Comparison and evaluation of seven different bench-top flow cytometers with a modified six-plexed mycotoxin kit

Authors


Abstract

Many bench-top flow cytometers (b-FCs) are compatible with microsphere-based multiplexed assays. Disciplines implementing b-FCs–based assays are expanding; they include monitoring and validating food quality. A multiplexed platform protocol was evaluated for poly-mycotoxin assays, which is compatible with a variety of b-FC models. The seven instruments included: BD FACSCalibur, BD FACSArray Bioanalyzer, Accuri C6, Partec CyFlow® Space, Beckman Coulter FC 500, Guava EasyCyte Mini, and Luminex 100 . Current reports related to the food industry describe fungal co-infections leading to poly-mycotoxin contamination in grain (Sulyok M, Berthiller F, Krska R, Schuhmacher R, Rapid Commun Mass Spectrom 2006;20:2649–2659). It is imperative to determine whether b-FC–based assays can replace traditional single-mycotoxin enzyme-linked immunosorbent assay (ELISA). A six-plexed poly-mycotoxin kit was tested on seven different b-FCs. The modified kit was initially developed for the BD FACSArray Bioanalyzer (BD Biosciences) (Czeh A, Mandy F, Feher-Toth S, Torok L, Mike Z, Koszegi B, Lustyik G, J Immunol Methods 2012;384:71–80). With the multiplexed platform, it is possible to identify up to six mycotoxin contaminants simultaneously at regional grain collection/transfer/inspection facilities. In the future, elimination of contaminated food threat may be better achieved with the inclusion of b-FCs in the food protection arsenal. A universal protocol, matched with postacquisition software, offers an effective alternative platform compared to using a series of ELISA kits. To support side-by-side evaluation of seven flow cytometers, an instrument-independent fluorescence emission calibration was added to the protocol. All instrument performances were evaluated for strength of agreement based on paired sets of evaluation to predicate method. The results suggest that all b-FCs were acceptable of performing with the multiplexed kit for five of six mycotoxins. For OTA, the detection sensitivity was consistent only for five of the seven instruments. © 2013 International Society for Advancement of Cytometry

Back in 2003, two historians of contemporary medicine, Keating and Cambrosio declared flow cytometry (FC) a “biomedical platform” [3]. They acknowledged the impact FCs had in the field of hematopathology in the late 20th century. The FC platform is penetrating a variety of disciplines; currently, it includes applications as diverse as monitoring life in the oceans and in space [4, 5]. Food safety is another area where there are breakthroughs in FC applications. Worldwide implementation of food safety practices is critical to reduce the constant threat of food borne disease outbreaks. To guard against mycotoxins contaminated grain, b-FCs may become part of a global food safety monitoring network deployment. Much of the world's grain/crop supply is susceptible to fungal infections, which in turn can contaminate food supplies with dangerous levels of mycotoxins. The annual global grain consumption is over 1,000 million metric tons. Most of it is transported over vast distances, as the bulk of grain production is limited to nine countries [6]. During growth, harvesting, shipping and storage, a significant portion of grain crops is at some point vulnerable to mould attacks. Both single and multiple fungal infections can lead to mycotoxin(s) release [7]. The six mycotoxins individually or as co-contaminants, represent a serious global threat to human health. The detrimental health effects of tainted food are well documented [8, 9]. Accurate, robust assays are essential to detect contaminated food before it reaches consumers. During the past decade, for routine mycotoxin detection, a competitive enzyme-linked immunosorbent assay (ELISA) was considered the method of choice [9, 10]. Currently, the reference method is high-pressure liquid chromatography with tandem mass spectroscopy [11]. For the first time, a report in 2008, described poly-mycotoxin contamination in grain [11]. More recently, a competitive fluorescent microsphere immunoassay (CFIA) technology compatible with FCs entered the poly-mycotoxin detection field [2]. With increasing numbers of reports of poly-mycotoxin infections in stored crops, the introduction of an affordable simultaneous multiple mycotoxin assay capacity is timely [12]. For over a decade, multiplexed assays are available on flow cytometers for a wide variety of biomarkers [13, 14]. Luminex Corporation introduced commercial multiplexing using a b-FC. Their first dedicated instrument with dual-laser was the Luminex® 100™ System FC [13, 15]. Subsequently, this strategy became available on other b-FCs such as BD FACSArray Bioanalyzer and Guava® Personal Cell Analysis (PCA) System (Merck, Millipore). The first commercial CFIA kit, the Fungi-Plex, was developed and validated on the BD FACSArray Bioanalyzer. The objective here is to determine whether a modified version of a CFIA kit for poly-mycotoxins is compatible with most of the currently available bench-top model FCs already deployed around the world. Critical compatibility questions related to switching a competitive immunoassay from ELISA to FC using same reagents were addressed in a previous publication [2]. Sensitivity, specificity and robustness of mycotoxin assays all improved with conversion from ELISA to b-FC [2]. The multiplexed kit was incorporated into a universal platform with a four-component protocol. The question is: will a multiplexed kit work with most modestly priced b-FCs? Although all instruments considered for this evaluation meet basic criteria to quantify any of the six mycotoxins, it is yet to be demonstrated that each of them will provide satisfactory performance with a universal multiplexed reagent/software platform. The challenge is to demonstrate, how effective will a universal poly-mycotoxin detection platform work using seven flow cytometers with diverse optical, signal processing and fluidic configurations. Minimum performance inclusion criteria for a b-FC to be considered, were: (a) the ability to resolve six distinct microfluorosphere populations based on morpho-spectral attributes [2]; (b) data files for multiplexed assays are available in a standard FC format that the postacquisition software can read; and (c) all instruments must have minimum fluorescent emission sensitivity to detect as low as 480 PE molecules per microfluorosphere, see Table 1.

Table 1. QuantiBRITE PE with four fluorescence intensity values for instrument evaluation
  Evaluated instruments
IIIAaBCDEFaG
  1. The use of one calibration bead system for all seven instruments permitted instrument independent fluorescence intensity cross calibration. Column I lists the four intensities of QuantiBRITE PE microfluorospheres. Column II lists average amounts of PE molecules expressed on the four populations per microfluoropheres (data from manufacturer). In columns A–G, all values are given in mean fluorescence intensity (MFI) as calculated by the postacquisition software.

  2. a

    On instruments A and F, PMT's gain adjustment were locked for blue and green lasers, respectively; therefore, the values for those two instruments differ from the others where the gains were adjusted manually.

Low474522.0014.3918.6018.7720.1719.1721.00
Med-low53596133.00183.52196.32209.08198.1198.05238.74
Med-high2384326374.00802.23881.68905.80913.98689.141038.97
High6233673598.002187.062267.092436.232502.871476.782744.58

Materials and Methods

A comprehensive universal protocol was developed to address the wide functional diversity challenges posed by the seven b-FCs selected. The objective was to determine whether there was sufficient compatibility across instruments when measuring poly-mycotoxins with a multiplexed assay. The four components of the universal protocol are: (a) unified analytical capacity to streamline toxin extraction; (b) instrument independent cross calibration (IICC) to overcome fluorescent signal amplification diversity; (c) postacquisition software; to integrate seamlessly data generated by a variety of instrument-dedicated software; (d) a statistical component to assess the strength of agreement between instruments for poly-mycotoxin quantitation. A workflow chart was created to summarize the sequential steps, which chronologically links the four components of the platform.

Unified Analytical Section

The efficacy of a one-step extraction is critical to the success of a multi-instrument poly-mycotoxin kit evaluation. The protocol for sample preparation and the details of unified mycotoxin extraction have been previously reported [2]. It must be sensitive enough to be able to detect all mycotoxins at the maximum acceptable concentration levels where the critical cut-off points are [2]. A single extraction step was selected with the required efficacy to accommodate the side-by-side evaluation of seven instruments [2].

IICC to Manage Fluorescent Signal Diversity

There is considerable operational diversity between the seven b-FCs investigated. Traditionally, FCs have used analog signal processing, however most current generation instruments, have digital signal processing (DSP), as they can process signals faster. DSP is an advantage with multiplexing protocols. In this study, the Luminex 100 is the only b-FC sold as dedicated platform for microfluorospheres. It is designed to resolve up to 100 different microfluorosphere clusters in two-dimensional morpho-spectral space. The six other b-FCs are all hybrid instruments, meaning that they are capable of analyzing both cells or microfluorospheres. However, they are frequently sold to perform only polychromatic cell-based immunophenotyping. For some hybrid instruments, to perform microfluorosphere based assays, additional software may be required. The numerous combinations of optical filters, signal processing systems and lasers create a precarious condition where matrix effects may have impact on multiplexed poly-mycotoxin analysis. Therefore, a universal poly-mycotoxin quantitative platform is designed to yield results with uniform accuracy and precision across instruments. Certified reference materials (CRM) for all six mycotoxins are unavailable at this time. An alternative strategy had to be implemented to secure effective fluorescence emission comparability between instruments. Most instrument manufacturers market their own calibration beads, however in a situation with seven different instruments, synchronization of quantitative assay calibration is critical. Further complications with consistent fluorescent measurements were caused by the significant DSP diversities among various instruments. By cross calibrating the reporter molecule's fluorescence, all b-FCs measured PE under comparable conditions. IICC protocol for PE signal quantitation were implemented to accommodate management of signal diversity, laser excitation wavelength conditions, optical filter combinations, efficacy of fluorescent scatter detection and microfluorosphere propulsion method variations. The objective was to reduce instrument specific differences related to reporter molecule fluorescence emission quantitation. Two of the seven instruments used green and the rest were equipped with blue lasers for reporter molecule excitation. In most immunophenotyping applications, fluorescence intensity is reported as mean relative fluorescent (MRF) units. However, data generated for quantitative measurements such as in this case, mycotoxins quantities are reported in μg/kg units. Therefore, it was important to eliminate artifacts caused by fluorescent excitation variations that are due to differences between lasers at 488 nm wavelength and lasers at 532 nm as they exist on FCs such as BD FACSCalibur and BD FACSArray Bioanalyzer respectively. QuantiBRITE (QB), a lyophilized four levels, fluorescence quantitation bead system from BD Biosciences Ltd. (Erembodegem, Belgium) was used [16]. With IICC, it was possible to cross calibrate reporter antibodies bound per microfluorospheres (ABμS) for most of the instruments. The fluorescent intensity ranges expressed as PE molecules per microfluorospheres are listed for QB PE. All instruments were able to detect bellow 480 molecules of PE per microfluorosphere. Control preparations were measured at the same instrument settings, as were the labeled mycotoxins. Calculations are based on known ratios of PE to antibodies, and converting PE molecules per microfluorosphere to ABμSs. Cross calibration of results across various instruments is achieved only when the reference and quantitative calibration standards have similar excitation and emission spectra as the QBs used in the evaluation. The standardized universal protocol was designed to adjust automatically for configurational variations and to obtain comparable signals across all instruments when analyzing the same sample. This procedure is called IICC [16]. There are configurational variations among the seven b-FCs. Ranges for the seven standard curves had to be adjusted for each instrument's signal processing. Because of significant variation between dynamic ranges of fluorescence, each standard curve required adjustment to synchronize detection ranges, without compromising mycotoxin detection sensitivity. The selected mycotoxin range always incorporates the limits of acceptable values as established by EU legislation [17, 18] and/or other food safety jurisdictions. Operator selectable options are incorporated into the postacquisition software to add flexibility to the evaluation. Throughout the study, all standard curves were generated from a commercially available Fungi-Plex kit's stock standard (Soft Flow Hungary R&D Ltd., Pecs, Hungary). All standard curves are based on six levels of mycotoxin concentrations. The modified six-plexed mycotoxin kit included: ochratoxin A (OTA), aflatoxin B1 (AFB1), fumonisin B1 (FB1), T2-toxin (T-2), deoxinivalenol (DON), and zearalenone (ZEA). All acquisition protocols were executed with original software furnished by vendors. Acceptable interassay variations were established as P intercept ≥0.05 and P slope at ≥0.05.

Postacquisition Software to Convert All Data into a Consistent Format

To tabulate multiplexed results, FCAP Array v3.0 software, (Soft Flow Hungary R&D Ltd., Hungary) was developed. It permits seamless handshake between any of the seven b-FCs list-mode data files. The software is designed to detect all mycotoxins included in that specific assay up to the maximum of six. If an assay is a four-plexed run, it will automatically deal with it accordingly. There is also automatic selection of the appropriate dynamic range for standard curve generation. The postacquisition software also performs a curve-fitting algorithm to report each standard curve using a format that adjusts for instrument dependent variance found in the list-mode files. The custom software can also generate as many concentration points per standard curve as required by the kit manufacturer.

Strength of Agreement Assessment of Poly-Mycotoxins Results Between Instruments

The first commercial CFIA based poly-mycotoxins kit was developed with a hybrid b-FC [2]. This multi-instrument evaluation included a Luminex 100 System, which is the only dedicated bead based multiplexed b-FC. It was decided to consider two predicate methods in this study. Paired Bland-Altman bias plots were calculated with the BDs FACSArray and again with the Luminex 100 System, both served as reference b-FCs for the remaining six instruments. The parallel evaluations are referred to as Phase 1 and Phase 2, respectively. As an example, AFB1 mycotoxin was selected to illustrate the evaluation protocol for all seven b-FCs. Bias plot analysis is used to determine the strength of agreement between predicate and each test instrument, one instrument paired at a time.

Results

Unified Analytical Section

Based on results tabulated in Table 2, the unified extraction protocol was working in most instances. It is clear that OTA extraction was problematic as seen with instrument A and E. Aside mycotoxin OTA; all other toxins were detected with acceptable extraction efficacy for all seven instruments.

Table 2. Summary of strength of agreement evaluation for six instruments
   Instruments
 MycotoxinsABCDEG
  1. Table 2 displays paired Bland-Altman bias plot evaluation for six instruments compared to predicate method F. Columns A–G lists a summary of slope and intercept data (pIntercept: ≥0.05; pslope: ≥0.05). There are two shaded rectangles highlighting situations where the agreement was statistically weak: in Rows 3 and 4. In both instances, it was with mycotoxin OTA. Instrument A had laser gain adjustment locked. Instrument E was the only flow cytometer equipped with a single laser.

1AB1Intercept0.810.130.290.360.240.19
2Slope0.800.680.720.750.770.99
3OTAIntercept0.010.120.770.060.010.62
4Slope0.040.070.990.050.010.29
5T-2Intercept0.080.070.090.110.070.30
6Slope0.370.470.520.870.290.84
7ZEAIntercept0.900.810.220.830.070.59
8Slope0.510.500.310.720.650.33
9FB1Intercept0.380.840.910.250.530.57
10Slope0.520.720.750.750.380.64
11DONIntercept0.120.090.850.560.050.63
12Slope0.380.120.650.700.780.73

IICC to Manage Fluorescent Signal Diversity

To achieve inter-instrument comparability, the laser for reporter mAb's fluorescence emission was cross-calibrated among the seven instruments. Comparability of standard curves among instruments was achieved only for five of the seven instruments. Manual adjustment for gain on the laser exciting the reporter molecule was not possible for instruments A and F. Figure 1A illustrates the overall standard cure conformity among the seven b-FCs. Performance characteristic variations among instruments are listed in Table 3. Significant diversity manifests among average residual percentages, coefficient of response, log decades, and zero/maximum channel settings. Spectrally matched microfluorosphere reference standards adjust for variation due to these factors. Microfluorosphere populations with standard PE concentration are fixed to a channel number using a common reporting scale. Performance parameters at zero and maximum channel values are summarized in Table 3. They were all calculated from regression equations. The average residual percent is the mean of the difference in absolute percentage as the line differs from the actual data points. Coefficient of response is the slope of the regression line. The dynamic range is the ratio of channel numbers and the coefficient of response. Zero channel value is the intercept on the x-axis. The maximum channel value can be calculated at the y point = 256, using the equation related to intercept on the line. PE labeled BD QuantiBRITE microspheres served as calibration standard [19]. When using a blue laser, there is an additional requirement for adjusting fluorescence signals to eliminate the impact of RAMAN shift, which is observed with 488 nm laser emission in all aqueous solutions. It is accomplished by adjusting the gain to increase the PE signal above 50% intensity level on the relative mean fluorescent scale. Table 2 lists the strength of agreement statistics for six instruments when compared to the predicate b-FC. The impact of instrument independent cross-calibration with respect to the dynamic variation in reporter molecule's log fluorescence range is illustrated in Figure 1A. Instruments A and F, have different intensity slope characteristics, see Figure 2. These are the same instruments where there were problems with OTA detection as well. Mycotoxin AFB1 was selected as a representative mycotoxin to illustrate combined percentage reduction linear response for six concentration levels for all seven instruments, see Figure 1B.

Figure 1.

(A) AFB1 Mycotoxin standard curves with 256 chanel scale conversion for all instruments. The mean fluorescent intensity (MFI) of AFB1 mycotoxin concentration slopes are standardized with the 256 channel scale conversion. Instrument A has over seven logs compared to instrument B or E with 3.9 and 4.2 range logs, respectively. The mycotoxin concenration valuse are: 0.75, 1.5, 3.0, 6.0, 12.0, and 24.0 μg/kg representing standards 1–6, repectively. The seven instruments are split into two groups only to improve legibility. (B) The combined percentage AFB1 reduction for all instruments. Because this is a competitive assay evaluation, as mAb concentrations increase, the B/BO % deminishes. There is some impact on performance between instruments based on relative log scale compresion. The concenration valuse are: 0.75, 1.5, 3.0, 6.0, 12.0, and 24.0 μg/kg representing standards 1–6, repectively. The seven instruments are split into two groups only to improve legibility.

Figure 2.

Instrument independent cross calibration protocol with QuantiBRITE microfluorospheres. In the case of instruments A and F, the gains adjustment for PMT's were locked with acqusition software, therefore it was not possible to achieve manual adjustment/optimization. Please note that the overall performance of A and F were relatively consistant when compared with the other instruments for all four QB concentrations. The seven instruments are split into two groups only to improve legibility.

Table 3. IICC-based primary performance parameters for all instruments
  Evaluated instruments
 Performance parametersABCDEFG
  1. Because of IICC protocol implementation, it was possible to collect calibrated primary performance statistics. Critical performance parameters include; Row 1: average residual (it varied between 0.5 and 5.5%), Row 2: coefficient of response varied between instruments (35.9 and 65.9), Row 3: the log range variation was remarkably wide (3.9–7.1), Row 4: zero value settings were between 1 and 45, Row 5: maximum channel values were also diverse.

1Average residual percent0.52.01.02.25.52.10.6
2Coefficient of response35.965.963.463.563.056.964.1
3Dynamic range in log decades7.13.94.04.04.04.54.0
4Zero channel value1312425234523
5Maximum channel value137136272344182662812638792670071423786227193

Postacquisition Software to Convert All Data into a Consistent Format

Several novel features were included in postacquisition software; they are listed in Table 4. For a specific region or country the acceptable and/or legal mycotoxin limits can be added including relevant references. The software can also accommodate adjustments to add as many points as the kit manufacturer requires for generating standard curves. In this evaluation, six points are used for standard curves; see Figure 3, graphs 3A–3F. The third row in Table 4 specifies application or intended use. It can be adjusted to animal feed or human food (including children or adult), as maximum toxin level limitations may differ. The software automatically detects which mycotoxins were included in a specific poly-mycotoxin assay panel. Once the software recognizes all included toxins, it applies the suitable dynamic scale for the calculations see Figure 3. The schematics of the postacquisition software are illustrated in Figure 4.3.

Figure 3.

Software selected multiplexed standard cures using six concentration points with instrument F. Figure displays the postacquisition software generated standard curves with instrument F. Graphs 3A, 3B, 3C, 3D, 3E, and 3F represent AFB1, FB1, T-2, DON, OTA, and ZEA, respectively. All standard curves were generated with six concentration points. The Y-axes denote mean fluorescence intensity values. The X-axes denotes the concentrations of the mycotoxin standards given in ng/mL.

Figure 4.

Workflow schedule for seven instrument evaluation. The evaluation protocol is based on four components.They are described in the Materials and Methods section. Fig. 4.1 refers to the unified extraction procedure. Fig. 4.2 covers two elements: they are 4.2.1 and 4.2.2. They are.descriptons of the kit used and the calibration beads for IICC protocol respectively. Fig. 4.3 refers to postacqusition software. Fig. 4.4 describes the fourth component: strength of agreement assessment, it has two elements: Figs. 4.4.1 and 4.4.2. They include the bias plot generation and the calculations of primary performance parameters respectively. Fig. 4.4.1 includes Phase 1 and 2 of the evaluation.

Table 4. Postacquisition software for multiplexed poly-mycotoxin assay management
  Mycotoxins measured
 Software featuresAB1OTAFB1T-2ZEADON
  1. The postacquisition software includes customized features. There is an option to add applicable EU maximum permitted mycotoxin units in row 1. In row 2, it is possible to add region or country specific acceptable and/or legal mycotoxin limits. In row 3, intended application of kit can be specified. It can be adjusted to animal feed or human food (children or adult), as tolerated maximum toxin limits may vary. Row 4: is designed to accommodate the number of points used to generate standard curves. Row 5: identifies which mycotoxins were included in that specific multiplexed assay. The last row indicates the dynamic fluorescence detection range available on the b-FC.

1Indication of EU legal limits (μg/kg)2514003001001250
2Optional local or national limitsFeasibleFeasibleFeasibleFeasibleFeasibleFeasible
3Child/adult/animal use      
4Plotted points on standard curves666666
5Restriction of mycotoxins # in kitNoNoNoNoNoNo
6Identification of available log scaleYesYesYesYesYesYes

Strength of Agreement Assessment of Poly-Mycotoxins Results Between Instruments

In evaluations without certified reference materials, where a predicate instrument protocol is compared to other candidate instruments, the Bland-Altman strength of agreement statistics provides some objectivity.

Figure 2 shows instrument independent cross calibration channel positions for QuantiBRITE with individual histograms for all seven instruments. Figure 5 displays the bias plots for six paired instruments evaluated against the predicate flow cytometer. Table 5 outlines the pairing protocol used for the evaluation. The data displayed in Table 2 indicate strong agreement between instruments as they were defined in earlier in the Method section. The two shaded rectangular boxes contain the only two statistically weak data sets for the entire evaluation. They indicate that instrument A and E failed in rows three and four, respectively measuring OTA. Please note that on instrument A, manual green channel gain adjustment was not possible and instrument E was the only single laser system in the evaluation. These were two situations where the IICC protocol adjustments were incomplete. Also, the OTA standard curve had the least linear slope spread for the six-concentration range (Fig. 3E). The relationship between the two predicate methods in the evaluation is characterized in Figure 6. It is clear that there are some differences between instruments. Concerns with reliability of mycotoxin quantitation are at low-end concentrations where the critical cut-off points are. If a specimen is above acceptable maximum concentration as stated by health protection authorities, the grain is rejected. At low concentrations P values ≥0.05 for slope and intercept were obtained with two already noted exceptions, see Table 2. Although there are some differences between bias plot characteristics between Phase 1 and Phase 2, the results with either predicate method yielded acceptable agreements, see Figure 6. After completion of Phase 1 and 2, as outlined in Figure 4.4.1, workflow for statistics are calculated as indicated in Figure 4.4.2. Figure 6, indicated a horizontal line that represents a mean bias (Mean) of 4.3. It is apparent that there are bias differences between instruments B and F, however it seems that the bias remains constant with increasing mycotoxin concentrations. In Figure 5, the x-axis represents B/B0%, the horizontal lines (y) are the mean difference. The width of the line is limited to the mean difference ±2 SD. Overall, the results are consistent with previously reported quantitative efficiency for all six mycotoxins [2].

Figure 5.

Bland-Altman bias plot agreement between instrument F and the other six instruments. These graphs illustrate values obtained with mycotoxin AFB1. The multiplexed assay included all six mycotoxins, however to conserve space, only AFB1 was selected for graphical representation. The bias plots presented were generated with three sets of experiments.

Figure 6.

Bland-Altman bias plot agreement between the two predicate methods. The plot represents the mean values collected from three sets of experiments; instrument B was compared to F. It is clear that there are some performance differences between instruments; however, there is acceptable strength of agreement between them.

Table 5. Bland-Altman bias plot agreement pairing for all instruments for AFB1
Instrument codeInstrument code
 ABCDEFG
  1. Mycotoxin AFB1 was selected as a representative example for all six mycotoxins to save space. The objective is to illustrate the impact of IICC protocol on calibrated instrument performance. OK represent pairs of results where there was no statistically significant difference (P ≥ 0.05).

AOKOKOKOKOKOK
BOKOKOKOKOKOK
COKOKOKOKOKOK
DOKOKOKOKOKOK
EOKOKOKOKOKOK
FOKOKOKOKOKOK
GOKOKOKOKOKOK

Discussion

The recent discovery of frequent manifestation of multiple fungal infections in grain has compounded the global challenge of keeping grain supplies free of mycotoxins [7]. When mycotoxins in food/feed exceed acceptable concentrations, the infected crop should be destroyed [8].There is also in-vitro evidence that the presence of certain poly-mycotoxins have a synergistic toxic effect on mammalian intestinal epithelial cells [20]. Contamination with poly-mycotoxins represents a compounded threat to the food supply chain as it can lead to significant additional crop shortages. With cost effective monitoring strategy single and poly-mycotoxosis outbreaks can be avoided. In this study, there was consideration given to include parallel evaluation of various ELISA kits. However, there are significant challenges when considering side-by-side ELISA kit evaluations with a multiplexed assay. Most ELISA kit manufactures do not use identical reagents [13, 14]. Most distributors of ELISA kits will loan or lease an ELISA reader with dedicated software control and the readers are often equipped with proprietary optical filters. Such practices restrict the optical reader use to a specific ELISA kit. Because of the above listed technical challenges, ELISA kits were excluded from this evaluation. Some b-FCs are equipped with blue, whereas other brands have green as the primary laser to detect reporter molecules. This discrepancy between type of lasers onboard represents another significant obstacle for parallel multiple-instruments evaluation. PEs RAMAN scattering in aqueous solutions interferes with blue laser emission significantly. However, when blue is replaced with green laser excitation, the RAMAN scatter interference shifts to out-of-range when reading emission signal at the PE spectral zone [21]. This explains why when detecting PE fluorescence emission, instruments using less powerful green laser can match the performance of more powerful blue lasers. In Table 6, row 7 lists various maximum laser outputs. When dual laser instruments are compared, the average green laser operates with lower energy compared to blue lasers, the energy output averages are 15 and 28 mW respectively (Table 6). When the excitation is with blue lasers at 488 nm in aqueous solutions, PE generates a RAMAN scatter appearing 14 nm higher, towards the yellow range, which is the PE emission spectral zone [21, 22]. With green laser, the Stokes RAMAN shift because of aqueous interference is in the range of 80 nm upstream [23]. Therefore, the PE emission spectrum is unaffected by the RAMAN scatter shift. In this evaluation, all b-FCs with blue lasers had manually adjustable gain, see Table 6. The implementation of IICC is also critical in laboratories with more than one model of FCs operating side-by-side. The routine use of a cross calibration protocol permits switching analysis between instruments if required during normal daily operation. Additional customized postacquisition software was required to address the wide variation of log decade scales among instruments. The postacquisition software liberates the b-FC operator from dealing with restrictions, which are instrument specific. For example, regardless of laser configuration, the software performs all necessary calculations adjusting for variations automatically. FCAP Array v3.0 software recognizes both FCS 2.0 and 3.0 data files, locates microfluorosphere clusters, and determines the mean fluorescence intensity (MFI) of reporter antibodies for each mycotoxin. In summary, the postacquisition software combined with all the other features of the platform was able to generate suitable standard curves with six points for all six mycotoxins (see Figs. 3A–3F) and provided satisfactory performance with six mycotoxins for five instruments.

Table 6. Salient features of B-FCs evaluated
 FeaturesInstruments
  1. Table 6 lists available features on the seven b-FCs. To simplify the description of instruments throughout the study, alphabetical codes are introduced in row 1. Although all instruments had the capacity to work with microfluorospheres, instrument F is sold exclusively for multiplexing, whereas instrument E was developed for cell analysis only. All other instruments are designed for either application, hence the description “hybrid” instruments. Most instruments are manufactured with digital signal processing chips, only instrument C has analog signal processor, see row 5. Laser configuration is a significant variable: instrument E has single laser. B and F are equipped with green/red laser combinations; all others have blue/red lasers (row 6).Laser excitation wavelength combinations are listed in row 7. Maximum laser energy outputs are listed in row 8. Data for log scale dynamic ranges were compiled from sources provided by manufacturers or calculated as described in the method section (row 9). There are established standards for FC file structure; however as instrument F was not designed for cell analysis, the list-mode files needed some modification (row 10). Traditional flow cytometers use hydrodynamic focusing (hdf), however there are other options available such as peristaltic and capillary action for fluid transport, see A and E instruments respectively (row 11). Most instruments can be equipped with automated sample processing front-end, however instruments C, E, and G were not, see row 12.

1Instrument codeABCDEFG
2Brand/NameBD Accuri™ C6BD FACSArray BioanalyzerBD FACSCalibur™Beckman-Coulter FC-500 AnalyzerGuava® Personal Cell Analysis (PCA) SystemLuminex® 100™ SystemPartecCyFlow® Space
3Acquisition softwareBD Accuri C6BD FACS ArrayCellQuest™ ProCXPGuava Express® ProLuminex 100 IS 2.3WindowsFloMax® 2.60
4Analysis capacity (beads, cells or hybrid)HybridHybridHybridHybridCellsBeadsHybrid
5Signal processing (digital or analog)DigitalDigitalAnalogDigitalDigitalDigitalDigital
6Laser configuration (green/red, green or, blue/red)Blue/redGreen/redBlue/redBlue/redGreenGreen/redBlue/red
7Laser/wavelength488/640532/635488/635488/635532532/635488/638
8Laser/efficiency50/30 mW15/15 mW15/9 mW20/25 mW80 mW15/10 mW20/25 mW
9Log scale range7.143.884.004.024.154.503.99
10List-Mode File (standard or modified)StandardStandardStandardStandardStandardModifiedStandard
11Fluidics (hdf, peristaltic, or capillary)PeristaltichdfhdfhdfCapillaryhdfhdf
12Sample processing (manual or automated)AutomatedAutomatedManualAutomatedManualAutomatedManual

Conclusion

The need for monitoring poly-mycotoxin contamination in grain containing food products is well documented [8, 9]. In this evaluation, instruments A and E failed with the OTA assay. Based on statistics presented in Table 2, it is advisable to avoid single laser instruments and instruments without access to manual gain adjustment for poly-mycotoxin assays requiring capacity for OTA assay. IICC protocol was successfully implemented for five of the seven b-FCs. In the case of instrument A, the lack of manual adjustment caused the failure. In the case of instrument E, the powerful single green laser (80 mW) excitation may have contributed to spectral interference with PE detection of low concentrations from bead address fluorescent tags. In the case of instrument F, although the laser gain was locked, it was set not far from the required setting, therefore the OTA quantitative performance was acceptable. Five of seven b-FCs were performing well throughout the entire six-plexed poly-mycotoxin evaluation. This study focused on multiplexed poly-mycotoxin measurements with various affordable b-FCs. With the integration of IICC protocol, a level of quality-control across instruments was achieved. Instrument specific variations were recognized and managed with customized postacquisition software. It was able to analyse data from both analogue and DSP b-FCs with either blue or green lasers. In the future, elimination of poly-mycotoxins contamination and associated outbreaks threatening global food supply will possibly include the deployment of affordable flow cytometers. This study demonstrated that with comprehensive postacquisition software it is possible to achieve “handshake” application with a variety of compact b-FCs. This study presents evidence how applications of diverse instrumentation with a universal protocol is possible. A rigorous platform protocol with stringent quality control elements can address instrument diversity effectively. Based on the results presented here, it is important to consider in the future, multi-centric studies to demonstrate true robustness of this platform.A universal assay platform with capacity for further diversification may widen acceptance of b-FC as an indispensable global tool to maintain the safety of grain-containing food supplies. Evidence is increasing that there is bio-transfer of mycotoxins through the food chain threatening the safety of meat and dairy products [24]. The most vulnerable groups are infants, and pre-pubertal girls [25]. Mycotoxicoses caused by DON and ZEA can affect the health of several species of fowl and non-ruminant mammals, including turkeys and pigs [26, 27]. The current multiplexed poly-mycotoxin assays could be modified so that in the future it will detect poly-mycotoxicoses in birds and mammals by analyzing blood, urine and organs such as liver to detect and eliminate poly-mycotoxicoses.

Ancillary