How helpful are Holter recordings at predicting risk of sudden death ?
Andrew Staniforth Nottingham University Hospital NHS Trust
Science 1961. 134; 1214-1220 New Method for Heart Studies: Continuous electrocardiography of active subjects over long periods is now practical Norman J Holter I (propose) that … electrocardiography be implemented, both for research and medical purposes, by the use of long-period, continuous recording of heart potentials with a portable, self-contained instrument
SCD • • • • •
SD important – 80k / YR UK, 350K / YR USA 50% of cardiac deaths are sudden 50% of sudden deaths are 1st cardiac presentation Survival rate 5% Now we have an effective Rx (ICD) – but it’s expensive & invasive SO – we want need to risk stratify
BUT • It’s not a single disease (although most data relates to CAD) • It’s not all arrhythmia • VT / VF is not the only arrhythmia – and that might be changing SO – single test as magic bullet - naive
EARLY SCD RISK PROGNOSTICATION IN AMI Began with the Holter
• Moss Circ 1974 & 1979 •
Described that in post-MI patients VT / VF often preceded by increase VPB activity and complexity
SCD RISK PROGNOSTICATION - CAD VPB / NSVT found to be a marker of increased risk SCD
Buxton SCD. Heart Rhythm 2009; 6: 836-47
Sensitivity is low 6-53% & acting as marker for SCD and non-SCD risk
EF cut off adopted from ICD trials has low sensitvity
ICD have limited impact on population SCD
Target
Tool
Depolarisation QRSD SAECG Repolarisation QT TWA Autonomics
HRV HRT
BRS Triggers
VPB / NSVT EPS
Anatomy
EF
Target
Tool
Depolarisation QRSD SAECG Repolarisation QT
Limitations Only MUSTT (LBB at increased risk) & MADIT II (LBB benefited more) Identifies non-sudden rather than SCD Sens 22-75% and Hi NPV for SCD No consistent relationship
TWA Autonomics
HRV HRT BRS
Triggers
Anatomy
Some studies identify risk of SCD greater than total mortality. Sens 5-73%. Best > 1-month post-MI
VPB / NSVT
Predicts increased risk sudden and non-sudden death Sens 12-53%.
EPS
Identified risk for SCD greater than for total mort (esp for better EF and MMVT). Sens 30-80%
EF
Sens 22-72%. Identifies total mort & SCD. Lo PPV, Hi NPV
2001 Analysis of Screening Tests for SCD following MI
SVA = serious ventricular arrhythmia
2014 Meta-Analysis of Screening Tests for SCD in NICM Goldberger. JACC 2014; 63: 1879-89
Limitations • Must be SR / edit out VEs • Longer recordings preferable • What’s best measurement? Frequency Domain Analysis
• Complex math to describe the underlying basic frequency oscillations that drive variations in heart rate • • • •
HF – PANS LF - ? SANS VLF ULF
• Similar prognostic ability for time and frequency domain data • Most studies similar predictive value for SCD and non-S-CD
• Compared with other markers BRS in ATRAMI study fared poorly in patients with low EF ?-Tends to drop-out on multi-variate analyses • Large Finnish (Huikuri 2003) registry – marker for non-sudden mortality only • Doesn’t seem to be useful marker in NICM
Heart Rate Turbulence
• A manifestation of BRS. Following VPB there is HR acceleration • Not in AF, need a VPB • A flattened acceleration slope was identified as markers of death in the EMIAT and MPIP study cohorts Schmidt et al. Lancet 1999; 353: 1390-6
Buxton SCD. Heart Rhythm 2009; 6: 836-47
Buxton SCD. Heart Rhythm 2009; 6: 836-47
Micro-Volt T-wave Alternans • • • • • •
Micro-voltage beat-to-beat variation in T-wave voltage Must be in SR B-blockers withheld as diminish alternans voltage Complex math beat-to-beat spectral analysis of T-wave template Most studies are ETT based Holter method exists (REFINE)
• Gehi. JACC 2005; 46: 75-82 Meta-analysis likely significant marker for SCD • Ikeda. JACC 2006; 48: 2268-74 May also be a marker of SCD with preserved LV post MI • Then ..
MASTER Trial Chow. JACC 2008;52:1607–15
n575 MADIT-II type patients in 50 US centres 1-EP: SCD + ICD Rx