Articles, Blog

Better, Faster, Stronger: Clinical Microbiology in the Era of MALDI-TOF Mass Spectrometry

August 24, 2019


Thank you so much it’s really an honor
to be here and standing up here I guess I’ve done pretty well if I’ve managed to
not have to give reigns until six years after my completion my fellowship but
I’m really am delighted to be here today so although i was born in limerick i am
not going to give my seminar in the form of a limerick I apologize it’ll be a
regular seminar so I only have one disclosure that’s pertinent to this talk
and that is honorary and research funding from BM are you the rest of my
disclosures are in the pieces of paper that many of you have so depending on
your age the Better Faster Stronger you might be thinking of the Million Dollar
Man or daft punk or if it got some really we ones in the audience this
gentleman here and I put this up because I think that that is a very pertinent
quote that we should all be telling ourselves every day to believe in your
finest and conquer you’re showing us so these are my learning objectives they’re
listed here and I’m going to begin by giving you an overview of a timeline of
diagnostic microbiology so the point here is not to be all inclusive with
respect to major discoveries in clinical microbiology or in Diagnostics within
the field of clinical microbiology but mainly to just show you some of major
discoveries and also really to point out that despite the fact that bacterial
culture and gram stains and other stains were first invented in the 1800s we’re
still using these techniques today and I’ll hope to convince you that that’s
not a bad thing just to put into context of what I’m talking about here today in
2006 is when the first maldi microbial identification system was launched okay
so obviously knowing that we have pc or and i will say my Irish accent makes me
say or instead of our so I apologize so PC or why can’t we just do pc or for
everything well there’s a couple of reasons that we don’t really want to do
piece your for everything and that is because just something as simple as a
Gram stain can be very advantageous it’s very fast for example if a patient is
suspected of having bacterial meningitis we
epicly will turn those gram-stain results around within 30 minutes or so
it can also provide very important information on the quality of the
specimen and importantly a presumptive diagnosis can often be made if organisms
are observed why do we want to continue to do bacterial culture and culture of
other pathogens bottom line is that it’s more sensitive than microscopy it’s
relatively cheap compared to molecular methods and this is really key that you
have to have the organism growing in culture to be able to perform
susceptibility testing so you might say well look we could just do piece your
find the target and be done but the problem is is that the books are clever
and they’re continuing to evolve new resistance mechanisms so an example of
this is a paper that was published in 2011 where they basically discovered
this new resistance marker that was associated with methicillin-resistant
Staph aureus called Maxie now regular staph aureus that were used to has a
mech ageing and so if you were only looking for the presence of the gene you
would have missed the fact that these isolates were methicillin-resistant and
if beta lactam therapy with the exceptions of tara lee would have failed
for patients with these infections ok so that’s the reason we want to continue to
do culture and gram stain but just to put it into context what are some of the
disadvantages of doing culture ok so this is a pretty typical scenario we get
the specimen in the lab we do a gram stain and we played it to a mixture of
solid and liquid media and we on the first day observe and a couple hours and
report it there are three plus Polly’s present as well as grand positive cocks
and clusters so the next day because the bacteria have to grow up and to be able
to work with them we see that there’s growth on the plate and we can do some
spot tests and reported a staph aureus because of Staph aureus we can also do
some spot test to look for PvP 2a which is the protein that’s encoded by MEK a
that confers methicillin resistance but the full susceptibility results are not
going to be available until day later now that’s kind of the best-case
scenario for a bug like staph aureus where we can use spot test to identify
it but the problem is is that for many most gram-negative rods you’re now
delaying your identification by 18 to 24 hours using standard identification
methods and then if it’s a mixed if there’s mixed colony morphology
separated those all present they all have to be separated out in order to
have enough growth to be able to work with to do identification and
susceptibility testing so bottom line is is that unfortunately this is how we’re
rocking in microbiology land and i’ll get back to why this is significant
later okay so I’m not here to tell you about traditional culture I’m here to
tell you about maldito so for all you mass spec aficionados in the audience
I’m going to be keeping this pretty basic so I apologize to all the folks
that are really down with the mass spec I’m going to keep this simple so as we
apply this to microbiology what we’re dealing with is a target plate to which
the colony is directly applied there are different steel plates versus plastic
disposable plates depending on which system that you use these are then
covered with a matrix that imparts a positive charge to the proteins that are
present and then these proteins then fly through a field free vacuum where they
reach a detector and so the time of flight refers to the time they’re
traveling through this tube and what results is the presence of the
production of characteristic spectrum that’s unique to that organism and then
this can then be queried against a database to result in an identification
so since the introduction of these commercially available systems for
identification you can see that they’re just have been an ever-increasing number
of publications that have been out there regarding maldi-tof and identification
okay so what does what difference does this make and why is this such a big
deal in the field of clinical microbiology so this is not for the most
part changing the initial specimen and requirement to grow it on a plate but
what it is doing is dramatically reducing the amount of time it takes for
an identification to be produced by the lab so in contrast to traditional
methods that are biochemically based that require at least overnight growth
the MALDI tough produces results in mere men
compare two hours or days and so although this is not actually real I
would encourage the MALDI manufacturers to consider including a black box
morning on their product because initial use of it can be associated with
symptoms of hyperventilate ventilation hysteria amongst lab personnel including
lab directors okay so just to summarize the platforms that are available what
I’m showing you right here is the bruker maldi bio typer that’s the first
instrument that came to market that’s actually the instrument that we use here
and so they have flavors of their databases they have an fda-approved
database that covers gram-negative organisms and then they have a separate
research use only database that has as you can see over 2,000 different species
present they also sell separate or uo micro bacterial and fungal databases and
we’ll get into that a little bit later in terms of how it makes its ID the
Brooker bio typer basically matches spectra it looks for things that are in
common things that are different and it produces a score value and so these
thresholds are important to score value of greater than or equal to 2 is a
definitive identification to the species level and a score greater than or equal
to 1 point 7 but less than 2 is considered identified to the genus level
I’m going to also talk about the second platform in my talk so I just want to
introduce it so you can understand why it’s a little bit different the first
thing it’s a little bit different in terms of size the broker lives on a
bench top the beemer you vitek ms is basically looks like a Coke machine and
so the poor guy who came to install it is really tired of people cracking jokes
about putting in money and getting out coke machines so it’s the truth epic so
the BMR you both systems got FDA clearance in 2013 so the ivd hefty a
clear database includes both gram-positive gram-negative organisms
east and anaerobes and it also has a research database but their research
database includes mycobacteria and filamentous fungi and so in contrast to
the Brooker platform which uses spectral matching the beer me you platform uses
what they refer to as advanced spectra classifier an advanced spectra crossfire
and basically all the spectra that are generated or translated into numbers and
sorted into bins and the different bins are
based on sensitivity and specificity for calling certain organisms so instead of
getting a score value you get a percent probability or a confidence value so
obviously the fact that it’s fast is great but if it’s not accurate than who
cares and so there are many many papers that have been published looking at this
but this is a table that was published by Robin Patel in 2013 and bottom line
is without getting into the details is that moldy Tov overall displays
equivalent or if not superior performance compared with traditional
identification methods but does it in a much faster manner so I will so I
mentioned that we evaluated this maybe I didn’t quite a long time ago on one of
the technologists who was involved in the initial evaluation when was susan
turner who is our lead in bacteriology and the other was Bruce illness and so
he used to refer to this as if you had a bug and you didn’t know what it is what
would God say what does God say it is so there’s this I think perception amongst
lab staff that the MALDI is somehow infallible and so I want to present a
case of what happened here to show you how it is not indeed infallible okay so
we had a 62 year old male who previously had been on holidays in ‘the island he
suffered heart attack when he was there which is a total bummer because you’re
in thigh land and you’ve only been there for three days and you’re hiking around
waterfalls and stuff and all of a sudden you’ve a heart attack which might be
kind of what led to the heart attack and he was hospitalized there for a week one
month after he came back to see Apple he developed symptoms of a urinary tract
infection and his primary care physician prescribed him nitrofurantoin there was
no improvement in his symptoms after four days of taking antibiotic and so he
went back to his doctor and a urine culture urine specimen was obtained for
culture and sent to our lab here at UWM see okay so after 24 hours pinpoint
growth was noted on blood agar so that’s the sort of common situation that we
would see for things that are gram neg gram-positive were not seeing growth on
our ground- mediate sort of pinpoint looking but then after 48 hours we saw
that there was a pure growth of what clearly now was the gram-negative the
had grown up sufficiently and it was growing on
blood auger and macconkey agar which are stamped standard microbiological media
it was oxidase positive and the MALDI kicked out an ID of burkholderia Thielen
densest and it gave a score value that although wasn’t sufficiently high to be
identified to the species level would be identified to the genus level however
it’s important to point out that many clinical labs that are using these
systems have validated lower score thresholds so depending on the organism
some labs will actually accept a score value of greater than or equal to 1.8 as
a species level identification so the problem is is that this was not what it
really was and so at the time there’s a joke in the lab that when things when I
could leave the country is when really bad stuff happens and there’s an end of
3 to prove this but in this case Tanis Dingell a former fellow was doing our
bench rounds because I was out of the country and the text said hey you check
this out this is so cool we have a burkholderia thailand dentists on a
urine culture she said close the plates and bag him up and thanks just tennis is
a student’s she rightly guessed that we shouldn’t believe the ID and this was in
fact burkholderia pseudomallei i which is a select agent it’s the causative
agent of meloidogyne web with a box 21 employees were exposed both high-risk
and low-risk certain employees were required to do things that maybe didn’t
make the most sense like come in on their holiday days off to give blood
cultures and all that stuff so it was a big mess and so the bottom line is that
these instruments are not infallible so in addition to this there are other
problematic organisms so some really commonly encountered misidentification
issues or known identification issues are that maldi will over call strep
pneumo so varied and stretch their are many it’ll tend to call them strep
pneumo it can miss identify shigella as e.coli and vice versa depending on what
database you’re using and also misidentification of a saprophytic
neisseria species as neisseria meningitidis
obviously select agents are an issue in the standard brooker database which we
use there are other than there are no select agents at all so there’s no
burkholderia pseudomallei there’s no brucella be anthracis any of that and so
these are huge gaps you don’t have a database that you just get out of the
box that’s going to be able to to help you identify these bugs some other
issues are that it’s not reliable for serotyping salmonella and it also has
difficulty resolving members of different complexes or highly closely
related groups for example the bacillus cereus group or beast patient complex
okay so I wanted to give you that’s really I guess one way to think of my
talk is it’s kind of like the new micro board exam it’s like a little bit of
everything but not in great detail so I’m just going to really talk about how
I’m all these change or practice and I want to give you one example of how it
has changed our workflow and allowed us to dramatically improve our turnaround
times so stool culture for anyone has ever been in a micro lab is an involved
process we look for a variety of stool pathogens we have to look for these
using a variety of different media that are both selective and differential and
so previously we had been using pretty standard media and tarek media used for
the recovery of salmonella and shigella species so hecto and Tareq and McConkey
augers but the problem with these is that there’s lots and lots of normal
flora that can mimic pathogens on these media so for example Proteus or
citrobacter and so what happens in labs is that you basically have all these
mimics that you have to rule out so you have to set up additional tests that
require additional growth in vitro and it’s just there’s lots and lots of work
lots of work associated with these cultures so the question was could we
combine maldito with a better media for the recovery of salmonella and shigella
species and how would that impact what we do so technologists turn in lab Katie
Schwartz did a study and when she presented ASM where she took two stool
specimens and compared the performance on chroma genic media that we chose to
study the hardy chrome Salmonella shigella auger compared with our
traditional workflow and so rather than soap culturing the organisms that grow
on the chromogenic media and identifying those we did perform maldi straight off
of the chroma genic auger and we didn’t know if that would work or not work
because chroma Jenna kharghar basically the colonies take on a color just like
they do on the heck to an auger and she compared this with our routine workflow
and so obviously if we got an identification of a coli as I’ve
mentioned brooker maldi isn’t very good for scent of a cola versus shigella we
would have to do traditional idea at that point but we cut out if this were
to work we would be able to cut out all these tubes that we have to do for all
of these suspicious colonies and so what we found was by using this auger it
actually had improved performance relative to our traditional method we
recovered more seminal and shigella I slits and what we found also was that we
had improved specificity meaning that we had less normal flora that we even had
to go to the trouble of doing a maldi on and the key of this was that basically a
negative culture could be a result of two days earlier than with our
traditional methodology the turnaround time for reporting a salmonella which is
a reportable organism was reduced by a day and we found almost fifty percent
cost reduction associated with using this new approach okay so I could I’ve
showed you hopefully how we can harness mont mall d to improve what we’re doing
but the bottom line is how much does this instrument cost and so the
instruments themselves are expensive they’re in the order of 200 to 250
thousand dollars now for people that are used to using plates and platinum loops
that we flame because we don’t want to pay for disposable oops that’s a lot of
money so is this something that’s really feasible for labs to bring in and so the
bottom line is that maldi-tof is cost effective for routine organism
identification so in this study published by Karen Carroll’s group at
Johns Hopkins they found that they had an annual cost savings of rennes hundred
thousand dollars per year and in our data what we found was that
we had annual savings of ninety four thousand dollars per year and we reduced
the number of islip so we had to send for sequencing I want to point out that
although we had reduced labor costs this ability to reduce the amount of work
involved it’s just basically silly work overworking allowed us to take on
additional testing without having to add FTE and it allows us to also bring on
new tests without having to add ft ok so up to now I’ve just been telling you
about the pretty routine stuff that we’re doing and so anyone who has small
children will know who this lady is over here but this was my Bible because it
was all about establishing the routine for small children but we’re going to
break the routine now for the rest of the talk we’re going to talk about three
different things we’re going to talk about maldi as it pertains to direct
blood culture identification maldi and its ability to I to identify micro
bacterium species and lastly I’m going to tell you about some very preliminary
work that we’ve been doing to use mouldy to potentially detect vancomycin
intermediate staph aureus so I’m going to begin with blood culture
identification so why are blood culture so important well they’re so important
because they’re what we use to diagnose the etiologic agent of blood stream
infections and so obviously bacteremia is a presence of bacteria in your
bloodstream but what what makes the news is sepsis which basically of a dis
dysregulated inflammatory response to infection and so steps is associated
with very high mortality rates and unfortunately no zap substitute unos the
co meal pathogens or infections are required in the hospital are associated
with higher mortality than those that are required in the community the number
of patients that are being hospitalized in this country with sepsis is
increasing and if for patients who who are septic they have 75% longer hospital
stays than those with other conditions so in 2008 is estimated that there was a
14 point six billion dollars being spent on hospitalizing patients with
septicemia and so obviously that’s unlikely to go down in terms of the
number at its current rate okay so one thing
that’s key though is the impact of delayed therapy so this is a paper
that’s widely cited that’s looking at a specific issue and this was a large
retrospective cohort study done across 10 hospitals 14 different icu’s between
1989 and 2004 and so bottom line is that for each hour that appropriate
antimicrobial therapy is delayed your mortality increases by seven point six
percent okay so think about that that’s really staggering especially in the
context of what I showed you before where we’re rocking it like
microbiologist and so just to show you how what microbiology does is not really
consistent with the clinical needs of managing sepsis we do blow culture which
is great because we get the bugs and we can test them but we have to wait for
them to grow up and then when they’ve grown up we do a gram stain and then we
have to wait in the standard algorithm you have to wait for it to grow on the
plate and then if you’re not using maldi you would have to also wait longer to
get a definitive identification now there are some tricks that can be done
here phenotypically low-cost to try and get a faster turnaround time there are
also importantly molecular tests that are available now that can identify
organisms directly from blood culture but I’m not going to talk about those
today instead I’m going to focus on moldy and discuss whether it has the
ability to also do direct identification from positive blood cultures okay so the
answer is that it can in that brooker the manufacturer maldi-tof instrument
has a blood culture specific module and so the difference between this and the
regular module is that it looks over a slightly different mass range instead of
looking from two to twenty thousand Dalton’s it looks between four thousand
and twenty thousand and also has a lower threshold for being considered
identified to the species level and brooker also sells a kit known as the
sexy thai / kit which basically is has washing and lysis steps and takes about
five minutes to perform the the processing and then you do a standard
extraction which is just ethanol formic acid acetonitrile so bottom line you 20
it’s or so processing time and so Raquel Martinez a former fellow of ours and
Elizabeth barley a master student who now works in the lab as a mls embarked
on looking at the ability of maldi to directly identify bacteria from positive
blood cultures this was done as part of a larger study with dr. Fong Harborview
looking at molecular methodologies to but I’m just going to focus on the part
that has to do with maldi so Raquel and Elizabeth looked at a hundred and
fifty-nine positive aerobic blood cultures so we didn’t include anaerobic
blood cultures in this study because of the study design and other assays we
were looking at and so what we found is pretty standard with what most lab see
in terms of our organism distribution staff Lacock fire what we see most
commonly followed by a colon and so what we what they did was compared the
identification that they got directly from the blood culture to that which
they got from the bacteria when they were growing in colonies and so just to
highlight the performance for broken down by ground positives and gram
negatives what you can see here is that if you look to the right where you’ve
the correct identification and these are all to the species level it does really
well for staphylococci we saw for co- staff eighty-four percent correctly
identified which is actually higher than was published in other studies with this
method possibly because we did some additional washing steps and it seemed
to do very well with the exception of as we would expect veradun strip so I’m not
the numbers don’t add up because I’m not showing you all the organisms I’m just
showing you the ones where there were issues so the veradun strep affected the
performance restrepo cock side but the non viridans streptococci worked and so
for the others so the micrococcus and some other gram-positive organisms
micrococcus didn’t seem to work that well and so I would say that we don’t
feel terribly confident about how well it performs for these types of organisms
but bottom line is it does very well for staff and very well for enterococci
which are very commonly encountered in clinical labs in terms of its
performance for gram negatives it did excellently for enterobacteriaceae so at
eric organisms again i’m just showing ecoli
but it really performed excellently for enterobacter citrobacter other common
organisms unfortunately it didn’t seem to do quite so well for the non
enterobacteriaceae and these are the ones that would really like to be able
to identify quickly because for example stenotrophomonas maltophilia is
intrinsically resistant to carbapenems and it’s not unusual for patients that
come in to be put on carbapenems as part of empiric therapy and so we want to be
able to catch those and let the physicians know that that’s what’s
causing their infection so it did not perform as well for non
enterobacteriaceae okay so what’s the upshot of using maldi directly to
identify blood cultures so obviously the pros we found in our study and this has
been consistent other studies that when you do get a species identification with
the exception of arid and strep it’s pretty accurate so we didn’t have any
other misidentifications in our study other than with a verdant strip if it
didn’t work it just didn’t work it didn’t it didn’t call it the wrong bug
so overall we found in our study taking the ground positives gram negatives and
yeast which I didn’t talk about together we found that eighty percent of
organisms were correctly identified to the species level just to put that into
perspective that’s an organism ID within a half an hour to an hour compared to
waiting you know at least 24 hours potentially longer to have a definitive
ID it’s very cheap to perform right these are the reagents are really low
there’s fairly minimal tech time involved compared to say some of the
molecular methods that are available but the problem is is that it struggles with
polymicrobial cultures so meaning if there’s more than one bacterial species
present it struggles to identify them and so 8.2 percent of our cultures in
our study were polymicrobial and although the MALDI was able to identify
one of the books that was present in around seventy six percent of the time
it’s not able to identify the second organism so it doesn’t do well for
polymicrobial cultures and this is really key it doesn’t provide any
information on antimicrobial resistance or susceptibility so even if we tell you
it’s a staph aureus with this method we can’t tell you whether it’s mrs a or
MSSA which has important room for therapy however I think that it can
be used as part of a an algorithm that includes molecular testing that maybe
this could be something to reflex to and I know in some very large medical
centers that you know eight center systems it becomes prohibitively
expensive to do molecular methodologies to assess this so they’re they’re often
using maldi okay so that’s great and all but does any of this make a difference
so one way to think about it is you might have a preference about what
method you’d like to do but is it really all just the same thing so the answer is
it does make a difference but it only makes a difference with the intervention
of antimicrobial stewardship and so we’re very lucky here we’ve great into
microbial stewardship team here at the U rupali Jana’s idea pharmacists uh Paul
Pottinger is the infectious disease a medical director and so antimicrobial
stewardship is key if there’s multiple studies that show that if you do rapid
identification of positive blood cultures it does nothing unless you have
stewardship involved if you just put the results out there you can’t expect that
anyone’s going to act on it and so the bottom line is it does make a difference
and so these are some studies that have been published with regards to using
maldi-tof performed directly on positive blood cultures so the first study was by
sarah cosgrove group hopkins and this was a theoretical study and so looking
at their retrospectively they calculated that they’d be able to save avoid 58
days of therapy per 1,000 patient days and that there would be a forty-five
percent or so of patients would have improved time to procure p but that’s
theoretical was actually been shown so there’s a really nice study that was a
performed by Gemma Sarah’s group and it showed that using direct identification
of organisms in combination with antibiotic and this was performed using
multi obviously that they found a reduction in all calls 30-day mortality
from twenty one percent down to nine percent around 20,000 or so has a dollar
saved per patient which if you multiply it out in their large system would
result in a cost savings of 2.4 million dollars which are really
huge numbers but just to show you that it’s not just necessarily all having to
do it at the direct identification level if you partner with stewardship closely
which Dwayne Newton’s group did at Michigan they found that even when they
just did maldi off of the colonies that were growing they didn’t process the
bottles or do any of what I just described they just did their
identification off the colonies but made sure that they engaged stewardship and
and they were involved in changing therapy they observed again pretty big
reductions in optimal and antibiotic therapy decreased mortality although not
to the same extent as that was observed here possibly because of the difference
between most likely due to the difference of doing it directly versus
waiting for the colonies to grow and a decreased length of intensive time in
the ICU so really it does make a difference okay so switching gears I’m
going to talk about now going from a specimen a positive blood culture back
to identifying an organism and so I’m going to tell you about mycobacteria now
obviously is dr. shirts mentioned I did a postdoc working on TV so mycobacteria
have a special place in my heart but why do we need to identify these organisms
so the mycobacterium genus actually has over 150 different species and although
there are some really great molecular detection strategies for this for
designed really for tuberculosis they’re not perfect and so the bottom line is we
still need to perform culture from like bacteria we can’t just rely on molecular
detection of these organisms is the reason you have to identify these is
that they’re important treatment differences between what we refer to as
non-tuberculous mycobacteria and TB and so for example there’s differences
between infection treatment fruit perfections you to micro bacterium for
to atone versus I’m chill Oni as a result the American Thoracic Society and
the infectious disease Society of America recommends that all clinically
significant micro bacterial isolates be identified to the species level wherever
possible okay so how do labs typically identify mycobacterium species well
traditionally ob’s would have performed phenotypic testing these are based on
growth characteristics biochemical pigment production and when you think
about mycobacteria there are rapid growers that can grow within a week of
being subcultured and then there are slow growers that can take weeks and
weeks to grow so trying to do phenotypic testing on something that takes weeks to
grow is just not really ideal hplc has also been used it’s not widely available
but by all accounts labs that have it find it to be quite good and then
there’s molecular assays now there are probes that can be done on positive
cultures and those are those don’t take days those are rapid they’re great but
what you can see is that they only cover these species these five species here
although these are actually complexes technically so that leaves us with
having to do sequencing which is fantastic it’s really accurate but
bottom line still takes days and it’s very expensive and so as much as our
clinical colleagues might like to think that these are the types of folks that
work in the lab we have real people working in relapse and we don’t want to
stick them with having to do a ton of manual labor for no game where if it’s
in inaccurate so bottom line is that i think there’s a real need to have rapid
cheap identification of mycobacteria so bottom line is if you could use maldi
you could do this very cheaply because mouldy so quick you could do it very
rapidly and we know from maldi and other organisms just very specific and has
high accuracy but with regards to mycobacteria there are some important
considerations first is if you have a slowly growing mycobacteria species you
don’t know if it’s a TB or not and tuberculosis is a college it’s as
important cause of laboratory acquired infection so the infectious dose for TB
is very very low and if it’s not handled appropriately you can become infected
the cell wall of mycobacteria are different than the cell wall of many
other pathogens they have very lipid-rich cell wall so that’s also a
factor how well would it work on a maldi when you have this really big thick cell
wall around the book how reproducible is the method how much organism do you
actually need when you got this big thick soil wall around
what’s the story with databases because and I’ll get into this a little bit
later with regards to what’s available from the manufacturers so bottom line is
when we looked at this and I was really really lucky to have just it’s
phenomenal resident Cheryl neither work on this project this was this was her
baby and she just did a fantastic job and and when Cheryl started looking at
this what we found was that there was really very little little published on
the performance of maldi to identify mycobacteria and so these are this is
really it when she started these are really the only publications that were
there so I’ll draw your attention to two salient features the first is that the
number of strains in the databases were pretty small like less than 50 in most
cases and the second thing was just the diversity of different extraction
methods were being used now somebody back in 2006 decided it was a good idea
to put mycobacteria directly onto a plate and just stick on your matrix and
off you go that’s not a good idea because matrix doesn’t actually kill
mycobacteria mycobacterium tuberculosis and although actually they managed to
get it to work that’s not something that we want to do similarly you can have
sort of hit or miss and activation in ethanol so that’s not necessarily
approach that we want to do so when Cheryl looked at the literature which
she saw was just this smorgasbord of methods including things like micro
pistols and all sorts of stuff and so what were the manufacturer saying well
the manufacturer of the instrument that we use had a procedure which is great
they have a procedure they have a database let’s just use that but when
you actually look at what their recommended procedure was it involved
you to do extensive manipulation of the culture before you ever inactivated it
and that’s just not something i think that is safe or practical for labs to do
you don’t want to be spinning and washing a bunch of times before you
actually kill off your TV so what we decided to do was to build our own
database and so what Cheryl did was rather than trying to build a database
that covers all 150 different organisms instead she elected to choose organisms
that were specifically highlighted by the American Thoracic Society IDSA
guidelines as being most clinically significant obviously that those
guidelines were about non-tuberculous mycobacteria you have to include
TB and so we also included some other organisms particularly and lanta flavum
that we commonly encounter in the lab now everything with a line through it or
bugs that we didn’t have access to and I just want to take this chance to this
opportunity to thank the molecular section of the lab because all of these
isolates that Cheryl was able to use had been definitively identified using DNA
sequencing and we’re in the stocks in the molecular section ok so what Cheryl
did was she basically developed her own extraction method and so it’s shown here
on the right and so the first step here is to basically inactivate the organism
which is really what any good method should be doing just kill the TV before
you start monkeying around with it and so what she did was to transfer only one
loop full of mycobacterial biomass and importantly some other procedures had
used up to 10 micro litre loop full so it was quite a small amount in deionized
water with silica beads and then she heated those for 30 minutes and then
added ethanol did a bead beating procedure and then did a standard
ethanol formic acid acetonitrile extraction so compared to other
published methods and particularly compared to the manufacturer recommended
method this was a remarkably simple Cheryl then used this method to build a
database which I’ve shown here and so you don’t have to pay attention to much
to everything that’s in here but the bottom line is that we she was able to
build a database containing 123 different strains across the species
that she targeted in some case we didn’t have very many of them in some cases we
were able to dramatically increase the numbers and so what she chose to do was
to combine our database with the existing manufacturer database and see
if it works but there’s no point in doing all of that unless you know that
your extraction method is going to inactivate the organisms and
particularly this is really focused on TV and so what Cheryl did was after her
an activation step she inoculated both solid media as well as mycobacterial
liquid media with a hundred microliters of sample and incubated her plate and
her brats for six weeks I believe or eight weeks wonder I kara member which
one it was the appropriate time for recovering mycobacteria and so she
tested five strains of TB and the other organisms shown here and what she found
was that there was no growth so this is great because we have
a method that’s effectively and activating our organisms and so then the
important thing to do is to say does it actually work so it’s all well and good
having a simplified extraction method but if it’s not going to work then
what’s the point and then at the same time we happen to have an instrument
from the other manufacturer the other commercially available manufacturer of
the beer murray vitek MS in the lab and so Cheryl decided would be interesting
to compare the performance of her database with that of the other system
now importantly that other system actually also includes a relatively
simple simplified extraction procedures well that has some differences with ours
but she wanted to see how they stacked up so the bottom line is that both our
Cheryl’s u-dub procedure and the beer me you extraction procedure produce spectra
of similar quality so down here in the x-axis are the different parameters that
are commonly used to assess spectral quality and what you can really see is
that between the u-dub protocol and the BMR you protocol was that these were
fairly equivalent the only place we noticed a difference was in terms of the
maximum intensity of when you ran both Pro X tracks from both protocols on the
vitek ms instrument and we don’t really know what the significance of that is
but bottom line is that share at the UW action procedure was generating spectra
of high-quality equivalent to that of a manufacturer recommended protocol okay
so how did this stack up so Cheryl ran many isolates and this is a busy slide
so I’m just going to highlight the pertinent features in red I’m bottom
line is that when Cheryl tested her database and so she combined her
database that she built with the manufacturer database and ran her
isolates using her extraction method she observed that nine almost ninety-five
percent of strains were identified correctly to the species level the other
manufacturers system and by extraction performed very very similarly Cheryl
also took her extraction her extractions and random just against the Brooker
recommended a bruker database and what you can see is that the identification
numbers were there was a poorer identifications obtained
however interestingly regardless of what database you used or what extraction
method you use we observed a hundred percent correct identification of
tuberculosis and so there’s just a highlight one last pertinent feature we
were correctly able to identify chillona from the M obsesses group for both
databases and both protocols okay so lastly I’m going to talk about something
completely different and this is as I said very preliminary work and this is
we’re trying to see if we can use multi tough to detect vancomycin intermediate
staph aureus and again I have Cheryl who’s expressed interest in working on
this and is this is really her baby and she’s running with this and we’ve been
really lucky to partner with Brian worth who’s a fabulous ID pharmacist in the
pharmacy department here and show beanie who’s a ID fellow from Australia so
there’s a nice little group of us working on this so vancomycin and why do
we care about it so I’m sure everyone here is heard of ANCA meissen so
vancomycin is really the mainstay in terms of treating serious infections due
to mrs a it’s a glycopeptide antibiotic that inhibits cell wall synthesis and
this is an antibiotic it’s active against gram-positive organisms and so
in terms of what breakpoints we use to call something susceptible the clinical
laboratory standards institute has a break point set of susceptible bugs or
isolates that have em ICS less than or equal to 2 micrograms per mil and this
number this less than or equal to 2 is important and it’ll come into play as I
continue through a range of my talk it sets a intermediate criterias mi ceza
four to eight and resistance criteria of greater than or equal to 60 so the
reason the two is important is that it has implications with regards to the
pharmacodynamics of this drug so the optimal pharmacodynamic parameter that
has been shown to correlate with treatment success for vancomycin is the
area under the curve over the mi and that that ratio has to be greater
than 400 so I know that not people are really versed in this and the MG’s an
audience at some point in their med school learns about this but I know it’s
not something that you guys think about a lot but the bottom line is that you
just have to remember that this number greater than 400 is what seems to
correlate with success the problem is is that when they model this the
probability of target attainment so the probability that you’re going to have an
auc over mi SI ratio of greater than 400 is very very variable when your mi c is
too so remember when you’re mi SI is to you’re still considered susceptible and
so you’re probably a target attainment can reach from anywhere from zero to
only fifty-seven percent and that fifty-seven percent is that very high
doses of vancomycin and if you administer those doses of vancomycin up
to thirty-five percent of patients getting those doses will have
nephrotoxicity so and this is even when the vancomycin trough is looking peachy
like it’s supposed to and so to show how mi see correlates with vancomycin
failure or success there’s this very nice meta-analysis that was published
not too long ago comparing the results of all these studies may stratified they
looked a number of parameters but I’m going to focus on vancomycin failure or
success they stratified it based on the mi see that was obtained and so what you
can see is that for patients who have mi seas that they characterized as high
which is greater than or equal to 1.5 and so 1.5 usually in a lab would be
rounded up to 2 because in for mi see testing you usually report to the next
doubling dilution and they compare those two patients who had mi sees that
they’ve considered to be low which are mi c is less than equal to 1 and what
they found was that patients with high mi sees the odds ratio was 24 vancomycin
failure okay so on top of that you can then
layer this issue with vancomycin intermediate staph aureus so I told you
that there’s there’s a susceptibility criteria for vancomycin intermediate
stuff where is 48 I’m not going to talk about vancomycin resistant staph aureus
or V or si these are bugs that are very very rare there’s only been three kind
of clusters of infection and they are due to staff worries that has acquired a
van a gene from enterococcus faecalis I’m not talking about that instead i’m
talking about vancomycin intermediate stuff where is so it’s really
interesting about these strains is that they look different than regular stuff
than regular stuff where is so they have this thick and cell wall they can also
have decreased autolytic activity and other features and so the problem is
that in terms of visa it’s not just you do to a single change at the genetic or
transcriptional level so for mrs a we can just detect mac a for certain gram
negatives we can detect the presence of certain carbapenem Aizaz we can’t do
that for vancomycin intermediate staph aureus but they do many of them most of
them in fact have this thick and cell wall and what’s that that’s believed to
do is to basically reduce access of antibiotic to the target so in
particular at the tips of the the septum that’s forming has there’s reduced
ability of the drug to reach that target because of the thick and so wha so
that’s vancomycin intermediate staph aureus so you should think ok well
there’s clsi criteria if its intermediate that’s great we just report
it and problem solved well the problem is is this problem with H visa so H is
for heterogenous and so the issue with H visa is that the resistant population is
present at a frequency of 10 to the minus 50 10 to the minus sixth sometimes
even less frequently than that but when we do susceptibility testing in the lab
we use an inoculum size of five times ten to the fourth so what that means is
you could potentially be missing this subpopulation of resistance when you’re
doing your susceptibility testing and so there’s really only one standard
method that has been proposed and proven as the gold standard for detecting H
visa and that is this population analysis profile and as the name
suggests it is one of these it is time consuming it’s labor intensive it’s not
something that labs can actually really be doing in the clinical setting and
essentially what they’re looking at is the population analysis profile over the
area under the curve compared to a control strain and if it has a ratio of
greater than or equal to 0 point 9 it’s considered to meet the criteria for
being H visa Brian in his lab does this and it’s it’s a lot a lot of work okay
so so far I’ve told you that really why would I even care about H visa because
we already know that if you have a vancomycin in my sea of two that you’re
more likely to fail vancomycin therapy well perhaps it’s not just a numbers
game after all so this is a retrospective multicenter study that was
published looking at patients between 2004 and 2012 and they had a hundred and
twenty-two patients with mrs a bloodstream infection and they had 61
matched pairs and so they did pop analysis on all of these retrospectively
and looked at the outcomes and compared the outcomes in people that were
confirmed to have h visa and those that are confirmed to have true vancomycin
susceptible staph aureus 0 vs SA and so bottom line is for every parameter they
looked the patients that had h visa did worse so in contrast the two fold odds
ratio for vancomycin failure that you saw if the MSC was too if h visa was
present those patients had 11 times greater odds of failing vancomycin
therapy they had an increased length of stay of eight days so remember it’s
about 1500 to 2000 dollars for each day you’re in the hospital multiply that by
eight that adds up I think this is really key in this study they didn’t
observe any relationship between the AUC over mi see and vancomycin failure and
so when they looked at patients that had a auc over mi c of less than 400 they
had the same rates mice and failure as those patients that
had a ucs over em ICS auc over m IC of greater than or equal to 400 so that
would argue that perhaps it’s not the mi see that matters so much it’s actually
the presence of h visa so what do we know about h fees in terms of its
prevalence it varies widely between studies and part of that is how these
studies are performed some of these studies they freeze the eye slits and
they don’t try and enrich for these when they’re pulling out the bugs and then
they make the conclusion that there’s no h visa here some studies have claimed
that they have prevalence rates of up to fifty percent but when as we know
unfortunately when h visa strains are tested and using standard cept ability
methods they can have low mi seas so we can’t just use a higher mi cita as a
criterion for looking for these strains so there’s a real need to have a way to
identify visa h visa and there are lots of different culture based methods that
have been published but they have very variable performance and can require up
to 48 hours of culture incubation similarly the infection C society
guidelines of America’s state that they don’t endorse changing therapy if mi cs2
but on the other hand they say if you detected visa that you should change
therapy so what was the goal here and I just have a few slides showing you some
pilot date on this the goal is to develop a rapid maldito based tool to
detect visa h visa from blood streamers eye slits and we believe that if we
could do this and we could do it through maldi in a rapid manner and be able to
report that we would be very interested to see if those patients have a more
positive outcome so Cheryl did a little pilot study a while back where she took
13 known visa strains and 12 presumed vs-si isolates we hadn’t confirmed them
at this point and compared the spectra on blood agar compared to Muller hinton
with vancomycin and the reason she chose Muller Hinton was because Muller hinton
is something that most labs have and so what she observed was for despite
looking at many many Peaks there was really only there was no unique peak
that was observed to be different in terms of presence or absence between
vs-si and v’s on blood agar but on Mueller hinton with point 5 micrograms
per million camosun she observed a one peak that was present
at a little over 6,000 Dalton’s but the problem was that the peak was absent
from twenty-five percent of the visas that were tested so it looks like it
could have something to do with it but it’s not going to be it’s not going to
be all based on this one peak but when she was looking at these spectra she saw
that there was differences with respect to the intensity of the peaks between
vs-si and visa so very very recently and these are really hot off the press data
so we have looked at 16 confirmed visas and 47 confirmed vs SAS and these were
all confirmed in Brian worth slab by the gold standard method and so Cheryl
basically tried to cast a very very broad net and so her criteria were very
not too strange in to begin with she basically observed that there were three
hundred ninety four peaks that were present in at least one of the spectra
that she obtained for each eye slit on both pieces of media and so then she
tried to narrow that down and so ultimately she found that there were 168
Peaks that were shared between each media type within a class so within visa
or within vs-si and so when she looked at these and she picked a threshold of
having a intensity that’s one point three fold greater what she observed was
between visa and vs-si that there were Peaks that changed in intensity between
these but for vs-si and h Fiza there weren’t any peaks that differed in
intensity and so that suggests that what were that really just speaks to the
issues with trying to detect each visa but regardless she built a training set
and she picked 10 visa eye slits and 10 vs sa isolates and she chose these
isolates based on mi c and their differences in their mi cita p a p value
and she I’m not going to go into details too much because I’m short on time she
picked criteria that were consistent with what had been shown in literature
for other examples of trying to use peak presence or absence or intensity and so
she tested the remainder of the visa and vass AI slits and what she found was she
had saw a hundred percent sensitivity for visa detection an eighty-four
percent specificity so meaning that it called visa correct
whole time and was mostly right about calling the SAS fee of the ssas and
interestingly when she ran forty seven percent of the H visa strains even
though they weren’t in the testing algorithm the training algorithm it
actually called forty-seven percent of these visa even though she didn’t really
when she looked at Peaks like I showed you in the previous slide there wasn’t a
huge but it wasn’t huge difference for the blood auger it didn’t work quite as
well but interestingly even with blood auger she was able to observe that
twenty-seven percent of hey Jesus trains were called as visa so in conclusion I
hope I’ve convinced you today that maldi has revolutionized clinical microbiology
it’s allowing us to change how we do routine identification it’s allowing us
to look for things we never thought we could look forward before like visa H
visa it’s that made dramatic impact in terms of our ability to identify
organisms that we routinely see it has the ability to reduce the time to
identification of blood culture eye slits accurately and rapidly identify
micro bacterial and just really quickly my acknowledgments it would be remiss of
me not to point out the immense contributions of dr. cookson and
hoofnagle without whom we would not have had maldi here they were instrumental in
getting them all the instrument here for the initial eval I think I’ve mentioned
most of these folks and last but not least everybody who’s involved in the
uwc micro lab and with that I will take I think of a few minutes left for
questions Jenny excellent talk my question is
what’s the potential for high-volume rap for
that I ask any one of the we might look there
reply to culture a forest platelets poop they need to be 24 hours old typically a
flood Center will hold on to them for a minimum of 12 hours for some cases 24
ass before that issue them to hospitals to so the question is it’s a great question
and I just like to say that I’m not here to answer credit market is crewed up I
was going to say I’m just here so I don’t get fined by dolly screwed it up
that’s a great question so the question is what’s the ability to do high
throughput on this because there may be a need and blood banking to turn in to
release units faster and so bottom line is there’s great potential for
high-throughput the system that we use you can run up to 96 organisms at a time
and so on other platforms they have the capacity to actually run even more at
that time so there’s very high throughput and it’s very short essay 20
so how many oh gosh I don’t think we’ve looked at that but there’s data that
I’ve seen and I want to say it’s a couple hundred but the question sorry
was how many could run an error i believe it’s up to a couple hundred but
it all depends if you have more instruments you can run more so some big
labs have like two three instruments for instruments thanks so one of the reasons
60 necessary useful target for education we have so
many sequences and there we have resources for reference data are there
any prospects for a public database of mass effect spectra or is it just going
to be sort of a bunch of clothes platforms that vendors provide or the
individual labs developed for themselves so that’s a great question the question
is what’s the potential for having open-source platforms open source
databases for microbial ID by maldi or we just sort of limited to what the
manufacturer gives us or what we all develop and i would say that so anilao
who’s at the NIH has developed the NIH themselves have developed a number of
identified at abase a–‘s including for fungi and mycobacteria and those are
publicly available there is no reason that you can’t share the database that
you build with anybody else if you have the or you owe platforms so there is
potential I just don’t think there’s a whole lot of that happening to date but
there are NIH because it taxpayer money built them they’re freely available to
whoever wants them so great talk jevlin machine you have a lot of methods and
thousands of species so if the FDA got their way and regulated lab developed
test how would you even begin to comply with such a versatile
sort of wide-ranging so the question is how would one comply
with proposed new regulations by proposed by the FDA where this would be
considered an LD t because this is recorded I’d like to plead the fifth is
that the terminology is here that’s what I’m going to do but we have to talk to
you about it afterwards if there’s not any more questions and thank you guys so
much for your time you you

1 Comment

  • Reply amol bajaj April 15, 2016 at 5:58 pm

    Just an awesome talk..

  • Leave a Reply