By Michael Crawford, MD, Editor
SYNOPSIS: The echocardiographic diagnosis of left ventricular non-compaction is difficult, and experienced readers disagree frequently. Careful attention to suggested criteria and the use of other imaging modalities in difficult cases resolves most diagnostic disagreements.
SOURCES: Stöllberger C, et al. Interobserver agreement of the echocardiographic diagnosis of LV hypertrabeculation/noncompaction. JACC Cardiovasc Imaging 2015;8:1252-1257.
Pinto FJ. When and how to agree in disagreeing on the diagnosis of noncompaction by echocardiography? JACC Cardiovasc Imaging 2015;8:1258-1259.
The diagnosis of left ventricular (LV) non-compaction has profound implications for patients, yet no uniform criteria for its diagnosis by echocardiography exist. Investigators from Austria and Germany studied the interobserver agreement in two laboratories using the same criteria. Potentially afflicted patient echoes were mixed with controls who were age- and LV systolic function-matched. Three readers from two labs were blinded to the identity of the echoes and the other readers’ initial opinion. Patients with other congenital cardiac conditions, recent myocardial infarction, and aortic valve disease were excluded. In brief, the criteria were: more than three prominent trabecular formations; two layered myocardiums with a heavily trabeculated endocardial layer and a dense epicardial layer; a non-compacted to compacted myocardium ratio of > 2:1; and contrast or color flow evidence of perfusion in the intratrabecular spaces. The three reviewers discussed discordant opinions. Echoes from 100 patients, of which 51 had received the diagnosis of LV non-compaction (NC), were reviewed. LV ejection fraction ranged from 4-88%. Agreement between the three reviewers occurred in 65% of the cases. Review of the 35 discordant cases resulted in agreement in 24 while 11 remained questionable. In those with an initial clinical diagnosis of LVNC (51 patients), only 27 had agreement between the three observers. Of the remaining 24, all agreed there was no LVNC in three, there was LVNC in eight, and 10 remained questionable. Among the 49 not diagnosed as LVNC, only in 33 did all three observers agree there was no LVNC. Of the remaining 16, all agreed that two had LVNC. The remaining 14 were reviewed and LVNC was excluded in 13 and one remained questionable. The authors concluded that there is considerable interobserver disagreement in the diagnosis of LVNC using uniform criteria, which can be reduced by mutual review, but in 11% of cases, the diagnosis remains questionable.
COMMENTARY
Since it was first described in 1984, the diagnosis of this rare condition has increased. Some believe it is being overdiagnosed now. This is an important issue, because this condition has been associated with malignant arrhythmias, sudden death, and heart failure. It may affect patients’ careers, e.g., competitive athletes and their subsequent management or, e.g., defibrillator placement. Even in those with mild cases with normal global LV function, there can be psychological harm with this diagnosis.
Small studies in single labs have suggested considerable variability between observers in the diagnosis of LVNC. These investigators sought to assess the variability between two experienced labs in different countries who used the same criteria. They found that in those in whom LVNC was diagnosed on their routine clinical echo, the reviewers agreed with the diagnosis in 53%. Even more disturbing, there was only 67% agreement with the diagnosis of the absence of LVNC. After a re-review and discussion among the reviewers, 11% remained questionable. Agreement was hampered by poor images, small LVs with normal function, aberrant bands, false chordae, and unusual papillary muscles. Not only is it difficult to distinguish from normal variants, LVNC can be seen in any cause of LV hypertrophy or dilatation. In this study, the authors excluded patients with aortic valve disease and recent myocardial infarction.
There were limitations to this study. The reviewers only had access to the recorded studies, so they couldn’t request additional views, contrast use, or 3-D images. There are no data on intraobserver reproducibility. Also, the anatomical location of the observed abnormalities was not considered. In addition, other imaging modalities use was not reported and could have improved diagnostic accuracy. However, there is no accepted gold standard for the diagnosis of LVNC and there are insufficient pathological data to validate imaging methods.
European Society of Cardiology President Faust Pinto’s accompanying editorial suggests that in questionable cases, contrast echo, 3-D transesophageal echo, or MRI should be used to help clarify. However, there are no universally accepted criteria for LVNC by these studies. One MRI study of normal subjects found LVNC in 91% at the apex and 78% at the mid-wall. This suggests that there is a spectrum from normal variants to severe LVNC. The latter subjects usually have low LV systolic performance, so clinical decisions are easier in them. The subject with normal LV systolic function is a greater challenge, and care must be taken not to overdiagnose this condition.