1. An analysis of 22,672 patients from the CLARIFY registry
How can we explain the J-curve in hypertensive patients with coronary artery disease?
An analysis of 22,672 patients from the CLARIFY registry
Emmanuelle Vidal-Petiot*, Nicola Greenlaw, Ian Ford, Roberto Ferrari, Kim M Fox, Jean-Claude Tardif, Michal Tendera, Alexander Parkhomenko, Deepak L Bhatt, and Philippe Gabriel Steg, for the CLARIFY Investigators
*DHU-FIRE, Hôpital Bichat, Assistance Publique - Hôpitaux de Paris Université Paris-Diderot, Inserm U1149
Paris, France

2. Emmanuelle Vidal-Petiot, MD-PhD
Disclosures
Non financial support from Servier
The CLARIFY registry is supported by Servier.
The sponsor had no role in the study design or in data analysis, and interpretation, but assisted with the set-up, data collection and management of the study in each country.

3. How can we explain the J-curve between BP and CV events?
Vidal-Petiot et al. Lancet 2016; 388:

4. How can we explain the J-curve between BP and CV events?
CLARIFY (Vidal-Petiot et al, Lancet 2016)
HR of CV events
Diastolic blood pressure ? ? ?
Low myocardial perfusion
Reverse causality
Arterial stiffness (↑PP, ↓DBP)

5. Does increased pulse pressure explain the J-curve?
Elastic arteries
Stiff arteries
no causal link
↑ PP
↓ DBP

6. Objective of the study
Is the increased cardiovascular risk associated with low diastolic BP in hypertensive patients with CAD the result of elevated pulse pressure ̶ associated with low diastolic BP in patients with stiffened large arteries?

7. Study population
22,672 CAD patients treated for hypertension from the CLARIFY registry
prospective longitudinal registry of 32,703 outpatients with stable CAD receiving standard care.
Enrolled in 45 countries
Age 65±10 years
Male 75%
Diabetes 33%

8. Outcomes and statistical analyses
Primary outcome:
Composite of cardiovascular death, myocardial infarction
Relationships between BP categories and outcome:
Cox proportional hazards model (unadjusted and adjusted)
Covariates:
Model 1: age, sex, geographic region, smoking status, MI, percutaneous coronary intervention, coronary artery bypass grafting, diabetes, LDL- and HDL-cholesterol, BMI, GFR, peripheral artery disease, hospitalization for or symptoms of heart failure, LVEF, ethnicity, stroke, transient ischaemic attack, and baseline medications (aspirin, statins, ACE inhibitors, ARB, beta-blockers, CCB, diuretics, and other antihypertensive medications)
Model 2: ± mean arterial pressure (PP analysis), ± PP (diastolic BP analysis)

9. Relationship between diastolic BP and primary outcome
J-curve relationship which persists after adjustment for multiple covariates
…and after additional adjustment for pulse pressure

10. Relationship between pulse pressure and primary outcome
18%
16%
14%
12%
10% 8% 6% 4%
Increased risk for high PP (≥ 65mmHg)
Incidence rate (%)
… and for low PP (< 45 mmHg)
<45 mmHg
45-54 mmHg
55-64 mmHg
65-74 mmHg
≥75 mmHg

11. Cross-classification of diastolic BP and pulse pressure
DBP<70 mmHg
DBP mmHg (ref) DBP≥80 mmHg
Cross-classifications in 9 subgroups
Pulse pressure
PP<45 mmHg PP mmHg (ref)
PP≥65 mmHg

12. Cross-classification of diastolic BP and pulse pressure
Total population
The J-curve persists in the lowest
-risk PP range

13. Cross-classification of diastolic BP and pulse pressure
Total population
The J-curve persists in the lowest
-risk PP range

14. Cross-classification of diastolic BP and pulse pressure
Total population
The J-curve persists in the lowest
-risk PP range
P-value for interaction 0.005

15. Diastolic BP and pulse pressure: subgroup analysis
The J-curve is attenuated as PP increases.
A deleterious effect of a lower DBP may be compensated by
the beneficial effect of a lower SBP.

16. Cross-classification of diastolic BP and systolic BP
DBP<70 mmHg
DBP mmHg (ref) DBP≥80 mmHg
Cross-classifications in 9 subgroups
Systolic BP
SBP<120 mmHg
SBP mmHg (ref) SBP≥140 mmHg
Vidal-Petiot et al. Lancet 2016; 388:

17. Cross-classification of diastolic BP and systolic BP
Total population
The J-curve persists in the lowest
-risk SBP range
P-value for interaction 0.24

18. Conclusion
Elevated pulse pressure is independently associated with increased cardiovascular risk
However, elevated pulse pressure is not the major determinant of the increased risk associated with low diastolic BP in this large cohort of CAD patients

19. Discussion and Perspectives
Observational studies: association in not causation
Methods to attenuate influence of reverse causality 1/excluding frail patients
2/adjusting for confounders
3/sensitivity analyses (age, heart failure…) 4/time-updated analyses 5/pathophysiological plausibility for causation …
Satter Circulation 2017; 135:
Observational studies help generate hypotheses. Only future RCTs will provide definite proof on potential harmfulness of excessive pharmacological lowering of diastolic BP.

20. Thank you for your attention!

21. (additional slides)

22. Baseline characteristics of the patients

23. Optimal BP targets: the J-curve controversy
SPRINT
CLARIFY
risk of cardiovascular events
systolic blood pressure
HR of primary outcome (95% CI 0.64 – 0.89)
diastolic blood pressure
Wright et al. NEJM 2015; 373:
Vidal-Petiot et al. Lancet 2016; 388:

24. Optimal BP targets: the J-curve controversy
151 mmHg
risk of cardiovascular events
systolic blood pressure
136 mmHg
unattended BP
HR of primary outcome (95% CI 0.64 – 0.89)
diastolic blood pressure
Wright et al. NEJM 2015; 373:
Vidal-Petiot et al. Lancet 2016; 388:

25. Does low myocardial perfusion explain the J-curve?
Rabkin et al. Am J Hypertens 2016; 29: