Abstract
The proposed causal chain — improved daytime cardiac output → better N3 and sleep efficiency, mediated by an evening sympathovagal shift — has never been formally tested in an LBBAP cohort and only partially in CRT populations. Borrowed CRT evidence suggests the sleep benefit of resynchronisation runs predominantly through reduced central sleep apnoea and shortened circulation time, with objective deep-sleep architecture moving least and the autonomic correlate being baroreflex sensitivity rather than a global spectral index. In a high-vagal-tone, PICM-upgrade phenotype the sleep-disordered-breathing pathway is largely absent, which paradoxically makes any deep-sleep gain more attributable to the direct autonomic mechanism. Testing that mechanism, however, fails with conventional LF/HF HRV, which is invalid during atrial-pacing epochs. We propose a phase-rectified deceleration-capacity (PRSA-DC) mediator, gated to atrial-sensed beats, as the defensible way to decompose the loop, and outline an n-of-1 crossover using lower-rate-limit manipulation as the intervention.
01 / THE QUESTIONStating the hypothesis precisely
The clinical intuition is attractive: restore physiological activation with LBBAP, recover stroke volume and synchrony by day, and the heart should be "calmer" by night — lower sympathetic drive at bedtime, deeper slow-wave sleep, fewer arousals, better efficiency. If true, sleep would become both a marker and a mechanism of successful resynchronisation.
But that intuition bundles three distinct claims into one sentence. First, a correlation: does the magnitude of daytime output improvement track polysomnographic N3 and sleep efficiency? Second, a mediation: is that relationship carried by a favourable evening sympathovagal shift, rather than by some parallel pathway? Third, an implicit direction: that daytime haemodynamics drive night-time sleep, and not the reverse. Each claim needs separate evidence, and the three variables are entangled in a 24-hour feedback loop that makes naïve correlation nearly uninterpretable.
02 / THE EVIDENCEWhat the borrowed data actually support
There is no LBBAP-specific sleep-mediation dataset. The nearest mature evidence is the CRT / heart-failure literature, and it delivers a mixed verdict for the hypothesis. Cardiac resynchronisation robustly reduces central sleep apnoea and Cheyne–Stokes respiration and improves subjective sleep quality and daytime sleepiness. When investigators looked at objective architecture, however, the signal largely evaporated: sleep-stage distribution showed little significant change, with at most a trend toward increased slow-wave sleep under overdrive pacing.
Crucially, the dominant mediator of the CRT sleep benefit was not autonomic in the sense the hypothesis assumes. The fall in apnoea–hypopnoea index correlated tightly with a shortened circulation time — a loop-gain / chemoreflex mechanism. Where autonomics did enter, the relevant variable was baroreflex sensitivity, with the strongest correlations tied to reductions in the central apnoea index — not a global LF/HF ratio. The honest reading: resynchronisation improves sleep mostly by stabilising ventilatory control, the autonomic arm that matters is baroreflex-mediated, and deep-sleep percentage specifically is the endpoint that moves least.
Transplanting CRT-HFrEF-CSA findings onto a preserved-EF, physiologically-paced patient is an extrapolation, not evidence. It is used here to bound plausibility, not to establish it.
03 / THE CHAINThe three links, taken individually
- Daytime output → evening sympathovagal shift. The strongest link. Restored synchrony raises stroke volume and improves baroreceptor loading; LBBAP produces a larger acute systolic-BP reduction than RV pacing, consistent with reduced sympathetic outflow and better baroreflex buffering. Combined with the reverse-remodelling and AF-suppression signal that separates LBBAP from RVP, there is a coherent rationale for a downward evening sympathetic drift. Direct HRV data in LBBAP cohorts, however, remain essentially absent — this is inference from haemodynamics.
- Evening sympathovagal shift → deep sleep. Grounded independently of pacing. N3 generation depends on the sleep-onset withdrawal of sympathetic tone and the nocturnal vagal surge; elevated pre-sleep sympathetic activity fragments sleep and suppresses slow-wave. A genuine favourable shift should therefore facilitate N3 and efficiency — but the CRT data suggest this link is quantitatively small once ventilatory control is accounted for.
- Deep sleep → next-day output. Here the framing turns circular. N3 is itself a period of blood-pressure dipping, parasympathetic dominance and autonomic restoration; it feeds back onto next-day tone and haemodynamics. The three variables are not a clean X→M→Y chain but a closed loop, all co-determined by the underlying degree of resynchronisation. Cross-sectional correlation cannot assign direction.
04 / THE INVERSIONWhy a high-vagal athlete phenotype flips the reading
This is the part worth dwelling on. The CRT evidence attributes most of its sleep benefit to the central-sleep-apnoea / circulation-time pathway. A PICM-upgrade athlete — chronically low resting rate, high vagal tone, an ejection fraction that dipped into the mid-40s and recovered after physiological pacing — does not carry that severe sleep-disordered-breathing substrate. The dominant confounding pathway of the entire CRT literature is therefore largely absent.
The consequence is counter-intuitive: because the competing ventilatory-control pathway is not in play, any genuine post-upgrade gain in N3 or efficiency becomes more attributable — not less — to the direct haemodynamic/autonomic mechanism the hypothesis proposes. Such a phenotype is, in effect, a cleaner (if n-of-1) test of the hypothesis than the entire CRT dataset, precisely because it strips out the loop-gain confounder.
05 / TWO OUTCOMESEfficiency and deep sleep have different mediators
The question bundles two polysomnographic outcomes that almost certainly do not share a pathway, and separating them sharpens the whole analysis.
| Outcome | Most plausible mediator | Pathway class |
|---|---|---|
| Sleep efficiency (TST/TIB) | Reduced nocturnal congestion, orthopnoea, PND; improved renal perfusion and reduced nocturia; fewer respiratory arousals | Haemodynamic / congestion |
| N3 / slow-wave % | Withdrawal of evening sympathetic tone; nocturnal vagal surge; reduced arousal pressure | Autonomic |
Collapsing both into a single "sleep quality" correlate blurs a congestion mechanism and an autonomic mechanism that may move in different directions and on different timescales. The autonomic mediation hypothesis is properly a hypothesis about N3 specifically; efficiency should be modelled separately, with a congestion covariate.
06 / THE FRAMEWORKA PRSA-DC–gated mediation model
Here is the crux, and it is specific to the paced heart. To test the mediation you must measure "evening sympathovagal balance." The reflexive choice — frequency-domain HRV — is the wrong instrument here, and understanding why is the whole methodological point.
6.1 Why LF/HF fails in a paced heart
Spectral HRV is interpretable only during atrial-sensed, ventricular-paced (As-Vp) epochs, where RR variability still reflects sinus-node autonomic modulation. On any night when the intrinsic atrial rate falls below the programmed lower rate limit — highly likely in a high-vagal athlete during the nocturnal rate trough — atrial pacing engages and the RR series becomes metronomic. LF/HF is then not merely noisy but invalid, and it fails precisely during the deepest-sleep, highest-vagal windows the hypothesis cares most about. A naïve overnight spectral mediator is therefore contaminated exactly where it must be clean.
6.2 Gate the analysis to atrial-sensed beats
The first requirement is epoch gating: restrict all autonomic computation to As-Vp segments, excluding any window containing atrial-paced beats, mode-switch episodes, ventricular ectopy, or PVARP artefact. This demands beat-level annotation aligned to device marker channels — not just the surface ECG — so the atrial channel behaviour (sensed vs paced) is known for every cardiac cycle entering the model.
6.3 Why PRSA / deceleration capacity survives
Phase-rectified signal averaging solves what spectral analysis cannot. It selects anchor points at RR decelerations, aligns fixed-length segments around them, and averages — extracting quasi-periodic, vagally-weighted structure while suppressing non-stationarity and artefact by design. Deceleration capacity (DC) is the resulting vagal index; acceleration capacity (AC) its counterpart. Because DC is phase-rectified and averaged over many anchors, the occasional paced or excluded beat degrades it far less than it destroys a spectral ratio — and gated to As epochs it becomes the most defensible autonomic mediator available in this setting. It is also a validated post-MI risk marker, so the metric carries independent prognostic meaning.
Use evening/early-night PRSA-DC, gated to atrial-sensed epochs, as the mediator variable — not LF/HF. Pair it with DFA α1 as a complexity index and, where instrumentable, pre-ejection period or skin sympathetic activity as the sympathetic arm.
6.4 The mediation model
With a robust mediator in hand, the decomposition becomes tractable. The variables:
Because the three principal variables form a closed loop, ordinary cross-sectional mediation is under-identified. Two design features break the deadlock: within-subject repeated measures across many nights, and an exogenous manipulation that perturbs the pathway without touching the outcome directly.
6.5 The n-of-1 crossover: lower rate limit as the intervention
The natural manipulation is the programmed lower rate limit (50 vs 60 bpm), alternated in randomised blocks. It is ideal because it moves two things at once in a known direction: the nocturnal atrial-paced fraction (the mediator-measurement confounder) and the night-time haemodynamic floor. A block design — several nights at each setting, washout between, blinded scoring of PSG — approximates the device-crossover logic used in CRT sleep trials, but at n-of-1 resolution and with the autonomic mediator measured correctly.
| Element | Specification |
|---|---|
| Design | Randomised block crossover, LRL 50 ↔ 60 bpm, ≥5 nights/block, ≥2 washout nights |
| Primary outcome | N3 % (PSG, AASM-scored, blinded) |
| Secondary | Sleep efficiency; arousal index; DC/AC; DFA α1 |
| Mediator | Evening PRSA-DC, As-Vp–gated, pre-sleep + first NREM cycle |
| Exposure | Standardised daytime CO/SV surrogate, fixed clock window |
| Analysis | Within-subject causal-mediation with LRL block as instrument; sensitivity analysis on paced-fraction |
6.6 Confounders that must be gated, not adjusted
Some confounders cannot be handled statistically because they corrupt the measurement itself. The atrial-paced fraction is the chief example: it does not merely bias the mediator estimate, it invalidates spectral computation for the affected beats — which is why it is handled by epoch exclusion, not covariate adjustment. Respiratory rate and position modulate DC and must be logged at beat resolution. Core-temperature phase sets the circadian envelope of both autonomic tone and slow-wave propensity, so PSG timing should be anchored to habitual sleep onset, not clock time.
07 / SYNTHESISWhat can honestly be claimed
To answer the two questions directly. The correlation between daytime output gain and polysomnographic sleep is biologically plausible and partially supported for sleep efficiency, but weakly supported for deep sleep specifically. The mediation by evening sympathovagal balance is plausible but, on the best available evidence, likely secondary to a ventilatory-control / baroreflex pathway in typical CRT patients — while in a preserved-EF, high-vagal, physiologically-paced phenotype the autonomic mediator is more likely to dominate, simply because the competing pathway is absent.
No one has measured this cleanly, and the reason is instrumental: the mediator is genuinely hard to measure in a paced heart. A PRSA-DC index gated to atrial-sensed epochs, embedded in a within-subject crossover that uses lower-rate-limit manipulation as the exogenous lever, is the most defensible way to move the question from mechanistic story to measured effect.