OMEF biochip for evaluating red blood cell deformability using dielectrophoresis as a diagnostic tool for type 2 diabetes mellitus

Abstract

Type 2 diabetes mellitus (T2DM) is a prevalent and debilitating disease with numerous health risks, including cardiovascular diseases, kidney dysfunction, and nerve damage. One important aspect of T2DM is its association with the abnormal morphology of red blood cells (RBCs), which leads to increased blood viscosity and impaired blood flow. Therefore, evaluating the mechanical properties of RBCs is crucial for understanding the role of T2DM in cellular deformability. This provides valuable insights into disease progression and potential diagnostic applications. In this study, we developed an open micro-electro-fluidic (OMEF) biochip technology based on dielectrophoresis (DEP) to assess the deformability of RBCs in T2DM. The biochip facilitates high-throughput single-cell RBC stretching experiments, enabling quantitative measurements of the cell size, strain, stretch factor, and post-stretching relaxation time. Our results confirm the significant impact of T2DM on the deformability of RBCs. Compared to their healthy counterparts, diabetic RBCs exhibit ∼27% increased size and ∼29% reduced stretch factor, suggesting potential biomarkers for monitoring T2DM. The observed dynamic behaviors emphasize the contrast between the mechanical characteristics, where healthy RBCs demonstrate notable elasticity and diabetic RBCs exhibit plastic behavior. These differences highlight the significance of mechanical characteristics in understanding the implications for RBCs in T2DM. With its ∼90% sensitivity and rapid readout (ultimately within a few minutes), the OMEF biochip holds potential as an effective point-of-care diagnostic tool for evaluating the deformability of RBCs in individuals with T2DM and tracking disease progression.

---

Introduction

The Global Burden of Disease study highlighted that in 2021, over 0.5 billion adults, aged between 20 and 79, were diagnosed with diabetes, and 6.7 million died due to diabetes.¹ Projections for the near future are even more concerning, with estimates indicating an increase to 1.3 billion diabetic patients by 2050.² Among the various forms of diabetes, type 2 diabetes mellitus (T2DM) stands as the predominant one, accounting for over 90% of the diagnosed cases.³ T2DM is characterized by insufficient insulin production, which is primarily attributed to pancreatic β-cell malfunction, coupled with simultaneous insulin resistance in target organs.⁴ These metabolic abnormalities significantly increase the risk of macro- and microvascular complications among older adults with T2DM, thus contributing to increased mortality rates.⁵

In diabetic individuals, the persistence of elevated blood glucose levels and the presence of glycated hemoglobin in the bloodstream have detrimental effects on the mechanical properties of red blood cells (RBCs).⁶ These alterations disrupt the normal functioning of RBCs, leading to adverse health complications such as cardiovascular diseases,⁷ kidney dysfunction,⁸ nerve damage,⁹ retinopathy,¹⁰ and atherosclerosis.¹¹ The relationship between these health outcomes and diabetes points out the significant impact on the mechanical properties of RBCs and their broader implications for overall health. Consequently, gaining a thorough understanding of these effects becomes paramount for the development of effective diagnostic and therapeutic approaches that aim to help manage diabetes and prevent associated complications.

RBCs, also known as erythrocytes, are unique biconcave disk-shaped, nucleus-free cells enveloped by a plasmatic membrane.¹² With diameters ranging from 6 to 8 μm, RBCs circulate in the bloodstream, where they transport oxygen from the lungs to various tissues and return carbon dioxide back to the lungs. During the process, RBCs utilize hemoglobin, an iron-rich protein within the cells.¹² This vital task often requires these cells to undergo deformation and navigate through capillaries narrower than their diameter, typically around 3 μm.¹³ Furthermore, during the blood filtration in the spleen, RBCs must squeeze through gaps ranging from 0.5 to 10 μm between endothelial cells.¹⁴

The ability of RBCs to deform while retaining their characteristic shape is pivotal for their optimal functioning. This deformability is determined by three key cell properties:¹⁵ cell shape, which is characterized by a significant surface area-to-volume ratio; cytoplasmic viscosity, which is largely influenced by the hemoglobin content; and membrane viscoelasticity, which contains proteins that facilitate maximal recovery of the cell shape after deformation.

The shape of RBCs is highly adapted to their primary function, which is the efficient transport of oxygen throughout the body. Meanwhile, the deformability of RBCs primarily depends on their flexible membrane, which can change shape without jeopardizing its integrity. This flexibility is crucial for RBCs to pass through the smallest capillaries and effectively deliver oxygen to tissues. Several factors contribute to RBC deformability, and their distinctive shape is a pivotal factor in this process.^16,17

Pathological alterations in the RBC deformability can result in increased rigidity, which may potentially cause severe vascular complications.⁷ The increase in rigidity can be triggered by hereditary conditions such as sickle cell anemia or acquired diseases such as diabetes.¹⁸ In individuals with T2DM, the deformability of RBCs is altered due to glycation, where excess blood glucose binds to hemoglobin.¹⁹ This leads to the formation of glycated hemoglobin and increased rigidity of the RBC membrane.²⁰ Elevated levels of glycation have detrimental effects on RBC mechanical properties, resulting in reduced protein fluidity²¹ and decreased deformability.²² Therefore, based on these facts, there is a great opportunity for developing a new approach for studying how T2DM affects RBC mechanical characteristics. Such an approach can be utilized for developing a point-of-care diagnostic tool for predicting disease progression and its impact on bodily functions.

The mechanical properties of the RBC membrane have been characterized through various properties, including shear, area compressibility, bending, and elasticity.^14,23–25 Currently, there are several techniques available to analyse the alterations in RBCs at the single-cell level, such as glass microneedles,²⁶ optical and magnetic tweezers,²⁷ converging–diverging microfluidic channels,²³ and atomic force microscopy (AFM).²⁸ While these single-cell approaches provide high precision, they have limitations including low throughput and slow processing speed. In contrast, techniques that involve whole blood and diluted RBC suspensions, such as viscometry and ektacytometry, offer better throughput compared to single-cell methods. However, they do not account for heterogeneity or size differences within the cell population in a single sample and lack the capability for single-cell level analysis.²⁹

Dielectrophoresis (DEP), on the other hand, is an emerging technique that allows high-throughput manipulations on biological samples without causing irreversible damage.³⁰ Example applications include trapping enzymes, enriching proteins, concentrating and detecting viruses, separating yeast and bacteria, and manipulating cancer cells, among others.³¹ DEP has also been utilized in several other studies to investigate RBCs under different experimental conditions. For example, Nakashima et al.³² developed a microfluidic device to selectively separate RBCs from white blood cells. Guido et al.³³ demonstrated, through development of microchips, that variations in stretching responses are attributed to the distinct cell-specific mechanical properties. Doh et al.³⁴ employed a single-cell microchamber array chip technology to correlate RBC deformation with the cell surface and cytosolic characteristics and define the distribution of individual cellular characteristics in diverse populations. Qiang et al.³⁵ introduced a microfluidic method where cyclic deformation led to a significantly greater loss of membrane deformability in RBCs compared to static deformation under the same load. Qianqian et al.³⁶ developed a microchip to examine the viscoelastic properties of cryopreserved RBCs. Their approach offered a non-invasive and label-free method to assess the quality of cryopreservation by measuring changes in shear modulus and viscosity. Although these studies provided valuable insights into RBC physiology and dynamics, DEP has not been previously employed to assess the mechanical properties of RBCs from diabetic patients.

In this study, we investigate the deformability of RBCs in T2DM patients by introducing an open micro-electro-fluidic (OMEF) biochip that utilizes DEP to capture and elongate (stretch) RBCs of interest. Fabricated using photolithography, the biochip serves as a platform on which RBCs undergo stretching when subjected to a non-uniform electric field. The process is conducted under an inverted microscope, which enables the recording of stretched cells for subsequent analysis. The OMEF biochip allows us to study several mechanical properties in RBCs, including strain, stretch factor, diameter, and relaxation time post-stretching, and compare these values between two distinct groups: healthy individuals and diabetic patients.

Materials and methods

The DEP method

DEP, an ideal technique for particle manipulation,³⁰ relies on the interaction between a dielectric particle and a non-uniform electric field. For a cell subjected to a non-uniform electric field, positive charges move along the direction of the field, while negative charges move in the opposite direction. This phenomenon leads to the separation of opposite charges within the cell, resulting in its polarization. As the cell becomes polarized, it generates an internal electric field that interacts with the external field, leading to the establishment of an electric field gradient between the two. As a result, the quantifiable polarizability of the cell creates a dipole moment that exerts a dielectrophoretic force (F_DEP) on the cell. Depending on its orientation, the F_DEP can either attract or repel the polarized cell. The time-averaged F_DEP on a cell in three dimensions is given by:

F_DEP = 2πε_mr³_c Re(α)∇(E⃑·E⃑) (1)

where ε_m is the permittivity of the suspending medium, r_c is the radius of the cell, Re(α) is the real part of the Clausius–Mossotti factor α (a complex number that characterizes the polarizability of the cell), and ∇(E⃑·E⃑) is the gradient of the square of the electric field strength. The magnitude of the F_DEP depends on the frequency of the applied voltage. Meanwhile, the direction of the F_DEP depends on the α, which is a ratio defined as:

α = (ε_cell − ε_medium)/(ε_cell + 2ε_medium) (2)

where ε_cell is the normal permittivity of the cell and ε_medium is the normal permittivity of the surrounding medium. Positive α indicates positive DEP (p-DEP), where cells are attracted to regions of higher electric field. This occurs when the polarizability of cells is greater than that of the suspending medium. Meanwhile, negative α indicates negative DEP (n-DEP), in which cells are attracted to regions of lower electric field. This occurs when the polarizability of the cells is less than that of the suspending medium.

In addressing the polarization of a bioparticle, such as a viable cell, within a medium with differing dielectric properties, such as a DEP medium, in our study we considered the normal scenario, where an ideal dielectric sphere is suspended in an ideal dielectric medium without ohmic conduction of mobile charges (refer to eqn (2)). This scenario implies no energy loss in the form of heat due to the absence of conduction in both the particle and the medium. In contrast, the complex scenario considers both conduction and dielectric energy losses in the suspended particle and surrounding medium. This consideration becomes crucial when employing time-dependent electric fields.³⁷

It is noteworthy to mention that the complex permittivity of the cell is typically determined using a model that correlates the dielectric properties (e.g., membrane capacitance, cytoplasm conductivity) of the cell to its structure. Therefore, the choice of the model and its parameters can vary based on the characteristics of the cells and experimental conditions. While it is feasible to calculate the complex permittivity of RBCs,³⁸ its application to the assessment of RBC deformability is recommended for future studies.

DEP medium preparation

Prior to conducting the RBC stretching experiments, the DEP medium was prepared for iso-osmotic cell suspension by adding 3 g L⁻¹ of glucose, 85 g L⁻¹ of sucrose, and 3 mL of 1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES, Thermo Scientific) solution to deionized (DI) water.³⁹ The pH of the DEP medium was continuously monitored, with adjustments made using hydrochloric acid (HCl, Sigma-Aldrich) or sodium hydroxide (NaOH, Sigma-Aldrich) to maintain a value of 7.35 ± 0.10. Additionally, Dulbecco's phosphate-buffered saline (DPBS, Thermo Scientific) was used to adjust the ionic conductivity (i.e., the electrical conductivity due to the motion of ionic charge) of the DEP medium to 30 ± 1 mS m⁻¹, with a total osmolality of 310 mOsm kg⁻¹.

In our stretching experiments, where a DEP medium conductivity of 30 mS m⁻¹ and a frequency range of 1–10 MHz were utilized, the AC electro-osmotic velocity tends to be negligible due to the insufficient time for ionic charges to follow the electric field.⁴⁰ Additionally, electrothermal flow, influenced by AC frequency and ionic conductivity, is unlikely to be significant given the low DEP medium conductivity of 30 mS m⁻¹.³⁹ Therefore, we considered the normal permittivity of the suspended cell and the surrounding medium, as defined by eqn (2), rather than utilizing complex permittivities.

Building on our previous work,³⁹ the normal permittivity of the DEP medium was approximated to be 79 × 8.85 × 10⁻¹². In cases of p-DEP observed in viable RBCs suspended in a DEP medium with an ionic conductivity of 30 mS m⁻¹, the real part of the α in eqn (2) was estimated to be ∼0.65, assuming that RBCs were spherical in shape.⁴¹ However, employing a more accurate model would require consideration of the biconcave discoid shape of RBCs. Investigating this aspect, however, is left for future studies.

Blood sample collection and RBC fractionation

Venous blood samples were collected from 46 volunteers, including 15 healthy individuals and 31 T2DM patients, following informed consent and in accordance with protocols approved by the institutional review boards of New York University Abu Dhabi and Cleveland Clinic Abu Dhabi, United Arab Emirates.

The coauthors of this study, who are medical professionals at Cleveland Clinic Abu Dhabi, informed their patients about the research, and those interested in participating expressed their willingness to the doctor and signed the informed consent form. Additionally, the 46 volunteers approached for blood donation also completed a survey with details on biometric data and medical history, such as age, kidney health, glycemia status, gender, family history of T2DM, details related to smoking habits, and other relevant information (Table S1†). All data, including the medical records, were reviewed to establish correlations with the mechanical properties of RBCs.

After blood collection in ethylenediaminetetraacetic acid (EDTA, VWR) tubes, continuous mixing on a digital rocker was employed immediately to maintain sample homogeneity. Subsequently, 1 mL of whole blood was transferred into a sterile Eppendorf tube and centrifuged at 2200 rpm for 5 min to separate the plasma from the RBCs (Fig. 1a). Afterwards, the plasma was aspirated, leaving a sediment of RBCs at the bottom of the tube. RBCs were then washed in another Eppendorf tube with 1 mL of DEP medium, followed by another centrifugation step at 2200 rpm for 5 min. The DEP medium was then aspirated, and 1 mL of fresh DEP medium was added to resuspend the RBCs at a 1 : 1500 ratio (volume/volume). All blood samples in our study were utilized within 3 h of collection to follow the timeframe guidelines established in transfusion practices.⁴²

Fig. 1 Experimental procedure, OMEF biochip, and setup for assessing the deformability of RBCs in type 2 diabetes mellitus (T2DM) using dielectrophoresis (DEP). (a) The flowchart illustrates the sample preparation steps of RBCs and their deposition onto one of the stretching areas of the developed open micro-electro-fluidic (OMEF) biochip for their follow-up stretching. (b) The micrograph displays the developed OMEF biochip, featuring 6 independent stretching areas, each of which offers complete functionality for stretching. (c) The flowchart illustrates the individual steps involved in stretching RBCs using DEP. (d) The micrograph shows the experimental setup, along with the electronic devices and the microscope, used in this study.

Fabrication and optimization of the open micro-electro-fluidic (OMEF) biochip

The developed OMEF biochip, shown in Fig. 1b, is designed for cell capture and stretching via the DEP method. It is an open biochip that consists of an array of interdigitated electrodes equally spaced apart. To achieve optimal design of the biochip, COMSOL Multiphysics simulations were employed to assess the strength of the electric field, as depicted in Fig. S1a.† The goal was to determine the ideal dimensions for effectively stretching RBCs, which are about 8 μm in diameter. The simulations revealed that the optimal electrode dimensions were a width of 20 μm with a 20 μm spacing.

To enhance the experimental throughput, the optimized design of the OMEF biochip incorporated multiplexing, with 6 independent deformation areas, each of which provides complete functionality for cell stretching (Fig. 1b). Each OMEF biochip deformation area is connected to two gold pads. These pads are bonded to metallic pins and subsequently linked to a voltage source. This configuration enabled the consecutive execution of different stretching scenarios without the need for chip cleaning between the tests.

The developed device represents an advanced version of a stretching chip we previously developed for studying dendritic cells.³⁹ Its fabrication followed standard microfabrication methods, which involved photolithography, metal deposition, and subsequent lift-off procedures previously described in detail.⁴³ Briefly, computer-aided design software (Autodesk) was used for designing a photoresist mask. Afterward, 100 nm thick gold and 5 nm thick chrome layers were deposited onto the glass through sputtering (Cressington 108 Auto Sputter Coater). The electrodes were designed with a width and spacing of 20 μm each to facilitate cell stretching, as depicted in Fig. 1b and c. When the electrodes are biased using an alternating current (AC) source, F_DEP is generated between the electrodes due to the non-uniform electric field.

In the fabrication and optimization of the OMEF biochip, the decision to use electrode dimensions of 20 μm width and 20 μm spacing was guided by three key considerations. Firstly, to accommodate the typical diameters of RBCs, which range from 6 to 8 μm, we opted for a 20 μm electrode width to ensure sufficient margin for cells to be fully stretched. Secondly, an electrode spacing of 20 μm was chosen to minimize the pearl-chain effect,⁴⁴ where cells form a chain-like formation under an applied electric field. Thirdly, these dimensions enabled the analysis of a greater number of cells in a single field of view. By combining these considerations, our electrode dimensions not only facilitated effective and reliable stretching of RBCs but also minimized potential artifacts, thereby improving overall experimental throughput and ensuring statistical significance in the analysis.

Experimental setup

For each experimental set, the OMEF biochip was immersed in a glass beaker filled with isopropyl alcohol and placed in a shaker for 10 min to remove any residue and dust. Following that, it was washed with DI water and dried with a nitrogen gun. Subsequently, the cleaned OMEF biochip was placed on the stage of an inverted fluorescence microscope (Nikon Eclipse Ti2) and connected to the AC voltage wires, as depicted in the experimental setup shown in Fig. 1d.

Throughout the experiments, the microscope was operated in brightfield mode using 20× air objective lenses. A low-noise monochrome camera (Nikon DS-Qi2) was employed to capture high-resolution videos of the stretched RBCs. The AC signal at frequencies exceeding 1 MHz was supplied by a function generator (Keysight N5172B EXG X-Series). The signal from the function generator passed through an amplifier (Mini Circuits ZHL-5 W-1), which was powered by a DC power source set to 24 V, to generate the desired electric field. The output of the amplifier was connected to the terminals (pins) of the OMEF biochip.

In our study, blood samples were maintained at 4 °C from collection until the stretching experiments. Following blood fractionation and separation of RBCs, all stretching experiments were conducted at room temperature (25 °C), under stable humidity conditions, and within a 1 min timeframe, including the deposition of the cell suspension onto the OMEF biochip and the stretching of RBCs for 5 s. This minimized potential variations in RBC mechanical properties associated with temperature changes.⁴⁵ Notably, temperatures beyond 4 °C ± 2 °C for various durations are not expected to significantly affect the quality of the RBCs or lead to substantial bacterial contamination.⁴²

For each stretching experiment, 10 μL of the prepared RBC solution was applied to the connected OMEF biochip to initiate the stretching of RBCs for 5 s. The location of the droplet containing the cells was carefully identified, and its centre was aligned with the centre of the objective lens to ensure that the video recording is conducted away from the curved edges of the droplet.

DEP capture and stretching forces

When suspended RBCs in the DEP medium are deposited on the OMEF biochip, the cells exhibited random movement within the liquid due to convection. When cells settled in proximity to the microelectrode substrate (within a few tens of seconds), F_DEP was applied to pull them towards the nearest electrode edge, effectively immobilize them on the substrate, and polarize them resulting in a dipole moment. The dipole moment led to either attraction or repulsion between the electrodes and cells depending on their electrical properties. Electrodeformation and cell stretching can only be observed under p-DEP. This requires that ε_cell > ε_medium to maintain a positive α value.

To determine the crossover frequency of the AC voltage at which the DEP force changes polarity from negative to positive, we conducted simulations using a MATLAB script to plot the α, as shown in Fig. S1b.† The crossover frequency, at which the polarity reversal occurs, is 0.6 MHz. Consequently, any frequency higher than 0.6 MHz results in a p-DEP force, which leads to the stretching of RBCs. These values were consistent with the findings in the literature.³⁹ Subsequent characterization experiments conducted with the OMEF biochip at frequencies exceeding 0.6 MHz revealed that, for a DEP medium with low conductivity (30 mS m⁻¹), RBCs experience p-DEP effects only when subjected to frequencies greater than 1 MHz.

Optimization of voltage and frequency settings

Following the identification of the minimum frequency required for stretching based on the α, we optimized the “safe” frequency and magnitude ranges of the AC signal needed to achieve maximum stretching of RBCs. Initially, we explored a range of input voltages from 3 to 15 V and adjusted the frequencies within the range of 1 to 10 MHz to determine the optimal stretching frequency, as depicted in the top panel of Fig. 2a. It was observed that at frequencies below 5 MHz, RBCs displayed minimal to no stretching, and at frequencies exceeding 8 MHz, RBCs exhibited lateral movement along the electrode edges (see Video S1†). Considering these undesired effects, we proceeded with subsequent experiments at 5 MHz (see Video S2†). RBCs were captured at the edges of the electrodes at a voltage of 3 V without inducing any noticeable cell stretching. To stretch the cells, the voltage was increased to 7 V for 5 s and subsequently turned off to release the cells or lowered to 3 V to allow the cells to relax from stretching, as depicted in the bottom panel of Fig. 2a.

Fig. 2 RBC stretching characterization of voltage and frequency. (a) Top panel: microscopy images of captured RBCs in elongated (stretched) and relaxed states. Bottom panel: microscopy images showing an RBC captured on the electrode at 3 V and depicting the four states of elongation under increasing AC voltage from 4 V to 9 V with 5 s stretching intervals. Scale bars: 5 μm. (b) Assessment of RBC viability using calcein AM during (top panel) and after (bottom panel) elongation with the optimized voltage to confirm cell viability and integrity. There was no change in the fluorescence (green) intensity after deformation indicating non-invasive stretching. Scale bar: 20 μm. (c and d) Bar charts displaying the average strain and stretch ratio of RBCs subjected to increased voltages (4–9 V) and frequencies (4–10 MHz), respectively. The analysis included 10 healthy RBCs (each bar) from a single donor. Error bars represent the mean ± SD.

Our experimental design focused on establishing uniform conditions to ensure that RBCs from different donors or patients, characterized as the same cell type and falling within the measured size range, experience similar DEP-based stretching under the adjusted ionic conductivity of the DEP medium (30 ± 1 mS m⁻¹). However, recognizing the significance of exploring and characterizing individual cell responses, especially at the single-cell level, we acknowledge the potential variations in DEP force factors (α) or crossover frequencies among RBCs from sampled donors or patients. Therefore, for more accurate RBC deformability measurement, future studies will explore the influence of RBC types, sizes, and shapes, along with varying DEP medium conductivities, to validate the universality of DEP-based RBC stretching at 5 MHz. Additionally, we will consider the translational potential of the findings in clinical diagnostics.

Evaluation of cell viability

To ensure that the selected parameters for electrodeformation did not mechanically damage the cells, we conducted viability assessment of RBCs after stretching using different frequencies and voltages. The staining process was conducted using a Live/Dead cell assay kit (Invitrogen) and calcein AM cell-permeant dye (Invitrogen), which serves as an indicator of intracellular esterase activity.⁴⁶ We followed the protocol provided by the manufacturer to assess RBC survival following the stretching procedure (as shown in Fig. 2b). The results demonstrated that the RBCs maintained their integrity, with ∼99% of the cells remaining viable after being subjected to the stretching.

Assessment of cell deformation and relaxation time

Frame-wise analysis was conducted to monitor the deformation of the RBCs and quantify the geometric alterations in cell shape. We evaluated the cell deformation by employing the strain and stretch factor. Here, strain (δ) is defined as the ratio of the cell's elongation (ΔL) to its original length (L):

δ = ΔL/L (3)

On the other hand, the stretch factor, denoted as β, is defined as the ratio of the major cell axis (R_major) to the minor cell axis (R_minor) for a stretched RBC:

β = R_major/R_minor (4)

The values for the strain and stretch factor corresponding to different voltages and frequencies are shown in Fig. 2c and d, respectively. Additionally, we assessed the relaxation time, which is the duration required for RBCs to return to their initial state after the release of the DEP stretching forces. To record the relaxation time, RBCs were allowed to fully relax before concluding the video recording. In each recorded video, we analyzed a set of RBCs to obtain the necessary mechanical and temporal parameters. Videos were recorded at a rate of 100 frames per second and then were subjected to in-depth analysis to extract the relevant mechanical properties of RBCs, as explained below.

Video analysis

After conducting the RBC stretching experiments, we employed Kinovea image and video processing software (https://kinovea.org) that provided a semi-automated approach for analysis of the video content. It was utilized to trace and quantify the stretching and relaxation of cells observed in the recorded videos using the object tracking function. However, cell selection was manual and guided by the user to omit analysis of artefacts or non-single cells. We measured and recorded the R_major and R_minor axes of the RBCs before, during, and after deformation. To track the deformation of RBCs, we conducted frame-wise analysis. This analysis provided valuable data on the mechanical behavior of the RBCs during the stretching experiments. However, the manual approach we followed does not support high-throughput cell analysis.

Statistical analysis

The values are presented as the mean ± standard deviation (SD). Plots were generated using OriginPro data analysis and graphing software (OriginLab). Where appropriate, statistical analysis was conducted with OriginPro using Student's t test. A p value of <0.0001 was considered statistically significant.

The analysis includes n = 420 for healthy RBCs, averaging 28 cells per healthy donor for a total of 15 healthy donors. Additionally, the analysis includes n = 1208 for diabetic RBCs, averaging ∼39 cells per diabetic patient, with a total of 31 diabetic patients. Although this dataset enabled a thorough examination of RBC mechanical properties in both healthy and diabetic cohorts, the relatively low sample sizes were due to manual image processing. In this regard, we acknowledge the importance of a larger sample size for enhanced statistical accuracy. In future studies, as we integrate deep learning into our analysis, we anticipate a more extensive dataset, which will contribute to a more comprehensive and robust reporting of our findings.

Variation in the RBC deformability

The glass slide of the OMEF biochip accommodated four deformation areas (Fig. 1b), and all standard deviations were presented for the data collected using these areas. Thus, for a single patient, if four experiments were conducted, each would be performed on a separate deformation area on the same slide, and the entire slide would be thoroughly washed after the four experiments.

The high fabrication precision of the OMEF biochip is a key factor for minimal variation in the RBC deformability, as evidenced by small margin errors in photolithography procedures. It is also noteworthy that all our measurements are qualitative rather than quantitative, with a focus on percent elongation per cell rather than the total elongation of each cell. This consideration led us to normalize the strain and stretch factor with respect to cell diameters, taking advantage of the precision of the photolithography process to ensure accurate and reproducible results.

Regarding concerns on diurnal variation and reproducibility, our experimental design was controlled for potential variations. Stretching experiments were consistently conducted with the same time, temperature, and humidity to minimize external factors that might influence RBC mechanical properties. Further investigations, possibly incorporating multiple time points, are suggested for future studies to explore diurnal variations and assess reproducibility over different intervals for RBC deformability within the same participant.

Reusability of the OMEF biochip

The OMEF biochip provides reusability for numerous experiments. To ensure the biochip's usability and prevent nonspecific adsorption, a thorough cleaning protocol—with isopropanol, ethanol, and DI water washes, followed by nitrogen gas drying—is implemented after each experiment. Additionally, each of the four deformation areas of the OMEF biochip can be reused an average of 20 times during stretching experiments. After each use, microscopy observations confirmed the absence of nonspecific adsorption during the duration of experiments. However, it is essential to note that for commercial purposes, the disposal of the device after single use is recommended. In mass production, the cost-effectiveness of disposable devices is likely to be significantly higher compared to those fabricated in a laboratory setting. This aspect highlights practical considerations for broader-scale applications of the OMEF biochip.

Results

Our primary objective of conducting the RBC stretching experiments was to compile a comprehensive record of the mechanical properties of RBCs in T2DM patients. In each experiment, mechanical parameters were averaged across cells, as shown in Fig. S2a.† Cells that exhibited the pearl chain formation,⁴⁴ where coinciding axes caused cells to touch each other (Fig. S2b and Video S3†), were excluded from the analysis.

After completion of the RBC stretching experiments using the OMEF biochip, the microscopy video recordings of the process, and the subsequent video analysis, our conclusions were based on the provided demographic information and the observed properties of RBCs from the two participant groups, namely healthy (control) and diabetic. We obtained four key parameters from the analysis of RBC elongation: the original cell size before elongation, the strain during elongation, the stretch factor during elongation, and the relaxation time required after elongation. Moreover, we categorized the participants based on their medical reviews and questionnaire responses to analyse the impact of each parameter on different subgroups. Initially, the participants were divided into the healthy group (control) and diabetic group. This allowed us to investigate the influence of T2DM on individuals and compare its effects on patients. Subsequently, we performed a more detailed examination of diabetic patients by further subdividing them according to various influencing factors (see Table S1†). These subgroups shed light on the effects of kidney health, the glycemia status, genetics, and smoking habits on the health of RBCs. We also included additional categories based on age and gender, which allowed for a differentiated study of how T2DM is moderated across various age groups and genders.

RBCs in control and diabetic groups

Cell size and shape. The size of RBCs was measured before stretching and averaged across cells. The average diameter for T2DM patients was 8.80 ± 0.03 μm, while healthy individuals had a lower average diameter of 6.80 ± 0.02 μm (Fig. S3a†). This indicates that RBCs from T2DM patients have significantly larger diameters, ∼27% larger than those from healthy individuals. This enlargement in RBC size is attributed to hyperglycemia, which leads to increased glucose influx into the RBCs. This, in turn, disrupts the characteristic shape of RBCs and, consequently, impairs their deformability.⁴⁷

Moreover, to explore the variations in elasticity between the RBCs of healthy individuals and diabetic patients, we applied a static loading test (Fig. 3a and b). This test is analogous to a standard mechanical tensile test, where a specimen is clamped at two ends and pulled until it fails. In such tests, the outcome typically produces a stress–strain graph from which the elasticity of the specimen could be determined.⁴⁸ Importantly, we wanted to determine whether the observed deformation in our study is plastic (resulting in a permanent change in cell size) or elastic (allowing the cell to return to its original size), as this is crucial for understanding the bio-nano-mechanical properties of the investigated RBCs.

Fig. 3 Deformation of healthy and diabetic RBCs. (a and b) The graphs represent static loading on 5 healthy and 5 diabetic RBCs of single donors, respectively, for a duration of 25 s at 7 V. (c and d) The scatter plots differentiate between the strain and stretch factor in both healthy and diabetic RBCs. As shown by the solid purple and cyan plots for healthy and diabetic RBCs, respectively, there is a linear decrease in strain as the cell diameter increases (solid purple and cyan fits, respectively). In contrast, the stretch factor displays a linear increase in both cases. The averages of these slopes (dashed black lines) enable the differentiation of RBC deformability between healthy individuals and those with T2DM, achieving ∼90% sensitivity. The analysis includes n = 420 for healthy RBCs and n = 1208 for diabetic RBCs, averaging 28 cells and ∼39 cells per healthy donor and diabetic patient, respectively. The total number of healthy donors who participated in this study was 15, and the total number of diabetic patients was 31. (e) The bar chart displays the averages of the strain, stretch factor, and relaxation time for comparison between healthy individuals and T2DM patients who participated in this study. Error bars represent the mean ± SD.

In our initial experiments, we monitored the elongation of the unrestrained side of RBCs for 25 s during static loading at 7 V (see Video S4†). Specifically, we examined 5 RBCs from a healthy volunteer and 5 RBCs from a patient with T2DM. The findings revealed a significant difference in the behavior of RBCs between the two groups (Fig. 3a and b). Following 25 s of static loading, the healthy RBCs exhibited an elastic behavior with a complete return to their original size, while the diabetic RBCs showed limited recovery (∼30%), which was indicative of a plastic behavior. This difference in behavior is crucial for distinguishing between healthy and diabetic RBCs based on their mechanical characteristics.

These experiments also revealed the rapid stretching of cells, and thus we used the stretching-duration of 5 s in all following experiments, which is sufficient for stable stretching of all cells and yet rapid and involves less post-processing time in result analysis. Furthermore, to minimize the influence of variation in cell size, we further normalized the obtained mechanical properties with respect to cell diameters for the strain and stretch factor, respectively. Significantly, our analysis of the normalized strain and stretch factor against individual cell diameters revealed that averaged linear fits enable differentiation of healthy RBCs from diabetic RBCs with ∼90% sensitivity (Fig. 3c and d). Notably, healthy RBCs, being generally smaller in size, demonstrated greater stretchability, resulting in higher strain and stretch factor values compared to diabetic RBCs. This correlation is visually presented, revealing a distinct linear decrease in strain with an increase in cell diameter for healthy and diabetic RBCs. Conversely, the stretch factor exhibited a linear increase in both cases. Quantitatively, the slopes of the strain decrease were determined to be −0.07 ± 0.02 strain per μm and −0.05 ± 0.01 strain per μm for healthy and diabetic RBCs, respectively. Meanwhile, the slopes for the increase in the stretch factor were 0.11 ± 0.01 stretch factor per μm and 0.07 ± 0.01 stretch factor per μm for healthy and diabetic RBCs, respectively.

Our study, while acknowledging the significance of size variability, primarily focused on obtaining an overall understanding of RBC behavior under DEP-based stretching. For more comprehensive analysis, future investigations could explore the relationship between size variations of RBCs and their mechanical parameters in greater detail. This might involve categorizing RBCs based on size ranges and systematically studying the strain and stretch factor variations within these groups. Additionally, considering the potential impact of the RBC size on clinical implications and disease states, further investigations on how size-related variations might relate to health conditions or pathological states could enhance the depth of the findings.

These findings align with a study by AlSalhi et al.⁴⁹ that utilized AFM to analyse the effects of T2DM on the shape, size, and mechanical properties of RBCs. Their results, which were grouped based on hemoglobin A1c (HbA1C) levels into healthy (<6%), prediabetic (between 6% and 7%), and diabetic (>7%) patients, showed similar trends. Imaging by AFM revealed that healthy RBCs have an average diameter of 8.13 ± 0.81 μm, while prediabetic cells have a diameter of 6.87 ± 0.56 μm, and diabetic cells have a diameter of 9.11 ± 0.81 μm. Additionally, the same study reported differences in the concave depth, which is the depth at the centre of the cell. Healthy cells exhibited a concave depth of 267.1 ± 66 nm, while prediabetic cells had a shallower concave depth of 113 ± 46 nm. Interestingly, prediabetic cells displayed a more swollen structure, whereas diabetic cells exhibited a convex, balloon-like shape. These provide valuable insights into the relationship between the cell shape of RBCs and their ability to deform under various diabetic conditions. Indeed, the biconcave discoid shape enables RBCs to maintain a large surface area-to-volume ratio, essential for efficient oxygen exchange. Additionally, the unique cytoskeletal structure of RBCs, composed of a network of spectrin and other proteins, provides both stability and flexibility to the cell membrane. This unique structure allows RBCs to deform and subsequently return to their original shape as they traverse the circulatory system.^16,50

Strain and stretch factor. The elasticity of RBCs from healthy and diabetic patients was investigated by calculating the strain and stretch factor. The data indicated significant variation in these parameters between the two groups (Fig. 3e and S3b and c†). RBCs from healthy individuals stretched by a ratio of 0.65 ± 0.07, while those of diabetic patients stretched by 0.36 ± 0.04. Thus, the strain of RBCs from healthy individuals was significantly higher than that of RBCs from T2DM patients (∼55%). Meanwhile, the average RBC stretch factor for T2DM patients was 1.00 ± 0.02, ∼29% less than that of healthy individuals, which stretched by 1.41 ± 0.03.

The considerable difference in the strain and stretch factor of RBCs from the two groups can be attributed to physiological changes resulting due to the glycation of RBCs in T2DM patients.^6,51 These RBCs experience elevated levels of blood glucose and HbA1c, which lead to reduced deformability and compromised biological function in RBCs.¹⁸ Abnormal glycation levels in hyperglycemia are also shown to negatively affect the mechanical properties of RBCs due to disruption of hemoglobin synthesis and polypeptide synthesis,⁵² thus reducing membrane fluidity, increasing stiffness, and impairing deformability.^21,22

Relaxation time. The results regarding the relaxation time of RBCs in both healthy individuals and diabetic patients, as presented in the histogram (Fig. S3d†), show that there is no significant difference between the two groups. Healthy individuals exhibited an average relaxation time of 0.65 s, while diabetic patients had a slightly longer average relaxation time of 0.69 s. The similarity in relaxation times between healthy and diabetic individuals suggests that this specific mechanical property may not be a highly sensitive marker for distinguishing between the two groups. However, it is crucial to recognize that the deformability of RBCs is a multifaceted phenomenon, and other mechanical properties and parameters may exhibit more distinct differences in response to diabetes. Future studies should explore various aspects of RBC mechanics to better understand their relationship with diabetes and its complications.

RBCs and demographic factors

Age. We examined a random age pool comprising healthy individuals and T2DM patients. Healthy individuals were between 26 and 62 years old, while T2DM patients ranged from 35 to 80 years old. Analysing the results with respect to age allowed us to understand how T2DM correlates with the age of the participant and its impact on the mechanical properties of RBCs across different age groups.

Moreover, analysing the strain and stretch factor of RBCs across different age groups, we found no correlation between age and stretchability (Fig. S4†). Thus, among all diabetic patients, RBCs from older individuals did not exhibit significantly different stretching behavior compared to RBCs from younger individuals.

Kidney health. A total of 30 diabetic patients and 14 healthy individuals provided responses regarding their kidney health. Healthy individuals reported no kidney complications, while 15 out of 30 diabetic individuals reported kidney issues. When analyzing the strain of RBCs in these two groups, we utilized the IQR to mitigate the influence of extreme values and outliers (Fig. S5†). The average strain values of individuals with kidney complications displayed greater dispersion compared to the group without kidney issues, indicating wider variability and spread in strain values.

Median values for both groups showed that individuals without kidney complications had a higher average strain value, ∼0.37, compared to the group with kidney complications, which averaged ∼0.33 (Fig. S5a†). This suggests that RBCs in individuals without kidney complications are ∼11% more flexible than those in individuals with kidney complications. Regarding the RBC cell size in both groups, the data in Fig. S5b† demonstrated greater variability and spread in cell diameter among healthy kidneys. The group without kidney complications had a higher average diameter, ∼9.1 μm, compared to the group with kidney complications, with an average diameter of ∼8.6 μm. The percent difference in diameter between these groups was relatively small, ∼5.5%. Therefore, the kidney status did not appear to significantly affect the RBC size. Thus, our analysis of strain and RBC diameter in patients with and without kidney complications revealed that kidney complications primarily impact the stretchability of cells, while having no substantial effect on the RBC size.

Glycemia status. All diabetic patients underwent HbA1c tests to assess their average blood sugar levels over a 2- to 3 month period. The participants were categorized into two subgroups based on their HbA1c levels: patients with controlled glycemia (<7%) and patients with uncontrolled glycemia (>7%). These classifications were made independent of the specific diabetes treatment or medications patients were receiving, whether it involved insulin injections or other blood sugar control medications. The comparison of the strain and stretch factor of RBCs from these two groups with those of healthy individuals revealed notable differences (Fig. 4a and d). Patients with HbA1c levels below 7%, indicative of better glycemic control, exhibited ∼15% higher strain than patients with HbA1c levels above 7%. Additionally, the strain and stretch factor of healthy RBCs were higher than those of diabetic RBCs.

Fig. 4 Scatterplots of individual patients from various groups and healthy volunteers show the impact of diabetes control and gender on the mechanical properties of RBCs. (a–f) Panels demonstrate the influence of diabetes control and gender on the RBC strain and stretch ratio. These visual representations contribute significantly to our understanding of how these factors affect the health of individuals with T2DM and play a crucial role in the development of effective interventions. Each data point represents an individual healthy/patient donor of total 15 healthy individuals and 31 diabetic patients.

In Fig. S6,† separate fitting lines are plotted to distinguish healthy RBCs from diabetic RBCs for both strain and stretch factor assessments. In the case of strain, the line accurately grouped ∼92% of healthy RBCs and ∼87% of diabetic RBCs (Fig. S6a†). Meanwhile, the analysis of the stretch factor revealed that all healthy RBCs were positioned above the line, while ∼94% of diabetic RBCs were below it (Fig. S6b†). This difference indicates the higher sensitivity of the stretch factor compared to strain in effectively distinguishing between healthy RBCs and diabetic RBCs. Nevertheless, these outcomes were consistent with the averaged fitting lines shown in Fig. 3c and d, confirming the sensitivity of the developed OMEF biochip.

Gender. Analysis of the scatter plots in Fig. 4b and e, and c and f, and did not indicate any significant association between the strain and stretch factor of RBCs and the gender of the participant. This finding suggests that the observed differences in these mechanical properties, particularly regarding RBC deformability, were not influenced by gender.

In medical research, it is essential to investigate potential gender-related differences in various health parameters such as sport conditioning, as there are often physiological differences between males and females.⁵³ However, in the context of RBC deformability and its relationship with T2DM, our findings suggest that the mechanical properties of RBCs, as assessed through strain and stretch factor measurements, are more likely influenced by the presence of diabetes rather than gender.

Family history. Our data revealed that 90% of diabetic patients had a family history of T2DM. This finding strongly suggests a significant association between genetic factors and the diabetic status of an individual. It underscores the importance of genetic factors and familial history in both the development and progression of T2DM.⁵⁴

Smoking habits. All diabetic patients, along with most of the healthy individuals, provided information about their smoking status, which was categorized into three groups: N (never smoked), E (ex-smoker), and S (current smoker). Analysis of the average RBC strain for each group and the average stretch factor did not reveal significant differences, indicating that the smoking status did not seem to have a discernible impact on diabetes-related changes in RBC mechanical properties. However, it is important to note that the sample size for smokers was small, and further investigations with larger sample sizes may be necessary to draw definitive conclusions regarding the influence of smoking on these mechanical properties in the context of T2DM.

Discussion

Over the years, several techniques have been employed to investigate the effect of T2DM on the mechanical properties of RBCs, as these properties serve as biomarkers for monitoring disease progression and treatment effectiveness. For example, Agrawal et al.²⁷ conducted a study using the dual optical tweezers stretching technique to investigate RBC deformability in T2DM patients. This method split and recombined a single laser beam to stretch a cell while recording the process on a microscope for measuring initial and final RBC lengths. It enabled the calculation of the deformability index, which indicated the individual cell deformation. The study revealed significantly lower deformability of RBCs in T2DM patients compared to healthy controls. However, the approach had limitations, such as low throughput and reliance on a complex setup. It was also sensitive to environmental conditions, including temperature and humidity. Conversely, Tan et al.⁵⁵ developed a technique that correlates RBC deformability with hematological and serum biochemical markers in T2DM patients. They used a microfluidic channel to align cells at the midline of the suspension. At the crossslot's stagnation point,⁴¹ they generated an extensional flow field to uniaxially stretch the cells and measure their deformability. Their data demonstrated significantly reduced RBC deformability in diabetic patients compared to non-diabetic individuals. Nevertheless, the technique was complex due to the multiscale nature of fluid dynamics, and there was limited control over the stretch parameters, which introduced uncertainties into the stretching process.⁵⁶

To address the need for a rapid and cost-effective method of studying RBC deformability in T2DM, this study introduces the OMEF biochip technology based on the DEP single cell stretching. The developed technology allows for high-throughput RBC stretching experiments, which enable the measurement and analysis of important mechanical properties of RBCs, including the cell size, strain, stretch factor, and relaxation time. This biochip technology presents a promising solution to overcome the limitations of existing techniques, providing a comprehensive understanding of how T2DM influences the deformability of RBCs.

Our findings underscore the significant impact of T2DM on RBC deformability. In comparison with their healthy counterparts, diabetic RBCs exhibited ∼27% increased size, ∼55% lower strain, and ∼29% reduced stretch factor. However, in an extended stretching experiment, diabetic RBCs exhibited plastic deformation, whereas healthy RBCs displayed elastic behavior. These findings highlight the potential of employing the size and deformability of RBCs as biomarkers for monitoring T2DM, especially using the OMEF biochip.

Importantly, our findings also revealed that gender does not significantly impact the observed differences in RBC deformability, emphasizing that these changes are primarily T2DM-related rather than gender-dependent. Both diabetic patient groups (i.e., patients with controlled glycemia, HbA1c < 7%, and patients with uncontrolled glycemia, HbA1c > 7%) showed higher HbA1c levels compared to healthy individuals. This suggests that HbA1c alters the mechanical properties of RBCs. Interestingly, the observed slower relaxation times of diabetic RBCs returning to their initial state after elongation for 25 s in Fig. 3a and b may be attributed to the complex interplay of various factors influencing the mechanical properties of diabetic RBCs. These factors could include altered membrane composition, cytoskeletal changes, or variations in the viscoelastic properties of diabetic RBCs. In contrast, the lack of significant differences in relaxation time between healthy and diabetic RBCs in Fig. 3e might be influenced by the relatively shorter stretching time of 5 s. Further investigation and detailed analysis, potentially exploring individual cell responses and variations, are essential to reveal the underlying mechanisms contributing to the observed paradoxical results. This, however, is left for future studies.

Clearly, by utilizing the health and demographic data presented in Fig. 5 and detailed in Table S1,† further investigation could be carried out to explore additional correlations related to the mechanical properties of RBCs. For example, parameters such as serum creatinine levels,⁵⁷ the presence of thalassemia and sickle cell anemia,⁵⁸ and information regarding physical activity⁵⁹ could offer valuable insights. However, to advance our understanding of the relationship between diabetes and RBC health comprehensively, future studies should include additional work that may impact these findings. For example, further validation of RBC mechanics as a biomarker for T2DM should be conducted with larger and more diverse patient populations to ensure the reliability and generalizability of our findings. Through simple modifications to the OMEF biochip, such as employing specialized AFM liquid cells that enclose sample and maintain physiological environment,⁶⁰ AFM-based imaging and force measurements can also play a significant role in verifying changes in the morphology and nanomechanical properties (e.g., elasticity and deformation) of diabetic RBCs at the single-cell level.

Fig. 5 Health and demographic analysis of T2DM patients. The cross-analysis data show the distribution of 31 T2DM patients with diverse health and demographic conditions. For a more detailed review, including information on the size, strain, stretch factor, and relaxation time of RBCs analyzed for each patient, refer to Table S1.†

Moreover, with the OMEF biochip, longitudinal studies can be conducted to track changes in RBC mechanics over time in individuals with T2DM to offer a more dynamic perspective on the disease progression, complications, and treatment outcomes. Investigating how interventions, such as lifestyle modifications, medications, and insulin therapy, impact RBC mechanics could provide insights into the mechanisms of treatment efficacy. Importantly, exploring molecular and cellular mechanisms underlying changes in RBC mechanics in T2DM and incorporating additional parameters such as inflammatory markers or insulin resistance to improve diagnostic accuracy and comprehensiveness may uncover novel therapeutic targets for the disease. It is also worth noting that investigating RBC deformability in prediabetic individuals would be essential, as early alterations could signify hyperglycaemia-related changes occurring before the onset of T2DM.⁶¹ This would provide insights into early markers for the disease and potential interventions to moderate its progression.

Employing deep learning techniques for RBC deformability assessment through image processing represents another highly promising future research study in the hemorheology field.⁶² This holds significant potential for various applications, particularly in automation and big data processing. Automation, facilitated by precise image analysis, promises enhanced efficiency in RBC deformability assessments. Additionally, the ability to process big data implies the potential for analyzing and integrating extensive datasets for comprehensive and in-depth investigations into RBC deformability.

The OMEF biochip technology not only holds promise for advancing our understanding of RBC mechanics in T2DM but also extends its potential applications to various disease research areas. This innovative technology can be employed to explore RBC mechanics in cardiovascular diseases, thus offering insights into cardiovascular health and conditions such as anemia.⁶³ Additionally, the OMEF biochip can study the deformability of cancer cells, thus contributing to cancer diagnosis and treatment monitoring.⁶⁴ Its versatility can extend further to investigating the mechanical properties of cells relevant to neurodegenerative disorders, infectious diseases, autoimmune conditions, and rare genetic disorders.⁶⁵ Moreover, the technology can also be used to assess the impact of pharmaceutical compounds on cell mechanics during drug development.⁶⁶

Conclusions

The developed OMEF biochip introduces a novel approach for investigating RBC mechanics in T2DM. Our findings underscore its significant potential as a point-of-care diagnostic tool for T2DM. The biochip demonstrates promising results, including distinct stretching values of RBCs, high sensitivity, efficient throughput, and rapid data acquisition. With its capability to process clinical samples of varying sizes, the OMEF biochip offers a versatile solution for practical application.

The device functions by capturing RBCs at a voltage of 3 V, with subsequent stretching achieved at 7 V for 5 s. This approach provides a quick and efficient means of assessing RBC mechanical characteristics. The streamlined process, from obtaining a blood sample to analysis and reporting, ensures a time-effective diagnostic procedure, typically concluding within a few minutes.

Future work will focus on refining and expanding the capabilities of the OMEF biochip, exploring additional diagnostic applications, and assessing its performance in larger clinical cohorts. The technology exhibits promising potential for advancing point-of-care diagnostics and contributing to our understanding of cellular mechanics in various pathological conditions.

Author contributions

A. M., H. S., and M. A. Q. designed the study. A. M. and M. A. Q. developed the OMEF biochip. D. S. A., S. O. S., and M. A. Q. developed the experimental procedures. D. S. A. and S. O. S. performed all the experiments. P. S. performed the blood handling experiments and staining of RBCs. D. S. A., M. D., and M. A. performed the data analysis. D. S. A., M. A., and M. D. prepared the figures. D. S. A., M. A., S. Y., and M. D. wrote the paper. C. R., R. G., and R. H. recruited patients/volunteers, obtained consent, and performed phlebotomy procedures. R. H. and H. S. provided clinical insights and analysis. M. A. Q. supervised the study and revised the paper.

Conflicts of interest

The authors report no conflicts of interest.

Acknowledgements

D. S. A. acknowledges the NYU Global PhD Fellowship. M. A. Q. acknowledges the financial support from the Seed Grant (AJF2018085) from Al Jalila Foundation, Dubai, UAE, and NYU Abu Dhabi Research Fund. We acknowledge the technical support from Core Technology Platforms at NYU Abu Dhabi. We thank Dr. Ayoub Glia for cross-analysis of T2DM patients' data. We also thank Mr. Vijay Dhanvi for the technical support.

References

1. M. A. B. Khan, M. J. Hashim, J. K. King, R. D. Govender, H. Mustafa and J. Al Kaabi, J. Epidemiol. Glob. Health, 2020, 10, 107–111.
2. S. Bellary, I. Kyrou, J. E. Brown and C. J. Bailey, Nat. Rev. Endocrinol., 2021, 17, 534–548.
3. U. Galicia-Garcia, A. Benito-Vicente, S. Jebari, A. Larrea-Sebal, H. Siddiqi, K. B. Uribe, H. Ostolaza and C. Martín, Int. J. Mol. Sci., 2020, 21, 6275.
4. E. Ahmad, S. Lim, R. Lamptey, D. R. Webb and M. J. Davies, Lancet, 2017, 400, 1803–1820.
5. T. R. Einarson, A. Acs, C. Ludwig and U. H. Panton, Cardiovasc. Diabetol., 2018, 17, 83.
6. Y. Q. Wang, P. Y. Yang, Z. L. Yan, Z. Liu, Q. Ma, Z. H. Zhang, Y. X. Wang and Y. Su, J. Diabetes Res., 2021, 2021, 6656062.
7. J. Pernow, A. Mahdi, J. N. Yang and Z. C. Zhou, Cardiovasc. Res., 2019, 115, 1596–1605.
8. R. Bissinger, T. Nemkov, A. D'Alessandro, M. Grau, T. Dietz, B. N. Bohnert, D. Essigke and M. Wörn, et al. , Kidney Int., 2021, 100, 1227–1239.
9. R. H. Hong, J. Zhu, Z. Z. Li, J. Yuan, P. Zhao, J. Ding, Q. L. Fan and J. Yang, et al. , Aging, 2020, 12, 9855–9867.
10. H. C. Cho, Diabetes Metab. J., 2011, 35, 595–601.
11. C. Turpin, A. Catan, O. Meilhac, E. Bourdon, F. Canonne-Hergaux and P. Rondeau, Int. J. Mol. Sci., 2021, 22, 5843.
12. V. Pretini, M. H. Koenen, L. Kaestner, M. H. A. M. Fens, R. M. Schiffelers, M. Bartels and R. Van Wijk, Front. physiol., 2019, 10, 945.
13. R. T. Yen and Y. C. Fung, Am. J. Physiol., 1978, 235, 251–257.
14. I. V. Pivkin, Z. L. Peng, G. E. Karniadakis, P. A. Buffet, M. Dao and S. Suresh, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 7804–7809.
15. N. Mohandas, M. R. Clark, M. S. Jacobs and S. B. Shohet, J. Clin. Invest., 1980, 66, 563–573.
16. S. Chien, S. Usami, R. J. Dellenback and M. I. Gregersen, Science, 1967, 157, 827–829.
17. S. Chien, Scand. J. Clin. Lab. Invest., 1981, 41, 7–12.
18. W. H. Reinhart and S. Chien, Blood, 1986, 67, 1110–1118.
19. S. I. Sherwani, H. A. Khan, A. Ekhzaimy, A. Masood and M. K. Sakharkar, Biomarker Insights, 2016, 11, 95–104.
20. I. Juhan, P. Vague, M. Buonocore, J. P. Moulin, R. Jouve and B. Vialettes, Lancet, 1982, 1, 535–537.
21. C. Watala, M. Kordacka, A. Loba, Z. Jozwiak and S. Nowak, Acta Diabetol. Lat., 1987, 24, 141–148.
22. Q. Li and L. Z. Yang, Acta Endocrinol., 2018, 14, 66–75.
23. G. Tomaiuolo, Biomicrofluidics, 2014, 8, 051501.
24. N. Mohandas and E. Evans, Annu. Rev. Biophys. Biomol. Struct., 1994, 23, 787–818.
25. S. Himbert and M. C. Rheinstädter, Front. physiol., 2022, 13, 953257.
26. F. Pan, S. X. Chen, Y. Jiao, Z. Y. Guan, A. Shakoor and D. Sun, IEEE Robot. Autom. Lett., 2020, 5, 1167–1174.
27. R. Agrawal, T. Smart, J. Nobre-Cardoso, C. Richards, R. Bhatnagar, A. Tufail, D. Shima, P. H. Jones and C. Pavesio, Sci. Rep., 2016, 6, 15873.
28. A. Loyola-Leyva, J. P. Loyola-Rodríguez, M. Atzori and F. J. González, Micron, 2018, 105, 11–17.
29. C. Trejo-Soto, G. R. Lázaro, I. Pagonabarraga and A. Hernández-Machado, Membranes, 2022, 12, 217.
30. C. Qian, H. B. Huang, L. G. Chen, X. P. Li, Z. B. Ge, T. Chen, Z. Yang and L. N. Sun, Int. J. Mol. Sci., 2014, 15, 18281–18309.
31. B. Sarno, D. Heineck, M. J. Heller and S. D. Ibsen, Electrophoresis, 2021, 42, 539–564.
32. Y. Nakashima, S. Hata and T. Yasuda, Sens. Actuators, B, 2010, 145, 561–569.
33. I. Guido, M. S. Jaeger and C. Duschl, Eur. Biophys. J., 2011, 40, 281–288.
34. I. Doh, W. C. Lee, Y. H. Cho, A. P. Pisano and F. A. Kuypers, Appl. Phys. Lett., 2012, 100, 173702.
35. Y. H. Qiang, J. Liu, M. Dao, S. Suresh and E. Du, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 19828–19834.
36. H. Qianqian, W. Zirui, S. Lingxiao and Z. Gang, Anal. Chem., 2022, 94, 10245–10255.
37. R. Pethig, in Dielectrophoresis: Theory, methodology and biological applications, ed. R. Pethig, John Wiley & Sons, Online book, 1st edn, 2017, ch. 6, pp. 119–144.
38. F. Bordi, C. Cametti and T. Gili, J. Non-Cryst. Solids, 2002, 305, 278–284.
39. A. Menachery, J. Sapudom, A. Vembadi, A. Alatoom, J. Teo and M. A. Qasaimeh, BioTechniques, 2021, 70, 29–36.
40. H. Morgan and N. G. Green, AC Electrokinetics: Colloids and Nanoparticles, Research Studies Press, 2003.
41. A. Menachery, C. Kremer, P. E. Wong, A. Carlsson, S. L. Neale, M. P. Barrett and J. M. Cooper, Sci. Rep., 2012, 2, 775.
42. S. Brunskill, S. Thomas, E. Whitmore, C. P. McDonald, C. Dorée, S. Hopewell, J. Staves, R. Cardigan and M. F. Murphy, Transfus. Med. Rev., 2012, 26, 209–223.
43. L. Guo and S. P. DeWeerth, Small, 2010, 6, 2847–2852.
44. J. L. Griffin and C. D. Ferris, Nature, 1970, 226, 152–154.
45. Z. S. Xu, Y. Zheng, X. Wang, N. Shehata, C. Wang and Y. Sun, Microsyst. Nanoeng., 2018, 4, 17103.
46. D. Bratosin, L. Mitrofan, C. Palii, J. Estaquier and J. Montreuil, Cytometry, Part A, 2005, 66, 78–84.
47. B. N. Alamri, A. Bahabri, A. A. Aldereihim, M. Alabduljabbar, M. M. Alsubaie, D. Alnaqeb, E. Almogbel and N. S. Metias, et al. , Riv. Eur. Sci. Med. Farmacol., 2019, 23, 2139–2150.
48. D. J. Celentano, E. E. Cabezas, C. M. García and A. E. Monsalve, Modell. Simul. Mater. Sci. Eng., 2004, 12, S425.
49. M. S. AlSalhi, S. Devanesan, K. E. AlZahrani, M. AlShebly, F. Al-Qahtani, K. Farhat and V. Masilamani, Int. J. Environ. Res. Public Health, 2018, 15, 2368.
50. D. Kabaso, R. Shlomovitz, T. Auth, V. L. Lew and N. S. Gov, Biophys. J., 2010, 99, 808–816.
51. Z. H. Huang, Y. J. Liu, Y. F. Mao, W. W. Chen, Z. G. Xiao and Y. Y. Yu, J. Breath Res., 2018, 12, aa9081.
52. B. Hardesty, R. Miller and R. Schweet, Proc. Natl. Acad. Sci. U. S. A., 1963, 50, 924–931.
53. S. Ivan, O. Daniela and B. D. Jaroslava, Physiol. Behav., 2023, 259, 114038.
54. M. Mambiya, M. K. Shang, Y. Wang, Q. Li, S. Liu, L. P. Yang, Q. Zhang and K. L. Zhang, et al. , Front. Public Health, 2019, 7, 00349.
55. J. K. S. Tan, X. Wei, P. A. Wong, J. Fang, S. Kim and R. Agrawal, Microcirculation, 2020, 27, e12649.
56. A. T. Brimmo and M. A. Qasaimeh, RSC Adv., 2017, 7, 51206–51232.
57. N. Harita, T. Hayashi, K. K. Sato, Y. Nakamura, T. Yoneda, G. Endo and H. Kambe, Diabetes Care, 2009, 32, 424–426.
58. S. M. Kosecki, P. T. Rodgers and M. B. Adams, Ann. Pharmacother., 2005, 39, 1557–1560.
59. D. Aune, T. Norat, M. Leitzmann, S. Tonstad and L. J. Vatten, Eur. J. Epidemiol., 2015, 30, 529–542.
60. M. Deliorman, F. K. Janahi, P. Sukumar, A. Glia, R. Alnemari, S. Fadl, W. Chen and M. A. Qasaimeh, Microsyst. Nanoeng., 2020, 6, 20.
61. S. Z. Jiang, J. L. Young, K. Wang, Y. Qian and L. Cai, Mol. Med. Rep., 2020, 22, 603–611.
62. M. H. Liang, J. W. Zhong, C. S. Shannon, R. Agrawal and Y. Ai, Sens. Actuators, B, 2024, 401, 135056.
63. P. K. Barrera-Reyes and M. E. Tejero, Ann. N. Y. Acad. Sci., 2019, 1450, 32–46.
64. J. Chen, J. Li and Y. Sun, Lab Chip, 2012, 12, 1753–1767.
65. M. Gutierrez-Arcelus, S. S. Rich and S. Raychaudhuri, Nat. Rev. Genet., 2016, 17, 160–174.
66. R. Krishnan, J. A. Park, C. Y. Seow, P. V. S. Lee and A. G. Stewart, Trends Pharmacol. Sci., 2016, 37, 87–100.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3lc01016c

---