Ultrasound Elastography for Differentiating Benign from Malignant Thickened Greater Omentum: Reality or artifact?
by Leonhard Gruber, Hannes Gruber, Thomas Auer, Alexander Loizides (Alexander.Loizides@i-med.ac.at)Zhang Y, Wang X, Tao C, et al. (2016) Ultrasound Elastography for Differentiating Benign from Malignant Thickened Greater Omentum. Eur Radiol 26: 2337-2343
With great interest we read the article by Zhang et al.  about the use of strain elastography (SE) to differentiate between benign and malignant causes for thickening of the greater omentum.
Aside from a curious way to calculate strain ratios in a retrospective fashion, two crucial aspects have to be mentioned, which cast strong doubt on the validity of the presented data. If the examples in Fig. 1c/d and 3 allow for any inference on the actual study cases, then the method of SE must be assumed to have been applied incorrectly.
SE is successfully used in examinations of the prostate and breast [2,3] and increasingly tested in other fields. SE is based on the principle of tissue elasticity during repeated compression and release by the examiner . Software algorithms calculate a non-quantified compression modulus through comparison of relative pixel shifts within the tissue examined, which is then expressed as a colour-coded overlay . This means that force has to be transmitted vertically through the tissue of interest, ideally in a very homogeneous way. To achieve pseudo-quantification the stiffness of two regions of interest (ROI) can be compared and a strain ratio can be calculated [3,5]. To reduce the influence on pressure inhomogeneity, the examiner should select ROIs in the same depth to ensure comparable conditions . This obviously has not been considered, as is illustrated by Fig. 2 and 3. Here reference ROIs were placed very superficially into subcutaneous fat and the abdominal wall muscles, while target ROIs can be found deep within the abdominal cavity.
The second, more crucial flaw disregards the physical laws of fluid dynamics: as long as fluids are able to flow freely (i.e. are not confined to rigid containers as syringes) they will not transmit directional pressure. There will only be an undirected increase of pressure within the fluid. Thus no directed force can be exerted on any body immersed in fluid without direct contact. This misleads the above mentioned software algorithms to interpret the lack of compression of masses surrounded by fluid (such as the omental lesions surrounded by ascites in Fig. 1c/d and 3) as having high stiffness!
Although all patients were described as having ascites, the figures curiously illustrate direct contact of the abdominal wall with the suspicious omental mass only in benign cases (Fig. 1a/b and 2), while a clear separation of the two compartments by intraperitoneal fluid can be observed in malignant cases (Fig. 1c/d and 3). Interestingly the authors even mention the one case of peritoneal tuberculosis where a false positive classification was caused by massive ascites. It thus may be argued that the observed stiffness in malignant omental lesions is rather an expression of a greater amount of ascites.
Malignant tumours may be ‘stiffer’ than their benign counterparts, but it is doubtful that the findings by Zhang et al. reflect actual pathologic changes rather than methodical errors. In fact a very strong suspicion of malignancy should already be expressed with conventional ultrasonography in such cases: Tumorous omental formations accompanied by significant free intraperitoneal fluid are highly suggestive of a malignant process . On the other hand, smaller nodular lesions of the peritoneal surface as in diffuse peritoneal carcinomatosis might completely be missed by examinations focused on SE, as the spatial resolution of SE is low and prone to artifacts especially at interfaces of high and low tissue stiffness [2,3] (see Fig. 1c and d).
Furthermore several entities may not follow the rule that greater stiffness is linked to malignancy, such as fibrotic tuberculous peritonitis, malignant pseudomyxoma peritonei or cystic metastases of e.g. adenocarcinomas.
This study illustrates the innate dangers of methods like SE in regard to observer bias. As conventional ultrasound has to be done before (strain) elastography, the two examinations are not independent and the observer may be tempted to fit SE findings to prior features observed in B-mode or Doppler-sonography. This is further complicated by the need to select the ‘best’ still image within a clip of compression-release cycles. Unfortunately no studies have looked into this so far to our knowledge.
If you would like to comment on this letter, please send your comments together with full contact data to firstname.lastname@example.org. Replies and comments will be reviewed by the Editorial Office and published on this website.