The influence of pectus excavatum on biventricular mechanics: a systematic review and meta-analysis.

Abstract

Introduction: During the last decade, a small number of studies have used feature tracking (FT) cardiovascular magnetic resonance imaging (CMR) and speckle tracking echocardiography (STE) to investigate the effect of pectus excavatum (PE) on biventricular mechanics. The present systematic review and meta-analysis has been primarily designed to summarize the main findings of these studies and to examine the overall influence of PE on both left ventricular (LV)- and right ventricular (RV)-global longitudinal strain (GLS).

Evidence acquisition: All imaging studies assessing conventional indices of biventricular size and function and myocardial strain parameters in PE individuals vs.. healthy controls, selected from PubMed and EMBASE databases, were included. The risk of bias was evaluated by using the National Institutes of Health (NIH) Quality Assessment of Case-Control Studies. Continuous data (LV-GLS and RV-GLS) were pooled as a standardized mean difference (SMD) comparing PE group with healthy controls. The overall SMDs of LV-GLS and RV-GLS were calculated using the random-effect model.

Evidence synthesis: The full-text of 7 studies with a total of 374 PE individuals and 141 healthy controls were analyzed. Both average LV-GLS (-17.1±3.5% vs. -18.9±3.0%, P<0.001) and RV-GLS (-17.9±5.2% vs. -20.9±3.7%, P<0.001) were significantly lower in PE patients than controls. Subtotal SMD was small and not statistically significant for CMR studies assessing LV-GLS (-0.23, 95%CI -0.92,0.47, P=0.52) and RV-GLS (-0.33, 95%CI -0.94,0.28, P=0.28), whereas subtotal SMD was large and statistically significant for echocardiographic studies measuring LV-GLS (-1.46, 95%CI -2.55,-0.38, P=0.008) and RV-GLS (-1.71, 95%CI -2.68,-0.74, P=0.001). The overall effect of PE was statistically significant on RV-GLS (SMD -0.72, 95%CI -1.24,-0.21, P=0.006), but not on LV-GLS (SMD -0.58, 95%CI -1.17,-0.00, P=0.05). Substantial heterogeneity was found for the studies assessing LV-GLS (I²=88.2%) and RV-GLS (I²=86.9%). Egger's test gave a P-value of 0.64 for LV-GLS and 0.47 for RV-GLS assessment, indicating no publication bias. On meta-regression analysis, none of the moderators was significantly associated with effect modification for both LV-GLS and RV-GLS (all P<0.05).

Conclusions: The influence of PE on RV mechanics is greater than on LV mechanics. STE and FT-CMR may detect subtle impairment in biventricular mechanics in PE individuals. The attenuation of myocardial strain indices revealed by STE may be enhanced by methodological issues.