1. Introduction
With the worldwide improvement in the living conditions of humans in general, accompanied by problematic changes in diet and life-style habits such as reduced exercise, diabetes mellitus (DM) has become a major epidemic disease. According to the latest report by the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC), there were 415 million people with diabetes around the world, of which 47% of the patients were unaware of their state of illness [
1].
Glycated hemoglobin (HbA1c) is a typical glycosylated protein in the body, and its abundance reflects the average blood glucose level over two to three months, corresponding to the 100- to 120-day lifespan of erythrocytes. The HbA1c level can not only be used by diabetes patients to monitor their long-term glucose management in a way that is not affected by fluctuations of the blood-glucose level, but also can be used by doctors to assess potential risks of diabetes complications of patients. In 2010 and 2011 the American Diabetes Association (ADA) and World Health Organization (WHO) have recommended a diagnostic cut-off point of ≥6.5% HbA1c as one of three diagnostic criteria for diabetes, respectively [
2,
3].
The glycosylation of hemoglobin occurs via a sequential two-step non-enzymatic reaction. Firstly, the neutral amino groups from N-terminal residues or from the side chain of Lys residues in hemoglobin (Hb) interact with the aldehyde or ketone groups in sugar molecules to form reversible Schiff base intermediates. Then the intermediate undergoes an irreversible intermolecular Amadori rearrangement to generate a more stable ketoamine structure [
4]. The probabilities of glycosylation are dependent on the local pKa of the amino groups as well as the charge and steric effects by neighboring residues [
5]. In general, the pKa values of α–amino groups of N-terminal residues, especially the one of βVal1 (close to 7), are lower than those of ε–amino groups of Lys residues in Hb. In addition, a positively charged cavity around the βVal1 moiety has a strong affinity to attract sugar molecules. Due to the high concentration of glucose in the blood stream, the most abundant form of the glycosylated hemoglobins (GHbs) is the glucose adduct named HbA1c.
The HbA1c level is defined as the ratio of HbA1c to the total Hb concentration, and the physiological range of HbA1c in the whole-blood samples is 3–13 mg/mL in terms of concentration or 5% to 20% in terms of percentage of total Hb. The ADA-recommended diagnostic criteria of HbA1c for diabetes and prediabetes are shown in
Table 1 [
6]. For pregnant diabetic individuals, it is strongly recommended that they have a stringent control on their HbA1c to minimize risks such as congenital malformations, overweight infants, and complications of pregnancy [
7]. Accurate and precise methods to detect HbA1c are thus required for better diagnosis and management control of DM.
In the past several decades, a variety of HbA1c detection methods has been developed, such as immunoassay [
8], ion-exchange chromatography [
9], boronate affinity chromatography [
10], electrophoresis [
11,
12], and a colorimetric method [
13]. Those methods can be generalized into three basic chemical principles based on charge differences, structural differences and chemical reactivity. Most of these methods are subject to at least one or more types of interferences from Hb variants including various forms of Hb (HbC, HbS, HbE, HbD, and HbF) and/or the modification of Hb (cabamylated Hb, acetylated Hb, labile HbA1c) [
14]. The global standardization of HbA1c was conducted by the IFCC and evaluated by different designated comparison methods developed by the National Glycohemoglobin Standardization Program (NGSP) and others, resulting in a laboratory-based gold reference method of HbA1c [
15]. Although the IFCC reference method can provide precision and accuracy for HbA1c measurement [
16], the needs for sophisticated equipment and the professional personnel to operate the system limit its accessibility to only large medical organizations or research institutions [
17]. Therefore, it is valuable to develop methods that are easy to operate and cost-effective, but robust enough for clinical use. HbA1c biosensors have great potential for the design of analytical devices with appropriate sensitivity, low cost, simplicity, and possibility for miniaturization. Recently, several very good review articles have summarized a wide range of methods being used for the determination of glycosylated proteins including HbA1c [
18,
19,
20,
21,
22,
23]. In this review, we summarize recent progress in the development of HbA1c biosensors from several aspects, including the prototypical methodology for HbA1c measurement, applications in whole-blood sample analysis, and the development of point-of-care (POC) technology in this research field, including non-invasive biosensors for diabetes.
3. Analysis of Samples from Human Blood
A good sensor system with potential for clinical use must be able to assess HbA1c level with the sensitivity and reproducibility while provide excellent operational and reagent-storage stability. The sensor system should be able to distinguish HbA1c from non-glycated hemoglobin (HbA0) or other interferences and demonstrate excellent agreement with a standard analytical method.
Table 5 summarizes some the applications of biosensors in HbA1c measurement from whole-blood samples.
Most of the examples described in the section above mainly focus on the method development for HbA1c determination. For practical application, HbA1c content rather than HbA1c concentration is used, which is the ratio of HbA1c concentration to total Hb concentration. Units for HbA1c content are commonly reported either in mmol/mol (used by the IFCC) or in percentage format (used by the NSGP). In general, a pretreatment of blood samples is needed to remove plasma interference followed by lysis of the RBCs. Hb and HbA1c are then measured separately using those biosensor systems.
An UV-Vis spectroscopy method can be used to quantify Hb concentration before the hemolysate sample is applied to biosensors for HbA1c readout [
36]. The HbA1c content can easily be calculated from these numbers. However, this kind of detection method is inconvenient in practice because the detection and separation are distinct processes.
Halámek and coworkers have developed FcBA-based biosensors for the detection of HbA1c [
25,
40]. Hbs are adsorbed to the surfactant-modified surface first, and then followed by using the mass-sensitive quartz crystal balance, and also using voltammetry to monitor total Hb and HbA1c, respectively. HbA1c binding to the APBA-modified electrodes was reversible, thereby providing a reusable sensing system. As such, Halámek applied the feature to develop a novel HbA1c biosensor based on flow injection [
72]. They used a reticulated vitreous carbon (RVC) electrode modified with 3-APBA to separate and detect the HbA1c concentration, followed by the detection of Hb using the screen-printed electrode modified with a sol-gel film involving chitosan, and tetraenthoxyl silica entrapped into carbon nanotubes. This sensor showed a good relation between signal and Hbs concentrations.
Even though gold has been commonly used in the biosensor system, carbon or membrane-based materials have been adopted into for their biocompatibility and cost effective feature. Booyasit constructed an 3-APBA-modified ESMs electrode to detect HbA1c and a Hp-modified ESMs electrode to determine the Hb [
32]. These sensors can simultaneously detect total Hb and HbA1c with excellent precision and also an acceptable reproducibility of fabrication. The chemical stability of the Hp-modified ESMs is good (98.84% over a shelf-life of 4 weeks), however, for the APBA-modified ESMs, the stability is not so good (92.35% over a one-week period). Thus one needs to improve the stability of the APBA-modified ESM for clinical application.
In addition to the electrochemical signal transduction pathway, optical detection methods are also used in the HbA1c biosensor systems. Ahn and coworkers fabricated a HbA1c-capturing interface made of carboxy-EG6-undecanethiol SAM coupled with 3-APBA on a gold thin-film substrate, and used an 11-amino-1-undecanethiol SAM modified gold substrate for Hb immobilization [
73]. The chemical luminescence (CL) response in the Luminol/H
2O
2 system in which the four heme groups play a role as the catalyst is linearly proportional to the amount of Hbs. The luminol CL method not only provides a high sensitivity for HbA1c detection without the need for signal amplification, but also can be used for Hb detection on an amino-SAM based interface, which means this method can be applied on whole blood sample analyses. The linear dynamic range of HbA1c level is from 2.5% to 17% which covers the clinical concentration range with negligible interference from other carbohydrates.
One of the main advantages of the boronate affinity-based method is that the measurement does not interfere with Hb variants such as HbF, HbS, HbC, formayl-Hb etc., which makes it applicable to a large population. In addition, boronate-based reagents are relative stable and cost-effective compared to HbA1c antibodies. Due to the nature of the boron-cis-diol covalent bond, this method actually measures GHb (HbA1c and Hb glycated at other Lys sites) and also interferes with other endogenous sugar molecules in the sample.
Because of its high specificity and selectivity, the HbA1c antibody is the best candidate for capturing HbA1c. Xia and coworkers developed a micro immunosensor consisting of an ISFET integrated chip (see
Section 2.2.3 above) and an electrode array based on the micro-electro-mechanical systems (MEMS) technology, modified with anti-Hb and anti-HbA1c antibodies [
51]. The sensor can simultaneously detect the concentration of HbA1c and Hb to obtain HbA1c level. The responses of the immunosensor are linear over the concentration range of 166.7–570 ng/ml Hb and 50~170.5 ng/ml HbA1c which are 10
5-fold lower than physiological Hb concentrations. To apply the method for whole blood analysis, the samples were lysed and diluted 150,000-fold prior to measurements. This micro immunosensor exhibited a low relative deviation of measured HbA1c level. By reducing cost, enhancing the shelf-life of the antibody, and invoking lab-on-a-chip (LOC) technology, electrochemical immunosensors will have a great potential to be translated into clinical use.
Moon and coworkers developed a disposable microfluidic amperometric dual-sensor for the detection of HbA1c and total Hb separately [
74]. The concentration of total Hb was measured by the cathodic currents of total Hb catalyzed by a TBO/pTTBA@MWCNT-modified working electrode. On the other surface of the dual-sensor, aptamer was used for HbA1c immobilization. After removal of unbound Hb by washing with PBS, the cathodic current of HbA1c captured on the surface was determined by the aptamer/TBO/pTTBA@ MWCNT-modified working electrode in the fluidic channel.
Given the advantages of minimal sample volume, rapid analysis time and wide accessibility to diagnosis, the global POC diagnosis market is expected to reach US
$36.93 billion by 2021 (
http://www.marketsandmarkets.com/PressReleases/point-of-care-diagnostic.asp). More than a dozen commercial HbA1c POC devices are currently used in clinics, of which the principal detection method is either based on boronate affinity separation or immunoassay. Among most cited devices, In2it
TM (Bio-Rad, Hercules, CA, USA), Alere Afinion
TM (Alere Technologies AS, Oslo, Norway), Nycocard (Alere Technologies AS) and Clover A1c (Inforpia, Kyunggi, Korea) are built based on the affinity method, while DCA Vantage (Siemens Medical Solutions Diagnostics, Tarrytown, NY, USA) and A1CNow series (Bayer HealthCare, Sunnyvale, CA, USA) are based on the immunoassay method. In a very recent review, Hirst and coworker independently reviewed 1739 records published before June 2015 in several databases (Medline, Embase and Web of Science) and carried out a meta-analysis on sixty-one studies to compare the accuracy and precision of eleven HbA1c POC devices [
75]. The analysis results show that the majority of devices (9 out of 11) have a negative mean bias compared to laboratory comparator methods, as well as a large variability in bias within devices. The implication of using HbA1c POC testing results on medical treatment decision-making and patient outcomes needs to be evaluated further.
4. Conclusions
With the number of diabetic patients increasing each year, it has caused a huge burden on public medical resources. A HbA1c POC device with good analytical performance can help patients with diabetes to monitor their long-term glycemic status. More importantly, the characteristics of immediate feedback can avoid delaying the diagnosis and treatment of diabetes. The boronate affinity-based method and the immunoassay method are the two methodologies currently used in commercially available HbA1c POC devices. Given the advantages of being relatively simple to operate, easy to miniaturize as well as its high sensitivity, the electrochemical HbA1c biosensors can be potentially translated for clinical use.
Other noninvasive technologies have emerged to supplement current blood glucose and HbA1c measurements for diabetes diagnosis and management. For example, Saraoglu developed a QCM sensor by an electronic breath-analysis system to detect the amount of acetone in exhaled breath for indirectly determining HbA1c levels and blood glucose for 30 patients’ samples [
76]. The average accuracy rates for HbA1c level and blood glucose predictions are approximate 83% and 75%, respectively. With an expected global market of up to US
$12.2 billion by the end of 2017, blood glucose monitoring occupies the top share in the POC market [
77]. Raman scattering has also become a promising tool for noninvasive, continuous tracking of blood glucose level [
78,
79]. By exploiting an improved concentration independent calibration approach, Spegazzini and coworkers performed a longitudinal tracking of blood glucose which exhibited a 35% reduction in error compared to a conventional calibration method [
79].
The integrations of affinity- and immunoassay-based HbA1c biosensors with emerging cell phone-based technology and LOC platforms, supplemented by noninvasive blood glucose monitoring provide a highly promising interface for the design of new generation of personalized HbA1c POC devices [
77].