Parasternal intercostal muscle recruitment and its resolution after unilateral phrenic nerve injury during mediastinal tumour resection: a prospective observational study

Abstract Background Unilateral phrenic nerve injury during extended mediastinal tumour resection can cause post-operative respiratory failure. How the parasternal intercostal muscles respond, and whether respiratory support alters this, is unknown. We followed diaphragmatic excursion and parasternal intercostal thickening fraction (TFic) during the intensive care unit (ICU) stay. Methods Single-centre prospective observational study (October 2023–July 2024). We enrolled consecutive adults admitted to the ICU after sternotomy for mediastinal tumour resection with an intra-operatively documented phrenic nerve injury or resection and ultrasound-confirmed ipsilateral diaphragmatic dysfunction. Bilateral excursion, TFic, diaphragmatic thickening fraction (TFdi) and respiratory rate were measured within 12 h of ICU admission, before any respiratory support, and at ICU discharge, after weaning. Patients receiving high-flow nasal oxygen or non-invasive or invasive ventilation formed the respiratory support (RS) group; the others, the conventional oxygen (COT) group. Trajectories were analysed with mixed-effects models and analysis of covariance (ANCOVA), with the contralateral hemithorax as an internal control. Stored images were re-measured by a blinded second observer. Results Of 40 patients, 19 (47.5%) required respiratory support. Median affected-side excursion was 0.30 versus 1.58 cm contralaterally ( P < 0.001) and barely changed. TFic did not differ between sides at either examination. On the affected hemithorax, TFic fell in the RS group and rose in the COT group (group × time ratio 0.41, 95% CI 0.26–0.64); adjusted for the admission value and ICU stay it remained lower at discharge in the RS group (ANCOVA ratio 0.44, 0.28–0.70), with no difference contralaterally (0.97, P = 0.919), although the test for a between-side difference was inconclusive ( P = 0.119). TFdi on the denervated side was not reduced and was uncorrelated with excursion. Inter-observer ICC(2,1) was 0.97–0.99 for excursion and intercostal thickness and 0.90 (0.74–0.95) for TFic. Conclusions TFic did not identify the side of the lesion; its relative fall in supported patients, more marked on the affected side, is compatible with reduced rib-cage muscle recruitment as respiratory failure resolved, but effort was not measured and causality cannot be inferred. Excursion, not TFdi, should be used to assess the denervated hemidiaphragm. Trial registration ClinicalTrials.gov NCT06375538; retrospectively registered in April 2024.