Skip to main content

Predicting the Evolution of Antibiotic Resistance in Streptococcus pneumoniae

Periodic Reporting for period 1 - PEARS (Predicting the Evolution of Antibiotic Resistance in Streptococcus pneumoniae)

Reporting period: 2015-06-01 to 2017-05-31

The principal objective of this project is to develop mathematical models to better understand the dynamics of antibiotic-resistant and antibiotic-sensitive strains in the major human pathogen Streptococcus pneumoniae, predict the frequency of resistance as a function of antibiotic usage, and infer key parameters such as the cost of resistance.

The pneumococcus is a human commensal inhabiting the nasopharynx, with a prevalence in children < 5 years old ranging of 30 to 60%. Colonization lasts a week to a few months and is mainly asymptomatic, but occasionally causes infections responsible for the death of about 800, 000 children per year. Multiple genotypes exhibiting resistance to antibiotics have emerged worldwide in past years. Intriguingly, despite strong selection pressure due to antibiotic use, sensitive and resistant strains have coexisted at a stable frequency over the last 15-20 years. This suggests very strong evolutionary forces maintain the stable frequency of resistance.

The project is important for society as it addresses a major public health issue. The emergence of resistance to antibiotics is a very pressing public health issue worldwide and probably one of the greatest challenges that humanity is facing in the 21st century. In Europe alone, the cost of antimicrobial resistance is estimated at 25,000 deaths per year and €1.5 billions. Epidemiological-evolutionary modelling is an adequate framework to understand these phenomena, as it provides a rigorous mechanistic description of the biological system and allows explaining what has happened and forecast what will happen. Mathematical models allow predicting the future evolution of antibiotic resistance depending on the rate at which the antibiotic are prescribed, and predicting the impact of public health interventions, for example reducing antibiotic treatment in specific risk groups.
1. Evolution of antibiotic resistance under seasonally changing antibiotic use:

I developed a model that describes the evolution of antibiotic resistance under selection by multiple antibiotics prescribed at seasonally changing rates. This model was inspired by, and fitted to, published data on monthly antibiotics prescriptions and frequency of resistance in two communities in Israel over 5 years. Seasonal fluctuations in antibiotic usage translate into small fluctuations of the frequency of resistance around the average value. These dynamics were described using a generic model encapsulating all ecological and evolutionary forces. Fitting the model to the data revealed a strong stabilizing force, typically two to five times stronger than direct selection due to antibiotics, explaining that resistance fluctuates in phase with usage. While most antibiotics selected for increased resistance, intriguingly, the cephalosporin class of antibiotic selected for decreased resistance to penicillins and macrolides, an effect consistent in the two communities. One extra monthly prescription of cephalosporins per 1000 children decreased the frequency of penicillin-resistant strains by 1.7%. This model emerges under minimal assumptions, quantifies the forces acting on resistance and explains up to 43% of the temporal variation in resistance.

2. Evolution of antibiotic resistance in a structured host population:

I formulated and analysed a mathematical model describing the epidemiological and evolutionary dynamics of drug resistance in a bacterial species. The host population was structured in several classes, and I delineated analytical conditions under which both the resistant and sensitive strains coexist. The major results are that persistence of the resistant strain is facilitated by invasion of the resistant strain in the niche formed by the treated uncolonised individuals; while coexistence is favoured by the existence of different semi-isolated classes of hosts treated at different rates.

Next, I simulated more complex versions of the mathematical model parameterised with available data from Western countries, describing pneumococcal evolution in a host population structured into age classes (age 0 to 6, as mainly young children are infected by the pneumococcus), and structured in the type of child care arrangement (children placed in collective day care centres experience higher prevalence of S. pneumoniae, and higher rates of antibiotic treatment), two strong determinants of antibiotic consumption. Younger children in day care centres form a ‘core’ group with high prevalence and frequency of resistance. Reducing inappropriate antibiotic prescription in priority in younger children going to collective daycare centers is predicted to lead to a 14-fold larger reduction in resistance compared to treating older children not going to collective daycare centers. The host population structure – subdivision of the human host population in different classes using antibiotics at different rates - was an important driver of resistance. In particular, this structure, with preferential transmission within classes of hosts, facilitated coexistence of sensitive and resistant strain.

The work was disseminated in two open-access publications in peer-reviewed journals and two more in preparation, and 16 international seminars and conferences.
The project represents progress beyond the state of the art along three directions:

(i) The project represents a fundamental contribution to evolutionary biology and to the topic of adaptation to heterogeneous environment.
(ii) The evolution of resistance has inspired an abundant theoretical literature, but few studies have linked the theoretical work with empirical data and validated these theories. The analytical solutions to complex epidemiological models provide general insights into the evolution of resistance.
(iii) This is complemented by an effort to develop more complex models that remain tractable, can be parameterised will available data and used to predict the frequency of resistance and the impact of potential public health interventions.

Potential impacts of the results:

Resistance frequency reacts very quickly to temporal changes in antibiotic use. In consequence, any reduction in antibiotic will immediately reduce resistance. Usage does not have to be lowered below a threshold before the impact on resistance is seen. This is in accordance with observations in campaigns aimed at reducing inappropriate antibiotic usage. Thus, the model provides a strong theoretical justification for these campaigns.

The model of resistance evolution in a structured host population allows predicting the impact of specific interventions, for example reducing antibiotic treatment in specific risk groups. These predictions could enhance the impact of interventions. They may also inform on which groups focus first when a vaccine is deployed in the population.

These contributions represent a first step towards a better understanding of resistance evolution in common bacteria such as S. pneumoniae, Escherichia coli or Staphylococcus aureus, which are mainly commensal species but occasionally cause infections responsible millions of deaths every year. Future work will focus on more complex forms of host population structure such as household structure, the association between resistance loci and other loci such as those controlling virulence of the bacteria, and the emergence of resistances to multiple antibiotics.