Multiplexed Plasmon-enhanced Lateral Flow Assay for Early Diagnosis of Acute Kidney Injury

Abstract

Early diagnosis of acute kidney injury (AKI) remains a major clinical challenge due to the delayed and insensitive nature of conventional markers such as serum creatinine. Urinary protein biomarkers, including neutrophil gelatinase-associated lipocalin (NGAL) and cystatin C (CysC), provide earlier and complementary information on kidney injury, but their clinical translation is limited by the lack of rapid, quantitative, and multiplexed point-of-care (POC) diagnostic tests. Here, we report a multiplexed plasmonic-fluor-based lateral flow assay (p-LFA) that enables sensitive, quantitative, and simultaneous detection of NGAL and CysC. By harnessing ultrabright plasmonic-fluor nanolabels and ratiometric fluorescence analysis, the p-LFA achieves picogram-per-milliliter analytical sensitivity comparable to that of the enzyme-linked immunosorbent assay (ELISA). In addition, the simplified assay format enables sample-to-answer times as short as 10 min, while maintaining sensitivity sufficient for accurate AKI detection. Clinical validation using pediatric urine samples demonstrates a strong correlation with ELISA measurements for both biomarkers. Combined analysis of NGAL and CysC significantly improves AKI diagnostic performance, yielding an area under the receiver operating characteristic curve of 0.973 with a clinical sensitivity of 90.9% and specificity of 91.2% in the studied cohort. Furthermore, a low-cost, portable fluorescence reader enables quantitative performance equivalent to benchtop imaging while dramatically reducing instrument cost and complexity. Collectively, this work establishes a scalable and adaptable p-LFA platform that bridges laboratory-grade immunoassays and practical POC diagnostics, offering a promising solution for early AKI detection and multiplexed biomarker analysis in decentralized healthcare settings.

1. Introduction

Acute kidney injury (AKI) is a common and serious clinical condition that affects up to 30–50% of critically ill patients, particularly those admitted to intensive care units (ICUs) (Kaddourah et al.). AKI is associated with increased morbidity, mortality, prolonged hospitalization, and a heightened risk of chronic kidney disease (Hessey et al. 2021; Meena et al. 2023; Sutherland and Kwiatkowski 2017; Zarei et al. 2025). Despite its clinical significance, early diagnosis of AKI remains challenging (Jamaludin et al. 2022; Robinson et al. 2023; Xiao et al. 2022). Current clinical practice relies primarily on changes in serum creatinine (SCr) and urine output, which are delayed, indirect, and insensitive indicators of biological kidney injury (Hooper et al. 2025; Yuan 2019). As a result, opportunities for early intervention are often missed, underscoring the urgent need for rapid, sensitive, and clinically actionable diagnostic tools for AKI. In recent years, urinary protein biomarkers such as neutrophil gelatinase-associated lipocalin (NGAL) and cystatin C (CysC) have emerged as promising early indicators of kidney injury (Goldstein et al. 2024; Nejat et al. 2010a; Westhuyzen 2006). NGAL is rapidly upregulated and released into urine and blood following tubular injury, while urinary CysC reflects proximal tubular dysfunction and provides complementary information to NGAL(Chen et al. 2023; Wang et al. 2023). However, translation of these biomarkers into routine clinical use has been limited by the lack of diagnostic platforms that can simultaneously deliver high sensitivity, quantitative accuracy, multiplexing capability, and rapid turnaround at the point of care.

Lateral flow assays (LFAs) are widely used for POC diagnostics owing to their low cost, ease of use, and minimal infrastructure requirements (Liu et al. 2025; Liu et al. 2021; Posthuma-Trumpie et al. 2009; Quesada-González and Merkoçi 2015; Yetisen et al. 2013). Conventional LFAs rely predominantly on colorimetric gold nanoparticle labels, which suffer from an intrinsically limited optical signal intensity. This limitation constrains analytical sensitivity and poses significant challenges for quantitative analysis and detection of low-abundance protein biomarkers. Various nanolabels have been developed to improve the sensitivity of LFAs and enhance compatibility with simple readout methods for quantitative detection (Hang et al. 2025; Liang et al. 2026; Lou et al. 2022; Luo et al. 2026; Tu et al. 2022; Wei et al. 2022). Among these nanolabels, plasmonic-fluors (PF) have recently emerged as a promising strategy to enhance analytical sensitivity by coupling molecular fluorophores to plasmonic nanostructures, enabling strong fluorescence amplification through plasmon-enhanced excitation and emission (Gandra et al. 2014; Guo et al. 2023; Liu et al. 2026a; Luan et al. 2020). Recent advances in plasmon-enhanced fluorescence have further demonstrated that rational engineering of fluorophore-plasmon interactions through plasmonic and photonic-plasmonic nanostructures can substantially enhance fluorescence signals and improve analytical sensitivity for biomarker detection (Bhaskar et al. 2026; Bhaskar et al. 2025; Bhattacharya et al. 2025). Plasmon-enhanced fluorescence arises from two primary mechanisms: excitation enhancement and radiative decay rate enhancement. Localized surface plasmon resonance (LSPR) in metallic nanostructures generates intensified near-field electromagnetic fields that increase fluorophore excitation efficiency, while coupling between fluorophore dipoles and plasmonic modes enhances radiative decay rates and fluorescence emission (Kant et al. 2024; Kravets et al. 2018; Petryayeva and Krull 2011). Prior studies have demonstrated that PF nanolabels can generate detectable fluorescence signals at nanolabel densities approximately four orders of magnitude lower than those required for conventional colorimetric detection based on gold nanoparticles (Gupta et al. 2023). While these advances highlight the potential of PF nanolabels to enhance LFA sensitivity, their integration into multiplexed, quantitative, and clinically validated POC platforms remains unexplored (Gupta et al. 2023; Liu et al. 2026b).

Here, we report a multiplexed plasmonic-fluor-based lateral flow assay (p-LFA) for rapid, quantitative, and point-of-care (POC) detection of AKI biomarkers. The p-LFA integrates ultrabright PF nanolabels and a ratiometric fluorescence readout, defined as the ratio of test-line to control-line fluorescence intensity, to enable sensitive and multiplexed detection of NGAL and CysC within clinically relevant concentration ranges. We optimize assay parameters to achieve analytical sensitivity comparable to enzyme-linked immunosorbent assay (ELISA) and to enable simplified operation with assay times as short as 10 min. We further demonstrate minimal cross-reactivity in multiplexed operation and validate diagnostic performance using pediatric urine samples, showing that combined NGAL and CysC measurements significantly improve AKI classification compared with single-biomarker analysis. To enable POC deployment, we evaluate the performance of the p-LFA using a low-cost, portable fluorescence reader and demonstrate quantitative equivalence to a benchtop fluorescence scanner.

2. Methods

2.1. Materials.

Whatman FF120HP nitrocellulose membrane with polystyrene backing, Whatman standard 14 glass fiber membrane, and Whatman CF6 absorbent pad were purchased from Cytiva. NGAL and CysC antibodies were purchased from Hytest. Microplates, human NGAL ELISA kit, human CysC ELISA kit, recombinant NGAL, horseradish peroxidase (HRP)-labelled streptavidin, color reagents, and stop solution were purchased from R&D Systems. Goat immunoglobulin G (IgG) and anti-goat IgG were purchased from Jackson Immuno Research Labs. Phosphate-buffered saline (PBS, 10×) was purchased from Invitrogen. Bovine serum albumin (BSA) was purchased from Sigma-Aldrich. Tween-20 was purchased from Fisher Scientific.

2.2. Synthesis of detection antibody-plasmonic-fluor 800CW conjugates.

The detection antibody-plasmonic-fluor 800CW conjugates (dAb-PF800) for NGAL, CysC, and goat-IgG were synthesized following the previous report (Gupta et al. 2023; Liu et al. 2026b). Shortly, 1–3 μL 1 mg/mL biotinylated dAb was mixed with 1 mL streptavidin-PF 800CW and incubated for 30 minutes. For goat-IgG, 1–5 μL 2 mg/mL biotinylated goat-IgG was used. The solution was washed for 3 times before mixing with target solutions or depositing on conjugate pads.

2.3. Fabrication and testing of NGAL and multiplexed p-LFAs.

0.5–2 mg/mL NGAL capture antibody for test line and 200 μg/mL anti-goat IgG for control line were printed on FF120HP NC membrane at a dispense rate of 1 μL/cm and a speed of 50 mm/s using a XYZ Dispenser. After vacuum drying at room temperature (RT) for 30 min, the NC membrane (300 mm × 25 mm) was blocked by 9 mL 3 wt% BSA in 1x PBS for 60 minutes and rinsed by 6 mL PBST (0.05% Tween-20 in 1x PBS) for 10 minutes. The NC membrane was then vacuum dried at RT overnight. dAb-PF800 (OD: 5) and goat-IgG-PF800 (OD: 5) were mixed in a 2:1 ratio and then air-jet sprayed on a conjugate pad with a dispense rate of 3 μL/cm. The conjugate pad was then vacuum dried at RT overnight. To prepare multiplexed p-LFAs, 2 mg/mL NGAL antibody and 2 mg/mL CysC antibody for test lines and 200 μg/mL anti-goat IgG for the control line were printed on FF120HP NC membrane at a dispense rate of 1 μL/cm and a speed of 50 mm/s. dAb-NGAL-PF800 (OD: 5), dAb-CysC-PF800 (OD: 5), and goat-IgG-PF800 (OD: 5) were mixed in a 2:2:1 ratio and then air-jet sprayed on a conjugate pad with a dispense rate of 5 μL/cm. The NC membranes after capture antibodies printing, conjugate pads after dAb-PF800 and goat-IgG-PF800 conjugates printing, and absorbent pads were assembled with 5 mm overlaps and cut into 2.5 mm wide strips. For p-LFAs without conjugate pads, 1 μL dAb NGAL-PF800 (OD: 0.06), 1 μL dAb CysC-PF800 (OD: 0.06, for multiplexed p-LFA only), and 1 μL goat-IgG-PF800 (OD: 0.06) were mixed with 97 μL of samples and incubated for 30 minutes. The p-LFAs were scanned 30 min after introducing the sample-conjugate mixture to the NC membranes. For p-LFAs with a conjugate pad, 25 μL samples were introduced to the conjugate pads. The p-LFAs strips were scanned after assay times of 10, 20, or 30 min. For fluorescence analysis, absolute fluorescence was defined as the mean fluorescence intensity measured within a fixed rectangular region of interest (ROI) centered on each line. The ROI was 500 μm wide along the flow direction and 2 mm long perpendicular to the flow direction. The same ROI dimensions and analysis procedure were applied to all test and control lines, and no local background subtraction was performed. Ratiometric fluorescence was calculated by dividing the mean fluorescence intensity of each test-line ROI by that of the control-line ROI on the same strip. For multiplexed p-LFAs, NGAL and CysC fluorescence ratios were calculated separately using the shared control line. The logistic5 fitting and calculation of LODs and LOQs were performed using Origin. The ROC analysis was performed using Python.

2.4. NGAL and CysC ELISA.

ELISAs were performed according to the manufacturer's instructions. Briefly, 96-well plates were coated overnight at room temperature with capture antibodies diluted in 1× PBS (2 μg/mL for NGAL and 4 μg/mL for CysC). The plates were washed three times with PBST and blocked with 3 wt% BSA in 1× PBS for 1 h. After washing, NGAL and CysC standards and urine samples diluted in reagent diluent (1 wt% BSA in 1× PBS) were added and incubated for 2 h. The plates were then washed and incubated with biotinylated dAb (25 ng/mL for NGAL and 250 ng/mL for CysC in reagent diluent) for 2 h, followed by washing and incubation with HRP-conjugated streptavidin (40-fold dilution for NGAL and 200-fold dilution for CysC in reagent diluent) for 20 min. After washing, substrate solution (1:1 mixture of color reagents A and B) was added and allowed to react for 20 min before the reaction was stopped with 2 N H₂SO₄. All incubation steps were carried out at room temperature without agitation. Absorbance was measured immediately at 450 nm with background correction at 540 nm using a microplate reader.

2.5. Urine sample collection and storage.

Urine samples were collected via urinary catheters from patients within 24 h of the onset of organ failure. Samples were stored at 4 °C for no longer than 24 h, supernatants after spinning were aliquoted into 1 mL microcentrifuge tubes, and frozen at −70 °C until analysis. On the day of p-LFA and ELISA testing, samples were thawed at 4 °C within 1 h and diluted as required. All urine samples were diluted more than 10-fold prior to analysis, substantially reducing sample-to-sample differences in urine viscosity and matrix composition, thereby minimizing their potential impact on assay performance. Human sample collection and testing were conducted under approval from the Baylor College of Medicine Institutional Review Board (IRB H-45195) and the Texas A&M University Institutional Review Board (IRB 2020–0949).

2.6. Characterization.

A LI-COR Odyssey M imaging system and a portable reader from Brightest Bio LLC were used to collect the fluorescence images. Using the Odyssey M Imaging System, p-LFAs were imaged using the 785Ex/820Em channel, which employs a 785 nm solid-state diode laser for excitation and an 816–840 nm emission filter for fluorescence detection. Images were acquired using a complementary metal-oxide-semiconductor (sCMOS) image sensor at a spatial resolution of 100 μm. A Tecan Infinite M Nano+ plate reader was used to measure the absorbance after ELISA. LSPR spectra were collected using a Shimazu UV-1800 spectrophotometer. Transmission electron microscopy (TEM) image was obtained using a JEOL JEM-2100F FE TEM.

3. Results and discussion

3.1. Design of multiplexed p-LFA

Figure 1a illustrates the design and operating principle of the plasmonic-fluor-based LFA. The assay comprises an integrated sample and conjugate pad, a nitrocellulose (NC) membrane containing spatially defined test lines and a control line, and an absorbent pad. NGAL and CysC are selected as target analytes to improve diagnostic accuracy at early AKI onset(Basu Rajit et al. 2014; Feng et al. 2026; Murray et al. 2014). Although several LFAs have been reported for AKI biomarker detection, most focus on a single biomarker such as NGAL or CysC (Feng et al. 2026). Consequently, they do not capture complementary information from multiple kidney injury pathways. The multiplexed p-LFA reported here enables simultaneous quantitative measurement of NGAL and CysC within a rapid POC format, potentially improving diagnostic accuracy through combined biomarker assessment. The test lines are functionalized with capture antibodies specific to NGAL and CysC, respectively. The control line is formed using anti-goat IgG to capture goat IgG-conjugated PF nanolabels, validating proper capillary flow and providing an internal reference signal for ratiometric fluorescence analysis. Upon sample introduction, target analytes bind to antibody-conjugated PF nanolabels within the integrated sample and conjugate pad and are subsequently captured at the corresponding test lines through specific immunorecognition. This configuration facilitates effective interaction between the sample and nanolabel-conjugated antibodies prior to transport through the NC membrane, enabling reliable quantitative detection of NGAL and CysC. The absorbent pad serves as a capillary sink to sustain continuous fluid flow across the strip.

Figure 1.

Design of plasmonic-fluor-based lateral flow assay (p-LFA) for AKI diagnosis. (a) Schematic illustration of the multiplexed p-LFA, comprising an integrated sample/conjugate pad, nitrocellulose membrane with spatially defined NGAL and CysC test lines and a control line, and an absorbent pad. (b) Schematic illustration of the antigen-antibody binding at the test line and the fluorescence intensity measurement using a portable imager. (c) Normalized extinction spectra of AuNR@Ags before and after bioconjugation. (d) TEM image of dAb-PF800 conjugate.

In the multiplexed p-LFA, PF nanolabels serve as ultrabright fluorescent reporters by coupling molecular fluorophores to plasmonic nanostructures, enabling strong fluorescence amplification. Such strong signal at low PF nanolabel densities enhances analytical sensitivity without compromising capillary flow or increasing nonspecific background in multiplexed LFA formats, enabling sensitive detection of NGAL and CysC within clinically relevant concentration ranges. Following sample introduction, signal generation is completed in approximately 10 minutes and is subsequently analyzed using a low-cost, portable fluorescence imager for quantitative detection of NGAL and CysC (Figure 1b).

The antibody-conjugated PF nanolabels were prepared following a previously reported protocol(Gupta et al. 2023). We first synthesized silver-coated gold nanorods (AuNR@Ag) with an LSPR wavelength centered at approximately 800 nm to maximize fluorescence enhancement by matching both the excitation and emission wavelengths of the 800CW fluorescent dye. Compared with AuNR, AuNR@Ag can generate stronger localized electromagnetic fields and reduced plasmon damping, resulting in greater fluorescence enhancement (Bauch et al. 2014). The AuNR@Ag nanostructures were conjugated with the 800CW dye and detection antibodies as biorecognition elements, yielding the dAb-PF800 conjugates. A thin siloxane copolymer spacer layer was introduced between the AuNR@Ag surface and the 800CW dye to maintain an optimal fluorophore-metal separation and minimize metal-induced quenching, as established in our previous studies (Luan et al. 2020). Previous characterization of related PF nanolabels demonstrated substantial increases in quantum yield relative to the free dye, indicating that plasmon-enhanced excitation and radiative decay can dominate over quenching when an appropriate spacer layer is employed. Following bioconjugation, a red shift in the longitudinal LSPR peak was observed, consistent with an increase in the local refractive index at the nanoparticle surface (Figure 1c). Notably, the symmetry and full width at half maximum of the longitudinal LSPR peak remained unchanged, indicating the absence of nanoparticle aggregation during the bioconjugation process. To assess whether bioconjugation affected fluorescence enhancement, the fluorescence intensities of PF and dAb-PF conjugates were compared under identical optical density conditions. An approximately 15% reduction in fluorescence intensity was observed following antibody conjugation, indicating that bioconjugation did not substantially compromise plasmon-enhanced fluorescence. Notably, our previous study demonstrated that PF nanolabels exhibit fluorescence intensities approximately 6,700-fold greater than those of individual IRDye 800CW fluorophores (Luan et al. 2020), suggesting that the fluorescence enhancement provided by the plasmonic construct remains substantial a antibody conjugation. TEM imaging and size analysis confirm the uniform morphology and narrow size distribution of the synthesized dAb-PF800 conjugates (Figure 1d). The AuNR@Ag nanostructures exhibit an average length of 114.8 ± 12.5 nm and a width of 38.4 ± 3.6 nm, with a silver coating thickness of 12.8 ± 1.2 nm (Figure S1). The controlled size, shape, and surface functionalization of the PF nanolabels are critical for achieving reproducible plasmonic enhancement while maintaining consistent capillary-driven transport through the porous nitrocellulose membrane. This nanolabel uniformity supports stable p-LFA performance required for reliable quantitative detection.

3.2. Quantitative p-LFA for NGAL

We first designed and optimized a singleplex half-strip p-LFA for quantitative detection of NGAL to establish the baseline analytical performance of the p-LFA platform. The half-strip configuration, consisting of only the nitrocellulose membrane and absorbent pad, enables direct introduction of premixed analyte and detection nanolabels while eliminating variables associated with conjugate pad release and multiplexed reagent competition. Establishing the analytical sensitivity and quantitative performance in this controlled single-analyte format provides a benchmark for subsequent integration of the conjugate pad and development of the multiplexed p-LFA for simultaneous NGAL and CysC detection. NGAL capture antibody and anti-goat IgG were printed on an NC membrane to form the test line and control line, respectively (Figure 2a). Each NC strip is 2.5 mm wide, and the distance between test and control lines is 4 mm. The NC membrane was subsequently blocked with BSA to minimize nonspecific binding. NGAL dAb-PF800 and goat IgG-PF800 conjugates were used as fluorescent nanolabels. For assay evaluation, dAb-PF800 and goat IgG-PF800 conjugates were mixed with a series of NGAL standard solutions ranging from 7.8 pg/mL to 1 ng/mL in two-fold dilution, along with a blank control, and then applied to the half-strips. Upon assay completion, sandwich complexes consisting of NGAL capture antibody, NGAL, and dAb-PF800 formed at the test line, while goat IgG-PF800 conjugates were captured at the control line. Fluorescence intensities at both the test and control lines were measured 30 min after sample introduction using a benchtop fluorescence imager (Figure 2a). As expected, the fluorescence intensity at the test line increased monotonically with NGAL concentration, whereas the control line consistently exhibited strong and stable fluorescence intensity across all conditions. The control-line fluorescence intensity varied by ~14% across the tested NGAL concentration range, demonstrating relatively stable control-line performance over the analytical range.

Figure 2.

Singleplex half-strip p-LFA for quantitative NGAL detection. (a) Schematic of the half-strip p-LFA configuration and representative fluorescence images showing NGAL test and control lines. (b) Absolute fluorescence standard curves obtained using NGAL half-strip p-LFA with capture antibody concentrations of 0.5 and 1 mg/mL. (c) Absolute fluorescence standard curves obtained using NGAL half-strip p-LFA with a capture antibody concentration of 2 mg/mL, and (d) corresponding ratiometric fluorescence standard curve obtained using NGAL half-strip p-LFA. (e) Standard curve obtained using a standard NGAL ELISA for comparison.

To optimize the analytical performance of the singleplex p-LFA, we investigated the effect of NGAL capture antibody concentration on assay sensitivity. Calibration curves were generated using capture antibody concentrations of 0.5, 1, and 2 mg/mL under identical conditions (Figures 2b, 2c). Increasing the capture antibody concentration led to enhanced fluorescence intensities, improved assay sensitivity, and decreased variation in fluorescence intensities, reflecting increased capture efficiency and uniformity at the test line. The calibration curves were fitted using a five-parameter logistic model, and the lower limit of detection (LOD) and lower limit of quantification (LOQ) were calculated as the mean of the blank plus three times and ten times the standard deviation of the blank, respectively. The LOD represents the lowest analyte concentration that can be reliably distinguished from the blank signal, whereas the LOQ represents the lowest analyte concentration that can be quantified with acceptable accuracy and precision. Using an NGAL capture antibody concentration of 2 mg/mL, the p-LFA achieved high analytical sensitivity, with an LOD of 6.2 pg/mL and an LOQ of 12 pg/mL, compared with an LOD of 15 pg/mL and an LOQ of 31 pg/mL obtained using a capture antibody concentration of 1 mg/mL. In addition to validating proper assay function, the control line facilitated localization of low-signal test lines and served as an internal reference for ratiometric analysis. Using an NGAL capture antibody concentration of 2 mg/mL, the p-LFA achieved an LOD of 4.6 pg/mL and an LOQ of 7.6 pg/mL based on ratiometric fluorescence analysis (Figure 2d). The reported LOQ was confirmed by a relative error (RE) of 13.9% and a coefficient of variation (CV) of 8.4%, both satisfying the recommended acceptance criterion of ≤20% for assay accuracy and precision (DeSilva et al. 2003). The ratiometric approach provided slightly improved LOD and LOQ compared with absolute fluorescence analysis. The improved LOD and LOQ achieved by ratiometric analysis are attributed to normalization of the test-line signal to the control-line signal, which compensates for strip-to-strip variations in membrane properties, sample flow, and imaging conditions, thereby reducing measurement variability and improving analytical sensitivity. Under the optimized conditions, no evidence of fluorescence reduction attributable to excessive PF nanolabel packing was observed within the tested concentration range. We further evaluated ratiometric normalization using p-LFAs operated under different relative humidity conditions. The resulting ratiometric calibration curves remained highly consistent across the tested conditions, supporting the robustness of ratiometric analysis (Figure S2). For comparison, a standard ELISA yielded an LOD of 4.2 pg/mL and an LOQ of 4.9 pg/mL, demonstrating comparable analytical performance between the p-LFA and ELISA (Figure 2e). Based on these results, an NGAL capture antibody concentration of 2 mg/mL was selected for all subsequent experiments.

To reduce sample-to-answer time and simplify assay operation while maintaining quantitative performance, we developed a p-LFA configuration incorporating an integrated sample/ conjugate pad. In this design, the integrated p-LFA consists of an absorbent pad, an NC membrane, and a sample/conjugate pad, eliminating the need for off-strip premixing of detection reagents. As illustrated in Figure 3a, NGAL dAb-PF800 conjugates were pre-immobilized onto the conjugate pad and released upon sample introduction, enabling on-strip formation of target-detection antibody complexes prior to capture at the test line. For assay evaluation, NGAL standard solutions spanning concentrations from 125 pg/mL to 16 ng/mL, along with a blank control, were applied directly to the p-LFAs. Upon capillary flow, released dAb-PF800 conjugates bound to NGAL in the sample and were subsequently captured at the test line through sandwich immunorecognition, while excess conjugates migrated to the control line. Fluorescence images of the p-LFAs show increasing fluorescence intensity at the test lines with increasing NGAL concentration, while the control lines exhibit stable fluorescence intensity across all NGAL levels (Figure 3a).

Figure 3.

Integrated p-LFA for Quantitative NGAL detection. (a) Schematic of the integrated p-LFA configuration and representative fluorescence intensity images of NGAL integrated p-LFA strips showing NGAL test and control lines. (b) Absolute fluorescence standard curve obtained using the NGAL integrated p-LFA with a 30 min assay time, and (c) corresponding ratiometric fluorescence standard curve. (d) Absolute fluorescence standard curves obtained using the NGAL integrated p-LFA with 20 and 10 min assay times, and (e) corresponding ratiometric fluorescence standard curves obtained with 20 and 10 min assay times.

With a total assay time of 30 min, calibration curves were generated using both absolute and ratiometric fluorescence analyses (Figures 3b and 3c). The LODs and LOQs were determined to be 85 pg/mL and 0.18 ng/mL based on absolute fluorescence intensity, and 32 pg/mL and 33 pg/mL using ratiometric fluorescence analysis. For absolute fluorescence analysis, the RE and CV were 3.4% and 8.6%, respectively, whereas ratiometric fluorescence analysis yielded an RE of 7.3% and a CV of 4.1%, indicating acceptable accuracy and precision for both approaches. For comparison with a conventional LFA format, a colorimetric NGAL LFA was prepared using NGAL dAb-functionalized ~60 nm gold nanosphere (AuNS) labels deposited on the conjugate pad at a nanolabel loading comparable to that used in the PF-based LFA (Figure S3). Despite the larger size and more complex geometry of the PF nanolabels, the PF-based LFA exhibited an approximately two-orders-of-magnitude lower LOQ than the colorimetric AuNS-LFA, indicating that effective target recognition and capture were preserved following PF conjugation. We further shortened the p-LFA assay time to 20 min and 10 min to evaluate the trade-off between assay time and analytical sensitivity (Figures 3d and 3e). As expected, shortening the assay duration resulted in a progressive decrease in sensitivity, reflected in higher LOD and LOQ values relative to the 30 min assay. For the 10 min assay, the absolute calibration curve yielded an LOD of 0.35 ng/mL and an LOQ of 1.4 ng/mL, whereas the ratiometric calibration yielded an LOD of 0.18 ng/mL and an LOQ of 1.2 ng/mL. This sensitivity reduction can be attributed to reduced release of dAb-PF800 conjugates from the conjugate pad and residence time for antigen-antibody binding and capture on the nitrocellulose membrane. The improved sensitivity observed at longer assay times suggests continued accumulation of captured complexes beyond 10 min. However, quantitative NGAL detection was achieved within 10 min under the selected assay conditions. Despite the reduced sensitivity, the NGAL LOQ remained substantially below the clinical cutoff concentration of 50–500 ng/mL (Haase et al. 2009; Matarneh et al. 2025), supporting the feasibility of rapid, simplified p-LFA operation for time-critical applications. Overall, incorporation of the sample/conjugate pad significantly simplifies assay operation and reduces sample-to-answer time compared with the half-strip format, while preserving quantitative detection capability. In addition, we evaluated the stability of assembled p-LFA strips stored at 4°C for up to 7 days and observed comparable analytical sensitivity over the testing period (Figure S4). Although these results support short-term storage stability under refrigerated conditions, future work will focus on improving long-term storage stability for POC deployment. In particular, we will investigate advanced stabilization strategies based on organosiloxane and metal-organic framework materials that have shown promise in our previous studies (Guo et al. 2022; Li et al. 2021; Yin et al. 2020).

3.3. Cross-reactivity Evaluation in Multiplexed p-LFA

We next evaluated potential cross-reactivity between NGAL and CysC in the multiplexed p-LFA format to validate assay specificity prior to full multiplexed quantification. NGAL and CysC capture antibodies were patterned as two spatially separated test lines on the same NC membrane, with a fixed interline distance of 4 mm (Figures 4a, 4b). Fluorescence intensity profiling confirmed that the effective width of each test line was approximately 1 mm, ensuring no physical overlap between adjacent test lines. To assess cross-reactivity, multiplexed p-LFA strips were challenged with single-analyte samples while containing a mixture of NGAL and CysC dAB-PF800 conjugates together with goat IgG-PF800 conjugates. Serial dilutions of NGAL (500 pg/mL to 64 ng/mL) and CysC (125 pg/mL to 16 ng/mL), each prepared with a two-fold dilution and including a blank control, were applied to the strips. This experimental configuration represents a stringent test for cross-reactivity, as both detection antibodies are present while only one target analyte is supplied.

Figure 4.

Cross-reactivity evaluation in multiplexed p-LFA for NGAL and CysC detection. (a) Representative fluorescence images of multiplexed p-LFA strips exposed to varying concentrations of NGAL in the presence of both NGAL and CysC detection antibody-PF nanolabels. (b) Representative fluorescence images obtained upon exposure to varying concentrations of CysC. (c) Ratiometric fluorescence response of the NGAL test line upon exposure to NGAL and CysC. (d) Ratiometric fluorescence response of the CysC test line upon exposure to NGAL and CysC, demonstrating minimal cross-reactivity.

Fluorescence images of the p-LFAs showed a concentration-dependent increase in fluorescence intensity at the NGAL test line, with no detectable response at the CysC test line, upon introduction of varying NGAL concentrations (Figure 4a). Similarly, when varying concentrations of CysC were introduced, a monotonic increase in fluorescence intensity was observed exclusively at the CysC test line, with no measurable response at the NGAL test line (Figure 4b). Figures 4c and S5 compare the fluorescence responses at the NGAL test line when exposed to varying concentrations of NGAL and CysC. The ratiometric fluorescence intensity increased systematically with NGAL concentration and closely matched the behavior observed in singleplex p-LFAs, whereas exposure to CysC produced responses comparable to or lower than the blank control. The ratiometric fluorescence signals obtained from singleplex and multiplexed NGAL p-LFAs were compared at 1 ng/mL NGAL. The difference between the two formats was less than 2%, demonstrating that incorporation of the CysC test line did not measurably affect NGAL quantification. Under CysC-only conditions, the NGAL test-line signal showed a slight decrease at higher CysC concentrations, potentially due to reduced nonspecific PF accumulation following specific capture of CysC-dAb-PF800 complexes at the CysC test line. Based on ratiometric calibration, the LOD and LOQ for NGAL in the multiplexed format were determined to be 80 pg/mL and 0.10 ng/mL, respectively. These values are approximately 3-fold higher than those obtained in the singleplex configuration, reflecting the added complexity of multiplex operation and shared reagent transport. At the reported LOQ, the RE and CV were 11.3% and 3.9%, respectively. Similarly, Figures 4d and S6 show the fluorescence responses at the CysC test line under NGAL-only and CysC-only exposure. The CysC test line exhibited a concentration-dependent fluorescence increase exclusively in response to CysC, with minimal response to NGAL across the tested concentration range. The calculated LOD and LOQ for CysC were 0.14 ng/mL and 0.27 ng/mL, respectively, confirming quantitative detection with minimal interference from NGAL. At the reported LOQ, the RE and CV were 11.3% and 8.5%, respectively. Collectively, the fluorescence images and standard curves demonstrate negligible cross-reactivity between NGAL and CysC in the multiplexed p-LFA format. To assess signal uniformity across the membrane width, a fluorescence intensity profile was first analyzed across the entire width of a representative test line. The fluorescence intensity exhibited a broad plateau across most of the test-line width, with reduced signal observed only near the membrane edges (Figure S7a). Fluorescence intensities were subsequently quantified using regions of interest (ROI) spanning widths of 1.0, 1.5, and 2.0 mm (Figure S7b). Comparable absolute and ratiometric fluorescence responses confirmed good signal uniformity across the ROI widths (Figure S7c, S7d). The high specificity is attributed to the selectivity of the antibody pairs, spatial separation of test lines, and effective blocking conditions.

3.4. Multiplexed p-LFA for AKI diagnosis

We next tested the multiplexed p-LFA for simultaneous quantitative detection of NGAL and CysC toward AKI diagnosis. All multiplexed p-LFA measurements used for analytical characterization and clinical validation were performed using the 30-min assay protocol. To establish analytical performance in the multiplexed format, the p-LFAs were first exposed to a series of mixed NGAL and CysC standard solutions, with NGAL concentrations ranging from 500 pg/mL to 64 ng/mL and CysC concentrations ranging from 125 pg/mL to 16 ng/mL in a two-fold dilution series. Fluorescence intensities at both test lines increased monotonically with increasing concentrations of their corresponding biomarkers, confirming effective multiplexed signal generation (Figures 5a, S8). Ratiometric calibration curves were constructed for each biomarker (Figure 5b). The LOD and LOQ for NGAL were determined to be 0.12 ng/mL and 0.30 ng/mL, while the LOD and LOQ for CysC were 52 pg/mL and 91 pg/mL, respectively. To evaluate quantitative analytical capability and clinical applicability, NGAL and CysC concentrations in 45 urine samples collected from pediatric patients within 24 h of the onset of organ failure (respiratory and/or circulatory failure and/or shock) were quantified using the multiplexed p-LFA and standard ELISA assays. To assess nanolabel stability in urine, extinction spectra of dAb-PF conjugates incubated in urine were measured immediately after mixing and after 1, 2, and 20 h. No significant shift or decrease in the LSPR bands was observed, indicating that the dAb-PF conjugates remained stable under the tested conditions (Figure S9). Linear regression analysis comparing multiplexed p-LFA results with ELISA measurements showed strong correlations for both NGAL and CysC, with Pearson correlation coefficients (r) of 0.980 and 0.984, respectively (Figures 5c, 5d). To evaluate potential high-dose effects, we extended the NGAL concentration range beyond the primary calibration range and observed signal saturation but no evidence of a high-dose hook effect within the tested concentration range (Figure S10). Increasing the concentration of antibody-conjugated PF nanolabels shifted the signal saturation point toward higher NGAL concentrations, indicating that further optimization of nanolabel loading may expand the dynamic range of the assay. To assess quantification of samples above the calibration range, we performed dilution-recovery testing using a high-NGAL sample diluted from 100× to 1600×. The measured recovery rates ranged from 86.1% to 111.3%, supporting accurate quantification following dilution (Table S1). All clinical samples were diluted to fall within the validated calibration range to ensure accurate quantification.

Figure 5.

Multiplexed p-LFA for AKI diagnosis using NGAL and CysC. (a) Representative fluorescence images of multiplex p-LFA exposed to mixed NGAL and CysC standards. (b) Ratiometric calibration curves for NGAL and CysC obtained using the multiplexed p-LFA. (c) Correlation between NGAL concentrations quantified by multiplexed p-LFA and ELISA in patient urine samples. (d) Correlation between CysC concentrations quantified by multiplexed p-LFA and ELISA in patient urine samples. (e) Receiver operating characteristic (ROC) curves for AKI classification using NGAL alone, CysC alone, and combined NGAL and CysC measurements obtained by multiplexed p-LFA.

AKI adjudication was based on serum creatinine (SCr) criteria evaluated over 96 h after enrollment, with 11 of 45 patients (24.4%) classified as AKI positive. Receiver operating characteristic (ROC) analysis was performed to assess diagnostic performance using NGAL alone, CysC alone, and the combination of both biomarkers measured by the multiplexed p-LFA (Figure 5e). The area under the curve (AUC) was 0.939 (95% CI: 0.836–1) for NGAL and 0.818 (95% CI: 0.655–0.981) for CysC. The optimal cutoff concentrations were determined to be 333 ng/mL for NGAL and 702 ng/mL for CysC. Both biomarkers exhibited significantly higher concentrations in AKI patients compared with non-AKI patients within the first 24 h of enrollment (Figure S11). Importantly, combining NGAL and CysC measurements markedly improved diagnostic performance, yielding an AUC of 0.973 (95% CI: 0.905–1), with a clinical sensitivity of 90.9% and a clinical specificity of 91.2% at the optimal cutoff. These results indicate superior diagnostic performance of the multiplexed p-LFA compared with single-biomarker analysis. For comparison, ROC analysis based on ELISA quantification (Figure S12) yielded AUC values of 0.952 (95% CI: 0.862–1) for NGAL and 0.922 (95% CI: 0.809–1) for CysC, while the combined biomarker analysis achieved an AUC of 0.979 (95% CI: 0.917–1) with the same sensitivity of 90.9% and specificity of 91.2% at the optimal cutoff as multiplexed p-LFA. These results demonstrate the ability of the multiplexed p-LFA to achieve diagnostic performance comparable to ELISA in this pediatric proof-of-concept cohort while reducing sample-to-answer time in a simplified POC format. It is important to note that serum creatinine may not fully capture early structural kidney injury. Therefore, biomarker-positive/SCr-negative cases may represent kidney injury that is not captured by the reference classification. Although the multiplexed p-LFA achieved LOQs well below reported clinical cutoff concentrations (Haase et al. 2009; Nejat et al. 2010b), larger cohort studies are needed to validate its diagnostic performance and establish optimal biomarker cutoff concentrations across diverse patient populations. While the present study focused on pediatric urine samples, the analytical detection mechanism is not expected to be inherently age-dependent, as NGAL and CysC are quantified through target-specific immunorecognition. Future studies involving adult patient populations would be valuable to further validate assay performance across diverse urine matrices and clinical settings. Furthermore, longitudinal studies involving individuals at risk for AKI are needed to determine whether the assay can detect early or subclinical kidney injury prior to conventional clinical diagnosis. Such studies would also enable characterization of NGAL and CysC trajectories and provide insight into how biomarker dynamics relate to AKI onset and progression.

3.5. Evaluation of a Portable Fluorescence Reader for POC p-LFA Measurements

All p-LFA results discussed thus far were acquired using a LI-COR benchtop fluorescence scanner, which costs approximately $90,000 and is not compatible with POC settings due to its size, cost, and operational complexity. An inexpensive, portable fluorescence reader is therefore essential for translating the p-LFA platform to POC applications. To address this need, we evaluated a compact, low-cost portable fluorescence reader for quantitative analysis of p-LFA strips. The reader has dimensions of 21.5 × 16.3 × 14.7 cm and a weight of 2 kg, features an integrated touchscreen interface, delivers results within approximately 3 s, and supports up to 12 h of battery operation (Figure 6a). The projected cost of the reader at scale is approximately $200, making it suitable for deployment in resource-limited and decentralized clinical settings.

Figure 6.

Evaluation of a portable fluorescence imager for p-LFA measurements. (a) Optical images illustrating sample introduction into the p-LFA cassette and subsequent fluorescence scanning using the portable imager. (b) Representative fluorescence images of multiplexed p-LFA strips acquired using the portable imager after exposure to varying NGAL and CysC concentrations. (c) Absolute calibration curves for NGAL and CysC obtained using the portable reader, and (d) corresponding ratiometric calibration curves. (e) Correlation between ratiometric fluorescence intensities measured by the portable reader and a benchtop fluorescence imager for NGAL and (f) CysC. (g) Correlation of NGAL and (h) CysC concentrations measured in clinical urine samples using the portable reader and benchtop imager.

To demonstrate POC applicability, multiplexed p-LFAs were assembled in commercial cassettes and sequentially analyzed using both the benchtop imager and the portable reader under identical assay conditions (Figure 6a). Representative fluorescence images acquired with the portable reader (Figure 6b) show clearly resolved control line and two test lines, indicating sufficient optical sensitivity and spatial resolution for multiplexed fluorescence detection. Quantitative performance of the portable reader was assessed by generating standard curves using both absolute and ratiometric fluorescence analyses (Figures 6c, 6d). For NGAL, the LOD and LOQ obtained using the portable reader were 0.16 ng/mL and 0.54 ng/mL based on absolute fluorescence analysis, and 0.054 ng/mL and 0.12 ng/mL using ratiometric analysis. For CysC, the corresponding LOD and LOQ values were 0.18 ng/mL and 0.44 ng/mL based on absolute fluorescence analysis, and 0.17 ng/mL and 0.28 ng/mL using ratiometric analysis. The calibration curves obtained with the portable reader closely matched those of the benchtop imager (Figure S8). The LOD and LOQ achieved using the portable reader were comparable to those obtained with the benchtop imager, confirming that assay sensitivity was preserved despite the simplified optical configuration. To further validate measurement consistency, we compared ratiometric fluorescence intensities obtained from standard samples between the benchtop and portable readers. We observed excellent correlations for both the NGAL and CysC test lines, with Pearson correlation coefficients of 0.998 (Figures 6e, 6f). When applied to 10 clinical urine samples, quantification results obtained using the portable reader showed excellent agreement with benchtop measurements, yielding Pearson correlation coefficients of 0.997 for NGAL and 0.999 for CysC, respectively (Figures 6g, 6h). Collectively, these results demonstrate that the portable fluorescence reader provides quantitative performance equivalent to that of a benchtop scanner while dramatically reducing cost, size, and operational complexity. The comparable analytical performance achieved by the portable reader and benchtop scanner can be attributed in part to the strong fluorescence amplification provided by the PF nanolabels, enabling sensitive detection with both platforms. The combination of multiplexed p-LFA strips with a low-cost portable fluorescence reader enables sensitive and quantitative detection of NGAL and CysC in POC settings, supporting rapid AKI screening and monitoring outside centralized laboratories. All measurements in this study were performed under controlled laboratory conditions. Future studies will evaluate the performance of the multiplexed p-LFA and portable reader across a broader temperature range representative of decentralized testing environments.

4. Conclusion

We developed and validated a multiplexed p-LFA for rapid, quantitative, and POC detection of AKI biomarkers. By integrating ultrabright nanolabels with a ratiometric fluorescence readout, the p-LFA achieves analytical performance comparable to standard ELISA. Incorporation of an integrated sample/conjugate pad significantly reduced sample-to-answer time while providing clinically sufficient sensitivity, enabling assay completion within 10 min. The multiplexed p-LFA enabled simultaneous quantitative detection of NGAL and CysC and demonstrated strong correlation with ELISA measurements in pediatric urine samples. Importantly, combined analysis of NGAL and CysC improved diagnostic performance for AKI compared with single-biomarker assessment, achieving high clinical sensitivity and specificity and diagnostic accuracy comparable to that of ELISA-based testing. These results highlight the clinical value of multiplexed biomarker measurement for early AKI risk stratification. Finally, integration of the p-LFA with a low-cost, portable fluorescence reader enabled accurate quantitative analysis with performance equivalent to a benchtop scanner, while dramatically reducing instrument cost, size, and operational complexity. The combination of multiplexed p-LFA strips and a portable reader establishes a practical pathway toward true POC AKI diagnostics. Collectively, this study demonstrates a scalable and adaptable p-LFA platform that bridges the gap between laboratory-grade immunoassays and rapid, low-cost POC testing. Beyond AKI, the multiplexed p-LFA strategy presented here is broadly applicable to other clinical scenarios requiring sensitive, quantitative, and multiplexed biomarker detection in decentralized healthcare settings.