Development of bead based multiplexed immunoassay for evaluation of midkine, syndecan-1, and ANGPTL4 in patient serum

Abstract

Background:

Angiogenesis is associated with tumor progression in a range of malignancies. Herein, we develop custom immunobead assays for several mechanistically important targets and evaluated these against sera from cohorts of non-small cell lung cancer (NSCLC) patients.

Methods:

Antigen “capture” antibodies for midkine, syndecan-1, and ANGPTL4 were independently conjugated to MagPlex^® Microspheres using standard carbodiimide/NHS-based chemistry. These reagents served as the basis for quantitative sandwich assay assembly using biotinylated detection antibodies and R-phycoerythrin-conjugated streptavidin reporter system. Standard curves were created using dilution series of recombinant target proteins with assay performance characteristics calculated, accordingly. Finally, we evaluated a range of serum samples from NSCLC patients (n = 32) to verify assay performance.

Results:

Multiplexed assays for midkine, syndecan-1, and ANGPTL4 were developed with three orders of magnitude in dynamic range, excellent intra- and inter-assay precision, and accuracy parameters (<10%, and <15% variability, respectively). Detection and quantifications limits were suitable for the three assays to efficiently evaluate sera across a range of disease stages with a four-fold dilution factor.

Conclusion:

We successfully developed and analytically validated a 3-plex immunobead assay for quantifying midkine, syndecan-1, and ANGPTL4 in patient sera. This multiplexed assay will provide an important tool for future studies delineating the role of angiogenesis in lung cancer progression.

Introduction

Disease pathogenesis involves interplay of different cytokines to induce the multiple biological phenomena of disease. In cancer, cytokines can also create tumor microenvironments that contribute to tumorigenesis and enhance tumor cell survival throughout disease progression. Therefore, the ability to efficiently and accurately evaluate these factors can provide important mechanistic information as well as potential diagnostic and prognostic tools.

Angiogenesis factors are linked to cancer progression in a wide range of malignancies, including breast, colorectal cancers, and lung cancers. Although many angiogenesis biomarkers have been discovered and characterized as to their association with tumor progression,^[1–4] there are a limited number of commercial assays available to evaluate these targets in patient sera. With this, the purpose of this study was to develop and analytically validate tools to evaluate angiopoietin like protein-4 (ANGPTL4), midkine, and syndecan-1 given the connection these cytokines have to mechanisms such as angiogenesis, cancer metabolism, and epithelial to mesenchymal transition (EMT).^[5–12]

A multi-biomarker approach is generally a more robust means of understanding biological events than a single biomarker approach. Due to an increased interest in integrating more pathways-based approach to understand biological processes, multiplex immunoassays for detection of range of analytes of interest is gaining popularity in multiple areas of biomedical research.^[13–15] Bead-based immunoassays allow parallel quantitative and semi-qualitative analysis of multiple targets with a unique combination of features, including rapid data acquisition, excellent sensitivity and specificity, cost effectiveness, and multiplexed comparative analysis capabilities.^[14,16]

In this study, we develop and analytically validate multiplexed immunobead assays in the Luminex platform to provide a means to evaluate circulating levels of ANGPTL4, midkine and syndecan-1 in patient sera, given the lack of commercially available assays at the time this study was initiated. We also demonstrate how to evaluate cross-reactivity in multiplexed assay development, and contrast performance of these assays in multiplexed and single-plex formats.

Materials and methods

Materials

Three distinct bead regions (MC10025-01, MC10072-01, and MC10056-01) of MagPlex^® Microspheres were purchased from the Luminex Corporation (Austin, TX). All immunoreagents and recombinant proteins were obtained from R&D Systems (Minneapolis, MN), which included the following: “capture” antibodies—Human Midkine Antibody (AF-258-PB), Human Syndecan-1/CD138 Antibody (AF2780), and ANGPTL4 (AF3485); recombinant proteins (as standards)—Midkine (258-MD-010), Syndecan-1 (2780-SD-050), and ANGPTL4 (4487-AN-050); and biotin-conjugated “detection” antibodies—Human Midkine Biotinylated Antibody (BAF258), Human Syndecan-1 Biotinylated antibody (BAF2780), and Human ANGPTL4 biotinylated antibody (BAF3485). The following reagents were obtained from ThermoFisher Scientific (Waltham, MA), including sulfo-N-hydroxysuccinimide (sulfo-NHS), 1-ethyl-2-(dimethylaminopropyl) carbodiimide (EDC), R-phycoerythrin-conjugated streptavidin (SAPE), and 2-(N-morpholino) ethanesulfonic acid (MES). All remaining reagents were obtained from Sigma-Aldrich (St. Louis, MO).

Magnetic bead activation and “capture” antibody conjugation

Each polyclonal “capture” antibody was conjugated independently to a distinct set of MagPlex^® Microspheres (Luminex Corporation, Austin, TX) using conventional carbodiimide/NHS chemistry for amine-specific reactivity and Luminex recommended protocols.^[17] Briefly, 1.25×10⁶ beads from each set of microspheres were washed twice with HPLC grade water and re-suspended in 20 μL of 100 mM monobasic sodium phosphate (pH 6.2) with gentle shaking for 20 s. Then, 2.5 μL each of 50 mg/mL sulfo-NHS and 50 mg/mL of EDC were added sequentially and incubated at room temperature for 20 min with gentle mixing. Mixtures were then placed on magnetic separator and the supernatants removed, followed by three cycles of washing and resuspension in 100 μL of 50 mM MES, pH 5.0 with gentle mixing. Beads were then resuspended in 25 μL of 50 mM MES, pH 5.0 and had 5 μg of the appropriate capture antibody added and the total mixture made up to 125 μL of 50 mM MES, pH 5.0 and incubated in dark overnight with rotation at 4°C. After the conjugation reaction was complete, the beads were washed and resuspended in 200 μL blocking buffer (PBS containing 1% BSA (w/v) and 0.05% sodium azide (w/v)) and incubated for 30 min. The beads were then washed three times in this manner and finally placed in 200 μL of PBS containing 1% BSA and 0.05% sodium azide, beads counted, and stored in the dark at 4°C.

Confirmation of coupling efficiencies and assay performance

The efficiency of capture antibody conjugation was directly evaluated for each batch of beads using the appropriate biotin-conjugated, anti-species antibody. Briefly, this consisted of incubating 5,000 coupled microspheres (resuspended in 100 μL of PBS containing 1% BSA) per well with 50 μL of 4 μg/mL of the biotin-conjugated, anti-goat antibody (Santa Cruz Biotechnology, Dallas, TX) in PBS containing 1% BSA in a 1.2 μm polyvinylidene difluoride filter 96-well microtiter plate (Thermo Fisher) for 4 h at room temperature. This incubation was followed by washing each well three times with PBS containing 1% BSA. Beads were then resuspension in 100 μL of PBS containing 1% BSA and incubated for 30 min with 50 μL of 4 μg/mL SAPE, with constant agitation. The beads were then all washed, as before, and resuspended in PBS containing 1% BSA and read on our FlexMap 3-D^® system (Luminex Corporation, Austin, TX) with a minimum of 50 bead reads per well. Median fluorescence intensity (MFI) values are reported for the conditions to permit the ongoing evaluation of coupling efficiencies.

For assay performance, all single-plex assays were conducted in 1.2 μm polyvinylidene difluoride filter 96-well microtiter plates (Thermo Fisher) using 5,000 coupled microspheres per well with continuous plate vortexing for all incubations. All assays (single-plex and multiplex) were conducted with a minimum of triplicate reads for determination of analytical performance characteristics. Also, between all steps detailed below the beads were washed 3x with 100 μL of PBS containing 1% BSA, as described above. A seven point serial four-fold dilution series for each protein was performed on each plate as a standard curve, starting from 2 μg/mL for ANGPTL4, 0.05 μg/mL for syndecan-1, and 0.5 μg/mL for midkine; with 1× PBS-1% BSA serving as the background control and diluent. After capture of the analytical standard, each well was incubated with 50 μL of 4 μg/mL of the appropriate biotin-conjugated detection antibody (as specified in “Materials”) for 30 min at room temperature. The plates were then developed upon incubation with 4 μg/mL SAPE for 30 min with constant agitation and read, as indicated above. All assays were conducted a minimum of three times per batch for assay performance characteristic determinations.

For multiplexed assay, each well was loaded with mixture of beads with 5,000 beads represented for each assay. Assays were performed identically to that performed above, with exception to the use of mixtures of standard proteins and detection antibodies. “Leave one out” procedures were conducted as modifications of this procedure and, again, and consistent with the recommended by the Luminex Corporation.

Determination of assay performance characteristics

A series of standard curves were performed no less than three times for each batch of beads, both in single-plex and in multiplex, in order to calculate and assess assay performance characteristics according to the Clinical and Laboratory Standards Institute guide;^[18–20] including precision, limits of quantitation/detection (assay range), and accuracy. Using the concentrations calculated from the standard curves, accuracy and precision of the assays were evaluated for each analyte. Precision was investigated by assessing variation within categories of assay performance variation: intra-assay variation and inter-assay variation. Coefficients of variation (CV) were calculated for estimated concentrations within each category of variation using standard methods. Assay range was determined by Upper Limit of Detection (ULOD) and Lower Limit of Detection (LLOD), and Upper Limit of Quantitation (ULOQ) and Lower Limit of Quantitation (LLOQ). ULOD is defined here as the highest MFI signal produced by assay. The LLOD was defined as the lowest MFI signal that distinguish assay from the blank and was calculated by LLOD = LOB + 1.645(SD_{low conc}.), where lower limit of blank (LOB) is the highest apparent analyte signal expected to be found in replicate of blank.^[22] The ULOQ and LLOQ values were defined as the upper and lower instances where the accuracy between an expected concentration and measured concentration were within 10% of the actual concentration. To determine LLOQ, for each analyte, we defined a precision profile, which is a plot of the CV (defined as the ratio of the estimated standard error over the estimated concentration) versus the estimated concentration. A maximum CV of 10% was our internal threshold for the limit of quantitation. Starting from the lowest concentration for each analyte, the lower LLOQ was defined to be the point at which the %CV dips below 10%, while the ULOQ was defined to be the point at which the %CV rises above the 10% threshold.^[22]

Cross-reactivity was calculated as percentage of difference in signal produced by an analyte in multiplex from signal produced in single-plex format on the same run and using same concentration, as follows: Cross-reactivity % = 100X change in MFI signal of target in multiplexed format/MFI signal from same target in single-plex assay in the same plot. In this experiment we used the highest and the lowest concentrations from each curve. In order to evaleuate any effect of cross reactivity, or changes in MFI values between single-plex and multiplex formats, we compared the accuracy of both formats in measuring levels of cytokines in samples with known consternations of analytes.

Comparison of assay performance: “in-house” Luminex versus commercially ELISA and Luminex

An evaluation of the performance characteristics for the assays developed here (as described above) with the published parameters for commercially available ELISA and Luminex assays was performed. For this, we contrasted the performance of the identical immunoreagents we used in the development of the Luminex assays in the ELISA format (i.e. Human Angiopoietin-like 4 DuoSet ELISA^® (DY3485), Human Syndecan-1 DuoSet ELISA^® (DY2780), and Human Midkine DuoSet ELISA^® (DY258) assays by R&D systems), as well as include parameters from similar assays in the ELISA format for comparison. These include the Human Midkine ELISA Kit (ab193761), Human ARP4 ELISA Kit (Angiopoietin-Like 4) (ab99974), Human Syndecan-1 ELISA Kit (CD138) (ab46506) by Abcam; all comparisons for ELISA assays with the novel Luminex assays are shown in Table 4. Similarly, published performance characteristics from commercial Luminex format assays manufactured by R&D Biosystems and EMD Millipore were also contrasted to the assays developed in this study, as shown in Table 5.

Table 4.

Comparison of performance characteristics for our “in house” multiplexed assays with commercial Luminex platform assays.

	“in-house” multiplexed assay		R&D		EMD Millipore
	Sensitivity (ng/mL)	Range (ng/mL)	Sensitivity (ng/mL)	Range (ng/mL)	Sensitivity (ng/mL)	Range (ng/mL)
ANGPTL4	0.489	1.9–2000	0.086	1.8–438	1.316	5.49–4000
Syndecan-1	0.017	0.012–12.5	0.015	0.173–42.1	NA	NA
Midkine	0.174	0.122–500	0.006	0.0428–10.4	0.0137	0.0137–10

Table 5.

Comparison of performance characteristics for our “in house” multiplexed assays with commercial ELISA platform assays.

	“In-house” multiplexed assay		R&D		ABCAM
	Sensitivity (ng/mL)	Range (ng/mL)	Sensitivity (ng/mL)	Range (ng/mL)	Sensitivity (ng/mL)	Range (ng/mL)
ANGPTL4	0.489	1.9–2000		1.25–80	0.02	0.0274–20
Syndecan-1	0.017	0.012–12.5		0.125–8	2.56	8–256
Midkine	0.174	0.122–500		0.0781–5	0.0035	0.0156–1

Evaluation of assay performance in patient serum samples

Serum samples from 32 patients representative of control populations from our lung cancer screening program and several discrete stages in lung cancer progression, including, benign nodule (n = 8), T_1–2N₀M₀ (n = 8), T_1–2N_1–3M₀ (n = 8), and T_1–2N_1–3M₁ (n = 8) were accessed from the Rush University Cancer Biorepository. All specimens were collected with written informed consent and with full IRB approval. All clinical specimens were evaluated using methods defined in the “Assay performance and microsphere validation” section above. We evaluated a dilution series of these specimens to assess the optimal dilution factor, as follows: 1/2, 1/5, 1/10, and 1/50. We determined the optimum dilution factors based on which dilution provided the highest MFI signal that falls within the range of assay quantitation while minimizing amount of serum consumed. Then, the distributions observed for these groups at the optimized assay conditions were illustrated through “Box and Whisker” plots using SPSS v22.0 (IBM Analytics, Armonk, NY), as previously described.^[1,3,21]

Results

Assay range and limitations

Multiple concentration curves were performed for each target analyte in order to identify the optimum assay range for in single-plex format before we multiplex all curves. Results of defined assay standard concentrations, range, and limitations are listed in Table 1 and illustrated in Figure 1. ULOD was measured at 2,000, 50, and 500 ng/mL for ANGPTL4, syndecan-1, and midkine, respectively, in both single- and multiplexed format. These figures are defined by the maximum signal obtained before plateau phases of the standard curves were observed (see Figure 1). The LLOD values were calculated as described in the ‘Methods’ section and determined to be 0.449, 0.01744, and 0.174 ng/mL for ANGPTL4, syndecan, and midkine, respectively, in multiplexed format, compared to 0.488, 0.01143, 0.1715 ng/mL, respectively in single-plex format. Detection ranges were almost identical between two formats. As expected, the MFI values weren’t identical between two formats, though each achieved approximately three orders of magnitude.

Table 1. Multiplex and single-plex assay ranges and limitations.

Lower limit of detections LLOD, upper limit of detections ULOD, lower limit of quantification LLOQ, and upper limit of quantification ULOQ are showin in ng/mL with corresponding average MFI values (in parenthesis).

	Multiplex ng/mL (ave MFI)
	LLOD	ULOD	LLOQ	ULOQ
ANGPTL4	0.489 (133)	2000 (24,686)	1.953	2000
Syndecan	0.0174 (200)	50 (22,250)	0.0122	12.5
Midkine	0.174 (117)	500 (16,755)	0.122	500
	Single-plex ng/mL (ave MFI)
	LLOD	ULOD	LLOQ	ULOQ
ANGPTL4	0.4880 (121)	2000 (20,737)	1.953	2000
Syndecan	0.01143 (198)	50 (21,564)	0.0122	12.5
Midkine	0.17150 (100)	500 (14,553)	0.122	500

Figure 1.

Single-plex and multiplex standard curves. Single-plex curves for ANGPTL4 (A), midkine (B), and syndecan-1(C). Standard curves in multiplexed format (D) average MFI is depicted corresponding to seven serial dilution concentration of each target. Standards in multiplexed format (STDs) constitute standard concentrations of each of the three analytes.

Assay accuracy

To evaluate the accuracy of the assays two known concentrations of each analyte in 1× PBS-1% BSA were measured by both single-plex and multiplex formats. The overall averaged accuracy was 98.81%, with figures ranging between 95% and 105.6% (see Table 2). There was an agreement between the accuracy of two assay formats, with no effect of multiplexing on the accuracy of measurement as compared to single-plex assay.

Table 2. Evaluation of accuracy in single-plex and multiplex assay.

For each analyte two samples with known concentrations evaluated by single-plex and multiplex assays. Accuracy in percent measured as percentage of measured values to actual values of concentrations.

	Reference Concentration (ng/mL)	Single-plex Value (ng/mL)	%	Multiplex Value (ng/mL)	%
Midkine	1.465	1.416	96.69	1.485	101.39
	23.437	22.613	96.48	23.222	99.08
ANGPTL4	5.859	6.1914	105.67	5.632	96.13
	93.750	96.465	102.90	89.140	95.08
Syndecan	0.586	0.590	100.69	0.566	96.65
	9.375	9.011	96.544	9.226	98.41

Assay precision

Relative standard deviation was calculated based on two concentrations from each analyte at a low (LC) and high (HC) concentration value to determine intra- and inter-assay variabilities. These reference values were 1.953 and 500 ng/mL for ANGPTL4; 0.488 and 12.5 ng/mL for syndecan-1; 0.488 and 125 ng/mL for midkine. Precision was examined by determining the coefficient of variation (CV) for each sample which is defined as the (standard deviation/average) × 100. CVs ≤15% were considered acceptable for intra- and inter-assay precision with an overall CV of 9.59% observed for intra-assay and 5.80% observed for inter-assay variations, as shown in Table 3.

Table 3. Assay precision represented by intra and inter assay coefficient of variance CV%.

Low (LC) and high (HC) concentration values correspond to 1.953and 500 ng/mL for ANGPTL4, 0.488and 12.5 ng/mL for syndecan-1, 0.488 and 125 ng/mL for midkine and CV% measured in intra- and inter- assay variabilities.

	Multiplex CV%
	intra-assay		inter-assay
	LC	HC	LC	HC
ANGPTL4	4.90	6.50	11.00	11.30
Syndecan-1	6.50	4.50	7.20	6.50
Midkine	7.00	5.40	9.96	11.60

Cross-reactivity between each protein and captured beads

Cross-reactivity between analytes (i.e. recombinant proteins) and other capture antibodies in each multiplex assay were evaluated in this section. For this, we incubated each analyte separately with multiplexed capture beads followed by incubation with the corresponding detection antibody. MFI values produced by each analyte specific run were then compared to MFI values produced from the single-plex version of the assay. There were negligible changes (<3%) in MFI signal in case of syndecan and midkine when incubated with multiplexed beads compared to their single-plex assays. For ANGPTL4 there was 33% decrease in signal intensity in the single-plex format. However, this alteration was accompanied by only minor increases in the signal intensities for syndecan and midkine (2.4% and 3.5%, respectively), indicating the cross-reactivity is not likely to impact the analytical robustness of the multiplex assays. That is, this observation was only observed with high end of assay curve, which as will be seen below, are outside of the working range of the assay for clinical specimens. Signals in this range were within acceptable threshold of differences (i.e. <2%) at lowest point of each curve.

Cross-reactivity between detection antibodies and assay complexes

Cross-reactivity between target cytokines and other detection antibodies were determined by incubatating multiplex capture bead-analyte complexes with each detection antibody individually and compared the MFI produced by each assay to its respective blank, single-plex. No significant changes in MFI values were observed in case of syndecan detection antibodies when incubated across other multiplexed analytes, indicating negligible cross-reactivity values. In case of midkine detection antibody, there was 14% decrease in the MFI signal accompanied with 4.5% increase in MFI resulted from its reaction with blank ANGPTL4 compared to ANGPTL4 blank. There was very little decrease in the MFI signal in ANGPTL4 with 2% change and 4.6% increase of signal from midkine compared to its blank. With this, again we determined this observation to be of little importance to the analytical robustness of the multiplex as this observation was only observed with high end of assay curve, which is outside of the working range of the assay for clinical specimens.

Evaluation of “in house” developed multiplex assay against other commercially available assays

In terms of assay performance, the “in house” developed multiplexed assays had very comparable sensitivity figures, overall, as shown in Table 4. The dynamic range for the “in house” developed assays, however, were generally broader, with similar LLOQ figures (i.e. 1.9, 0.012, and 0.122 ng/mL for ANGPTL4, syndecan-1, and midkine, respectively), but with ULOQ being 25-fold higher than the R&D Systems figures and 100-fold higher than the EMD Millipore figures in the case of ANGPTL4. In the ELISA format the assay performance figures were again comparable to the “in house” assays, as shown in Table 5. Interestingly, the identical immunoreagents from R&D Systems had much broader dynamic ranges in the Luminex format than the ELISA format, and either comparable or slightly better LLOD figures.

Preliminary evaluation of multiplex assay against patient specimens

In order to evaluate the potential utility of the developed assay in measuring the target analytes in clinical specimens, sera from 32 cases of non-small cell lung cancer (NSCLC) and relevant controls were obtained from our institutional biorepository for evaluation. Upon initial evaluation of a series of dilution factors, we determined 1:5 as our optimum dilution factor—based on that being the value where we achieved MFI values that fall within the working assay range while minimizing sample expenditure. The use of dilution factors less than 1:10 are highly discouraged as very low signal intensities were observed that were frequently below the calculated LLOQ. On the other hand, 1:2 dilution still fit in the middle of assay curve which gives flexibility to choose which dilution to use. Using neat samples was not in our scope as we aim for optimum signal while saving patient sample, therefore we did not use neat sample concentrations. Patient groups (n = 8 each) representative of patients with solitary (benign) pulmonary nodules, stage I (T_1–2N₀M₀) NSCLC, stages 2–3 (T_1–3N_1–3M₀) NSCLC, and stage IV (T_1–3N_1–3M₁) patients. The range of measured levels of three targets in ng/mL in serum samples with reproducibility values are shown in Table 3 and depicted with corresponding standard curves in box in Figure 2. All of these ranges fit within quantification range of assays as in figure. The results from 1:5 dilution generally fits in the 2^nd quadrantile of assay range and provides excellent linearity with the range of specimens observed from the patient groups.

Figure 2.

Range of measured concentrations in serum. Standard curves showing standard concentrations (X axis) in each assay with measured levels in serum (Y axis). The highlights in the middle of each curve represent the range of measured cytokines levels in patients serum as following: (A) ANGPTL4 (99–952 ng/mL); (B) midkine (1.7–22.4 mg/mL); and (C) syndecan-1 (0.042–6.6 ng/mL).

Discussion

Multiplexed immunobead assays provide a robust platform for quantifying multiple protein analytes in a range of clinically-relevant biological matrices. In circumstances where commercially available assays are not available, the Luminex platform is superior to many other multiplex immunoassay platforms given the relative ease in which custom assays can be developed and without the need for additional specialized instrumentation necessary to do so. In this study we detail the development of a 3-plex assay to detect angiogenesis-associated analytes in serum.

Although multiplexing allows comparative profiling of biologically related cytokines, interference driven by reagents or sample components may lead to misleading results that would be difficult to identify without proper analytical controls and, thus, should be carefully addressed in all bioanalytical laboratories.^[15] Herein, we addressed the issue of cross-reactivity between some of assay components and defined the degree and source of cross-reactivity observed in these assays. Across the assays being developed, nonspecific binding was never found to be an issue (i.e. ≤1% of total MFI) within same assay components (i.e. between detection and capture antibodies of same assay) or between detection antibodies and capture antibodies across multiplexed assay. However, there was general decrease of MFI signal in multiplexed assay compared to single-plexed. This led to focus on the recombinant target protein interaction with antibodies. A higher MFI signal resulted from running ANGTPL4 protein across other two capture antibodies compared to their background with 4.5% and 3.5% fold increase in case of syndecan and midkine, respectively. We determined this effect to be of negligible importance, overall, since concentration determinations in the multiplex format were compared to those in the single-plex, as shown in Table 3.

Published product sheets from the immunoreagent manufacturer, R&D Systems, indicate that ANGPTL4 does not have appreciable cross-reactivity with angiopoietin-1, angiopoietin-2, angiopoietin-4, and angiopoietin-like 3;^[22] whereas syndecan-1 was similarly tested across ADAMTS4, CD44/Fc chimera, COL23A2, eotaxin, GROα, HGF, MCP-3, midkin, MMP-7, RANTES, syndecan-2, syndecan-3, syndecan-4, and TARC with no observed cross-reactivity;^[24] and finally, midkine was tested against ALK-1/Fc chimera, LRP-1 cluster II/Fc chimera, LRP-1 cluster III/Fc chimera, LRP-1 cluster IV/Fc chimera, and pleiotrophin with no reported cross-reactivity.^[24]These published assessments provide some appreciation of the analyte specificity for the immunoassays developed herein.

Although both ANGPTL4 and midkine showed some degree of cross-reactivity between their components, these observations did not lead to significant change in signals from nonspecific reactions while signal from target assay diminished. However, this effect was only observed at high end of detection range (i.e. at the ULOD, and ULOQ)—which were values of negligible impact to the range in which clinical specimens at the 1:2 or 1:5 dilution factors were observed. Furthermore, we showed that the change in MFI did not affect the accuracy of measurement in multiplexed assay for samples fall within the detection range (see Table 2). With this, we determined the assays to be suitable for concentration determinations of clinical specimens in the multiplex format. As a side note on this idea, commercially available kits for the target analytes have been introduced to the market in the Luminex format since the time of project inception by R&D and EMD Millipore.^[25–30] Overall, our “in house” assays showed a generally broader dynamic range, which provides the ability to detect circulating levels of the target analytes across a larger spectrum of clinical specimens compared to the commercial assays. Further, the assay featured in this report also used standard curves aimed at evaluating serum specimens based on four-fold dilutions versus the twofold dilution used commercially-marketed assays,^[25–27] thus advantages in sensitivity also translates to greater conservation of clinical specimens (see Table 4 and Table 5).

Conclusions

A custom 3-plex assay of ANGPTL4, syndecan, and midkine was successfully developed and analytically validated in this study. The multiplexed assay was capable to measure circulating levels of these target analyses in samples with a high degree of accuracy, precision, and over 3 orders of magnitude with low interference. Further, this multiplexed assay was also qualified for the evaluation of these analytes in patient sera, in preparation for future studies evaluating their clinical significance.