ABSTRACT

Modifications to pathogen detection kits to accomplish simplified protocols with reduced time to results may impact method performance, particularly when combining shortened enrichment times and simplified enrichment procedures. We used Salmonella detection in dark chocolate as a model to test the impact of different enrichment times (minimum and maximum validated times) and procedures on detection of low levels of difficult-to-detect Salmonella strains, for three PCR kits that were AOAC International Performance Tested Method certified for detection of Salmonella spp. in dark chocolate. Initial inclusivity studies with pure cultures showed that all three kits detected 70 of 70 Salmonella spp. strains at 1 log above the theoretical limit of detection, with some strains yielding later cycle threshold values or having variable detection among technical replicates, indicating reduced assay performance for these strains. Based on these data, we selected a S. enterica subsp. enterica serovar Poona strain as well as three non-subsp. enterica strains to test the ability of the three kits to detect Salmonella in dark chocolate inoculated at low levels (0.06 to 1.18 most probable number per 25 g). With primary enrichment in skim milk at 35°C, detection frequency for all assays did not significantly differ from the reference method for both the minimum and maximum validated enrichment times. However, a pilot study that used primary enrichment in buffered peptone water at 42°C yielded significantly fewer positive samples (13 of 80) than were obtained with the U.S. Food and Drug Administration Bacteriological Analytical Manual method using enrichment in skim milk at 35°C (40 of 80 positive samples); strains representing subsp. houtenae and salamae were detected in significantly fewer chocolate samples than enrichment with skim milk. Our data indicate that continued efforts to simplify rapid pathogen detection kits may reduce kit performance in a way that can only be detected with stringent evaluation protocols that are designed to identify kit failure modes.

HIGHLIGHTS
  • Chocolate is a challenging matrix for PCR-based detection of Salmonella.

  • Difficult-to-detect strains should be included in matrix validation studies.

  • Our data support shortening the enrichment time of skim milk–enriched chocolate samples.

  • Alternative enrichment procedures should be carefully evaluated by the end user.

  • Alternative confirmations via chromogenic agar provide shortened time to results.

Nontyphoidal Salmonella is one of four key global causes of bacterial diarrheal disease according to the World Health Organization (27). Food-implicated sources alone are estimated to cause 78 million illnesses (95% uncertainty interval [UI]: 31 million, 211 million) and nearly 28,000 deaths (95% UI: 17,000, 49,000) annually worldwide (27). The Center for Disease Control and Prevention's preliminary 2018 incidence and trends report lists Salmonella as the second leading cause of bacterial foodborne disease, with a 9% increase in the number of reported human clinical infections compared with 2015 to 2017 data (5). This consistently high incidence of human clinical infections with Salmonella is due in part to the diverse nature of this pathogen, which includes more than 2,600 serovars (23). Furthermore, multiple serovars have been shown to persist for long periods of time in suboptimal environments, such as in powdered infant formula for 15 months (9) or in chocolate for up to 2 years (17). Salmonella is therefore able to survive a number of food-associated stresses, such as low water activity in dried foods and the presence of inhibitory compounds in foods such as spices and cocoa. Compared with other bacterial foodborne pathogens (e.g., Campylobacter spp., Shiga toxin–producing Escherichia coli, Listeria monocytogenes, Vibrio spp., Shigella spp.), Salmonella has been isolated from a broader variety of foods (12, 45, 47). Of particular concern are ready-to-eat, low-water-activity foods (aw < 0.70) (14) such as nuts and nut butters, spices, cereals, chocolate, and dairy powders, all of which have been implicated in U.S. and global Salmonella outbreaks (15, 17, 25, 39). Together, Salmonella's tremendous strain diversity and its ability to survive in a variety of different foods present a multifaceted challenge that regulatory preventive measures, such as those required of domestic and international food manufacturers by the Food Safety Modernization Act, have yet to successfully address to reduce the incidence of Salmonella infections (5).

Given the global incidence of Salmonella and the increasingly international food supply, implementation of a reliable pathogen detection method represents a critical component of any food safety program, especially as ready-to-eat, low-water-activity foods are often consumed without further processing steps such as heating. Therefore, reliable rapid detection methods are essential to verify and validate food safety systems designed to control Salmonella. The desire for a rapid method that offers faster time to results, without compromising the accuracy of the method, has led to the adoption of rapid technologies such as PCR and automated immunoassays and to method simplification by reducing the number of enrichments performed (i.e., no secondary enrichment), or to the use of novel enrichment media. As part of the method selection criteria, food manufacturers and regulatory agencies typically require method approval from a third-party standards organization, such as AOAC International. Currently, there are >100 methods for Salmonella detection with AOAC approval (52 with Official Methods of Analysis; 53 with Performance Tested Method). Although AOAC approval requires both a robust laboratory validation study and inclusivity and exclusivity data, the validation study design does not specify which Salmonella strains should be included. Furthermore, AOAC approval is specific to the food matrices included in the validation, and only a limited number of Salmonella strains are tested, which may not account for the genetic and phenotypic diversity of Salmonella associated with a given matrix. Given both the wide diversity of Salmonella strains and the wide variety of food matrices tested for Salmonella, a rational approach to select species and serovars for use in verification and validation of methods for a given matrix is important.

Validation of rapid method performance for detection of Salmonella in low-water-activity foods is especially important because these products typically contain injured cells that could possibly fail to reach detectable levels when using simplified, novel enrichment procedures. Low-water-activity, cocoa-containing products present added challenges for method developers as the detection method must address (i) sublethal cell injury (due to heat treatment and/or low moisture stress), (ii) the inhibitory properties of the polyphenolic compounds present in cocoa, and (iii) PCR inhibition (for PCR-based rapid methods) to reduce the risk for false negative results. Salmonella contamination of chocolate has previously been associated with multiple outbreaks involving several different serovars of Salmonella (6, 11, 18, 19, 49). The ability of Salmonella to survive for extended periods of time in chocolate has been attributed to the fat and sugar components of chocolate providing a protective effect in the low-water-activity environment (17). Because product involved in outbreaks has been shown to harbor low levels of Salmonella (i.e., as low as 1 to 6 CFU/g) (15), rapid detection methods must reliably detect low levels of injured Salmonella, despite these potentially inhibitory matrix effects. The antimicrobial properties of cocoa may not only interfere with recovery and enrichment but may also present challenges for PCR-based technologies because these compounds have been shown to cause PCR inhibition (8, 13, 41). To recover Salmonella, enrichment with skim milk has traditionally been used to address the antimicrobial properties of cocoa-containing foods and to facilitate recovery and prevent PCR inhibition. Given the previous outbreaks of Salmonella-contaminated chocolate (6, 11, 18, 19, 49), rapid methods including novel, simplified enrichment procedures need to be strategically assessed through a carefully designed matrix study to ensure all the noted challenges with this matrix have been addressed.

To assess the effects of strain diversity and different enrichment procedures on the recovery of Salmonella from a challenging, low-water-activity matrix, we challenged three AOAC Performance Tested Method–approved PCR methods previously validated for detection of Salmonella in chocolate by testing dark chocolate samples inoculated with fractional positive levels of four diverse Salmonella strains. Together, our study demonstrates that (i) use of a diverse Salmonella strain set including multiple non-subsp. enterica strains identify differences in recovery of some non-subsp. enterica strains, suggesting that rapid methods need to be challenged with sufficiently diverse strain sets, and (ii) by subjecting the rapid methods to “worst-case” challenges we identified enrichment procedures that could lead to false negatives and/or the potential for PCR inhibition.

MATERIALS AND METHODS

Strain collection

For this study, a Salmonella strain set used in previous studies (4, 42) was modified to include Salmonella enterica subsp. indica strain ATCC 43976, yielding a set of 70 Salmonella strains. The strain collection includes strains representing both Salmonella species (i.e., bongori and enterica) and six S. enterica subsp.: enterica (I), salamae (II), arizonae (IIIa), diarizonae (IIIb), houtenae (IV), and indica (VI). As previously detailed (42), this strain collection was assembled to include serovars of international importance to human health (e.g., Typhimurium, Enteritidis, Typhi), serovars known to be associated with specific geographical regions, and serovars that had been previously implicated in food contamination (4). In addition, five serovars (S. enterica subsp. enterica serovars Mbandaka, Montevideo, Oranienburg, Senftenberg, and Typhimurium) that had been previously associated with outbreaks in which chocolate was implicated as the source were included in the strain collection (39). All strains were stored at −80°C in brain heart infusion (BHI; BD, Sparks, MD) broth with 15% glycerol. Salmonella strains were routinely cultured in BHI broth or on BHI agar plates at 37°C.

Rapid methods evaluated

Three rapid methods designated as kit A, kit B, and kit C were evaluated. All of these kits have AOAC Performance Tested Method approval for detection of Salmonella spp. in dark chocolate. All three kits are PCR based, have next-day time to results, and allow for up to 96 samples to be analyzed simultaneously.

Inclusivity study

Inclusivity studies were performed by screening all assays for their ability to detect all 70 Salmonella strains in pure culture at ∼1 log above the theoretical limit of detection (LOD) calculated for each kit. The theoretical LOD was defined as the bacterial concentration (CFU per milliliter) in enriched samples needed postincubation to achieve transfer of one Salmonella cell into each tube for PCR amplification.

For the inclusivity studies, frozen cultures (for all 70 strains) were streaked for isolation onto BHI agar, followed by incubation at 37°C for 18 to 24 h. An isolated colony was selected and substreaked onto a fresh BHI agar plate, with subsequent incubation for an additional 18 to 24 h at 37°C to generate plates with confluent growth. Sterile cotton-tipped swabs were used to collect Salmonella from agar plates with confluent growth, followed by resuspension in 5 mL of phosphate-buffered saline (PBS). A spectrophotometer was used (OD600) to adjust the concentration to 108 CFU/mL as described previously (42). Salmonella suspensions were subsequently serially diluted in PBS to achieve approximate concentrations of LOD + 1 log; these diluted suspensions were used as inputs for the inclusivity studies for each of the three kits tested. Strains were tested as three technical replicates, instead of the recommended one sample per strain (i.e., AOAC guidelines for method validation (1)), to provide additional resolution regarding the repeatability of the inclusivity assessments for each strain and kit. Concentrations were verified by serially diluting and spiral plating onto BHI agar using the Eddy Jet 2 spiral plater (IUL Micro, Barcelona, Spain), followed by incubation at 37°C for 18 to 24 h. Colonies were then enumerated using the SphereFlash automatic colony counter (IUL Micro).

Determination of die-off rate for Salmonella spp. strains inoculated onto dark chocolate

Four Salmonella spp. strains (Salmonella bongori ATCC 43975 [S. bongori], S. enterica subsp. enterica serovar Poona FSL R8-0115 [Salmonella Poona], S. enterica subsp. houtenae ATCC 43974 [subsp. houtenae], and S. enterica subsp. salamae ATCC 700148 [subsp. salamae]) were selected for further evaluation of detection in dark chocolate. Strains were selected based on (i) performance in the inclusivity study and to (ii) ensure broad representation of both species and multiple subspecies of Salmonella. Specifically, data from the inclusivity study were used to identify strains that showed (i) later cycle threshold (CT) values in the PCR (indicating that more PCR cycles were required for detection) or (ii) at least one negative (i.e., undetected) out of the three technical replicates, both indicating that a given strain is more likely to not be detected by a given method.

Fractional Salmonella-positive samples in the range of 25 to 75% is a required criterion for successful validation of an AOAC method (1). To achieve fractional positive levels, AOAC recommends achieving a final level of 0.2 to 2.0 CFU per sample following acclimation for 14 days at room temperature for low-water-activity matrices (1). To facilitate low-level inoculation of Salmonella, the die-off rates of the four selected Salmonella strains were evaluated over 14 days of acclimation on dark chocolate held at 25°C. For the die-off studies, dark chocolate was cut into 1-g pieces that were then surface inoculated at a concentration of 106 CFU Salmonella per 1-g dark chocolate piece. Salmonella suspensions (109 CFU/mL) were prepared in PBS via the methods described above. Dark chocolate pieces were surface inoculated with three 1-μL aliquots of the 109 CFU/mL suspension (final inoculum of ∼3 × 106 CFU). Inoculations were performed as three biological replicates (with four chocolate pieces inoculated for each biological replicate). Inoculated chocolate pieces were held at room temperature for 1 h, after which two technical replicates from each biological replicate were used for enumerating Salmonella levels. The remaining inoculated dark chocolate pieces were incubated at 25°C for 14 days after which enumerations were preformed (two technical replicates for each biological replicate). Enumeration of Salmonella spp. was performed by diluting 1-g dark chocolate pieces in 9 mL of ultrahigh temperature skim milk (hereafter skim milk) prewarmed to 35°C and performing subsequent serial dilutions followed by spiral plating in duplicate onto xylose lysine deoxycholate (XLD; BD) agar plates. After incubating plates at 35°C for 18 to 24 h, colonies were enumerated using the SphereFlash automated colony counter. Log reductions for each Salmonella strain were calculated by subtracting the log(CFU/g) recovered from dark chocolate at 14 days postinoculation from the log(CFU/g) recovered from dark chocolate at 1 h postinoculation. Die-off rates were used to calculate the starting concentration required for surface inoculation of chocolate to achieve the targeted low cell level (0.2 to 2.0 CFU/1-g dark chocolate piece) after equilibration for 14 days at 25°C.

Low-level inoculation of select Salmonella strains onto dark chocolate

Rapid methods were assessed for their ability to detect low levels of all four Salmonella spp. strains inoculated onto dark chocolate pieces based on AOAC evaluation methodologies, including inoculation at levels expected to yield fractional positive levels (i.e., 0.2 to 2.0 CFU/g) (1). For each serovar, three to four sets of chocolate samples, each inoculated with different Salmonella levels, were prepared to yield predicted final concentrations (following acclimation at 25°C for 14 days) ranging from 0.01 to 4.0 CFU/g. Although predictions of final levels were based on the results of the die-off study, multiple inoculation levels were used to account for potential variation between the die-off observed with the inoculated samples and the predicted die-off rates. Each target concentration level included inoculation of (i) 20 (1-g) dark chocolate pieces for fractional positive evaluation, (ii) 18 (1-g) pieces for most-probable-number (MPN) enumeration of Salmonella on dark chocolate to verify the achieved inoculation, and (iii) 10 (1-g) pieces for prescreening of samples for selecting the inoculation level predicted to yield 25 to 75% Salmonella-positive dark chocolate samples. An additional set of 20 (1-g) dark chocolate pieces was prepared for each level to evaluate an unpaired, alternative enrichment procedure. In addition, two dark chocolate pieces were inoculated with ∼100 CFU/g Salmonella Typhimurium (FSL S5-0536) as positive control samples, and two dark chocolate pieces were processed as uninoculated negative controls for the 10-day (prescreen assessment) and 2-week (paired and unpaired studies) analyses. Bacterial suspensions were prepared in PBS as described above, except that the suspensions were held overnight at 4°C to allow for plating and enumeration of the inoculum on BHI agar plates before inoculation. After confirmation of the inoculum concentration, suspensions were used to inoculate 1-g dark chocolate pieces with 1 to 8 μL of the Salmonella suspension. Inoculated dark chocolate pieces were held at room temperature for 1 h and were subsequently incubated at 25°C. Immediately after inoculation, aliquots of the bacterial suspensions were serially diluted and enumerated by spiral plating in duplicate onto BHI agar. Kit specificity was not evaluated because of a lack of naturally occurring background flora in the chocolate used here (i.e., no colonies recovered from uninoculated control samples on XLD or Hektoen Enteric [HE]), and because potential cross-reacting organisms were not included as part of the matrix inoculations.

Prescreen assessment of inoculated dark chocolate pieces

Inoculated dark chocolate pieces were prescreened after 10 days of equilibration at 25°C by using a modified U.S. Food and Drug Administration (FDA) Bacteriological Analytical Manual (BAM) method. The modified procedure did not include a selective enrichment step; rather, the primary enrichment was directly streaked onto a single selective-differential plating agar, HE (BD). Ten days postinoculation, ten 1-g dark chocolate pieces were enriched in skim milk (prewarmed to 35°C, 9 mL per chocolate piece). Positive (inoculated with Salmonella Typhimurium) and negative (uninoculated dark chocolate) matrix controls were also enriched. Enrichments were homogenized by hand and set at room temperature for 1 h after which the pH was adjusted to 6.8 ± 0.2 with 1 N NaOH. Brilliant green dye was added to a final concentration of 0.002% (v/v), followed by incubation at 35°C for 24 h. After incubation, a 10-μL sterile loop was used to streak primary enrichments onto HE agar plates, followed by incubation at 35°C for 24 h. Levels that yielded (30 to 70%) presumptive Salmonella (i.e., blue-green colonies with black center) were selected for method evaluations on day 14.

Sample enrichment in skim milk and culture confirmation for paired enrichment evaluation

Evaluation of all three rapid methods used the same primary enrichment media (skim milk containing 0.002% [v/v] brilliant green dye) as the FDA BAM method; hence, analysis from a single paired FDA BAM enrichment was used for all three methods (Fig. 1). At 14 days postinoculation, dark chocolate samples (n = 20 pieces) and matrix positive and negative control samples were prepared by combining 1 g of inoculated chocolate with 24 g of uninoculated chocolate, in Whirl-Pack filter bags (Nasco, Fort Atkinson, WI). The resulting 25-g samples were diluted 1:10 (sample-to-enrichment ratio) with 225 mL of skim milk prewarmed to 35°C. An enrichment positive control was prepared by directly inoculating 225 mL of skim milk with 100 to 1,000 CFU/mL Salmonella. An uninoculated skim milk sample was included as an enrichment negative control. Samples were homogenized with a Seward Stomacher 400 (Seward Limited, West Sussex, UK) for 1 min to resuspend chocolate pieces. After being held at room temperature for 1 h, the pH was adjusted to 6.8 ± 0.2 using 1 N NaOH, and 450 μL of 1% brilliant green dye solution (final concentration, 0.002% [v/v]) was added, followed by incubation at 35°C for a total of 24 h. Primary enrichments were sampled throughout the 24-h incubation time in accordance with the minimum and maximum enrichment times specified by each rapid method (i.e., 16, 20, and 24 h; see Fig. 1). In brief, ∼7 mL of enriched sample was transferred to a sterile culture tube at each sampling time point, and the remaining primary enrichments were returned for further incubation at 35°C until the maximum enrichment time. The subsampled enrichments were then subjected to (i) lysis for PCR as per kit manufacturer's instructions for each rapid method, (ii) alternative confirmation via plating on XLD and a chromogenic agar (details described in the alternative confirmation procedure below), and (iii) reference method culture confirmation (representing the “paired BAM method”) by transferring 1.0 and 0.1 mL to sterile culture tubes containing 10 mL of tetrathionate (TT; BD) and 10 mL of Rappaport-Vassiliadis (RV; BD) broth, respectively. TT-inoculated tubes were incubated at 35°C for 24 h, and RV-inoculated tubes were incubated in a water bath at 42°C for 24 h. RV- and TT-enriched samples were subsequently streaked onto XLD and HE agar plates, followed by incubation at 35°C for 24 h. In addition to XLD and HE, the FDA BAM cultural confirmation method also requires use of a third plating medium (bismuth sulfite [BS] agar) for the isolation of Salmonella. Because the primary purpose of BS is to aid in the recovery of Salmonella with atypical morphologies and specifically Salmonella Typhi, BS was not included as our challenge study strains did not include serovar Typhi. For each sample tested, at least one colony with morphology typical of Salmonella was confirmed with PCR amplification of invA; different invA primers were used for S. enterica and S. bongori, as described below.

FIGURE 1

Schematic flow diagram representing one sample analyzed by the paired and unpaired matrix studies. Following 14 days of acclimation of Salmonella on the dark chocolate matrix, the primary enrichments were prepared by diluting 25 g of dark chocolate at a 1:9 sample-to-enrichment medium ratio. For the paired matrix study, primary enrichments were prepared and incubated following the FDA BAM method. For the unpaired matrix study, primary enrichments were prepared following an alternative protocol using BPW and 42°C incubation. After incubation, enrichments proceeded through detection with the three different rapid methods and the FDA BAM method at both the minimum and maximum enrichment times specified by each rapid method. Kits A and B included a BHI secondary enrichment before PCR lysis, whereas kit C proceeded directly to the lysis step from the primary enrichment. An additional alternative confirmation was performed with kit C alongside the FDA BAM method.

FIGURE 1

Schematic flow diagram representing one sample analyzed by the paired and unpaired matrix studies. Following 14 days of acclimation of Salmonella on the dark chocolate matrix, the primary enrichments were prepared by diluting 25 g of dark chocolate at a 1:9 sample-to-enrichment medium ratio. For the paired matrix study, primary enrichments were prepared and incubated following the FDA BAM method. For the unpaired matrix study, primary enrichments were prepared following an alternative protocol using BPW and 42°C incubation. After incubation, enrichments proceeded through detection with the three different rapid methods and the FDA BAM method at both the minimum and maximum enrichment times specified by each rapid method. Kits A and B included a BHI secondary enrichment before PCR lysis, whereas kit C proceeded directly to the lysis step from the primary enrichment. An additional alternative confirmation was performed with kit C alongside the FDA BAM method.

Sample preparation and enrichment for unpaired alternative enrichment evaluation

An alternative enrichment procedure using buffered peptone water (BPW) and incubation at 42°C was also assessed, using one of the kits. In brief, Salmonella-inoculated dark chocolate samples (n = 20) were processed as stated above, except that BPW prewarmed to 42°C was used instead of skim milk, and pH adjustment of the enriched samples was not required (21). Matrix positive and negative and enrichment positive and negative controls were also prepared. BPW enrichments were incubated at 42°C for a total of 24 h; samples for testing were collected at a minimum enrichment time of 18 h and maximum of 24 h for Salmonella PCR detection and culture confirmation by the FDA BAM method (i.e., enrichment with RV and TT followed by plating on HE and XLD).

MPN quantification of low levels of Salmonella on dark chocolate

A three-level, five-tube MPN test was performed to quantify the achieved inoculation level of Salmonella on the dark chocolate pieces. The levels represented enrichment of (i) 2 g of chocolate and 18 mL of skim milk (high level), (ii) 1 g of chocolate and 9 mL of skim milk (medium level), and (iii) 0.5 g of chocolate in 9.5 mL of skim milk (low level). Enrichment in skim milk was done as described by the FDA BAM method, including pH adjustment and brilliant green dye addition (0.002% [v/v]), followed by incubation at 35°C. After 24 h, culture confirmation was performed as outlined above, beginning with subculturing into TT and RV. MPN levels were estimated with the MPN calculator provided in the FDA BAM Appendix 2 (46).

Alternative culture confirmation method without selective enrichment and using chromogenic agar

For kit C, an alternative confirmation using a chromogenic agar was assessed in parallel with the FDA BAM culture confirmation method. The alternative confirmation method consisted of streaking 10 μL of the primary enrichment onto XLD and a chromogenic agar, followed by incubation at 35°C for 24 h. The alternative confirmation procedure was performed for three of four sets of inoculated chocolate (method was not performed for subsp. salamae–inoculated samples due to product shipment delay). Isolated colonies having morphologies typical of Salmonella, as described by manufacturer's instructions, were confirmed by PCR amplification of invA as described below.

PCR amplification of invA

Confirmation of Salmonella spp. through PCR detection of invA was selected for isolate confirmation as it is one of the approved confirmation protocols listed in the FDA BAM (48). Cell lysis, PCR, and detection of invA were performed as described in a previous study (4), with some modifications. The forward primer for S. bongori was AB9intINVAF (4); primers invAF and invAR were used for the confirmation of the other Salmonella strains studied here (26). Other modifications included a lysis for 5 min at 95°C, an annealing temperature of either 52°C (Salmonella Poona, subsp. houtenae, and subsp. salamae) or 60°C (S. bongori), and an elongation time of 30 s at 72°C (4).

Statistical analyses

Pearson's chi-square with Yate's continuity correction was used to compare Salmonella detection frequency in the 10-day prescreen and the 14-day screen. McNemar's exact test was used for assessing paired enrichment results (i.e., enrichment with skim milk). All statistical analyses were performed using R version 3.5.1 (R Core Team, Vienna, Austria). The AOAC difference in probability of detection (dPOD) (1) was used to analyze unpaired data obtained with the alternative enrichment procedure (BPW incubated at 42°C).

RESULTS

Three rapid methods successfully detected all 70 Salmonella strains in the inclusivity analysis, but certain strains were more difficult to detect

Inclusivity analyses conducted with 70 diverse Salmonella spp. strains at ∼1 log above the calculated LOD were used to identify difficult-to-detect strains, defined as strains that either (i) had later CT values compared with other strains in the set and/or (ii) were not detected in at least one of the triplicate analyses. The theoretical LODs for kits A, B, and C were calculated to be 7 × 102, 6 × 102, and 2 × 103 CFU/mL, respectively. For kits A, B, and C, a total of 70, 64, and 69 of the 70 total strains tested yielded a positive result for all three technical PCR replicates (Supplemental Tables S1 through S3). For kit C, subsp. houtenae was not detected in one of three replicates. For kit B, three strains were not detected in two of three replicates (serovar Wien, subsp. houtenae, sp. bongori), and three strains were not detected in one of three replicates (serovars Adelaide and Putten, and subsp. arizonae).

Average CT values obtained using kit A ranged from 33.2 to 37.8 cycles; the 10 isolates with the latest CT values for this kit included serovars Adelaide, Anatum, Cerro, Ealing, Lille, Muenchen, Oranienburg, subsp. arizonae, subsp. houtenae, and sp. bongori (all of these isolates had CT values > 36.2). Kit B generated CT values ranging from 31.2 to 32.9 cycles. The CT values obtained using kit C ranged from 31.2 to 34.6 cycles. The non-subsp. enterica and S. bongori strains CT data from kit C had values ranging from 33.0 to 34.6 cycles. Overall, our data indicated that S. bongori as well as subsp. houtenae and arizonae could also be classified as difficult to detect (see Table S1 through S3 for details); for example, the subsp. houtenae strain was difficult to detect across all three kits; for kits B and C, only two and one technical replicates generated a CT value, respectively, whereas for kit A, this strain yielded the second latest CT value.

Ten-day prescreen results do not necessarily predict day 14 fractional positive results

For each strain tested, 10 inoculated chocolate pieces were “prescreened” at 10 days to select the inoculation level that would most likely yield between 25 and 75% Salmonella-positive samples. Although the fraction of positive samples obtained at day 10 (based on testing of 1 g) was numerically higher or lower for two strains each, compared with the fraction of positive samples obtained at day 14 (based on testing of 25 g; Table 1), the proportion of day 10 and 14 positive samples only differed significantly for S. bongori (with a significantly lower proportion of positives observed for day 14). This indicates that for some strains (e.g., S. bongori in our study) die-off rates for Salmonella cultured under suboptimal conditions, such as on dark chocolate, may have considerable variability, which may make matrix studies with S. bongori more challenging than studies done with other Salmonella strains. This was also a challenge in our study where multiple separate experiments did not yield any replicates that met the criterion of 25 to 75% of samples positive and thus why we reported results that only yielded four positives out of 20 samples (20%) with the FDA BAM method.

TABLE 1

Proportion of positives obtained from the 10-day prescreen compared with results after 14 days of incubation at 25°C

Proportion of positives obtained from the 10-day prescreen compared with results after 14 days of incubation at 25°C
Proportion of positives obtained from the 10-day prescreen compared with results after 14 days of incubation at 25°C

Recovery of Salmonella did not vary significantly between the minimum and maximum validated enrichment incubation times for any of the methods

To compare the three rapid methods' performance compared with that of the FDA BAM, a total of 80 inoculated dark chocolate samples (20 samples per Salmonella strain) were enriched in skim milk and tested at both the minimum and maximum validated enrichment times: (i) 16 and 20 h (for kits B and C) and (ii) 20 and 24 h (for kit A). The paired “gold standard” reference method (i.e., the FDA BAM culture method, which uses primary enrichment in skim milk with 0.002% brilliant green dye for 24 h at 35°C), yielded 40 Salmonella-positive 25-g dark chocolate samples among the total 80 samples tested (see Table 2 for details, including breakdown by serovar). The achieved inoculation levels, as determined by MPN, were all below the AOAC validation guidelines' maximum recommended level of 2.0 MPN/25 g for a matrix study. For one strain, S. bongori ATCC 43975, the inoculation level (i.e., 0.06 MPN/25 g) was just below the minimum recommended level.

TABLE 2

Total number of Salmonella-positive dark chocolate samples by strain for the minimum and maximum enrichment times

Total number of Salmonella-positive dark chocolate samples by strain for the minimum and maximum enrichment times
Total number of Salmonella-positive dark chocolate samples by strain for the minimum and maximum enrichment times

In total, three false-negative results were obtained using kit A and one false negative was obtained for kit C. For kit A, 38 of 80 samples tested positive at the minimum enrichment time (sensitivity = 0.95; Table 3), whereas 39 of 80 samples tested positive at the maximum enrichment time (sensitivity = 0.98; Table 3). Specifically, one sample inoculated with subsp. salamae was not detected at the minimum enrichment time, whereas one sample inoculated with subsp. houtenae was not detected at either the minimum or maximum enrichment times (Table 2). The same subsp. houtenae–inoculated sample that was negative at both enrichment time points for kit A was only positive at the maximum enrichment time for kit C. Overall, 39 of 80 samples (sensitivity = 0.98; Table 3) and 40 of 80 samples (sensitivity = 1.00; Table 3) tested positive at the minimum and maximum enrichment times for kit C, respectively. Of the three kits, only kit C showed perfect sensitivity using the maximum enrichment time across all four Salmonella strains compared with the paired FDA BAM method.

TABLE 3

Comparison of the rapid method false negatives, false positives, and method sensitivity for Salmonella-positive dark chocolate samples tested at the minimum and maximum enrichment times

Comparison of the rapid method false negatives, false positives, and method sensitivity for Salmonella-positive dark chocolate samples tested at the minimum and maximum enrichment times
Comparison of the rapid method false negatives, false positives, and method sensitivity for Salmonella-positive dark chocolate samples tested at the minimum and maximum enrichment times

For kit B, the overall sensitivity was the same (0.98; Table 3) for both the minimum and maximum enrichment times (i.e., 39 rapid method Salmonella-positive results compared with 40 with the reference method at each time point). However, some test results were classified by kit B's PCR software program as “repetition.” As per the manufacturer's instructions, samples would be classified as repetition due to either a late CT value (>33) or due to an atypical amplification curve; this classification could indicate PCR contamination, low levels of Salmonella, or partial PCR inhibition. Samples that were classified as repetition included two subsp. houtenae–inoculated samples (tested at the maximum enrichment time); both of these samples were negative with the paired FDA BAM reference method and were negative when tested with kit B at the minimum enrichment time (Table 2). For samples inoculated with S. bongori, kit B yielded two and eight samples classified as repetition at minimum and maximum enrichment times, respectively; none of these samples tested positive with the FDA BAM method. In addition, kit B yielded a total of nine positive samples at the maximum enrichment time for samples inoculated with S. bongori, five of which were classified as false positives because Salmonella was not detected in these samples by using the FDA BAM method (Table 2). For samples inoculated with Salmonella Poona and subsp. salamae strains, kit B yielded the same number of positive samples at minimum and maximum incubation times; results obtained perfectly matched the paired culture-based reference method (i.e., FDA BAM). For subsp. houtenae, kit B yielded six positive samples at both enrichment times (excluding two samples classified as repetition), compared with seven positive samples with the paired culture-based reference method; one subsp. houtenae–inoculated sample was negative for all enrichment time–kit combinations, except for the combination of kit C and the associated maximum incubation period, indicating low Salmonella levels postincubation.

Overall, none of the results obtained using any of the kits (at either minimum or maximum enrichment times) were significantly different from the reference method (Table 3). These results support acceptable recovery of Salmonella from dark chocolate enriched in skim milk when samples are tested at the minimum range of the validated incubation time, even for difficult-to-detect strains. Because FDA BAM culture confirmation was performed after enrichment for the specified minimum and maximum enrichment time for each kit, our study also generated data on the performance of the complete BAM method (which includes primary enrichment in skim milk followed by selective enrichment in TT at 35°C and RV at 42°C) for primary enrichment periods of 16, 20, and 24 h. The FDA BAM method yielded the same number of positives (40 of 80) at all three incubation times, suggesting that incubation times as low as 16 h can successfully recover fractional positive levels of Salmonella from dark chocolate, when using skim milk as the primary enrichment medium and an incubation temperature of 35°C.

Enrichment in BPW incubated at 42°C is associated with a significantly reduced ability to detect subsp. houtenae and subsp. salamae compared with enrichment in skim milk incubated at 35°C

Because some rapid methods allow for the use of BPW and elevated temperature as part of an alternative enrichment procedure, Salmonella-inoculated dark chocolate samples were also enriched with BPW incubated at 42°C to assess potential differences between the alternative and reference enrichment procedures. Among 80 inoculated dark chocolate samples enriched in BPW, both the rapid method and the standard culture-based method detected Salmonella in 13 (16%) of 80 samples; all 13 samples were positive after both 18 and 24 h of enrichment with BPW at 42°C. As described above, 40 of 80 samples enriched with skim milk for 24 h at 35°C tested positive with the FDA BAM method. The observed difference in the number of Salmonella-positive chocolate samples was strain specific, as recovery of Salmonella Poona (dPOD = −0.10; 95% confidence interval [CI]: −0.37, 0.19) and S. bongori (dPOD = −0.15; 95% CI: −0.37, 0.07) from dark chocolate was not significantly different when BPW and incubation at 42°C was used as the enrichment procedure, as opposed to enrichment with skim milk at 35°C (Table 4). For subsp. houtenae (dPOD = −0.35; 95% CI: −0.57, −0.12) and subsp. salamae (dPOD = −0.75; 95% CI: −0.88, −0.12), the number of Salmonella-positive dark chocolate samples was significantly lower with the BPW and 42°C enrichment procedure than with the reference method enrichment procedure (skim milk incubated at 35°C). Overall, these results suggest that certain combinations of enrichment media and incubation temperatures may reduce the recovery of certain non-subsp. enterica Salmonella strains in the specific dark chocolate matrix used here.

TABLE 4

Difference in probability of detection (dPOD) calculations for Salmonella-positive dark chocolate samples between an alternative enrichment procedure using BPW and 42°C incubation temperature compared with FDA BAM (skim milk and 35°C incubation temperature)

Difference in probability of detection (dPOD) calculations for Salmonella-positive dark chocolate samples between an alternative enrichment procedure using BPW and 42°C incubation temperature compared with FDA BAM (skim milk and 35°C incubation temperature)
Difference in probability of detection (dPOD) calculations for Salmonella-positive dark chocolate samples between an alternative enrichment procedure using BPW and 42°C incubation temperature compared with FDA BAM (skim milk and 35°C incubation temperature)

PCR inhibition from dark chocolate is dependent on both the primary enrichment medium and whether a secondary growth step is performed

Cocoa-containing matrices present additional challenges for PCR-based rapid detection methods because these matrices contain PCR inhibitory substances such as polyphenolic compounds (13, 31, 41). Two of the kits incorporated a dilution of the primary enrichment in BHI (1:10 or 1:200) followed by a regrowth step (i.e., incubation at either 37 or 41.5°C for 3 h). None of the 384 samples tested (i.e., samples and matrix controls across two rapid methods, for both minimum and maximum incubations for all four Salmonella strains) with these regrowth protocols showed evidence of matrix inhibition, as determined by successful amplification of the PCR internal positive controls. For one assay, lysates were performed directly from skim milk– and BPW-enriched dark chocolate samples, without a dilution and regrowth step. Of the 192 dark chocolate samples tested with this enrichment procedure, 2 (1%) of 192 skim milk–enriched samples had PCR inhibition compared with 53 (28%) of 192 BPW-enriched samples. All PCR inhibition was successfully resolved with either a 1:10 dilution of the primary enrichment with BPW or a 1:3 dilution of the sample lysates with kit control buffer, followed by PCR reanalysis. Together, these results suggest that direct PCR analysis of dark chocolate enriched with BPW may be prone to PCR inhibition, although dilution of the primary enrichment can be used to resolve the inhibition.

Use of an alternative confirmation method incorporating a chromogenic agar suggests that secondary selective enrichment procedures may not be necessary for culture confirmation of Salmonella from foods with low microbial loads

The incorporation of a chromogenic agar as part of an alternative confirmation method offers the advantages of both improved specificity and reduced time to results compared with the FDA BAM cultural confirmation method. Traditional culture confirmation methods require a selective secondary enrichment in RV and TT broth. To test whether Salmonella could be detected directly from primary enrichments from foods with a low microbial load, we tested the use of an alternative confirmation method that involved directly plating primary enrichments onto chromogenic and XLD agar plates (without the use of RV and TT secondary enrichments). The 60 samples tested with this approach yielded 20 of 60 and 21 of 60 positive samples after 16 and 20 h of enrichment time, respectively; the same samples were classified as positive by the kit C PCR, whereas the reference FDA BAM method (with 24-h enrichment) yielded 21 of 60 positives. The one false negative result was obtained from the same sample containing subsp. houtenae discussed above (i.e., the sample likely had low levels of Salmonella postincubation), which tested negative at the minimum enrichment time. The alternative confirmation test results for Salmonella Poona and S. bongori agreed with the FDA BAM for all enrichment time points. The subsp. salamae–inoculated dark chocolate samples were not tested due to shipment delays with the chromogenic agar. Although the S. bongori strain used here did not have typical growth on the chromogenic agar, and would therefore have been classified as “not Salmonella” according to the manufacturer's instructions for interpretation, the overall alternative confirmation procedure (which included both XLD and chromogenic agar) yielded the same number of Salmonella-positive samples as the FDA BAM (4 of 20 samples), as S. bongori showed typical growth on XLD, and therefore confirmed the Salmonella detection. Importantly, these results suggest that for low microbial load foods, such as dark chocolate, direct plating of enriched samples may represent a potential alternative culture confirmation method, offering faster time to results.

DISCUSSION

The ability to rapidly and accurately detect foodborne pathogens in ready-to-eat foods represents an important component in achieving the overall goal of reducing morbidity and mortality associated with foodborne illness. In response to industry demands for rapid methods with improved ease of use, the modification of enrichment methods (e.g., media, incubation times) is often adopted to reduce the time to results (or confirmation) and to simplify the workflow. To evaluate the impact of these modifications, we assessed the ability of three AOAC-approved, PCR-based methods for Salmonella spp. detection under deliberately designed worst-case scenarios: (i) use of strains that are difficult to detect, (ii) minimum enrichment incubation times, (iii) a low-water-activity matrix known to contain PCR and bacterial growth inhibitors (i.e., dark chocolate), and (iv) alternative enrichment schemes that include elevated temperatures and a nonskim milk enrichment medium. Our data indicate that (i) careful selection of Salmonella strains used for assay evaluation can reveal method weaknesses; (ii) use of the minimum enrichment times validated for different assays has limited impact on Salmonella detection in chocolate; (iii) use of alternative enrichment procedures in rapid methods requires optimization and can lead to reduced method performance, including PCR inhibition; and (iv) alternative simplified confirmation methods that include a chromogenic agar represent a promising alternative to current more time-intensive Salmonella confirmation protocols required by the FDA BAM.

Careful selection of Salmonella strains used for assay evaluation can reveal method weaknesses

Because Salmonella represents a highly diverse genus with >2,600 serovars, selection of strains used for inclusivity assessment and matrix studies (i.e., assay evaluation on inoculated “matrix” samples) can have a considerable impact on the outcome of validation studies, as has been observed in previous studies (4, 42). This challenge is also specifically addressed by AOAC guidelines (1) that require the use of at least 100 Salmonella strains in inclusivity assessments, as opposed to 50 strains for other foodborne pathogens (e.g., Listeria monocytogenes). In our study, the inclusivity assessment with a smaller but carefully assembled set of 70 Salmonella strains was able to identify Salmonella strains that can be designated as “difficult-to-detect” with a given assay. For example, some strains only yielded positive PCR results for one of three replicates (using concentrations 10-fold higher than the calculated assay LOD). In addition, some strains showed consistently later CT values, suggesting reduced PCR performance for these strains. These findings are consistent with previous studies (4, 42) that have also shown that PCR performance was not uniform across a diverse set of Salmonella strains. In general, S. bongori and S. enterica subsp. other than subsp. enterica are more likely to yield weak or negative results, particularly at low inoculum levels, which is consistent with the observation that invA, a target gene for many Salmonella PCR assays, shows considerable sequence diversity in these strains relative to S. enterica subsp. enterica (4). Although S. bongori and non-subsp. enterica strains are less commonly associated with human illness, they are still capable of causing human clinical infections (23), and they have been isolated from food (29). Importantly, the AOAC Guidelines for Validation of Microbiological Methods for Foods and Environmental Surfaces and the ISO 16140-2 protocol for the validation of alternative (proprietary) methods against a reference method both stipulate the use of a diverse strain set, but they do not specify a set of standard strains nor do they require an inclusivity panel that includes all species and subspecies (1, 20). Hence, Salmonella spp. test kits could achieve AOAC or ISO approval (with reported assay inclusivity of 100%), even if a given kit may not (reliably) detect certain species, subspecies, or serovars. Our findings suggest that standards organizations should require the use of a core standardized strain set that, at a minimum, includes both known Salmonella species, all the S. enterica subsp., and certain difficult-to-detect strains (e.g., slow-growing strains or strains with unusual phenotypic characteristics). In addition, it is important for end users to review the strains used for inclusivity testing for validating a given assay to determine whether additional inclusivity verification or validation is needed. For example, in some cases inclusivity verification with S. bongori and all S. enterica subspecies may be needed, whereas in other cases inclusivity may need to be verified with specific serovars that are regularly found in a given region or supply chain but that were not included in the inclusivity panel. Strain selection for validation studies can also draw on existing data that report variation in growth characteristics among Salmonella serovars (10, 30, 43, 44). For example, previous studies have identified strains of serovars Enteritidis (30), Dublin (10), and Gallinarum (44) to have longer lag-phase growth rates, lower optimal growth temperatures, or variable growth in response to different environmental stresses (e.g., pH and NaCl concentration).

Although a large number of strains need to be included in the inclusivity evaluations, typically only a single strain is used to evaluate a method with a given matrix (the so-called matrix study). Hence, the selection of the specific strains used in matrix studies will also have a considerable impact on the outcome of the overall method validation. Similar to inclusivity studies, there is no requirement for specific strains to be used in matrix studies, and it would hence be possible to exclude strains that may be difficult to detect. Our study showed that inclusion of some difficult-to-detect Salmonella strains (based on the initial inclusivity study data) showed reduced detection relative to gold standard methods (i.e., FDA BAM); however, these differences were not statistically significant (i.e., P > 0.05 for all comparisons). In contrast, other studies (30, 42) have found wide variation among several strains of S. enterica subsp. enterica when subjected to suboptimal growth conditions (e.g., unfavorable NaCl and pH concentrations, low water activity). One of these studies (42) reported that several AOAC-approved methods showed significantly reduced detection of Salmonella in low-water-activity foods relative to gold standard methods (e.g., assay false-negative rates >40% with two S. enterica serovars) (42), indicating that selective use of specific strains in validation studies may be a problem. Inadvertent use of “easy-to-detect” strains in matrix studies with low-water-activity foods (such as chocolate) may also occur because some strains of Salmonella Montevideo, Salmonella Senftenberg, and Salmonella Oranienburg, which have demonstrated both prolonged survival in dry environments and a higher heat tolerance (facilitating matrix inoculation and predictable die-off rates), also show faster growth rates than strains from other serovars (3, 25, 28, 44), which may improve their ability to be detected with rapid methods, particularly those using shortened enrichment times. Overall, our data as well as those of previous studies indicate that selection of strains used in the matrix studies represents an important component of a robust method evaluation, particularly because interactions between strains and the matrix tested may lead to situations where only specific strains show high false negative rates with a given matrix. End users thus need to carefully assess the strains used in initial matrix validation studies and consider whether additional matrix validation or verification studies with specific strains of concern or importance are needed. In the future, it may also be possible for standards agencies to either define specific Salmonella strains to be used with different matrices or perform an initial review of inclusivity test data to select appropriate challenge strains for matrix studies.

Use of minimum enrichment times validated for different assays has limited impact on Salmonella detection in dark chocolate

The findings from our study support that shortening the FDA BAM–specified 24-h skim milk primary enrichment time by up to 8 h (i.e., to 16 h) has no significant impact on detection of Salmonella from dark chocolate with either PCR-based or cultural methods. Specifically, the 80 samples tested (20 samples for each of four strains) yielded the same number of positives (n = 40) with the FDA BAM method with 16- and 24-h enrichment. Use of a 16-h FDA BAM method enrichment, however, yielded a false-negative result for one sample, with both kits allowing for 16-h primary enrichment; this enrichment appears to have had very low levels of S. enterica subsp. houtenae postincubation. In agreement with our findings, an equivalent reference method, ISO 6579-1:2017, also stipulates a minimum incubation time of 16 h for dark chocolate enriched with skim milk (21). The ISO 6579-1:2017 method has utilized a minimum incubation time of 16 h for the primary enrichment of multiple matrices, including dark chocolate for over 15 years, suggesting the robustness of this enrichment method (21). Consequently, several Association Française de Normalisation–certified PCR methods (33), which are validated against the ISO method instead of the FDA BAM, adhere to a minimum enrichment incubation time of 16 h. Overall, our results suggest that use of minimum enrichment times (in skim milk incubated at 35°C) does not compromise the performance of rapid PCR-based methods for detection of Salmonella in dark chocolate.

Alternative enrichment procedures require optimization and can lead to reduced method performance, including PCR inhibition

Cocoa-containing matrices have historically required enrichment with skim milk for the detection of Salmonella spp. (16, 22, 36, 48) because the casein present in milk is proposed to mitigate the antimicrobial properties of the polyphenolic compounds found in cocoa (51). Alternative enrichment procedures that use BPW are an attractive option because they would enable the use of a single enrichment medium for detection of multiple foodborne pathogens. In this study, we challenged a simplified alternative BPW enrichment procedure with worst-case scenarios, including (i) difficult-to-detect strains as described above and (ii) an elevated incubation temperature at 42°C. The 42°C incubation temperature has been used historically for recovering noninjured Salmonella cells from samples containing high levels of competitive microbes (10, 50). This alternative enrichment procedure yielded results that were not significantly different from the FDA BAM results with the only S. enterica subsp. enterica strain tested (serovar Poona) as well as the S. bongori strain tested. Published AOAC and Association Française de Normalisation validation studies (2, 34, 35) that incorporated six subsp. enterica strains and six cocoa-containing product types also reported that the alternative BPW enrichment protocol using incubation at 42°C was not significantly different from the reference methods (both FDA BAM and ISO) with respect to detection of S. enterica subsp. enterica from cocoa-containing products. Our study however found that the non-subsp. enterica strains representing subsp. houtenae and subsp. salamae yielded significantly fewer Salmonella-positive samples with this alternative enrichment than with the reference method. Other studies also support that use of BPW as the enrichment medium for detection of Salmonella from chocolate (both dark and milk) can be challenging (24, 50). For example, a study by Jasson et al. (24) did not recover any low-level, sublethally injured subsp. enterica strains directly inoculated into BPW-enriched chocolate incubated at 35°C. However, this study only included three replicates and the inoculation procedure (i.e., direct inoculation of enrichment media with added chocolate) is not reflective of low-water-activity environmental stress, which would have physiological effects on the Salmonella (15). Together, these data suggest that for some combinations of chocolate type and Salmonella serovar, the combination of BPW media and incubation at elevated temperatures may not be as effective as skim milk for Salmonella enrichment. Some studies however suggest strategies that may allow for improved Salmonella enrichment when using BPW as the enrichment medium. For example, Wilson et al. (50) demonstrated enhanced recovery of Salmonella from chocolate (n = 5 samples) with enrichment in BPW by reducing the incubation temperature from 43 to 35°C. They also found that enrichment with BPW at 43°C recovered fewer Salmonella spp. cells from chocolate and was considered to be “inhibitory” in that study (50). Furthermore, an AOAC study that validated enrichment procedures for chocolate by using BPW supplemented with powdered skim milk (100 g of skim milk per 225 mL of BPW) and incubation at 35°C showed no statistical difference compared with the FDA BAM in the number of detected Salmonella-positive chocolate samples (40). Overall, these data suggest that further optimization of a BPW enrichment method for chocolate is needed, because non-subsp. enterica strains may have reduced recovery from BPW-enriched dark chocolate, as observed in our study. Future studies assessing the effects of lowering incubation temperatures of BPW-enriched chocolate to 35°C and/or including skim milk powder–supplemented BPW for recovery of difficult-to-detect strains, such as the non-subsp. enterica strains tested here, may be beneficial for improving the use of BPW enrichment procedures for the recovery of diverse Salmonella spp. from matrices with potentially inhibitory compounds, such as polyphenolic compounds in dark chocolate.

PCR analysis of chocolate samples enriched in BPW also resulted in higher rates of PCR inhibition compared with PCR analysis of samples enriched with skim milk. PCR inhibition was defined in this study as a failure to amplify the internal control, which was incorporated in the master mix of the kit, to sufficient levels. When lysis was performed from the primary enrichment, 28% of BPW-enriched chocolate samples resulted in PCR inhibition compared with just 1% of skim milk enriched samples tested with the same kit and lysis procedure. Although the exact cause of the observed PCR inhibition was not determined in this study, previous studies have shown that when polyphenolic compounds present in chocolate are not “neutralized” by the addition of skim milk, PCR inhibition can occur (13, 31, 41). Furthermore, transfer of noticeably higher amounts of chocolate particulate material into the lysis reaction could also explain the observed inhibition, because the amount of chocolate particulate in BPW-enriched samples was noticeably greater than with the skim milk–enriched samples, despite the use of Whirl-Pak bags with filters. To address the PCR inhibition, we repeated the PCR analysis after diluting the lysate or the primary enrichment; both dilution approaches were successful at resolving the PCR inhibition in these samples. In addition, allowing particulate matter to settle to the bottom of the tube (performed for a set of seven subsamples) before lysis procedures also successfully resolved inhibition (data not shown). Regardless of the mechanism of inhibition, our data showed that incorporating a dilution step resolved PCR inhibition without impacting the sensitivity of the method. Furthermore, our data support that further studies should be conducted to optimize the processing of BPW-enriched dark chocolate to determine the best method for reducing PCR inhibition in these samples.

Alternative confirmation methods that include a chromogenic agar represent a promising alternative to current more time intensive Salmonella confirmation protocols

In contrast to the traditional selective and differential Salmonella agars included in the FDA BAM method (e.g., XLD, HE), chromogenic agars contain substrates that facilitate uniquely colored, target-specific growth, which improve the differential capacity of the agar (32, 37, 38). When cultured onto traditional selective and differential agars, non-Salmonella enteric organisms are often indistinguishable from atypical Salmonella strains (e.g., hydrogen sulfide negative, lactose positive), thus requiring additional identification procedures for confirmation. The improved specificity of chromogenic agars allows for shortened time to results due to the elimination of subsequent subculturing steps currently specified by the FDA BAM method (48). In our study, we were able to identify presumptive Salmonella-positive samples 24 h faster than the FDA BAM cultural confirmation method, with no significant difference in the number of confirmed Salmonella-positive samples, demonstrating the feasibility of alternative confirmation methods that allow for direct analysis of enriched samples. The reliability of the alternative confirmation examined in this study is further supported by the ISO 6579-1:2017 method, which also incorporates the use of both a traditional selective and differential agar (e.g., XLD) and a chromogenic agar (21), although the use of a secondary selective enrichment is still required by this method (21). The incorporation of chromogenic agars in standard cultural confirmation procedures has also demonstrated an improved ability to detect strains with atypical growth. For example, in an outbreak involving flour contaminated with a difficult-to-detect strain of Escherichia coli O121, the chromogenic agar correctly identified the strain as Shiga toxin–producing E. coli, whereas the traditional confirmation procedure indicated the isolate was not E. coli (7), suggesting that the incorporation of chromogenic agar media into standard culture confirmation methods should be considered by regulatory agencies. Our results combined with those from the flour outbreak, as well as the ISO method, support that a culture confirmation procedure that combines a chromogenic agar and a traditional selective and differential agar (e.g., XLD) is a robust and simplified (i.e., fewer steps, faster time to results) alternative to the FDA BAM confirmation method.

In conclusion, our study reinforces a critical role for strain selection when validating the applicability of a method for detection of a diverse group of bacteria such as Salmonella spp. in a specific food matrix. Furthermore, challenging detection methods by using worst-case scenarios such as the recovery of difficult-to-detect Salmonella strains from matrices such as dark chocolate supports the accuracy of shortened primary enrichment times (16 versus 24 h) when using skim milk as the enrichment medium and the value of incorporating a chromogenic agar as part of a simplified confirmation method. Although alternative enrichment procedures may offer a simplified approach, the potential for reduced recovery of specific strains from food matrices containing potentially inhibitory compounds should be considered when performing validation studies.

ACKNOWLEDGMENTS

This project was supported by contracts from Mars, Inc. (McLean, VA) and bioMérieux (Marcy-l'Étiole, France) to M. Wiedmann.

SUPPLEMENTAL MATERIAL

Supplemental material associated with this article can be found online at: https://doi.org/10.4315/JFP-20-066.s1

REFERENCES

REFERENCES
1.
AOAC International
.
2012
.
Methods committee guidelines for the validation of microbiological methods for food and environmental surfaces
.
AOAC International
,
Gaithersburg, MD
.
2.
AOAC Research Institute Performance Tested Methods
.
2017
.
GENE-UP Salmonella 2 (SLM 2)
.
Manufactured by bioMérieux. AOAC PTM certification no. 121802
.
AOAC International
,
Gaithersburg, MD
.
3.
Baylis,
C. L.,
MacPhee
S.,
Robinson
A. J.,
Griffiths
R.,
Lilley
K.,
and
Betts
R. P.
2004
.
Survival of Escherichia coli O157:H7, O111:H- and O26:H11 in artificially contaminated chocolate and confectionery products
.
Int. J. Food Microbiol
.
96
:
35
48
.
4.
Buehler,
A. J.,
Wiedmann
M.,
Kassaify
Z.,
and
Cheng
R. A.
2019
.
Evaluation of invA diversity among Salmonella species suggests why some commercially available rapid detection kits may fail to detect multiple Salmonella subspecies and species
.
J. Food Prot
.
82
:
710
717
.
5.
Centers for Disease Control and Prevention
.
2019
.
Preliminary incidence and trends of infections with pathogens transmitted commonly through food — Foodborne Disease Active Surveillance Network, 10 U.S. sites, 2015–2018
.
6.
Craven,
P. C.,
Mackel
D. C.,
Baine
W. B.,
Barker
W. H.,
Gangarosa
E. J.,
Goldfield
M.,
Rosenfeld
H.,
Altman
R.,
Lachapelle
G.,
Davis
J. W.,
and
Swanson
R. C.
1975
.
International outbreak of Salmonella Eastbourne infection traced to contaminated chocolate
.
Lancet
1
(7910)
:
788
792
.
7.
Crowe,
S. J.,
Bottichio
L.,
Shade
L. N.,
Whitney
B. M.,
Corral
N.,
Melius
B.,
Arends
K. D.,
Donovan
D.,
Stone
J.,
Allen
K.,
Rosner
J.,
Beal
J.,
Whitlock
L.,
Blackstock
A.,
Wetherington
J.,
Newberry
L. A.,
Schroeder
M. N.,
Wagner
D.,
Trees
E.,
Viazis
S.,
Wise
M. E.,
and
Neil
K. P.
2017
.
Shiga toxin-producing E. coli infections associated with flour
.
N. Engl. J. Med
.
377
:
2036
2043
.
8.
De Boer,
S. H.,
Ward
L. J.,
Li
X.,
and
Chittaranjan
S.
1995
.
Attenuation of PCR inhibition in the presence of plant compounds by addition of BLOTTO
.
Nucleic Acids Res
.
23
:
2567
2568
.
9.
Finn,
S.,
Condell
O.,
McClure
P.,
Amezquita
A.,
and
Fanning
S.
2013
.
Mechanisms of survival, responses and sources of Salmonella in low-moisture environments
.
Front. Microbiol
.
4
:
331
.
10.
Goodman,
L. B.,
McDonough
P. L.,
Anderson
R. R.,
Franklin-Guild
R. J.,
Ryana
J. R.,
Perkins
G. A.,
Thachil
A. J.,
Glaser
A. L.,
and
Thompson
B. S.
2017
.
Detection of Salmonella spp. in veterinary samples by combining selective enrichment and real-time PCR
.
J. Vet. Diagn. Invest
.
29
:
844
851
.
11.
Greenwood,
M. H.,
and
Hooper
W. L.
1983
.
Chocolate bars contamination with Salmonella Napoli: an infectivity study
.
Br. Med. J
.
286
:
1394
.
12.
Greig,
J. D.,
and
Ravel
A.
2009
.
Analysis of foodborne outbreak data reported internationally for source attribution
.
Int. J. Food Microbiol
.
130
:
77
87
.
13.
Gryson,
N.,
Dewettinck
K.,
and
Messens
K.
2007
.
Influence of cocoa components on the PCR detection of soy lecithin DNA
.
Eur. Food Res. Technol
.
226
:
247
254
.
14.
Gurtler,
J. B.,
Doyle
M. P.,
and
Kornacki
J. L.
2014
.
The microbiological safety of low water activity foods and spices: correcting historic misassumptions
,
p.
3
13
.
In
Gurtler
J. B.,
Doyle
M. P.,
and
Kornacki
J. L.
(ed.),
The microbiological safety of low water activity foods and spices.
Springer
,
New York
.
15.
Gurtler,
J. B.,
Keller
S. E.,
Kornacki
J. L.
Annous
B. A.,
Jin
T.,
and
Fan
X.
2019
.
Challenges in recovering foodborne pathogens from low-water-activity foods
.
J. Food Prot
.
82
:
988
996
.
16.
Health Canada
.
2009
.
MFHPB-20—Isolation and identification of Salmonella from food and environmental samples. Compendium of Analytical Methods, vol. 2
.
Health Canada
,
Ottawa
.
17.
Hiramatsu,
R.,
Matsumoto
M.,
Sakae
K.,
and
Miyazaki
Y.
2005
.
Ability of shiga toxin–producing Escherichia coli and Salmonella spp. to survive in a desiccation model system and in dry foods
.
Appl. Environ. Microbiol
.
71
:
6657
6663
.
18.
Hockin,
J. C.,
D'Aoust
J. Y.,
Bowering
D.,
Jessop
J. H.,
Khanna
B.,
Lior
H.,
and
Milling
M. E.
1998
.
An international outbreak of Salmonella Nima from imported chocolate
.
J. Food Prot
.
52
:
51
54
.
19.
Huedo,
P.,
Gori
M.,
Amato
E.,
Bianchi
R.,
Valerio
E.,
Magnoli
L.,
and
Pontello
M.
2016
.
A multischool outbreak due to Salmonella enterica serovar Napoli associated with elevated rates of hospitalizations and bacteremia, Milan, Italy, 2014
.
Foodborne Pathog. Dis
.
13
:
417
422
.
20.
International Organization for Standardization
.
2016
.
Microbiology of the food chain—method validation—part 2: Protocol for the validation of alternative (proprietary) methods against a reference method. ISO 16140-2:2016
.
International Organizations for Standardization
,
Geneva
.
21.
International Organization for Standardization
.
2017
.
Microbiology of the food chain—horizontal method for the detection, enumeration and serotyping of Salmonella—part 1: Detection of Salmonella spp. ISO 6579-1
.
International Organization for Standardization
,
Geneva
.
22.
International Organization for Standardization
.
2017
.
Microbiology of the food chain—preparation of test samples, initial suspension and decimal dilutions for microbiological examination—part 4: Specific rules for the preparation of miscellaneous products. ISO 6887-4:2017
.
International Organization for Standardization
,
Geneva
.
23.
Issenhuth-Jeanjean,
S.,
Roggentin
P.,
Mikoleit
M.,
Guibourdenche
M.,
de Pinna
E.,
Nair
S.,
Fields
P. I.,
and
Weill
F. X.
2014
.
Supplement 2008–2010 (no. 48) to the White-Kauffmann-Le Minor scheme
.
Res. Microbiol
.
165
:
526
530
.
24.
Jasson,
V.,
Baert
L.,
and
Uyttendaele
M.
2011
.
Detection of low numbers of healthy and sub-lethally injured Salmonella enterica in chocolate
.
Int. J. Food Microbiol
.
145
:
488
491
.
25.
Jueve,
B. J.,
Cox
N. A.,
Bailey
J. S.,
Thomson
J. E.,
Charles
O. W.,
and
Shutze
J. V.
1984
.
Survival of Salmonella in dry food and feed
.
J. Food Prot
.
47
:
445
448
.
26.
Kim,
J. S.,
Lee
G. G.,
Park
J. S.,
Jung
Y. H.,
Kwak
H. S.,
Kim
S. B.,
Nam
Y. S.,
and
Kwon
S.
2007
.
A novel multiplex PCR assay for rapid and simultaneous detection of five pathogenic bacteria: Escherichia coli O157:H7, Salmonella, Staphylococcus aureus, Listeria monocytogenes, and, Vibrio parahaemolyticus
.
J. Food Prot
.
70
:
1656
1662
.
27.
Kirk,
M. D.,
Pires
S. M.,
Black
R. E.,
Caipo
M.,
Crump
J. A.,
Devleesschauwer
B.,
Dopfer
D.,
Fazil
A.,
Fischer-Walker
C. L.,
Hald
T.,
Hall
A. J.,
Keddy
K. H.,
Lake
R. J.,
Lanata
C. F.,
Torgerson
P. R.,
Havelaar
A. H.,
and
Angulo
F. J.
2015
.
World Health Organization estimates of the global and regional disease burden of 22 foodborne bacterial, protozoal, and viral diseases, 2010: a data synthesis
.
PLoS Med
.
12
:
e1001921
.
28.
Kumar,
A.,
and
Kumar
S.
2003
.
Survival kinetics of Salmonella enterica serotype Senftenberg (S. Senftenberg) after heat and acid stress
.
World J. Microbiol. Biotechnol
.
19
:
985
987
.
29.
Lamas,
A.,
Miranda
J. M.,
Regal
P.,
Vasquez
B.,
Franco
C. M.,
and
Cepeda
A.
2018
.
A comprehensive review of non-enterica subspecies of Salmonella enterica
.
Microbiol. Res
.
206
:
60
73
.
30.
Lianou,
A.,
and
Koutsoumanis
K. P.
2011
.
Effect of the growth environment on the strain variability of Salmonella enterica kinetic behavior
.
Food Microbiol
.
28
:
828
837
.
31.
Luck,
G.,
Liao
H.,
Murray
N. J.,
Grimmer
H. R.,
Warminski
E. E.,
Williamsom
M. P.,
Lilley
T. H.,
and
Haslam
E.
1994
.
Polyphenols, astringency and proline-rich proteins
.
Phytochemistry
37
:
357
371
.
32.
Manafi,
M.
2000
.
New developments in chromogenic and fluorogenic culture media
.
Int. J. Food Microbiol
.
60
:
205
218
.
33.
NF Validation by AFNOR Certification
.
n.d
.
Food industry/Salmonella spp
.
4
December
2019
.
34.
NF Validation by AFNOR Certification
.
2018
.
GENE-UP Salmonella method for the detection of Salmonella spp. in a broad range of foods, pet foods and production environmental samples
.
Holder: bioMe´riex.
Certificate no. BIO 12/38-06/26. Summary report (41.5°C)
.
Paris, France
.
35.
NF Validation by AFNOR Certification
.
2018
.
GENE-UP Salmonella method for the detection of Salmonella spp. in a broad range of foods, pet foods and production environmental samples
.
Holder: bioMe´rieux.
Certificate no. BIO 12/38-06/26. Summary report (37°C)
.
Paris, France
.
36.
Norma Oficial Mexicana
.
2014
.
NOM-210-SSA1-2014, Productos y sericios. Métodos de pruueba microbiológicos. Determinación de microorganismos indicadores. Determinación de microorganismos patógenos
.
Direccion General de Normas
,
Mexico City, Mexico
.
37.
Park,
S. H.,
Ryu
S.,
and
Kang
D. H.
2012
.
Development of an improved selective and differential medium for isolation of Salmonella spp
.
J. Clin. Microbiol
.
50
:
3222
3226
.
38.
Perez,
J. M.,
Cavalli
P.,
Roure
C.,
Renac
R.,
Gille
Y.,
and
Frediere
A. M.
2003
.
Comparison of four chromogenic media and Hektoen agar for detection and presumptive identification of Salmonella strains in human stools
.
J. Clin. Microbiol
.
41
:
1130
1134
.
39.
Podolak,
R.,
Enache
E.,
Stone
W.,
Black
D. G.,
and
Elliott
P. H.
2010
.
Sources and risk factors for contamination, survival, persistence, and heat resistance of Salmonella in low-moisture foods
.
J. Food Prot
.
73
:
1919
1936
.
40.
Qiagen
.
2013
.
Product insert for the AOAC-RI PTM certified and NF Validation certified mericon automated and manual Salmonella detection workflows
.
Qiagen
,
Hilden, Germany
.
41.
Rowlyk,
K.,
Zhang
Z.,
and
Kermekchiev
M.
2014
.
Composition to overcome inhibitors in PCR and growth cultures
.
U.S. provisional patent application series 61/425,913
.
42.
Ryan,
G.,
Roof
S.,
Post
L.,
and
Wiedmann
M.
2015
.
Evaluation of rapid molecular detection assays for Salmonella in challenging food matrices at low inoculation levels and using difficult-to-detect strains
.
J. Food Prot
.
78
:
1632
1641
.
43.
Scallan,
E.,
Hoekstra
R. M.,
Angulo
F. J.,
Tauxe
R. V.,
Widdowson
M. A.,
Roy
S. L.,
Jones
J. L.,
and
Griffin
P. M.
2011
.
Foodborne illness acquired in the United States–major pathogens
.
Emerg. Infect. Dis
.
17
:
7
15
.
44.
Stokes,
J. L.,
and
Bayne
H. G.
1957
.
Growth rate of Salmonella colonies
.
J. Bacteriol
.
74
:
200
206
.
45.
U.S. Food and Drug Administration
.
2001
.
Investigation of food implicated in illness, chap. 25
.
In
Jackson
G. J.
and
Klontz
K. C.
(ed.),
Bacteriological analytical manual
.
46.
U.S. Food and Drug Administration
.
2011
.
Most probable number from serial dilutions, app. 2
.
In
Blodgett
R.
and
Zhang
G.
(ed.),
Bacteriological analytical manual
.
47.
U.S. Food and Drug Administration
.
2015
.
Guidelines for the validation of analytical methods for the detection of microbial pathogens in foods and feeds, 2nd ed., app. 3
.
Bacteriological analytical manual
.
48.
U.S. Food and Drug Administration
.
2018
.
Salmonella, chap. 5
.
In
Andrews
W. H.
and
Hammack
T.
(ed.),
Bacteriological analytical manual
.
49.
Vella,
J.
2015
.
Case of Salmonella Montevideo in Cadbury chocolate products in 2006
.
University of Central Lancashire
,
Preston, UK
.
50.
Wilson,
C. R.,
Andrews
W. H.,
and
Poelma
P. L.
1980
.
Recovery of Salmonella from milk chocolate using a chemically defined medium and five nondefined broths
.
J. Food Sci
.
45
:
310
313
.
51.
Zapatka,
F. A.,
and
Varney
G. W.
1977
.
Neutralization of the bactericidal effect of cocoa powder on Salmonella by casein
.
J. Appl. Bacteriol
.
42
:
21
25
.

Supplementary data