Vascular Medicine Institute Chairman Department of Medicine University of Pittsburgh Institute for Transfusion Medicine Hemophilia Center of Western Pennsylvania and UPMC Personalized science and medicine ID: 760821
Download Presentation The PPT/PDF document "Mark T. Gladwin, MD Director, Pittsburgh..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Mark T. Gladwin, MD
Director, Pittsburgh Heart, Lung, Blood, and Vascular Medicine InstituteChairman, Department of MedicineUniversity of Pittsburgh, Institute for Transfusion Medicine,Hemophilia Center of Western Pennsylvania, and UPMC
Personalized science and medicine
UPMC Heart and Vascular Institute
Slide2Drivers for BIG data and analytics in Medicine
Population Health Management
:
Affordable care act and payer/provider mergers drive care models from volume to value; moving medicine to analytics-based population management
High cost/low volume care must be managed or prevented bringing “hot-spotting” to consumer medicine
Personalized (precision) medicine
:
Revolution in Genomic knowledge base has introduced an entire set of tools for refined Phenomes for clinical investigation and therapy
Expanding list of high cost, targeted therapeutics require greater precision in Phenome characterization to select high yield targets
Slide3Slide45%
High
Risk
15-35%
Rising Risk
60-80%
Low Risk
ActiveChronic Conditions
Healthy Minor Acute
4
COST
POPULATION HEALTH CARE PARADOX- Hot Spotting Risk
Analytics Needed to Distinguish
Small Volume / High Cost Population
vs
Large Volume / Low Cost Population
Analytics Critical to Target
Complex Interventions to High Risk
Cost Efficient Interventions to Low Risk
Prevention Strategies Across Risk
Slide5Drivers for BIG data and analytics in Medicine
Personalized (precision) medicine
:
Revolution in Genomic knowledge base has introduced an entire set of tools for refined Phenomes for clinical investigation and therapy
Expanding list of high cost, targeted therapeutics require greater precision in Phenome characterization to select high yield targets
Slide6What is personalized medicine?
Slide7THE NEW PHENOME
RIGHT INTERVENTION to RIGHT PATIENT
Slide8Slide9Can Personalized Medicine accelerate this timeline?
Slide10How personalized medicine will accelerate discovery to therapy
A “clean” phenotype (LDL-C levels) with variable response to therapeutic interventions
Slide112003
Slide12Slide13Slide14Slide15Slide16Slide17Structure and SNPs identified in proprotein convertase subtilisin/kexin type 9 (PCSK9)
Slide189 Years:
2012
Slide19Slide20Slide21Slide22How personalized medicine will accelerate discovery to therapy
A “clean” phenotype (LDL-C levels) with variable response to therapeutic interventions
A “complex” phenotype using integrated EMR data
Slide23Pulmonary hypertension: Deadly vascular disease with enigmatic molecular origins
Enlarged right heart
UPMC Heart and Vascular Institute
Slide24Slide25Can we explore new therapies for PH-HFpEF without doing an expensive clinical trials?
First define the EHR phenome: Key to link hemodynamic definition to defined outcomes, mortality and hospitalization
Assess the EHR phenome for outcomes based upon existing interventions (metformin) or new interventions in refined clinical trials
Slide26Aim : To examine the effect of nitrite and metformin on PH-HFpEF
**
P
< 0.01 vs. Ln-Su;
#
P
< 0.05,
##
P
< 0.01 vs. Ob-Su;
n
= 8
Slide27Can we explore new therapies for PH-HFpEF without doing an expensive clinical trials?
First define the EHR phenome: Key to link hemodynamic definition to defined outcomes, mortality and hospitalization
Project lead by Melissa Saul and Rebecca
Vanderpool
Slide28Mining Electronic Health Records
Jensen PB et al., Nature Reviews
Genetics 2012
Patient encounter
Patient linked Data
Electronic Health Records
Time
Specific Databases
Right Heart Catheterizations
1/19/2005
–
9/26/2012
20,200 incidences10,577 subjects
Project’s starting point
Slide291.) Right Heart Catheterizations
Data Sources
What about additional clinical data? Mortality?
Hospitalizations?
1/19/2005
–
9/26/2012
20,200
RHCs10,577 subjects
Starting point
Slide301.) Right Heart Catheterization
4.) Labs
5.) Pulmonary Function Tests
2.) Ejection Fraction
MRI
CT
Echo
3.) Administrative data, admin, discharge, ICD 9 diagnosis and procedure codes,
CPT codes
Patient linked data Sources
Slide311.) Right Heart Catheterization
4.) Labs
5.) Pulmonary Function Tests
2.) Ejection Fraction
3.) Administrative data, admin, discharge, ICD 9 diagnosis and procedure codes, CPT codes
Data Files
1/19/2005
–
9/26/2012
20,200 incidences10,577 subjects
Echo headings11/1/1993 – 7/6/2015252,097 entriesVolumetric derived EF (CT, MRI)3/13/1992 – 8/18/20155359 entries
12/13/1990 – 10/15/2014 652,425 entries
1/5/1998 – 5/29/2015819,163 entries
1/1/1998 – 12/31/2014>8.1 million entries
MRI
CT
Slide32I WANT ALL THE DATA!!
Clinician scientists
provide key clinical characteristics and (hopefully!)
specific
research questions
Data scientists
build upon strong bioinformatics tools and a strong knowledge base in medical data structure and relationships to extract accurate data and link disparate information
Data validation requires a strong
collaborative
process between the clinical scientist and the data scientist
Slide33Right Heart Catheterizations UPMC Presbyterian
Study Characteristics: Includes all RHCs between 1/19/2005 and 9/26/2012All patients were followed until death (last visit) or 10/15/2014
20200 RHC Incidences(1/2005 – 9/2012)
20128 RHC incidences
72 incidences excluded due to no follow-up
10,023 subjects
No PH
mPAP
< 25 mmHg
PAH
mPAP
≥ 25 mmHgPCWP ≤ 15 mmHg PH-LHDmPAP ≥ 25 mmHgPCWP > 15 mmHg
866 incidences excluded due missing PA pressure or cardiac output
19262
RHC incidences
Slide34Sources of Ejection Fraction
All reported ejection fractions in Echocardiography reports and Volumetric derived EF (MRI, CT) were tabulated. Ejection Fraction closest the date of the RHC was used. Priority was given to volumetric derived EFs. Search text/progress note for a reported EF in the 13.6% with no Echo or Volumetric EFPerformed preliminary analysis on subjects with a measured EF
Echo headings
11/1/1993 – 7/6/2015252,097 entriesVolumetric derived EF (CT, MRI)3/13/1992 – 8/18/20155359 entries
MRI
CT
No EF
2615
RHCs
Echo. EF
14,455
RHCs
Vol. EF2192 RHCs
19262
RHC incidences
Slide35Subtype PH-LHD
Subjects with an Ejection Fraction ≥ 45% were defined as having a preserved Ejection Fraction
(
pEF
)
or diastolic dysfunction based on ESC 2012 and ACC/AHA 2013 heart failure guidelines.
McMurray JJ, et al. EHJ, 2012.
Yancy
CW, et al. JACC, 2013.
Subjects with an ejection fraction < 45% were classified as having a reduced ejection fraction
(
rEF
)
or systolic dysfunction
Slide36Distribution of Ejection Fractions in the whole population
2141 subjects have PH-
HFpEFcurrently no exclusion for valvular disease
pEF – preserved Ejection Fraction (≥ 45%)rEF – reduced Ejection Fraction (< 45%)
1475 subjects have PH-HFrEF
1005 subjects have not been linked to an ejection fraction
Slide37With a defined and validated phenotype in hand we can ask question:
Does pulmonary hypertension affect outcomes (hospitalization and mortality) in patients with
HFpEF
or
HFrEF
?
Does exposure to metformin modulate this outcome?
Slide38Diffusion Capacity and Mortality in Patients With Pulmonary Hypertension Due to Heart Failure With Preserved Ejection Fraction
Hoeper
MM, et al., JACC Heart Failure. 2016.
Diastolic Pressure Gradient Predicts Outcome in Patients With Heart Failure and Preserved Ejection Fraction Zotter-Tufaro C, et al. J Am Coll Cardiol. 2015; 66:1308-1310.Diastolic Pressure GradientUnivariable Analysis: HR (95%CI): 1.078 (1.018–1.141), P-value: 0.01Multivariable Analysis: HR (95%CI): 1.057 (1.017–1.097), P-value: 0.004
Right Ventricular Function in Heart Failure With Preserved Ejection Fraction
Mohammed SF, et al., Circulation. 2014; 130:2310-2320
Pulmonary Arterial Capacitance Is an Important Predictor of Mortality in Heart Failure With a Preserved Ejection Fraction
Al-
Naamani
N, et al., JACC Heart Failure. 2015; 3:467-474
PA Ca
≥ 1.1
PA Ca
< 1.1
Slide39Univariate predictors of mortality in PH-HFpEF
Slide40Elevated TPG, PVR and PVR associate with increased mortality in PH-HFpEF
Slide41Drivers for BIG data and analytics in Medicine
Population Health Management
:
Affordable care act and payer/provider mergers drive care models from volume to value; moving medicine to analytics-based population management
High cost/low volume care must be managed or prevented bringing “hot-spotting” to consumer medicine
Slide42Example: Where are PAH patients managed and what are their outcomes?
Can we use clinical analytics
to find patients outside of specialty medical homes and hot spot them?
Use to compare outcomes between specialty and primary care
models to
evaluate need for specialty
vs
general medical homes?
Slide43Manage the PH Population
43
330 Patients
538 Visits
3857 Patients7236 Visits
364 Patients1608 Visits
Clinic*
Non-Clinic
*Clinic defined as CVI PUH HBC & PULM HBC HYPERTENSION
EPIC pts between1/2014 to 12/2014Any office visit where a patient had one of the 3 dx codes on that visit:416.0, 416.8, 416.9PA Pressure: Looked at max PA Pressure of any patient found in initial population
PA Pressure>=6019960.3%<6011534.8%N/A164.8%
PA Pressure>=60108028.0%<60185448.1%N/A92323.9%
PA Pressure
>=60
255
70.1%
<60
104
28.6%
N/A
5
1.4%
Slide44Patients with PH but not seen in the clinic
44
Office
n
CVI GRNVL OFC
318
GMC INTERNAL MED
308
BMA SMC FAM PRAC
299
HORIZON PULMONOLOGY
297
BMA SMC CARDIOLOGY
297
PGH CARDIOLOGY SHDYS
197
RFP PENN HILLS
197
MEDICOR ASSOC ERIE
184
CVI PASSAVANT HBC
177
BMA SMC INTERNAL MED
175
BMA SMC PULMONOLOGY
172
PULM HBC OAKLAND
165
CVI LATROBE OFFICE
164
HORIZON FMLY HLTHCARE
138
MURRYSVILLE EXTENDED
124
CVI SHADYSIDE
121
GMC FAMILY PRACTICE
105
RHMS PRIMARY CARE PRT
103
METRO ENT WEXFORD
103
FPN PULMONARY
102
PIMA CASTLE SHANNON
101
PULMONARY PART PASSVNT
97
CCP-HAMOT WEST
97
PARTNERS LEVEL GREEN
94
NORTHERN MED WEXFORD
93
Slide45Pulmonary Hypertension--Outcomes
45
Slide46Healthcare is at a pivotal junctureAffordable care act and payer/provider mergers will drive care models from volume to value; movement to analytics-based population managementPersonalized (precision medicine) is critical for successful population management and discovery and translation of new therapeutics for the right patientPitt and UPMC are ground zero for discovering the way forward and serving as a national model to provide better care for the right patient at a lower cost - with personalized care
46
Slide47We want you!