USP Equivalency Study: PathoSEEK®
Salmonella, E. coli & S. aureus for Dietary Supplements

An independent contract research laboratory compared the PathoSEEK® Salmonella, E. coli, and S. aureus Multiplex Detection Assay with MaGiC Lysis against USP <2022> compendial plating in six dietary supplement matrices. The study reports results for all six USP <1223> parameters for qualitative methods: specificity, limit of detection, repeatability, ruggedness, robustness, and equivalency. Equivalency, the one parameter tested against a pass/fail criterion, was met.

The qPCR method agreed with compendial results on every inoculated test portion for all three organisms, at every inoculation level, on both qPCR instruments, with zero false positives. Non-inferiority was demonstrated at low, intermediate, and high levels tested (no discordant test portions; δ = 0.20, α = 0.05). A separate dilution series using certified reference materials put the LOD₅₀ at 0.18 to 0.40 nominal CFU.

There were no discordant results to explain. The one practical note concerns botanical powder that has been autoclaved, covered in the Robustness section.

Study at a glance

ReportUSP <1223> Alternative Method Equivalency Study for Detection of Salmonella spp., Staphylococcus aureus, and Escherichia coli in Dietary Supplements (P0103-RPT-001-03), Version 3 (October 2026)
MethodPathoSEEK® Salmonella, Escherichia coli, and Staphylococcus aureus Multiplex Detection Assay with MaGiC Lysis
Target organismsSalmonella enterica NCTC 6676, Staphylococcus aureus NCTC 10788 / ATCC 6538, Escherichia coli NCTC 12923 / ATCC 8739
Off-target challengeKlebsiella oxytoca NCTC 8167, Citrobacter koseri ATCC 25408
MatricesEnergy shot, vitamin gummy, vitamin powder, botanical powder, MCT oil, chocolate bar
Reference methodUSP <2022> (harmonized with USP <62>): MacConkey Agar for E. coli, Xylose Lysine Deoxycholate (XLD) Agar for Salmonella, Mannitol Salt Agar for S. aureus
ConfirmationIndole and Citrate tests (E. coli, Salmonella); DNase Agar and Coagulase Test (S. aureus)
LOD reference materialCertified reference materials (CRMs), matrix-independent dilution series, direct lysis
Conditions variedTwo qPCR instruments; one analyst and one extraction method (automated)
InstrumentsBio-Rad CFX Opus, BioMolecular Systems Mic, Myra liquid handler for automated extraction
Conducted byCambium Analytica

Results by matrix

All six matrices showed complete agreement with compendial methods at every level tested, for all three organisms. Full data for each parameter follows below.

Matrix

Low-level spike, CFU/test portion (Salmonella / E. coli / S. aureus)

Low-level detection, qPCR vs. compendial

Intermediate and high-level detection

Specificity

Highest average RSD (Cq)

Relative trueness

Energy shot

2 / 8 / 6

8/8 vs. 8/8

100%

100%

2.27%

100%

Vitamin gummy

2 / 5 / 3

8/8 vs. 8/8

100%

100%

0.83%

100%

Vitamin powder

4 / 2 / 2

8/8 vs. 8/8

100%

100%

2.80%

100%

Botanical powder

1 / 2 / 4

8/8 vs. 8/8

100%

100%

0.76%

100%

MCT oil

6 / 1 / 1

8/8 vs. 8/8

100%

100%

2.55%

100%

Chocolate bar

1 / 3 / 3

8/8 vs. 8/8

100%

100%

2.46%

100%

Low-level detection counts apply to each organism on each instrument. Relative trueness is measurement-level (each test portion and target counted once per instrument). Source: Tables 3, 6 to 12, and 17 of P0103-RPT-001-03.

Chocolate is the most demanding matrix in the set, with high fat, cocoa polyphenols, and methylxanthines. Per the User Guide, its lysate is diluted 1:100 before qPCR and its internal control is added at 1:5,000. With those steps, the internal amplification control amplified in every chocolate sample, and every low-level sample was detected for all three organisms, including those carrying 1 CFU of Salmonella per test portion.

Study design

Each sample was tested side by side using the qPCR method and USP <2022> compendial plating, with compendial colonies confirmed by biochemical tests. Before the matrix work, a matrix-independent LOD was determined using certified reference materials.

Inoculation levelSamples per matrixPurpose
Uninoculated blank (0 CFU)2Confirm no false positives from matrix
Off-target (K. oxytoca + C. koseri)2Confirm no cross-reactivity
Low level, 1 to 10 CFU per test portion8Challenge sensitivity
Intermediate level8Confirm detection at moderate contamination
High level2Confirm complete detection

All three target organisms were co-inoculated into every positive sample, so each channel had to find its target in a mixed contamination. The low level was set within 1 to 10 CFU per test portion and was not adjusted to produce fractional results. Samples were enriched in buffered peptone water for 20 to 24 hours at 37 °C. DNA was extracted with MaGiC Lysis on the Myra automated liquid handler, and each extract was run on both the Bio-Rad CFX Opus and the Mic. Each reaction detected E. coli (FAM), Salmonella (ROX), S. aureus (Cy5), and an internal amplification control (HEX) in a single well.

Per the User Guide, a target is positive at Cq < 40. In this study a microbiologist also reviewed every amplification curve and accepted only clean exponential curves, which generally crossed at Cq ≤ 37. A result is valid only if the internal control amplifies at Cq ≤ 35.

The CRM series ran at 5, 2.5, 1, 0.5, 0.25, and 0.1 nominal CFU, five replicates per level, extracted directly without matrix or enrichment.

Specificity

The assay produced zero false positives across all six matrices and both instruments. Specificity was 100% for the qPCR method and the compendial method alike.

Specificity here means no positive results in uninoculated blanks or in samples spiked with off-target organisms (K. oxytoca NCTC 8167, C. koseri ATCC 25408). Recovery of the target organisms is covered under Limit of detection.

Matrix

Instrument

True negatives

False positives

Specificity, qPCR

Specificity, compendial

Energy shot

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

Vitamin gummy

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

Vitamin powder

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

Botanical powder

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

MCT oil

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

Chocolate bar

Bio-Rad / Mic

4 / 4

0 / 0

100%

100%

True negatives are counted per sample; all three target channels were negative in every one. Source: Table 3. The internal amplification control met its validity criterion (Cq ≤ 35) in every sample, so the negatives were valid. Inclusivity and exclusivity across strain panels were established separately by Medicinal Genomics and were not re-tested in this study.

Limit of detection

A CRM dilution series put the LOD₅₀ at 0.18 to 0.40 nominal CFU, and in the matrices the qPCR method detected every low-level sample, the same as compendial plating. LOD was evaluated two ways: a CRM dilution series run by direct lysis without matrix or enrichment, and detection in each matrix reported descriptively as Probability of Detection (POD) and dPOD (qPCR POD minus compendial POD). At 8 samples per level, per-matrix data cannot establish non-inferiority on their own, so the method comparison rests on the pooled analysis under Equivalency.

Nominal CFU

E. coli, Bio-Rad

E. coli, Mic

Salmonella, Bio-Rad

Salmonella, Mic

S. aureus, Bio-Rad

S. aureus, Mic

5

5/5

5/5

5/5

5/5

5/5

5/5

2.5

5/5

5/5

5/5

5/5

5/5

5/5

1

5/5

5/5

5/5

4/5

5/5

5/5

0.5

4/5

5/5

3/5

4/5

4/5

4/5

0.25

2/5

3/5

1/5

1/5

3/5

1/5

0.1

0/5

0/5

0/5

0/5

0/5

0/5

Target

Instrument

LOD₅₀ (95% CI)

LOD₉₅ (95% CI)

E. coli

Bio-Rad CFX Opus

0.27 (0.14, 0.51)

1.16 (0.62, 2.19)

E. coli

Mic

0.18 (0.09, 0.34)

0.78 (0.41, 1.48)

Salmonella spp.

Bio-Rad CFX Opus

0.38 (0.20, 0.71)

1.63 (0.87, 3.05)

Salmonella spp.

Mic

0.40 (0.22, 0.76)

1.75 (0.93, 3.27)

S. aureus

Bio-Rad CFX Opus

0.23 (0.12, 0.43)

0.99 (0.53, 1.87)

S. aureus

Mic

0.31 (0.17, 0.59)

1.36 (0.72, 2.55)

Five replicates per level. Estimates computed with the PODLOD program referenced in ISO 16140-2. Source: Tables 4 and 5.

Values are in nominal CFU, as stated on the CRM certificates. Detection at 0.5 and 1 nominal CFU ran higher than a one-culturable-cell model predicts, which the report attributes to the assay picking up DNA from cells not counted as culturable CFU. So these figures are not a per-cell detection limit. With five replicates per level the confidence intervals are wide, and they overlap between instruments for every target.

In-matrix detection

Matrix

Organism

CFU/test portion

POD, qPCR

POD, compendial

dPOD (95% CI)

Energy shot

Salmonella spp.

2

1.00

1.00

0.00 (−0.31, 0.31)

Energy shot

E. coli

8

1.00

1.00

0.00 (−0.31, 0.31)

Energy shot

S. aureus

6

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin gummy

Salmonella spp.

2

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin gummy

E. coli

5

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin gummy

S. aureus

3

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin powder

Salmonella spp.

4

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin powder

E. coli

2

1.00

1.00

0.00 (−0.31, 0.31)

Vitamin powder

S. aureus

2

1.00

1.00

0.00 (−0.31, 0.31)

Botanical powder

Salmonella spp.

1

1.00

1.00

0.00 (−0.31, 0.31)

Botanical powder

E. coli

2

1.00

1.00

0.00 (−0.31, 0.31)

Botanical powder

S. aureus

4

1.00

1.00

0.00 (−0.31, 0.31)

MCT oil

Salmonella spp.

6

1.00

1.00

0.00 (−0.31, 0.31)

MCT oil

E. coli

1

1.00

1.00

0.00 (−0.31, 0.31)

MCT oil

S. aureus

1

1.00

1.00

0.00 (−0.31, 0.31)

Chocolate bar

Salmonella spp.

1

1.00

1.00

0.00 (−0.31, 0.31)

Chocolate bar

E. coli

3

1.00

1.00

0.00 (−0.31, 0.31)

Chocolate bar

S. aureus

3

1.00

1.00

0.00 (−0.31, 0.31)

Low inoculation level, n = 8 per row. Results were identical on the Bio-Rad CFX Opus and the Mic. At 0 CFU every POD was 0.00; at the intermediate and high levels every POD was 1.00. Source: Tables 6 to 11.

At the low level, compendial plating recovered 1 to 8 CFU per test portion, and the qPCR method detected every sample for all three targets in every matrix. No matrix-related loss of detection was seen at the levels tested.

Repeatability

Qualitative results agreed across replicate samples in every matrix and inoculation level. As supporting data, every average RSD of log-transformed internal control Cq values was at or below 2.80%, and no individual sample exceeded 3.52%.

RSD was measured on the internal amplification control (HEX), which is present in every valid sample. Each DNA extract was run on both the Bio-Rad CFX Opus and the Mic, and RSD was calculated for each sample from its two readings.

Matrix

Highest average RSD

Highest individual RSD, inoculated samples

Overall average RSD (n = 22)

Energy shot

2.27%

3.52%

1.85%

Vitamin gummy

0.83%

0.90%

0.42%

Vitamin powder

2.80%

3.52%

2.58%

Botanical powder

0.76%

1.46%

0.54%

MCT oil

2.55%

3.02%

1.32%

Chocolate bar

2.46%

2.68%

1.92%

Source: Table 12. The report gives averages only for uninoculated samples. The Mic generally read HEX earlier than the Bio-Rad, typically within 3 cycles; the offset did not change with inoculation level and did not affect any detect/non-detect result.

Ruggedness

The qPCR method returned identical detect/non-detect results on both instruments for every sample and all three targets, in every matrix: 396 paired determinations (132 samples × 3 targets), 100% concordant. Ruggedness under USP <1223> measures whether qualitative results hold when operating conditions change.

In this study the varied condition was the qPCR instrument. All extractions were run on the Myra automated liquid handler. Every sample detected on one instrument was detected on the other, every uninoculated sample was negative on both, and results on both instruments matched the compendial method.

Robustness

The internal amplification control amplified in every sample across all six matrices, both instruments, and all inoculation levels. No systematic PCR inhibition was observed. Deliberate changes to method parameters were not part of this study; robustness here means tolerance to variation between matrices.

Matrix

Challenge it presents

Energy shot

Aqueous liquid with organic acids, vitamins, caffeine

Vitamin gummy

Semi-solid gelatin with sugars, colorants, vitamins

Vitamin powder

Fine powder with mineral and vitamin components

Botanical powder

Plant matrix with polyphenols, tannins, and other potential PCR inhibitors

MCT oil

Hydrophobic lipid

Chocolate bar

High-fat confection with cocoa polyphenols, methylxanthines, and sugars

Sample prep followed the PathoSEEK® User Guide, including pre-qPCR lysate dilutions for botanical powder (1:10), energy shot (1:10), and chocolate (1:100), and a 1:5,000 internal control dilution for chocolate.

Note on heat-treated botanicals

The first botanical powder lot tested had been autoclaved to lower its native bacterial load. It gave invalid results in all three pathogen channels, most likely because autoclaving leaves a large amount of cell-free DNA that competes in the reaction. Re-extracting those samples with the optional Grim Reefer free DNA removal step from the User Guide gave valid results with no inhibition.

The botanical powder data on this page come from a new, non-autoclaved lot. For heat-treated, autoclaved, or irradiated material, the report recommends adding Grim Reefer or an equivalent DNA cleanup before lysis, and verifying low-level detection before routine use.

Equivalency

Non-inferiority to compendial plating was demonstrated at all three inoculation levels tested using USP <1223> Approach 1 (paired samples, one-sided Z-test, δ = 0.20, α = 0.05).

All three organisms were co-inoculated into every test portion, so each portion gives one paired outcome; a portion counts as discordant if the two methods disagree on any organism. No discordant portions were found. The Bio-Rad and Mic results are repeated measurements of the same extract, so the analysis uses independent test portions: 48 at the low and intermediate levels (8 per matrix across six matrices) and 12 at the high level.

Level

Test portions

Positive rate, qPCR

Positive rate, compendial

Z-statistic

Critical value

Result

Low

48

1.000

1.000

∞ (no discordant pairs)

1.645

Pass

Intermediate

48

1.000

1.000

∞ (no discordant pairs)

1.645

Pass

High

12

1.000

1.000

∞ (no discordant pairs)

1.645

Pass

Source: Tables 13–16

 

USP <1223> recommends testing non-inferiority at a level where 50 to 75% of samples are positive, with at least 75 samples per procedure. This study used 48 test portions per level at levels where both methods detected everything, so non-inferiority holds at the levels tested, and relative sensitivity near the detection threshold was not characterized.

Metric

Low level

Intermediate level

High level

Overall

Sensitivity, qPCR

100%

100%

100%

100%

Sensitivity, compendial

100%

100%

100%

100%

Relative trueness

100%

100%

100%

100%

False positive rate

0%

0%

0%

0%

Measurement-level figures, counting each test portion and target once per instrument. Source: Tables 17 and 18.

Conclusion

Non-inferiority of the PathoSEEK® Salmonella, E. coli, and S. aureus Multiplex Detection Assay with MaGiC Lysis to USP compendial plating was demonstrated per USP <1223> Approach 1 at the inoculation levels tested, for qualitative detection of Salmonella spp., S. aureus, and E. coli in energy shot, vitamin gummy, vitamin powder, botanical powder, MCT oil, and chocolate. Per USP <1223>, laboratories adopting the method should verify method suitability for each product they test.

Download the full report (PDF), USP <1223> Alternative Method Equivalency Study for Detection of Salmonella spp., Staphylococcus aureus, and Escherichia coli in Dietary Supplements

Related Resources:

Frequently asked questions

Is PathoSEEK® equivalent to USP / plating for Salmonella, E. coli, and S. aureus in dietary supplements?

In this study, the assay demonstrated non-inferiority to USP <2022> compendial plating (harmonized with USP <62>) at the low, intermediate, and high levels tested across six supplement matrices, evaluated against USP <1223> Approach 1. The two methods agreed on every test portion for all three organisms. Labs adopting the method should verify suitability for their own products.

Energy shot, vitamin gummy, vitamin powder, botanical powder, MCT oil, and chocolate bar. They span aqueous liquids, gelatin semi-solids, powders, plant material, oils, and a high-fat confection.

Using certified reference materials without a matrix, LOD₅₀ ranged from 0.18 to 0.40 nominal CFU and LOD₉₅ from 0.78 to 1.75, depending on target and instrument. These are nominal CFU from the CRM certificates, not a per-cell detection limit. In the matrices, every low-level sample was detected, with compendial plating recovering 1 to 8 CFU per test portion.

Yes. Every positive sample carried Salmonella, E. coli, and S. aureus together, and each channel detected its target at the low level with no loss of sensitivity.

Yes. Each extract was run on both the Bio-Rad CFX Opus and the BioMolecular Systems Mic, and detect/non-detect results matched on every sample. All extractions used the Myra automated liquid handler.

The study found that autoclaved botanical powder can carry enough cell-free DNA to interfere with the reaction, and adding the Grim Reefer free DNA removal step resolved it. The report recommends that step, or an equivalent cleanup, for heat-treated or irradiated material, along with in-house verification at low inoculation levels.

If you have questions, please complete our “Contact Us” form and a Medicinal Genomics representative will contact you with details about how to get started with PathoSEEK® for Dietary Supplements.

MaGiC™ Lysis

Simplified DNA Extraction Process

SenSATIVAx®

Purify Plant and Microbial DNA

Grim Reefer®

Eliminate Dead DNA

PathoSEEK®

qPCR Detection Assays

PathoSEEK® ID

Identify Species in Failed Samples

FemINDICAtor®

Plant Sex Detection Assay

Mic & Myra

Automated Liquid Handling & qPCR

LIDAR

Breakthrough Multiplex Technology

Sequencing

Valuable Insights and Actionable Data

Fill form to watch Video