Clinical Systems Pathology Integrated approach for drug development and customized therapy

1
Clinical Systems Pathology(Integrated approach
for drug development and customized therapy)

• Zoltan N. Oltvai
• Department of Pathology
• University of Pittsburgh
• School of Medicine

2
Outline

• Role of informatics and computational approaches
  in medicine and pathology
• Drug discovery in bacterial metabolic networks
• 3P medicine
• Personalized 6 million nucleotide difference
  in each person
• Predictive (in a probabilistic fashion) biannual
  multiparametric blood measurements, etc.
• Preventative targeted vaccines and therapy,
  including preventative drugs

3
SingleView distributed digital imaging within a
healthcare system

• Seamless access to imaging studies and priors
  across the entire organization
• Flexible search criteria
• Report and study preview
• Global timeline

4
Advanced Visualization
Advanced 3D Visualization available to an
unlimited number of user virtually from every PC
in the healthcare network
5
Digital Orthopedic Templating
To address issues with orthopedic processes based
on two-dimensional x-ray
6
Pathology workflow today
7
Digital Pathology
Pathologist Cockpit
8
The challenge of pathology today one diagnosis
does not fit all
Need to identify disease subtypes systematically
and in a standardized way, in part using data
that is incomprehensible for the pathologist.
9
Challenges in pathology diagnosis
1. Vastly increased amount of data 2. Their
underutilization in diagnosis

• Missing capabilities.
• Disease diagnoses are not standardized
• Disease trajectories, including co-morbidities,
  are not quantitatively predicted
• Treatment modalities are not calculated

10
Copy number analyses of three morphologically
challenging kidney tumors using SNP arrays
Monzon et al, Modern Pathol., 2008, Hagenkord et
al, Diagn. Pathol., 2008
11
Diagnostic decision support (potentially
standardized)
Prototype software tool in C (Avenzoar) Openslide
(CMU) high performance scanner-neutral library
and viewer for whole slide images Agar open
source GUI development library
Renal tumor cases
Gilliland et al., in preparation
12
Analytic Result Graph for Renal Tumor
Classification
Clear cell RCC (75) Papillary RCC
(10) Chromophobe RCC (5) Oncocytoma (benign)
(5)
Computational morphometry
13
Performing comparative analysis based on SNP
array data
WEKA collated machine learning algorythms-based
14
Network representation-based modeling of disease
co-morbidities (possible disease progression)
Rzhetsky et al , PNAS 2007 Hidalgo et al., PLoS
CB 2009
15
Co-occurance of diseases at various thresholds
16
Network biology-based treatment modeling
(forward simulation of best therapies)

• Know the system
• Develop drugs that can modify elements of the
  system
• Know how the drugs modify the system

17
Microbiology diagnosis and treatment
Third and next generation sequencing
technologies Cheap and rapid bacterial genome
sequencing allowing routine sequencing of
clinical isolates
18
Long-term goal Combining the diagnosis of
infectious agent with bacterial isolate-specific
therapy
Staphylococcus aureus
19
Basis of FBA the conservation of mass of
intermediary metabolites
1. System identification
Flux balance analysis (FBA) of metabolic networks
Schilling Palsson, P.N.A.S., 1998
20
Global and local flux organization in the
Escherichia coli metabolic network
Almaas et al, Nature, 2004
21
Reorganization of the high-flux backbone upon
shifting E. coli from glutamate- to
succinate-limited growth media involves a limited
number of reactions
22
Individual reactions with high flux display mono-
or multi-modal flux distribution under different
growth conditions
23
FBA-identified metabolic core reactions of
Escherichia coli
Enzymes Red essential Green
dispensable Metabolites Blue directly to
biomass formation E. coli 81 indispensable,
unconditionally essential 9 optimal growth
Almaas et al, PLoS Comput. Biol., 2005
24
Step 1 Antimicrobial hit discovery targeting
bacterial metabolic networks
Shen et al., submitted