Conduction system pacing · Device troubleshooting

Phrenic vs. left bundle capture thresholds in LBBAP: strength-duration physiology and nocturnal diaphragmatic twitching

The requested ratio does not exist in the literature. What does exist is a strength-duration argument that explains when pulse width can help, a hard ceiling that explains when it cannot, and a far more likely explanation for why the twitching happens only at night.

The honest starting point: this dataset does not exist

No published dataset reports phrenic-to-left-bundle capture threshold ratios across pulse widths, stratified by posture, in left bundle branch area pacing. Producing a strength-duration table that looks authoritative would be a disservice, because the numbers would be invented.

Phrenic nerve stimulation is well characterised in the context of left ventricular and cardiac resynchronisation leads. In LBBAP it is described mainly in case reports and remains under-characterised. The clinically meaningful ratio is lead-specific, position-specific, and patient-specific — it is a measurement, not a lookup.

What follows is what is genuinely established, where the reasoning holds, and where it becomes speculative.

Strength-duration physiology: why pulse width is a lever at all

Both nerve and myocardium obey the same hyperbolic relationship, described by Lapicque:

• Rheobase is the asymptotic threshold amplitude at infinitely long pulse width. It reflects how excitable the tissue is at the prevailing lead-to-tissue distance.
• Chronaxie is the pulse width at which threshold equals twice rheobase. It is a time constant — a measure of how quickly the tissue integrates charge.

The entire manoeuvre rests on one fact: large myelinated nerve has a substantially shorter chronaxie than working myocardium. Roughly 0.1–0.2 ms for phrenic nerve against roughly 0.3–0.7 ms for ventricular myocardium. Skeletal muscle sits between the two.

Geometrically, this means the neural curve flattens early. Beyond approximately 0.4 ms it already sits near its rheobase and gains little from further widening. The myocardial curve is still descending at 1.0–1.5 ms. As the pulse widens, myocardial threshold continues to fall while phrenic threshold has largely bottomed out, and a gap opens between them.

The rheobase ceiling: the cautionary half of the argument

The ratio at any pulse width is:

As pulse width tends to infinity, the second term tends to 1 and the ratio converges on the ratio of the two rheobases. This yields three consequences that matter more than any tabulated number.

• If phrenic rheobase lies below myocardial rheobase — that is, the lead is effectively closer to the diaphragm than to the bundle — then no pulse width in existence resolves the problem. The curves cross at every width. This is a lead position problem, not a programming problem.
• If phrenic rheobase is only modestly higher, the resulting window is real but frequently too narrow to programme with a safety margin. A 0.2 V window is unusable against day-to-day threshold drift.
• The gains are front-loaded. Most of the achievable separation appears by approximately 0.8–1.0 ms. Moving from 1.0 to 1.5 ms buys very little while meaningfully increasing current drain and shortening device longevity.

Two caveats to the model

1. It assumes neural capture. If the culprit is direct diaphragmatic muscle capture rather than phrenic nerve capture, the chronaxie gap narrows considerably — skeletal muscle sits far closer to myocardium — and the pulse-width manoeuvre largely evaporates.
2. The real curves are not stationary. Posture and respiration displace the phrenic curve vertically by changing distance. A window measured supine at rest may close during a particular phase of the respiratory cycle. This is precisely why a bench-derived ratio would not help even if one existed.

Posture and the nocturnal pattern

The postural component of the question is well grounded and mirrors what is documented for left ventricular and resynchronisation leads. Postural phrenic capture is a recognised phenomenon: diaphragmatic stimulation appears on lying down that is absent when upright.

Supine and lateral positions alter the geometry between lead, phrenic nerve, and diaphragm. When supine, abdominal contents displace the diaphragm cephalad, functional residual capacity falls, and the diaphragm sits higher — closer to a septal pacing source.

Respiration does modulate the threshold dynamically: capture can wax and wane across the respiratory cycle as anatomy shifts millimetre by millimetre. Whether the net inspiratory effect raises or lowers threshold depends on whether the culprit is true phrenic nerve capture, which is relatively fixed anatomically, or direct diaphragmatic capture, which is far more geometry-dependent.

An anatomical objection worth raising first

A mid-septal LBBAP lead is an unlikely place from which to capture the phrenic nerve. The right phrenic nerve descends along the superior vena cava and lateral right atrium. The left phrenic courses over the left ventricular lateral wall and pericardium. Neither approaches the mid-interventricular septum.

Where diaphragmatic twitching genuinely arises from an LBBAP lead, the likelier explanations, in rough order, are:

• The lead sits more apically and inferiorly on the septum than intended, near the diaphragmatic surface of the heart. This is direct diaphragmatic capture, not phrenic capture.
• Septal perforation into the left ventricular cavity.
• A different lead is responsible. In a resynchronisation system the left ventricular lead is a far likelier source and should be excluded first.
• Insulation breach or conductor fracture producing extracardiac stimulation.
• The twitching is not device-related at all.

A programming decision tree

Step 1 — Establish that the device is responsible

• Confirm the twitch is 1:1 with pacing output against the surface ECG or telemetry.
• Drop output below threshold briefly, with backup available, or inhibit pacing. Does it stop?
• If it does not track the paced rate, the differential is elsewhere entirely: hiccups, diaphragmatic myoclonus, or diaphragmatic flutter.

Step 2 — Localise the lead and the electrode

• With multiple leads, disable each in turn.
• Switch polarity. LBBAP leads are frequently programmed bipolar with the ring in the right ventricular septum or cavity, and anodal ring capture is real — sometimes intentional. Testing tip-to-can unipolar against tip-to-ring bipolar can identify the culprit and supply the fix in a single move.
• Interrogate impedance and threshold trends. Rising or erratic impedance suggests insulation breach; a step change in threshold suggests micro-dislodgement.
• Chest radiography or fluoroscopy compared against implant imaging, assessing lead depth and how inferior or apical the septal entry site truly is.

Step 3 — Programming levers, in order of yield

1. Reduce output. Devices are commonly programmed with generous headroom. A bundle threshold of 0.6 V paced at 2.5 V leaves considerable room to move.
2. Change the vector. The highest-yield single manoeuvre where anodal capture is involved.
3. Only then exploit the strength-duration curve. Measure the actual threshold pair at two or three pulse widths, in the posture that provokes the symptom, and determine whether a window exists. This is done in the patient, never from a table.

Step 4 — The step most often skipped

Where atrioventricular conduction is intact, the elegant answer is not to pace:

• Extend the AV delay or enable an intrinsic-conduction-preference algorithm (AAI↔DDD mode switching, AV search hysteresis) so native conduction resumes.
• Lower the night rate or rest rate. Nocturnal twitching frequently reflects a simple mechanism: the intrinsic sinus rate falls below the programmed lower rate limit overnight, so ventricular pacing begins at night and ceases by morning. The postural narrative may be a red herring — the twitching does not worsen when supine because of diaphragm geometry, it appears because that is when the device took over.

This hypothesis should be tested before any investment in strength-duration work. Review the percentage of ventricular pacing by time of day in the device diagnostics. If ventricular pacing is minimal during waking hours and near-continuous between midnight and 06:00, the answer lies in lower rate limit and AV delay programming rather than in capture physics.

Step 5 — Where programming fails

Lead revision. Where the rheobase relationship is unfavourable — the crossed-curves scenario — no amount of programming will separate the two thresholds, and continued attempts merely postpone the inevitable while sleep remains disrupted.

Summary

• The requested ratio does not exist in published form and would not be useful if it did, because it is lead-, position-, and posture-specific.
• Widening the pulse width favours myocardial over neural capture, because nerve chronaxie is shorter than myocardial chronaxie — but the rheobase ratio is an absolute ceiling.
• Most of the achievable separation arrives by 0.8–1.0 ms. Widths beyond that cost longevity and buy little.
• The inspiratory hypothesis is likely reversed: inspiration moves the diaphragm away from a septal source. The tonic supine change is the better explanation.
• True phrenic capture from a mid-septal lead is anatomically improbable; direct diaphragmatic capture from an inferoapical septal position is likelier.
• Check the ventricular pacing histogram by time of day first. Nocturnal-only symptoms often mean nocturnal-only pacing.