Rapid and low-cost quantitative detection of creatinine in human urine with a portable Raman spectrometer

Journal Pre-proof Rapid and low-cost quantitative detection of creatinine in human urine with a portable Raman spectrometer Wei Zhu, Bao-Ying Wen, Ling-Jun Jie, Xiang-Dong Tian, Zhi-Lin Yang, Petar M. Radjenovic, Shi-Yi Luo, Zhong-Qun Tian, Jian-Feng Li PII:

S0956-5663(20)30064-6

DOI:

https://doi.org/10.1016/j.bios.2020.112067

Reference:

BIOS 112067

To appear in:

Biosensors and Bioelectronics

Received Date: 18 October 2019 Revised Date:

12 December 2019

Accepted Date: 30 January 2020

Please cite this article as: Zhu, W., Wen, B.-Y., Jie, L.-J., Tian, X.-D., Yang, Z.-L., Radjenovic, P.M., Luo, S.-Y., Tian, Z.-Q., Li, J.-F., Rapid and low-cost quantitative detection of creatinine in human urine with a portable Raman spectrometer, Biosensors and Bioelectronics (2020), doi: https://doi.org/10.1016/ j.bios.2020.112067. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier B.V.

CRediT authorship contribution statement Wei Zhu: Investigation, Methodology, Writing - original draft, Writing - review & editing. Bao-Ying Wen: Investigation, Writing - original draft, Writing - review & editing. Ling-Jun Jie: Investigation, Writing - original draft. Xiang-Dong Tian: Formal analysis, Writing - original draft, Writing - review & editing, Conceptualization, Methodology. Zhi-Lin Yang: Conceptualization, Methodology, Supervision. Petar M. Radjenovica: Writing - original draft, Writing - review & editing, Conceptualization. Shi-Yi Luo: Formal analysis, Investigation. Zhong-Qun Tian: Conceptualization, Methodology, Writing - review & editing. Jian-Feng Li: Conceptualization, Writing - original draft, Writing - review & editing, Methodology, Supervision.

Rapid and Low-Cost Quantitative Detection of Creatinine in Human Urine with a Portable Raman Spectrometer Wei Zhua,1, Bao-Ying Wena,1, Ling-Jun Jieb, Xiang-Dong Tianb,*, Zhi-Lin Yanga,*, Petar M. Radjenovica, Shi-Yi Luoa, Zhong-Qun Tiana, Jian-Feng Lia,* a

Department of Physics, State Key Laboratory of Physical Chemistry of Solid Surfaces, iChEM,

College of Chemistry and Chemical Engineering, College of Energy, Xiamen University, Xiamen 361005, China b

Xiamen Cardiovascular Hospital, Xiamen University, Xiamen 361005, China

*Corresponding authors: [email protected], [email protected], [email protected] 1

Wei Zhu and Bao-Ying Wen contributed equally to this work.

Abstract The creatinine concentration of human urine is closely related to human kidney health and its rapid, quantitative, and low-cost detection has always been demanded. Herein, a surface-enhanced Raman spectroscopic (SERS) method for rapid and cost-effective quantification of creatinine concentrations in human urine was developed. A Au nanoparticle solution (Au sol) was used as a SERS substrate and the influence of different agglomerating salts on its sensitivity toward detecting creatinine concentrations was studied and optimized, as well as the effect of both the salt and Au sol concentrations. The variation in creatinine spectra over time on different substrates was also examined, demonstrating reproducible quantitative analysis of creatinine concentrations in solution. By adjusting the pH, a simple liquid−liquid solvent extraction procedure, which extracted creatinine from human urine, was used to increase the SERS detection selectivity toward creatinine in complex matrices. The quantitative results were compared to those obtained with a clinically validated enzymatic “creatinine kit (CK).” The limit of detection (LOD) for the SERS technique was 1.45 mg L-1, compared with 3.4 mg L-1 for the CK method. 1

Furthermore, cross-comparing the results from the two methods, the average difference was 5.84% and the whole SERS detection process could be completed within 2 minutes compared with 11 minutes for the CK, indicating the practicality of the quantitative SERS technique. This novel quantitative technique shows promises as a high-throughput platform for relevant clinical and forensic analysis. Keywords: Creatinine; Urine; SERS; Quantitative analysis; Portable Raman spectrometer

1. Introduction Global incidences of chronic kidney disease (CKD) are high and rapidly increasing, with billions of people suffering from CKD induced diseases, such as diabetes, hyperlipidaemia, hypertension and anaemia. Thus, it is vital to diagnose CKD at an early stage. Creatinine, in blood and urine, is the key biomarker for assessing CKD and monitoring its progress in clinical medicine (Randviir and Banks, 2013). Moreover, urinary creatinine concentrations are often used as internal standards to quantitatively analyze drugs and xenobiotics in forensic toxicology (Liotta et al., 2009). Therefore, to meet growing demands, the development of an accurate, low-cost and rapid urinary creatinine detection technique for routine clinical and forensic analysis is highly desirable (Lad et al., 2008; Pundir et al., 2013; Randviir and Banks, 2013). To date, the most commonly used protocols for determining urinary creatinine in both clinical and toxicological laboratories are the Jaffe reaction (Randviir et al., 2013; Sergeyeva et al., 2013) and enzymatic assays (Hanif et al., 2016; Nguyen Vu et al., 1991; Yamato et al., 1995). In the Jaffe reactions creatinine forms a colored complex with picric acid in a basic solution and is then detected with spectrophotometry. However, the Jaffe reaction suffers from low specificity due to the interference from other molecules in bodily fluids. An enzymatic method referred to as a “creatinine kit (CK)” (Yamato et al., 1995) has improved specificity, however, the addition of enzymes increases detection costs. Other traditional analytical techniques, including HPLC (Burton et al., 2013; Chen et al., 2012; Devenport et al., 2014; Jen et al., 2002; Park et al., 2008; Zhao et al., 2011), and 2

isotope dilution gas chromatography/mass spectrometry (ID-GC/MS) (Devenport et al., 2014; Hewavitharana and Bruce, 2003; Mora et al., 2007), provide excellent detection limits for quantifying creatinine concentrations but require expensive equipment and lengthy sample preparation times as well as trained personnel (Table. S1). Consequently, a new, simple, low-cost and highly sensitive analytical tool is still demanded in improving the clinical diagnosis of metabolic disease. Surface enhanced Raman spectroscopy (SERS) as a vibrational spectroscopic technique can realize fast and trace level detection of analytes in complex media (Li et al., 2010). Presently, SERS is widely used in interface and surface science (Dong et al., 2019; Li et al., 2019; Wang et al., 2019), material analysis (Oakley et al., 2011; Wang et al., 2019), biomedicine detection (Ji et al., 2013; Park et al., 2017; Tang et al., 2015; Wang et al., 2017; Xu et al., 2015), forensics ( Deriu et al., 2019; Dong et al., 2015; Kline et al., 2016), food safety (Yan et al., 2018; Zhang et al., 2016), and environmental monitoring (Hardy et al., 2014; Ma et al., 2013). SERS has been used to detect creatinine in body fluid previously, however, detection specificity can be affected by the complexity of biological samples. Interferents such as small molecules, organic salts, nucleotides and proteins generate large amounts of background noise and can prevent the analyte from reaching the enhancement “hot spots” of the SERS substrate, leading to low specificity and sensitivity (Sun et al., 2016). Stosch et al. (Stosch et al., 2005) used isotopically labelled creatinine as an internal standard, combined with multivariate data analysis, to achieve the quantitative SERS determination of creatinine in human serum. However, low accessibility to the isotopic analogue of the analyte greatly limits the clinical utility of this method. Also, high operational requirements and time-consuming extraction procedures hinder the rapid detection in clinical and law enforcement settings. Tadele Alula et al. (Tadele Alula and Yang, 2014) prepared Ag@ZnO/Fe3O4 composites for the magnetic separation and quantitative SERS measurements of creatinine in urine. This method detected the urine creatinine in a concentration range up to 0.8 µM. However, the results are not compared to the standard creatinine detection methods. Li et al. (Li et al., 2015) realized the detection of 3

creatinine in urine using the stamping SERS technique where the PDMS thin film accommodating the target works as the ink pad and the nanopore Au disks substrate functions as the stamp. Although the urine sample can be directly used for the SERS analysis without the pretreatment, the sensitivity was deteriorated by about four orders of magnitude due to the interferents (for example: the trace proteins in urine), compared to the pure water sample. Moreover, the manufacturing of the Au disks requires delicate instruments and time-consuming multi-step process. Thus, robust SERS method capable of adaptation to the fast, simple and quantitative detection of urine creatinine are still urgently need, where more research should be focused on preventing the interference of contaminants in urine, improving the sensitivity and achieving rapid result readout.

Fig. 1. Schematic illustration for the SERS detection of creatinine in human urine.

Herein, a simple liquid-liquid solvent extraction process combined with a portable Raman spectrometer-based method for quantifying creatinine concentrations in human urine was developed with potential for applications in clinical and forensic medicine (Fig. 1). First, the SERS performance of the Au nanoparticle solution (Au sol) substrate was optimized by studying the effect of different aggregating salts as well as the concentration of both the aggregating salt and Au sol. Due to the complexity of the human urine matrix, a 4

simple liquid-liquid solvent extraction method was developed to increase the selectivity and sensitivity of the SERS technique by inhibiting interferents in urine from influencing the analyte. With the optimized properties of the SERS detection technique, a linear calibration curve was obtained from the log concentration of creatinine and the log peak area (PA) of the characteristic creatinine Raman peak, giving a correlation factor of 0.969. Finally, the SERS method was cross-compared with the clinically validated CK method to demonstrate the feasibility of the SERS technique in a clinical setting.

2. Material and methods 2.1 Chemicals Creatinine (99%) was purchased from Aladdin, chloroauric acid (HAuCl4· 3H2O, 99.99%), and Sodium citrate (CA, Na3C6H5O7, 99.0%) were purchased from Alfa Aesar., HCl (36.0%~38.0%), NaOH (99.9%), Na2HPO4 (99.0%), n-Butanol (C4H10O, 99.5%), and NaCl (99.5%, AR) were purchased from Sinopharm Chemical Reagent Co., Ltd. NaHCO3 (99.5%) was purchased from Guangdong Guanghua Sci-tech. All water used in experiments was ultrapure Milli-Q water (~18.2 MΩ·cm, 3 ppb). All actual urine samples in the present study were provided by different healthy volunteers. 2.2 Instrumentation A Hitachi scanning electron microscope (SEM, HITACHI S-4800, Japan), a custom-made portable Raman spectrometer (BTR115-785S, B&WTEK, LLC), and a UV-visible spectrometer (UVmini-1280, SHIMADZU) were used. 2.3 Preparation of Au Nanoparticle Solution The preparation of the Au sol substrate was based on the Frens method (Frens, 1973). 2.424 mL of 0.825 wt% chloroauric acid solution was diluted 20 times with water, continuously stirred and heated to boiling for 10 min, then 1.5 mL of 1 wt% sodium citrate solution was quickly added. The solution visibly turned from colorless to black then to brick red, and was heated for a further 20 min before cooling naturally before use.

5

2.4 Preparation of Aqueous Creatinine Solutions and Artificial Urine A 5 wt% aqueous solution of creatinine was prepared by dissolving 1 g of creatinine powder in Milli-Q water within a glass vial. The solution was then diluted to make-up creatinine samples with 25, 20, 15, 10, 7.5, 5, 2, 1.5, 1, and 0.5 mg L-1 concentrations. The concentrations of each component in artificial urine were: 1.55 mM NaCl, 2.67 mM KCl, 2.6 mM CaCl2, 29.6 mM Na2SO4, 3.2 mM MgSO4, 19.8 mM NaH2PO4, 9.8 mM creatinine (1109 mg L-1) and 310 mM urea (18,618.6 mg L-1). Different concentrations of creatinine aqueous solution were then added to the artificial urine to give samples with 10, 15, 30, 55, 130, and 260 mg L-1 final creatinine concentrations. For quantification measurements, a human urine sample with a starting creatinine concentration of 300 mg L-1 (Measured through the CK method in hospital) from a healthy child volunteer was used and spiked with different creatinine concentrations. The final creatinine concentrations of the urine samples were 6.5, 14.0, 26.5, 51.5, 126.5, and 251.5 mg L-1. 2.5 Optimization of Experimental Detection Conditions 20 mg L-1 creatinine aqueous solutions were used as standards for exploring experimental conditions. 200 µL of the standard creatinine solution and 50 µL of the agglomerating agent solution were mixed in a glass tube, followed by 300 µL of the Au sol. To improve the NP agglomeration in the Au sol SERS substrate, it was repeatedly mixed with suction using a pipette. The color of the Au sol gradually changed from brownish red to brownish grey, indicating particle agglomeration. The portable Raman spectrometer was then used to detect creatinine in the Au sol. The excitation wavelength was 785 nm, laser power was 500 mW, and the spectral acquisition time was 1 s. To optimize the enhancement of the SERS substrates, the impact of volumetric ratio of the agglomerating agent to the Au sol was studied. 200 µL creatinine aqueous solution (20 mg L-1) and 300 µL Au sol were mixed together. Then 10, 20, 30, 40, 50, and 100 µL of aqueous NaCl (1 M) was added to agglomerate the Au sol.

6

After determining the optimal NaCl concentration, the volume of Au sol added was then changed from 100, 200, 300, 400, to 500 µL, respectively, to study its impact on the SERS enhancement. 2.6 Solvent Extraction of Creatinine from Artificial and Real Urine 250 µL of artificial urine with a final creatinine concentration of 130 mg L-1 was used as the standard for testing detection conditions of creatinine in a complex matrix. 250 µL of NaHCO3-NaOH buffer (pH 10) and 500 µL of different solvents for creatinine extraction (n-butanol, n-hexane, ethyl acetate, methyl tert-butyl ether, benzene, cyclohexane or petroleum ether) were mixed with the standard artificial urine solution. After two-phase separation, the upper organic layer was taken for SERS detection. To study the effect of pH, 250 µL of different pH buffers (Na2HPO4-CA buffer at pH 2, and NaHCO3-NaOH buffer at pH 10 and 11) and 500 µL of solvent were mixed with 250 µL of creatinine spiked artificial urine. After two-phase separation, the upper organic layer was taken for SERS detection. To study the effect of the added volume of buffer on the extraction process, 50, 100, 150, 200, 250, 300, and 350 µL of NaHCO3-NaOH (pH 10) buffer and 500 µL of the optimal solvent extractant (n-butanol) were mixed with 250 µL of creatinine spiked artificial urine. After two-phase separation, the upper organic solvent layer was taken for SERS detection. Under the optimized conditions, creatinine concentrations in five prepared samples of real human urine (diluted by 10-fold）were measured using SERS. For comparison, these five samples were then sent to the hospital for creatinine concentration determination using the CK method.

3. Results and discussion 3.1 Raman Characterization of Creatinine Creatinine is a waste product of body metabolism released into the urine through the kidneys. Environmental changes can cause the configuration of creatinine to undergo 7

reversible changes (inset in Fig. 2a), such as, dehydrogenation of the ring which leads to the formation of a new double bond and amino group (Randviir et al., 2013). Reference Raman spectra of bulk solid and aqueous (5 wt %) creatinine were collected using a portable Raman spectrometer, respectively (Fig. 2a). Characteristic Raman bands at 669 cm-1 and 834 cm-1 for the solid sample shift to 681 cm-1 and 845 cm-1 in the aqueous creatinine solution, respectively. Several other bands also noticeably shift, indicating that the configuration of the creatinine molecular ring changes to some extent in the aqueous solution. The SERS spectrum of creatinine differs from the Raman spectrum of aqueous creatinine (Fig. 2a). However, the slightly shifted bands at 685 cm-1 and 847 cm-1 show the presence of creatinine. The SERS peak at 685 cm-1 is attributed to the lactam ring inner vibration mode, while the peak of 847 cm-1 is denoted as the out-of-annular deformation vibration mode (Bispo et al., 2013). 3.2 Detection Sensitivity Optimization Herein, Au sol prepared by the Frens method was used as the SERS substrate. Using SEM and UV-Vis, the average Au nanoparticle (NP) size is around ~ 50 - 60 nm with a plasmon resonance absorption wavelength at 547 nm, respectively (Fig. S1a and b). The SERS signal strength is related to the extent of NP agglomeration in the Au sol substrate. The degree of agglomeration of Au NPs affects the SERS quantitative performance due to fluctuations in the numbers of "hot spots" in the laser illumination area causing differences in SERS signal intensities. Insufficient or excessive agglomeration will decrease the SERS detection sensitivity. Therefore, it is necessary to vary certain experimental parameters, such as the types of the aggregating agent, its concentration and its aggregation time, to investigate and optimize the performance of the SERS substrate toward creatinine detection. First, the effect of aggregating agents on creatinine SERS signal strengths were investigated using NaCl, KCl, Na2SO4, MgSO4, HCl, NaOH, KBr, and KI (Fig. S2a). The SERS response was measured using the PA of the 685 cm-1 band of creatinine. Keeping all other conditions constant, the chlorine salts generated the strongest creatinine SERS 8

signals, with the PA of the salt agglomerating agents varying in the order Cl- > SO42- > I- > Br- > OH- (Fig. 2b and S2a). This much stronger SERS intensity response is likely due to the inductive

Fig. 2. (a) Normal Raman (solid powder: black line; aqueous solution: red line) and SERS spectrum of creatinine. Inset depicts the structural formula of creatinine. Plots showing the effect of varying experimental parameters on the SERS sensitivity towards the characteristic 685 cm-1 band of creatinine: (b) different types of agglomerating agent; (c) different volume of NaCl agglomerating agent; and (d) different volume of Au sol SERS substrate added.

effect of Cl- beneficially increasing the dipole moments of creatinine more than other anions. Moreover, we think Cl- can improve the adsorption of creatinine on the AuNPs is another possible reason. The pH of the citrate-capped Au sol is about 6.0, meaning all carboxylate groups of citrates are deprotonated (pKa1: 3.13, pKa2: 4.76, pKa3: 6.40). The protonated and neutral creatinine coexists at pH 6. It is reported that each creatinine can form two hydrogen bonds with the carboxylate group of each surface-adsorbed citrate molecule (Alula et al., 2018). Cl- can interact with the protonated creatinine (creatininium 9

cation) and stabilize the hydrogen bonds by reducing the electrostatic attraction between the creatininium cation and citrate anion. However, SO42- has two net electrons, which can neutralize the positive charge but one electron remains, leading to the repulsive interaction between the creatinine and the surface-adsorbed citrate. The hydrogen-bond induced adsorption process can also explain the results of KBr and KI. It well known that the Brand I- can interact strongly with the AuNPs. Therefore, the surface-adsorbed citrate molecule is replaced, which makes the creatinine and even other molecules difficult to adsorb on the Au surface. There are nearly no distinguishable SERS peaks in the spectra, demonstrating the surface cleaning effect of Br- and I- (Fig. S2a). Next, the effect of the concentration of both NaCl and Au sol on SERS performance was investigated. As shown in Fig. 2c and S2c, the optimal volume of added NaCl is 20 µL (38.5 mM). Lower NaCl volumes lead to insufficient aggregation of Au NPs in solution, while larger amounts cause the excess NPs to rapidly over aggregate. From Fig. 2d and S2d, 300 µL of Au sol (33 pM) is the optimal volume of Au NPs. Further addition of Au sol decreased the number of creatinine molecules in the hot spots, thus reducing the observed SERS intensity (Fig. 2d). 3.3 Reproducibility and Stability Characterization of the SERS Substrates For effective quantitative measurements, the SERS signal from the substrate must be strong, stable and reproducible. Thus, SERS reproducibility was studied by measuring the variation in the SERS signals between 20 different Au sol substrates under optimized conditions. Across the 20 substrates, the SERS intensity of the creatinine band at 685 cm-1 shows a relative standard deviation (RSD) of 4.93%, exhibiting remarkable SERS reproducibility (Fig. 3a). After addition of the aggregating agent, intensity fluctuations over time due to NP aggregation were repeated for three different sample Au sol substrates. The substrates show excellent stability throughout the spectral collection period with an average RSD of 4.35% (Fig. 3b and S2b). 3.4 Quantitative Analysis of Aqueous Creatinine Solution

10

To demonstrate the feasibility of quantitatively measuring creatinine levels using SERS, aqueous solutions with different creatinine concentration gradients were added to the Au sol and the SERS response monitored. At and above 0.5 mg L-1 the characteristic Raman bands of creatinine can be clearly identified (Fig. 3c). Above 25 mg L-1 the signal becomes constant and does not lead to any significant signal increases. The PA of the 685 cm-1 creatinine band as a function of concentration is shown in Fig. 3d, showing a clear linear intensity dependence with a correlation coefficient of 0.973. According to the 3σ principle, the limit of detection (LOD) of this method can reach 0.042 mg L-1 (Table S2).

Fig. 3. (a) SERS substrate reproducibility measurements for detection of creatinine (peak at 685 cm-1) from different substrate batches. (b) SERS reproducibility measurement over time using the 685 cm-1 spectral band intensity fluctuation over 80 s (measurements recorded 40 s after addition of the NaCl aggregating agent). (c) SERS spectra of creatinine aqueous solution with different concentrations. (d) The logarithmic plot of the SERS peak at 685 cm-1 versus the creatinine concentrations. Error bars equal to the standard deviations obtained from three measurements. 11

3.5 Optimizing Creatinine Detection Using Organic Solvent Extraction Detection of creatinine in human urine is complicated by the presence of interferents such as organic salts, nucleotides, and proteins which can generate large amounts of background noise and inhibit creatinine from reaching the enhancement hot spots of the SERS substrate (Sun et al., 2016). To determine the main interferents, the major components of urine were studied with SERS (Fig. S3). With the optimized substrate, there is nearly no SERS response for glucose and urea at concentrations similar to those in urine. Whereas, uric acid shows a SERS response but its characteristic peaks do not overlap with those of creatinine. Thus, these urine components can be excluded as interferents. Therefore, the presence of a trace proteins was considered as the main SERS signal interferents. To overcome any protein signal interference issues with detecting creatinine, a pretreatment steps for the extraction of creatinine from artificial urine was performed. By adjusting its pH organic solvent separation was used to extract creatinine from artificial urine. In the liquid−liquid creatinine extraction process, three different buffer solutions, namely Na2HPO4-CA (pH 2), and NaHCO3-NaOH (pH 10, and 11), were respectively added to solutions of artificial urine to tune its pH. In Fig. 4a, the NaHCO3-NaOH buffer solutions with both pH 10 and 11 showed the best signal response for extracting aqueous creatinine into the organic solvent (n-butanol) layer. Due to the pKa value of creatinine being 4.8 in water (Sittiwong and Unob, 2015), at low pH, the amino groups on the creatinine molecule are easily protonated, making the creatinine charged in aqueous solutions. However, at high pH values creatinine molecules are neutral and can be readily extracted into the organic solvent layer. After adding different volumes of the NaHCO3-NaOH buffer salt solution (pH = 10, and 11), the final pH of the solution was about 9.5 (Fig. S4). Besides n-butanol, the extraction performance of other organic solvents such as ethyl acetate, methyl tert-butyl ether, benzene, cyclohexane, and petroleum ether are also evaluated.

12

However, n-butanol has the best performance for improving the SERS detection sensitivity to creatinine (Fig. 4b). By combining a solvent extraction step and SERS, the quantitative detection of creatinine concentrations in artificial urine was realized. Different concentrations of creatinine in artificial urine were extracted into the organic solvent phase which was then extracted and transferred to the optimized Au-sol SERS substrate for the quantitative analysis (Fig. 4c). In Fig. 4d, a linear response from 10 mg L-1 to 260 mg L-1 for the creatinine SERS intensity is demonstrated. According to the 3σ principle, the LOD of this method can reach 1.45 mg L-1 (Table S2).

Fig. 4. SERS characterization of the impact of (a) pH buffers and (b) types of organic solvents on the extraction of creatinine from artificial urine. (c) SERS spectra of different concentrations of creatinine after solvent extraction from artificial urine. (d) Log plot of the SERS peak at 685 cm-1 vs. the Log creatinine concentration. Error bars equal to the standard deviations obtained from three measurements. 13

3.6 Rapid Quantitative Analysis of Creatinine in Human Urine Following this, the creatinine level in the urine from a healthy volunteer was analyzed to test the applicability of this combined solvent extraction and SERS detection technique for quantitatively analyzing creatinine concentrations in real human urine. First, using a clinical CK test routinely used in hospitals, the creatinine concentration of a human urine sample was determined to be 300 mg L-1. Then additional creatinine was added to the urine sample which was then split and diluted to different concentrations using ultrapure Milli-Q water to make-up multiple standards for solvent extraction and SERS analysis. In Fig. 5a and b, the linear response between the SERS PA of creatinine and the creatinine concentration in solution clearly demonstrates the potential for this technique to quantitatively analyze the human urine.

Fig. 5. (a) SERS spectra of creatinine with different concentrations after solvent extraction from human urine. (b) Plot of the log creatinine SERS PA at 685 cm-1 vs. the log of sample creatinine concentrations. Error bars depict the RSD obtained from three measurements. (c) SERS spectra

14

of 5 blind samples. (d) Linear plot comparison of the creatinine detection between the CK and SERS methods.

To further demonstrate the clinical potential of this SERS technique for routine clinical analysis, using the calibration curve obtained from Fig 5b and c, blind samples from five volunteers were analyzed. The quantitative results were compared with those obtained through the CK method. A clear strong correlation between the SERS and CK results is shown in Fig. 5d. The average percentage difference between the two analytical approaches is 5.84%, demonstrating this SERS method can be a highly attractive alternative technique for meeting the clinical application requirements (Table S3). Finally, besides comparable quantitative detection accuracy, the cost and operation time of the SERS analysis technique was also improved compared with the CK, further showing its potential for application in a clinical setting (Table 1).

Table 1. Comparison between CK and SERS method for the detection of creatinine in urine.

Readout

CK/clinical lab

SERS/our lab

Time-to-result Measurement range LOD Price Instrument

11 min 3.4~199 mg L-1 3.4 mg L-1 ¥5 each sample Integrated large biochemical analyzer

2 min 6.5 ~251.5 mg L-1 1.45 mg L-1 ¥0.2 each sample Portable Raman spectrometer

4. Conclusions There is a clear need for rapid, simple and low-cost quantitative techniques capable of effectively detecting CKDs using creatinine concentrations in human urine. Herein, using a simple solvent separation step, creatinine concentrations in human urine were accurately detected with a portable Raman spectrometer in an optimized Au-sol SERS substrate. By comparing different types of salt agglomerating agents, the detection sensitivity of the Au-sol substrate was optimized and improved using NaCl. The Au sol SERS substrate shows good signal reproducibility with less than 10% variation between different 15

substrates, meeting quantitative analysis application requirements. Spectral interference from proteins and other molecules in human urine was overcome by a simple solvent extraction creatinine which increased the selectivity for SERS analysis. The quantitative results measured using SERS were confirmed by the clinical CK method. The average percentage difference of the two methods was less than 10% demonstrating the excellent detection accuracy. It is believed that this rapid and low-cost SERS analytical approach holds great potential for applications in clinical and forensic settings. Acknowledgements This work was supported by NSFC (20720170102, 21775127, 21427813 and 21503231), the Fundamental Research Funds for the Central Universities (20720190044), and the Open Fund of the State Key Laboratory of Luminescent Materials and Devices (South China University of Technology). Declaration of Competing Interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Appendix A. Supporting Information Supplementary data to this article can be found online. References Alula, M.T., Karamchand, L., Hendricks, N.R., Blackburn, J.M., 2018. Analytica Chimica Acta 1007, 40-49. Bispo, J.A.M., Vieira, E.E.d.S., Jr., L.S., Fernandes, A.B., 2013. J. Biomed. Opt. 18(8), 1-8, 8. Burton, C., Shi, H., Ma, Y., 2013. Anal. Chem. 85(22), 11137-11145. Chen, S., Chen, L., Wang, J., Hou, J., He, Q., Liu, J.-A., Wang, J., Xiong, S., Yang, G., Nie, Z., 2012. Anal. Chem. 84(23), 10291-10297. Deriu, C., Conticello, I., Mebel, A.M., McCord, B., 2019. Anal. Chem. 91(7), 4780-4789. Devenport, N.A., Blenkhorn, D.J., Weston, D.J., Reynolds, J.C., Creaser, C.S., 2014. Anal. Chem. 16

86(1), 357-361. Dong, J.-C., Zhang, X.-G., Briega-Martos, V., Jin, X., Yang, J., Chen, S., Yang, Z.-L., Wu, D.-Y., Feliu, J.M., Williams, C.T., Tian, Z.-Q., Li, J.-F., 2019. Nat. Energy 4(1), 60-67. Dong, R., Weng, S., Yang, L., Liu, J., 2015. Anal. Chem. 87(5), 2937-2944. Frens, G., 1973. Nat. Phy. Sci. 241(105), 20-22. Hanif, S., John, P., Gao, W., Saqib, M., Qi, L., Xu, G., 2016. Biosens. Bioelectron. 75, 347-351. Hardy, M., Doherty, M.D., Krstev, I., Maier, K., Möller, T., Müller, G., Dawson, P., 2014. Anal. Chem. 86(18), 9006-9012. Hewavitharana, A.K., Bruce, H.L., 2003. J. Agr. Food Chem. 51(17), 4861-4865. Jen, J.-F., Hsiao, S.-L., Liu, K.-H., 2002. Talanta 58(4), 711-717. Ji, W., Chen, L., Xue, X., Guo, Z., Yu, Z., Zhao, B., Ozaki, Y., 2013. Chem. Commun. 49(66), 7334-7336. Kline, N.D., Tripathi, A., Mirsafavi, R., Pardoe, I., Moskovits, M., Meinhart, C., Guicheteau, J.A., Christesen, S.D., Fountain, A.W., 2016. Anal. Chem. 88(21), 10513-10522. Lad, U., Khokhar, S., Kale, G.M., 2008. Anal. Chem. 80(21), 7910-7917. Li, C.-Y., Le, J.-B., Wang, Y.-H., Chen, S., Yang, Z.-L., Li, J.-F., Cheng, J., Tian, Z.-Q., 2019. Nat. Mater. 18(7), 697-701. Li, J.-F., Huang, Y.-F., Ding, Y., Yang, Z.-L., Li, S.-B., Zhou, X.-S., Fan, F.-R., Zhang, W., Zhou, Z.-Y., Wu, D.-Y., Ren, B., Wang, Z.-L., Tian, Z.-Q., 2010. Nature 464, 392. Li, M., Du, Y., Zhao, F., Zeng, J., Mohan, C., Shih, W.-C., 2015. Biomed. Opt. Express 6(3), 849-858. Liotta, E., Gottardo, R., Bonizzato, L., Pascali, J.P., Bertaso, A., Tagliaro, F., 2009. Clin. Chim. Acta 409(1), 52-55. Ma, W., Sun, M., Xu, L., Wang, L., Kuang, H., Xu, C., 2013. Chem. Commun. 49(44), 4989-4991. Mora, L., Sentandreu, M.A., Toldrá, F., 2007. J. Agr. Food Chem. 55(12), 4664-4669. Nguyen Vu, K., Wolff, C.M., Seris, J.L., Schwing, J.P., 1991. Anal. Chem. 63(6), 611-614. Oakley, L.H., Dinehart, S.A., Svoboda, S.A., Wustholz, K.L., 2011. Anal. Chem. 83(11), 3986-3989. 17

Park, E.-K., Watanabe, T., Gee, S.J., Schenker, M.B., Hammock, B.D., 2008. J. Agr. Food Chem. 56(2), 333-336. Park, M., Jung, H., Jeong, Y., Jeong, K.-H., 2017. ACS Nano 11(1), 438-443. Pundir, C.S., Yadav, S., Kumar, A., 2013. Trac-Trend. Anal. Chem. 50, 42-52. Randviir, E.P., Banks, C.E., 2013. Sensor. Actuat. B-Chem. 183, 239-252. Randviir, E.P., Kampouris, D.K., Banks, C.E., 2013. Analyst 138(21), 6565-6572. Sergeyeva, T.A., Gorbach, L.A., Piletska, E.V., Piletsky, S.A., Brovko, O.O., Honcharova, L.A., Lutsyk, O.D., Sergeeva, L.M., Zinchenko, O.A., El'skaya, A.V., 2013. Anal. Chim. Acta 770, 161-168. Sittiwong, J., Unob, F., 2015. Spectrochim. Acta A. 138, 381-386. Stosch, R., Henrion, A., Schiel, D., Güttler, B., 2005. Anal. Chem. 77(22), 7386-7392. Sun, F., Hung, H.-C., Sinclair, A., Zhang, P., Bai, T., Galvan, D.D., Jain, P., Li, B., Jiang, S., Yu, Q., 2016. Nat. Commun. 7, 13437. Tadele Alula, M., Yang, J., 2014. Talanta 130, 55-62. Tang, L., Li, S., Han, F., Liu, L., Xu, L., Ma, W., Kuang, H., Li, A., Wang, L., Xu, C., 2015. Biosens. Bioelectron. 71, 7-12. Wang, Y.-H., Le, J.-B., Li, W.-Q., Wei, J., Radjenovic, P.M., Zhang, H., Zhou, X.-S., Cheng, J., Tian, Z.-Q., Li, J.-F., 2019b. Angew. Chem. Int. Edit. 58(45), 16062-16066. Wang, Y.-H., Wei, J., Radjenovic, P., Tian, Z.-Q., Li, J.-F., 2019a. Anal. Chem. 91(3), 1675-1685. Wang, Z., Zong, S., Wu, L., Zhu, D., Cui, Y., 2017. Chem. Rev. 117(12), 7910-7963. Xu, L.-J., Lei, Z.-C., Li, J., Zong, C., Yang, C.-J., Ren, B., 2015. J. Am. Chem. Soc. 137(15), 5149-5154. Yamato, H., Ohwa, M., Wernet, W., 1995. Anal. Chem. 67(17), 2776-2780. Yan, L., Dou, L., Bu, T., Huang, Q., Wang, R., Yang, Q., Huang, L., Wang, J., Zhang, D., 2018. Food Chem. 261, 131-138. Zhang, Y., Zou, Y., Liu, F., Xu, Y., Wang, X., Li, Y., Liang, H., Chen, L., Chen, Z., Tan, W., 2016. Anal. Chem. 88(21), 10611-10616. Zhao, J., Chen, H., Ni, P., Xu, B., Luo, X., Zhan, Y., Gao, P., Zhu, D., 2011. J. Chromatogr. B 18

879(26), 2720-2725.

19

Highlights Rapid and Low-cost Quantitative Detection of Creatinine in Human Urine with a Portable Raman Spectrometer A highly sensitive surface-enhanced Raman spectroscopic (SERS) technique was developed for the rapid quantification of creatinine levels in human urine. The SERS detection selectivity was improved through a simple liquid−liquid solvent extraction procedure. This novel approach promises a high-throughput platform for clinical and forensic analysis.

Author contribution section These authors contributed equally (Wei Zhu and Bao-Ying Wen). Experiments were designed by Jian-Feng Li and Zhi-Lin Yang. The experiments and data processing were executed by Wei Zhu, Bao-Ying Wen, Xiang-Dong Tian and Jian-Feng Li. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Declaration of interests ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. ☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: