Transcript Prediction and Prevention of Emergence of Resistance of Clinically

Prediction and Prevention of Emergence of
Resistance of Clinically Used Antibacterials
Fernando Baquero
Dpt. Microbiology, Ramón y Cajal Hospìtal
Madrid, Spain
The basic process
Variation: mutation rate
Environment
Selection of variants
Evolution of Antibiotic Resistance
Baquero, ICC 1999
House-keeping gene
C
O
M
p
Genetic variation
C
O
S
T
A1
Antibiotic selection
- selective compartments
Genetic variation
- gene recombination
C
O
M
p
C
O
S
T
A1
Genetic variation
- gene recombination
- accessory genetic elements
X
C
O
M
p
Antibiotic selection
- selective compartments
C
O
S
T
A1
A2
Antibiotic selection
(Multiple)
Genetic variation
- linkage colonization factors
Host
NEW HOUSE-KEEPING GENE?
Elements for Prediction
• Antimicrobial agent (A)
• Bacterial population/s (B)
• In-host environment of A/B interaction
• Ecology of host population
Emergence of mutational resistance
• Resistance is a function of the product of
original inoculum, rate of reproduction and
the mutation rate, divided by the negative
growth rate (reduction in susceptibles).
If high inoculum size  resistance
If no starting mutants, best S killer  resistance
If starting R mutants, best S killer  resistance.
(Lipsitch and Levin, AAC 1997; Austin et al., J. Theor. Biol., 1999)
Complexity in prediction of mutation rate
Target access mutations
Target protective mutations
Target structural mutations
Target structural mutations (1)
Antibiotic target-based mutation rate depends on:
• Target gene/s structure
Base composition determines possibility of mutation
The higher the gene size,  possibility mutation
• Target permissivity
Wide functional domains in the gene  mutation rate
• Target diversity
Multiple targets  mutation rate
• Target cooperativity
If inhibition of multiple targets are required for effect,  mutation
Target structural mutations (2)
• Target determination
If target is determined by multiple genes  mutation
• Target density
High number of target molecules  mutation
• Target redundancy
Multiple redundant genes encoding the target  mutation
• Target dominance
If modified target is recessive  mutation
• Target essentiality
Low cost target functional modifications  mutation
Prediction of antibiotic-resistance
theoretical mutation rate
• Mutation rate results from a
multifactorial set of conditions
• In-vitro mutation rate is only
mutation rate in vitro
Process of sequential selection of intermediate and
resistant variants
100
100
90
80
70
10
S
I
R
60
%S
50
%R
40
%I
30
1
1
2
3
4
5
6
7
8
9
20
10
0
0.1
1
2
3
4
5
6
7
8
Reduction in viability after exposure to different antibiotics or
concentrations. Effect on final proportion of different bacterial
subpopulations
9
Antibiotic Gradients in
Compartmentalized Habitats
Concentration-Dependent Selection of
TEM-12 over TEM-1 (mixed cultures1:100)
Selection coefficient
7
6
5
4
3
2
1
0
0
0.004 0.008 0.015 0.03
0.06
cefotaxime (µg/ml)
0.12
0.25
0.5
Time-dependent Selection of TEM-12 and TEM12/OmpF over TEM-1 in mixed cultures
Selection coefficient
10
8
6
4h
4
2
0
0
0
0.01 0.02 0.03 0.06 0.12 0.25 0.5
-2
cefotaxime (µg/ml)
TEM-12 selection over TEM-1 in mice treated with
cefotaxime: change in log TEM-12/TEM1
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0 mg/k
.25 mg/k
1 mg/k
4 mg/k
16 mg/k 64 mg/k 256 mg/k
P. aeruginosa mutation rates in cystic fibrosis
and bacteremic patients
Mutation-rates
Mutation-rates
1x10-5
1x10-5
3,6x10-6
1x10-6
1x10-6
1x10-7
1x10-7
2,9x10-8
2,4x10-8

<1x10-8
0
2
4
6
8 10 12 14 16 18 20 22 24 26 28 30
CF-Patients
<1x10-8
0
5
10
15
20
25
30
35
40
45
50
Bacteremic-patients
Antibiotic Resistance in mutator phenotype
P. aeruginosa from cystic fibrosis patients
% Resistance
90
80
70
60
50
40
30
20
10
0
Ticarcillin
Imipenem Tobramycin Norfloxacin
Ceftazidime Gentamicin
Amikacin Fosfomycin
Concentration-dependent E. coli mutS
mutation rate (rifampicin-resistance)
4,00E-05
)
0.5
0.4
0.3
0.2
0.1
37º/18 hours
2,00E-05
Mutation rate
0
1,20E-05
4,00E-7.5
0.5
0.4
0.3
0.2
0.1
CEFTAZIDIME
(µg/ml)
CAZ (µg/ml)
0
Why mutators do not predominate?
mutator
non-mutator
Stressful Environment
Exploitable Environment
Biological Cost of Low-level Resistance may be
Compensated before Evolution to High-level
Resistancel
HLR
LLR
Biological Cost
Sörensen and Andersson, 1999
Conditions that increases the rate of
antibiotic-R mutants (I)
1. High number of bacterial cells
2. Low antibiotic concentrations of the selective
agent, exerced during a prolonged period
3. Antibiotic degradation or inactivation
(spontaneous-binding-enzymatic)
4. Slow killing kinetics of the selective agent
5. Many different genes leading to resistance
Conditions that increases the rate of
antibiotic-R mutants (II)
6. Mutator phenotype (methyl-mismatch repair
defficiencies and other mutator mechanisms)
7. Up-recombination systems
8. Bacterial stress; Slow bacterial growth
9. No significant decrease in fitness of R mutants
10. Physically structurated habitat
Hungry predictive
mathematical models
• Models require the inclusion of important
parameters for which no quantitative estimates
are available for most host-bacteria-antibiotic
interactions.
• The use of models to design/evalute drug
treatment regimes will depend on the
availability of such data, and on how well the
models predict observed outcomes.
Hungry models for resistance:
what do we need?
Most models are based on:
1. Duration of infectiousness of infected individuals
2. Incidence of drug treatment
3. Extent to which treatment of susceptible population
reduces the transmission of the infection
4. Degree of reduction in fitness of the resistant bacteria
in the absence of treatment (cost)
5. Probability of acquisition of resistance during therapy.
(Science, 283:808, 1999)
The 15 essential components in the predictive
modeling of development of antibiotic resistance
(1)
.
.
.
.
.
.
.
.
R0
f
ß
µ
z0
w

y0
transmissibility of S or R genotypes
rate of loss of carriage
secondary cases per unit of time
removal or death of cases
initial frequency of R genotype
fitness of S or R genotypes
probability of selection of R genotype during therapy
endemic prevalence as a function of antibiotic use
The 15 essential components of the predictive
modeling of development of antibiotic resistance
(2)
.
.
.m
.a
.
.
. TR
erradication (lengh colonization/lengh therapy)
superinfection fitness (colon. of S/R hosts with R/S)
adquisition of resistance (mutation rate)
prescription rate x lengh of treatment
prescription rate per unit of time
change in consumption of antibiotics
time to reach a given frequency of resistance
Some parameters used in the study of Iceland S.
pneumoniae pen-R
. R0
.f
.µ
. z0
.
.m
.a
.
.
transmissibility R
loss of carriage
removal cases
initial R frequency
superinfection fitness
mutation rate
antibiotic pressure
prescription/time
change in consumption
2.1 cases per dase
2.6 months of carriage (1/f)
84 months of maintenance
-3.1 (log10 z0)
1 (R  S)
not considered
38 DDDs/1,000 children
10 days
-12.7 %
(Austin, Kristinsson & Anderson, PNAS 96:1152, 1999)
The patient and the community: the unified view
Patient
a. R proportional to total amount of antibiotic
b. R proportional to multiple sequential treatments
c. R proportional to persistance of R organism
Community
a'. R proportional to total usage of antibiotic
b'. R proportional to number of treated patients
c'. R proportional to endemicity of R organism