Hb variants are a group of prevailing inherited genetic defects caused by point mutations in the globin gene, eventually resulting in amino acid substitution (24). Previous studies have demonstrated that Hb variations can result in HbA1c values that do not correspond to blood glucose levels in the same patient, while the degree of interference is dependent on the method used and the specificity of the variation (15-17,25). Therefore, for each variant, this interference can be divided into method-specific, where some, but not all HbA1c determination methods are affected and variant-specific, where HbA1c levels are affected by an altered erythrocyte lifespan and glycation rate.

Method-specific interference of Hb variant. The charge and mass of Hb can change when an amino acid at a particular location on the globin peptide chain is altered. Therefore, detection methods based on the physical properties of Hb, such as IEC and CE, are vulnerable to interference, since variants interfere with the elution of the peak of interest, thus resulting in false to glycated Hb values and even invalid HbA1c measurements. For the IEC method, the %A1c is measured by excluding the A1c peak area, divided by the area sum of A1a, A1b, LA1c, A1c, A0 and HbX (both HbX1c and unmodified HbX) from the calculation. The effect of Hb variants on IEC is mainly due to the inability of IEC to completely separate the HbA from HbA1c variants. Therefore, when the peak of HbA1c cannot be completely separated from that of HbX, a proportion of HbX value is calculated into that of HbA1c, thus resulting in falsely elevated HbA1c values (Fig. 2C). Conversely, when HbX is calculated into the denominator of the formula, a spuriously reduced HbA1c value is obtained (Fig. 2B). However, the IEC-generated interference can vary in different systems and is mainly dependent on the position of the elution peaks of HbX, HbA, HbX1c and HbA1c, and whether or not a clear separation occurs. IEC, the most commonly used method for HbA1c detection, is also susceptible to interference from Hb variants. From another perspective, IEC exhibits more advantages compared with other methods, since it is easier to identify abnormal Hb by examining the chromatogram. Therefore, it is of utmost importance for the laboratory to routinely inspect the chromatogram prior to issuing the report. In terms of CE, the reported HbA1c is derived from the ratio (A1c)/(A0+A1c), where A1c indicates HbA glycated at the N-terminal valine of the β chain and A0 the not glycated one. Consistent with the IEC method, any mutations that can cause changes in charge and mass, can potentially co-migrate with A1c or A0, thus resulting in erroneous results or even invalid HbA1c measurements. However, CE runs considerably longer and is more capable of distinguishing between HbX and HbA, as well as between HbX1c and HbA1c, compared with IEC, when dealing with the majority of variants. Yun et al (16) suggested that the changes in the glycosylation rate can be estimated by comparing the results of the CE method with those obtained using immunoassay or BAC. Since immunoassays rely on antibodies that specifically recognize and bind to the first 4-10 amino acids of the N-terminal of the β chain, the use of this technique can be limited when bases at this site are mutated. However, the main disadvantage of the immunoassay is its inability to distinguish between HbA and HbX (26). By using an immunoassay, %HbA1c is calculated by dividing A1c + X1c by total HbA + HbX. Therefore, when HbA is absent or erythrocyte biology is altered, using an immunoassay may lead to errors in clinically reported HbA1 levels. This issue is discussed in further detail below. Ideally, repeated analyses using methods based on different analytical principles need to be performed, since the effect of a particular Hb variation on HbA1c readings could be associated with the method sensitivity. The benefits and challenges of each method are summarized in Table I.

Variant-specific erythrocyte lifespan changes. Interpretation of HbA1c values in terms of glycemic control is affected by a combination of factors. For example, a HbA1c value of 7%, generally corresponds to an average plasma glucose value of 154 mg/dl in the majority of individuals (27). However, for individuals with a shorter RBC lifespan or higher glycation rates, a HbA1c value of 7% may be associated with different average glucose levels. A normal RBC has a lifespan of ~100-115 days on average, although the range is much wider, ~70-140 days (28). It has been reported that the levels of A1c are most commonly affected by the levels of blood glucose over the past 30 days, accounting for ~50% of the total A1c levels. However, only 10% of A1c levels are influenced by blood glucose levels from the past 90-120 days (29). When the glycation phase of RBCs changes, and more specifically their lifespan, HbA1c can no longer accurately represent glycemic control. HbS (β6Glu>Val) and HbC (β6 Glu>Lys) are the most common Hb variants (30,31). It has been demonstrated that ~75% of patients with sickle cell disease are found in Sub-Saharan Africa (32). Additionally, significant prevalence rates have been also recorded in the Middle East, India and the Mediterranean area (33). In West Africa, the prevalence of HbC has reached 40-50%, while that in Benin, the United States and North Africa is estimated to 20, 3 and 1-10%, respectively (31). Since heterozygous forms of both variants do not cause hemolytic disease, they cannot, therefore, affect glycosylation. Sickle cell disease, and more particularly its homozygous clinically severe condition (HbSS), where the lifetime of RBCs is decreased to <20 days, is accompanied by severe hemolysis (12). Therefore, HbA1c values in those patients need to be interpreted with caution, taking into consideration factors, such as anemia, an enhanced RBC turnover, increased blood transfusion needs and increased HbF levels, which may all have a negative impact on HbA1c as a long-term glycemic control indicator. Additionally, the heterozygous type of the disease, HbSC, exhibits the same confounding issues as HbSS when it comes to determining HbA1c values. However, HbSC causes less severe anemia compared with sickle cell disease. In a large cohort study published by Lacy et al (34) in 2017, African-Americans with sickle trait had a decrease of 0.3% in HbA1c values compared to those without the sickle trait. In addition, patients with the sickle trait had a decrease of 0.29% in HbA1c levels at the same fasting blood glucose levels.

Variant-specific glycation rate alterations. The biological question is whether the glycation rates of HbA and Hb variations are equivalent. When they are not equal, the results of the method used to measure total glycated Hb (affinity chromatography) may be biased. Therefore, the interpretation of glycemic control based on the glycated Hb levels results may be incorrect. Although none of the first codon variants in the β-globin gene can cause any major clinical condition, such mutations are of interest due to their potential interference with co-translational modifications, such as acetylation at the same site during β-globin synthesis. The type of N-terminal amino acid can determine the degree of acetylation. Therefore, valine can substantially inhibit this process, resulting in the slight acetylation of α- and β-globin. However, it has been reported that the N-terminal glycine of γ-globin is less inhibitory, thus leading to ~15% acetylation (35). A previous study demonstrated that in Hb Raleigh (β1Val>Ala), the N-terminal amino acid of its β chain could be substituted to produce acetylated alanine, thus producing a large amount of acetylated Hb, which could not be glycosylated normally (36). It was hypothesized that the glycation process occurred close to the N-terminal valine and the 140th amino acid in the β chain (37). Mutants located near valine at the N-terminus of the β chain could cause changes in the glycation rate, such as reduced Hb Görwihl (β5Pro→Ala) levels, thus suggesting attenuated glycation reaction (38). Hb Himeji (β140Ala→Asp), a rare variant occurring on the β chain, is characterized by an enhanced glycation (37). In heterozygous carriers, HbA1c values determined by immunoassay or BAC have been found to be significantly higher compared with those determined using cation exchange chromatography (37). Mutations in this region can either increase or decrease glycosylation. For example, a previous study demonstrated that individuals with Hb Sagami (β139Asn→Lys) exhibited low HbA1c levels, as assessed using immunoassay, thus indicating a reduction in the glycation response (39). Glycosylation rates can even have an effect on HbA1c measurements of more common variants, such as those associated with sickle traits. Therefore, the glycosylation rate of β^S appears to be higher than that of β^A. However, Kabytaev et al (40) concluded that the clinical interpretation was relatively unaffected by the tiny net difference between HbAS (sickle phenotype) and uncharacterized total glycosylation. A summary of the altered glycosylation rates presented in mutants located in the first 10 amino acids of the β chain and at amino acid positions 139-140 is presented in Table II (41-44).

Elevated HbF can cause related problems. Certain thalassemias and genetic persistence of fetal Hb (HPFH) can lead to elevated HbF. Increased HbF levels can be also caused by hydroxyurea, a medication commonly used to treat sickle cell disease and several hematological malignancies (45). In the IEC measurement method, the HbF and A1c peaks are eluted adjacently. Therefore, high HbF may overlap and distort the shape of the A1c peak, thus resulting in falsely high values (Fig. 2D). Elevated HbF can also cause other effects in immunoassay and BAC (10,18). This interference could be caused by the slower rate of glycation compared with that of HbA. HbF lacks the β chain and encompasses glycine instead of valine at the N-terminus of the γ chain. HbF can only be glycosylated on the lysine residue at the N-terminus of α chain. A previous study showed that the rate of glycosylation at the N-terminus of the α chain was indeed 8-10 times lower compared with that of the β chain's N-terminus (6). Therefore, for the boronate affinity assay, the glycation fraction in individuals with elevated HbF levels would be lower compared with those without increased HbF value due to the lower degree of glycation of HbF. During immunoassay, HbA1c antibodies cannot bind with the glycated portion of HbF and, therefore, total Hb measurements also include HbF value, eventually leading to a significant reduction in the measured HbA1c. From a clinical perspective, in patients with HbF levels of <20%, the use of immunoassays or BAC could reduce HbA1c levels by 1-2% (18). The Diabetes Control and Complications Trial clearly showed that a 1% reduction in HbA1c levels was equivalent to a reduction in the risk of diabetes complications by approximately 30% (2). A spurious reduction of 1-2% in HbA1c value could result to undertreatment of hyperglycemia, which in turn could lead to a significantly increased risk of complications.

When separation techniques based on charge differences are utilized, HbA1c measurements may also be affected by the chemical changes of Hb, which may physically and chemically imitate HbA1c, thus leading to an incorrect assessment of HbA1c. Carbamylated Hb is more common in uremic patients and is the most common derivative (46). This is due to the decomposition of urea nitrogen into ammonia and cyanate in the body. In turn, cyanate is protonated to form isocyanic acid, which combines with the α and ε amino groups of proteins to form carbamoyl moieties (46,47). Valine at the N-terminus of the Hb β chain reacts specifically with isocyanic acid to form stable carbamyl-Hb (CHb). The isoelectric points of CHb and HbA1c are similar. Therefore, the peak times of both CHb and HbA1c are close in the IEC system based on the detection principle of Hb species with different charges. When CHb reaches a certain concentration, the peak time of LA1c/CHb is delayed or the peak shape increases, thus resulting in the overlap of the LA1c/CHb peak with that of HbA1c. The overlap increases with the enhanced CHb concentration (Fig. 2D). If the peak is large, depending on the system, it can lead to biased results in the HbA1c levels (either increase or decrease) (19,48). In vitro, the carbamylation of Hb at concentrations up to 5.4% can result in erroneous readings in glycated Hb levels, when different cation-exchange techniques are used (19). However, studies evaluating the in vivo effects of carbamoyl Hb have revealed several differences, ranging from insignificant to significant (49-51). The studies by Little et al (50) and Dolscheid-Pommerich et al (51) demonstrated that the effects of CHb were statistically, yet not clinically significant. However, the assessment of HbA1c in this population should be always performed using the same measuring technique to ensure longitudinal comparability and produce comparable readings. Furthermore, the association between chronic kidney disease (CKD) and A1c is complex. Therefore, previous studies have demonstrated that patients with CKD exhibit lower levels of erythropoietin, possible increased glycation and enhanced carbamylated Hb levels, while dialysis in such patients may shorten the RBC lifespan and decrease A1c levels (50-53). The study by Little et al (50), comparing the levels of glycated albumin (GA) with HbA1c in patients with chronic renal failure, demonstrated that the levels of HbA1c in such patients were decreased by ~1.5% compared with those of GA. Additionally, the results of a clinical trial revealed that the treatment of 15 individuals with type 2 diabetes mellitus (T2DM) and CKD (3B/4) with erythropoietin resulted in a clinically meaningful decrease of ~0.7% in HbA1C readings (52). Therefore, HbA1c testing should be interpreted cautiously in patients with renal failure. Long-term aspirin use can also induce the acetylation of Hb, thus resulting in falsely elevated levels of HbA1c, due to its interference with some of the assays used (54,55). More specifically, aspirin promotes the acetylation of Hb to change its surface charge. In turn, this acetylated product can co-migrate with A1c, eventually leading to a falsely elevated HbA1c value (56). Although the exposure of normal Hb to aspirin in vitro results in falsely high acetylated Hb values, as verified using different cation-exchange methods, the results of in vivo studies have contradicted this finding (56,57). Camargo et al (56) demonstrated that treatment with lower doses of aspirin (200 mg/day) for 4 months did not result in a clinically relevant increase in HbA1c levels. In clinical practice, the effect of aspirin on HbA1c levels could not be palpable until long-term and high-dose therapy (55). However, it has been reported that the BAC method is not affected by carbamylated and acetylated Hb (50,57).

High doses of vitamins C and E have also been shown to be associated with decreased A1c levels mediated by the inhibition of Hb glycosylation. Vitamin C can form ionic bonds with several biomolecules. In turn, ionic interactions of ascorbate and its free radical with proteins can affect the reactivity of molecular complexes via altering the local redox potentials and charge transfer reactions. It has been suggested that the potential for non-enzymatic glycosylation is reduced when vitamin C reacts directly with the glucose-binding site (lysine residue) (58). A previous study revealed that vitamin E supplements could prevent the glycosylation of Hb by blocking glycation in the early stages of the Maillard reaction or by partially preventing the development of advanced glycosylation end-products (59). However, the inhibition of Hb glycosylation by vitamins C and E is clinically controversial. An in vitro study demonstrated that vitamins C and E could prevent the development of protein glycosylation (60). However, the outcomes of in vivo experiments are contradictory with the aforementioned finding. While Ceriello et al (61) demonstrated the inverse association between vitamin E consumption and glycated Hb levels, a later meta-analysis (59) was unable to reveal a discernible difference. However, further subgroup analysis revealed that following vitamin E supplementation, HbA1c was noticeably reduced in patients with T2DM in the group with low vitamin E levels. However, the small datasets used in this subgroup analysis, suggested that further research should be conducted to support this conclusion (59). Likewise, conflicting information on vitamin C intake and glycated Hb levels can be found in the literature (56,62,63). Additionally, other drugs, such as dapsone, sulfasalazine, antiretrovirals and ribavirin can increase the hemolysis rate, thus reducing glycated Hb levels (64-68). In a retrospective review of 49 individuals with T2DM and Hansen's illness, 35 patients (71%) had HbA1c readings lower than the mean blood glucose levels (65). At the same fasting blood glucose concentration, the HbA1c discordant group had a significantly lower hemoglobin A1c value (mean HbA1c, 4.4±1.8%) compared with the HbA1c consistent group (mean HbA1c, 7.9±2.1%). During the first 3 months of dapsone therapy, the HbA1c levels decreased considerably (65).

Particular pathologies can alter the lifespan of RBCs, thus affecting the HbA1c levels. The average lifetime of RBC increases when erythropoiesis is suppressed, due to the lack of iron and vitamin B12, leading to high HbA1c levels (21-23,69). Kim et al (70) demonstrated that women with an iron deficiency without anemia (n=1,150) exhibited a small increase in HbA1c levels (<5.5% to ≥5.5%), independent of fasting glucose levels. In another study, comparing the use of both HbA1c and fasting blood glucose as diagnostic criteria, Attard et al (71) demonstrated that males with an iron deficiency alone or with iron deficiency anemia (IDA) exhibited a greater relative risk of developing pre-diabetes. However, other studies have yielded inconclusive results. According to a meta-analysis, IDA and iron deficiency had no effect on the HbA1c levels (72). In addition, patients with IDA have been found to have a higher glycation rate, which may be due to the higher malondialdehyde levels, a lipid peroxidation metabolite, observed in this population, thus enhancing Hb glycation (73,74). Additionally, it has been reported that splenectomy can promote RBC survival, thus enhancing glycated Hb levels (75). Conversely, a decrease in the mean age of RBCs can reduce glycated Hb levels. However, in the absence of fibrosis and splenomegaly, this can be observed in chronic liver disease, although the cause remains unknown (76). Furthermore, splenomegaly and rheumatoid arthritis can also increase the rate of hemolysis, thus resulting in a decrease in glycated Hb value (77).

It has been reported that HbA1c levels can be affected by age and race. Previous studies have indicated that after the age of 30, the glycated Hb value can be increased by ~0.1% every 10 years (78,79). The effects of race on HbA1c values are controversial, however. Accumulating evidence has suggested that differences in HbA1c levels can be observed among different races. Therefore, several studies have demonstrated that African Americans and Hispanics have higher glycated Hb levels compared with Caucasians at the same blood glucose levels (79-81). Additionally, Selvin et al (82) demonstrated that the mean HbA1c value of African Americans was 0.4% higher compared with that of non-Hispanic whites. However, race did not alter the association between HbA1c concentrations and adverse cardiovascular outcomes or mortality (82).