Cardiac Resynchronization Therapy (CRT)

Details

CRT restores mechanical synchrony in heart failure with dyssynchronous ventricular contraction by delivering coordinated biventricular pacing (RV + LV via coronary sinus lead) or, increasingly, via conduction system pacing (His bundle pacing/LBBAP). CRT devices are implanted as CRT-P (pacing only) or CRT-D (combined with ICD). The primary indication is symptomatic HFrEF (LVEF ≤35%) with LBBB morphology and broad QRS (≥150ms), with the strongest evidence in NYHA class II–IV on optimal GDMT.

Major RCTs: CARE-HF (NYHA III–IV, QRS ≥120ms; CV death/unplanned CV admission HR 0.63); COMPANION (NYHA III–IV, LBBB QRS ≥120ms; all-cause mortality/hospitalisation RR 0.81/0.76 for CRT-P/CRT-D); MADIT-CRT (NYHA I–II, QRS ≥130ms; HF events 34% RRR); RAFT (NYHA II–III; death/HF hospitalisation 25% RRR).

QRS morphology matters: LBBB benefits established across all major trials; non-LBBB benefit is significantly attenuated — response rates substantially lower.

Key Facts

CRT Appropriateness — AUC 2025 Ratings

• LVEF ≤35% + LBBB + QRS ≥150ms + NYHA II–IV (sinus rhythm): A(9) — strongest evidence, all major RCTs support (sources/icd-crt-auc-2025, rating: high)
• LVEF ≤35% + LBBB + QRS 120–149ms + NYHA II–III: A(7) — evidence base includes MADIT-CRT and RAFT subgroups (sources/icd-crt-auc-2025, rating: high)
• LVEF ≤35% + non-LBBB + QRS ≥150ms + NYHA III–IV: A(7) (sources/icd-crt-auc-2025, rating: high)
• LVEF ≤35% + non-LBBB + QRS ≥150ms + NYHA II: M(5) (sources/icd-crt-auc-2025, rating: high)
• LVEF ≤35% + non-LBBB + QRS 120–149ms: M(4–5) — weak evidence base; echo-guided optimisation may assist patient selection (sources/icd-crt-auc-2025, rating: high)
• LVEF ≤35% + QRS <120ms (narrow QRS): R(1) — ECHO-CRT and RethinQ demonstrated no benefit and possible harm (sources/icd-crt-auc-2025, rating: high)
• LVEF 36–50% + LBBB + QRS ≥150ms: M(4–6) — limited RCT evidence; pacing-induced cardiomyopathy subgroup may benefit most (sources/icd-crt-auc-2025, rating: high)

CRT in Atrial Fibrillation

• AF reduces CRT appropriateness by approximately 1 tier (e.g., A→M) vs sinus rhythm counterpart (sources/icd-crt-auc-2025, rating: high)
• AF + CRT + AVJ ablation (guaranteeing biventricular pacing): Restores A(7–8) — high pacing percentage required for CRT efficacy (sources/icd-crt-auc-2025, rating: high)
• Without rate control achieving near-100% biventricular pacing, CRT response substantially reduced

CRT Upgrade Scenarios

• RV pacing-induced cardiomyopathy (LVEF decline ≥10% or LVEF <50% attributable to RVP): A(7) for upgrade to CRT (sources/icd-crt-auc-2025, rating: high)
• RV pacing burden >40% + LVEF ≤50%: A(7) — pre-emptive upgrade to avoid pacing-induced dyssynchrony (sources/icd-crt-auc-2025, rating: high)

CRT Alternatives: Conduction System Pacing

• CSP-CRT (HBP/LBBAP replacing LV lead): M(5–6) in AUC 2025 — insufficient RCT data for A rating (sources/icd-crt-auc-2025, rating: high)
• LBBAP preferred over HBP for CSP due to better electrical parameters and lower threshold instability (sources/icd-crt-auc-2025, rating: high)
• Ongoing trials (PROTECT-HF n=2,600; Left vs Left n=2,136) will determine whether CSP-CRT should be elevated to A

Randomised Trial Evidence

• COMPANION (Bristow et al., NEJM 2004) — first large-scale CRT mortality RCT; CRT-D significance established: n=1,520; NYHA III/IV; LVEF ≤35%; QRS ≥120ms + PR >150ms; sinus rhythm; ischemic or non-ischemic CM; ≥1 HF hospitalisation in prior 12 months; 1:2:2 randomisation to OPT alone (N=308), CRT-P (N=617), or CRT-D (N=595); 128 US centres; median follow-up ~12–16 months; Guidant-funded; stopped early when pre-specified stopping boundaries crossed. Background OPT: diuretics + ACEi/ARB + beta-blockers + spironolactone. Implant success: 87% CRT-P, 91% CRT-D; 30-day mortality not significantly different across groups. Primary composite (death or any-cause hosp): CRT-P HR 0.81 (95% CI 0.69–0.96; P=0.014); CRT-D HR 0.80 (95% CI 0.68–0.95; P=0.010) — both reduced primary endpoint by ~20%; 12-month rate 68% OPT vs 56% CRT-P/CRT-D. All-cause mortality: CRT-D HR 0.64 (95% CI 0.48–0.86; P=0.003) — 36% reduction, significant; CRT-P HR 0.76 (95% CI 0.58–1.01; P=0.059) — borderline NON-significant. HF death/hosp: CRT-P 34% ↓ (HR 0.66; P=0.002); CRT-D 40% ↓ (HR 0.60; P<0.001). CV death/hosp: CRT-P 25% ↓; CRT-D 28% ↓. NYHA class, 6MWT, and QoL (Minnesota) all significantly better in both CRT groups at 3 and 6 months. Subgroups: wider QRS → greater primary endpoint benefit; non-ischaemic CM CRT-D mortality HR 0.50 (P=0.015); ischaemic CM CRT-D mortality HR 0.73 (P=0.082, NS; interaction NS). Key limitation: 26% OPT withdrawal (disproportionately ischaemic CM patients following MADIT-II publication mid-trial) required "reconsent" approach; open-label; stopped early. Clinical conclusion: CRT reduces hospitalisation and improves function in NYHA III/IV with prolonged QRS; adding ICD to CRT provides statistically significant incremental mortality benefit that CRT-P alone does not achieve at this follow-up duration. (sources/crt-companion-nejm-2004, rating: very high)

• CARE-HF (Cleland et al., NEJM 2005) — definitive CRT-P mortality trial: n=813; NYHA III/IV; LVEF ≤35%; QRS ≥120ms (echo dyssynchrony criteria required for QRS 120–149ms: ≥2 of aortic pre-ejection delay >140ms, interventricular mechanical delay >40ms, or delayed posterolateral LV activation); CRT-P (no defibrillator) vs medical therapy; 82 European centres; mean 29.4 months. Primary composite (death or unplanned CV hospitalisation): HR 0.63 (95% CI 0.51–0.77; P<0.001); CRT 39% vs medical 55%. All-cause mortality: HR 0.64 (95% CI 0.48–0.85; P<0.002); 20% vs 30%; NNT≈9 (9 devices to prevent 1 death and 3 HF hospitalisations). HF hospitalisations: 18% vs 33%. Core-lab echo at 3m and 18m: ↑LVEF, ↓LVEDVI, ↓mitral regurgitation, ↓interventricular mechanical delay, ↓NT-proBNP. First trial to demonstrate significant CRT mortality benefit — prior COMPANION showed CRT-P mortality trend only (P=0.06 at 16 months). Residual SCD: 7% of CRT patients (29/409; 35% of CRT deaths) died suddenly despite biventricular pacing — the mechanistic basis for preferring CRT-D. (sources/crt-carehf-nejm-2005, rating: very high)

• MADIT-CRT (Moss et al., NEJM 2009) — CRT benefit extended to NYHA I/II: n=1,820; LVEF ≤30%; QRS ≥130ms; NYHA I/II ischaemic or NYHA II non-ischaemic; sinus rhythm; all ICD-eligible; CRT-D vs ICD alone; 3:2 ratio; 110 centres; mean 2.4 years; Boston Scientific-funded; stopped early at 9th of 20 planned analyses. Primary composite (death or nonfatal HF event): HR 0.66 (95% CI 0.52–0.84; P=0.001); 17.2% vs 25.3%. Benefit driven entirely by 41% reduction in HF events; no mortality difference (3% annual mortality each arm). Critical QRS interaction (prespecified; P=0.001): QRS ≥150ms HR 0.48 (0.37–0.64) — strong benefit; QRS 130–149ms HR 1.06 (0.74–1.52) — no benefit. Sex interaction (prespecified; P=0.01): women HR 0.37 vs men HR 0.76. No aetiology interaction (ischaemic = non-ischaemic). LV reverse remodeling at 1 year: ↓LVEDV, ↓LVESV, ↑LVEF all significantly greater in CRT-D. Foundational trial for Class I CRT in NYHA II; QRS ≥150ms interaction underpins the modern ≥150ms threshold. (sources/crt-maditcrt-nejm-2009, rating: very high)

LVAD + CRT-D

• De novo CRT-D at LVAD implantation: R(2–3) — LV lead mechanical function is redundant with LVAD support (sources/icd-crt-auc-2025, rating: high)

Contradictions / Open Questions

• CRT-P vs CRT-D — COMPANION borderline CRT-P mortality vs significant CRT-D; CARE-HF residual SCD: COMPANION (median ~15 months) showed CRT-P mortality HR 0.76 (P=0.059, NS) while CRT-D achieved HR 0.64 (P=0.003, significant) — hospitalisation benefit was equivalent for both. CARE-HF (29 months follow-up; CRT-P only) subsequently confirmed a significant CRT-P mortality benefit (HR 0.64; P<0.002). The COMPANION CRT-P result is likely a follow-up duration artefact rather than a fundamental CRT-P inefficacy — CARE-HF's longer follow-up allowed the CRT-P mortality benefit to emerge. However, CARE-HF also showed that 7% of CRT-P patients still died suddenly (SCD = 35% of all CRT deaths). The inference is that CRT resynchronisation reduces all-cause mortality over time (confirmed by CARE-HF) but does not eliminate the arrhythmic fraction — the ICD component of CRT-D addresses this residual risk. A direct head-to-head RCT of CRT-D vs CRT-P adequately powered for all-cause mortality has not been conducted. Current guidelines preferentially recommend CRT-D in patients also meeting primary prevention ICD criteria (LVEF ≤35%; NYHA II–IV). (sources/crt-companion-nejm-2004, sources/crt-carehf-nejm-2005, rating: very high)
• MADIT-CRT QRS 130–149ms: no benefit despite broad entry criterion: MADIT-CRT enrolled QRS ≥130ms but the critical prespecified interaction (P=0.001) showed benefit only at QRS ≥150ms (HR 0.48) with no benefit at QRS 130–149ms (HR 1.06; point estimate favours ICD alone). The trial entry criterion was broader than the population that benefited. This interaction is the principal evidence basis for the modern QRS ≥150ms Class I threshold; QRS 120–149ms recommendations rely on MADIT-CRT subgroups, RAFT data, and meta-analyses — not direct primary-endpoint evidence. The benefit at QRS 130–149ms may be further attenuated in non-LBBB morphology (not separately analysed in the primary paper). (sources/crt-maditcrt-nejm-2009, rating: very high)
• MADIT-CRT stopped early — no mortality benefit and potential effect overestimation: Trial stopped after mean 2.4 years at 9th of 20 planned analyses; early stopping typically overestimates the true effect magnitude. The primary benefit is HF event reduction, not survival; the 3% annual mortality was identical in both arms at the time of stopping. Long-term follow-up analyses (Goldenberg et al. 2014) subsequently showed mortality benefit emerging with longer follow-up — suggesting the 2.4-year primary data underrepresent the full clinical effect rather than overestimate it, but the stopping rule remains a methodological caveat. (sources/crt-maditcrt-nejm-2009, rating: very high)
• CARE-HF NYHA III/IV only — extension to NYHA II based on subsequent trials: CARE-HF enrolled NYHA III/IV; NYHA II benefit was established by later trials (MADIT-CRT 2009; REVERSE). Current Class I guidelines recommend CRT for NYHA II–IV, but the CARE-HF foundational mortality evidence applies to NYHA III/IV; NYHA II extrapolation rests on MADIT-CRT and REVERSE, not CARE-HF. (sources/crt-carehf-nejm-2005, rating: very high)
• CARE-HF QRS criterion vs modern LBBB requirement: CARE-HF required QRS ≥120ms with echo dyssynchrony criteria — not LBBB morphology specifically. The current Class I LBBB requirement derives from meta-analyses and post-hoc trial analyses showing attenuated CRT response in non-LBBB; modern trial evidence distinguishes LBBB from non-LBBB more sharply than CARE-HF did. The AUC-based LBBB-first framework extends beyond what CARE-HF directly tested. (sources/crt-carehf-nejm-2005, sources/icd-crt-auc-2025, rating: very high / high)
• Non-LBBB CRT benefit: Benefit in non-LBBB is substantially attenuated across all RCTs; AUC rates NYHA III–IV non-LBBB QRS ≥150ms as A(7), but individual response is heterogeneous. Echo-based dyssynchrony assessment and EPS-guided QRS classification may improve patient selection.
• QRS 120–149ms non-LBBB: Rated M(4–5) but represents a large clinical population. Mechanical dyssynchrony assessment tools (3D echo, CMR feature tracking) have not been validated in RCTs for CRT selection.
• CSP-CRT vs biventricular CRT: Physiologically superior conduction system activation does not yet have RCT outcomes data. The superiority assumption is based on acute haemodynamic and QRS narrowing data only. (concepts/Conduction-System-Pacing)
• CRT in AF: Near-universal biventricular pacing required for efficacy, yet AF rate control is imperfect. AVJ ablation mandatory for A-level appropriateness — this is a significant intervention not all patients or centres pursue.

Connections

• Related to entities/ICD — combined as CRT-D; most HFrEF patients with CRT indication also meet ICD criteria
• Related to entities/Heart-Failure — primary indication; HFrEF with broad QRS
• Related to entities/Atrial-Fibrillation — AF reduces CRT appropriateness; AVJ ablation restores it
• Related to concepts/Conduction-System-Pacing — CSP-CRT as alternative to biventricular pacing; HOT-CRT/LOT-CRT
• Related to concepts/His-Bundle-Pacing — HBP as CRT alternative
• Related to concepts/Left-Bundle-Branch-Area-Pacing — LBBAP as CRT alternative; LOT-CRT
• Related to concepts/Secondary-Mitral-Regurgitation — CRT can reduce functional MR in ventricular SMR

Sources

• sources/icd-crt-auc-2025 — ACC/AHA/HRS 2025 AUC (primary source)
• sources/crt-companion-nejm-2004 — COMPANION; first large CRT mortality RCT; CRT-D HR 0.64 (P=0.003); CRT-P HR 0.76 (P=0.059 NS); NYHA III/IV; QRS ≥120ms
• sources/crt-carehf-nejm-2005 — CARE-HF; definitive CRT-P mortality RCT; HR 0.64 (P<0.002); NNT≈9; NYHA III/IV
• sources/crt-maditcrt-nejm-2009 — MADIT-CRT; CRT-D in NYHA I/II; QRS ≥150ms threshold; HF events HR 0.66; no mortality difference at 2.4 years