Transcript Slide 1

Measurement Considerations In Rheumatology:
Integrating Biomarkers, Technology, Safety, and Comorbidities
to Assess Risks and Benefits of Treatment
Jeffrey Curtis, MD MS MPH
University of Alabama at Birmingham
Director, Arthritis Clinical Intervention Program (ACIP)
Co-Director, UAB Center for Education and Research on Therapeutics (CERTS)
of Musculoskeletal Diseases
Acknowledgements & Disclosures
Funding
• AHRQ R01-HS018517
• AHRQ U18-HS016956-01
• NIH AR053351
• Doris Duke Charitable Foundation
Research / Consulting
Centocor, Amgen, Abbott, UCB, CORRONA,
Crescendo, BMS, Roche/Genentech, Pfizer
Overview
• More on Measurement
– Biomarker-Based Assessment of RA
Disease Activity
– Technology-based approaches
• Safety & Relationship with Comorbidities
– Infections
– GI Perforations
– CV Events
• Putting It All Together
Which Biomarkers Might be Important in RA?
Interleukins
IL1A
IL1B*
IL1RA *
IL2
IL3
IL4
IL5
IL6*
IL7
IL8*
IL9
IL10
IL12
IL12B
IL13
IL15
IL17
IL18*
IL23
Selectins
Selectin E
Selectin L
Selectin P
Receptors
AGER
EGFR
IL2RA
IL4R
IL6R*
IL-1 receptor, type I
IL-1 receptor, type II
KIT
sFLT4
sKDR
TNFRI*
Hormones
Follicle stimulating hormone
Gastric inhibitory polypeptide
ghrelin
GLP-1
Growth hormone 1
insulin
Leptin*
NT-proBNP
Pancreatic polypeptide
POMC
Prolactin
PTHrP
PYY
Resistin *
TNF Superfamily
TNFR Superfamily
APRIL
CD30
FAS
BAFF*
LIGHT
Osteoprotegerin
LTA
TNFRSF1A
RANKL
TNFRSF1B
TNF-alpha
TNFRSF9
TNFSF18
TWEAK
Growth Factors
FGF2
EGF*
HGF
NGF
PDGF-AA
PDGF-AB
PlGF
TGFA
VEGFA*
Adhesion Molecules
Enzymes
ICAM1*
Alkaline phosphatase
ICAM3
Lysozyme
VCAM1*
Myeloperoxidase
Thyroid peroxidase
Apolipoproteins
APOA1*
APOA2
APOB
APOC2*
APOC3
APOE
*Indicates biomarkers selected for development; 25 total were selected
Bakker et al. Presented at ACR 2010; Poster #1753.
Curtis et. al. Manuscript under review.
Skeletal
Aggrecan
C2C
CS846-epitope
COMP
ICTP*
Keratan sulphate
Osteocalcin
Osteonectin
Osteopontin
PIIANP
PYD*
Other Cytokines
EPO
GCSF
GMCSF
IFNA1
IFNA2
IFNG
LIF
MCSF
CCL22*
Matrix Metalloproteinases
MMP1*
MMP10
MMP2
MMP3*
MMP9
Others
Adiponectin
Adrenomedullin
Amyloid P component, serum
Bone morphogenetic protein 6
c5a
c5b-9
CALCB
Calprotectin*
CD40 ligand
CRP*
Cystatin C
DKK
Fibrinogen
FLT3 ligand
Glial cell derived neurotrophic factor
gp130
Haptoglobin
HSP90AA1
IGFBP1
Neurotrophin 4
Pentraxin 3
S100A12
SAA1*
sclerostin
SERPINE1
sFLT1
SLPI
Thrombomodulin
YKL40*
Biomarker
Screening
• Identify
candidate
biomarkers
396
Candidate
Biomarkers
DEVELOPMENT
Select biomarkers
Build prototypes
> 500 patients
> 700 samples
• Finalize algorithm
• ~800 patients
• > 800 samples
Feasibility
II
Feasibility I
• Qualify
assays
Feasibility
III
Feasibility
IV
Assay
Optimization
Training
• Select top • Build
• Optimize
• Develop
• Prepare for
candidates prototypes
analytical
algorithm
development
performance
of individual
assays
137
Candidate
Biomarkers
Adapted from: Bakker et al. Presented at: ACR 2010; Poster #1753.
Curtis et. al. Manuscript under review.
25
Candidate
Biomarkers
>300 patients
>300 samples
•
•
•
•
FEASIBILITY
VALIDATION
SCREENING
Vectra™ DA: Development Studies
Verification
Validation
• Refine
• Evaluate in
algorithm
independent
and validate cohort
analytically
12
Final
Biomarkers
Validated
Vectra DA
Cohorts Used in Vectra™ DA
Development
BRASS (n=637)
Oklahoma
(n=288)
InFoRM
(n=685)
Leiden EAC
(n=77)
CAMERA (n=74)
Description
Brigham and
Women’s RA
Sequential Study
(Massachusetts)
Oklahoma City
Community
Cohort
(Oklahoma)
Index For RA
Measurement Crescendo
Bioscience
study (N Amer)
Leiden Early
Arthritis Cohort
(Netherlands)
Computer Assisted
Management in
Early RA
(Netherlands)
Type
Observational
Observational
Observational
Inception Cohort
Randomized Open
Label (Tight control)
Inclusion
criteria
Patients with RA >
18 yrs
Patients age 1890 with RA
Patients age
18-90 with RA
Patients with early
arthritis (all
arthritis; <2yrs)
Patients age >16
with early RA (<1
yr)
Patients
>1100
>800
>1300
>1800 all arthritis
299
Sample and
clinical exam
schedule
Annual clinical
exam and samples
One clinical
exam and
sample per
patient
3 visits/patient,
~3 months
apart, with
clinical exam
and samples
Baseline and 3
months then yearly
sample and clinical
exam
Clinical exam and
sample at every
visit: Conventional
group every 3
months, intensive
group every 4 wks
Therapies
DMARDs, biologics
DMARDs,
biologics
DMARDs,
biologics
DMARDS,
analgesics
MTX +/cyclosporine
Timeline
2003 - ongoing
2007-ongoing
2009-2010
1993-ongoing
1999-2003
InFoRM Fleischmann et al. Presented at EULAR 2010. Poster #SAT0518. BRASS Iannaccone et al. Rheumatology (Oxford). 2010
Sep 16. [Epub ahead of print] Leiden van Aken et al. Clin Exp Rheumatol. 2003;21(5 suppl 31):S100-S105. van der Linden et al.
Arthritis Rheum. 2010;62:3537–46. CAMERA Verstappen et al. Ann Rheum Dis. 2007:1443-49.
6
RA: A Disease with a Diverse Biology
IL-6, TNF-RI
VCAM-1
bone
EGF, VEGF
MMP-1, MMP-3
osteoclasts
T
cells
YKL-40
res
leptin, resistin
TNFRI
YKL40
res
IL-6
EGF
res
lep
TNFRI
EGF
chondrocytes
MMP1
MMP1
VEGF
YKL40
VEGF
VCAM1
IL-6
endothelial
cells
IL-6
EGF
innate
immunity
IL-6
MMP3
EGF
YKL40
IL-6
VCAM1
SAA
MMP1
VEGF
VCAM1
YKL40
EGF
VCAM1
peripheral blood
cartilage
res
VEGF
leukocyte recruitment
& angiogenesis
CRP
SAA
YKL40
YKL40
res
VCAM1
VCAM1
res
IL-6
TNFRI
TNFRI
osteoblasts
EGF
IL-6
TNFRI
bone erosion
lep res
IL-6
IL-6 TNFRI
IL-6
TNFRI
lep
monocytes,
macrophages,
dendritic cells
IL-6
TNFRI
res
B, plasma
cells
lep
SAA
IL-6
TNFRI
adaptive
immunity
peripheral
VEGF
VEGF
SAA, CRP
systemic
inflammatory
response
VCAM1
res
IL-6
cartilage
degradation
EGF
IL-6
MMP3
MMP1
fibroblastlike
synoviocytes
IL-6
IL-6
hyperplasia
neutrophils
Vectra™ DA Algorithm
• Includes 12 biomarkers and uses a formula similar to DAS28CRP
• Different subsets and/or weightings of biomarkers are used to
estimate SJC28, TJC28, and PG
DAS28CRP=0.56√TJC + 0.28√SJC + 0.14PG + 0.36log(CRP+1) + 0.96
TJC=tender joint count; SJC=swollen joint count; PG =patient global health
Vectra DA Score =(0.56√PTJC + 0.28√PSJC + 0.14PPG + 0.36log(CRP+1) + 0.96) * 10.53 +1
PT JC=predicted TJC, PSJC=predicted SJC, PPG =predicted PG
TJC28
Biomarkers Used To
Estimate Each DAS
Component
YKL-40
SJC28
Leptin IL-6
SAA
VEGF-A EGF
VCAM-1 TNF-RI
MMP-1
MMP-3
Resistin
Patient
Global
Bakker et al. Presented at: ACR 2010; Poster #1753.
Curtis et. al. Manuscript under review.
CRP
CRP
Vectra™ DA Validation and Performance
• The Vectra DA score was significantly associated with disease activity
categories compared to the gold standard of the DAS28CRP* (p<0.001)
RF- and Anti-CCP• AUROC = 0.70*
True Positives
True Positives
RF+ and/or Anti-CCP+
• AUROC = 0.77*
False Positives
*low versus moderate/high disease activity using DAS28CRP = 2.67 as the threshold
Curtis et al. Presented at ACR 2010; Poster #1782
False Positives
9
Vectra™ DA algorithm score tracks disease
activity over time
• Studies demonstrate that change in Vectra DA
algorithm score is significantly correlated with
change in DAS28 (p<0.001)
• In the BeSt Study:
– Vectra DA algorithm score
significantly correlated with
change in DAS28
(0.54, p < 0.0001)
.
.
Hirata S,et al. Ann Rheum Dis 2011;70(Suppl3):593;
Vectra™ DA algorithm score discriminates
low disease activity from remission
• Vectra DA algorithm score was significantly associated with
remission by ACR/EULAR Boolean criteria (by AUROC, p<0.001)
• Similar AUROCs were seen for CDAI, SDAI, DAS28CRP and
DAS28ESR remission (p≤0.001)
0.4
Sensitivity
0.6
0.8
1.0
ROC curve for Vectra DA algorithm score classification of
Boolean-defined remission vs. non-remission.
0.0
0.2
AUROC = 0.74
95% CI = [0.60,0.85]
p<0.001
1.0
0.8
0.6
0.4
0.2
0.0
Specificity
Ma MH, et al. EULAR Annual Meeting 2011; Presentation SAT0047;
11
Vectra™ DA algorithm score was not
affected by common comorbidities in a study
of 512 patients
Ratio of Disease Activity Measure’s Median Value Between RA Patients With and Without
Common† Comorbidities
n (%)
CRP
CDAI
DAS28CRP
Vectra DA
Algorithm
Score
Hypertension
223 (44)
0.98
1.32*
1.14*
1.05
Osteoarthritis
Osteoporotic
bone fractures
Degenerative
joint disease
172 (34)
0.88
1.17
1.13
1.05
131 (26)
0.91
1.05
1.02
1.05
113 (22)
1.20
1.18
1.11*
1.07
Diabetes
73 (14)
1.01
1.09
1.04
1.07*
67 (13)
1.46
1.45*
1.17*
0.91
50 (10)
1.28
1.11
1.05
1.05
Subgroup
Current
smoker
Asthma
† Present in ≥10% of the study population
* Nominal p < 0.05; adjusted for age and gender. When adjusted for multiple comparisons, none were
statistically significant
Shadick NA, et al. EULAR Annual Meeting 2011; Presentation FRI0305
Exploratory Analysis: Fibromyalgia had smaller
observed effects on the Vectra™ DA algorithm score
than on other disease activity measures
Measures of Disease Activity in RA Patients With and Without Fibromyalgia
FM (n=33)
Non-FM (n=475)
Ratio
47
42
1.1
4.3
3.3
1.3
18
11
1.6
COMPONENTS
Mean swollen joint count
4.7
4.3
1.1
Mean tender joint count
9.1
5.2
1.8
Mean patient global
50
33
1.5
Median CRP (mg/L)
7.0
4.2
1.7
INDICES
Median Vectra DA
algorithm Score
Median DAS28CRP
Median CDAI
• The slight elevation of the Vectra DA algorithm score was of
similar magnitude to the elevation in the swollen joint count
Shadick NA, et al. EULAR Annual Meeting 2011; Presentation FRI0305
13
Vectra™ DA significantly associated with
radiographic progression in the BeSt study
• In the BeSt study, the Vectra DA algorithm score had greater
observed correlations with 12 month change in total Sharp-van der
Heijde score (DTSS) than measures available in routine clinical
practice* (n=89)
Spearman Correlation
Relative performance of variables measured at Year 1 that predict TSS change from Year 1 to Year 2
0.4
0.34
0.31
0.3
0.2
0.25
0.23
0.20
0.1
0.15
0.12
0.10
0.10
0.09
0.05
0
Allaart CF, et al. EULAR Annual Meeting 2011; Presentation THU0319
14
High Vectra™ DA algorithm score in DAS28CRP
remission indicates increased joint damage risk
Risk of Progression
Risk of radiographic progression in a subset of the Leiden EAC.
All patients in DAS28CRP Remission (<2.32)
100%
RR=1.5* 87%
*p<0.05
80%
60%
58%
RR=2.3* 47%
RR=3.1*
40%
33%
20%
20%
11%
0%
>0
>3
>5
Δ TSS Threshold for Progression
DAS28CRP Remission (n=83)
DAS28CRP Remission and High Vectra DA algorithm score (n=15)
• Patients in DAS28CRP remission had a significantly higher risk of
progression if they also had a high Vectra DA algorithm score
EAC = Early Arthritis Cohort; TSS = total van der Heijde sharp score; DAS CRP remission=(< 2,32); High Vectra DA algorithm score= (> 44)
Van der Helm-van Mil, ACR Annual Meeting 2011 Presentation SUN323
16
Significant change in the mean Vectra™ DA
algorithm score occurred as early as 2 weeks
after initiation of therapy
Change in Vectra DA algorithm score (in both responders and non responders)
Bold Line indicates Median and Boxes Indicate the IQR
Δ BL
to:
n
Mean Δ
(95% CI)
p value
Wk 2
43
-8.0 (-12 to -4.1)
<0.001
Wk 6
43
-7.9 (-11 to -4.6)
<0.001
Wk 12
29
-8.4 (-13 to -3.7)
0.001
• The majority of the decrease in the Vectra DA Algorithm Score
occurred during the first 2 weeks
Weinblatt M, et al. EULAR Annual Meeting 2011; Presentation THU0339. BL, baseline
Change in Vectra™ DA Score significantly
discriminates between ACR50 responders vs.
non-responders; Change in CRP does not
• The change in Vectra DA algorithm score at the last study visit
was significantly associated with ACR50 (AUROC=0.69, p=0.03)
• The %change in CRP was not significantly associated with ACR50
(AUROC=0.60, p=0.30)
Weinblatt M, et al. EULAR Annual Meeting 2011; Presentation THU0339
18
Potential Uses of
Measuring Biomarkers in RA
• Assist in clinical management when more
information is needed
• Allow for more rapid switching of therapies in
Phase 2/3 studies & clinical practice
• Impact patient-physician communication
• Predict
– Successful therapy withdrawal
– Flare
– Radiographic progression
• Proxy for synovitis on MSK US & MRI
19
Overview
• More on Measurement
– Biomarker-Based Assessment of RA
Disease Activity
– Technology-based approaches
• Safety
– Infections
– GI Perforations
– CV Events
• Putting It All Together
Electronically Collected PROs:
One Example at UAB
Also and optionally collects MDHAQ, RAPID3, Patient Acceptable
Symptom State (PASS), EQ5D, SF-12, SF-6D, RADAI, patient
preferences…
Physician Collected Data
Final Scoring Page
RAPID3 Score
Longitudinal Trends In Disease Activity
Predicting Response with Clinical Data Collected Early
Curtis JR. Ann Rheum Disease 2011; epub ahead of print
Overview
• More on Measurement
– Biomarker-Based Assessment of RA
Disease Activity
– Technology-based approaches
• Safety
– Infections
– GI Perforations
– CV Events
Increased Infection Due To RA Itself
and Active Disease
• 609 RA patients and 609 controls matched on age,
residence, sex* residing around Rochester, Minnesota
– Greater than 12 years of follow-up, Pre-biologic era
– Risk for hospitalized infection associated with RA:
hazard ratio = 1.83 (1.52-2.21)
• CORRONA registry**
– More than 25,000 RA patients
– More active RA  higher rate of infection
* adjusted for smoking, diabetes, chronic lung disease, steroid use, and leukopenia
* Doran et al. Arthritis Rheum 2002; 46(9):227-2293
** Au et. al. Ann Rheum Disease May 2011;70(5):785-91
Potentially Confounding Factors:
Concomitant Glucocorticoid Use
Mean daily dose of glucocorticoids
(no. of treatment episodes), outcome
≤5 mg (n = 1,781)
Pneumonia
Any bacterial infection
6-9 mg (n = 1.510)
Pneumonia
Any bacterial infection
10-19 mg (n = 4,435)
Pneumonia
Any bacterial infection
≥20 mg (n = 2,891)
Pneumonia
Any bacterial infection
Propensity score
adjusted rate ratio (95% CI)
0.88 (0.37-2.12)
1.34 (0.85-2.13)
2.01 (0.87-4.66)
1.53 (0.95-2.48)
2.97 (1.41-6.23)
2.86 (1.80-4.56)
6.69 (2.83-15.8)
5.48 (3.29-9.11)
Schneeweiss S. Arthritis Rheum. 2007 Jun;56(6):1754-64
Schneeweiss, S. et al., Arthritis Rheum 2007;56:1754-64.
Effect of Anti-TNF Therapy on the Incidence of
Serious Infections in RA Patients:
Results from Clinical Trials
Summary Relative Risk of Infection = 2.0 (1.3 – 3.1)
Bongartz T et al, JAMA, May 17 2006, Vol 295: No. 19, 2275-2285
Results from Observational Studies:
Serious infections under anti-TNF treatment
Incidence of serious infections in anti-TNF treated
patients (per 100 patient years)
RABBIT: Listing et al., Arthritis Rheum 2005;52:3403-12
6.3
BSRBR: Dixon et al., Arthritis Rheum 2006;54(8):2368-76
5.3
ARTIS: Askling et al., Ann Rheum Dis 2007;66:1339-44
5.4*
Curtis JR, et al., Arthritis Rheum 2007; 56(4):1125-33
2.9**
Schneeweiss S, et al., Arthritis Rheum 2007; 56(6):1754-64
2.2
*only prior hospitalized patient, first year
** in the first six months after biologic use
Rates of Serious Infections Largely
Driven by Disease, Comorbidities and
Patient Factors, Not Biologics
Rate of Serious Infections
per 100 person-years
PBO + DMARD
3.8
Combination MTX + TCZ, Overall
5.2
RR= 5.2 / 3.8 = 1.4
PBO = placebo; TCZ = tocilizumab
Kremer et. al. ACR 2008, abstract 1668; Smolen et. al., ACR 2008, abstract 1669;
Genovese 2008 (TOWARD); Emery 2008 (RADIATE)
Rates of Serious Infections Largely
Driven by Disease, Comorbidities and
Patient Factors, Not Biologics
Rate of Serious Infections
per 100 person-years
PBO + DMARD
3.8
Combination MTX + TCZ, Overall
5.2
TOWARD (DMARD failure, biologic naive)
(MTX +TCZ 8mg/kg) vs. (MTX + PBO)
5.9 vs. 4.7
PBO = placebo; TCZ = tocilizumab
Kremer et. al. ACR 2008, abstract 1668; Smolen et. al., ACR 2008, abstract 1669;
Genovese 2008 (TOWARD); Emery 2008 (RADIATE)
Rates of Serious Infections Largely Driven
by Disease, Comorbidities and Patient
Factors, Not Biologics
Rate of Serious Infections
per 100 person-years
PBO + DMARD
3.8
Combination MTX + TCZ, Overall
5.2
TOWARD (DMARD failure, biologic naive)
(MTX +TCZ 8mg/kg) vs. (MTX + PBO)
5.9 vs. 4.7
RADIATE (TNF Failures, refractory RA)
(MTX +TCZ 8mg/kg) vs. (MTX + PBO)
9.9 vs. 9.6
Risk difference for patients on MTZ + TCZ who have
diabetes compared to those who don’t is ~ 4 / 100py
PBO = placebo; TCZ = tocilizumab
Kremer et. al. ACR 2008, abstract 1668; Smolen et. al., ACR 2008, abstract 1669;
Genovese 2008 (TOWARD); Emery 2008 (RADIATE)
Applying Research Results
to Clinical Care
How much should a ~1.5 to 2-fold
increased risk of infection
matter to my patients?
Putting Relative Risks into
Context:
Two Examples
Example Patient
#1: 42 yo, severe RA
MTX, HCQ
no other medical problems
Hypothetical
Baseline
Serious Infection
Rate
Hypothetical RR
of Infection
Associated with
Biologic Use
Resulting
Infection
Rate
1% / yr
2.0
2% / yr
Putting Relative Risks into Context:
Two Examples
Hypothetical
Baseline
Serious Infection
Rate
Hypothetical RR
of Infection
Associated with
Biologic Use
Resulting
Infection
Rate
#1: 42 yo, severe RA
MTX, HCQ
no other medical problems
1% / yr
2.0
2% / yr
#2: 65 yo, moderate RA
MTX, prednisone 7.5 mg/day
Diabetes, COPD, hosp. for
pneumonia last year
10% / yr
2.0
20% / yr
Example Patient
Safety Assessment of Anti-TNF
Agents Used in Autoimmune
Disease (SABER)
Sponsored by FDA / AHRQ
THE UNIVERSITY OF
ALABAMA AT BIRMINGHAM
CCEB
Specific Aims
• Aim #1: To estimate incidence rate ratio (RR) of SAEs associated
with each biologic agents among users and comparable nonusers
– To estimate the RR of SAEs after considering time since first use,
duration of use, concomitant drug use and relevant comorbidities
• Aim #2: To estimate the RR of SAEs in vulnerable populations
including
(1) low income groups;
(2) minority groups;
(3) women (especially pregnant women);
(4) children;
(5) the elderly;
(6) individuals classified as disabled;
(7) patients with co-morbidities;
(8) patients living in rural or inner city areas who may have reduced
access to health care.
SAE = serious adverse events
Centers, Working Groups, & Datasets
Center
Working Group
(Outcomes Lead)
Infections (including
Opportunistic, TB)
Datasets used for
Each Outcome
Medicare Standard
Analytic Files & MAX,
1999-2006
HMORN
Death,
Pulmonary Fibrosis
KPNC, 1998-2007
Univ Penn
Malignancies
-
UAB
Vanderbilt
Congenital anomalies & TennCare, 1998- 2007
pregnancy outcomes
Fractures
Brigham and
Cardiovascular
PACE, PAAD ,’ 98- ’06
Women’s
BCLHD, ’96-’06
DEcIDE Center
Horizon BCBSNJ, ’96-’07
New Paradigms to Pool Data to
Study Rare Adverse Events
Rassen J. Med Care. 2010 Jun;48(6 Suppl):S83-9.
SABER Results for Serious
Bacterial Infections
Figure 3. Incidence Rates and hazard Ratios for Specific
TNF-a Antagonists and Serious Infections Among
Patients with Rheumatoid Arthritis
Is Serious Infection Risk Additive,
or Multiplicative, for anti-TNF Users?
20
18
16
14
12
DMARD Only
TNF Users
TNF Users2
10
8
6
4
2
0
Low Risk
Medium Risk
High Risk
Assumptions for this hypothetical scenario
DMARD rate of infection is 3 per 100 patient years; TNF user rate is 6 per 100
patient years. Rate ratio = 6 / 3 = 2.0; Rate difference is 6 - 3 = 3.0 per 100 py
Is Serious Infection Risk Additive,
or Multiplicative, for anti-TNF Users?
20
18
16
14
12
DMARD Only
Multiplicative
Additive
10
8
6
4
2
0
Low Risk
Medium Risk
High Risk
Assumptions for this hypothetical scenario: multiplicative risk doubles the
rate of infection, additive risk increases it by 3 per 100 patient years
Infection Risk Constant for High
Risk and Low Risk Patients
Curtis JR, ACR 2011 annual meeting, manuscript under review
TB Risk for those on
Anti-TNF Therapy
UK Biologic Registry
Cochrane: TB rate 200/100,000 persons receiving drug
Dixon WG et al. Ann Rheum Dis 2010:69:522-528
Drug-Specific Risks of Other Opportunistic
Infections from French RATIO registry
• 45 cases of opportunistic infections
• Most common infections were zoster, PCP, listeria,
nocardia, non-tuberculosis mycobacteria
• Overall absolute event rates 1.5 / 1000 py
Adjusted
Odds Ratio (95% CI)
Most recent TNF
Etanercept
Adalimumab
Infliximab
1.0 (referent)
10.0 (2.3 – 44.4)
17.6 (4.3 – 72.9)
Prednisone > 10mg/day or bursts
No
Yes
1.0 (referent)
6.3 (2.0 – 20.0)
Salmon-Ceron et. al. Ann Rheum Dis 2011; 70:616–623
Incidence of PML in SABER
• Among 712,708 unique individuals with RA, PsA, PsO,
JIA, IBD, or AS, a total of 55 hospitalizations with PML
diagnoses identified
• 55 suspected cases
– 29 had insurance coverage for > 6 months prior to the PML
case date and > 1 physician diagnoses of a rheumatic
disease that occurred before PML case date
– 82% with HIV; 10% with malignancy
• Overall case rate = 7.7 per 100,000 individuals
• Among biologic users, 1 cases among inflixumab
users, 2 among rituximab users
• Case rate among patients with autoimmune diseases
on biologics w/o HIV or cancer ~0.2 per 100,000
Bharat A, Curtis JR. Arthritis Care & Research, in press
What About Infections
for Which We Can Vaccinate?
• Patients with rheumatic and autoimmune
diseases are at increased risk of herpes zoster
(HZ), also known as shingles
• A live zoster vaccine reduces risk by 51%
– Treatment-related contraindication
– Safety concern: vaccine might trigger HZ in
these patients within 4-6 weeks
– Safety and efficacy not clear
Strangfeld et al., JAMA. 2009;301(7):737-744.
Oxman et al., N Engl J Med. 2005;352(22):2271-2284.
Harpaz et al., MMWR Recomm Rep. 2008;57(RR-5):1-30; quiz CE32-34.
Study Design
Retrospective cohort study using 100% sample of
Medicare data
– age >= 60
– RA, psoriasis, PsA, AS, or IBD based upon >= 2
MD diagnoses
Vaccination
Start of
Follow-up
Safety analysis:
≤ 42 days after vaccination
End of
Follow-up
Effectiveness analysis:
> 42 days after vaccination
Unvaccinated Person-time
Results
• 463,104 eligible patients with at least one of the
5 autoimmune diseases of interest
– Mean age 74 years
– 72% women
– 86% Caucasian
– 20,570 (4.4%) received zoster vaccine
– 10,032 developed HZ during follow-up
– Patients with RA contributed over half (65.3%)
of the total person-years during follow-up
Herpes Zoster Incidence Rates,
Unvaccinated, by Steroid Exposure
Medications (exclusive groups)
Any anti-TNF (regardless of non-biologic
DMARDs use)
Adalimumab
Etanercept
Infliximab
Other anti-TNFs
Any non-TNF biologics (regardless of nonbiologic DMARDs use)
Abatacept
Rituximab
Non-biologic DMARDs without biologics
Exposure to
Glucocorticoids
No
Yes
‡
‡
HZ IR
HZ IR
IR Ratio
95% CI
12.6
22.4
1.8
1.6-2.0
11.8
11.5
13.2
15.6
14.3
21.7
20.7
23.2
26.2
18.6
1.3
1.0-1.7
12.1
17.5
11.0
17.1
20.4
18.6
1.7
1.6-1.7
Methotrexate (regardless of other non10.4
18.2
biologic DMARDs use)
All other non-Methotrexate DMARDs alone
11.9
19.3
or in combination
*HZ, Herpes Zoster; IR, Incidence Rate per 1,000 Person-Years; 95% CI, Confidence Interval
Herpes Zoster Incidence Rates by
Vaccination Status and Medication
Exposure
Safety Endpoint:
≤ 42 Days Following Vaccination Unvaccinated
Infections, Vaccinated,
IR*
IR*
n
n
Overall
Drug Exposure
Biologics (regardless of
concomitant DMARDs or oral
glucocorticoids)
<11
7,781
7.8
11.6
0
636
-
15.8
Anti-TNF therapies
DMARDs (without biologics
but regardless of oral
glucocorticoids)
Oral glucocorticoids alone
0
<11
556
1,817
14.6
15.7
13.8
<11
1,215
21.2
17.1
*HZ, Herpes zoster; IR, incidence rate per 1,000 person-Years
Reduced Risk of Zoster
Associated With Vaccination,
Varying Case Definitions
Outcome Definition
Hazard
Ratio*
95% CI
Diagnosis code + anti-viral
0.69
0.56-0.86
medications
Diagnosis code only
0.72
0.71-0.84
*Controlling for age, gender, race, concurrent
medications (anti-TNF, non-TNF biologics, non-biologic
DMARDs, oral glucocorticoids), and health care
utilization (hospitalization and physician visits)
TNF Inhibitors and Risk of
Post-Op Infections: Impact of Stop Time
SPOI and Influence of Stop Time
On
Infections, N
(%)
Adjusted OR
(95% CI)
49
(3.0)
Ref.
Off
15
(3.5)
1.15
(0.622.12)
On/Off 28 Days
Before Surgery
On 28
Days
59 (3.4)
Ref.
Off 28
Days
5 (1.4)
0.38
(0.180.93)
On/Off
at Surgery
Adjusted OR (95% Cl)
On/Off at Time
of Surgery
On/Off 28 Days
Before Surgery
2
"On"
1.15 "On 28"
1.0
0.6
0.4
"Off"
0.2
Conclusions
• Patients off TNF inhibitor >28 days before surgery had ~60% reduction in
infections
• Data support discontinuing TNF inhibitor at least 4 weeks prior to surgery
Dixon W, et al. Presented at: 2007 EULAR Annual Meeting. Barcelona, Spain. Abstract OP0215.
0.38
"Off 28"
Are Anti-TNF Users at Higher Risk
for Recurrent Malignancies?
Person-years of followup
Median (IQR) follow-up time, yrs
Incident malignancies, no.
Rate per 1,000 person-years
IRR (95% CI)
DMARD
(n =117)
Anti-TNF
(n =177)
235
515
1.9 (1.3–2.7)
3.1 (2.0–3.9)
9
13
38.3 (17.5–72.7) 25.3 (13.4–43.2)
1.0 (referent)
0.56 (0.23–1.35)
Dixon WG et al. Arthritis Care Res (Hoboken). 2010 June; 62(6): 755–763
Rates of GI Perforations for Patients on
Biologics and DMARDs
Drug Exposure Group
Rate/1000 PYs (95% CI)
Biologics with glucocorticoids
1.87 (1.46–2.35)
Biologics w/o glucocorticoids
1.02 (0.80–1.29)
Methotrexate with glucocorticoids
2.24 (1.82–2.74)
Methotrexate w/o glucocorticoids
1.08 (0.86–1.35)
Other DMARDs* with glucocorticoids
3.03 (2.34–3.85)
Other DMARDs* w/o glucocorticoids
1.71 (1.34–2.16)
Glucocorticoids w/o any DMARD or biologic
2.86 (2.27–3.56)
No DMARDs, biologics, or glucocorticoids
1.68 (1.44–1.96)
Total
1.70 (1.58–1.83)
DMARD=disease modifying antirheumatic drug; PYs=person years.
*Azathioprine, chloroquine, hydroxychloroquine, cyclosporine, D-penicillamine, leflunomide, sulfasalazine, gold compounds. 124
Curtis JR et. al. presented at EULAR 2011, London
124
Exposure On or After Index
Relative Risk of GI Perforation During
Follow-up–Adjusted Results
Diverticulitis
Diverticulosis w/o diverticulitis
Other DMARDs w/ glucocorticoids
Glucocorticoids w/o any DMARD
Methotrexate w/ glucocorticoids
Biologics w/ glucocorticoids
Biologics w/o glucocorticoids
Other DMARDs w/o glucocorticoids
No DMARD or glucocorticoid
NSAID
Baseline CCI
Age 65+
Age 40-64
Female
Urban
0
1
2
3
4
11
13
15
17
19
Hazard Ratios With 95% Confidence Intervals
Results of Sensitivity Analysis that Varied Definition of GI Perforation
• Exclusion of diverticulitis/diverticulosis + GI surgery decreased incidence rate to
1.25 (95% CI, 1.12–1.34) per 1000 PYs
• Hazard ratio for diverticulitis ranged from 3.6 to 14.5
Reference groups are as follows: for all drug groups except NSAIDs = methotrexate without steroids; for NSAIDs = the absence of NSAIDs; for all binary variables
= the absence of the condition or status.
CCI=Charlson Comorbidity Index; DMARD=disease-modifying antirheumatic drug; NSAIDS=Non-Steroidal Anti-Inflammatory Drug.
125
125
Incidence Rate
(per 1000 person-years)
RA Is an Independent Risk Factor
for MI, Stroke
70
60
50
40
30
20
10
0
Patients With RA (n=25,385)
Patients Without RA (n=252,976)
18-49
50-64
65-74
Age Range (y)
Solomon DH et al. Ann Rheum Dis. 2006;65:1608-1612.
75+
Changes in Lipids Associated
with Tocilizumab (IL-6Ra)
Mean Change From Baseline
in 6-Month Controlled Period
30
25
25
ACT 8 + DMARD (n = 1582)
20
ACT 8 (n = 288)
20
ACT 4 + MTX (n = 774)
13
15
10
5
0
5
4
HDL (mg/dL)
3
LDL (mg/dL)
* From tocilizumab prescribing information (PI)
127
Change from Baseline (mg/dL)
Increase in Total Cholesterol
associated with Anti-TNF therapy
Infliximab*
30.0 28.0
Adalimumab
25.0
20.0
20.0
15.0
13.0
n = 32
10.0
7.2
5.0
n = 52
1.4
n = 10
n = 56
9.0
n = 97
6.7
n = 19
5.8
n = 55
n = 45
n = 80
0.7
0.4
n = 69
6.0
n = 33
n = 50
0.0
1
2
3
4
5
6
7
Study
2.5
8
9
10
11
12
*Two additional studies with total n of 35 had a mean change in total cholesterol of -5.4 (Popa, et al. Ann rheum Dis
64(2):303-305) and -2.3 (Perez-Galan, et al. Med Clin (Barc) 126(19): 757) mg/dL.
Pollono EN. Clin Rheumatol. 2010; 29(9):947-55.
3.6
n=8
13
TNF Inhibitor Therapy in
RA and CV Outcomes
•
Examined 10,870 patients with RA
from CORRONA registry
CV Events
– Median RA duration: 7 years
– Median follow-up: 2 years
Conclusions
– Compared with non-biologic therapies
excluding methotrexate (MTX)
• Substantial reduction in CVD risk
for patients treated with TNF
inhibitors (RR 0.3)
• Intermediate reduction in CVD
risk for patients treated with MTX
(RR 0.6)
– Prednisone an independent risk factor
for CVD
Greenberg JD. Ann Rheum Dis. 2011 Apr;70(4):576-82.
1.5
HR
•
2.0
1.0
0.6
0.5
0.3
0
TNF
MTX
Putting It All Together:
Applying Research Results
to Clinical Care
Communicating Risk
Know What Your Patients are
Reading about Safety
• “The most common side effects of Prolia®
are back pain, pain in your arms and legs,
high cholesterol, muscle pain, and bladder
infection.”
(manufacturer website at www.prolia.com)
Denosumab* (n = 3886)
Placebo* (n = 3876)
1347 (34.7%)
1340 (34.6%)
Pain in extremity
453 (11.7%)
430 (11.1%)
Musculoskeletal pain
297 (7.6%)
291 (7.5%)
Hypercholesterolemia
280 (7.2%)
236 (6.1%)
Cystitis
228 (5.9%)
225 (5.8%)
Back pain
* As observed in pivotal 3 year trial
Communicating Benefits
and Risks of Biologics to Patients
• “Ms. Jones, there’s a good chance that
you will respond to this medication, but…
• “It may increase your risk of infection by
50 to 100%”
OR
“There is an extra 2 out of 100 chance over
the next year of having a serious infection
OR
100 patients, Active Disease, on MTX










10










20










30










40










50










60










70










80










90










100
Likelihood of Achieving an Good
Clinical Response, Remaining on MTX
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
























































































10
20
30
40
50
60
70
80
90
100
Likelihood of Achieving a Good
Clinical Response, Adding a Biologic
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻
☻






























☻





☻





☻











10
20
30
40
50
60
70
80
90
100
Likelihood of a Serious Bacterial
Infection, Remaining on MTX




































































































10
20
30
40
50
60
70
80
90
100
Likelihood of a Serious Bacterial
Infection, After Adding a Biologic




































































































10
20
30
40
50
60
70
80
90
100
Risk:Benefit Curve of Aggressive Therapy
Need for Aggressive Rx
Risk of Therapy
increased toxicity
+/- benefit
limited
toxicity
+ benefit
(control of
inflammation
lowers risk)
……older age,
disability,steroids, etc
Death
Severity of Comorbidities
Summary & Conclusions
• Biomarkers appear useful to assess disease activity in an objective
manner and may predict future outcomes (e.g. structural damage,
CV risk, future response to tx)
• Clinical data, perhaps in conjunction with biomarkers, may be
maximally useful; technology may assist in collecting this data
• Infections
• Increased risk of infections, largely early after starting
• Risk difference compared to non-biologic therapies low
(~1-4 / 100py)
• Appears similar for low vs. high risk patients
• No greater than risk for moderate dose glucocorticoid use
• Risk for zoster does not appear to be increased with vaccination,
even for biologic users
• No apparent increase in primary or recurrent malignancy except
possibly non-melanoma skin cancer
Summary & Conclusions
• Increases in lipids but neutral or even reduced CV risk
• Low absolute rates of other SAEs (e.g. gastrointestinal
perforation)
• Lots of data, new methods needed to study rare SAE
• Overall risk-benefit profile of biologic therapy likely to be
favourable for almost all patients who need it
• Communicating Risk to Patients Challenging, Better
Tools Needed
• Absolute risk (not relative risk) likely to be most
informative
Acknowledgements &
Collaborators
• UAB
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
John Baddley, MD MPH
Tim Beukelman, MD MSCE
Aseem Bharat, MPH
Lang Chen, PhD
Elizabeth Delzell, ScD
Mary Melton
Paul Muntner, PhD
Ryan Outman, MS
Nivedita Patkar, MD MPH
Kenneth Saag, MD MSc
Monika Safford, MD
Jas Singh, MD MPH
Fenglong Xie, MS
Shuo Yang, MS
Jie Zhang, PhD
• OHSU
– Kevin Winthrop, MD
• U Nebraska
– Ted Mikuls, MD MSPH
• U Utah
– Grant Cannon, MD
– Scott Duvall, PhD
• Vanderbilt University
– Carlos Grijalva, MD
– Marie Griffin, MD
• Brigham and Women’s
Hospital
– Dan Solomon, MD MPH
– Jeremy Rassen, ScD
– Sebastian Schneeweiss, ScD