Handmade Paper as a Paper Analytical Device for Determining the Quality of an Antidiabetic Drug.

Paper analytical devices (PADs) are a class of low-cost, portable, and easy-to-use platform for several analytical tests in clinical diagnostics, environmental pollution monitoring, and food and drug safety screening. These devices are primarily made from cellulosic paper. Considering the importance of eco-friendly and local or distributed manufacturing of devices realized during the COVID-19 pandemic, we systematically studied the potential of handmade Nepali paper to be used in fabricating PADs in this work. We characterized five different handmade papers made from locally available plant fibers using an eco-friendly method and used them to fabricate PADs for determining the drug quality. The thickness, grammage, and apparent density of the paper samples ranged from 198.6 to 314.8 μm, 49.1 to 117.8 g/m2, and 0.23 to 0.43 g/cm3, respectively. The moisture content, water filtration, and wicking speed ranged from 5.8 to 7.1%, 35.7 to 156.7, and 0.062 to 0.124 mms-1, respectively. Furthermore, the water contact angle and porosity ranged from 76.6 to 112.1° and 79 to 83%, respectively. The best paper sample (P5) was chosen to fabricate PADs for the determination of metformin, an antidiabetic drug. The metformin assay on PADs followed a linear range from 0.0625 to 0.5 mg/mL. The assay had a limit of detection and limit of quantitation of 0.05 and 0.18 mg/mL, respectively. The average amount of metformin concentration in samples collected from local pharmacies (n = 20) was 465.6 ± 15.1 mg/tablet. When compared with the spectrophotometric method, PAD assay correctly predicted the concentration of 90% samples. The PAD assay on handmade paper may provide a low-cost and easy-to-use system for screening the quality of drugs and other point-of-need applications.

Introduction

Paper-based analytical devices (PADs) are a class of low-cost, portable, and easy-to-use point-of-need assay platforms. Assays on PADs need significantly smaller volumes of reagents and samples, thus generating lower volumes of waste.[1] In recent years, various efforts have been made to improve the performance, design, and applicability of PADs by fabricating devices for multiplex assays,[1] three-dimensional devices,[2] fully enclosed paper devices,[3] programmable diagnostic devices,[4] and enzymatic biofuel cells.[5]

The PADs can be integrated with various methods of signal detection for assay analysis. The most commonly used methods are colorimetry and electrochemical detection.[6] In colorimetry, analysis is performed by adding a reagent(s) to the reaction zones within the paper device along with the analyte of interest.[7] Change in the color of the assay zone in the paper device is identified or measured visually or using a camera and scanner. This minimizes the need for expensive and sophisticated instrumentations and facilities.[8] Because of these advantages, several interesting applications of PADs have been demonstrated[7] in environmental analysis, clinical diagnosis, pharmaceutical analysis,[9] and chemical and biological testing using colorimetric assays such as for proteins,[10] glucose,[11] uric acid,[12] drugs,[9] and biomarkers.[1]

Several methods are available for the fabrication of (micro)fluidic channels or assay zones in PADs[1] such as photolithography,[13] plasma treatment and inkjet printing,[14] wax printing,[15] screen printing,[16] wax dipping,[17] flexographic printing,[18] and laser cutting.[19] Filter paper, blotting paper, and chromatography papers are among the most widely used paper substrates for fabricating PADs.[9,20] Paper is a low-cost and ubiquitous material with a wide range of choices. Whatman grade 1 chromatographic paper and Whatman no. 1 filter paper have been widely applied for the development of PADs.[20] These papers are made of cellulose (>98%). Whatman grade 1 chromatography paper has a clean surface, uniform thickness, high hygroscopic properties, wicking properties, flow rate, and cost effectiveness.[21] Some other papers include Whatman grade 4 chromatography paper that has a pore size of 20–25 μm[14] and nitrocellulose (NC) paper that has been used as a substrate for protein immobilization as it provides high protein-binding capacity due to charge–charge interactions and weak secondary forces.[22] Similarly, grade 3 chromatography paper,[23] Whatman P81,[24] paper towel,[21,25] and office paper[26] have been used as a suitable platform in the fabrication of paper-based sensors. Although various types of paper substrates are currently being used for the fabrication of PADs, researchers are still looking for paper substrates having unique properties or locally made or manufactured following eco-friendly methods. The need for local or distributed manufacturing has been highlighted during the recent COVID-19 pandemic to overcome the global shortage of diagnostic tools and personal protective equipment.[27]

In this work, we characterized five different locally made handmade papers known as Nepali kagaj, in the local language, for their potential use in fabricating PADs. Nepali kagaj is made from the fibrous bark of Daphne bholua and Daphne papyracea or other similar plant species following the traditional eco-friendly method of fiber processing and pulping.[28] Nepali handmade paper is considered to be highly resistant to germs such as mildew, paper crawlers, and termites.[29] It has been used traditionally for recording government records and religious texts. However, in modern days, it is used as wrapping papers, paper lamps, restaurant menus, greeting cards, and photo frames.[29] We characterized commercially available Nepali handmade papers by measuring several physical and fluid flow characteristics. The best type of paper was chosen to make PADs for two representative assays. As a proof of concept, we used the PADs made from local handmade paper for the colorimetric protein assay. We then developed an assay for screening drug quality analysis and measured the active pharmaceutical ingredient (API) in metformin drugs purchased from the local market.

Experimental Section

Materials and Reagents

Tetrabromophenol blue (TBPB), citric acid, trisodium citrate, sodium hydroxide, potassium dichromate, sodium nitroprusside (SNP), and sodium hypochlorite (NaOCl) were bought from Thermo Fisher Scientific India Pvt. Ltd., India. The metformin standard was bought from Accord Healthcare Pvt. Ltd., India, and was standardized according to the Indian Pharmacopoeia. Bovine serum albumin (BSA) or Fraction V purchased from Himedia Laboratories Pvt. Ltd., India, was used as a standard protein. All chemicals were used as received without further purification. We purchased five different Nepali handmade paper samples (hereunder named as P1, P2, P3, P4, and P5) from local handmade paper enterprises and stored them in airtight Ziplock bags until performing experiments.

Characterization of Handmade Paper Sheets

The grammage of the paper samples was measured following the TAPPI T410 test method with slight modifications.[30] Handmade paper sheets were cut into rectangular shapes of different sizes to measure the grammage of the paper sample. We measured the area and weight of the paper (±0.001 g) at ∼23 °C temperature and ∼50% relative humidity. Grammage was estimated as the ratio of the weight of paper (g) to area (m2). We measured the thickness of papers using an optical microscope (Amscope) by imaging them along their thickness. The field of view of the microscope was calibrated using a linear calibration grid (grid size 10 μm), and the image pixels were converted to micrometers in ImageJ software to get thickness information.[31] Apparent density was calculated by dividing grammage (g/m2) by its thickness (μm). Five measurements were taken of each sample.

The wicking speed of the paper samples was measured using paper strips of different widths. The strips were kept vertically in a beaker containing potassium dichromate solution, and the flow of colored solution was monitored by taking images every minute using a smartphone. The distance traveled by the dichromate solution on the paper strip per unit time was considered as the wicking speed.[32]

The porosity of paper samples was calculated by measuring the volume of water absorbed by rectangular-shaped paper pieces of different sizes.[33] At first, we measured the dry weight of each paper sample, and then, they were soaked in distilled water for 2–3 min, and the wet weight was measured. The porosity of paper samples was calculated by dividing the absorbed weight of water by total weight of the sample.[34]

The moisture content was determined following the TAPPI T412 test method.[35] To measure the moisture content, 2.0 g of paper sample was oven-dried at 105 ± 2 °C for 24 h. The sample was then cooled in a nonhygroscopic desiccator and weighed. The difference in the weight before and after drying was used to calculate the moisture content of the paper.[36]

To measure the water filtration coefficient, paper samples were cut using a circular cutter into circles of different diameters. They were folded to make a 60° cone. The folded papers were then wet thoroughly with distilled water. A total of 25 mL of distilled water was then poured into the samples, and the time taken by the samples to filter half its volume was noted using a stopwatch.[37] The water filtration coefficient was calculated using the following equation[37]where K is the water filtration coefficient and t is the time (s) taken to filter out half volume of water.

The water contact angle (WCA) was measured using a smartphone.[38] A drop of water (50 μL) was put onto the surface of paper using a micropipette, and images of water drops were taken using a smartphone. The images were analyzed using polynomial fitting with the dropsnake plugin in ImageJ software to measure the contact angle. The drop image was cropped to make the left and right interface of the drop clearly visible. Few knots (5–10) were added on the drop contour until the spline was finalized and the plugin displayed the contact angle. The WCA measurement for each type of paper surface was repeated ten times. See the Supporting Information for the step-by-step procedure to measure the WCA using ImageJ software.

Fabrication of the Paper Device

Paper-based devices were fabricated using a wax printing method.[15] We used Adobe Illustrator software to design assay regions as an array of circles with an inner diameter of 5.3 ± 0.2 mm with 2.7 mm line thickness. The pattern designs were transferred onto paper using the wax printing method (Xerox ColorQube 8580, Japan).[15] The wax-patterned paper was heated using a heating iron for 40 s during which the wax melted and penetrated through the paper to form hydrophobic barriers across the thickness of paper. Finally, one side of the device was covered with transparent tape to keep the reagents contained in the assay region.

Colorimetric Assays

Protein assay was performed using the TBPB method and BSA as a standard protein. The testing zone of PADs was prepared by loading 3 μL of 250 mM citrate buffer (pH 2.0) and was allowed to dry at room temperature for 2 min. Then, 3 μL of 3 mM TBPB in 95% ethanol was added in the testing zone followed by the addition of 5 μL of protein standard. The signal of the assay was recorded after 8 min because it got saturated afterward.[10]

Metformin assay was performed by allowing 4 μL of each 0.4% (w/v) of NaOCl, 2.0 M NaOH, and 0.04 M sodium nitroprusside (SNP) and 5 μL metformin react.[39] The color of the assay was recorded after 15 min with a smartphone. We also tested metformin samples purchased from local pharmacy stores in Kathmandu.

Image Analysis

We used a Samsung Galaxy M30s smartphone to image the assay device and ImageJ image analysis software to measure the color intensity of the assay. The images were converted into 8-bit and inverted. After this, the images are analyzed in three color spaces (R, G, and B). The average signal value of all pixels in the assay zone was measured. Triplicate measurements were carried out for each assay. Blank assays were performed along with sample assays. The net signal of the assay was obtained by subtracting the mean signal of the sample assay from the signal of the blank assay. We chose the green color channel because it gave a higher net intensity.[10] A schematic of the assay procedure on the paper device is outlined in Figure .

Figure 1

Photograph of handmade paper (on left) and a general procedure for performing colorimetric assays on the paper device (middle and right).

Spectrophotometric Detection of Metformin

The spectrophotometric signal of metformin standard solution (5–50 μg/mL) was measured at 236 nm in an LVS-A20 UV–visible spectrophotometer (LABTRON, U.K.).[40] Drug samples (30 μg/mL) were prepared in distilled water from 500 mg tablets. One tablet from each sample was crushed into fine powder using a pestle and was dissolved in 5mL of water. The concentration of metformin in samples was estimated using the regression equation of calibration curves.

Results and Discussion

A photograph of Nepali kagaj is shown in Figure . The physical parameters of all handmade papers are shown in Table . The average thickness of the paper samples was 230.30 ± 55.05 μm, ranging from 198 μm (P2) to 314 μm (P4). The thickness of P5 was similar to that of Whatman grade 1 paper (180 μm), and the remaining four samples were thicker than the Whatman grade 1 paper.[20] However, P1 and P2 had similar thickness to that fo Whatman grade 4, P5 was thinner, and P3 and P4 were even thicker than Whatman grade 4 paper (205 μm).[20]

Table 1

Physical Properties of Nepali Handmade Papers

paper type	thickness (μm)	GSM (g/m2)	apparent density (g/cm3)	porosity (ε %)	moisture content (%)	contact angle (°)	filtration coefficient (per sec)
P1	207.4 ± 11.1	49.1 ± 3.5	0.23	0.83	5.9 ± 0.4	106.2 ± 3.8	156.7 ± 2.7
P2	198.6 ± 7.1	75.2 ± 2.5	0.38	0.79	6.1 ± 0.7	102.0 ± 3.4	132.2 ± 2.6
P3	254.1 ± 9.4	108.3 ± 7.4	0.43	0.82	5.8 ± 1.0	101.9 ± 3.9	45.3 ± 1.5
P4	314.8 ± 9.3	117.8 ± 8.9	0.37	0.81	7.1 ± 0.2	112.1 ± 4.2	35.7 ± 1.3
P5	176.6 ± 14.3	54.4 ± 16.3	0.30	0.81	6.5 ± 1.0	76.6 ± 1.3	104.8 ± 4.3
grade 1a	180	88	0.40	0.68	N/A	N/A	150
grade 4a	205	96	0.46	0.64	N/A	N/A	37

Whatman filter paper for comparison.[20]

The gram per square meter (GSM) of papers, which is also known as grammage, ranged widely from 49 g/m2 (P1) to 117.8 g/m2 (P4). For reference, the grammage and thickness of Whatman grade 1 and 4 filter papers[20] were reported to be 88 and 96 g/m2, respectively. A good positive correlation (R = 0.89) between the grammage and thickness was observed. The apparent density of paper samples ranged from ∼0.23 (P1) to 0.43 (P3) g/cm3. The low apparent density of most of the paper samples suggests that these are lightweight papers.

The optical microscopy images of the handmade papers show variable fiber networking and pores (Figure ). The pores in the images may have been impacted by light penetration through the paper, which depends on the thickness of the papers. The thickness of the paper substrate determines the penetration of wax while making hydrophobic barriers on paper analytical devices. Similarly, the optical path length, scattering, assay sensitivity, and volume of reagents for an assay are also affected by thickness in the paper device.[9]

Figure 2

Optical microscopy images of five handmade papers.

Wicking speed affects the contact time between the sample and reagents and distribution of reagents in the reaction zone. This eventually may have an impact on the intensity and homogeneity of the color.

We measured the wicking speed of the aqueous solution on the paper strips. The experimental setup of the measurement is given in Figure A. The dichromate solution wicked upward on the strips with time. The wicking speed of P1, P2, P3, P4, and P5 was found to be 0.12, 0.123, 0.062, 0.089, and 0.124 mm/s (see Table S1). The thicker substrates (P3 and P4) transferred solution at a slower rate in comparison with the thinner substrates (P1, P2, and P5). Different types of handmade Nepali papers provided variable wicking speed of aqueous solution (Figure B). The graph representing the distance with time for each paper was fitted with a linear function to estimate the speed. We also explored the impact of the width of the paper strip on wicking speed. The wicking speed of P5 samples in the strips of width 4, 3, 2, and 1 mm was 0.124, 0.119, 0.115, and 0.112 mm/s, respectively (see Table S2). We found that wicking speed decreased in the channels of smaller width. A narrow channel provides more resistance for the fluid flow in paper and therefore slows down the flow.[41] The fluid flow also depends on the thickness of the paper strip as thinner strips provide lower resistance to flow, leading to a faster fluid flow.[20]

Figure 3

(A) Demonstration of flow visualization using a color solution at different times. The fluid front is indicated with arrows; (B) variation of distance traveled by color solution in different paper samples with times; and (C) variation of distance traveled by the color solution in paper strips (P5) of different width.

The porosity of handmade paper samples was in the range of ∼79% (P2)–83% (P1). The porosity of Whatman qualitative filter papers is reported in the range of ∼64% (grade 2)–68% (grade1).[20] Papers with high porosity increase the absorbance of ink and help the ink to dry quickly. High porosity is caused by stiff fibers, excessive flocculation of fibers, or insufficient calendaring.[42] Porosity is useful in calculating the total volume of the liquid reagent required to wet the substrate.[43]

The amount of water contained in the paper is expressed as a percentage of the paper’s weight. The equilibrium moisture content (EMC) of the paper samples ranged from ∼5.2% (P3) to 7.13% (P4). The water filtration coefficient of handmade paper samples was found in the range of 35.67 (P4)–156.74 (P1). The filtration efficiency is affected by the density, thickness, and size of water-permeable pores in the paper. The paper with high efficiency has a high filtering speed and resolution.

We started the water contact angle (WCA) measurement experiments on the poly(tetrafluoroethylene) (PTFE) substrate as a reference. The contact angle between distilled water and PTFE was found to be 106.8°, which is close to the reported value of 108°, which was measured by atomic force microscopy.[44] The WCA of handmade paper ranged from ∼76° (P5) to 112°(P4) (Table ). Unlike commercial filter papers, most of the handmade papers had a contact angle greater than 90°, which indicates the hydrophobic nature of the papers, while P5 is hydrophilic. The hydrophobicity of handmade papers may have arisen from possible additives such as wax or oil added during the paper making process. The WCA values reported in our case may have some measurement variations due to the surface roughness and inhomogeneity of paper surfaces, making the drop not perfectly axisymmetric.[38]

Colorimetric Assays on Paper Devices

Among five different types of NepaliKagaj, sample P5 had a contact angle less than 90°, making it hydrophilic in nature. We selected the P5 sample as an appropriate platform to fabricate the paper device, considering its thickness, wicking speed, and color uniformity.

As a proof of concept, we performed a protein assay on the paper device. The quantification of BSA is based on its ability to interact with TBPB indicator dye through a combination of electrostatic and hydrophobic interactions to form a concentration-dependent bluish-green complex (Figure A). To demonstrate the viability of the method, levels of protein in the BSA standard were quantified. The results showed that the net intensity increased the concentration of BSA, and the signal was proportional to the logarithmic value of BSA concentration in the range of 0.5–50 mg/mL (Figure B). The linear range was found to be 0.5–6 mg/mL (inset in Figure B). The limit of detection (LoD) and limit of quantification (LoQ) of this assay were 1.33 mg/mL and 4.91 mg/mL, respectively, which are similar to the literature-reported values of 0.9 and 2.9 mg/mL, respectively.[10] LoD was calculated using the 3.3(Sy/S) equation, and LoQ was calculated using the 10(Sy/S) equation. Sy in the equations is the standard deviation of response in the calibration curve, and S is the slope of the calibration curve.[45]

Figure 4

(A) Images of PADs showing the color development in the protein assay. Triplicate images for each BSA concentration are shown. The first row shows the triplicate images for the blank. Numbers indicate the concentration of the standard protein (BSA); (B) response curve of BSA on PADs at different concentrations (0–50 mg/mL). The inset shows a linear region of the response curve.

We developed a more useful assay to expand the applicability of handmade papers. This assay determined the concentration of metformin in tablet forms. Drug quality is a serious issue as low-quality drugs pose social, economic, and health burden to the society. Recent reports have suggested that as high as 10.5% drugs worldwide are either substandard or falsified. The problem is more adverse in low- and middle-income countries (13.6%), especially in Africa (18.7%) and Asia (13.7%).[46] Having a low-cost, easy-to-use, point-of-need drug quality screening technology such as PADs would contribute toward solving the widespread prevalence of low-quality drugs.

The metformin assay relied on the addition of sodium hypochlorite solution to an alkaline solution of metformin hydrochloride that produced β-diketone. β-diketone is an oxidized product of metformin. Sodium nitroprusside (SNP) in alkaline medium reacts with β-diketone to give a green-colored product.[39] The photographs of assay zones after the color development are shown in Figure A. Triplicate experiments were performed for each concentration. The assay signal responded to the concentration of metformin tested from 0.0625 to 40 mg/mL in the logarithmic fitting (Figure B). The linear range obtained for metformin in our system is 0.0625–0.5 mg/mL (Figure B inset). The LoD and LoQ of this assay were 0.05 and 0.18 mg/mL, respectively.

Figure 5

(A) Images of PADs to showing the green color development of the metformin standard. Triplicate images for each metformin concentration are shown. The first row shows the triplicate images of the blank. Numbers indicate the concentration of metformin; (B) response curve of metformin analysis using PADs. The inset shows the linear region of the response curve.

Quality of Metformin Samples

After developing the PAD method for determining the amount of metformin, we collected metformin samples (n = 20) from local pharmacies and tested them using PADs and spectrophotometric methods. The average amount of metformin concentration in samples was 465.6 ± 15.1 mg/tablet (range: 429.3–482.39 mg, see Table ). The label claim of these samples was 500 mg. The PAD-determined value of metformin samples was slightly lower than the label claim. According to the Indian Pharmacopoeia,[47] the acceptable range for metformin hydrochloride tablets is 450–550 mg/tablet.[47] The PAD assay found that three samples (S2, S10, and S16) did not meet the regulatory standards, all slightly lower than 450 mg.

Table 2

Concentration of Metformin in Locally Collected Samples

sample ID	metformin (mg/tablet)
S1	470.3
S2	444.3
S3	473.8
S4	466.2
S5	480.2
S6	480.3
S7	475.5
S8	468.1
S9	482.4
S10	429.3
S11	475.5
S12	452.9
S13	458.5
S14	482.3
S15	479.7
S16	445.6
S17	477.8
S18	463.5
S19	451.1
S20	455.9
ave.	465.6
stand. dev.	15.1

To compare the performance of the PAD method for the determination of metformin in tablet forms, we also tested the same samples using a spectrophotometric method. The calibration curve of metformin determination using the spectrophotometric method is given in Figure S1. The LoD and LoQ of the spectrophotometric method were 0.75 and 2.45 μg/mL, respectively.[40] The amount of metformin in each tablet determined using the spectrophotometric method was 483.8 ± 21.1 mg/tablet (range: 419.0–516.0 mg, see Figure ). Based on the spectrophotometric measurement, only one sample was found to be not within the acceptable range suggested by the Pharmacopoeia (Table S3). The paper device method underestimated the concentration of metformin samples by 18.2 ± 23.6 mg/tablet; p < 0.001 when compared with the spectrophotometric method. Additionally, the paper device correctly predicted 18 out of 20 samples. In Figure b, we plotted the difference between two methods against the mean of two methods, which shows that 19 out of 20 samples were within the 10% acceptable range, indicating the good agreement of paper devices with the spectrophotometric method.

Figure 6

Box plot of determination of metformin using two different methods (A). A comparative plot is shown in (B) in which difference in the amount of API from two methods is plotted. The top and button two horizontal lines indicate the maximum and minimum acceptable amount of API as indicated by the Pharmacopoeia, respectively.

Conclusions

In this paper, we reported the characterization of five Nepali kagaj for their potential in fabricating PADs. The handmade kagaj showed a wide range of properties such as thickness, grammage, wicking speed, contact angle, etc. The best sample was chosen to make PADs for two different applications. At first, we demonstrated a protein assay on the PADs fabricated on handmade paper. Then, the PADs were used to develop an assay to determine the amount of metformin in tablets. Both assays performed satisfactorily. We showed that the PADs fabricated on handmade Nepali kagaj can be a low-cost and easy-to-implement sensor for screening the quality of metformin, an antidiabetic drug. A similar assay can be developed, with the proper choice of colorimetric reaction, to detect and quantify other types of analytes of interests including counterfeit drugs. Our future work is to look into combining PADs made from Nepali kagaj and smartphone application for data reading, analyzing, and reporting, which could be beneficial to users, policy makers, and regulating agencies.