Antibiotic Resistance and Hospital-Acquired Infection Prof. Carl T. Bergstrom

Transcription

1 Antibiotic Resistance Carl T. Bergstrom Department of Biology University of Washington 1 In the first nine months of 2005, Pennsylvania hospitals reported: 13,711 hospital acquired infections Pennsyl vania Health Care Cost Containment Council 1,456 deaths 227,000 extra days of hospitalization $2.3 billion in extra expenses 2 PHC4 Research Brief, March 29, 2006 Pennsyl vania Health Care Cost Containment Council Scaling these numbers up to the entire United States, we expect each year approximately: 450,000 hospital acquired infections 50,000 deaths 7,500,000 extra days of hospitalization $75 billion in extra expenses 3 The screen versions of these slides have full details of copyright and acknowledgements 1

2 Antibiotic resistance After decades of heavy antibiotic use in hospitals, many hospital-associated strains of bacteria are resistant to multiple antibiotics Infection with resistant strains: Increases the probability of treatment f ailure Extends the duration of hospital stay Increases the mortality rate Increases the economic cost of treatment 4 Resistance in the intensive care unit National Nosocomial Infections Surveillance System Report, 2003 Klebsiella pneumoniae 10 % Pseudomonas aeruginosa 23 % Enterococcus sp. 28 % Staphylococcus aureus 52 % 5 Playing catch-up ball Methicillin against macrolide resistan ce % Resistance Vancomycin used against MRSA MRSA VRE Linezolid against VRE Linezolid? Year 6 The screen versions of these slides have full details of copyright and acknowledgements 2

3 Combating antibiotic resistance is a problem in applied evolution 7 How evolution works Variation: Heritability: Selection: Time: dif f erent indiv iduals hav e dif f erent traits of f spring tend to be somewhat like their parents indiv iduals with certain traits surv iv e better or reproduce more successf ul v ariations accumulate ov er many generations 8 Natural selection, in a nutshell From Battling bacterial evolution: The work of Carl Bergstrom Understanding Evolution, University of California 9 The screen versions of these slides have full details of copyright and acknowledgements 3

4 1 2 3 Antibiotic-sensitive Antibiotic-resistant Dead 10 Transformational evolution versus Variational evolution 11 Transformational process Variational process 12 The screen versions of these slides have full details of copyright and acknowledgements 4

5 Where does the variation come from? 2. What is the structure of selection? 3. How can we intervene? 13 Mutation 14 Mutation Macrolide antibiotics block protein synthesis by binding to bacterial ribosomes From Hanson et al., (2002) Molecular Cell 15 The screen versions of these slides have full details of copyright and acknowledgements 5

6 Mutation A single point mutation in the green binding region can prevent macrolide binding and confer resistance Modified from Hanson et al., (2002) Molecular Cell 16 Genome size: Mutation rate: Population size: ~ 5 x 10 6 base pairs ~ 2 x 10-3 per genome Mutation to per g fecal matter A single gram of f ecal matter is likely to contain a nov el point mutation conf erring macrolide-resistance! 17 The E. coli efflux pump AcrB Edward Yu, Iowa State More complex mechanisms 18 The screen versions of these slides have full details of copyright and acknowledgements 6

7 Nature makes penicillin I only found it Natural ecology of antibiotics - Alexander Fleming 19 Natural ecology of antibiotics Soil microbes liv e in highly structured env ironments with intense competition f or space and nutrients Many microbes produce antibiotics to kill of f their competitors Antibiotic producers must be resistant to their own products; this generates a v ast reserv oir of resistance genes in bacterial populations 20 Lateral gene transfer Transduction Transf ormation Conjugation 21 The screen versions of these slides have full details of copyright and acknowledgements 7

8 Lateral gene transfer Unknown A. orientalis Van R,S Van A,H,X Enterococcus Where does the variation come from? 2. What is the structure of selection? 3. How can we intervene? 23 Most resistant strains are commensals 24 The screen versions of these slides have full details of copyright and acknowledgements 8

9 Extremely high rate of drug use 25 Hospital staff act as disease vectors 26 High rate of patient turnover Community 27 The screen versions of these slides have full details of copyright and acknowledgements 9

10 Resistance in the community Antibiotic use by non-hospitalized patients leads to resistance in the community at large 28 Agricultural use 25 million pounds per year into animal feed Union of Concerned Scientists, Agricultural use 30 The screen versions of these slides have full details of copyright and acknowledgements 10

11 Where does the variation come from? 2. What is the structure of selection? 3. How can we intervene? 31 A model of a hospital Community Hospital Lipsitch, Bergstrom, and Levin (2000) Proc. Nat. Acad. Sciences USA 32 Translate our model into equations ds/dt=mµ+βsx (τ 1 +τ 2 +γ+µ)s dr i /dt=β(1 c)r i X (µ+τ ~i +γ)r i dx/dt=(1 m)µ+(τ 1 +τ 2 +γ)s+(τ 2 +γ)r 1 +(τ 1 +γ)r 2 βsx β(1 c)r i X µx S: patients colonized with sensitiv e bacteria R i: patients colonized with bacteria resistant to drug i X: uncolonized patients Lipsitch, Bergstrom, and Levin (2000) Proc. Nat. Acad. Sciences USA 33 The screen versions of these slides have full details of copyright and acknowledgements 11

12 We can study the dynamics using numerical solution Things change fast Non-specif ic control does appreciably reduce resistance* *When resistance is rare in the community Formulary changes can rapidly eradicate resistant bacteria Fraction resistant Time (days) Infection control (70% transmission reduction) Infection control + switch antibiotics 34 Odds ratios can be misleading Patients treated with drug 2 have a higher chance of carrying drug 1 resistance Drug Y resistance Treated with X Untreated... but this is a poor measure of efficacy In fact the net level of drug 1 resistance drops as the use of drug 2 increases Drug Y resistance Drug X usage in hospital Treated with X Untreated 35 "The `crop rotation' theory of antibiotic use [suggests] that if we routinely v ary our `go to' antibiotic in the ICU, we can minimize the emergence of resistance because the selectiv e pressure f or bacteria to dev elop Antibiotic resistance to a specif cycling ic antibiotic would be reduced as organisms become exposed to continually v ary ing antimicrobials." - M. Niederman (1997) Am. J. Respir. Crit. Care Med. 36 The screen versions of these slides have full details of copyright and acknowledgements 12

13 Based on sound ecological principles: Populations have a hard time tracking rapidly fluctuating environmental conditions 37 gentamicin piperacillin/tazobactam ceftazidime Cycling Control Cycling in a neonatal ICU gentamicin piperacillin/tazobactam ceftazidime Toltzis et al. (2002) Pediatrics 38 Clinical consequences Toltzis et al. (2002) Infection Control Cycling (n = 548} (n = 514) Resistant colonization 7.7% 10.7% Blood stream Meningitis 4 1 Pneumonia 7 13 Urinary tract 7 9 Necrot. enterocolitis 8 7 No significant difference in nosocomial infection rate 39 The screen versions of these slides have full details of copyright and acknowledgements 13

14 Modelling the efficacy of cycling Total resistant inf ections: R 1 + R 2 Baseline f or comparison: in each case, compare the outcomes under cy cling to an approximation of the status quo: Mixing of the two drugs, in which at any giv en time half of the patients receiv e drug 1, the other half drug 2 Bergstrom, Lo, and Lipsitch (2004) Proc. Nat. Acad. Sciences USA 40 Total resistant infections Three month cycling period Fraction resistant Cycling Time in days Mixing 41 Total resistant infections by cycle length One year Three months Two weeks Cycling Mixing 42 The screen versions of these slides have full details of copyright and acknowledgements 14

15 Average total resistance increases with cycle period Cycling Mixing 43 Why doesn't cycling work? T Time Bed Why doesn't cycling work? Time Bed The screen versions of these slides have full details of copyright and acknowledgements 15

16 Mixing creates more heterogeneous environment than does cycling! T Time Time Bed Bed US infectious disease mortality throughout the 20th century 1918 flu pandemic Sulfonamides Penicillin HIV Armstrong et al., Acknowledgements Marc Lipsitch Harvard School of Public Health Bruce Levin Emory University Diane Genereux University of Washington 48 The screen versions of these slides have full details of copyright and acknowledgements 16

17 49 The screen versions of these slides have full details of copyright and acknowledgements 17