1
Exploding Head Zone – The Interface of Molecular and Growth‐Based
Drug Susceptibility Testing
Ed Desmond
Diplomate, American Board of Medical Microbiology
Good:‐ New technologies improve turnaround times and accuracy
‐ CA lab now offering reference DST by culture‐based and molecular methods to 12 states with low test volumes
Bad:‐ New knowledge shows our understanding of DST was naïve
‐ Inconsistencies in media and techniques lead to inconsistent and/or conflicting results
Good News & Bad News:
DST Services at CA State Lab for Other States
• Funded by CDC through APHL
• States with lower volumes of TB DST (<50/yr) are eligible
• Now doing work for CO, DE, IA, ME, MT, NE, NH, NM, RI, SC, VT, WY
• Routine I,R,E,P DST by MGIT on submitted isolates, reflex to 2nd line DST for resistant isolates or on request
• Molecular detection of resistance to INH and RIF directly on smear + specimens (with prior approval)
2
New DST Service at CA Lab, cont’d
• All rifampin‐resistant strains are sent to CDC MDDR service for sequencing
• When there is suspicion of drug resistance, please request pyrosequencing for rapid result (1 day)
– Country of origin, previous treatment, exposure to drug‐resistant (DR) case, etc.
– When a culture is received for DST, a subculture is first made and DST is performed on subculture.
– Range of TAT 10‐20 days, mean 15 days, but remember DR strains may grow more slowly!
What happens when you don’t tell us there is a suspicion of drug resistance
• Aliquot of MGIT broth received 8/28/15 & subculturedon that day
• 9/11, growth first observed in subculture and first line DST set up (drug‐resistant strains may grow slowly)
• 9/21, DST for first line completed: MDR!
• 9/22, PSQ performed and indicates resistance to FQ and injectables: XDR!
• TAT: 25 days to detect XDR
• TAT for detecting XDR by PSQ if you request molecular testing because of suspicion of drug resistance: 1 day
Exploding Head Zone: Molecular and Culture‐Based DST for
Ethambutol & Pyrazinamide
3
Ethambutol
Culture‐based ethambutol DST for TB strains with wild‐type embB sequence (expanded MPEP)
StrainembB
sequence
Agar proportion result Lab A
Agar proportion result Lab B
MGIT MIC Lab C
2 WT R R >5 (R)
3 WT R R >5 (R)
6 WT S S <3 (S)
7 WT R S 4 (S)
8 WT S S <3 (S)
10 WT R R >5 (R)
16 WT R S 4 (S)
21 WT S S <3 (S)
27 WT S S <3 (S)
Some ethambutol resistance is not detected by embB sequencing
MGIT detects less ethambutol resistance than agar proportion (AP)
Ethambutol: November 2010 MPEP– duplicate isolates with Met306Val mutation
–90% of labs using AP detected drug®–23% of labs using MGIT detected drug®Note: for most cultures, ethambutol results by AP and MGIT agree. The frequency of discrepancies is unknown. But there is a tendency for AP to detect more resistance than MGIT
4
Culture‐based ethambutol DST for TB strains with met 306 ile mutation in embB gene
StrainembB
mutation
Agar proportionresult lab A
Agar proportion result lab B
MGIT MIC (lab C)
4 met 306 ile R S 5 (S)
13 met 306 ile R S >5 (R)
15 met 306 ile R * 4 (S)
22 met 306 ile R R >5 (R)
23 met 306 ile S * 4 (S)
28 met 306 ile R S >5 (R)
29 met 306 ile R S <3 (S)
30 met 306 ile R * <3 (S)
Isolates with the same mutation may differ widely in suscept. vs. resistanceLabs with good QC, doing AP DST for EMB may differ significantly in results
Findings from Expanded MPEP Regarding Ethambutol
• Some strains resistant to ethambutol do not have mutations in the embB region which is routinely sequenced
• Laboratories with strong QC programs may differ from each other in their (AP‐‐agar proportion) ethambutol results with some strains
• TB strains with the same embBmutation may differ significantly in their ethambutol MICs
• MGIT may detect less resistance than AP
Ethambutol—on what can we rely?
• Patient isolates that are susceptible to isoniazid are very unlikely to be resistant to ethambutol
• Madison, et al. 2002. J. Clin. Microbiol 40:3976, table 2– For isolates which are clearly resistant to ethambutol (by both BACTEC 12B and agar proportion), 96.6% of these are also resistant to isoniazid
• For isolates which test susceptible to both INH and ethambutol, the ethambutol result can be considered reliable (my opinion)
5
What about INH‐resistant strains—is a MGIT susceptible result for EMB reliable?
• Some strains (e.g., with Met306Val mutation) may test susceptible in MGIT and resistant by agar proportion– For evaluating treatment efficacy, it is unclear which one is correct
• One way to arrive at a decision: test by both molecular and culture‐based methods, and synthesize the two results
• Conflicting results may indicate diminished susceptibility but EMB might still be a useful part of the regimen
Synthesizing ethambutol (EMB) results
Lab findings Interpretation
INH susceptible, EMB susceptible
EMB susceptible
INH susceptible, EMB resistant
Questionable result, repeat
INH resistant, EMB resistant by any method (agar, MGIT, or embB sequencing)
Not fully susceptible to EMB
EMB susceptible by MGIT,resistant by agar
Not fully susceptible but clinical relevance is unknown
We have a de facto intermediate or grey zone.
Ethambutol Research Questions
• TB strains with wildtype sequences in embBmay have high EMB MICs: what other genes and mutations are associated with resistance?
• TB strains with the same embBmutation may have significantly different MICs: how do combinations of mutations affect susceptibility or resistance quantitatively?
• Linking of whole genome sequencing data with MIC values may provide answers
6
Where is the clinical correlation? We can work to make different lab methods agree, but are we really doing a better job of predicting treatment success or failure?How a rodent study might help: Develop a set of strains with a range of EMB MICs Infect rodents with these strains, then treat with EMB doses equivalent to human therapy
Sacrifice mice following treatment and evaluate whether EMB had been effective in inhibiting growth in lungs
Establish an MIC cut point, above which EMB treatment is not likely to be effective
Research Questions Re EMB testing in MGIT
Pyrazinamide
Pyrazinamide
Chedore, et al. 2010. JCM 48:300: of 57 isolates which tested PZA resistant in MGIT, only 33 (58%) were resistant on repeat testing.
Zhang, et al. 2014: Current phenotype‐based susceptibility testing is not reliable due to false resistance; sequencing of the pncA gene represents a more rapid, cost‐effective, and reliable molecular test for PZA susceptibility testing… (Microbiol. Spectrum 24(4).)
7
o If inoculum is too heavy, pH may be raised and drug activity will decrease (false resistance)
o If inoculum is too light, it may not grow because low pH medium does not grow M. tbwell (false susceptibility)
o If culture being inoculated is old, metabolically inactive cells may have increased susceptibility to PZA
So hard to get it right!
Zhang Int J Tuberc Lung Dis 2003; 7: 6–21
Challenges of testing PZA in MGIT
19
• Use fresh cultures for inoculation to avoid false susceptibility
• Avoid over‐inoculation to avoid false resistance
• When a culture tests PZA monoresistant, repeat and investigate− Investigation—is it from an extrapulmonary site or other reason to suspect M. bovis? This would support resistance
• When repeat result becomes available:
− Resistant: confirmed. Report resistance
− Susceptible: sequence pncA gene including promoter region
Proposal for PZA
20
Molecular basis of resistance to PZAZhang 2015 IJTLD 19:1276
• Mutations in the pncA gene are the main mechanism of PZA resistance in M. tuberculosis.
• pncAmutations are highly diverse and scattered along the gene, which is unique to PZA resistance.
• Most PZA‐resistant M. tuberculosis strains (72–99%, average 85%) have mutations in pncA, but some do not
• Those that do not have mutations in the drug target RpsA. RpsA target mutations are usually associated with low‐level PZA resistance (MIC 200–300 μg/ml PZA)
• More recently, panD was found to be involved in PZA resistance. panDmutations were identified in naturally PZA‐resistant M. canettii strains and some PZA‐resistant MDR‐TB strains
8
Lot to lot variability in culture media sometimes affects drug susceptibility
testing results
Guthertz, L., et al. 1988 JCM 26:2338: “This study demonstrates that individual lots of components of this basal medium may vary significantly in their suitability for susceptibility testing, and failure to detect such variation may dramatically affect susceptibility profiles.”
• Medium base and (OADC) supplement contain biological materials which may vary from lot to lot.
• This may affect growth support or drug activity in media made with these components
9
Does variability in culture‐based DST systems make them less reliable? The example of PZA
Supplementlot number
PZA druglot number
Mediumlot number
MIC for H37Rv Date
4262976 4262938 4262838 100 April
2015
4038491 4066064 3305456 50 August
2014
3101100 3143405 3305456 25 January
2014
Quality control MIC testing of stock strain with new batch of PZA drug and supplement
Lot to lot variability—what it says in the CLSI approved standard M24
• “…laboratories should choose a broth based system from a manufacturer that performs regular quality assurance (QA) testing to ensure that the media and drug performance are consistent from one lot to the next, and that the media and drug lots provide consistent results over long periods of time (e.g. years) without drift.”
• “When deciding which shorter incubation broth to use, it is reasonable to ask the manufacturers to describe the measures they take to ensure lot‐to‐lot consistency in drug susceptibility results, and to consider this information in the decision process.”
• Both mfrs of rapid broth systems (MGIT and VersaTREK) essentially abnegate this responsibility (test H37Rv at 100 μg/ml only).
Maintaining lot to lot consistency of DST media and components
• One approach: define medium components as carefully as possible, and do pre‐market testing of components to assure lot to lot consistency– A manufacturer’s responsibility—parameters for evaluation of ingredients must be carefully established
• Another approach: when a new batch of medium component is received, make up some media with new and some with old component and compare growth support and drug MICs between new and old lots
– DST laboratory’s responsibility (an onerous one)
1
New generation of Cepheid GeneXpert MTB/Rif (“Ultra”) expected to have greater sensitivity
with smear‐negative specimens.
Cepheid may not seek FDA clearance.
10
Take Home Points• DNA sequencing provides rapid results not altered by subtle changes in reagents– Correlation between sequence changes and drug susceptibility may be complex and multifactorial for some drugs. Elucidation may require extensive study
– Some mutations are clearly associated with resistance
• Culture‐based DST may be affected by subtle changes in culture medium ingredients– Is rigorous QA by the end user practical?– Would more rigorous pre‐market testing by manufacturer help?
• For ethambutol, clinical correlation of DST may need to be re‐established
• Testing algorithms may be designed to overcome weaknesses in current test methods
New generation of Cepheid GeneXpert MTB/Rif (“Ultra”) expected to have greater sensitivity
with smear‐negative specimens.
Cepheid may not seek FDA clearance.
Cepheid also working on a product to detect XDRTB by detecting resistance to
INH, FQs and injectables: not yet clear if/when it will be released.
11
• In DNA sequencing performed at the MDL & CDC labs, 18‐20% of the mutations found by GeneXpert are silent/ synonymous mutations which don’t confer resistance
• This differs from studies performed in some high‐burden countries where true resistance is much more prevalent– Globally, the incidence of MDR TB is about 5 X higher than it is in the USA
Why do APHL and CDC recommend DNA sequencing each time GeneXpert
detects an rpoBmutation?
Rifampin PhenotypeResistant Susceptible Total 98% Accuracy
Xpert Rifampin‐R 9* 6** 15 90% Sensitivity
Xpert Rifampin‐S 1 418 419 99% Specificity
Total 10 424 434 2% Prevalence60% PPV
100% NPV
32
GeneXpert Rifampin Resistance Performance DSHS‐Austin Nov 2012 – July 2014 (434 specimens)
THANKS TO KEN JOST FROM TEXAS STATE LAB
Specimen Probe rpoB MutationTexas RMP
CDC RMP
1 A* No Amplification S ‐
2 B Phe514Phe S ‐
3 B Phe514Phe S S
4 B Phe514Phe S S
5 B Phe514Phe S S
6 B Asp516Tyr & Asp549Asn S S
7 C Ser522Leu R/S S**
8 D His526Tyr R R
9 D His526Asp R R
10 E Ser531Leu R R
11 E Ser531Leu R ‐
12 E Ser531Leu R ‐
13 E Ser531Leu R R
14 E not tested R ‐
15 E Leu533Pro R/S S**
16 none Ile572Phe R S** 33
THANKS TO KEN JOST
FROM TEXAS STATE LAB
Note that GXP probe B mutation was not associated with rifampin resistance.
12
• Some rpoB mutations are clinically significant, but TB strains with these mutations may test susceptible in MGIT DST
• More about this later
Why do APHL and CDC recommend DNA sequencing each time GeneXpert detects an rpoBmutation? (cont’d)
Working definition for the purpose of this talk:
Presence of a mutation in rpoB, which causes a change in amino acid sequence, leading to an increase in MIC above that of the wild type, but <1 ug/ml in MGIT so that the culture will test as susceptible in MGIT
“Low level rifampin resistance”
• Williamson IJTLD 2012 16:216
• 3 New Zealand cases in which Cepheid GeneXpert detected rpoBmutation, but culture tested susceptible to rif in MGIT
• Treatment failed in these 3 cases (all were INH‐resistant)
• Mutations 516 Tyr and 526 Leu were assoc. w/ rifampin MICs of 0.25 and 0.5 respectively
Reports of treatment failure associated with “low level” resistance continue
13
• Greater use of GeneXpert MTB/RIF will increase discovery of low‐level resistance ‐‐(GeneXpert says resistant, MGIT says susceptible DNA sequencing)
• Expanded use of GeneXpert is recommended
• When there is low level resistance, e.g., MGIT/Xpert discrepancy or sequencing, treatment regimen may need to be modified
– NS Shah, et al. 2016. Open Forum Infect. Dis., 3 (3). DOI: 101093/ofid/ofw150.
Rifampin issues, cont’d
Category
Presence of rpoB mutation
ExpectedGeneXpert
result
Resistant by agar
proportionResistant by MGIT
Clinical expectation
Rifampin resistant
Yes Rif resistant
Resistant Resistant Rifampin not useful
Low level resistance
Yes Rif resistant
+/‐ Susceptible Rifampin activity reduced, but still may be useful. Expert consultation needed. Treatment may fail.
Susceptible No Rifampin susceptible
Susceptible Susceptible Reliablecontribution of Rif to treatment
regimen
Low level rifampin resistance recap
RBU MIC (µg/mL)
rpoB MutationNone 531 tcg
ttgRif >8
516 gacgtc
Rif>8
526 cac ctc
Rif 0.5-8
526cacggc
Rif =2
526 cacagc
Rif = 1
526 cactgc
Rif 2-8≥8 2
4 25
2 5
1.5 2
10.5 9 1
0.25 6 1
0.125 5 2 1 1 1 2
≤0.0625 37 1 (RIF S)
Total Samples
42 34 17 4 1 1 2
Rifabutin MICs associated with various rpoB mutations
Rifamycins: some mutations may lead to phenotype of rifampin resistant but some rifabutin activity is retained. 516 gtc or 526 ctcmutation: use rifabutin? Note rifampin MICs in red.
14
For new drugs, like bedaquiline, how do we establish test concentrations for drug susceptibility testing?
Test concentrations: “Discrimination approach”
Canetti, et al. 1969. WHO Bulletin 41:21‐43
Test MICs of:
A. Strains from untreated patients and of
B. Strains from treatment failures
Test concentration = concentration at which A. are susceptible and B. are resistant.
When establishing drug susceptibility testing methods for new drugs, it is important to include clinically resistant strains from
treatment failures
• It doesn’t matter what other drugs the patient was treated with. If a patient is still culture positive after a regimen of treatment with the new drug, the isolate did not respond to that drug and is clinically resistant
• Clinically resistant strains may be different from resistant strains generated in the test tube because they have to survive in human tissue in the presence of an immune system
• It can be challenging to obtain TB strains clinically resistant to a new drug
15
What info can we obtain from clinically resistant strains?
• What mutations do they have?– Particularly useful—having a strain from a specimen collected before therapy, and the clinically resistant strain if therapy fails.
Mutations found in this case are likely to be associated with drug resistance
• What are their MICs?– How do these MICs compare with those of wild type strains?
– Make sure the wild type strains have not been treated with a related drug.
Pan‐susceptible strains may be most likely to be truly wild‐type
What can we do if we have few or no clinically resistant strains?
• Can evaluate the MIC range of wild type strains
– Isolates with MICs higher than this are probably resistant
• Where to set the critical concentration?
– Just above the upper limit of MICs of wild strains?
Some variability in testing could lead to false resistance
– 4 fold higher than upper limit of MICs of wild strains?
If clinically resistant strains have MICs lower than this, false susceptible results are possible
Epidemiological cut‐off (ECOFF):Angeby, et al. Bull World Health Organ 2012;90:693–698
• The highest MIC within the wild‐type MIC distribution has been labelled the epidemiological cut‐off (ECOFF).
• The ECOFF is used, together with clinical, pharmacokinetic and pharmacodynamic data, to classify a given wild‐type MIC distribution as susceptible (S) (high likelihood of clinical success), intermediate (I) (clinical success uncertain) or resistant (R) (low likelihood of clinical success)
• The authors recommend, in short, that susceptibility breakpoints for antituberculosis agents be systematically reviewed and revised, if necessary, using the same modern tools now accepted for all other bacteria and fungi by the scientific community…
• In other words, they can’t be bothered with the chore of collecting strains from treatment failure cases and seeing what their MICS are.
16
“Those who do not remember the past are condemned to repeat it.”
George Santayana
Angeby, et al (2012): susceptibility breakpoints for antituberculosis agents should be based on the same modern tools now accepted for all other bacteria and fungi, but
• In tuberculosis, some bacteria are intracellular and some are extracellular, so the usual pharmacokinetic studies don’t apply
• Some TB bacilli are metabolically dormant and others are actively growing. The drugs and concentrations needed to kill these varying populations may differ
William Burman. 1997. The value of in vitro drug activity and pharmacokinetics in predicting the effectiveness of antimycobacterial therapy: a critical review. Am. J.
Med. Sci. 313(6):355‐63
• In vitro measures of drug activity ….do not predict the sterilizing activity of a drug or the activity of drug combinations.
• In vitro measures of drug activity do not allow reliable predictions of the level at which an organism should be considered resistant.
• Assays of drug penetration in tissues and activity against intracellular bacilli add modestly to the predictive value of in vitro measures but still do not predict sterilizing activity.
17
How do we decide on test concentrations for TB drug susceptibility testing?
On the shoulders of giants we can see a long way—the Canetti discrimination method.
Thank you!