vvEPA
United States
Environmental Protection
Agency
Detection and Quantification Limits
of EPA Enterococcus qPCR Methods
Office of Science and Technology (4303T)
Office of Water
Washington, DC 20460
EPA820-R-13-013
December 2013
-------
Disclaimer
Neither the United States Government nor any of its employees, contractors, or their employees
make any warranty, expressed or implied, or assumes any legal liability or responsibility for any
third party's use of apparatus, product, or process discussed in this document, or represents that
its use by such party would not infringe on privately owned rights. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
Questions concerning this method or its application should be addressed to:
Robin K. Oshiro
Engineering and Analysis Division (4303T)
U.S. EPA Office of Water, Office of Science and Technology
1200 Pennsylvania Avenue, NW
Washington, DC 20460
oshiro.robin@epa.gov or OSTCWAMethods@epa.gov
-------
Detection and Quantification Limits of EPA Enterococcus qPCR Methods
Introduction and Purpose of Document
The Environmental Protection Agency has recommended a quantitative polymerase chain
reaction (qPCR) method targeting Enterococcus spp. as an option for monitoring recreational
beach water quality and has published recreational water quality criteria (RWQC) geometric
mean and statistical threshold values (STV) for the method in its RWQC document (EPA,
2012a). The first objective of this report is to address the question of whether the analytical
sensitivity of the qPCR method is sufficient to support the site-specific alternate criteria values
provided in the RWQC. Analytical sensitivity is often defined in terms of limit of detection and
limit of quantification, i.e. lowest concentration of analyte that can be detected in a given
percentage of analyses (e.g. 95%) and quantified at a given level of precision (e.g. coefficient of
variation of 10%), respectively. Several previous studies have provided either indirect or rough
estimations of the limit of detection (LOD) and/or limit of quantification (LOQ) of the qPCR
method. Here we describe results from an extensive study that was designed to estimate the
values of these parameters from extracts of pure culture E. faecalis cells in a phosphate buffered
saline (PBS) reference matrix. The LOQ estimates are compared with the qPCR method values
provided in the RWQC document.
Analytical Procedures for LOQ Estimation
E. faecalis strain NCTC 12697 (equivalent to ATCC™ 29212) cells originating from Multishot
550 BioBall® cell preparations were suspended in phosphate buffered saline (PBS) buffer to a
concentration of 12 CPU (cells)/ml. This cell suspension was 2-fold serially diluted in PBS to
give predicted cell concentrations of 12, 6, 3, 1.5 and 0.75 /ml. Nine 50 ml subsamples of each of
these cell suspension dilutions, containing -600, 300, 150, 75 and 37.5 cells per subsample, were
filtered onto polycarbonate filters and the filters were extracted as described in EPA method
1611 (EPA, 2012b). As per EPA Method 1611, 5 |il aliquots of 5-fold dilutions of each of these
extracts were analyzed using the simplex qPCR assay described in this method as well as by a
multiplex qPCR assay for Enterococcus with an internal amplification control (EPA Method
1609). All analyses were performed in quadruplicate on an Applied Biosystems StepOnePlus
real-time PCR instrument.
Analysis of Results
Frequencies of analyses showing detection (positive Ct values) at each cell quantity were
determined from combined simplex and multiplex analysis results (Table 1). Ct values generated
from samples containing 37.5 cells were excluded from subsequent analyses due to significant
frequencies of non-detects. Ct values generated from the other samples were statistically
evaluated for distributional normality using procedures in the fitdistrplus function in R
(Maindonald and Braun, 2007). Distributions both within and across dilution levels were deemed
approximately normal according to the Kolmogorov-Smirnov statistic. Data generated via the
-------
combined simplex and multiplex assays were evaluated for their suitability for data pooling prior
to LOQ determination. Significant differences in mean Ct values among assays were evaluated
using the parametric 1-way ANOVA procedure (aov function in R), considering potential
sources of variation across dilutions (plate, dilution level, filter aliquot & sample replicate) and
dilution-specific sources of variation (plate, filter aliquot & replicate). In addition, analysis of
covariance of Ct on loglO cells/filter indicated that there was no significant difference between
the simplex and multiplex assays with respect to either slope (P=0.1562) or intercept
(P=0.1227). Differences in Ct variances among assays were evaluated using the robust
parametric Levene's Test procedure from the R package lawstat function. There were no
significant differences among assays (p-values > 0.05). The final dataset was created by
performing Ordinary Least Squares linear regression across all observed Ct values for loglO
cells/filter for all dilutions. For this data set, observations with Studentized residuals in excess of
2 (PROC REG procedure in SAS) were eliminated because they were considered to be overly
influential. The R routine "chemcal" by Johannes Ranke, was used to estimate the LOQ.l This
routine is based on estimating the point in the linear model at which the precision for loglO cells
/filter, expressed by the half-length of the 99% confidence interval, is some fraction of the
estimated loglO cells/filter from the standard curve, mainly 1/3, 1/5, or 1/10. These fractions are
equivalent to the coefficent of variation (i.e. 33.3%, 20%, 10%, respectively). For example, the
coefficent of variation (CV) of a loglO cells/filter value picked from the standard curve based on
observed CT would be +/- 20% at the LOQ20.
Results
Table 1 shows the frequencies of detection of different calibrator cell quantities/filter sample
determined in this study. Since target sequences are the analytes that are measured by the qPCR
technique, these results are specific to the mean target sequence/cell recovery ratio of 22.73 that
was determined for the lot of cells used in this study (Sivaganesan et. al., 2011).
LOQ estimates for the calibrator cells used in this study are shown in Table 2. Also shown are
the corresponding LOQ estimates of qPCR target sequences that were determined by multiplying
the calibrator cell LOQ estimates by the previously determined mean target sequences/cell
recovery ratio of 22.73 for this lot of cells (Sivaganesan et a/., 2011). Table 2 also shows LOQ
estimates for the calibrator cells used in EPA's National Epidemiological and Environmental
Assessment of Recreational (NEEAR) Water study that provided the basis for the RWQC values
(EPA, 2012a). These LOQ estimates were determined by dividing the qPCR target sequence
LOQ estimates from this study by the mean qPCR target sequences/cell recovery ratio estimate
of 15 for the calibrator cells used in the NEEAR Water Study (EPA, 2013).
Johannes Ranke. "chemCal: Calibration functions for analytical chemistry." 2013-06-14.
http://cran.r-project.org/web/packages/chemCal/index.html
4
-------
Table 1. Frequencies of detection of different calibrator cell quantities/filter sample
by Enterococcus qPCR method
600 Cells3
100%
300 Cells3
100%
150 Cells3
100%
75 Cells3
97%
37.5 Cells3
80%
3 Based on mean estimated recovery of 22.73 qPCR target
sequences/cell for the cell lot used in this study (Sivaganesan
et al., 2011)
Table 2. Limit of quantification (LOQ) estimates for Enterococcus qPCR method at 0.01
significance (alpha) level and CV values of 10, 20 and 33.3%.
Calibrator cells/filter from
this study
QPCR target sequences/filter
from this study 3
EPARWQC-adjusted
calibrator cells/filter15
LOQ
@ CV=10%
179
4069
271
LOQ
@ CV=20%
150
3409
227
LOQ
@ CV 33.3%
125
2841
189
3 Based on mean estimated recovery of 22.73 qPCR target sequences/cell for the cell
lot used in this study (Sivaganesan et al., 2011)
b Based on mean estimated recovery of 15 qPCR target sequences/cell from the EPA
NEEAR Water Study (EPA, 2013).
Discussion
In the 2012 RWQC document, EPA provides statistical threshold value (STV) values for the
qPCR method based on the estimated 90th percentile of the enterococci water quality
distributions from EPA's NEEAR study. It is also suggested that states use a beach action value
(BAV) as a conservative, precautionary tool for making beach notification decisions. The BAV
is not a component of EPA's recommended criteria, but a tool that states may choose to use,
without adopting it into their WQS as a "do not exceed" value for beach notification purposes
(such as advisories). The BAV was developed from the same water quality distribution as the
STV and corresponds to the estimated 75* percentile of the enterococci water quality
distributions. The RWQC further offers states the option to use STV and BAV values based on
two levels of gastrointestinal illness (NGI) rates: 36 estimated NGI illnesses/1000 primary
contact recreators or 32/1000. A summary of these criteria values is provided in Table 3.
-------
Table 3. Enterococcus qPCR criteria values from RWQC document
Criteria Type
STV (CCE per 100 mL)
BAV(CCE per 100 mL)
Estimated Illness
Rate (NGI):
36 per 1,000 primary
contact recreators
2,000
1,000
OR
Estimated Illness
Rate (NGI):
32 per 1,000 primary
contact recreators
1,280
640
Results from the current study indicate that the analytical sensitivity of EPA Method 1611 in
analyses of pure culture cells in a reference matrix (PBS) is sufficiently low in terms of calibrator
cell LOQ estimates to support each of these alternative values. It is noted that the 33% CV LOQ
estimate for target sequences obtained in this study is highly similar to the 99% probability LOD
estimate determined on the basis of prediction from the Poisson distribution and corroborated by
analysis results using Method 1609 reagents in a more recent study (Sivaganesan et a/.,
manuscript in review). Additional EPA studies (unpublished) indicate that in analyses of marine
and fresh surface water sample extracts that meet the method's control assay acceptance criteria
for demonstrating absence of sample matrix interference, the method shows similar analytical
sensitivity in terms of LOD and LOQ to the results obtained from the pure culture samples
analyzed in this study. While the RWQC are based on 100 ml water sample volumes, some
flexibility is provided in Method 1611 and forthcoming Method 1609 for the analysis of smaller
volumes of difficult to filter water samples. The LOQ estimates reported in this study pertain to
CCE per filter and hence will pertain to any water sample volume that is filtered. However,
analyses of 5-fold diluted extracts from smaller water sample volumes could lead to false
negatives or method results that have a higher degree of variability (e.g. CV > 33% from Table
2) from samples that are above some of the RWQC values listed in Table 3 in certain instances,
such as when low DNA recovery is encountered from the filter extracts. The LOQ of analyses of
undiluted extracts (recommended in Method 1609 only) should not be an issue with the smaller
water sample volumes specified in the EPA methods in terms of exceeding these RWQC values.
It also should be emphasized that these LOD and LOQ estimates are based upon the procedures
specified in EPA Methods 1611 and 1609 (see "Analytical Procedures" section above) and may
differ if other variations of these procedures are used, e.g. if different extract volumes or
dilutions are analyzed.
Acknowledgements
This document was developed by Rich Haugland, Manju Varma, and Larry Wymer of EPA's
Office of Research and Development (ORD) and Robin K. Oshiro of the Office of Water (OW).
Mano Sivaganesan and Kevin Oshima of ORD, and Jan Matuszko and Shamima Akhter of OW
are also gratefully acknowledged.
-------
References
Maindonald, J., Braun, J. 2007. Data Analysis and Graphics Using R. Cambridge University
Press, Cambridge, 2nd edition, 2007.
Sivaganesan, M., Siefring, S., Varma, M., Haugland, R.A. 2011. MPN estimation of qPCR target
sequence recoveries from whole cell calibrator samples. J. Microbiol. Meth. 87:343-349.
U.S. EPA 2012a. Recreational water quality criteria. U.S. Environmental Protection Agency,
Office of Water: Washington, DC. EPA-820-F-12-058.
U.S. EPA 2012b. Method 1611: Enterococci in Water by TaqMan® Quantitative Polymerase
Chain Reaction (qPCR) Assay. U.S. Environmental Protection Agency, Office of Water:
Washington, DC. EPA-821-R-12-008.
U.S. EPA 2013. Method 1609: Enterococci in Water by TaqMan Quantitative Polymerase
Chain Reaction (qPCR) with Internal Amplification Control (IAC) Assay. U.S. Environmental
Protection Agency, Office of Water: Washington, DC. EPA-821-R-13-005.
------- |