Validation of PhageDx™Salmonella Assay in Raw Ground Turkey and Powdered Infant Formula: AOAC Performance Tested MethodSM 121904

Abstract Background The PhageDx™Salmonella Assay is based on the infection of Salmonella spp. by specific bacteriophages and expression of a luciferase reporter gene. Results are generated in as little as 9.5 h for raw ground turkey and 18.5 h for milk-based powdered infant formula (PIF). Objective An AOAC Performance Tested MethodsSM (PTM) study was conducted to validate the PhageDx Salmonella Assay for the detection of Salmonella in 25 g raw ground turkey and 100 g PIF test portions. Method The performance of the PhageDx Salmonella Assay was compared to that of the U.S. Department of Agriculture (USDA), Food Safety and Inspection Service (FSIS) Microbiology Laboratory Guidebook (MLG) 4.10 for raw ground turkey and the U.S. Food and Drug Administration (FDA) Bacteriological Analytical Manual (BAM) Chapter 5 for PIF. Inclusivity/exclusivity, product consistency and stability, and robustness testing were conducted. Results There was no significant difference between the 25 g raw ground turkey and 100 g PIF PhageDx Salmonella Assay and the USDA/FSIS MLG 4.10 and FDA/BAM Chapter 5, respectively. The reporter bacteriophages were specific for Salmonella and infected 108 strains in inclusivity testing. They did not infect 30 non-Salmonella bacteria in exclusivity testing. Robustness testing showed that the method performed well with specific deviations from the standard protocol. Consistency and stability testing demonstrated that the recombinant phage gave consistent results across three production lots and was stable when stored under appropriate conditions for at least eight months. Conclusions The data collected in the validation study demonstrate that the PhageDx Salmonella Assay meets the qualifications for PTM status. Highlights The PhageDx Salmonella Assay is a rapid, specific, sensitive assay capable of detecting a wide range of Salmonella spp. with a significantly shorter turn around time than the USDA/FSIS and FDA reference methods.

in an estimated $3.7 billion in medical costs each year (4). The most common symptoms of a Salmonella infection include diarrhea, fever, and abdominal cramps and many recover without treatment. However, some cases can be so severe that they can result in hospitalization or death. The Centers for Disease Control and Prevention estimates that Salmonella causes about 1.2 million illnesses, 23 000 hospitalizations, and 450 deaths in the United States every year. Contaminated food accounts for about 1 million of these illnesses. In 2019, there were several outbreaks linked to papayas, tahini, raw tuna, melon, and ground turkey (5). In addition, the World Health Organization has stated that Salmonella contamination in powdered infant formula (PIF) remains a problem in many parts of the world (6).

Principle
The PhageDx Salmonella Assay is based on the infection of Salmonella spp. by bacteriophages and replication of the infecting bacteriophages within their specific hosts. Bacteriophages demonstrate a high specificity for their bacterial host and are capable of replicating within their host quickly to high numbers. The recombinant phages used in the PhageDx Salmonella Assay also express a luciferase reporter during replication. The presence of Salmonella spp. is determined by incubating the lysate with the appropriate luciferase substrate and detecting emitted light in a luminometer. An absence of detected light indicates that no Salmonella are present in that sample. An additional advantage of this system is that only viable bacteria cells are detected as bacteriophage only replicate in living cells.

Definitions
(a) Probability of detection (POD).-The proportion of positive analytical outcomes for a qualitative method for a given matrix at a given analyte level or concentration. POD is concentration dependent. Several POD measures can be calculated: POD R (reference method POD), POD C (confirmed candidate method POD), POD CP (candidate method presumptive result POD), and POD CC (candidate method confirmation result POD). (b) Difference of probabilities of detection (dPOD).-Difference of probabilities of detection is the difference between any two POD values. If the confidence interval of a dPOD does not contain zero, then the difference is statistically significant at the 5% level.

Test Kit Information
For raw ground turkey:

General Preparation
(a) Prepare BPW media according to manufacturer's instructions. (b) Before using the reagents, flick or spin the tube to collect all of the solution at the bottom of the tube. (c) Due to the short enrichment times, it is vital to maintain the temperature of the sample and BPW media used in the enrichment incubation. (d) Before adding the pre-warmed BPW to the sample, confirm that the media and incubator are warmed to 37 6 1 C (PIF) or 41 6 1 C (raw ground turkey). (e) Do not allow the pre-warmed media to cool before adding to the sample.
(f) Maintain the media at 37 6 1 C or 41 6 1 C in an incubator or water bath if preparing multiple samples. (g) Prepare the Promega luminometer by following the manufacturer's cleaning procedure and make sure there are no signal "hot spots" that will affect the sample results by reading an empty plate. All signals should be less than 20 relative light units (RLUs). Set up the luminometer to read the appropriate wells on the plate and set the signal integration to 1 second with a 180 second delay between starting the program and the signal read.

Sample Preparation
Raw ground turkey (25 g test portion): (a) Weigh 25 g of raw ground turkey and place into a filter sample bag. (b) Add 75 6 5 mL pre-warmed (41 6 1 C) BPW to the sample. (c) Homogenize sample in a Stomacher 400 or equivalent.
Alternatively, mix by hand. (d) Loosely close the sample bag and place in a static air incubator at 41 6 1 C for 7-18 h using a sample rack to keep the bags separate and allow heat transfer. (e) Remove the enriched samples from the incubator and mix thoroughly by hand for 15-30 s and immediately proceed to the next step after mixing is completed. If sample sits for 15 min or longer, mix sample again before proceeding to the next step. Note: It is critical that the enrichment is mixed well to ensure even distribution of target analyte before taking a sample aliquot. (f) Using PhageDx Salmonella Kit Cat. No. 5009; using a single channel pipettor and fresh sterile tip for each sample, transfer 150 mL of enriched sample to white 96-well breakapart plate taking care to avoid transferring fat and meat particles as much as possible.
For PIF (100 g test portion): (a) Weigh 100 g of PIF and place into a sample bag. (b) Add 300 6 5 mL pre-warmed (37 6 1 C) BPW to the sample.
(c) Homogenize sample in a Stomacher 3500 at the highest setting for 120 s (or equivalent homogenizer and setting). (d) Loosely close the sample bag and place in a static air incubator at 37 6 1 C for 16-24 h using a sample rack to keep the bags separate and allow heat transfer. (e) Remove the enriched samples from the incubator and mix thoroughly by hand for at least 30 s to ensure complete mixing. Note: Sample must be thoroughly mixed so that analyte is distributed evenly throughout the entire sample. We recommend vigorous shaking and massaging for at least 30 s and proceeding immediately to the next step after mixing is complete. If sample sits for 15 min or longer, mix sample again before proceeding to the next step. (f) Using a sterile tip/pipet, transfer 1 mL of the sample to a sterile 1.5 mL microfuge tube. (g) Mix contents in microfuge tube and dilute sample 1:10 in BPW (100 mL sample in 900 mL BPW) into a fresh sterile 1.5 mL tube. (h) Using a single channel pipettor and clean tip for each sample, transfer 150 mL of diluted sample to black 96-well break-apart plate.

For Both Matrixes
(a) After transferring samples to 96-well plates, using a single channel pipettor and clean tip for each sample, add 10 mL of the phage solution to the sample and gently mix by pipetting up and down. (b) Cover plate with sealing tape to prevent cross contamination and evaporation. Place the sample in the 37 6 1 C incubator for 2 h. (c) Remove one tube containing the lysis buffer, assay buffer, and substrate for each eight well strip used and thaw to room temperature. Flick or spin the tubes to collect all of the solution at the bottom of the tubes. (d) Prepare the lysis/luciferase master mix by transferring the entire contents of assay buffer (0.5 mL) and lysis buffer (150 mL) tubes to the substrate tube (10 mL) and mix well.

Interpretation and Test Result Report
(a) The luminometer program will display the results on the screen as RLU values corresponding to the well positions of the break-apart plate. (b) For raw ground turkey, samples positive for Salmonella will have a reading value of 750 RLU or greater for a 7-13 h enrichment or 50 000 RLU or greater for >13-18 h enrichment. Negative samples will be less than 750 RLU for a 7-13 h enrichment and less than 50 000 RLU for >13-18 h enrichment. (c) For PIF, samples positive for Salmonella will have a reading value of 500 RLU or greater. Negative samples will be less than 500 RLU. (d) Once all of the samples have been run and analyzed, remove the plate from the luminometer and follow the manufacturer's instructions for cleaning the instrument and shut down.
Note: In some cases,the PhageDx Salmonella Assay will generate a very high signal and result in adjacent wells reading as false positives due to the bleed over from the well with a high signal. If a sample well is positive and has a signal 1000 times lower than the adjacent sample well with a higher signal, this could be a result of bleed over. In these cases, we recommend that the contents of the potential false positive well (lower RLU sample) be transferred to a new well at least a 2-3 well distance from the high signal well or to a new strip and re-read to confirm that the signal is from the sample and not a result of bleed over signal.

Confirmation
We recommend that presumptive positives from the phage assay be confirmed.
(a) For raw ground turkey, confirmation for Salmonella can be performed on overnight enriched samples using immunomagnetic separation (IMS) particles coated with Salmonella

Method Developer Studies
(a) Inclusivity and exclusivity studiesInclusivity strains.-(Salmonella) were obtained from academic, governmental, and commercially available sources. Each strain was grown overnight to stationary phase in BPW media at 37 6 1 C. The overnight cultures were then diluted to 1000 CFU/mL in BPW. One-hundred microliters of diluted cells were used to inoculate 2 Â 9.9 mL of BPW to a concentration of 10 CFU/mL. Samples were then allowed to incubate at 41 6 1 C for 7 or 18h. At each time point, a 150 mL sample was taken for evaluation. To evaluate each strain, cells were infected with phage solution at 37 6 1 C for 2 h. Lysis/luciferase master mix was added, and the sample was read on the luminometer. Samples enriched for 7 h with signals >750 RLU were considered positive. Strains with <750 RLU were tested again using 18 h enriched samples. Samples enriched for 18 h with >50 000 RLU were considered positive (Table 1). Exclusivity strains were also obtained from commercially available sources and were grown to stationary phase overnight. Assays with exclusivity strains were done as with inclusivity strains except overnight cultures were assayed without dilution (Table 2). (b) Product consistency (lot-to-lot) and stability studies.-Three separate production lots of PhageDx Salmonella recombinant phage were prepared according to written manufacturing documents and tested according to quality control procedures. Quality control procedures verified that each lot when diluted to working concentration had the similar titer, background, and level of detection (LOD). Recombinant   phage reagents were aged between 1 and 6 months when assayed for stability. Consistency and stability were done according to AOAC guidance, where a sample was inoculated with S. typhimurium, American Type Culture Collection (ATCC) 19585, to give fractional positives. Ten replicates were run in the PhageDx Assay, and the RLU values analyzed. A set of stability studies was also conducted using the non-target bacterium Citrobacter freundii (ATCC 8090). Overnight cultures of C. freundii were used directly in the assay. Results are shown in Table  3. (c) Robustness study.-Three parameters were varied to demonstrate assay robustness: enrichment time (6.5 and 24 h), recombinant phage concentration (620%), and lysis/luciferase master mix amount (65 mL). Briefly, 25 g raw ground turkey samples were left unspiked or spiked with 0.2-2 CFU/25 g with S. Heidelberg SL476 and stored at 2-8 C for 48-72 h. The PhageDx Salmonella Assay protocol was followed with the variations in enrichment time, recombinant phage concentration, and lysis/substrate master mix amounts as indicated in Table 4. Samples with RLU values greater than 750 were considered positive for 6.5 and 7 h enriched samples and RLU values greater than 50 000 were considered positive for the 24 h enriched samples. Samples were confirmed by allowing samples to enrich overnight and performing IMS with anti-Salmonella coated particles and plating on chromogenic Salmonella selective plates. The presence of mauve colonies that are 1-3 mm in diameter on selective plates indicate a positive result for Salmonella. A summary of the testing is presented in Table 4.

Independent Laboratory Validation Study
(a) Inclusivity.-For the inclusivity study six strains of Salmonella were evaluated. Each Salmonella strain evaluated was cultured by transferring a single colony from trypticase soy agar with 5% sheep blood (SBA) to a 9 mL aliquot of BPW for 7 h at 41 6 1 C, and to a second 9 mL aliquot of BPW for 16 h at 37 6 1 C. After incubation each Salmonella strain at each culture condition was then diluted to 100Â the LOD of the PhageDx Salmonella Assay and analyzed. Tests results were reported as either positive or negative (Table 1) The raw ground turkey and milk-based PIF were purchased from a local supplier and prescreened for natural contamination of the analyte following USDA/FSIS MLG 4.10 and the FDA/BAM Chapter 5 reference methods, respectively. Total aerobic count was determined following FDA/BAM Chapter 3 Aerobic Plate Count reference method (9). Following the screening, the matrixes were inoculated with the indicated strains of Salmonella species. For raw ground turkey matrix, a liquid inoculum culture was used. The inoculum was prepared by transferring a single Salmonella colony from a stock culture stored at -70 C on SBA into brain heart infusion (BHI) broth and incubating the culture at 35 6 1 C for 24 6 2 h. Following incubation, the culture was diluted to a target level using BHI as the diluent to a low level expected to yield fractional positive results (5-15 positive results), and a high level expected to yield all positive results. Samples were spiked and held for 48-72 h post-inoculation at 2-8 C to allow for equilibration of the organism as per AOAC Guidelines. For the milk-based PIF matrix a lyophilized culture was used. Salmonella were cultured from stock stored at -70 C on SBA for 18 hr at 37 C. The lyophyilized culture was prepared by inoculating BHI broth with a single colony from SBA and incubating for 18-24 h at 35 6 2 C, diluting the culture into a sterile cryoprotectant, adding non-fat dried milk (NFDM), and freeze dried for 48-72 h. The culture was then diluted in a sterile cryoprotectant, reconstituted NFDM, and freeze dried for 48-72 h. A bulk lot of the matrix was inoculated with a lyophilized culture that was diluted in powdered NFDM to a low level expected to yield fractional positive results (5-15 positive results), and a high level expected to yield all positive results. After inoculation, samples were held for 2 weeks at room temperature (24 6 2 C) to allow for equilibration of the organism as per AOAC guidelines. For all 100 g test portions analyzed, 25 g of inoculated matrix at each level of contamination was transferred to sterile filter laboratory blender bags on the day of analysis, and then 75 g of uninoculated matrix added to create 100 g test portions.
The level of Salmonella in the low-level inoculum and highlevel inoculum was determined by most probable number (MPN) on the day of analysis. For the 25 g test portion samples, low-level inoculum MPN was determined by evaluating 5 Â 50 g, 20 Â 25 g reference method test portions from the study, and 5 Â 10 g inoculated test portions. The level of Salmonella in the high-level inoculum in 25 g test portions was determined by evaluating the 5 Â 25 g reference method test portions from the study, 5 Â 10 g, and 5 Â 5 g inoculated test portions. To the 50 g portions, 450 mL of the reference method enrichment broth was added; to the 10 g portions, 90 mL of the reference method enrichment broth was added; and to the 5 g portions, 45 mL enrichment broth was added. All 25 g portions were utilized from reference method test potions and analyzed following the FDA/BAM Chapter 5 reference method. The number of positives from the three test levels was used to calculate the MPN using the LCF MPN calculator (version 1.6) (10).

PhageDx Salmonella assay
All samples were analyzed by the PhageDx Salmonella Assay following enrichment with pre-warmed (41 6 1 C) BPW and incubated for 7 and 18 h at 41 6 1 C for raw ground turkey, and enrichment with pre-warmed (37 6 1 C) BPW and incubated for 16 and 24 h at 37 6 1 C for milk-based PIF. After enrichment, a 150 mL direct sample for raw ground turkey, or a 150 mL 1:10 diluted sample for PIF, was transferred to a 96-well plate. Ten microliters of the phage reagent were added and samples were incubated at 37 6 1 C for 2 h. Then, 65 mL of lysis/luciferase master mix was added and the samples read on a luminometer. Regardless of presumptive results, all samples were culturally confirmed by the USDA/FSIS MLG 4.10 or FDA/BAM Chapter 5 reference method. In addition, all samples were confirmed following the alternative confirmation described in Sample

USDA/FSIS MLG 4.10
For the USDA/FSIS MLG 4.10, 25 6 2.5 g of raw ground turkey portions were combined with 225 6 4.5 mL of BPW, homogenized by stomaching for 2 min and incubated 18-24 h at 35 6 2 C. After incubation of all test portions, 0.5 6 0.05 mL of the sample enrichment was transferred into 10 6 0.5 mL of tetrathionate (TT) broth Hajna, and 0.1 6 0.02 mL of the sample enrichment was transferred into 10 6 0.5 mL of modified Rappaport Vassiliadis (mRV) medium. The secondary enrichments were incubated in a circulating, thermostatic water bath at 42 6 0.5 C for 18-24 h. After 18-24 h, the contents in the TT and mRV enrichments were mixed by vortex and a loopful of each streaked to xylose lysine tergitol 4 (XLT4) agar and brilliant green sulfa agar (BGSA). All plates were incubated at 35 6 2 C for 18-24 h. After incubation, plates were observed for typical and well-isolated colonies. One typical colony for each positive sample was picked to triple sugar iron (TSI) agar and lysine iron agar (LIA) slants, along with tryptic soy agar (TSA) plates, and incubated for 24 6 2 h at 35 6 2 C. Following incubation, the slants were examined as a set and the biochemical reactions of the slants noted. Final confirmation was obtained from purified TSA isolates using the Bruker MALDI Biotyper following AOAC Method 2017.09.

FDA/BAM Chapter 5 Salmonella
For the FDA/BAM reference method, 25 g milk-based PIF portions were combined with 225 6 5 mL of lactose broth and homogenized by stomaching for 2 min. Following homogenization, test portions were allowed to stand at room temperature (24 6 2 C) for 60 6 5 min. If necessary, the pH of the enrichments for all matrices was adjusted to 6.8 6 0.2. Subsequently, all matrix enrichments were incubated at 35 6 2 C for 24 6 2 h.
Following incubation, 0.1 mL of primary enrichment was transferred into 10 mL of RV and 1.0 mL into 10 mL of TT medium. RV tubes were incubated at 42 6 0.2 C for 24 6 2 h. The milk-based PIF tested had a low microbial background (<10 4 CFU/g); therefore, the TT tubes were incubated at 35 6 2 C for 24 6 2 h. Following incubation, a loopful of the secondary enrichments were streaked to bismuth sulfite (BS), Hektoen enteric (HE) and xylose lysine deoxycholate (XLD) agar and incubated at 35 6 2 C for 24 6 2 h. If no visible colonies were present after 24 h of incubation on the BS plates, they were re-incubated for an additional 24 6 2 h at 35 6 2 C. A minimum of two suspect colonies from each selective agar were transferred to TSI and LIA slants and incubated at 35 6 2 C for 24 6 2 h. Following incubation, TSI and LIA slants were examined for typical reactions. Slants producing typical reactions were streaked to TSA and incubated for 35 6 2 C for 18-24 h.
Following incubation, isolates were serologically tested for both somatic O and flagellar H agglutination. Additionally, final confirmation was obtained from purified TSA isolates using the Bruker MALDI Biotyper following AOAC Method 2017.09.

Results
Inclusivity and exclusivity studies using the PhageDx Salmonella Assay demonstrate that the PhageDx Assay is specific for the detection of Salmonella spp. The PhageDx Salmonella Assay was able to detect 108/110 Salmonella strains tested (Table 1). In addition, the PhageDx Assay did not detect 30/30 non-Salmonella strains tested (Table 2).
Product consistency (lot-to-lot) and stability studies show that the PhageDx Salmonella recombinant phages can be manufactured consistently and are stable for at least 8 months when stored at 4 C. Manufactured lots were made on 12/18, 3/19, and 8/19 according to written manufacturing documents. Working solutions of each lot produced similar results when tested according to QC tests for bacteriophage concentration, background signal, and LOD. Stability tests of each lot were performed to determine the shelf life of the recombinant phage. These tests demonstrated that lots produced 1 month prior to testing showed no significant difference from lots produced at least 8 months prior to testing. Additionally, no variation in exclusivity was observed with these three recombinant phage lots in tests with C. freundii.
Robustness testing of the PhageDx Salmonella Assay demonstrated that variations in enrichment time, recombinant phage concentration, and lysis/luciferase master mix amounts do not alter the results compared to the standard protocol. Enrichment times of 6.5 and 24 h, recombinant phage volumes of 8 and 12 lL, and lysis/luciferase master mix volumes of 60 and 70 lL produced identical results to the standard protocol of 7 h enrichment, 10 lL of recombinant phage, and 65 lL of lysis/luciferase master mix in both uninoculated and low inoculum test samples (Table 4). These results indicate that these deviations from the PhageDx Salmonella Assay protocol did not alter the final results.
In an unpaired study, the presumptive results from the PhageDx Salmonella Assay for raw ground turkey (7 and 18 h enrichments) and PIF (16 and 24 h enrichments) were not significantly different from those of the USDA/FSIS MLG 4.10 and FDA/ BAM Chapter 5, respectively. In a paired study, the results from the PhageDx Salmonella Assay presumptive, PhageDx confirmation method, and the respective reference methods were identical (Table 6). In addition, no false positive or false negatives were detected in the matrix study. In summary, independent laboratory matrix tests demonstrated that the results from PhageDx Salmonella Assay and the USDA/FSIS MLG 4.10 and FDA/BAM chapter 5 reference methods for raw ground turkey and PIF, respectively, were not significantly different (Tables 5  and 6).

Discussion
The results of this validation study show that the PhageDx Salmonella Assay is an effective alternative to the USDA/FSIS MLG 4.10 for the detection of Salmonella in 25 g raw ground turkey and FDA/BAM Chapter 5 for the detection of Salmonella in 100 g of milk-based PIF.
In inclusivity and exclusivity testing, the method was shown to be specific for Salmonella, correctly identifying 108 Salmonella target strains across both species and six S. enterica subspecies and 30 non-target strains. The PhageDx Salmonella Assay was unable to detect two strains within the inclusivity panel, a strain of S. enterica, subsp. Arizonae and of S. enterica, subsp. Houtenae. It is unclear as to why these strains were missed since the PhageDx Salmonella Assay was able to detect other members of the subspecies. One explanation is that these strains do not have the receptor(s) required for recognition by the phage. With over 2600 serovars in the genus, it is not surprising that there is a range of diversity that is difficult to encompass entirely. Another explanation may be that the strain has a mechanism that prevents the phage from replicating, thus unable to produce the luciferase reporter (12).
The recombinant phage can be produced consistently and is stable for 8 months when stored appropriately. Robustness testing of the PhageDx Salmonella Assay indicated that the method works well when the assay parameters (enrichment time, recombinant phage concentration, and substrate amount) were varied from the stated protocol.
Independent laboratory testing demonstrated that the PhageDx Salmonella Assay was able to detect Salmonella at low levels in 25 g test portions of raw ground turkey and 100 g test portions of milk-based PIF, which also contained approximately 3.6 Â 10 6 CFU/g and 1.8 Â 10 3 CFU/g background flora, respectively. These studies also demonstrated that the performance of the PhageDx Salmonella Assay was not statistically different from that of USDA/FSIS MLG 4.10 for 25 g test portions of raw ground turkey or FDA/BAM Chapter 5 for 100 g test portions of milk-based PIF. An alternative confirmation procedure was shown to be identical to the reference method confirmation procedures.
The PhageDx Salmonella Assay has a number of advantages over the USDA/FSIS MLG 4.10 and FDA/BAM Chapter 5 reference methods. In addition to being a specific assay, the results are easy to interpret as an RLU endpoint is used to determine the outcome of the assay. Another advantage is that PhageDx provides a presumptive positive result in as little as 9.5 h for raw ground turkey or 18.5 h for PIF compared to >24 h in the case of the USDA/FSIS MLG 4.10 and FDA/BAM Chapter 5 reference methods, respectively. PhageDx is also a simple test that involves only five basic steps: enrichment, sampling, infection, substrate addition, and signal readout. Finally, PhageDx Assay is a rapid method that offers a considerable time savings

Conclusion
Results of this validation study support the claim that the PhageDx Salmonella Assay is a specific, sensitive, fast, and simple method for the detection of Salmonella in raw ground turkey and milk-based PIF and is statistically comparable to the USDA/FSIS MLG 4.10 and FDA/BAM Chapter 5 methods, respectively. By using a luciferase-expressing recombinant bacteriophage, the assay was able to detect a single, viable bacterium after 7 h enrichment and a 2 h infection for raw ground turkey and 16 h enrichment and 2 h infection for milk-based PIF. The PhageDx Salmonella Assay thus offers shorter time to results compared with the other validated Salmonella detection assays.