Talk given at the R0 conference in 2008.
Based on the article
http://dx.doi.org/10.1371/journal.pcbi.1000337
Evolutionary Epidemiology of Drug-Resistance in Space
Debarre, Lenormand and Gandon, 2009, PLOS Computational Biology
Harry Coumnas Thinks That Human Teleportation is Possible in Quantum Mechanic...
Spatial Evolutionary Epidemiology of Drug Resistance
1. Evolutionary epidemiology of drug resistance
A spatial model
Florence D´ebarre, Thomas Lenormand, Sylvain Gandon
CNRS UMR 5175, CEFE, Montpellier, France
October 30, 2008
3. Introduction Models Results Conclusion Supp
Introduction
Antimicrobial drugs and
life expectancy
But emergence of drug
resistance
Penicillin consumption in 2000
%ofdrug-resistantSpneumoniae
Goossens (2005)
4. Introduction Models Results Conclusion Supp
Introduction
Antimicrobial drugs and
life expectancy
But emergence of drug
resistance
Penicillin consumption in 2000
%ofdrug-resistantSpneumoniae
Goossens (2005)
5. Introduction Models Results Conclusion Supp
Introduction
Antimicrobial drugs and
life expectancy
But emergence of drug
resistance
Treatment is not applied
uniformly . . .
Penicillin consumption in 2000
%ofdrug-resistantSpneumoniae
Goossens (2005)
6. Introduction Models Results Conclusion Supp
Introduction
Questions:
What is the impact of the heterogeneity of treatment?
7. Introduction Models Results Conclusion Supp
Introduction
Questions:
What is the impact of the heterogeneity of treatment?
on the persistence of drug-sensitive parasites (WT)
8. Introduction Models Results Conclusion Supp
Introduction
Questions:
What is the impact of the heterogeneity of treatment?
on the persistence of drug-sensitive parasites (WT)
on the spread of drug-resistance
9. Introduction Models Results Conclusion Supp
Introduction
Questions:
What is the impact of the heterogeneity of treatment?
on the persistence of drug-sensitive parasites (WT)
on the spread of drug-resistance
Which spatial pattern best prevents the propagation of drug
resistant parasites?
10. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
11. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
β transmission
12. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
β transmission
γ recovery
13. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
β transmission
γ recovery
In a well-mixed population:
R0 =
Nβ
γ
14. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
β transmission
γ recovery
In a well-mixed population:
R0 =
Nβ
γ
Heterogeneous environment
TREATED UNTREATED
x
0
-qT L (1-qT) L
L
qT proportion of treated areas
L environmental grain
15. Introduction Models Results Conclusion Supp
Model: local changes
Epidemiology
S
IWT
Drug-sensitive
IR
Drug-resistant
Susceptible
individuals
Infected
individuals
β transmission
γ recovery
In a well-mixed population:
R0 =
Nβ
γ
Heterogeneous environment
TREATED UNTREATED
x
0
-qT L (1-qT) L
L
TREATED
x
UNTREATED UNTREATED
L
qT proportion of treated areas
L environmental grain
16. Introduction Models Results Conclusion Supp
Treatment and Resistance
Effects on parameters
UNTREATEDTREATED
Treatment's
effect
x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
WT,T
WT: drug-sensitive
R: drug-resistant
17. Introduction Models Results Conclusion Supp
Treatment and Resistance
Effects on parameters
Cost of
resistance
R
UNTREATEDTREATED
R
Treatment's
effect
x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
R
R0
WT,T
WT: drug-sensitive
R: drug-resistant
24. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Sasaki (2004)
Trivial when RWT,T
0 > 1 (always
persists)
UNTREATEDTREATED x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
WT,T
1
25. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Sasaki (2004)
Trivial when RWT,T
0 > 1 (always
persists)
When RWT,T
0 < 1 . . .
UNTREATEDTREATED x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
WT,T
1
26. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Sasaki (2004)
Trivial when RWT,T
0 > 1 (always
persists)
When RWT,T
0 < 1 . . .
UNTREATEDTREATED x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
WT,T
1
How to?
Study the stability of the IWT = 0 equilibrium
27. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
28. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
29. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
30. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
31. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
Proportion of treated area qT
Localdensityof
infectedindividuals
prediction
σ = 1
32. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
Proportion of treated area qT
Localdensityof
infectedindividuals
σ = 5
33. Introduction Models Results Conclusion Supp
Eradication of the drug-sensitive strain (WT)
Condition for the eradication
(1 − qT) L
√
2
σ
<
1
γU
WT
R
WT,U
0 − 1
arctan
−
γT
WT R
WT,T
0 − 1
γU
WT
R
WT,U
0 − 1
tanh
qTL
√
2
σ
−γT
WT
RT
0 − 1
Conclusion
Migration and local selection have antagonistic effects
34. Introduction Models Results Conclusion Supp
Evolution of drug-resistance
Suppose now that there are also drug-resistant parasites.
How does the heterogeneity of treatment affect the evolution
of drug-resistance?
35. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
36. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
When RWT,T
0 > 1
R
UNTREATEDTREATED
R
x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
R
R0
WT,T
37. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
When RWT,T
0 > 1
R
UNTREATEDTREATED
R
x
Basic Reproductive
Numbers (R0)
WT
WT
R0
WT,U
R0
R
R0
WT,T
1
38. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
Approximation
Real solution
qT L L
IWT/N
x
39. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
Approximation
Real solution
qT L L
IWT/N
1 - 1/R0
WT,T
1 - 1/R0
WT,U
x
40. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
Approximation
Real solution
qT L L
IWT/N
1 - 1/R0
WT,T
1 - 1/R0
WT,U
x
Low migration approximation
qT L L
IWT/N
1 - 1/R0
WT,T
1 - 1/R0
WT,U
x
41. Introduction Models Results Conclusion Supp
Invasion of the drug-resistant parasites
How to
Study the stability of the equilibrium with the drug-sensitive
strain only
Approximation
Real solution
qT L L
IWT/N
1 - 1/R0
WT,T
1 - 1/R0
WT,U
x
Low migration approximation
qT L L
IWT/N
1 - 1/R0
WT,T
1 - 1/R0
WT,U
x
τ =
1 − 1/RWT,U
0
1 − 1/RWT,T
0
46. Introduction Models Results Conclusion Supp
Conditions for coexistence
Find the criteria for the invasion of drug-sensitive parasites in
a population of drug-resistant parasites (exact)
47. Introduction Models Results Conclusion Supp
Conditions for coexistence
Find the criteria for the invasion of drug-sensitive parasites in
a population of drug-resistant parasites (exact)
Coexistence when reciprocal invasion
48. Introduction Models Results Conclusion Supp
Conditions for coexistence
Find the criteria for the invasion of drug-sensitive parasites in
a population of drug-resistant parasites (exact)
Coexistence when reciprocal invasion
Proportion of treated area qT
ScaledsizeoftheenvironmentL√2/σ
COEXISTENCE
DRUG-RESISTANT
ONLY
DRUG-SENSITIVE
ONLY
Invasion of the drug-
resistant parasites
Invasion of the drug-
sensitive parasitesdecreasingmigration
49. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
50. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
RR
0 =
N βR
γR
= 1.36
51. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
RR
0 =
N βR
γR
= 1.36
Costs on recovery γ
S
IWT
IR
Fast dynamics
52. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
RR
0 =
N βR
γR
= 1.36
Costs on recovery γ
S
IWT
IR
Fast dynamics
Costs on transmission β
S
IWT
IR
Slow dynamics
53. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
RR
0 =
N βR
γR
= 1.36
Costs on recovery γ
qTL L
x
Fast dynamics
Costs on transmission β
S
IWT
IR
Slow dynamics
Total prevalence IWT+IR
N
Proportion of drug-resistance IR
IWT+IR
54. Introduction Models Results Conclusion Supp
Beyond R0
R0 is a ratio . . .
Illustration with the drug-resistant parasites.
RR
0 =
N βR
γR
= 1.36
Costs on recovery γ
qTL L
x
Fast dynamics
Costs on transmission β
qTL L
x
Slow dynamics
Total prevalence IWT+IR
N
Proportion of drug-resistance IR
IWT+IR
55. Introduction Models Results Conclusion Supp
Beyond R0
RR
0 =
N βR
γR
= 1.36
Costs on recovery γ
qTL L
x
Fast dynamics
Costs on transmission β
qTL L
x
Slow dynamics
Conclusion
The R0s do not tell everything; the repartition of the effects between
transmission (β) and recovery (γ) rates also matters !
56. Introduction Models Results Conclusion Supp
Generalization: two-host system
Invasion of drug-resistant parasites
With a two-host system (e.g. vector-borne disease):
Drug-sensitive
Drug-resistant
S
IWT
IR
VWT
Vs
VR
HOST 1 HOST 2
62. Introduction Models Results Conclusion Supp
Take Home Message
Beyond R0: Costs repartition between transmission and
recovery
The local R0 are not enough !
What is the R0 at the whole population scale ?
63. Introduction Models Results Conclusion Supp
Take Home Message
Beyond R0: Costs repartition between transmission and
recovery
The local R0 are not enough !
What is the R0 at the whole population scale ?
Critical width of the treatment area
Migration selection balance
Wiping out the resistant strains
→ cline models in population genetics
64. Introduction Models Results Conclusion Supp
Acknowledgements
UMR 5175
Centre d’´Ecologie Fonctionnelle et
´Evolutive
Montpellier, France
UMR 2724
G´en´etique et ´Evolution des maladies
infectieuses
Montpellier, France
Minus van Baalen, Mark Kirkpatrick, Guillaume Martin, Andrew Park,
Oph´elie Ronce, Fran¸cois Rousset
71. Introduction Models Results Conclusion Supp
Two-host model
Equivalent migration σe
σ2
e =
σ2
V/νR + σ2
H/γR
1/νR + 1/γR
Initial density (WT)
Calculated either in the human (τH) or vector (τV) compartments
72. Introduction Models Results Conclusion Supp
Two-host model
Critical Size
With k =H or k =V:
qTL
√
2
σe
>
1
s
arctan τ2
k α tanh α
√
s
(1 − qT)L
√
2
σe
73. Introduction Models Results Conclusion Supp
Two-host model
Critical Size
With k =H or k =V:
qTL
√
2
σe
>
1
s
arctan τ2
k α tanh α
√
s
(1 − qT)L
√
2
σe
s ≈
RR
0
RWT,T
0
− 1
1
1/γR + 1/νR
α ≈ −
RR
0
RWT,U
0
− 1
RR
0
RWT,T
0
− 1
74. Introduction Models Results Conclusion Supp
Cost and evolution of resistance
γU
WT
βU
WT
Recovery rate γR
TransmissionβR
75. Introduction Models Results Conclusion Supp
Cost and evolution of resistance
γU
WT
βU
WT
Recovery rate γR
TransmissionβR