In 2007 a CLIA licensed diagnostic testing laboratory (Subject Laboratory) began offering a stool-screening test that uses a proprietary DNA method reported as “DNA-ELISA” (not Real-time Detection PCR) in effort to identify stool microbiota. Claims have been made that their DNA assessment is specific and accurate, and is more sensitive than traditional laboratory methods (e.g. culture).
To test the accuracy and specificity of the proprietary DNA method, an external proficiency analysis study was conducted by a highly respected independent Life Sciences research institution (IIT Research Institute, Chicago, IL). The results of the study were presented at the 112th general Meeting of the American Society of Microbiology (gingras Ba, Duncan SB, Schuyeller NJ, Schreckenberger PC. Assessment of the Diagnostic accuracy of Recently Introduced DNA Stool Screening Test. abstr. 112th gen. Mtg. am. Soc. Microbiol., San Francisco, Ca June 18, 2012).
Experimental Design and Methods
A human stool pool served as the consistent control matrix for all samples. The stool pool was tested extensively, using conventional methodologies, on two separate days and found to be free of entero pathogenic bacteria, yeast and parasites. Bacterial pathogens, authentic banked cell lines of Shigella sonnei, Salmonella typhi, Escherichia coli 0157:H7, Campylobacter jejuni, Vibrio parahemolyticus, Aeromonas caviae, Plesiomonas shigelloides, Edwardsiella tarda, Yersinia enterocolitica, and Clostridium difficile, were grown in culture and added in known concentrations to collection vials provided by the Subject Laboratory. Following the collection kit instructions, one gram of stool was added to each of three vials provided. The three vials contained either C&S Medium, 10% Formalin Fixative or a Nucleic acid Collection Solution. each vial was subsequently spiked with the bacterial preparations at final levels that are clinically significant, and at levels grossly higher in order to get into the Subject Laboratory’s stated cutoff for clinically significant pathogenic bacteria ( 1 x 10(3) CFU/g). Their stated cutoff value for clinical significance is highly questionable because a patient would likely be very symptomatic if an organism such as Campylobacter jejuni was detected by culture at < 1 x 10(2) (100) CFU/g. perhaps the Subject Laboratory is suggesting that their lowest level of detection is 1.0 x 10(3) (1,000) CFU/g for pathogenic bacteria.
Final concentrations of spiked bacteria were determined in quintuplicate by culture. Thirty-four samples were shipped tothe Subject Laboratory via overnight courier on the day of preparation, and the samples were also sent to reference laboratories for evaluation of entero pathogenic bacteria and parasites.
Bacteria — Thirty-one samples each containing one enteric pathogen (at different concentrations) and 3 unspiked samples (controls) were tested. all 31 samples spiked with enteric pathogens were reported negative for pathogenic bacteria- 100% false negative. All 10 intestinal pathogens were recovered by reference laboratories using high complexity clinical microbiology techniques. One each of the 34 stool samples were reported to contain the commensal bacteria Bacillus spp. (10(7)), Staphylococcus aureus (10(8)) or Klebsiella pneumonia (10(7))- when all samples were from same stool pool.
Complete results are presented here.
Parasites — Fifteen of 34 samples from the same stool pool (44%) were reported to contain No Ova or parasites, and 17 of
the 34 samples (50%) were reported as “Parasite Present, taxonomy unavailable.”
Three of 34 specimens were reported positive for Cryptosporidium sp. One sample that was spiked with P. shigelloides bacteria was reported negative for pathogenic bacteria but positive for Enterobius vermicularis (pin worm). All 34 samples contained the very same stool pool, and no parasites were detected by the reference laboratories using conventional O & P techniques and immunoassays for Cryptosporidium and Giardia lamblia.
Yeast/ Fungi — Two of the 34 samples (5.9%) from the same stool pool were reported to contain yeast, one with Candida spp. (2+, ? 10(6) pg DNA/g) and one with Geotricum (4+, 10(3) pg DNA/g). Yeast were not detected by microbiological methods nor were any of the samples spiked with yeast.
The results of the proficiency study indicate that the proprietary DNA methodology employed by the Subject Laboratory was not able to detect pathogenic bacteria at levels that are known to be clinically significant, and grossly higher.
An unanticipated finding was that the proprietary DNA methodology yielded random and non-specific results for parasites, and inconsistent results for yeast as well.
Claims made by the Subject Laboratory that their DNA assessment of stool specimens is specific and accurate could not be supported.