How Should We Target Prevention Interventions?
1 like578 views
The UC San Diego Antiviral Research Center holds weekly presentations on the latest infectious disease research, particularly focusing on HIV and other significant diseases. Presentations aim to enhance understanding of transmission networks, intervention strategies, and the role of genetic factors in controlling the spread of HIV. The document discusses various methodologies and findings related to molecular epidemiology and network analysis to improve public health interventions and better understand HIV transmission dynamics in specific populations.
![Legal:Take the ‘C’ out of CSI
The identification of individuals who have
transmitted HIV can have significant legal
implications.
Therefore for these data to be used for public
health purposes there must be:
◦ Decriminalization of unintended HIV transmission during
consensual exposure.
◦ Legal recognition that phylogenetic linkage using pol
sequences does not prove beyond a reasonable doubt
that transmission between partners occurred, such as
with a third party intermediary
◦ The legal precedent for admitting phylogenetic evidence
has set the bar extremely high [Scaduto PNAS, 2010], and
requires very complex analyses of multiple source and
recipient sequence isolates.
42](https://image.slidesharecdn.com/aidsrounds062813smith-130703144305-phpapp02/85/How-Should-We-Target-Prevention-Interventions-43-320.jpg)
How Should We Target Prevention Interventions?
- 1. The UC San Diego AntiViral Research Center sponsors weekly presentations by infectious disease clinicians, physicians and researchers. The goal of these presentations is to provide the most current research, clinical practices and trends in HIV, HBV, HCV, TB and other infectious diseases of global significance. The slides from the AIDS Clinical Rounds presentation that you are about to view are intended for the educational purposes of our audience. They may not be used for other purposes without the presenter’s express permission. AIDS CLINICAL ROUNDS
- 2. Network Informed Prevention Study? Davey Smith, MD Associate Professor of Medicine UCSD
- 3. Who knows who these guys?
- 5. Cholera Outbreak: space and time http://en.wikipedia.org/wiki/John_Snow_%28physician%29
- 6. Snow Veterinary Research 2011 42:56 doi:10.1186/1297-9716-42-56
- 7. How to define a network Contact versus Transmission ◦ Behavior ◦ Social ◦ Venue Molecular (genetic relatedness) ◦ Origins ◦ Timing ◦ Transmission linkages ◦ Transmission networks Retroviruses. Cold Spring Harbor Laboratory Press; c1997. Van Heuverswyn F. et al Nature. 2006 Nov 9;444(7116):164. Kober et al. Science 2000.
- 8. Why Father’s Day is Important
- 9. Do Genetic Backgrounds Matter? (Or how to blame everything on your parents) Smithsonian Natural History Museum 2011 www.hiv.lanl.gov/content/sequence/HIV/RE VIEWS/nomenclature/Nomen.html
- 10. Kosakovsky Pond and Smith Clin Infect Dis. 2009 May 1;48(9):1306-9. Viral Background Host Background (e.g. HLA) CTL How can background genetics influence HIV disease?
- 11. Defining Transmission Network and Clusters in San Diego Even in this geographically localized and aggressively sampled cohort, genetic distances <1.5% were very rare (0.25 %). Pairwise Nucleotide Distance Count 0.00 0.05 0.10 0.15 0100030005000 Distances <1.5% 0.000 0.005 0.010 0.015 020406080 Potential transmission Network N=921 unique seqs Non-subtype B
- 12. Switching Gears Networks are everywhere How many can you name is10 seconds?
- 14. SO CAN MOLECULAR EPIDEMIOLOGYTELL US ANYTHING ABOUT NETWORKS?
- 15. HIV is transmitted along a complex contact network Concept Contact Network Transmission network Node Individual HIV+ individual Edge A contact that could lead to HIV transmission, e.g. sexual, shared needle Transmission event Degree = edges connected to a node Number of contacts associated with a node Number of transmissions associated with a node HIV+ HIV- Contact w/o tranmission Transmission Degree = 7 Degree = 1 Degree = 3 Transmission network is a subset of the contact network
- 16. HIV Molecular Epidemiology Has Been Used Before Montreal and London: Primary infection drives local epidemics. (Brenner JID 2007, Fraser PLoS Med 2008, Leigh Brown JID 2011) San Diego: Partner tracing identifies hubs. (Smith AIDS 2009) Tijuana/San Diego: border offers some barrier (Mehta ARHR 2010) Europe: some countries serve as hubs (Retrovirology 2009)
- 17. United States: Clustering associated with ◦ Not using ARVs ◦ CD4 >350 cells/ml ◦ Viral load 10,000-100,000 copies/ml ◦ Women cluster with other women ◦ African Americans clustered with other African Americans (Aldous CID 2012)
- 18. Mol Epi-Network techniques are powerful Can readily infer patterns of world-wide HIV-1 transmission Found 4342 clusters using ~85,000 sequences from 141 countries. A large IDU and MSM cluster spanned 18 countries (primarily from the former USSR). Wertheim JID 2013
- 19. Defining Transmission Network and Clusters in San Diego Even in this geographically localized and aggressively sampled cohort, genetic distances <1.5% were very rare (0.25 %). Pairwise Nucleotide Distance Count 0.00 0.05 0.10 0.15 0100030005000 Distances <1.5% 0.000 0.005 0.010 0.015 020406080 Potential transmission Network N=921 unique seqs Non-subtype B
- 20. What kind of HIV network is San Diego? A random intervention would not work in a scale free network, but a targeted intervention would! Scale-free network Same size but one has Hubs Random network
- 21. San Diego Network is Scale Free and Deeply Sampled Count 50100150 2000 2002 2004 2006 2008 2010 2050100200500 Year Nodes Edges
- 22. So, can we use molecular epi and networks to do anything meaningful? Hypothesis:We can reduce R0<1 in the San Diego HIV epidemic by identifying and intervening on the “hubs” of the network.
- 23. Properties of SDPIC network 54% of individuals connected to someone else A small proportion of individuals have many connections (high degree) HIV transmission networks largely follow the model of preferential attachment – high degree nodes attract more new connections. Perhaps the network degree of the individual can be informative for intervention?
- 24. It pays to target highly connected nodes Degree = 7 Degree = 1 Targeting a low degree node has a local effect Targeting a high degree node has a global effect
- 25. Transmission Network Score DefineTNS as extent of the observed out-degree of the node and probability estimated from network degree distribution for the year of sampling. A highly connected node has a high likelihood to have future connections-TNS Cluster
- 26. Transmission Network Score For each participant, we determined the degree (i.e. ‘connectivity’) of their node within the network on the date that corresponded to the first HIV pol sequence. This was then used to predict how many new edges (i.e. contacts) the participant would acquire within the first year following their baseline sequence.
- 27. Early versus delayed treatment
- 28. LET USVISUALIZETHE PROBLEM Brought to you by Nadir Weibel and Sanjay Mehta
- 29. Applications Help us to understand the origins and dispersal of viral epidemics Reconstructing the history of epidemics Understanding the detailed epidemiology of an epidemic in real time may help us design, prioritize and direct prevention efforts
- 30. Obtaining a granular understanding of how the HIV epidemic occurred in a particular region will help us with future predictions on the how the epidemic will progress Example We find that HIV transmissions are occurring at a slow steady rate in IVDU, but occurring in bursts of related infections in MSM How? Suggests that we use this information to focus resources on contact tracing of acutely infected MSM
- 31. Obtaining a real-time picture of the HIV epidemic in a region complete with socio-demographic correlates of risk will help us choose and target our interventions Example Several new HIV infections are identified among heterosexual migrant farm-workers that are not IVDU Infections are linked together, and to several infections identified in another migrant farm-worker camp How? Suggests that we target our interventions toward sex workers servicing these camps
- 32. Clever Information Visualization helps in making hidden connections apparent New Methods? Finding the cause of a cholera outbreak John Snow’s map of London of 1854 allowed to trace back an unexplainable outbreak to a malfunctioning water pump on Broad Street … just by superimposing cholera cases (in red) on the street map.
- 33. Visualizing HIV Tracking: Challenges (1) (Too?) many variables ◦ Demographic information (e.g. age, gender, race, ethnicity, marital status, sexual orientation, education level, birth country and city) ◦ Clinical data (e.g. HIV stage at diagnosis, current CD4 levels, HAART use) ◦ HIV sequence and phylogenetic clustering ◦ Sexually transmitted infections and sexual risk (e.g. transaction sex, commercial sex, number of partners) ◦ Illicit drug use (e.g. needle sharing, use of heroin, meth, alcohol) ◦ Risk venues (e.g. bathhouses, adult bookstores, bars, internet sites used to meet partners for sex or drug use) ◦ Geography (e.g. residence, border crossing).
- 34. Visualizing HIV Tracking: Challenges (2) Continuous data over time ◦ How do we visualize trends? ◦ How can we identify significant snapshots? Unknown relationships between variables ◦ How do variable influence each other? ◦ Different views on data highlight different relationships Different kind of networks ◦ Phylogenetic ◦ Geographic ◦ Social
- 35. Approach Stanford’s Data-Driven Documents (D3) D3: Data-Driven Documents Michael Bostock, Vadim Ogievetsky, Jeffrey Heer IEEE Trans. Visualization & Comp. Graphics (Proc. InfoVis), 2011 http://d3js.org/ http://vis.stanford.edu/papers/d3
- 36. D3 Example / Inspiration
- 37. Preliminary work (1) HIV Datasets ◦ >1000 data entries collected ◦ 3 different sites: Primary Infection Cohort UCSD Owen Clinic San Diego County Epidemiology Department Phylogenetic cluster analysis ◦ Distance between single sequences ◦ 111 clusters found
- 38. Preliminary Work (2) Visualization ◦ Clusters grouped in circles ◦ Descriptive statistics for the cluster Number of subjects per cluster (size of the circle) average age (color of the circle) male/female percentage Ethnicity Sexual Orientation
- 39. Demo
- 40. Next Steps Multi-layered descriptions of phylogenetic clusters Network connectivity within the cluster and across clusters Time-based dynamic update of clusters Identification of specific important changes
- 41. In a (not too) far future? Dynamic Real-time Geographic visualizations Automatic detection of anomalies ◦ Machine learning algorithms Ultra-ScaleVisualizations ◦ Calit2 Wall Displays (VROOM)
- 42. Public Health Approach • Prevention measures cannot be a one-size fits all. • Use network analytical methods to characterize HIV transmission dynamics within a population • Use network and phylogenetic methods to identify cores, bridges and hubs • Focus outreach and prevention programs on identified cores, bridges and hubs • Since network dynamics can change (venues, susceptible populations, STIs, drug use, etc.) network surveillance should also be dynamic. • Epidemiologic surveillance • HIV genetic surveillance (genotypes) 41
- 43. Legal:Take the ‘C’ out of CSI The identification of individuals who have transmitted HIV can have significant legal implications. Therefore for these data to be used for public health purposes there must be: ◦ Decriminalization of unintended HIV transmission during consensual exposure. ◦ Legal recognition that phylogenetic linkage using pol sequences does not prove beyond a reasonable doubt that transmission between partners occurred, such as with a third party intermediary ◦ The legal precedent for admitting phylogenetic evidence has set the bar extremely high [Scaduto PNAS, 2010], and requires very complex analyses of multiple source and recipient sequence isolates. 42
- 44. Analyze HIV pol sequences in real time in the background of all sequences Generate TNS TNS high Generate Risk Profile and Intervention Plan DatAvailable for Risk Profile • Demographics • Risk behavior (MSM, IDU) • Venues (bars, bathhouses, shooting galleries, internet sites) • Estimated duration of infection (acute, early, chronic) • Clinical (CD4 and viral load) • No personally identifying information Ring Intervention Target identified venues HIV Testing, Counseling, STI testing, Needle Exchange, etc. Intervention Team • Nurse • Social worker • Test Counselor Intervention success can be measured by observing phylogenetic lineages dying out (Wertheim et al. 2011) TNS low= follow
- 46. Coverage- Acknowledgements Sanjay Mehta Nadir Weibel Sergei Kosakovsky Pond Susan Little Kimberly Bouwer Richard Garfein Igor Grant Tom Patterson Doug Richman Stephanie Strathdee Jason Young Joel Wertheim