reduced HbA₂+ HbE levels in HbE heterozygotes with co-inherited α⁰-thalassemia [9, 10, 16, 18, 26].

Based on the reduction in HbE and HbA₂+HbE levels in HbE heterozygotes with co-inherited α⁰-thalassemia, we focused on two parameters for optimal cutoff determination. Accurate screening of HbE heterozygotes with co-inherited α⁰-thalassemia is critical to avoid misclassification, which may result in severe clinical outcomes. Based on thresholds of HbE <19.0% and HbA₂+HbE <24.0%, PR-HPLC successfully identified all confirmed cases. Although both parameters demonstrated equal sensitivity in identifying double heterozygous HbE and α⁰-thalassemia, HbA₂+HbE showed a higher false-positive rate and lower specificity than HbE alone (89.7% vs. 91.1%). Specificity declined further to 85.6% when applying the original cutoff of HbA₂+HbE <25.0% [17, 18], corresponding to a false-positive rate of 14.4%. When the same cutoff was applied to CE-derived HbA₂+HbE, specificity decreased to 82.5%. These results indicate that isolated HbE quantification using PR-HPLC provides superior discriminatory precision, minimizes misclassification of simple HbE heterozygotes, and improves overall α⁰-thalassemia screening accuracy.

Validation of the proposed cutoffs in an 884-sample prospective cohort confirmed their diagnostic utility. The use of the HbE cutoff of <19.0% resulted in accurate identification of all cases of heterozygous HbE with co-inherited α⁰-thalassemia. The HbA₂+HbE cutoff of <24.0% showed slightly lower sensitivity because of a single false-negative case. This case has a combined value marginally exceeding the threshold (24.1%) because of a disproportionately elevated HbA₂ level (5.7%) despite a low HbE level (18.4%). This misclassification highlights a key limitation of using combined values, as elevated HbA₂ can mask suppressed HbE expression. Considerable variation in CE-based HbA₂ levels has been previously noted among HbE heterozygotes [10, 20, 27], supporting our findings. Collectively, these observations reinforce the value of isolated HbE quantification in improving diagnostic precision. While combined and isolated cutoffs yielded comparable specificity (93.73% vs. 93.60%), the isolated HbE cutoff demonstrated superior sensitivity, underscoring its utility as a reliable screening parameter.

Both cutoffs performed better in the health check-up cohort than with routine samples, which can be attributed to the higher prevalence of microcytic hypochromic anemia in routine samples, including samples from individuals with anemia, pregnant women, and couples undergoing hemoglobinopathy screening. Common contributing factors in these populations, such as iron deficiency and homozygous α⁺-thalassemia [28, 29], may suppress HbE expression [30], resulting in false positives. While these confounding variables were not systematically captured for all participants, some residual confounding may have remained and potentially affected the false-positive rate. The prevalence of HbE with co-inherited α⁰-thalassemia in our cohorts ranged from 13.4% to 16.6%, consistent with previous reports [9, 16]. In such high-prevalence settings, the relatively high PPV of the HbE cutoff substantially reduces the burden and cost of confirmatory DNA testing, underscoring the importance of maintaining high specificity in screening protocols to ensure efficient resource utilization and accurate carrier identification. To assess clinical utility across diverse populations, we calculated predictive values using the optimal HbE cutoff (<19.0%) at different α⁰-thalassemia prevalence rates (Supplemental Data Table S4). In our cohort with 13.4% prevalence, the PPV was 70.66%. PPV varied substantially across epidemiological settings: 45.14% at 5% prevalence, 63.46% at 10% prevalence, and 79.63% at 20% prevalence. NPV remained excellent (100%) across all scenarios because of the very high sensitivity of our cutoff. Hence, PPV varies substantially with prevalence, highlighting the importance of considering local epidemiology when implementing screening protocols.

This study had some limitations. First, our gap-PCR panel was designed to detect the most prevalent α⁰-thalassemia deletions (SEA, THAI, SA, and CR) in the Thai population. However, rare α⁰ variants or α⁺-thalassemia mutations may have been overlooked, possibly underestimating the actual burden of α-globin disorders. Second, we exclusively studied Southeast Asians, a population with high HbE and α⁰-thalassemia prevalence; therefore, the applicability of our HbE cutoff to other ethnicities or genetic backgrounds needs to be validated. Third, while PR-HPLC demonstrated excellent performance, HPLC platform-specific optimization may be required. Finally, our findings are based on adult Southeast Asian populations with specific epidemiological characteristics; hence, extrapolation to pediatric populations, other geographic regions, or ethnic groups with different α-globin and mutation spectra of HbE necessitates further validation.

In conclusion, PR-HPLC enables accurate quantification of HbE, establishing HbE as a novel and effective screening marker for α⁰-thalassemia in HbE heterozygotes. A HbE cutoff of <19.0% eliminates the need for HbA₂ quantification, simplifying interpretation and minimizing misclassification due to elevated HbA₂. This approach potentially improves diagnostic precision, reduces unnecessary molecular testing, and decreases laboratory workload and screening costs in thalassemia-endemic areas.

SUPPLEMENTARY MATERIALS

Supplementary materials can be found via https://doi.org/10.3343/alm.2025.0552.