Myocardial energy metabolism in heart failure: Systematic review and meta-analysis of Phosphorus-31 magnetic resonance spectroscopy phosphocreatine/adenosine triphosphate ratio

Figures

• 

  Fig. 1. Flowchart of study selection process

• 

  Fig. 2. (Top) Summary ORs with 95% CIs for HF per 1.0-unit decrease in PCr/ATP. Points show each study’s OR; horizontal bars are 95% CIs. The blue diamond is the pooled OR from the random-effects model; the vertical dotted line marks the pooled estimate. The red horizontal bar shows the 95% PI, i.e., the expected range of true effects in a new study. (Bottom) Summary of main results from meta-analyses. I² (percent of total variability due to between-study heterogeneity); τ² (between-study variance on the log-OR scale). For clinical scaling, the OR per 0.1-unit decrease equals 1.23. CI confidence interval, OR odds ratio, PI prediction interval, PCr/ATP phosphocreatine to adenosine triphosphate

• 

  Fig. 3. The funnel plot shows no significant publication bias, and the overall effect remains robust after trim-and-fill adjustment. (Top) Funnel plot of study OR. ORs are on the x-axis and their standard errors on the y-axis. Filled circles = observed studies; open circles = studies imputed by trim-and-fill. The vertical dotted line is the random-effects pooled OR; slanted lines are pseudo 95% limits. (Bottom) Tests of asymmetry (Egger’s test P = 0.21, Modified Egger’s test P = 0.62) did not show significant evidence of small-study effects. Trim-and-fill imputed five studies on the left; the bias-adjusted pooled OR remained significant at 5.32 (95% CI 3.15–8.97). Fail-safe N values were large, supporting the robustness of the overall effect. CI confidence interval, OR odds ratio

• 

  Fig. 4. Random-effects model—effects in two subgroups defined by magnetic field strength (1.5T vs. >1.5T). Squares show each study’s OR for heart failure per 1.0-unit decrease in myocardial PCr/ATP, with horizontal bars as 95% CIs; square size reflects random-effects weight. Diamonds indicate the pooled ORs for each subgroup and overall (random-effects); the vertical dotted line marks the overall pooled estimate. I² = % of total variability due to between-study heterogeneity within the subgroup; τ² = the between-study variance on the log-OR scale. The “test for subgroup differences” evaluates whether pooled effects differ between 1.5T and >1.5T (P shown in the plot). CI confidence interval, OR odds ratio, PCr/ATP phosphocreatine to adenosine triphosphate