Fully automatable, reproducible, noninvasive simple plethysmographic optimization: proof of concept and potential for implantability.

Background: Hemodynamic optimization of cardiac resynchronization therapy (CRT) can be achieved reproducibly and--with bulky, nonimplantable equipment--noninvasively. We explored whether a simple photoplethysmogram signal might be used instead.

Method: Twenty patients (age 65 ± 12) with CRT underwent automatic atrioventricular (AV) delay optimization, using a multiple-transitions protocol, at two atrially paced heart rates: just above sinus rate ("slow ApVp," 77 ± 11 beats per minute [bpm]) and 100 bpm ("fast ApVp"). We then retested to assess short-term reproducibility.

Results: All 80 optimizations identified an optimum (correctly oriented parabola). At 100 bpm, the simple photoplethysmogram had wider scatter between repeat optimizations than did Finometer: standard deviation of difference (SDD) 22 ms versus 14 ms, respectively, P = 0.028. The simple photoplethysmogram improved in reproducibility when slope (instead of peak) of its signal was used for optimization, becoming as reproducible as Finometer (SDD 14 ms vs 14 ms, P = 0.50). At slow heart rate, reproducibility of simple photoplethysmogram-based optimization worsened from 14 to 22 ms (P = 0.028), and Finometer-based optimization from 14 to 26 ms (P = 0.005). Increasing the number of replicates averaged improved reproducibility. For example, SDD of simple photoplethysmogram optimization (using peak) fell from 62 ms with two replicates to 22 ms with eight replicates (P < 0.0001). At 100 bpm, the eight-replicate protocol takes ∼12 minutes.

Conclusions: A 12-minute protocol of simple photoplethysmographic AV optimization can be processed fully automatically. Blinded test-retest reproducibility of the optimum AV is good and improves with more replicates. If benefits to some patients are not to be neutralized by harm to others, endpoint studies should first test check narrowness of "within-patient error bars."