1. NT-proBNP stratified follow up in outpatients heart failure clinics: A Randomized Danish Multicenter Study Morten Schou, MD, PhD Principal Investigator Hillerod University Hospital, Copenhagen The Danish Heart Failure Clinics Network On behalf on the NorthStar Steering Group: Finn Gustafsson, Lars Videbaek, Per R Hildebrandt and Morten Schou

2. Roche Diagnostics International, Basel, Schwitzerland Research grant Roche Diagnostics, Copenhagen, Denmark Research grant Merck, Sharp and Dohme, Copenhagen, Denmark Research grant

4. Objectives: To determine the effectiveness of a continued heart failure clinic intervention. To determine whether NT-proBNP could identify patients with particular benefit of continued follow up. Population: Predefined clinical stable systolic heart failure patients on optimal medical therapy. Primary outcome: Time to death or a CV hospitalization

5. GPEcho LabDept of Cardiology/ Internal Medicine Heart failure clinic (HFC): Education Exercise Etiologi ACE-I/ARB´s BB Aldosterone antagonists Adjusting doses of diuretics CRT and/or ICD Remain symptomatic Considered Stable HTX ? LVAD ? (2 centers) Scientific Question: Where should these patients be followed ? HFC ? GP (routine in DK) ? Only high risk patients in the HFC ? e.g identified by NT- proBNP ?

6. In the NorthStar Study the patients were on optimal therapy before randomization and only little room was left for a lowering strategy. (EMPHASIS, REVERSE and MADIT-CRT were initiated before NorthStar) We, therefore, created a clinical checklist, which should be used if NT-proBNP increased > 30 % compared to the randomization visit even the patient did not become more symptomatic.

7. Clinically stable systolic heart failure patients on optimal medical therapy benefit from long term follow up in a HFC. The benefit is driven by an effect only in high risk patients identified by NT-proBNP > 1000 pg/ml. (prespecified interaction analysis) High risk patients identified by NT-proBNP > 1000 pg/ml benefit further from NT-proBNP monitoring.

9. PROBE design (Prospective Randomized Open-labeled Blinded Endpoint) Multicenter (18 sites) Investigator Initiated

10. The patients visit the HFC with 1-3 months intervals based on the investigators discretion

11. M. Schou et al. (NorthStar DesignPaper): AHJ 2008

12. Simple randomization with strata* (*due to the prespecified interaction analysis (NT-proBNP*HFC))

14. N = 6180 patients had at least one visit in one of the HFC in the randomization period (registry) N= 1628 patients considered eligible N = 1120 patients randomized N = 460 patients allocated to GP N= 660 patients allocated to HFC (N=461) or HFC+NT-proBNP (N=199 ) N = 508 excluded Lost to follow up (N=0) Informed consent withdrawn due to traffiic accident (N=1) Lost to follow up (N=0) Informed consent withdrawn due to traffiic accident (N=1) N=659 patients included in final analyses N= 460 patients included in final analyses Registry Enrollment Allocation Follow up Analysis Lost to follow up (N=0)

16. Demografic variables, %GP (=460) N= 257 (NT-proBNP < 1000 ) and N=203 (NT-proBNP > 1000) HFC (=461) N= 253 (NT-proBNP < 1000) and N=208 (NT-proBNP > 1000) P-value Age, years69680.626 Female Sex, %27230.215 LVEF, %30320.123 NYHA II-III, %71750.174 Ischemic Cardiomyopathy, %57590.573 eGFR, ml/min/1.73 m266690.224 NT-proBNP, pg/ml8027930.640 Hosp with 12 months, %45410.196 ACE/ARB, randomization, %89860.100 ACE/ARB, target dose (of all patients), %68660.590 BB, randomization, %85840.726 BB, target dose (of all all patients), %49530.277 Aldo-anta, randomization, %33310.466 Loop diuretics, mg (median)40 0.773 ICD, % (only IHD in DK)9100.437 CRT, %430.728

17. * Only patients with NT-proBNP > 1000 pg/ml

18. Demografic variables, %HFC Usual Care N=208 HFC NT-proBNP-moni N=199 P value Age, years74720.197 Female Sex, %24 0.564 LVEF, %30 0.501 NYHA II-III, %80810.663 Ischemic Cardiomyopathy, %55590.661 eGFR, ml/min/1.73 m261640.151 NT-proBNP, pg/ml204218840.497 Hosp with 12 months44370.091 ACE/ARB, randomization, %83 0.222 ACE/ARB, target dose (of all patients), %65640.937 BB, randomization, %87840.665 BB, target dose (of all all patients), %52 0.861 Aldo, randomization, %32250.255 Loop diuretics, mg (median)60400.430 ICD, %1190.772 CRT, %460.668

20. HR: 1.17, 95 % CI : 0.45-1.45, P = 0.145 Blue: GP (N=460) HFC: Black (N=461) Events: 159 Events: 177

21. HR: 0.94; 95 CI:0.69-1.27; P = 0.680 HFC*NT-proBNP > 1000 pg/ml; P = 0.721 (test for heterogeneity) Blue: GP (N= 257 and N= 203) HFC: Black (N=253 and N=208)

22. * Only patients with NT-proBNP > 1000 pg/ml

23. HR: 0.94; 95 CI:0.69-1.27; P = 0.680 HR: 0.95, 95 % CI: 0.71-1.1.29, P = 0.776 HFC (N=208) Usual care: Black HFC (N=199) NT-proBNP: Red

24. Time to death Time to a cardiovascular hospitalization Time to a heart failure hospitalization Time to an over all-hospitalization Minnesota Living with Heart Failure Score NYHA class NT-proBNP Patients admitted, admissions and admission days P > 0.05 for all: -HFC vs. GP -NT-proBNP stratified hypothesis - HFC vs HFC+NT-proBNP

26. Clinically stable systolic heart failure patients on optimal medical therapy do not benefit from long term follow up in a HFC. A subgroup of high-risk patients identified by NT-proBNP do not benefit from long term follow up in a HFC. The present NT-proBNP monitoring concept does not improve long term clinical outcome.

28. Clinically stable HF patients can be referred back to their GP. Adherence in GP was suprisingly good – wrong hypothesis ? NT-proBNP identified the high-risk patients and did it´s job, but our intervention(s) could not improve outcome for these patients – wrong concept(s) ? We did not power the study to look at HF hospitalizations – wrong endpoint ? Our patients were primarily NYHA class I-II – will NT-proBNP monitoring only work in NYHA class III-IV – wrong patients ? Type II error - Bad luck ?