Ali K.
Yetisen§
*ab,
Nan
Jiang§
abc,
Ali
Tamayol
ab,
Guillermo U.
Ruiz-Esparza
ab,
Yu Shrike
Zhang
ab,
Sofía
Medina-Pando
a,
Aditi
Gupta
b,
James S.
Wolffsohn
d,
Haider
Butt
e,
Ali
Khademhosseini
abfg and
Seok-Hyun
Yun
*bh
aBiomaterials Innovation Research Center, Division of Biomedical Engineering, Brigham and Women's Hospital, Harvard Medical School, Cambridge, Massachusetts 02139, USA. E-mail: akyetisen@gmail.com
bHarvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. E-mail: syun@mgh.harvard.edu
cState Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, 122 Luoshi Road, Wuhan, 430070, China
dOphthalmic Research Group, School of Life and Health Sciences, Aston University, Aston Triangle, Birmingham B4 7ET, UK
eNanotechnology Laboratory, School of Engineering, University of Birmingham, Birmingham B15 2TT, UK
fDepartment of Physics, King Abdulaziz University, Jeddah, 21589, Saudi Arabia
gDepartment of Bioindustrial Technologies, College of Animal Bioscience and Technology, Konkuk University, Hwayang-dong, Gwangjin-gu, Seoul 143-701, Republic of Korea
hHarvard Medical School and Wellman Center for Photomedicine, Massachusetts General Hospital, 65 Landsdowne Street, Cambridge, Massachusetts 02139, USA
First published on 16th February 2017
The analysis of tear constituents at point-of-care settings has a potential for early diagnosis of ocular disorders such as dry eye disease, low-cost screening, and surveillance of at-risk subjects. However, current minimally-invasive rapid tear analysis systems for point-of-care settings have been limited to assessment of osmolarity or inflammatory markers and cannot differentiate between dry eye subclassifications. Here, we demonstrate a portable microfluidic system that allows quantitative analysis of electrolytes in the tear fluid that is suited for point-of-care settings. The microfluidic system consists of a capillary tube for sample collection, a reservoir for sample dilution, and a paper-based microfluidic device for electrolyte analysis. The sensing regions are functionalized with fluorescent crown ethers, o-acetanisidide, and seminaphtorhodafluor that are sensitive to mono- and divalent electrolytes, and their fluorescence outputs are measured with a smartphone readout device. The measured sensitivity values of Na+, K+, Ca2+ ions and pH in artificial tear fluid were matched with the known ion concentrations within the physiological range. The microfluidic system was tested with samples having different ionic concentrations, demonstrating the feasibility for the detection of early-stage dry eye, differential diagnosis of dry eye sub-types, and their severity staging.
The development of diagnostic devices for dry eye syndrome dates back to the 1900s. Otto Schirmer developed a semi-quantitative test for measuring tear volume on the ocular surface.19 Schirmer's test can be used without anesthetics for the measurement of reflex tear secretion in response to conjunctival stimulation.19 The pH-sensitive phenol red thread test is another semi-quantitative measurement device for detecting dry eye syndrome.20 The development of rapid diagnostic devices for analyzing tear fluid has been limited over the last two decades. An important development was a clinically-used benchtop osmometer (TearLab, San Diego, CA, USA).21 This device measures the conductivity in tear fluid (50 μL) and correlates the measurement with an osmolarity value, providing a quantitative readout to diagnose dry eye syndrome. However, clinical interpretation of osmolarity readings for the diagnosis of dry eye has been questioned.22 Lateral-flow assays have been also utilized for detecting analytes in tear fluid. In 2013, the Food and Drug Administration (FDA) approved a lateral-flow diagnostic test (InflammaDry, Rapid Pathogen Screening Inc., Sarasota, FL, USA) that measures the concentration of matrix metalloproteinase-9 (MMP-9) for the diagnosis of dry eye.23,24 The concentration of MMP-9 has been shown to be elevated in the tears of patients with dry eye disease.25 However, this assay requires multiple sample processing steps and gives a binary response limiting its value in determining severity. Additionally, Tearscan (Advanced Tear Diagnostics, Birmingham, AL, USA) is a lateral-flow assay that has a dynamic range from 0.25–2.50 mg mL−1 of lactoferrin which is an indicator of aqueous tear production.26 This test is potentially useful when combined with IgE (allergen testing) measurement.27 Recently, an alkaline microfluidic homogeneous immunoassay was demonstrated for the determination of the low-volume (<1 μL of tear) lactoferrin at clinically relevant concentrations.28 Another recent study utilized an inkjet-printed device for the measurement of lactoferrin concentration in tear fluid.29
The measurement of tear electrolytes can be used to identify dry eye at different severity stages and differentiate its sub-types such as MGD and LGD. The Na+ ion concentration in the tears of healthy individuals ranges between 120–165 mmol L−1.30 However, in dry eye syndrome caused by MGD or LGD, Na+ ion concentration in human tear increase was reported to be significant, which can be detected by a sensor sensitivity of ∼3.0 mmol L−1.31 Additionally, divalent metal ion concentration was found to be different in MGD and LGD. As compared to MGD, human tear Ca2+ ion concentration in LGD significantly increases and could be potentially detected by a sensor sensitivity of 20–40 μmol L−1.31 Hence, the accurate measurement of tear electrolytes can provide quantitative data for dry eye diagnosis. Therefore, the development of a minimally-invasive diagnostic system for the quantitative analysis of tear electrolytes is highly desirable for the detection of dry eye in different severity stages and classification of its subtypes. Such a quantitative device can be used to screen for at-risk dry eye patients, enable the early detection of dry eye, and improve management approaches.
This article reports on the development of a paper-based microfluidic system for application in the quantitative analysis of electrolytes in tear fluid (Fig. 1). Upon introducing a low-volume sample (e.g., tear fluid), the microfluidic device distributed the sample into sensing regions that were functionalized with fluorescent sensing agents. The microfluidic device was placed in a portable readout device that consisted of various LED illumination wavelengths for fluorescence excitation. A smartphone application (app) was utilized to capture the fluorescence assay images which were digitally processed to obtain concentration values for the quantitative analysis of electrolytes in artificial tear fluid.
Fig. 1 Principle of operation of the paper-based microfluidic system for the quantitative analysis of electrolytes in tear film. |
To design a paper-based microfluidic device that can operate with tear fluid samples (<10 μL) within a 3 min wicking time (point-of-care application), the geometry of the paper-based microfluidic device was optimized. Upon introducing DI water from their inlets at a constant width (2 mm), the fluid front in G1 strips reached 2 cm within 25 s (Fig. S3‡). As DI water (20 μL) was introduced to G1 strips with different widths at a constant length (3.5 cm), the wicking distances of strips with widths from 2–5 mm were comparable (Fig. 2a). Fig. S4‡ illustrates photographs of the G1 strips with varying channel widths. The G1 strips with 2 mm in width were chosen for assay optimization as it operated at low volume and it was easy to handle. The flow characteristics of fluid were fit using a modified Navier–Stokes equation (eqn (1)).
(1) |
The viscosity of human tear fluid ranges from 1 to 10 mPa s.35 The G1 matrix should have fast wicking time for application in rapid diagnostics. The viscosities of the solutions (10 μL) introduced to the G1 strips were varied from 1.0 to 10.0 mPa s. While a wicking time of 1 min saturated the fluid front at 1.0 mPa s, fluids with 10.0 mPa s required 3 min for fluid front saturation (Fig. 2b). The optimized G1 strip had fast sample wicking times at viscosities as high as 10.0 mPa s.
Fig. S5‡ illustrates the photographs of G1 microfluidic strips wicking fluids with different viscosities (1.0–10.0 mPa s). The effect of having branched channels on the movement of the fluid front was also tested. As the number of branches was increased from 1 to 4 at a constant main strip width of 2 mm, wicking distances among these strips were comparable within 20 s while and decreased by 14% with increasing number of branches within 20 to 60 s (Fig. 2c). Fig. 2d shows the fabricated paper-based microfluidic devices with different number of branches. The G1 matrix consists of cellulose with porous structure having a particle retention value of 11 μm (Fig. 2e). Fig. 2f shows the photographs of the paper-based microfluidic devices as the Rhodamine B solution (10 mmol L−1) was wicked from the main channel. The optimized paper-based microfluidic device with four branches allowed the transport of fluid samples (<20 μL) from the inlet to the branches under 1.5 min (Fig. S6‡).
Assay conditions of fluorescent chelating agents were optimized for functionalizing paper-based microfluidic devices. Crown ethers are cyclic chelating agents that are specific to monovalent metal ions.36 They form stable complexes with cations by ion–dipole interaction between a metal ion and negatively charged oxygen atoms in the polyether ring.37 The changes in fluorescence intensity on monovalent metal ion binding are caused by conformational or electronic changes that possibly occur in electron transfer between ground state and excited state of fluorophore due to electron density changes at the ion binding site.38,39 Fluorescent diaza-15-crown-5 (cavity size: 0.17–0.22 nm) and diaza-18-crown-6 (cavity size: 0.26–0.32 nm) were utilized for selectively sensing Na+ and K+ ions (Fig. 3a). In the presence of diaza-15-crown-5 (25 μmol L−1) in buffer solution (Tris, pH 7.4, 150 mmol L−1) at 24 °C, Na+ ions (100 mmol L−1) had the highest fluorescence intensity due to the cavity specific 1:1 chelation (Fig. 3b). The diaza-15-crown-5 response to Na+ ions was 3.5 fold higher as compared to K+ ions. In the presence of diaza-18-crown-6, K+ ions showed the highest selectivity (Fig. 3b). The diaza-18-crown-6 selectivity to K+ ions was 1.8 fold higher than Na+ ions. The interference due to Na+ ions could be due to 2:1 complexation with diaza-18-crown-6 cavity. To analyze ion binding affinity, the dissociation constants (Kd, the ion concentration at which 50% of crown ethers (receptors) are chelated by ions) of Na+ and K+ binding to fluorescent crown ether derivatives were determined as 14 mmol L−1 and 5.8 mmol L−1 respectively in Tris buffer solution (150 mmol L−1, pH = 7.4):
(2) |
Increasing pH from 7.0 to 8.0 enhanced the fluorescence intensities 0.2% and 2.6% for diaza-15-crown-5 and diaza-18-crown-6 respectively, showing stability within the physiological pH range (∼7.4) of tear fluid (Fig. 3c). The fluorescence intensities of both diaza-15-crown-5 and diaza-18-crown-6 depended on their concentrations. As their concentrations increased from 3 μmol L−1 to 50 μmol L−1 in buffer solutions (Tris-buffer, pH 7.4, 150 mmol L−1), their fluorescence intensities increased 10 and 13 fold, respectively (Fig. S7‡). When the temperature increased from 25 to 40 °C, the fluorescence intensity of diaza-15-crown-5 decreased ∼20% (Fig. S8‡). However, the diaza-18-crown-6 fluorescence intensity variation was ∼2% between 25 to 35 °C, decreasing ∼9% at 40 °C (Fig. S8‡).
The concentration of Na+ ions in tear fluid of healthy individuals is 120–165 mmol L−1.30 In dry eye syndrome caused by MGD and LGD, tear Na+ ion concentration increases by 2.2% and 6.8%, respectively.31 This requires a sensor sensitivity of ∼3.0 mmol L−1 and ∼9.0 mmol L−1. The presence of both MGD and LGD increases the Na+ ion concentration by 8.9%, requiring a sensitivity of ∼12.0 mmol L−1.40 G1 paper (32 mm2) was used as a reaction matrix to quantify concentrations of electrolytes. Probe solutions (2 μL, 25 μmol L−1 in DMSO) were immobilized on the G1 matrix, followed by adding ion solutions (2 μL, Tris-buffer, pH 7.4, 150 mmol L−1) to the G1 matrix. The fluorescent probes dissolved DMSO (2 μL) were inoculated on G1 paper, which was dried at 24 °C for 2 min (Fig. S9‡). As the concentration of Na+ ions in the presence of diaza-15-crown-5 on G1 matrix increased from ion-free buffer solution (Tris, pH 7.4, 150 mmol L−1) to 200 mmol L−1, the fluorescence intensity of the probe increased by 1.9 fold (Fig. 3d inset). To detect Na+ ions within the physiological concentration range of tear fluid (100–200 mmol L−1), the samples were diluted 16 fold (4 serial two-fold dilutions). After dilution, the fluorescence intensity of probes within the physiological Na+ ion range (100–200 mmol L−1) increased 13.3% (Fig. 3d and S10a‡). The sensitivity of the diluted Na+ ion sensor on G1 matrix was calculated to be 1.5 mmol L−1, which met the requirement for the diagnosis of dry eye. Sensitivity values from three/six independent measurements were calculated by averaging the standard error of the intensity ratio (I/I0) on the slope within the physiological disease detection range, followed by reading the corresponding electrolyte concentration values (mmol L−1) in the x-axis (ESI‡ Fig. S11).
The concentration of K+ ions in tear fluid of healthy individuals is 20–42 mmol L−1.30 However, in dry eye syndrome caused by MGD and LGD, tear K+ ion concentration increases by 2.5% and 3.8%, respectively.40 This requires a sensor sensitivity of ∼0.6 mmol L−1 and ∼0.9 mmol L−1. The presence of both MGD and LGD increases the K+ ion concentration by 5.8%, requiring a sensitivity of ∼1.4 mmol L−1. As the concentration of K+ ions increased from ion-free to 100 mmol L−1 at 24 °C, fluorescence intensity of diaza-18-crown-6 (25 μmol L−1) increased 97.7% (Fig. 3e inset). To detect K+ ions within the physiological range of tear fluid (20–50 mmol L−1), the samples were diluted 16 fold. After dilution, the fluorescence intensity of the probe within the physiological range increased 28.3% (Fig. 3e and S10b‡). The sensitivity of K+ ion sensor was 0.9 mmol L−1 which met the requirement for the diagnosis of dry eye.
To sense divalent metal ions, o-acetanisidide was utilized, where the N-(2-methoxyphenyl)iminodiacetate served as a generic chelation site (Fig. 4a). Among mono/divalent ions (1–100 mmol L−1) in the presence of o-acetanisidide (25 μmol L−1, Tris buffered, pH 7.4, 150 mmol L−1) at 24 °C, Ca2+ ions (100 mmol L−1) had the highest fluorescence intensity (1.2 fold higher than Ni2+ ions and 2.1 fold higher than Mg2+ ions), due to the site specific 1:1 chelation (Fig. 4b). The dissociation constant of Ca2+ ions was calculated to be 0.9 mmol L−1 (eqn (2)). The chelation of divalent metal ions depended on the pH; increasing the pH value from 5.5 to 8.0 enhanced the fluorescence intensity 2.6 and 1.5 fold for o-acetanisidide in the presence of Mg2+ and Ca2+ ions, respectively, showing stability within the physiological pH range (∼7.4) of tear fluid (Fig. 4c). As the concentration of o-acetanisidide was increased from 3 μmol L−1 to 50 μmol L−1, the fluorescence intensities for Mg2+ and Ca2+ ions increased 9.9 and 12.3 fold, respectively (Fig. 4d). The fluorescence intensity was affected by temperature; for example, when temperature increased from 25 °C to 40 °C, the fluorescence intensity of o-acetanisidide decreased ∼25% (Fig. S12a‡).
The concentration of Ca2+ ions in tear fluid of healthy individuals is 0.4–1.1 mmol L−1.42 However, in dry eye syndrome caused by MGD or LGD, tear Ca2+ ion concentration increases 2.5% and 5.0%, respectively. This requires a sensor sensitivity of 0.02–0.04 mmol L−1. The presence of both MGD or LGD increases the Ca2+ ion concentration 7.5%, requiring a sensitivity of ∼0.06 mmol L−1.40 As the concentration of Ca2+ ions increased from 0.25 mmol L−1 to 1.5 mmol L−1 at 24 °C, fluorescence intensity increased 3 fold in the presence of o-acetanisidide (25 μmol L−1) (Fig. 4e inset). The high sensitivity range of the fluorescent o-acetanisidide is 0.25–1.5 mmol L−1. To detect Ca2+ ions within the physiological concentration range of tear fluid, the sample does not need to be diluted. Even after 16-fold dilution, the fluorescence intensity of Ca2+ ion solution (0.25 mmol L−1 to 1.50 mmol L−1) increased 47% on the G1 matrix (Fig. 4e and S10c‡). The sensitivity of the Ca2+ ion sensor was calculated to be 0.03 mmol L−1 which met the requirements of dry eye diagnostic sensitivity.
pH changes can be quantified using seminaphtorhodafluor (pKa value ∼7.5), the fluorescence emission shifts from yellow-orange (λ = 580 nm) to deep red (λ = 640 nm) under acidic and basic conditions, respectively (Fig. 5a). As the concentration of seminaphtorhodafluor was increased from 3 μmol L−1 to 50 μmol L−1 at a constant pH value (7.4), the fluorescence intensity increased 15 fold (Fig. 5b). The tear fluid pH of a healthy individual is ∼7.4; however, in dry eye (MGD and LGD) the pH increases to ∼7.9.41 Changes in the composition and/or concentration of mucin secreted by the goblet cells increase the pH of the overlying aqueous layer.41 This requires a sensor sensitivity of ∼0.5 pH units. The seminaphtorhodafluor on G1 matrix exhibited a fluorescence intensity decrease of 2.9 fold as the pH increased from 7.0 to 9.0 (Fig. 5c inset). The sensitivity of seminaphtorhodafluor was 0.06 pH units after a 16-fold dilution, which met the requirement of the sensor sensitivity (Fig. 5c and S10d‡). Seminaphtorhodafluor also showed low interference in the presence of mono/divalent ions in solution (Fig. 5d). Additionally, the fluorescence intensity of seminaphtorhodafluor decreased ∼36% when the temperature increased from 25 °C to 40 °C (Fig. S12b‡).
To investigate potential ion interference in the quantification of electrolyte concentrations, the 16-fold diluted solutions containing two or more ions both in solution and on the G1 matrix were analyzed (Fig. S13‡). Fluorescence intensity changes of diaza-15-crown-5 for Na+ ion (30–180 mmol L−1) sensing in the presence of K+ (42 mmol L−1), Ca2+ (1.1 mmol L−1), and Mg2+ (0.4 mmol L−1) ions were evaluated. The maximum deviation for the interference of K+ ions in Na+ ion sensing was 2.4% in solution and 4.0% on the G1 matrix. For Na+ ion sensing interfered by K+ and Ca2+ ions, the deviation was 5.0% in solution and 2.8% on the G1 matrix. Additionally, in the presence of K+, Ca2+ and Mg2+ ions, Na+ ion measurement interference was 2.7% in solution and 4.7% on the G1 matrix. In the physiological range of human tear fluid (120–180 mmol L−1), the deviations of Na+ ion measurements were less than 1.4%. The deviations for Na+ ion sensing from the results of fluorescent crown ether derivatives were within the accuracy limit of target selectivities. Moreover, ion sensing in artificial tear fluid was evaluated and compared with electrolyte solutions in buffers (Fig. S14‡). Artificial tear fluid was prepared to mimic tear fluid composition. The maximum deviations for Na+ ions (0–200 mmol L−1), K+ ions (0–50 mmol L−1), Ca2+ ions (0–2 mmol L−1), pH (7.0–9.0) sensing on the G1 matrix were 5%, 5%, 6% and 5%, respectively, which were within the accuracy of target electrolyte sensing. Therefore, these fluorescent sensors can be used for ion sensing in artificial tear fluid on the G1 matrix.
To demonstrate the utility of the paper-based microfluidics for tear analysis, a microfluidic system including a sample collection device and a portable readout device was developed. 2 μL of each fluorescent sensor (Na+, K+, Ca2+ ions and pH sensors) was dispensed onto the tip of each branch of the microfluidic device (Fig. 6a and movie S1‡). The sample collection device was designed to be amenable to potential clinical use consisting of a dilution reservoir (∼75 μL DI water) connected to a capillary tube, which can sample tears (∼5 μL) (Fig. S15‡). The diluted sample was mixed thoroughly (30 s) and introduced to the paper-based microfluidic device which was connected to the opposite side of the reservoir (Fig. 6b and movie S2‡). A portable readout device was developed for blocking the ambient light and exciting the fluorescent probes impregnated into the branches of the paper-based microfluidic device (Fig. 6c). Four LEDs with different emission wavelengths (λem: 366, 460, 505, and 515 nm) illuminated the sensing regions from the rear. The four-channel paper-based microfluidic device was placed in a groove covered with a longpass filter, which was located in the interlayer of the readout device. Fig. S16‡ shows light attenuation of each LED light using the longpass filters (420, 495, 515, and 590 nm). The fluorescence images of probes at different artificial tear fluid compositions on paper-based microfluidic device were captured by an iPhone 6S camera positioned over a wide-angle lens in the readout device using a smartphone app (Shoot) (Fig. 6d–f and S17‡). A square (1 × 1 mm2) in the central of the captured fluorescence image was selected from the sensing region (2 × 2 mm2) at the end of each branch of the paper-based microfluidic device, which was used for image processing (Fig. 6f inset). Movie S3‡ shows the operation of the readout device for sample measurements.
The quantification of electrolyte concentrations in the artificial tear fluid was carried out by a portable readout device integrated with a smartphone camera. The fluorescence images were captured using a smartphone app and quantitatively analyzed using ImageJ. A square (1 × 1 mm2) in the central of the captured fluorescence image was selected from the sensing regions (2 × 2 mm2) at the end of each branch of the paper device for signal processing. The concentration-dependent fluorescence intensity ratio can be expressed as:
(3) |
We investigated the sensitivity of electrolyte sensing based on calibration curves (Fig. 7 and S18‡). The calibration data was compiled by subtracting the background (paper without fluorescent probe) (eqn S1‡). Increase in Na+ ion concentration from 100 mmol L−1 to 200 mmol L−1 in artificial tear fluid within the physiological range increased the fluorescence intensity of diaza-15-crown-5 by 22.4% on the paper-based microfluidic device (Fig. 7a). The sensitivity of diaza-15-crown-5 sensor on the microfluidic device was 2.7 mmol L−1, which met the requirement for Na+ ion sensing in dry eye diagnosis (∼3.0 mmol L−1). Fig. S19‡ shows the reproducibility of the sample measurement process using the sample collection device (Fig. 6b) that performed 2 to 64 fold dilutions. The average measurement error due to sample dilution in Na+ ions was 1.3 mmol L−1 (Fig. S19a‡). As the concentration of K+ ions was increased within the physiological range in artificial tear fluid (20 mmol L−1 to 50 mmol L−1), the fluorescence intensity of diaza-18-crown-6 sensor on the microfluidic device increased by 26.6% (Fig. 7b). The sensitivity of diaza-18-crown-6 sensor was 1.4 mmol L−1, which met the requirement for K+ sensing in dry eye diagnosis (∼1.4 mmol L−1). The average measurement error due to sample dilution in K+ ions was 0.8 mmol L−1 (Fig. S19b‡). Additionally, as the concentration of Ca2+ ions in artificial tear fluid was increased from 0.5 mmol L−1 to 2.0 mmol L−1, the fluorescence intensity of o-acetanisidide sensor on the microfluidic device increased 80.4% (Fig. 7c). The sensitivity of o-acetanisidide was 0.02 mmol L−1, which met the requirement for Ca2+ sensing (0.02–0.04 mmol L−1) in dry eye diagnosis. The average measurement error due to sample dilution in Ca2+ ions was 0.02 mmol L−1 (Fig. S19c‡). An increase in pH value from 7.0 to 8.0 in artificial tear fluid decreased the fluorescence intensity of seminaphtorhodafluor by 18.9% on the microfluidic device. The average measurement error due to sample dilution was 0.1 pH values (Fig. S19d‡). The sensitivity was calculated to be 0.06 pH units, which met the requirement for pH sensing in dry eye diagnosis (∼0.5 pH unit) (Fig. 7d). The fluorescence intensity measurements of paper-based microfluidic device using the readout system integrated with the smartphone app and ImageJ were consistent with the results from the microplate reader. The paper-based microfluidic devices remained exposed to air during measurements. We performed experiments to measure the effect of evaporation on the fluorescence intensity readouts in paper strips with different lengths (4, 8, 16, and 32 mm) during the wicking process. The evaporation from the electrolyte solutions during wicking process on paper device did not have significant effect on the fluorescence readouts (Fig. S20‡). The average standard errors due to evaporation among different strips were 0.08, 0.07, 0.02 mmol L−1, and 0.07 pH units for Na+, K+, Ca2+, and pH measurements, respectively. Additionally, batch-to-batch measurements of electrolytes in the paper-based microsystem showed that the average detection errors were 1.0 mmol L−1 (Na+ ions), 1.3 mmol L−1 (K+ ions), 0.02 mmol L−1 (Ca2+ ions), and 0.13 pH, indicating high reproducibility in independent trials (Fig. S21‡).
Fig. 7 Quantifications of electrolytes in artificial tear fluid using the smartphone readout system: (a) Na+ ions, (b) K+ ions, (c) Ca2+ ions and (d) H+ ions sensing. Scale bars = 2 mm. Insets in (a) and (b) show Na+ ion concentration in the range of 130–150 mmol L−1 and K+ ion concentration in the range of 24–26 mmol L−1. Error bars represent standard error of the mean (n = 6). Curves (red dashes) were fitted using eqn (3). Shadows show the physiological Na+, K+, Ca2+ ion concentration and pH ranges. |
Sub-types of dry eye (MGD and LGD) were simulated in artificial tear fluid by varying the concentrations of Na+, K+, and Ca2+ ions (Table 1). Fig. 8a–c shows the variation of inferred ion concentration in artificial tear samples. Fig. S22‡ shows the measurements of pH values. Ion concentrations in artificial tear sample 1–3 (simulated dry eye samples) were higher than that in control (simulated healthy sample). The maximum deviation of Na+ ion sensor was calculated to be 3% according to the standard curve, which was within the accuracy for dry eye diagnosis (∼3%). Moreover, the maximum deviation of the fluorescent sensors for K+, Ca2+ ions and pH were 3%, 0.4%, and 4%, respectively, which were within the accuracy for dye eye diagnosis (∼7%). Additionally, total electrolyte concentration of human tear fluid is correlated with tear osmolarity, which increases with dry eye severity.42 Different severity stages of dry eye were simulated by varying the concentrations of Na+, K+, and Ca2+ ions (Table 1) on the paper-based microfluidic device. The ion concentrations and pH value were measured in the portable readout device. The images were captured using a smartphone app and analyzed by ImageJ. The maximum deviation of the sensor in all samples was 1.4%, which was within the accuracy for dry eye diagnosis (∼2%) (Fig. 8d).
Subtype-differentiation | ||||
---|---|---|---|---|
Ions | Control (mmol L−1) | Sample 1 (MGD) (mmol L−1) | Sample 2 (LGD) (mmol L−1) | Sample 3 (MGD + LGD) (mmol L−1) |
Na+ | 133.2 | 136.1 | 142.2 | 145.1 |
K+ | 24.0 | 24.6 | 24.9 | 25.4 |
Ca2+ | 0.80 | 0.82 | 0.84 | 0.86 |
Severity stages | ||||
---|---|---|---|---|
Ions | Normal (mmol L−1) | Mild stage (mmol L−1) | Moderate stage (mmol L−1) | Severe stage (mmol L−1) |
Na+ | 135 | 145 | 155 | 165 |
K+ | 24 | 25 | 26 | 27 |
Ca2+ | 0.80 | 0.85 | 0.90 | 0.95 |
Footnotes |
† We declare no competing financial interests. |
‡ Electronic supplementary information (ESI) available: Microscopic images of G1 paper and G41 paper under brightfield; optimization of CO2 laser radiation fluence and beam speed for ablating filter paper-G1; photographs of DI water diffusion in microfluidic channels with different lengths, different widths, different viscosities of fluid and different numbers of channels; fluorescence intensity readouts of Na+ and K+ ions with varied concentrations of fluorescent probes; effect of variations in temperature on fluorescence intensity; photographs of DMSO on G1 paper dried in the air; calibration curves of electrolyte sensing on G1 paper using microplate reader measurement; calculation of sensitivity of the fluorescent sensors based on International Union of Pure and Applied Chemistry (IUPAC) guidelines; quantification of ion interference in buffer solution and artificial tear fluid; light attenuation of LED lights using different optical filters; the design of the sample collection device and its potential clinical use; calibration curves of electrolyte sensors using the paper-based microfluidic system; quantifications of evaporation effect on sampling process; design of the sample collection device and its potential clinical use; batch-to-batch variation experiments; equation for background subtraction; movies of sample collection and measurements. See DOI: 10.1039/c6lc01450j |
§ These authors contributed equally. |
This journal is © The Royal Society of Chemistry 2017 |