Tip:
Highlight text to annotate it
X
[ Music ]
>> My name is Amy Michaud,
and I'll be announcing the speakers this afternoon.
We will do all three presentations
and then we'll have a question and answer period afterwards.
So just hold your questions, write them down and keep those
in mind so we can ask some good questions at the end.
They did ask me to remind presenters to make sure you sign
in on the sign-in sheet this afternoon
if you haven't done that today.
And we'll be breaking at about 4:15 directly out back here.
So anyway, I'd like to announce our first speaker.
It's Jeffrey Dake from the US Army Crime Lab in Georgia.
And he'll be presenting on -- I know it's a long name;
so I'm going to have to read it.
He's going to do "Method Validation For the Analysis
of *** and *** Lubricants Using Direct Analysis
Mass Spectrometry".
[ Pause ]
>> Good afternoon.
I want to apologize.
The name sounded great until I tried to fit it
on to a PowerPoint slide.
So it's a little longer than I should have made it.
And despite the sophomoric implications,
unfortunately my presentation this afternoon is going
to be fairly dry.
But with all of the discussion about method validation
and instrument reproducibility,
I think you'll find the presentation itself
to be quite enjoyable.
This is our official disclaimer.
The Army and the Department
of Defense disavows any knowledge of my statements.
It is not to be construed as official military policy
or a recommendation by the military for any product
or chemical listed herein.
So what was the purpose of this validation?
Essentially, we were looking for a technique
that would involve rapid and easy discrimination
of two common *** lubricants.
Polydimethyl siloxane or PDMS, which is commonly employed
as a lubricant and it's present on a majority
of condoms produced, at least in the United States,
to my knowledge, and Nonoxynol-9,
which is very specific for condoms and *** lubricants
because it's often used as a spermicide.
So the instrumentation that we validated
for this method was direct analysis mass spectrometry
with time-of-flight mass spec. You can see instrument here,
particularly in the picture on your right,
that's the actual sample inlet port.
It's open to atmosphere.
It allows us to hold samples directly within the source
and perform real-time analysis of samples.
It can preclude steps like extraction, screening, cutting,
by directly stand pointing
and getting mass spectral information from your samples.
It involves disposable media which, of course,
is great for quality control.
And it gives us rapid analysis over a large mass range,
typically, somewhere between 100 and 1200 or 1500 AMU.
So the compounds of interest that we were analyzing
to get accurate mass information from this instrument,
we used internal tuning compound.
In this instance, we used polyethylene glycol
with an average mass of 600 AMU.
What this does is it gives us repeatable peaks
over the entire mass range so we can tune the instrument.
Our instrumental tolerance is plus
or minus 5 milli mass units.
Because the instrument doesn't involve a separation technique,
like GC or LC, to increase specificity,
we are using accurate mass as opposed to nominal mass.
We also use an internal QAQC.
Once we've tuned the instrument, we want to make sure
that our tune is accurate; so we used reserphine,
which is a turkey tranquilizer.
And it's got a mass of 609.281 in the protonated molecular ion.
So you can see on the bottom is polydimethyl siloxane.
It's got a repeating polymer unit of OSi-CH3, CH3.
And it has a mass range of about 200 to 600 on the Dart.
The nonoxynol polymer, because it actually is produced
as a polymer with the nine chain unit being a nonoxynol-9,
the spermicide.
It has a repeating O CH2-CH2 group.
And it covers a range of about 350 to 900 AMU.
So there is overlap between the two.
So already we can see that there may be a possibility
that discrimination in combination may not be possible.
Specifically, I want you to note the nonoxynol-9,
the actual spermicide compound of interest has a mass
of 617426, and it's a protonated molecular molecule.
So a little background.
Currently, in the laboratory, we are screening both
of these materials as they are polymers by FTIR.
This is a IR spectra of polydimethyl siloxane.
We've got five characteristic peaks all below
about 1280 wave numbers.
And polydimethyl siloxane is great for IR.
It's a strong IR absorber,
and it gives us a really good signal even
in small sample amounts.
What you'll see is the nonoxynol-9 has a very
distinguishable spectrum from PDMS.
However, it absorbs in a similar region,
and it doesn't have a very strong absorption in IR.
So when you talk about a *** or a *** lubricant
that is predominantly lubricant for PDMS and maybe 5
to 15 percent spermicide, you use this factor
that the PDMS is a strong absorber
and nonoxynol-9 is a weak absorber, and the fact
that there's much more PDMS than in nonoxynol-9.
It sometimes becomes difficult to detect nonoxynol-9
in the presence of PDMS.
So in our laboratory, what we'll end
up doing is further extraction and separation using C-18 tubes
and then reconstitution in methylene chloride,
which really nobody wants to work with.
So it becomes kind of a long process
to analyze and detect nonoxynol.
Additionally, because it is polymeric,
we have difficulty determining
that it's specifically an nonoxynol-9 present as opposed
to some other nonoxynol compound.
So when I told my boss that I was going
to be validating the Dart, he kind of got the impression
that it was going to look something like this;
kicking back at my desk, throwing darts.
But we designed a process
that would optimize our instrumental parameters
for both our tuning compound and our compounds of interest
over the entire mass range.
Then we wanted to establish our methodology, or our protocol
for analysis, establish limits of detections for our analytes
of interest, compare it to actual case samples
to get an idea of what type of background
or competition we might see and then, ultimately, implement it
in an analysis scheme.
And I guess the last step would be come here and talk about it.
So in the instrument, we used three parameters.
There are many parameters that you can vary,
but the three parameters that has the most effect
on your sample analysis are the orifice voltage,
which is that inlet which you can change the voltage.
A lower voltage will give you less fragmentation.
The system will perform similar to a CI mass spectrometer.
As you increase the voltage and it gets stronger and stronger,
you're going to see more EI spectra, a lot of fragmentation.
And again, since we're not doing any separation,
more fragmentation becomes a lot more difficult to interpret.
Orifice temperature is the most widely variable parameter.
It can go from somewhere around 100 degree C all the way
up to 500 degree C. So there's quite a range.
And changing that temperature is going to effect the types
of molecules that you volatilize and ultimately sample.
And your detector voltage, just like any other instrument,
as you increase the voltage on the detector, you're going
to get an increased signal with increased electronic noise.
Conversely, as you decrease that voltage, you're going
to get a weaker signal with much less noisy spectrum.
So we wanted to evaluate that to determine, again, if this lack
of separation, we want to decrease our noise to the point
to make our spectra much more interpretable.
In all, we ended up with 36 different parameter sets
that we created a matrix out of, and we ended
up testing all of these samples.
So to give you an idea, --
thankfully these colors showed up.
To give you a good idea what our data is looking like,
you can see in blue is our PDMS peaks;
and in orange, our nonoxynol peaks.
While they do overlap.
You can see that there's definite clear separation
between the two.
With PDMS, what we actually ended
up seeing is two different polymeric chains based
on where the PDMS molecule fragments.
The 355 and 371 fragments are actually
complementary fragments.
And the 429-445, et cetera, et cetera,
through different masses in the range.
With nonoxynol, even in a purchase standard
of supposedly an nonoxynol-9,
we get several other polymers present.
We're able to see all the way down to nonoxynol-5
up to nonoxynol-15, but, of course, you know,
there's our compound of interest that nonoxynol-9 at 617.
So yes, there's overlap but, Step One, we were able
to clearly distinguish them at the same time.
So in our parameter set,
essentially what you're looking at, on the Y axis is the number
of peaks we saw for our chemical compound.
So since we want to maximize our signal
over our entire sample range or mass range, the more peaks,
the better; the more pieces of information we can obtain.
And across the bottom is each discrete temperature
that we did.
And then we did four replicants
of our PEG internal tuning compound.
So what we were looking for was a high bar
and equivalent bars for each color.
For 15 volts, we had a problem with one of our sample sets.
Again, I said that 5 milli mass unit range was kind
of our limit, and one of our peaks fell out.
So clearly, those parameters weren't the best.
But as we look at 325 degrees, you can see that, you know,
not only do we get a high number of peaks at 15 peaks for PEG,
but we're seeing reproducibility.
We're getting the same result every time we put the sample in.
At 35 volts, our reproducibility wasn't as great,
but 275 looked to be our best.
So we wanted to take these forward, analyze our analytes
of interest and see which one was more effective for those.
Now I have no charts for PDMS
because PDMS is pretty much insensitive to all
of the parameters we used.
You can set it to almost anything
and the system is going to find PDMS.
With nonoxynol-9, again, the same thing.
We're looking for a large number of peaks.
So that would be our Y axis
and then our X axis is our varying temperatures.
And you can see again, that, 325 we saw a great number of peaks
of nonoxynol-9 which covered a wide range.
275, it wasn't quite as responsive.
We needed a little bit higher temperature to get
that nonoxynol-9 moving, which ultimately led us to decide
on 15 volts and 325 degrees for our parameters.
So moving on to limited detection.
Essentially, what we did was we started with serial dilutions
from 100 nanograms per microliter all the way
down to 1 nanogram per microliter of each anali.
We tested each solution in triplicate
to demonstrate reproducibility,
again using that 5 milli mass unit tolerance.
We had four different examiners perform it
to evaluate inter-examiner variability.
As you can see in this picture,
essentially the sampling techniques is as simple
as holding something in front of the source.
You know, obviously, the concern with that is well,
if you hold it to the left and I hold it to the right,
is one of us missing the spot.
So we wanted to get this data to see
if we could find inter-examiner variability.
We tested all these solutions in FTIR so we'd be able
to compare both of our instruments and get a clear idea
of how these two instruments perform these types of samples.
And we tried just dipping the rods in solution.
As you can see in the background,
there's that glass jar with the rod just hanging out.
We also used a one microliter syringe
to deposit one microliter of our solvents to, again,
increase our reproducibility and our specificity.
This is our typical data for PDMS.
On the top is our hundred nanograms per
microliter solution.
The bottom is 5 nanograms.
And you can see peaks for PDMS in each of them.
It's clear that there are several more peaks for PDMS
in the more concentrated sample.
And when you look at the abundances,
the overall abundance are the maximal abundance
in our top spectra was about 20,000 counts,
and the bottom, we're down to two.
So it's something that you would expect.
It's about 10 times -- a little less
than 10 times as concentrated.
So this is what we're looking for.
And ultimately, we found PDMS pretty efficient
on FTIR and on Dart.
The order of magnitude was essentially the same.
It was a little more sensitive by Dart than FTIR
in these pure samples.
But nonoxynol-9 had a strong preference for Dart.
The order of magnitude were more sensitive which, again,
leads us to a much stronger screening tool,
and it's very interesting because it complements the IR.
As I said earlier, the IR is not very preferential,
or the nonoxynol-9 is not a strong absorber for FTIR,
but it seems to love the Dart.
So for what our casework shadowing did was essentially we
took 86 case samples.
They involved actual case samples with all
of the case related information stripped out.
We also used lubricant standards, lotions,
cosmetics, skin care products.
We also had some people simulate case samples,
and that means exactly what you think it means.
So those daring souls went out and got us realistic samples
so we could evaluate how these instruments would perform.
Particularly, when we introduced things, like body oils,
detergents, and other types of things that will commonly be
in the environment that a case examiner is going
to have to deal with.
The work was performed by two different examiners;
myself and another examiner at the laboratory.
It was performed completely blind.
One examiner would test a sample on FTIR, transfer that sample
into a glass vial with a sample identifier.
No information was provided as to what the sample had in it
or what the results were.
And then the second examiner would test it by Dart.
To avoid any kind of bias, we flipped it halfway through;
and the person who was doing Dart now had
to go do IR and vice versa.
So this is what a typical sample came out to be.
We're looking for the PDMS in this spectrum.
This is a cutting from a pair of underwear,
and we found body oils by GCMS.
We found several compounds that we see in lotions
and cosmetic products.
So we knew that this was a pretty dirty sample.
But when we look at the results, you know,
we clearly can pull out these PDMS peaks.
And it's not just one peak or two peaks,
we're getting several peaks associated with PDMS
with this high mass accuracy.
So what this is telling us is that even though we're dealing
with complex mixtures in the absence
of a separation technique, we're still being very effective
at identifying our compound of interest.
Our competition or competitive ionization
from other molecules isn't hindering our ability
to detect this compound, even in realistic case samples.
And our ultimate results.
We had 80 percent agreement with PDMS.
And what I mean by that, is that approximately 25 percent
of our samples were positive,
both by FTIR and by Dart for PDMS.
So the IR saw it.
The Dart saw it.
About 50 percent of our samples were negative
on both instruments.
The IR didn't see it.
The Dart didn't see it.
When our disagreement came about,
we were actually fairly shocked.
We were expecting it to be one way or the other,
either the IR is going to be a much better technique
and thus we can use the IR to screen for Dart,
or the Dart is going to be a much more effective technique
and we can use the Dart to screen for IR.
But it turned out we had discrepancies going both ways.
We had 10 samples that were positive on Dart,
which means that Dart was able to identify compounds associated
with PDMS, and the IR came back negative.
Conversely, we had seven samples that were positive by FTIR,
and no result by Dart.
And we thought hard about these types of samples.
We compared our data, and we were finding is
that we had a lot of indications of weak samples.
So the most likely resolution for this is that these samples
that we were having contradiction were near the
limited detection of both instrument.
So as you reach the limit of detection, any minute error
such as holding that sample in front of the source,
any slight error is going to increase the possibility
that you're going to miss your analyte.
So it could be that we're dealing
with low limited detection issues
and that would be why we're having disparities
in both directions.
It's also possible that we could be dealing with a sample
that has IR competitive absorption with PDMS.
And it was difficult to identify PDMS in the sample
because of say a body oil or a lotion product,
but the Dart had no problem differentiating those two.
And conversely, you could have competitive ionization
with Dart allowing you --
or disallowing the ability to see PDMS
but they have two different IR absorptions and it's easy
to distinguish by FTIR.
The double bonus is the nonoxynol-9 had 100
percent agreement.
And all of the samples that we found in nonoxynol-9,
it was present by Dart.
It was present by FTIR.
And the majority of samples were negative for nonoxynol-9,
but we had strong agreement in the negatives as well.
So what we ultimately determined from this portion
of our validation is that the AccuTOF-Dart was effective
for the screening of the nonoxynol-9.
It eliminates all of the extraneous extractions
that we were doing before.
It saves us money by not having to buy costly C-18 tubes.
It prevents me from having to smell
like methylene chloride when I get home.
And ultimately, it is improving our efficiency and our accuracy
within the laboratory.
As I said, with FTIR, we're only able
to identify a polymer associated with nonoxynols.
But this exact mass information allows us
to identify an nonoxynol-9 being present.
We did have inconsistencies with PDMS.
As I said, it could be possible interference
from other compounds, issues with limited detection.
And what we're doing next in our next phase is attempting
to address these issues.
So we want to improve our detection abilities with PDMS.
We could do things such as increasing the volume.
What we found was that if we deposited 1 microliter of sample
into the Dart, it gave us say a response of 10,000.
And if we add 2 microliters, we would get a response of 20,000.
And then 3 microliters, then 30,000.
So it wasn't exactly linear but it was close to linear.
So we could always increase our sampling amount, move ourselves
above that limited detection, and evaluate whether
or not it is a limited detection issue that's giving
us discrepancies.
We're also considering using internal standards
to perform semi-quantitative analysis of PDMS.
PDMS is used in lotions, cosmetics, skin care products,
*** lubricants, and condoms.
But they are used in different concentrations
in all these different types of products.
One of the things we're interested
in evaluating is using say a known concentration,
internal standard, to develop information
as to whether the PDMS that we're seeing
in a sample is highly concentrated
and then thus more likely to have come
from a concentrated source of PDMS such as a ***
or a *** lubricant or in low concentrations and then likely
to be present from say a lotion or a cosmetic product.
Another examiner has also proceeded
in evaluating the analysis of lotions
and cosmetic products themselves by Dart,
and currently has characterized 40 different compounds commonly
used by the lotion and cosmetic industry.
This is giving us advantageous results.
We're able to see compounds that we've never seen by GCMS
because of issues with volatility and the separation.
The absence of a separation technique
in this instance has really benefited our laboratory.
We're able to see compounds that we were never able
to detect before giving us much more information
as we look at lotions.
And again, it's a very rapid technique.
So it allows us to rapidly screen these things and pull
out all of this information.
So I would just like to say thanks to the NIJ
for giving me the opportunity to speak here,
and Amy for being our moderator.
And I'd also like to specifically thank Derek
Dorrien, one of the other examiners at the laboratory
who was instrumental in performing this validation.
And he helped me put together a lot
of the slides of this presentation.
So if any of the slides were not to your liking,
those are the ones he made.
And according to Amy, we'll be waiting on questions.
So that will be it.
[ Applause ]