Evaluation of interference from 16 hemoglobin variants on hemoglobin A1c measurement by five methods

Abstract Hb variants prevalent in China are different from those in other countries. We aimed to assess the interference from Hb variants found in China on HbA1c measurement. All Hb variants were confirmed using Sanger sequencing. HbA1c was measured using a capillary electrophoresis method (Capillarys 3 OCTA), two cation-exchange high-performance liquid chromatography methods (ADAMS HA-8180V and HLC-723 G8 standard mode), an immunoassay (Cobas c501), and a boronate affinity chromatography method (Premier Hb9210). Premier Hb9210 was used as a comparative method. A total of 16 species of Hb variants were identified in 102 variant carriers. The most common variant was Hb E, followed by Hb Q-Thailand, Hb New York and Hb J-Bangkok. Clinically significant interference was observed for the Capillarys 3 OCTA (two Hb variants), ADAMS HA-8180V (seven Hb variants), HLC-723 G8 (14 Hb variants), and Cobas c501 (two Hb variants). The proportion of unacceptable HbA1c results was 13.7% for Capillarys 3 OCTA, 52.9% for HA-8180V, 83.3% for HLC-723 G8, and 3.9% for Cobas c501. Hb variants in China severely affect the accuracy of some commonly used HbA1c methods.


Introduction
Hemoglobin A1c (HbA 1c ) is widely accepted as an important biomarker in the management of diabetes. HbA 1c provides valuable information for long-term glycemic control and predicts the risk of chronic complications in patients with diabetes, and it has been recommended as a diagnostic test for diabetes for more than 10 years [1 -3]. However, its measurement is challenging in the presence of Hb variants, which remain a common cause of erroneous HbA 1c results. The interference from Hb variants on HbA 1c measurements depends on testing methods and the type of Hb variant [4,5]. Although over 1000 Hb variants have been described to date [6,7], most studies on the effects of Hb variants on HbA 1c assays have been limited to the four most common Hb variants worldwide, i.e. HbS, HbC, HbD, and HbE [8,9].
Data from the National Center For Clinical Laboratories (NCCLS; https://www.nccl.org.cn/) in China showed that 3227 laboratories participated in the HbA 1c external quality assessment in 2021. Of them, 2367 used cation-exchange high-performance liquid chromatography (HPLC) methods, 414 used immunoassay, 233 used boronate affinity method, and 23 used capillary electrophoresis (CE) method. The two most common HPLC-based analyzers were HLC-723 G8 standard mode and HA-8180V. Although these two methods were not intended to analyze samples with Hb variants, they are commonly used in southern China, where Hb variants are highly prevalent [10,11].
In this study, we evaluated the effects of 16 Hb variants found in China on a CE method (Capillarys 3 OCTA), two cation-exchange HPLC methods (HA-8180V and HLC-723 G8), and an immunoassay (Cobas c501), by comparing the results obtained with a boronate affinity method (Premier Hb9210).

Patients and samples
This study was approved by the Ethics Committee of Shenzhen Baoan District Songgang People's Hospital (No. SG202105), and informed consent was obtained from all participants. Between June 2020 and May 2021, among individuals who came to our hospital for the screening of hemoglobinopathies, 102 patients with suspected Hb variant were found during hemoglobin electrophoresis and their residual samples were collected. In addition, 30 whole blood samples without Hb variants (HbAA) were used as controls. All samples were collected in tubes containing ethylene diamine tetra acetic acid and stored rapidly at 4 C. Then, they were separated into aliquots and transferred to À80 C on the day.

Identification of Hb variants
Sanger sequencing was performed to determine the presence of Hb variants. Genomic DNA was extracted from the blood sample from the patient using a Genomic DNA Isolation Kit (Invitrogen, Carlsbad, CA). Then, HBA1, HBA2, and HBB genes were amplified using an S1000 Thermal Cycler PCR system (Bio-Rad, Foster, CA). The PCR products were sequenced using an ABI PRISM 3730 XL Sequencer (Applied Biosystems, Foster City, CA).

Data analyses
To eliminate any inherent calibration bias, variant sample results for each method were adjusted using regression of each method's results for non-variant samples against Premier Hb9210 results [4]. Variant results were considered acceptable or without clinically significant interference if the result was within the 99% prediction interval of the regression line. All data analyses and graphing were performed by using SPSS 16.0 (Chicago, IL).

Identification of Hb variants
Supplementary Figure Table 1 shows the frequency of Hb variants.
Warning flag and unacceptable HbA 1c result Table 2 shows the proportion of unacceptable results exceeding the 99% prediction interval of the regression line. Clinically significant interference Figure 1 shows the 99% prediction interval of the regression line for non-variant sample results. Table 3 shows the mean relative biases (NGSP units) between each evaluated method versus the comparison method for each Hb variant. The Capillarys 3 OCTA showed clinically significant interference from Hb J-Bangkok, Hb New York and Hb Pyrgos. The HA-8180V and HLC-723 G8 showed clinically significant interference from seven and 13 Hb variants, respectively. The Cobas c501 showed unacceptable results for 3 of the 11 samples containing Hb New York. Furthermore, all four methods showed a considerable positive bias for Hb New York, when compared with Premier Hb9210. If using HbA 1c ! 6.5% as the criterium of diabetes, as shown in Table 4, a significant number of individuals would be missed depending on the method used.

Discussion
Accurate reporting of HbA 1c results is critical for the precise diagnosis and effective management of diabetes. Many methods based on different analytical principles have been developed for HbA 1c assays, such as cation-exchange HPLC, CE, boronate affinity chromatography, and immunoassays [13,14]. Although the performance of these commercial HbA 1c analyzers has improved considerably, some Hb variants still adversely impact the accuracy of some HbA 1c methods [15,16]. Among the popular techniques, HPLC and CE can detect some of the Hb variants, but are susceptible to their interference. In contrast, boronate affinity chromatography and immunoassay will not show signs of Hb variants, as they both give only HbA 1c values. Our results demonstrated that Hb variants commonly found in China were different from those in other countries, with the most common Hb variant being Hb E, followed by  Hb Q-Thailand, Hb New York, Hb J-Bangkok, Hb G-Coushatta, Hb G-Honolulu, Hb G-Taipei, and so on, which matched the previous findings [17,18]. In the present study, the Capillarys 3 OCTA detected and flagged all tested Hb variants during HbA 1c measurements and showed acceptable results for most cases, and this could prevent the reporting of unreliable results. In contrast, the HA-8180V flagged only nine Hb variants, and the HLC-723 G8 standard mode did not flag any Hb variants, and a significant proportion of the reported results exceeded the 99% prediction interval, which means that there is a risk of reporting inaccurate HbA 1c results. Similar results were found in a study on Hb variants in Korea, where the most common Hb variants are Hb G-Coushatta and Hb Queens [15]. It should be noted that different versions of the same HbA 1c analyzer may lead to different HbA 1c results [19]. When using the G8 standard mode for HbA 1c measurement, laboratories must be cautious about reporting results in the presence of suspected variants. Clinically significant interference of Hb variants with HbA 1c measurements may lead to inappropriate treatment or affect the diagnosis of diabetes if clinical management of patients relies solely on HbA 1c results. In this study, Hb J-Bangkok, Hb New York and Hb Pyrgos produced clinically significant interference for the Capillarys 3 OCTA. The HLC-723 G8 standard mode and HA-8180V showed clinically significant interference from seven and 13 Hb variants, respectively. A previous study reported that several common Hb variants in southern China interfered with HbA 1c measurements by HLC-723 G8 variant mode [20]. Different from our study, they used the CE method (Capillarys 2 Flex Piercing) as a comparative method. However, the use of the CE method as a comparative method has not been validated, which could lead to potential errors in the evaluation. In addition, the study only determined statistically significant interference, not clinically significant interference [20]. For some variants with small numbers, we cannot accurately determine whether it produces clinically significant interference due to the presence of random error, especially for the variants with only one in this study, which requires a larger sample size for systematic evaluation.
In conclusion, Hb variants in China severely interfere with some popular HbA 1c methods. It is important to assess the interference from locally prevalent Hb variants on HbA 1c assays. Laboratories should be familiar with the limitations of the various methods concerning interference from Hb variants and should use a method (or version) that can flag these variants or is free from their interference.