CNICS: A Great Tool for Outcomes Research Michael Saag, M.D. December 2nd, 2016 UCSD HIV & Global Health Rounds

2. Michael Saag, M.D.,
Associate Dean for Global Health
Director, UAB Center for AIDS Research
University of Alabama at Birmingham
CNICS:
A Great Tool for Outcomes Research

3. Disclosures
• Received Research Funding
– BMS
– Merck
– ViiV
– Gilead
– Abbivie
• Scientific Advisor
– BMS
– Merck
– Gilead

4. A case: WEAU
• 23 yo Male
• 3 days of fever, myalgias, anorexia,
diarrhea, and sore throat
Exam
• Febrile, Skin rash, pharnygeal erythema,
lymphadenopathy
Lab
• WBC 2,300 – 43% lymphs, 12% mono,
45% polys
• Monospot / HIV EIA / WB: Negative

6. Which of the following is the most
likely diagnosis
1. Mononucleosis (EBV)
2. CMV infection
3. Toxoplama gondii infection
4. HIV infection
5. Group A Streptococcal infection

7. Natural History and Laboratory Staging of HIV Infection
Eclipse
Phase I II III IV V VI
Western + (p31+)
Western blot +/-
(Fiebig, AIDS 2003)
v RNA+
Western blot + (p31-)
Keele et al., PNAS 2008

8. Productive Infection by a Single Virus
Sequences compared to Consensus PRB956
59% identical to Consensus

9. Productive Infection by Two Viruses
Sequences compared to Consensus SC33
58% identical to consensus A
57% identical to consensus B

10. Donor
Mucosa Recipient
Abortive
Abortive
Abortive
Abortive
HIV-1 Transmission Model
Time (days)
0 7 2114 28
Conclusion: The consensus env sequence at peak viremia most often corresponds to the env sequence
of the transmitted virus (Keele et al., PNAS 2008).

12. Outcomes Research

13. MEDICAL
INFORMATICS
The FUTURE:

14. UAB 1917 Clinic Cohort
 1988 - 2004: Interval Cohort era
 Medical records department reviewed notes, entered all
problems and medications to database

16. Electronic Medical Records
• Historically existed as one of 2 types:
–Provider-oriented
–Research oriented
….And never the two shall meet.

18. UAB 1917 Clinic Cohort 1988 1995 1999 8/2004 2007
Demographics
Antiretroviral medications
Concurrent medications
Clinical – HIV/AIDS events
Clinical – Co-morbidities
Laboratory – HIV-associated
Laboratory – General
Laboratory – Resistance assays
Health Services Utilization
Patient Based Metrics (PBMs)

19. WHAT IS THE TRUTH?
Can You Handle
the TRUTH?

20. Documentation of Diagnosis;
Accuracy1: 53  78%
1Accuracy % = Made / Mentioned (changes)

22. CNICS - CFAR Cohorts
>32,000 HIV-infected Individuals
John Hopkins University
University of North Carolina
at Chapel Hill
Case Western Reserve
University
University of Alabama
at Birmingham
University of California
San Diego
University of California
San Francisco
University of Washington
Fenway Health

23. Data Domain Selected Data Elements
Demographics Sex, transgender, year of birth, race/ethnicity (HRSA standards), HIV transmission risk factors, enrollment
date
Laboratory test results:
normal min/max range, interpretations; vital
signs
T-cell subsets, HIV-1 RNA levels, hematology, chemistries, liver function, coagulation, lipids, cardiac
enzymes; serologic, virologic, and genotype testing (e.g. HCV, HBV, HSV, syphilis, GC, Chlamydia)
FibroSure; Systolic/Diastolic blood pressure , height, weight, BMI
Diagnoses:
source/reliability validated, adjudicated
AIDS-defining diagnoses, non-AIDS-defining malignancies, diabetes, dyslipidemia, hypertension, cardio-
cerebrovascular disease, kidney and liver disease, VTE (e.g. PE, DVT), mental health disorders, sexually
transmitted infections, substance use
Medications:
start and stop date, verified ART regimens
Antiretroviral, antimicrobial, antifungal, antiviral/direct acting antiviral, antihypertensive, diabetes, lipid
lowering, antianxiety/depressant, antipsychotic, anticoagulant, inhaled steroids and β agonists
Vital Status:
death date, cause of death
Death dates verified by Social Security Death Index (SSDI) semiannually; cause of death data from State
Death Certificates and the National Death Index (NDI)+ classified as underlying, contributory, or immediate
using ICD-9/10 coding
Drug resistance mutations Genotype, phenotype, tropism assays with expansion to new drug targets such as integrase
Healthcare utilization Primary care and specialty care encounters, appointments, hospitalizations, insurance (e.g. Medicaid,
Medicare, Ryan White, Other Public, Private, self-pay)
Procedures Coronary revascularization/surgical interventions, VTE procedures such as IVC filter placement, V/Q scans,
doppler/duplex exams, FibroScan, Spirometry, PFTs
Biologic specimens:
number/volume of aliquots
Plasma (e.g., for biomarkers), viably frozen PBMCs (e.g., for functional immunologic assays), snap frozen
PBMCs (for genetic and transcriptional analyses)
Patient Reported Outcomes (PRO):
>70,000 assessments completed by >14,000
patients
PRO domains assessed every 4-6 mos.: adherence (AACTG, VAS, self-rating item);68,83-86 smoking,
alcohol/drug use (AUDIT-C, AUDIT, MINI & ASSIST);87-90 sexual risk behaviour; HIV Symptom Index;91
depression/anxiety (PHQ-9, PHQ-5);65,92,93 physical activity level (LRCQ);94 body morphology (FRAM);95
Quality of Life (EuroQol, EQ-5D)96-98
Geospatial State, county, census tract, census block linkage to federal big data sources such as the US Census data to
collect social and structural determinants of health
Genetic Illumina LCG chip with >2.4 million variants
CNICS Data Domains / Data Elements

24. CNICS – PBMs
Domain Instrument
Medication adherence ACTU-4
Depression PHQ
Anxiety PHQ
Alcohol use AUDIT-C
Substance use ASSIST
HRQOL EuroQOL
Symptom burden HIV Symptoms Index
Body morphology FRAM

25. 2006: CNICS
funded as an R24
as the first EMR-
based resource
network that
contributes to the
contemporary HIV
research agenda
2009: First CNICS
annual meeting at UAB
2010: additional site
UNC
2013: Patient Reported
Outcomes (PROs) are at
7 of 8 CNICS sites
2015: Over 70,000
PROs collected and
798,098 aliquots in
the specimen
repository
18%
82%
Total External
funding to date:
$55,142,788
2007: First data
upload by all 7
sites:
Case Western
Fenway
JHU
UAB
UCSD
UCSF
UW
2012:
Mentoring
Program
established
CNICS is efficient,
costing only $83 per
participant annually.
* Number of CNICS
Participants (mid-2015)
2015: 82% of
concepts developed
by junior or mid-
career investigators
Cohort Diversity
Note: 2015 is a partial year.

26. Supported by: NIH R24 AI067039
Significance as a Unique Resource
• CNICS is available to any investigator
worldwide
• 123 concept proposals have been reviewed
by the RCC to date
31% of investigators are from
outside CNICS sites
84% of concepts are led by junior or
mid-career investigators

27. Supported by: NIH R24 AI067039
Significance as a Unique Resource
• 31,824 patients are enrolled in CNICS cohort
(mid-2015)
• Racially Varied
• Geographically Diverse
• Excellent sex and age representation

28. Supported by: NIH R24 AI067039
CNICS Demographics

29. Supported by: NIH R24 AI067039
Concept Proposal Review Flow
Submission
CNICS Collaborator / Sponsor
Study / Specimen Feasibility Request
RCC Review
Scientific / Content Leader Reviews
Epi/Bios Core Review
Data Core Review
Specimen Core Review
Incidental EC Review
Reporting and Follow-up

30. No. of Unique Patients Contributing ≥ 1
TimepointPlasma
Viable PBMC
Serum
Non-viable PBMCSaliva
0
500
1000
1500
2000
2500
3000
3500
0
500
1000
1500
2000
2500
0
500
1000
1500
2000
2500
0
200
400
600
800
1000
1200
0
500
1000
1500
2000
2500

31. History and specimens prior to clinical event
(e.g., MI)
Myocardial
infarction

32. Supported by: NIH R24 AI067039
Universal Definition of MI
Adapted from Thygesen K, et al. J Am Coll Cardiol. 2012
Plaque rupture with thrombus
Vasospasm
Supply demand mismatch
Type 1 / Primary
Type 2 / Secondary

33. Supported by: NIH R24 AI067039
Risk Factors for Atherosclerotic Primary MI
• Multivariate analysis - traditional CVD risk
factors are associated with T1MI
– Age
– Cigarette smoking
– Hypertension
– Diabetes
– Dyslipidemia
– Renal disease (eGFR<30)
• Lower CD4 count independently associated
with higher risk of T1MI
• Aggressive management of traditional and
HIV-related risk factors, including ART, could
reduce atherosclerosis and MI risk
• CNICS PRO measurement of smoking
intensity and duration improves prediction of
T1MI over smoking status (current/past) alone
Covariate HR [95% Cl]
Traditional CVD Risk Factors
Age (per 10 year) 1.80 [1.56-2.07]
Male 1.31 [0.88-1.97]
Race
Black 0.76 [0.54-1.06]
Hispanic 0.43 [0.23-0.81]
Other 0.52 [0.24-1.12]
Hypertension 1.5 [1.06-2.14]
Total cholesterol (per 10mg/dL) 1.05 [1.02-1.09]
HDL (per 10mg/dL) 0.82 [0.72-0.94]
Statin use 1.65 [1.10-2.47]
Smoking 1.58 [1.16-2.16]
GFR <30 4.71 [2.64-8.41]
Diabetes mellitus 1.81 [1.22-2.69]
HIV Associated Risk Factors
HIV transmission risk
MSM 0.99 [0.68-1.44]
IDU 1.02 [0.65-1.61]
Other 0.98 [0.48-2.00]
Time-updated CD4
350-499 1.09 [0.76-1.57]
200-349 1.35 [0.95-1.92]
100-199 1.71 [1.11-2.65]
<100 2.30 [1.45-3.67]

34. Supported by: NIH R24 AI067039Hunt, et al. CROI 2016 Abs 671

35. Supported by: NIH R24 AI067039

36. Supported by: NIH R24 AI067039

37. Mean Annual Total Patient Costs
by CD4 Count (cells/ul)

38. Mean Annual Total Patient Costs
by Component

39. www.positivethebook.com

40. Thanks
UAB 1917 Clinic Cohort supported by UAB CFAR (grant P30-AI27767), CNICS
(grant 1 R24-AI067039-1), and the Mary Fisher CARE Fund