31
Collaborative safety studies for rare outcomes 1. EULAR Registers and Observational Drug Studies Meeting Prag, 14.-15. November 2013 Joachim Listing Programmbereich Epidemiologie Deutsches Rheuma-Forschungszentrum Berlin

Collaborative safety studies for rare outcomes

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Collaborative safety studies for rare

outcomes

1. EULAR Registers and Observational Drug

Studies Meeting

Prag, 14.-15. November 2013

Joachim Listing

Programmbereich Epidemiologie

Deutsches Rheuma-Forschungszentrum Berlin

Outline

Focus on rare life-threatening SAEs

with an incidence rate: ≤1/1000 patient-years

General remarks

Two examples

Invasive malignant melanomas

lymphoma

Power calculation: Aim: Detection of an increase in

the hazard ratio with a power of 80%

25

45

65

85

105

125

145

165

185

1.75 2 2.5 3

RCT

Obs. cohort study

Obst. cohort studyunequal group sizes

Hazard ratio

Number of SAEs which are

required to observe Assumption: SAE rate group A: 0.5 /1000 pyrs

group B: 1.0/ 1000 pyrs

Sample size: 60,000 pyrs per group

DREAM RABBIT

SCQM

MONITOR-NET

RATIO AIR ORA REGATE

HBR GISEA

ARTIS NOR-DMARD

BSRBR

DANBIO

ATTRA

BIOBADASER

Even big registers:

Low power to detect an

increase in the incidence of

rare events

European biologics register

RHEUMA-PT

How should we start with the collaboration?

Discussion of general procedures

Collaborative analysis of one or two important

research questions

How should we start with the collaboration?

Discussion of general procedures

There are recommendations/points to consider

Collaborative analysis of one or two important

research questions

Dixon W et al.

Collaborative safety studies

Comparison of the incidence of invasive malignant

melanoma between biologics naϊve and biologics

exposed RA patients

Investigation of the lymphoma risk in RA patients not

exposed to biologics, exposed to anti-TNFs, or

exposed to non-anti-TNF biologics

Reasons for project 1

Invasive malignant melanoma is a rare life-threatening

malignancy

signals from observational studies of an increased risk

in patients treated with TNF inhibitors

biologically plausible that biologics, and in particular

TNF inhibitors, will increase the risk of melanoma

Possible to restrict the attention to a small number of

confounders

Age

Sex

(sun shine exposure)

Association between biologic treatment and

inicidence of melanona

N

(population)

N of

cases

Odds ratio

Biologics naive 1866 9 Referent

ETA,INF, ADA, or anakinra

exposed

1394 23 2.3 [0.9-5.4]

Total 3260 32

Wolfe F, Michaud K. Biologic treatment of rheumatoid arthritis and the risk of

malignancy. Arthr. Rheum 2007, 2886- 94

Increased inicidence of invasive melanoma

PYRS N of

cases

Incidence

/10,000 pyrs

Hazard ratio

P. Raaschou et al

Biologics naive 203,345 113 5.6 Referent

TNF exposed 57,223 38 6.6 1.5 [1.0-2.2]

L. Dreyer et al.

Biologics naive 9,219 3 3.3 Referent

TNF exposed 15,592 6 3.8 1.5 [0.4-6.3]

P. Raaschou et al. Rheumatoid arthritis, anti-tumour necrosis factor therapy, and

risk of malignant melanoma. BMJ 2013 , 1939

L. Dreyer et al. Incidences of overall and site specific cancers in TNFalpha inhibitor

treated patients with rheumatoid arthritis and other arthritides ARD 2013;79-82

Melanoma risk in biologics naïve RA patients

Studies with n>

5 cases

Country N (popul.) N (cases) SIR [95% CI]

Gridley 1993 Sweden 11683 12 0.9 [0.5-1.6]

Mellemkjaer

1996

Denmark 20699 37 1.1 [0.8 – 1.5]

Thomas 2000 Scotland 7080 26 (f) 1.2 [0.8 – 1.8] (f)

Askling 2005 Sweden 3703 124 1.2 [1.0-1.4]

53056 0.9 [0.2 – 2.2]

Buchbinder 2006 Australia 458 7 3.0 [1.2 – 6.2]

Hellgren 2010 Sweden 6745 11 RR: 1.0 [0.5-2.0]

Mercer 2013 UK 3771 9 2.1 [0.9 – 3.9]

Raaschou 2013 Sweden 42198 113 HR:1.2[0.9-1.5]

Perkins Meta-anal. 1,351,061 pyrs 601 1.0 [0.9 – 1.1]

Melanoma risk in biologics naïve RA patients

Studies with n>

5 cases

Country N (popul.) N (cases) SIR [95% CI]

Gridley 1993 Sweden 11683 12 0.9 [0.5-1.6]

Mellemkjaer

1996

Denmark 20699 37 1.1 [0.8 – 1.5]

Thomas 2000 Scotland 7080 26 (f) 1.2 [0.8 – 1.8] (f)

Askling 2005 Sweden 3703 124 1.2 [1.0-1.4]

53056 0.9 [0.2 – 2.2]

Buchbinder 2006 Australia 458 7 3.0 [1.2 – 6.2]

Hellgren 2010 Sweden 6745 11 RR: 1.0 [0.5-2.0]

Mercer 2013 UK 3771 9 2.1 [0.9 – 3.9]

Raaschou 2013 Sweden 42198 113 HR:1.2[0.9-1.5]

Perkins Meta-anal. 1,351,061 pyrs 601 1.0 [0.9 – 1.1]

Proposal for an EULAR collaborative project

Comparison of the melanoma incidence between

biologics naϊve

anti-TNF exposed

Rituximab (abatacept?) exposed RA patients

Primary outcome: standardized incidence rates (SIRs)

with the general population as reference

Proposal for an EULAR collaborative project

Comparison of the melanoma incidence between

biologics naϊve

anti-TNF exposed

Rituximab (abatacept?) exposed RA patients

Primary outcome: standardized incidence rates (SIRs)

with the general population as reference

Simple project??

Trends in the melanoma incidence per 100,000

(age standardized, world age population)

Erdmann, Friederike, et al. "International trends in the incidence of malignant

melanoma 1953–2008—are recent generations at higher or lower risk?."

International Journal of Cancer 132.2 (2013): 385-400.

Incidence of malignant melanoma in the general

population

Clear dependence on age and sex

Increasing incidence

Stronger trends in older subjects

Differences between the countries not explained by

sunshine exposure

Incidence of malignant melanoma in the general

population

Clear dependence on age and sex

Increasing incidence

Stronger trends in older subjects

Differences between the countries not explained by

sunshine exposure

Pooling of the data is not possible

Country-specific analyses need to be performed

Further steps

Working group meeting Friday 4 pm

Chairs: Louise Mercer, Joachim Listing

Determination of study procedures

Case definition

Exposure definition

Statistical analysis

Timelines, co-authorship

2. Investigation of the lymphoma risk

Lymphomas in Swedish RA patients

Comparison of 6,604 RA pts. exposed to biologics (ARTIS)

67,743 RA pts. from RA register

471,024 persons from the general population

Cross-linkage with cancer and cause-of-death register.

26 incident lymphomas in ARTIS = 96/100.000 person-years

0 1 2 3 4

vs. General

population

vs. biologics

naive RA

population

2.72 (1.82 – 4.08)

1.35 (0.82 – 2.11)

Relative risk for lymphoma

in anti-TNF treated RA

patients Askling et al., Ann Rheum

Dis 2009;68(5):648-53

Anti- TNF

Lymphomas in Swedish RA patients

Relative Risk for lymphoma in Swedish patients with RA,

by year of anti-TNF treatment start

0

0,5

1

1,5

2

2,5

3

3,5

1999-2001 2002-2003 2004-2006

Askling et al., Ann Rheum Dis 2009;68(5):648-53

Association between disease activity and incidence

of lymphoma

E. Baecklund et al. Arthr. Rheum. 2006, p. 692-701

Lymphoma risk in RA

Major concern: TNF inhibitors increase the lymphoma

risk

J. Askling et al., F. Wolfe et al: No clear signal

Meta-analysis of RCTs:

odds ratio (TNF vs. controls): 2.1 [0.6 – 8.4]

(only 10 RCTs contribute to this result!)

Still a need to re-evaluate this risk

Wolfe F, Michaud K. Arthr Rheum (2007) 2886-95.

Askling J et al. AnnRheumDis (2009) 648-53

Lopez-Olivo et al. JAMA (2012) 898-908

Open questions

Do we still have cases with persistently highly active

disease?

How should we deal with changing treatments in

these cases?

How should we deal with confounding by indication in

general?

Which risk factors need to be considered?

Which type of analysis should we perform?

Risk factors to be considered

Age, sex

activity of RA

Immunosuppressive treatment?

(specific infections e.g.hepatitis viruses (HBV, HCV))

(herbicides, pesticides, dioxin,…)

(x-ray, radiation)

Which type of analysis should be applied?

Calculation of SIRs?

Provides additional information

Cave: confounding by indication

Cox or Poisson regression

Nested case control studies

Pooling data and performing a Cox regression?

Only confounders assessed equally can be used

DAS28(ESR), DAS28(CRP), CDAI are not exchangable

Exclusion of patients with missing data

Impact of factors on treatment decisions likely

different in different countries

E.g. the impact of the number of nbDMARD failures

Adjustment for confounding by indication by PS

methods not possible

Contrast between high precision (narrow confidence

limits) of the results and a large portion of

uncontrolled confounding

Which type of analysis should be applied?

Calculation of SIRs?

Provides additional information

Cave: channeling bias

Cox or Poisson regression

No pooling of data

Disadvantage: Exclusion of registers with low

numbers of lymphoma in the case of register

specific analyses

Nested case control studies

Proposal to investige the risk of developing

malignant lymphomas in RA exposed to biologics

Performing nested case control studies

Agreement on factors which will be used to match

cases who developed a lymphoma with controls who

did not

Register specific matching factors are allowed

„Fuzzy overlap“ of these matching factors accross

registers is needed

Further steps

Working group meeting Friday 4 pm

Determination of study procedures

Case definition

Exposure definition

Statistical analysis

Timelines, co-authorship

Thank you for your attention