Visual classification of nocturnal transcutaneous CO2 and oximetry patterns in slowly progressive neuromuscular
Lucas Saulnier1, Marie Anne Melone2, Pierre Tankéré3
1Service de Physiologie et d'Explorations Fonctionnelles, AP-HP, GHU Paris-Saclay, Hôpital Raymond Poincaré, FHU UMANHYS, Garches, France.
Background:
Sleep-disordered breathing is highly prevalent in subjects with neuromuscular diseases (NMD). We hypothesized that visual analysis of nocturnal transcutaneous CO2 (TcCO2) monitoring could help physicians identify distinct respiratory phenotypes. This study aimed to derive a visual classification of nocturnal transcutaneous CO2 and oximetry patterns in patients with slowly progressive neuromuscular diseases and to assess its feasibility and inter-rater agreement.
Methods:
Retrospective study of 149 NMD subjects, naive to noninvasive ventilation, who underwent full-night TcCO2 monitoring. Three investigators independently reviewed each TcCO2 tracing and assigned a profile: absence of nocturnal hypercapnia (NH-), clusters of hypercapnic peaks only (cluNH), constant nocturnal hypercapnia without clusters (ConNH), and constant nocturnal hypercapnia associated with clusters (ConNH with cluNH). Profiles were also derived from the SpO2 tracing: absence or presence of an oxygen drop with a sawtooth profile (OD- or OD+). Overlap was defined as any form of nocturnal hypercapnia associated with OD+. Final classification was based on majority agreement, with consensus adjudication when no majority existed. Inter-rater agreement was assessed. Each profile was compared in terms of demographic data and parameters from arterial blood gases, pulmonary function testing, and respiratory polygraphy.
Results:
Overall agreement was kappa = 0.68 (95% CI, 0.61-0.75) for TcCO2 classification and kappa = 0.55 (95% CI, 0.44-0.65) for OD-/OD+, with complete agreement in 71.1% and 72.5% of recordings, respectively. Agreement was moderate for hypercapnic clusters and the sawtooth pattern. After correction, hypercapnic profiles remained associated with lower vital and supine vital capacities and higher daytime PaCO2 and bicarbonate than NH-, OD + remained associated with a higher apnea-hypopnea index.
Conclusions:
Visual classification of overnight TcCO2 and SpO2 recordings was feasible and showed moderate inter-rater agreement for hypercapnic clusters and the sawtooth oximetry pattern. The visual profiles were associated with selected physiological characteristics, but these associations primarily support face validity. Prospective external validation is required to determine whether this classification provides information beyond conventional quantitative indices or improves clinical decision-making.
