Fig 1.
Changes in relative serotype frequencies due to vaccination.
Each serotype is divided into vaccine types (VT) or non-vaccine types (NVT). The pre-PCV7 period is composed of three sample years from 1998–2000. The post-PCV7 period is from one sample period in 2009. These two sample periods had a roughly equal number of samples per region. We include 6A as a vaccine type due to the effectiveness of PCV7 at targeting this serotype [17].
Fig 2.
Pre- and post-PCV7 population structure in four US states.
Each node represents a sequence type and each community (shaded patch) represents a serotype. Nodes are coloured according to the era of maximum sampled relative frequency, either pre- or post-PCV7. Communities are coloured according to whether or not the given serotype was a vaccine type (VT) or nonvaccine type (NVT). Edges represent a shared sequence type. Note that we include 6A as a vaccine type due to the effectiveness of PCV7 at targeting this serotype [17].
Fig 3.
Co-colonisation of resistance and sensitivity at the within- and between-host scales.
The dynamics of co-colonisation between drug-resistant and drug-sensitive strains are shown for the four different competition types. (A) Illustration of the system for each of the four competition types (a)–(d). (a) No competition between the two strains; (b) The two strains share a metabolic type; (c) The two strains share a serotype; (d) The two strains share both a metabolic type and an antigenic type. (B) Trajectories show the within-host prevalence of each strain, assuming that the resistant strain invades 10 days following an initial infection with the sensitive strain. Trajectories for each competition type under treatment (left) and under no treatment (right) are shown. (C) Trajectories at the population-level, taking one strain sampled from each host. Simulations use 2,000 hosts and are repeated 72 times. Lines show the median frequency of each strain over all simulations, bands show 95% quantiles. Trajectories for each competition type are shown under low treatment coverage (left) and high treatment coverage (right). Parameters are fixed with c = 0.15, τ = 0.35 and β = 0.1. For each antigenic type, we set α = 0.03.
Fig 4.
Strain structure is preserved for low growth differences among metabolic types.
(A) Strain trajectories illustrating the pre- and post-vaccine dynamics when (a) there are no growth differences between MT1 and MT2; and (b) MT1 has a significant within-host advantage. Solid lines indicate median prevalence, shaded areas indicate 95% quantiles over 70 simulations. (B) Heatmaps summarizing trajectories over a range of parameters: α1 = 0.02, α2 ∈ [0.022, 0.025, 0.03], κ2 = 1.1, κ1 ∈ [1.1, 1.15, 1.2]. (Left) Median frequency of each AT/MT combination pre-vaccination. (Right) The fraction of simulations for which AT2-MT1 was present at the end of the simulation (post-vaccination). Simulations initialized with AT1-MT1 and AT2-MT2 only. Each simulation is run on 7000 hosts over 30 years, with vaccination at year 15. Additional parameters fixed: τ = 0.8, pτ = 0.22. Full trajectories are shown in Fig A4 in S1 Text.
Fig 5.
Pre- and post-vaccination dynamics with multiple NVTs.
We fix the robustness of each antigenic type, with the VT being the most, and NVT2 being the least, robust. Simulations are performed on 7000 hosts over 30 years with vaccination at year 15. Each simulation is repeated 70 times. All strains are drug-susceptible. (A) Change in metabolic composition of each antigenic type following vaccination. (B) Change in relative frequency of both NVTs following vaccination (including the VT), compared to the change in relative frequency among NVTs only. Fixed parameters: αVT = 0.02, αNVT1 = 0.025, αNVT2 = 0.03, κ1 = κ2 = κ3 = κ4 = 1.1, pτ = 0.22, τ = 0.8, β = 0.09, rt = 0.00005.
Fig 6.
Post-vaccine success of resistant variant depends on antigenic properties.
Each panel shows strain trajectories initialized with AT1-MT1-R and AT2-MT2-S, with vaccination against AT1 after 15 years. We only show here simulations in which MT1 was not eliminated following vaccination. The strength of immune response against AT2 is varied: α2 ∈ (0.026, 0.027, 0.028), with fixed α1 = 0.025. There is no difference in growth rates between MT1 and MT2. Resistance-specific parameters are fixed, with the growth cost of resistance c = 0.2, within-host treatment rate τ = 0.9 and population-level treatment coverage of pτ = 0.34. Simulations are on 7000 hosts and repeated 72 times. Lines denotes median value with bands representing 95% quantiles.
Fig 7.
Three AT, three MT system with cross immunity between AT1 (vaccine type) and AT2 (non-vaccine type).
Initialized with AT1-MT1, AT2-MT2 and AT3-MT3. MT1 is sometimes able to persist post-vaccination through switching events in either AT2 or AT3. (A) Relative frequency of each antigenic type (all strains of that type). (B) Absolute frequency of MT1 in each NVT (AT2 and AT3) provided a switching event occurred. Simulations on 10,000 hosts over 70 iterations. Fixed parameters: kg = 0.1, α1 = 0.023, α2 = 0.024, α3 = 0.024, κ1 = κ2 = κ3 = 1.0, pτ = 0.22, τ = 0.8, β = 0.09, rt = 0.00005.
Fig 8.
Pre-PCV7 sampled isolates across four US states.
Each node denotes a distinct lineage (GPSC) within a serotype (labelled, and coloured according to PCV7 status). Edges indicate a shared lineage.
Fig 9.
Pre-PCV7 sampled isolates across four US states.
Each node denotes a distinct sequence type within a serotype (labelled, and coloured according to PCV7 status). Edges indicate a shared sequence type.