multidrug-therapy and evolution of antibiotic resistance: when order
TRANSCRIPT
Multidrug-therapy and evolution of antibiotic
resistance: When order matters
Running title: Sequential Therapy and Evolution of Drug Resistance
Gabriel G. Perron*1, 2, Sergey Kryazhimskiy*1, Daniel P.
Rice1 & Angus Buckling2, 3
1FAS Center for Systems Biology & Organismic and Evolutionary Biology, Harvard University,
Cambridge, MA, 02138, USA. 2Department of Zoology, University of Oxford, Oxford, OX1 3PS, UK.
3Biosciences, University of Exeter, Tremough, Cornwall TR10 9EZ, UK
Corresponding author:
Gabriel G. Perron
FAS Center for Systems Biology
Harvard University
52 Oxford Street, Cambridge, MA 02138, USA
Email: [email protected]
*These authors contributed equally to this work
Copyright © 2012, American Society for Microbiology. All Rights Reserved.Appl. Environ. Microbiol. doi:10.1128/AEM.01078-12 AEM Accepts, published online ahead of print on 22 June 2012
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
2
ABSTRACT 1
The evolution of drug-resistance among pathogenic bacteria has led 2
public health workers to rely increasingly on multidrug therapy to treat 3
infections. Here, we compare the efficacy of combination therapy (i.e. using two 4
antibiotics simultaneously) and sequential therapy (i.e. switching two antibiotics) 5
in minimizing the evolution of multidrug-resistance. Using in vitro experiments, 6
we show that the sequential use of two antibiotics against Pseudomonas 7
aeruginosa can slow down the evolution of multiple-drug resistance when the two 8
antibiotics are used in a specific order. A simple population dynamics model 9
reveals that using an antibiotic associated with high costs of resistance first 10
minimizes the chance of multidrug-resistance evolution during sequential 11
therapy under limited mutation supply rate. As well as presenting a novel 12
approach to multidrug therapy, this work shows that costs of resistance not only 13
influences the persistence of antibiotic resistant bacteria, but also play an 14
important role in the emergence of resistance. 15
16
INTRODUCTION 17
With the rapid emergence of microorganisms resistant to one or many 18
antibiotics (24), clinicians worldwide have increasingly relied on multidrug therapy to 19
fight bacterial infections (21, 22). Although multidrug treatments proved successful in 20
reducing the prevalence of severe infections such as Mycobacterium tuberculosis (10, 21
14), the pervasive use of antibiotics has resulted in the evolution of multidrug-22
resistance (MDR) in many species of bacteria (1, 26). MDR is now frequent in many 23
healthcare-associated bacterial infections such as Clostridium difficile, 24
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
3
Staphylococcus aureus and Pseudomonas aeruginosa (19, 21, 25), raising serious 25
challenges regarding the optimal use of multidrug therapy (15). 26
Two key objectives of multidrug therapy are to minimize the rate of evolution 27
of multidrug-resistant bacteria and to limit the total amount of antibiotic used in 28
hospitals (17). These problems can be approached at two levels: hospital-wide level 29
and single patient level. At the hospital-wide level, two therapeutic approaches are 30
typically employed: mixing therapy (i.e. two or more drugs are used simultaneously in 31
the hospital where each patient receives a single drug) and periodic hospital-wide 32
rotation of antibiotics (i.e. two or more drugs can be alternated periodically within a 33
hospital). While clinical trials for different bacterial infections produced mixed results 34
(4, 8, 33, 34, 40), theoretical results typically suggest that a mixing strategy minimizes 35
the evolution of MDR compared to hospital-wide rotation (9, 11, 42). The mixing 36
strategy is believed to increase the rate at which bacteria are exposed to different 37
antibiotics relative to rotation, therefore minimizing the opportunity for resistance 38
evolution. Although recent theoretical results suggest that it should be possible to find 39
an optimal rotation strategy that minimize resistance evolution (6), the range of 40
optimal parameters is a matter of debate (5, 11, 20). 41
Despite the use of hospital-wide strategies, pathogenic bacteria such as P. 42
aeruginosa and M. tuberculosis often evolve resistance in the course of a single-host 43
treatment (16, 33). Therefore, clinicians also face decisions on how to best administer 44
one or many antibiotics to a single patient (18). For example, antibiotics can be used 45
simultaneously (i.e. combination therapy), or sequentially (i.e. sequential therapy), 46
where two or more different antibiotics are used one after the other. While 47
combination therapy has been used successfully, for example, to treat Helicobacter 48
pylori, the etiological agent of peptic ulcers, combination therapy can be associated 49
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
4
with uncomfortable side effects (13) or its effectiveness maybe limited as a result of 50
poorly-studied drug interactions (41). 51
Unlike hospital-wide cycling, sequential therapy within a single host exposes 52
bacterial infections to a rapid change in antibiotics. While the cycling of antibiotic 53
within a hospital system can take months to years to implement, it is possible to 54
switch antibiotics within a single host over a matter of days. Provided that the 55
antibiotics chosen for sequential therapy elicit no cross-resistance, mutants resistant 56
against one antibiotic are unlikely to reach high frequencies within the host before a 57
second antibiotic is applied. Furthermore, pleiotropic fitness costs associated with 58
resistance mutations are believed to affect the persistence of antibiotic-resistant 59
bacteria and to slow the spread of resistance (2). Therefore, a rapid switch in 60
antibiotic use has the potential to minimize multi-drug resistance while mitigating 61
negative clinical consequences of combination therapy. 62
Given that the rate at which resistance mutations are generated varies among 63
antibiotics (22) and that fitness costs associated with resistance mutations vary both 64
within and among individual antibiotics (2, 28), the success of sequential therapy is 65
likely to depend on both the antimicrobial activity of the antibiotics and the order in 66
which they are used. Building from the work presented in Perron et al. [2007], we 67
investigate the importance of antibiotic treatment order in determining the efficacy of 68
multidrug therapy using experimental and theoretical results. First, we compared the 69
effect of combination therapy and sequential therapy on MDR evolution in 70
experimental populations of P. aeruginosa maintained under different mutation 71
supply rate regimes. We show that using two antibiotics one after the other early in 72
the treatment can give results similar to using the two drugs simultaneously; but the 73
effect depends on which antibiotic is used first and on the mutation supply rate of 74
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
5
resistant mutations in the population. Second, we introduce a simple population 75
dynamics model identifying pleiotropic fitness costs associated with resistance 76
mutations as a plausible mechanism responsible for the order effect observed in our 77
experiment. 78
79
MATERIALS AND METHODS 80
Study organism and growth conditions. In this experiment, we cultured a total of 81
seventy-two experimental populations of Pseudomonas aeruginosa strain PAO1 (37). 82
P. aeruginosa is a non-recombinogenic Gram-negative bacterium that is an important 83
opportunistic pathogen of human. Each population was initially inoculated with 84
approximately 104 cells and was incubated at 37 ºC in 150 μL of M9KB media (per 85
litre: 20g proteose peptone #3, 12.8g Na2HPO4⋅7H2O, 10g glycerol, 3g KH2PO4, 0.5g 86
NaCl and 1g NH4Cl). Every 24 hours, each culture was mixed thoroughly by pipetting 87
50 μL in and out repeatedly before 1% of the overnight culture (approximately 106 88
bacterial cells) was transferred to a fresh microcosm. Growth was monitored daily for 89
ten days and was measured as optical density (OD600) using a universal microplate 90
reader (BioTek, Winooski, VT). Three antibiotic treatments were implemented as 91
described below. 92
93
Mutation supply rate. We looked at the effect of increasing the mutation supply rate 94
by manipulating the immigration rate of susceptible bacteria sampled from an 95
isogenic population of P. aeruginosa grown overnight. For each antibiotic treatment, 96
we grew twelve replicate populations with of the following mutation supply rates: no 97
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
6
immigration, 0.1% (approx. 1000 cells), 1.0% (approx. 10,000 cells) and 10% 98
(approx. 100,000 cells) immigration. 99
100
Antibiotic treatments. The antibiotics rifampicin and streptomycin were used in this 101
experiment. Rifampicin acts by lodging itself into the DNA/RNA tunnel of the 102
polymerase, and sterically blocking elongation of nascent mRNA molecules (12). 103
Streptomycin targets the S16 rRNA protein of the 30S ribosomal subunit, interfering 104
with the binding of tRNA, and therefore initiation of protein synthesis (36). Previous 105
studies of rifampicin- and streptomycin-resistant mutants of P. aeruginosa showed 106
that cross-resistance between the two drugs is unlikely in similar experimental 107
conditions (28, 39). 108
The antibiotics were supplemented to the media as follows: rifampicin (62.5 109
μg⋅ml-1) and streptomycin (16 μg⋅ml-1). These concentrations of antibiotic were 110
shown to completely inhibit the growth of P. aeruginosa in the absence of 111
immigration (28). Also, we established that rifampicin was more effective than 112
streptomycin to inhibit the growth of P. aeruginosa and that resistance to rifampicin 113
incurred a higher fitness cost (28). For each mutation supply rate treatment, three 114
different combinations of the antibiotics were used to treat populations of P. 115
aeruginosa: 1) the combination of streptomycin and rifampicin; 2) sequential therapy 116
with rifampicin used first; and 3) sequential therapy with streptomycin used first. 117
During sequential therapy, antibiotics were switched at every transfer. 118
119
Pleiotropic fitness costs of resistance. At the end of the experiment, the populations 120
were grown for three transfers in unsupplemented KB to test the heritability of 121
resistance and to estimate the pleiotropic cost of resistance of each population. The 122
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
7
cost of resistance was measured as the optical density (OD600) in unsupplemented KB 123
and was compared to that of PAO1 strains cultivated in parallel for the length of the 124
experiment to control for adaptation to normal laboratory conditions. 125
126
Statistical analyses. To analyze the evolution of antibiotic resistance during the 127
course of the experiment, we modeled the temporal dynamics of bacterial growth 128
using a hierarchical linear mixed model (lme function of the nlme package of the R 129
2.11.1 software), sometimes referred to repeated measure anova. We used the optical 130
density data (OD600) as the response. Time (number of transfers following the 131
beginning of the experiment) was considered as a random variable whilst antibiotic 132
treatment (3 levels) and the rate of immigration (4 levels) were considered fixed 133
effects. We also accounted for the non-linear growth dynamics over the bacteria 134
populations by computing the quadratic term for time. Because all replicates were 135
started under similar conditions we constrained the model to a unique intercept. 136
Replicates were taken to be random effects and were nested within treatments. We 137
began by fitting the full model that included all fixed effects and their interactions, 138
and then simplified it by sequential backward selection. We used an F-test to compare 139
the fit of different models. A variance function (varIdent of nlme library) that permits 140
different variances for each level of a stratification variable (here treatment) was used 141
to model heteroscedasticity when necessary. We also used the corAR1 function to 142
model the autocorrelation structure in the time series. Significance of fixed effects 143
was tested with F-tests. Differences between treatments were tested with pairwise 144
comparisons using log likelihood ratio tests. Model parameters and confidence 145
intervals were estimated with restricted maximum likelihood methods (31). 146
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
8
We also modeled the effect of antibiotic treatments and mutation supply rate 147
on the average growth (OD600) of all selection lines over the length of the experiment. 148
The model was fit in a Linear General Model using antibiotic treatments (3 levels) as 149
a fixed factor and mutation supply rate as continuous factor (this to control for the 150
non-normality of the data). Finally, we looked at the effect of antibiotic treatment (3 151
levels) and mutation supply rate (4 levels) on pleiotropic fitness costs of resistance 152
fitting the carrying capacity of each population in the absence of antibiotic as a 153
dependent variable. All analyses and model assumptions were performed and verified 154
using R 2.10.1 software (http://www.r-project.org). 155
156
RESULTS 157
Experimental evolution of multidrug-resistance under multidrug therapy 158
We investigated the efficacy of multidrug therapy and sequential therapy by 159
treating experimental populations of the bacterium P. aeruginosa with one of 160
following three treatments: 1) the combination of streptomycin and rifampicin; 161
sequential therapy of rifampicin and streptomycin with 2) rifampicin used as the first 162
antibiotic; and 3) streptomycin used as the first antibiotic. We observe that the three 163
treatments have a significantly different effect on MDR evolution and that this effect 164
depends on the mutation supply rate of the treated population (treatment × 165
immigration × time × time: F(6, 624) = 5.266; P < 0.0001; Figure 1). Under normal 166
growth conditions, we found that MDR evolved only in populations first exposed to 167
streptomycin during sequential therapy; all population treated with both antibiotics 168
went extinct after 24 hours while populations treated with rifampicin first went extinct 169
after being subsequently exposed to streptomycin (Figure 1a). Given that P. 170
aeruginosa can evolve resistance to both rifampicin and streptomycin when used 171
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
9
individually (28), our results confirm that, given a certain order, sequential therapy 172
can significantly decrease resistance evolution. 173
Bacterial populations rarely evolve in isolation (33). In previous studies, we 174
showed that immigration of susceptible bacteria cells from a larger source population 175
could favor adaptation to antibiotic environments by increasing the supply of 176
resistance mutations in local populations of bacteria (27-29). For this reason, we 177
simulated the effect of immigration by supplementing each experimental population 178
treated with antibiotics with susceptible bacterial cells at each transfer. As predicted, 179
immigration had a positive effect on MDR evolution (immigration: F(3, 60) = 44.306; P 180
< 0.0001): while small amounts of immigration (i.e. 1000 susceptible cells per 181
transfer) favored MRD evolution under sequential therapy (Figure 1b), higher level of 182
immigration (i.e. 100,000 cells per transfer), allowed MDR evolution in all treatments 183
(Figure 1c). 184
185
Pleiotropic fitness costs of resistance 186
To assess whether the type of multidrug therapy used impacted the pleiotropic 187
fitness costs associated with MDR, we measured the growth of each MDR population 188
in the absence of antibiotics. We found that the choice of therapy and the mutation 189
supply rate regime significantly affected the evolution of costs of resistance 190
(treatment × immigration: F(1,66) = 24.578; P < 0.001; Figure 2). While fitness cost of 191
resistance decreased as immigration rate increased, fitness costs were generally higher 192
under combination therapy than under both sequential treatments. At lower 193
immigration rates, costs of resistance were generally lower in sequential therapy 194
initiated with streptomycin than in sequential therapy initiated with rifampicin. As 195
immigration increased, both sequential treatments generated little or no cost of 196
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
10
resistance. These results indicate that the combination and the order in which two 197
antibiotics are used over the course of a treatment plays an important role in the 198
evolution of resistance costs in addition to resistance evolution. 199
200
Modeling the antibiotic order effect in sequential therapy 201
To understand the effect of antibiotic treatment order on the evolution of 202
resistance under sequential therapy, we considered a simple population dynamics 203
model of the evolution of MDR in a bacterial population that is sequentially exposed 204
to two antibiotics. Because most extinction events in our experiment happened within 205
the first two antibiotic switches, we focused on the population dynamic involved in 206
the first two transfers. In this model, we assume that bacteria acquire resistance 207
mutations against the first and the second antibiotics with rates μ1 and μ2 respectively 208
and that an antibiotic-resistant allele confers a cost of resistance in the absence of that 209
antibiotic. In particular, mutations that confer resistance against the first antibiotic are 210
deleterious with selection coefficient s1 in the absence of this antibiotic, while the 211
fitness cost of mutation conferring resistance to the second antibiotic is s2. We also 212
assume that si >> μj, i,j = 1,2, which reflects our knowledge of typical mutation rates 213
and costs of resistance (22). 214
In the first phase of our experiment, the population grows in the absence of 215
antibiotics and reaches population size N0. We assume that the fraction of mutants 216
resistant to the first antibiotic by the end of this phase reaches the mutation-selection 217
balance, μ1/s1, and that there are no MDR mutants. Thus, by the time the first 218
antibiotic treatment is applied, there are on average N0μ1/s1 bacterial cells that harbor 219
resistance to the first antibiotic. Starting with this (relatively small) number of 220
resistant cells, the population then grows for time T in the presence of the first 221
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
11
antibiotic, until the second antibiotic is applied and the first antibiotic is withdrawn. 222
We consider that MDR evolves in our experiment if on average there is at least one 223
cell harboring resistance to both antibiotics at the time the second antibiotic is applied. 224
Mathematically, if N12(t) is the expected number of MDR cells at time t during the 225
second growth phase, the condition for MDR evolution is 226
(1) N12 (T ) ≥1. 227
To express condition (1) in terms of evolutionary parameters μ1, μ2, s1 and s2, 228
we assume that in the presence of the first antibiotic the population stays below its 229
carrying capacity and grows exponentially for time T. If N1(t) is the expected number 230
of bacteria resistant against the first antibiotic but not against the second antibiotic at 231
time t, the average population dynamics is described by differential equations 232
(2) N1 = r1N1 − μ2N1
N 12 = (r1 − s2 )N12 + μ2N1
233
with the initial condition at the beginning of the second growth phase 234
(3) N1(0) = N0μ1 / s1
N12 (0) = 0 , 235
where r1 is the intrinsic growth rate of single-drug resistant bacteria in the presence of 236
the first antibiotic. The solution to system (2), (3) is 237
N1(t) = N0μ1 / s1( )e(r1−μ2 )t
N12 (t) =N0μ1 / s1( )μ2
μ2 − s2
e(r1−s2 )t − e(r1−μ2 )t( ). 238
Condition (1) then becomes 239
N0μ1μ2e
(r1−μ2 )T
s1
1− e−(s2−μ2 )T
s2 − μ2
≥1, 240
which we can rewrite as 241
(4) )( 21 sAfs ≤ , 242
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
12
where A = N0μ1μ2e(r1 −μ 2 )TT and f (s) =
1− e−(s−μ 2 )T
(s − μ 2)T. Notice that function f(s) is 243
positive (because, as mentioned above, in practice s2>>μ2) and is monotonically 244
decreasing. 245
We now ask which antibiotic should be applied first, given that the costs of 246
resistance to each are different. Consider two antibiotics, X and Y, for which 247
resistance incurs the costs sX and sY, respectively. Without loss of generality, we can 248
assume that sX > sY . We then say that antibiotic X is in this sense “strong” and 249
antibiotic Y is “weak”. From equation (4), the condition of MDR evolution in case X 250
is applied before Y (“strong-weak” treatment), is 251
(5) ( )YX sAfs ≤ , 252
and, vice versa, in case Y is applied before X (“weak-strong” treatment), MDR 253
evolves if 254
(6) )( XY sAfs ≤ . 255
As illustrated in Figure 3, inequalities (5) and (6), divide the (sX, sY) semi-256
plane into three regions. As we mentioned above, we only consider the area under the 257
diagonal (where sX > sY is respected). In region I, both conditions (5) and (6) are 258
satisfied implying that MDR is expected to evolve regardless of the order in which 259
antibiotics are applied. In region II, condition (6) is satisfied but condition (5) is not, 260
which implies that MDR is expected to evolve only under the weak-strong treatment. 261
Finally, in region III, both conditions (5) and (6) are violated, and MDR is not 262
expected to evolve under either treatment. In this analysis, the parameter region where 263
MDR evolves under the weak-strong treatment (regions I and II) is strictly larger than 264
the parameter region where MDR evolves under the strong-weak treatment (region I). 265
In other words, our model predicts that MDR is less likely to evolve under sequential 266
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
13
therapy when the cost of resistance to the first antibiotic used is higher than that of the 267
second antibiotic used. 268
Predictions of our simple model are consistent with our experimental 269
observations. As shown previously in Perron et al. (2007), resistance mutations 270
against rifampicin generally confer a larger cost on fitness than resistance mutations 271
against streptomycin. In the above terminology, rifampicin is the strong antibiotic and 272
streptomycin is the weak antibiotic. The fact that the MDR evolves under the weak-273
strong treatment (streptomycin first) but not under the strong-weak treatment 274
(rifampicin first) suggests that the costs of resistance against these antibiotics fall into 275
region II of the parameter space of our experiment (Figure 3). 276
277
DISCUSSION 278
In this study, we show that the order in which two antibiotics are used during 279
sequential therapy of a single bacterial “infection” has a significant effect on MDR 280
evolution and its fitness costs. When we grew our experimental populations in 281
isolation, without the influx in mutation provided by the immigration of cells from a 282
source population, the use of rifampicin before streptomycin precluded the evolution 283
of MDR just as successfully as the combination of the two antibiotics. Using a simple 284
population dynamics model, we demonstrate that the fitness cost associated with 285
mutations conferring resistance to the first antibiotic predominantly determines this 286
order effect. Specifically, the chance of MDR evolution is reduced by first using the 287
antibiotic for which resistant mutation confers the highest fitness cost. Our results 288
demonstrate that costs of resistance play an important role in reducing the frequency 289
of resistant mutants in bacteria populations and therefore minimizing the rate of MDR 290
evolution. Given that our model captures the fundamental population parameters 291
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
14
governing resistance evolution, we expect that our results are not be limited to the 292
specific antibiotics used, but instead may be general to many antibiotic combinations. 293
Our model provides a simple explanation for why antibiotics’ order in 294
sequential treatment can affect the evolution of resistance. Because of severe 295
competition in large bacterial populations within the host prior to the application of 296
any antibiotics, the frequency of resistant mutants is limited by the rate at which they 297
arise and the pleiotropic fitness costs they carry. Resistance mutations that arise at a 298
low probability and resistance mutations that incur a large cost on fitness will be less 299
frequent in a population that is not treated with antibiotics; a phenomenon known as 300
the mutation-selection balance (7). Importantly, at the moment the first antibiotic is 301
applied, (a) the frequency of a resistant mutant is inversely proportional to the cost of 302
resistance that it carries and (b) the probability that this mutant spreads in the 303
population after the antibiotic is applied is proportional to its initial frequency. This 304
implies that the cost of resistance against the first antibiotic directly modulates 305
bacterial population density during the antibiotic-free phase and, consequently, the 306
probability of MDR evolution. The application of the first antibiotic dramatically 307
reduces population density and, thus, competition between bacterial cells. With little 308
competition, the rate of resistance evolution against the second antibiotic is limited 309
primarily by the rate at which such resistance mutations arise; the cost of resistance 310
against the second antibiotic plays a relatively small role. Mathematically, this is 311
reflected in that the shape of function f(s2) in equation (4): it depends very weakly on 312
s2. Therefore, if resistance mutations against the two antibiotics incur different costs, 313
the order in which the antibiotics are applied critically determines the probability of 314
MDR evolution. 315
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
15
Although mutation supply rate is important factors in the evolution of MDR, it 316
does not contribute to the effect of antibiotic treatment order in our model. This result 317
means that changes in mutation rates over time within a population, such as 318
hypertmutability arising (29) or one of the drugs inducing the bacterium’s SOS 319
response (32), will not affect our conclusions. The model provides a striking match 320
with our experimental results: MDR evolve under sequential therapy when 321
streptomycin was first applied but did not when rifampicin was the first antibiotic 322
used. Since resistance to rifampicin generally confer a larger cost on fitness than 323
resistance to streptomycin (28), the order effect observed in this study is consistent 324
with our theoretical model and is likely to be independent of the target and the 325
function of the antibiotics. 326
Two important assumptions reflecting the settings of our experimental work 327
are crucial for our theoretical results to hold. First, the time scale between the use of 328
the two antibiotics determines the frequency of resistance mutations found at each 329
phase. While, the pre-antibiotic phase must be long enough for the population to reach 330
the mutation-selection balance with respect for resistance mutations, the population 331
size post-treatment must remain low relative to carrying capacity; otherwise the 332
resistance cost against the second antibiotic would become important. Although it 333
can be difficult to control the exact concentration of antibiotics that reach the site of 334
infection (3), it is relatively simple to switch between different antibiotics over a short 335
period of time when treating individual patients. 336
Second, we make the assumption that there is no cross-resistance, no 337
recombination between the resistance mutations, no possibility of compensatory 338
mutations, and no epistasis between resistance mutations. Cross-resistance between 339
the two antibiotics would cancel the order effect since evolution of resistance to one 340
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
16
antibiotic would confer resistance to the second antibiotic. With the accumulation of 341
data on resistance and cross-resistance for an increasing number of antibiotics, 342
however, it should be possible for clinicians to avoid the use of two drugs for which 343
cross-resistance is known. Horizontal gene transfer can bring together resistance 344
genes that evolved in different bacteria lineages (23) and can therefore promote MDR 345
evolution (30). The compensation of fitness costs associated with resistance could 346
also contribute to increasing the frequency of antibiotic-resistant mutants in natural 347
populations of bacteria by alleviating the competitive disadvantage of resistant 348
mutants (29, 35). Finally, epistatic interactions among different resistance mutations 349
could either promote or inhibit the evolution of MDR (38). 350
Given the availability of minimal information pertaining to fitness costs 351
associated with resistance to specific antibiotics, our results provide a simple 352
guideline for sequential multidrug-therapy: one ought to use antibiotic incurring the 353
stronger cost of resistance as a first line-antibiotic in order to minimize the change of 354
MDR evolution. Because of the different pharmacodynamics characteristics 355
associated with different antibiotics and the complex nature of bacteria infections in 356
the human body, it is practically impossible to accurately model the evolution of 357
antibiotic resistance within a human host. Still, our simple model captures the key 358
aspects of such evolution and makes a strong qualitative prediction: MDR is less 359
likely to evolve if the antibiotic that incurs larger resistance costs is applied first. 360
Although costs of resistance to specific antibiotics can change substantially between 361
bacterial species and between different environments (2, 27, 28), it usually follows 362
predictable distributions readily identified in laboratory experiments (22). 363
As the use of multidrug therapy is increasingly prevalent in the treatment of 364
healthcare-associated infections such as P. aeruginosa (21), it is crucial to understand 365
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
17
the full evolutionary consequences associated with drug deployments. Our results not 366
only demonstrate the influence cost of resistance has on the evolution of MDR, but 367
suggest a simple approach that could potentially improve multidrug-therapy. 368
369
370
ACKNOWLEDGMENTS 371
The authors would like to thank Sam P. Brown, R. Fredrick Inglis, and Luiz-Miguel 372
Chevin for their comments on earlier versions of this manuscript. This work was 373
funded by the European Research Council (ERC) and the Leverhulme Trust. G.G.P. 374
was funded during this work by the Clarendon Funds of the University of Oxford and 375
the National Science Engineering and Research Council of Canada (NSERC) and is 376
now funded by the Fond Québécois pour la Recherche sur la Nature et les 377
Technologies (FQRNT). The funders had no role in study design, data collection and 378
analysis, decision to publish, or preparation of the manuscript. 379
380
381
382
383
384
385
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
18
REFERENCES 386
1. Albert, T. J., D. Dailidiene, G. Dailide, J. E. Norton, A. Kalia, T. A. 387 Richmond, M. Molla, J. Singh, R. D. Green, and D. E. Berg. 2005. 388 Mutation discovery in bacterial genomes: metronidazole resistance in 389 Helicobacter pylori. Nat Methods 2:951-953. 390
2. Andersson, D. I., and D. Hughes. 2010. Antibiotic resistance and its cost: is 391 it possible to reverse resistance? Nat Rev Microbiol 8:260-271. 392
3. Andes, D., J. Anon, M. R. Jacobs, and W. A. Craig. 2004. Application of 393 pharmacokinetics and pharmacodynamics to antimicrobial therapy of 394 respiratory tract infections. Clinics in Laboratory Medicine 24:477-502. 395
4. Bal, A. M., A. Kumar, and I. M. Gould. 2010. Antibiotic heterogeneity: 396 from concept to practice, p. 81-91, Antimicrobial Therapeutics Reviews, vol. 397 1213. Wiley-Blackwell, Malden. 398
5. Beardmore, R. E., and R. Peña-Miller. 2010. Antibiotic cycling versus 399 mixing: The difficulties of using mathematical models to definitively quantify 400 their relative merits. Mathematical Biosciences and Engineering 7:923-933. 401
6. Beardmore, R. E., and R. Peña-Miller. 2010. Rotating antibiotics selects 402 optimally against antibiotic resistance, in theory. Mathematical Biosciences 403 and Engineering 7:26. 404
7. Bell, G. 2008. Selection: the mechanism of evolution. Oxford University 405 Press, Oxford, UK. 406
8. Bennett, K. M., J. E. Scarborough, M. Sharpe, E. Dodds-Ashley, K. S. 407 Kaye, T. Z. Hayward, and S. N. Vaslef. 2007. Implementation of antibiotic 408 rotation protocol improves antibiotic susceptibility profile in a surgical 409 intensive care unit. Journal of Trauma - Injury, Infection and Critical Care 410 63:307-311. 411
9. Bergstrom, C. T., M. Lo, and M. Lipsitch. 2004. Ecological theory suggests 412 that antimicrobial cycling will not reduce antimicrobial resistance in hospitals. 413 Proceedings of the National Academy of Sciences of the United States of 414 America 101:13285-13290. 415
10. Bhusal, Y., C. M. Shiohira, and N. Yamane. 2005. Determination of in vitro 416 synergy when three antimicrobial agents are combined against Mycobacterium 417 tuberculosis. Int J Antimicrob Agents 26:292-297. 418
11. Bonhoeffer, S., P. A. zur Wiesch, and R. D. Kouyos. 2010. Rotating 419 antibiotics does not minimize selection for resistance. Mathematical 420 Biosciences and Engineering 7:919-922. 421
12. Campbell, E. A., N. Korzheva, A. Mustaev, K. Murakami, S. Nair, A. 422 Goldfarb, and S. A. Darst. 2001. Structural mechanism for rifampicin 423 inhibition of bacterial RNA polymerase. Cell 104:901-912. 424
13. Chiba, N., B. V. Rao, J. W. Rademaker, and R. H. Hunt. 1992. 425 METAANALYSIS OF THE EFFICACY OF ANTIBIOTIC-THERAPY IN 426 ERADICATING HELICOBACTER-PYLORI. Am. J. Gastroenterol. 427 87:1716-1727. 428
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
19
14. Daniels, M., and A. B. Hill. 1952. Chemotherapy of pulmonary tuberculosis 429 in young adults; an analysis of the combined results of three Medical Research 430 Council trials. Br Med J 1:1162-1168. 431
15. Eber, M. R., R. Laxminarayan, E. N. Perencevich, and A. Malani. 2010. 432 Clinical and economic outcomes attributable to health care-associated sepsis 433 and pneumonia. Arch Intern Med 170:347-353. 434
16. Gagneux, S., C. D. Long, P. M. Small, T. Van, G. K. Schoolnik, and B. J. 435 Bohannan. 2006. The competitive cost of antibiotic resistance in 436 Mycobacterium tuberculosis. Science 312:1944-1946. 437
17. Gould, I. M., and J. W. M. van der Meer (ed.). 2007. Antibiotic Policies: 438 Fighting Resistance, 1st ed. Springer Science + Business Media, LLC., New 439 York, U.S.A. 440
18. Hauser, A. R. 2007. Antibiotics basics for clinicians: choosing the right 441 antibacterial agent. Lippincott Williams & Wilkins, Baltimore, MD, USA. 442
19. Kerr, K. G., and A. M. Snelling. 2009. Pseudomonas aeruginosa: a 443 formidable and ever-present adversary. Journal of Hospital Infection 73:338-444 344. 445
20. Kouyos, R. D., P. A. zur Wiesch, and S. Bonhoeffer. 2011. Informed 446 Switching Strongly Decreases the Prevalence of Antibiotic Resistance in 447 Hospital Wards. Plos Computational Biology 7. 448
21. Levy, S. B., and B. Marshall. 2004. Antibacterial resistance worldwide: 449 Causes, challenges and responses. Nature Medicine 10. 450
22. MacLean, R. C., A. R. Hall, G. G. Perron, and A. Buckling. 2010. The 451 population genetics of antibiotic resistance: integrating molecular mechanisms 452 and treatment contexts. Nat Rev Genet 11:405-414. 453
23. Ochman, H., J. G. Lawrence, and E. A. Grolsman. 2000. Lateral gene 454 transfer and the nature of bacterial innovation. Nature 405:299-304. 455
24. Palumbi, S. R. 2001. Humans as the world's greatest evolutionary force. 456 Science 293:1786-1790. 457
25. Pearson, A. 2009. Historical and changing epidemiology of healthcare-458 associated infections. Journal of Hospital Infection 73:296-304. 459
26. Perron, G. G., G. Bell, and S. Quessy. 2008. Parallel evolution of multidrug-460 resistance in Salmonella enterica isolated from swine. FEMS Microbiol Lett 461 281:17-22. 462
27. Perron, G. G., A. Gonzalez, and A. Buckling. 2008. The rate of 463 environmental change drives adaptation to an antibiotic sink. J Evol Biol 464 21:1724-1731. 465
28. Perron, G. G., A. Gonzalez, and A. Buckling. 2007. Source-sink dynamics 466 shape the evolution of antibiotic resistance and its pleiotropic fitness cost. Proc 467 Biol Sci 274:2351-2356. 468
29. Perron, G. G., A. R. Hall, and A. Buckling. 2010. Hypermutability and 469 compensatory adaptation in antiboitic-resistant bacteria. The American 470 Naturalist 176:303-3011. 471
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
20
30. Perron, G. G., A. E. Lee, Y. Wang, W. E. Huang, and T. G. Barraclough. 472 2012. Bacterial recombination promotes the evolution of multi-drug-resistance 473 in functionally diverse populations. Proc Biol Sci 279:1477-1484. 474
31. Pinheiro, J. C., and D. M. Bates. 2000. Mixed-effects models in S and S-475 PLUS. Springer, New York, NY, USA. 476
32. Radman, M. 1975. SOS repair hypothesis: phenomenology of an inducible 477 DNA repair which is accompanied by mutagenesis. Basic Life Sci 5A:355-478 367. 479
33. Reinhardt, A., T. Kohler, P. Wood, P. Rohner, J. L. Dumas, B. Ricou, and 480 C. van Delden. 2007. Development and persistence of antimicrobial 481 resistance in Pseudomonas aeruginosa: a longitudinal observation in 482 mechanically ventilated patients. Antimicrob Agents Chemother 51:1341-483 1350. 484
34. Sandiumenge, A., E. Diaz, A. Rodriguez, L. Vidaur, L. Canadell, M. 485 Olona, M. Rue, and J. Rello. 2006. Impact of diversity of antibiotic use on 486 the development of antimicrobial resistance. J Antimicrob Chemother 487 57:1197-1204. 488
35. Schrag, S. J., V. Perrot, and B. R. Levin. 1997. Adaptation to the fitness 489 costs of antibiotic resistance in Escherichia coli. Proceedings of the Royal 490 Society - Biological Sciences (Series B) 264:1287-1291. 491
36. Sharma, D., A. R. Cukras, E. J. Rogers, D. R. Southworth, and R. Green. 492 2007. Mutational Analysis of S12 Protein and Implications for the Accuracy 493 of Decoding by the Ribosome. Journal of Molecular Biology 374:1065-1076. 494
37. Stover, C. K., X. Q. Pham, A. L. Erwin, S. D. Mizoguchi, P. Warrener, M. 495 J. Hickey, F. S. L. Brinkman, W. O. Hufnagle, D. J. Kowallk, M. Lagrou, 496 R. L. Garber, L. Goltry, E. Tolentino, S. Westbrock-Wadman, Y. Yuan, 497 L. L. Brody, S. N. Coulter, K. R. Folger, A. Kas, K. Larbig, R. Lim, K. 498 Smith, D. Spencer, G. K. S. Wong, Z. Wu, I. T. Paulsen, J. Relzer, M. H. 499 Saler, R. E. W. Hancock, S. Lory, and M. V. Olson. 2000. Complete 500 genome sequence of Pseudomonas aeruginosa PAO1, an opportunistic 501 pathogen. Nature 406:959-964. 502
38. Trindade, S., A. Sousa, K. B. Xavier, F. Dionisio, M. G. Ferreira, and I. 503 Gordo. 2009. Positive Epistasis Drives the Acquisition of Multidrug 504 Resistance. PLoS Genet 5:e1000578. 505
39. Ward, H., G. G. Perron, and R. C. Maclean. 2009. The cost of multiple 506 drug resistance in Pseudomonas aeruginosa. J Evol Biol. 507
40. Warren, D. K., H. A. Hill, L. R. Merz, M. H. Kollef, M. K. Hayden, V. J. 508 Fraser, and S. K. Fridkin. 2004. Cycling empirical antimicrobial agents to 509 prevent emergence of antimicrobial-resistant Gram-negative bacteria among 510 intensive care unit patients. Critical Care Medicine 32:2450-2456. 511
41. Yeh, P., A. I. Tschumi, and R. Kishony. 2006. Functional classification of 512 drugs by properties of their pairwise interactions. Nat Genet 38:489-494. 513
42. zur Wiesch, P. A., R. Kouyos, J. Engelstädter, R. R. Regoes, and S. 514 Bonhoeffer. 2011. Population biological principles of drug-resistance 515 evolution in infectious diseases. The Lancet Infectious Diseases 11:236-247. 516
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
21
FIGURE LEGENDS 517
518
Figure 1. Evolution of multidrug-resistance in experimental populations of 519
Pseudomonas aeruginosa treated with multidrug therapy. Multidrug treatments 520
were as follows: rifampicin and streptomycin simultaneously (filled circles); 521
sequential therapy with exposure to streptomycin first (filled triangles); and 522
sequential therapy with exposure to rifampicin first (filled squares). For sequential 523
therapy, the antibiotics were switched on a daily basis. Populations were grown in the 524
absence of immigration, or in 1% b) and 10% c) immigration. Resistance was 525
measured as average density (OD600) of the selection lines at the end of each transfer. 526
Error bars represent standard error of the mean. 527
528
Figure 2. Pleiotropic fitness costs of experimental populations of Pseudomonas 529
aeruginosa that have evolved resistance under three multidrug treatments and four 530
immigration rates. Drug treatments are as follow: rifampicin and streptomycin 531
simultaneously (filled circle); sequential therapy with exposure to streptomycin first 532
(filled triangles); and sequential therapy with exposure to rifampicin first (filled 533
squares). Relative fitness is measured as the growth (OD600) of the resistant 534
populations in the absence of drug compared to that of control populations grown in 535
the absence of drug for the duration of the experiment. 536
537
538
539
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
22
Figure 3. Schematic illustration of regions where MDR is expected to evolve 540
under different treatments. Any pair of antibiotics X and Y with costs of resistance 541
sX and sY (such that sX > sY) represents a point in the lower semi-plane of the (sX, sY)-542
plane. The region to the left of the dashed line is the region where MDR is expected to 543
evolve under the strong-weak treatment (i.e., when antibiotic X is applied first). The 544
region below the solid line is the region where MDR is expected to evolve under the 545
weak-strong treatment (i.e., when antibiotic Y is applied first). The parameter region 546
where MDR evolves under the strong-weak treatment (region I) is strictly smaller 547
than the region where MDR evolves under weak-strong treatment (regions I and II). 548
Thus, the strong-weak treatment represents a safer bet against MDR evolution under 549
uncertainty. 550
551
552
553
554
555
556
557
558
559
560
561
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
23
Figure 1 562
A) 563
564
565
566
567
568
569
B) 570
571
572
573
574
575
576
577
C) 578
579
580
581
582
583
584
585
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
24
Figure 2
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Sequential Therapy and Evolution of Drug Resistance
25
Figure 3 607
608
609
on January 21, 2019 by guesthttp://aem
.asm.org/
Dow
nloaded from
Multidrug Therapy and Evolution of Antibiotic Resistance: WhenOrder Matters
Gabriel G. Perron, Sergey Kryazhimskiy, Daniel P. Rice, Angus Buckling
FAS Center for Systems Biology and Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, Massachusetts, USA; Department of Zoology,University of Oxford, Oxford, United Kingdom; Biosciences, University of Exeter, Tremough, Cornwall, United Kingdom
Volume 78, no. 17, p. 6137– 6142, 2012. The authors retract this paper because of unsatisfactory referencing of previously publishedexperimental data. Specifically, all data in Fig. 1 in this paper were presented in Fig. 1 in reference 28 in the original list of references(G. G. Perron, A. Gonzalez, and A. Buckling, Proc. Biol. Sci. 274:2351–2356, 2007), and all data points except for zero immigration ratein Fig. 2 were presented in Fig. 2 in the same reference. The authors maintain that all results and conclusions of the paper are correct.Sergey Kryazhimskiy and Daniel P. Rice were not involved in the reanalysis of previously published data.
Copyright © 2013, American Society for Microbiology. All Rights Reserved.
doi:10.1128/AEM.02554-13
RETRACTION
October 2013 Volume 79 Number 20 Applied and Environmental Microbiology p. 6521 aem.asm.org 6521