Update on PFO Trials Michael S. Kim, MD, FACC, FSCAI Assistant Professor of Medicine Director, Structural Heart Disease Program Division of Cardiology University of Colorado Denver July 11, 2012

Disclosures None

Outline Status of PFO closure in 2012 Scope of the problem Data Cryptogenic Stroke Migraine Headache Data ? Quantity vs. ? Quality Conclusion

PFO Closure Before Randomized Trials Devices were available, clinical expertise and experience developed, referral patterns formed, financial payment occurred, and highly motivated/engaged patients demanded treatment despite uncertainties The Genie was out of the bottle

PFO Closure 2012 The field of PFO related medical disorders remains full of controversies, some contradictory findings, and underfunding of new research that could lead to greater clarity For physicians: there are no professional society management standards for both evaluation and management of PFO related disorders AHA/ASA guidelines for secondary stroke prevention “PFO closure may be considered for patients with recurrent CS despite optimal medical therapy.” (Class IIb/LOE C) For patients: widespread disparities in both recognition of their disorders as well as the quality of care received O’Gara PT, et al. Circulation 2009

PFO and Cryptogenic Stroke

Scope of the Problem – Stroke View at 35,000 feet Stroke = leading cause of disability worldwide In United States 750,000 strokes/year 1/3 are recurrent 4th leading cause of death (as of 2009) Leading cause of acquired disability 25% of strokes occur in patients < 65 years of age ~40% of all strokes are of unknown cause, despite thorough evaluation (i.e., “cryptogenic”) “Dear Tedy, I went to Arizona State, and I always hated you. But I admire you for what you’ve overcome.” -Email to Tedy Bruschi, former linebacker for NE Patriots, after he suffered a presumed cryptogenic stroke through a patent PFO. Williams GR, et al. Stroke 1999 Sacco RL, et al. Ann Neurol 1989

PFO and Cryptogenic Stroke View at 10,000 feet Prevalence of PFO in the general population ~25% based on autopsy studies Prevalence of PFO in patients with CS ~40-50% in young (age < 55) patients ~20% in older patients Possible causal relationship between PFO and CS Relative risk of CS with PFO +3-4 compared to patients without PFO Homma S, et al. Circulation 2005

Paradoxical Embolism Definition Arterial embolism without evidence of a left-sided source Potential for RL shunting “Fly by” theory Thrombus in venous system transits the PFO “Lurking clot” theory Thrombus originates in PFO tunnel itself (in-situ propagation) Rachko Angiology 2001;52:793 Falk V, et al., Thorac Cardiovasc Surg 1997

PFO and Paradoxical Embolism View at 1000 feet PFO tunnel size and volume of RL shunt increases risk of paradoxical embolus Presence of atrial septal aneurysm (ASA) also increases risk (PICCS Study) Relative risk: +6 compared to patients without PFO

PFO and Cryptogenic Stroke Recurrence Rates The Lausanne Study Stroke recurrence: 2.4%/yr Stroke + TIA recurrence: 3.8%/yr Death: 1.2%/yr Low incidence rate of CS recurrence = challenge of statistical power in RCT

Data? Observational vs. RCT Quantity vs. quality? Multiple observational studies; One RCT Number of patients included in observational studies is nearly 10-fold higher than number of patients enrolled into CLOSURE I Observational data cannot simply be ignored Kitsios, GD, et al. Stroke. 2012

Data? Observational Data Multitude of evidence suggesting efficacy of PFO closure in significantly reducing incidence of recurrent CS over medical therapy Led to widespread adoption of “off label” use of closure devices for PFO closure Kitsios, GD, et al. Stroke. 2012

Data? CLOSURE I

Data? CLOSURE I Only RCT investigating PFO closure for secondary CS prevention with publically released data RESPECT closed in January 2012 – data not yet released Design Patients with documented, definite TIA or stroke and a PFO with 6 months of randomization 1:1 randomization to PFO closure with STARFlex occluder (NMT Medical) or best medical therapy (ASA, warfarin, or both) Primary endpoint: 2 year incidence of stroke or TIA, all-cause mortality for first 30 days, and neurologic mortality from ≥ 31 days Results 909 patients randomized No significant differences in primary endpoint of recurrent stroke (3.1% vs. 3.4%, p=0.77) or TIA (3.3 vs. 4.6%, p=0.39) Increased incidence of major vascular complication (3.2% vs 0.0%, p<0.001) and atrial fibrillation (5.7% vs. 0.7%, p<0.001) with closure

CLOSURE I What went wrong? CLOSURE I raises more questions than it answers ? Suboptimal device (86.7% closure rate) – 13.3% of patients were left with a residual shunt ? Inclusion of patients with TIA Clinical description, diagnosis, and etiology of TIA is a mixed bag (even more so than stroke!) Outcome rates in closure arm of study were nearly 4 times higher than event rate estimates based on previous nonrandomized data Criticism of patient screening techniques Clinical explanation other than paradoxical embolism could be found to explain recurrent TIA or stroke in close to 80% of patients studied Kitsios, GD, et al. Stroke. 2012

CLOSURE I It’s not dogma. CLOSURE I remains the only RCT investigating the safety/efficacy of PFO closure for recurrent CS Data from RESPECT (and PC Trial) is imminent CLOSURE I raises more questions than answers CLOSURE I is not representative of the general population The sheer volume of non RCT data CANNOT be ignored

Data? What is on the horizon? Kim MS, Carroll JD. Cardiac Interventions Today 2012

PFO and Migraine

Scope of the Problem – Migraine View at 35,000 feet Migraine prevalence Worldwide ~18% of women, 6% of men United States ~13% of general population between 20 and 64 years of age (~28 million people) Affects women in a 3:1 ratio Our understanding of migraine physiology and optimal therapy remains limited Sharma A, et al. Echocardiography 2011 Lipton RB, et al. Headache 1998

Scope of the Problem – Migraine View from 10,000 feet Migraine with aura (MA) – migraine variant characterized by transient neurological symptoms lasting from 5 to 60 min Visual, sensory, or speech disturbance Only 25-30% of migraineurs experience aura Case controlled studies report PFO prevalence of 40% to 72% in patients with migraine Highest reported prevalence in patients with MA

Scope of the Problem – Migraine View from 1000 feet Is there a causal relationship? MOA remains unknown Subclinical emboli, increased platelet activation & aggregation in response to serotonin, transient hypoxemia Size of PFO, degree of shunting, prominent Eustachian valve and/or Chiari network, +/- ASA all may play a role Genetic - ? Autosomal dominant inheritance of atrial shunts, Mendelian pattern inherence in rare forms of migraine, etc. White matter changes and gray matter abnormalities? Several non-randomized, case controlled studies have shown that closing a PFO in migraineurs for other reasons (i.e., stroke, decompression illness, etc.) reduced the frequency of migraine Wilmshurst PT, et al. Lancet 2000 Wahl A, et al. Heart 2010 Reisman M, et al. Circ Cardiovasc Intervent 2009

Data? Different Indication – Same Story Quantity vs. quality? Multiple case controlled, non-randomized, retrospective studies On average, PFO closure brought about complete resolution of migraine attacks in 57% (range, 29% to 84%) and improvement in 43% of patients (range, 8% to 83%) Migraineurs with aura are ~4.5 times more likely to have > 50% reduction in migraine frequency after PFO closure than migraineurs without aura and nonmigraineurs Placebo effect – 14% to 50% Two RCT’s in US – MIST I (published), PREMIUM (ongoing) Reisman M, et al. Circ Cardiovasc Intervent 2009

Data? MIST I

Data? MIST I Design Double-blind, prospective, randomized History of migraine with aura Frequent & disabling: ≥ 5 migraine headache days/month but at least 7 headache free days per month Drug resistant: Reported having failed at least 2 classes of preventive medications Primary endpoint: migraine headache cessation Secondary endpoints: change in migraine severity/ frequency/ characteristics, quality of life, and safety Dowson A, et al. NEJM 2008

Study Flow & Patient Disposition MIST I

Data? MIST I Results 432 migraine patients assessed for RLS by TTE 260 (60%) patients with RLS, 163 (38%) due to moderate or large PFO 147 patients divided into two study cohorts – device arm and control (sham procedure) No significant difference in primary endpoint No significant difference in secondary endpoints Reportedly, exclusion of 2 outliers in treatment arm resulted in statistical significance (P = 0.027) SAE in 16 patients in treatment arm; 3 patients in control arm Residual shunts detected in 4 patients Dowson A, et al. NEJM 2008

Exploratory Analysis MIST I Given the failure to achieve predefined endpoints, an exploratory analysis was conducted to “aid hypothesis generation and future study design” Two patients in the implant group were noted to account for > 1/3 of all migraine headache days throughout the study period and significantly differed from the rest of the population When these patients were excluded from the per-protocol population, a significant 2.2 d/mo (37%) reduction was noted in median total migraine headache days for the implant group compared to 1.3 d/mo (26%) reduction in the sham group

Conclusions MIST I Although a high frequency of R→L shunts in migraine with aura patients, no significant effect was found for the primary or secondary end points The exploratory analysis supports further investigation

Critical Analysis of MIST Why was it not successful? A completely negative study with regard to pre-defined primary and secondary efficacy endpoints Overly demanding (and potentially unrealistic) primary endpoint of complete migraine cessation Study was underpowered to adequately determine this endpoint Although the exploratory analysis is hypothesis generating, it provides NO PROOF as to a causal relationship between PFO and migraine with aura

Critical Analysis of MIST Why was it not successful? Was the follow-up period of 3-6 months adequate? Early analysis of PFO closure may have been confounded by a hangover effect of clopidogrel, incomplete device closure, residual shunts etc. Possible substandard quality of evaluation and treatment? PFO unable to be crossed in 5 patients (? initial evaluation)  no core echocardiography lab used 4 patients with residual shunts following closure (? device performance) Relatively high incidence of SAE’s - 16 patients (? quality of procedure execution)

Critical Analysis of MIST An “acceptable” risk-benefit ratio? After exploratory analysis demonstrating a 2.2 d/mo (vs 1.3 d/mo) reduction in migraine frequency Does an absolute reduction of 0.9 d/mo in migraine frequency justify the nearly 7% risk of major, potentially life-threatening procedural complication? Although the true procedural complication rate of PFO closure is much lower (~1.5% or less), what is an acceptable risk for < 1 d/mo benefit?

MIST I Lessons learned – Is the answer near? Kim MS, Carroll JD. Cardiac Intervntions Today 2012

PFO Closure – Beyond 2012 Will RCT’s bring us answers? RCT’s have been crippled by slow enrollment and (perhaps) unrealistic expectations Continued off-label use hampers enrollment Are RCT’s the ideal method for shedding light on the clinical impact of PFO? Creation of large, multicenter registries may be better suited to achieving “meaningful” answers

Food for thought

Two Middle Aged Patients with a CVA Double Standards Two Middle Aged Patients with a CVA An immediate treatment recommendation, the availability of closure with an FDA approved device, and insurance coverage. Hand-wringing by providers as to the best course of action, a prolonged informed consent process including discussions of off-label use of a device, and a potential insurance denial for an experimental procedure

Conclusions The issue of PFO closure in preventing CS and/or reducing migraine frequency remains unresolved No “reliable” data to suggest that PFO closure is NOT indicated No “reliable” data to suggest that PFO closure is beneficial Number of patients in observational data far outweighs number of patients in RCT’s Can this large quantity of observational data simply be ignored? Additional RCT data (RESPECT, PC Trial, REDUCE, PREMIUM) is pending and MAY shed additional light on the issue Unknown as to how large, multicenter registry data may/may not be used to bring clarity Until then, can we (in good conscience) deny patients a therapeutic option that is arguably just as safe and perhaps as effective as medical therapy?

Thank You michael.kim@ucdenver.edu