Limits...
Specificity and Strain-Typing Capabilities of Nanorod Array-Surface Enhanced Raman Spectroscopy for Mycoplasma pneumoniae Detection.

Henderson KC, Benitez AJ, Ratliff AE, Crabb DM, Sheppard ES, Winchell JM, Dluhy RA, Waites KB, Atkinson TP, Krause DC - PLoS ONE (2015)

Bottom Line: At present the most effective means for detection and strain-typing is quantitative polymerase chain reaction (qPCR), which can exhibit excellent sensitivity and specificity but requires separate tests for detection and genotyping, lacks standardization between available tests and between labs, and has limited practicality for widespread, point-of-care use.Here we demonstrate using partial least squares- discriminatory analysis (PLS-DA) of sample spectra that NA-SERS correctly identified M. pneumoniae clinical isolates from globally diverse origins and distinguished these from a panel of 12 other human commensal and pathogenic mycoplasma species with 100% cross-validated statistical accuracy.Furthermore, PLS-DA correctly classified by strain type all 30 clinical isolates with 96% cross-validated accuracy for type 1 strains, 98% cross-validated accuracy for type 2 strains, and 90% cross-validated accuracy for type 2V strains.

View Article: PubMed Central - PubMed

Affiliation: Department of Microbiology, University of Georgia, Athens, GA, United States of America.

ABSTRACT
Mycoplasma pneumoniae is a cell wall-less bacterial pathogen of the human respiratory tract that accounts for > 20% of all community-acquired pneumonia (CAP). At present the most effective means for detection and strain-typing is quantitative polymerase chain reaction (qPCR), which can exhibit excellent sensitivity and specificity but requires separate tests for detection and genotyping, lacks standardization between available tests and between labs, and has limited practicality for widespread, point-of-care use. We have developed and previously described a silver nanorod array-surface enhanced Raman Spectroscopy (NA-SERS) biosensing platform capable of detecting M. pneumoniae with statistically significant specificity and sensitivity in simulated and true clinical throat swab samples, and the ability to distinguish between reference strains of the two main genotypes of M. pneumoniae. Furthermore, we have established a qualitative lower endpoint of detection for NA-SERS of < 1 genome equivalent (cell/μl) and a quantitative multivariate detection limit of 5.3 ± 1 cells/μl. Here we demonstrate using partial least squares- discriminatory analysis (PLS-DA) of sample spectra that NA-SERS correctly identified M. pneumoniae clinical isolates from globally diverse origins and distinguished these from a panel of 12 other human commensal and pathogenic mycoplasma species with 100% cross-validated statistical accuracy. Furthermore, PLS-DA correctly classified by strain type all 30 clinical isolates with 96% cross-validated accuracy for type 1 strains, 98% cross-validated accuracy for type 2 strains, and 90% cross-validated accuracy for type 2V strains.

No MeSH data available.


Related in: MedlinePlus

PLS-DA of 32 M. pneumoniae clinical isolates, including reference strains M129 and FH.Each panel represents a cross-validated class prediction score for (A) class 1, substrate background; (B) class 2, growth medium control; and (C) class 3, all M. pneumoniae strains. For panels A-C, each individual shape represents a single pre-processed NA-SERS spectrum. The substrate background spectra are represented by gray diamonds, the growth medium control spectra by solid black squares, and the M. pneumoniae spectra by open shapes that differ by cluster to indicate the different individual strains and isolates. The red-dotted line indicates the classification threshold line for positive class prediction, and the black-dotted line indicates the 95% confidence interval. Cross-validated sensitivity, specificity, and class error for the panels were as follows: (A) nanorod substrate background: 1.00, 1.00, and 0, respectively; for (B) growth medium control: 1.00, 1.00, and 0, respectively; and for (C)M. pneumoniae: 1.00, 1.00, and 0, respectively. Cross-validated statistics were obtained using Venetian blinds with 10 data splits to represent the prediction performance of the PLS-DA model for M. pneumoniae detection.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4487258&req=5

pone.0131831.g002: PLS-DA of 32 M. pneumoniae clinical isolates, including reference strains M129 and FH.Each panel represents a cross-validated class prediction score for (A) class 1, substrate background; (B) class 2, growth medium control; and (C) class 3, all M. pneumoniae strains. For panels A-C, each individual shape represents a single pre-processed NA-SERS spectrum. The substrate background spectra are represented by gray diamonds, the growth medium control spectra by solid black squares, and the M. pneumoniae spectra by open shapes that differ by cluster to indicate the different individual strains and isolates. The red-dotted line indicates the classification threshold line for positive class prediction, and the black-dotted line indicates the 95% confidence interval. Cross-validated sensitivity, specificity, and class error for the panels were as follows: (A) nanorod substrate background: 1.00, 1.00, and 0, respectively; for (B) growth medium control: 1.00, 1.00, and 0, respectively; and for (C)M. pneumoniae: 1.00, 1.00, and 0, respectively. Cross-validated statistics were obtained using Venetian blinds with 10 data splits to represent the prediction performance of the PLS-DA model for M. pneumoniae detection.

Mentions: PLS-DA was applied here to determine statistically significant detection of M. pneumoniae by NA-SERS. PLS-DA is a full-spectrum, multivariate, supervised statistical method whereby prior knowledge of classes is used to yield more robust discrimination by minimizing variation within classes while emphasizing latent variables arising from spectral differences between classes [41, 42]. A PLS-DA model was generated to discriminate between three classes: the nanorod substrate background (Fig 2A); the growth medium control (Fig 2B); and all M. pneumoniae strains (Fig 2C). The inclusion of substrate background and growth medium controls allowed us to ensure that any differences in growth medium and nanorod background signal within the substrate did not affect the ability of the model to discriminate between the presence or absence of M. pneumoniae. Two nanorod substrates were used for these experiments, with each containing duplicate wells of the bare nanorod substrate, independently prepared M129, FH, and growth medium controls, and 15 additional clinical isolates of M. pneumoniae. A total of n = 390 pre-processed NA-SERS spectra collected from both substrates were included in the model, consisting of 20 nanorod substrate background spectra, 20 growth medium control spectra, 25 M129 spectra, 25 FH spectra, and 10 spectra per additional clinical isolate. The cross-validated statistics for the model show that NA-SERS correctly classified all 32 clinical isolates as M. pneumoniae regardless of global origin, year isolated, genotype, or macrolide susceptibility phenotype, and distinguished them from the substrate background and the growth medium control with 100% cross-validated sensitivity and specificity.


Specificity and Strain-Typing Capabilities of Nanorod Array-Surface Enhanced Raman Spectroscopy for Mycoplasma pneumoniae Detection.

Henderson KC, Benitez AJ, Ratliff AE, Crabb DM, Sheppard ES, Winchell JM, Dluhy RA, Waites KB, Atkinson TP, Krause DC - PLoS ONE (2015)

PLS-DA of 32 M. pneumoniae clinical isolates, including reference strains M129 and FH.Each panel represents a cross-validated class prediction score for (A) class 1, substrate background; (B) class 2, growth medium control; and (C) class 3, all M. pneumoniae strains. For panels A-C, each individual shape represents a single pre-processed NA-SERS spectrum. The substrate background spectra are represented by gray diamonds, the growth medium control spectra by solid black squares, and the M. pneumoniae spectra by open shapes that differ by cluster to indicate the different individual strains and isolates. The red-dotted line indicates the classification threshold line for positive class prediction, and the black-dotted line indicates the 95% confidence interval. Cross-validated sensitivity, specificity, and class error for the panels were as follows: (A) nanorod substrate background: 1.00, 1.00, and 0, respectively; for (B) growth medium control: 1.00, 1.00, and 0, respectively; and for (C)M. pneumoniae: 1.00, 1.00, and 0, respectively. Cross-validated statistics were obtained using Venetian blinds with 10 data splits to represent the prediction performance of the PLS-DA model for M. pneumoniae detection.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4487258&req=5

pone.0131831.g002: PLS-DA of 32 M. pneumoniae clinical isolates, including reference strains M129 and FH.Each panel represents a cross-validated class prediction score for (A) class 1, substrate background; (B) class 2, growth medium control; and (C) class 3, all M. pneumoniae strains. For panels A-C, each individual shape represents a single pre-processed NA-SERS spectrum. The substrate background spectra are represented by gray diamonds, the growth medium control spectra by solid black squares, and the M. pneumoniae spectra by open shapes that differ by cluster to indicate the different individual strains and isolates. The red-dotted line indicates the classification threshold line for positive class prediction, and the black-dotted line indicates the 95% confidence interval. Cross-validated sensitivity, specificity, and class error for the panels were as follows: (A) nanorod substrate background: 1.00, 1.00, and 0, respectively; for (B) growth medium control: 1.00, 1.00, and 0, respectively; and for (C)M. pneumoniae: 1.00, 1.00, and 0, respectively. Cross-validated statistics were obtained using Venetian blinds with 10 data splits to represent the prediction performance of the PLS-DA model for M. pneumoniae detection.
Mentions: PLS-DA was applied here to determine statistically significant detection of M. pneumoniae by NA-SERS. PLS-DA is a full-spectrum, multivariate, supervised statistical method whereby prior knowledge of classes is used to yield more robust discrimination by minimizing variation within classes while emphasizing latent variables arising from spectral differences between classes [41, 42]. A PLS-DA model was generated to discriminate between three classes: the nanorod substrate background (Fig 2A); the growth medium control (Fig 2B); and all M. pneumoniae strains (Fig 2C). The inclusion of substrate background and growth medium controls allowed us to ensure that any differences in growth medium and nanorod background signal within the substrate did not affect the ability of the model to discriminate between the presence or absence of M. pneumoniae. Two nanorod substrates were used for these experiments, with each containing duplicate wells of the bare nanorod substrate, independently prepared M129, FH, and growth medium controls, and 15 additional clinical isolates of M. pneumoniae. A total of n = 390 pre-processed NA-SERS spectra collected from both substrates were included in the model, consisting of 20 nanorod substrate background spectra, 20 growth medium control spectra, 25 M129 spectra, 25 FH spectra, and 10 spectra per additional clinical isolate. The cross-validated statistics for the model show that NA-SERS correctly classified all 32 clinical isolates as M. pneumoniae regardless of global origin, year isolated, genotype, or macrolide susceptibility phenotype, and distinguished them from the substrate background and the growth medium control with 100% cross-validated sensitivity and specificity.

Bottom Line: At present the most effective means for detection and strain-typing is quantitative polymerase chain reaction (qPCR), which can exhibit excellent sensitivity and specificity but requires separate tests for detection and genotyping, lacks standardization between available tests and between labs, and has limited practicality for widespread, point-of-care use.Here we demonstrate using partial least squares- discriminatory analysis (PLS-DA) of sample spectra that NA-SERS correctly identified M. pneumoniae clinical isolates from globally diverse origins and distinguished these from a panel of 12 other human commensal and pathogenic mycoplasma species with 100% cross-validated statistical accuracy.Furthermore, PLS-DA correctly classified by strain type all 30 clinical isolates with 96% cross-validated accuracy for type 1 strains, 98% cross-validated accuracy for type 2 strains, and 90% cross-validated accuracy for type 2V strains.

View Article: PubMed Central - PubMed

Affiliation: Department of Microbiology, University of Georgia, Athens, GA, United States of America.

ABSTRACT
Mycoplasma pneumoniae is a cell wall-less bacterial pathogen of the human respiratory tract that accounts for > 20% of all community-acquired pneumonia (CAP). At present the most effective means for detection and strain-typing is quantitative polymerase chain reaction (qPCR), which can exhibit excellent sensitivity and specificity but requires separate tests for detection and genotyping, lacks standardization between available tests and between labs, and has limited practicality for widespread, point-of-care use. We have developed and previously described a silver nanorod array-surface enhanced Raman Spectroscopy (NA-SERS) biosensing platform capable of detecting M. pneumoniae with statistically significant specificity and sensitivity in simulated and true clinical throat swab samples, and the ability to distinguish between reference strains of the two main genotypes of M. pneumoniae. Furthermore, we have established a qualitative lower endpoint of detection for NA-SERS of < 1 genome equivalent (cell/μl) and a quantitative multivariate detection limit of 5.3 ± 1 cells/μl. Here we demonstrate using partial least squares- discriminatory analysis (PLS-DA) of sample spectra that NA-SERS correctly identified M. pneumoniae clinical isolates from globally diverse origins and distinguished these from a panel of 12 other human commensal and pathogenic mycoplasma species with 100% cross-validated statistical accuracy. Furthermore, PLS-DA correctly classified by strain type all 30 clinical isolates with 96% cross-validated accuracy for type 1 strains, 98% cross-validated accuracy for type 2 strains, and 90% cross-validated accuracy for type 2V strains.

No MeSH data available.


Related in: MedlinePlus