Multidetector computed tomography assessment of venous invasion in hepatic alveolar echinococcosis

This study included 136 patients. The mean age was 38.08 ± 13.06 years, and the mean maximum dimension was 13.46 ± 4.27 cm. These results were different from Europe centers in patient age and maximum dimension of the lesion, Chinese patients were younger, and the lesion size were larger than European patients [36]. Possible reasons maybe the more frequent contact that children in China may have with infected dogs because of a pastoral lifestyle, higher environmental infection pressure. On the other hand, because of residence in remote rural areas, Chinese HAE patients commonly do not seek medical treatment until they are symptomatic, so HAE patients in China are diagnosed late and the lesion sizes are quite large. A total of 614 veins were estimated with a high invasion rate of 83.06%. In this study, we assessed veins within 1 cm of the HAE lesion edge, those veins more than 1 cm were not included in this study. Therefore, the invasion rate of this study was different from the study of vascular invasion based on MRI examination, in which the invasion rate of portal veins and hepatic veins were 51.88%, 43.28%, respectively [34].

Due to its high-density resolution and high scanning speed, MDCT has shown great advantages in the preoperative diagnosis of HAE, morphological classification, preoperative patient evaluation, PNM staging evaluation, postoperative follow-up, and determination of an appropriate treatment method. At present, there are several studies on CT vascular invasion, which mainly focus on pancreatic cancer and hilar cholangiocarcinoma [29,30,31, 37,38,39,40]. However, there has long been a lack of recognized CT criteria for the assessment of HAE vascular invasion. Although HAE and tumors have similar biological characteristics, the criteria for vascular invasion in pancreatic cancer or perihilar cholangiocarcinoma cannot be directly applied to HAE.

Using the ROC curve, we found that the cutoff value of the lesion-vessel contact angle for venous invasion was > 180° and the diagnostic efficiency was good, the sensitivity, specificity were 84.90%, 88.46%, respectively. Lu's criteria for diagnosing pancreatic cancer, with lesion-vessel contact angle > 180° as the threshold, the sensitivity was similar with ours, while the specificity was 98%, which was higher than ours, the possible reasons were different prevalence and characteristics of the two lesions [30]. The cutoff value was the same as Lu's criterion and NCCN's criterion for vascular invasion of pancreatic cancer. The NCCN guideline indicated that when the tumor was in contact with the SMV or PV of > 180°, the tumor was considered unresectable [41]. Although the PPV was as high as 97.30%, NPV was very low. The prevalence can affect both PPV and NPV. Usually, the higher the prevalence, the higher the PPV. PLR is not affected by the prevalence. In this study, the probability of the lesion-vessel contact angle > 180 in the invaded veins was 7.35 times higher than that in the non-invaded veins.

In single CT criterion and sign, the lesion-vessel contact angle with a cutoff value of > 180° performed better than other signs. The diagnostic performance of irregular wall and lumen stenosis were the lowest with low sensitivity and specificity. The possible reason is that the venous wall is thinner than the arterial wall, so lumen stenosis and irregular wall are easily observed when HAE lesion present. Even if lumen stenosis and irregular wall were assessed alone or combination with the lesion-vessel contact angle, the difference was not statistically significant compared with the latter. The possible reasons may be the lack of a systematic evaluation of these two grading standards, which were susceptible to the influence of observer experience. Interestingly, lumen occlusion had a good specificity of 100% but a very low sensitivity of 38.03%. In this study, there were 194 veins with lumen occlusion, which were confirmed by intraoperative exploration and postoperative pathology. The venous lumen is larger, and the blood stream is typically slow. HAE lesion can easily infiltrate into the lumen and form embolus, causing lumen occlusion gradually.

Although the diagnostic performance of the lesion-vessel contact angle > 180° was better than other CT signs, it was also associated with false-positive and false-negative findings. There were 89 veins that had been wrongly judged, so there was an urgent need to improve its diagnostic performance. When the lesion-vessel contact angle was > 180°, the diagnostic performance was significantly improved when accompanied by occlusion, and the difference was statistically significant. When the lesion-vessel contact angle was > 180° with irregular wall or lumen stenosis, the diagnostic efficiency can be significantly improved only when this metric was accompanied by occlusion. When combining these criteria and signs, the diagnostic efficiency was the highest.

In this study, we assessed the diagnostic performance of the CT criteria and signs for HAE venous invasion including lesion-vessel contact angle > 180°, irregular wall, lumen stenosis and occlusion. We found that the lesion-vessel contact angle > 180° was the most useful as a diagnostic criterion for HAE venous invasion. When combining these criteria and signs, the diagnostic performance reached the highest. These findings may aid decision-making for radiologists and clinicians regarding the assessment of venous invasion in HAE patients. Up to now, the only CT study on HAE vascular invasion in China had no clear criteria for vascular invasion, and vascular invasion was classified into three grades based on imaging findings [42]. In this study, the lesion-vessel contact angle was a semi-quantitative criterion, which can more accurately assess venous invasion status compared with other CT signs.

Vascular invasion criteria of HAE by MRI were as follows: incomplete low vascular wall signal in T1 weighted image (T1WI), vascular truncation, surrounding by lesion, lumen stenosis or occlusion. However, lumen stenosis but with a completely low vascular wall signal on T1WI was not considered vascular invasion [30]. Some of these criteria were similar to ours. However, CT examination could not accurately evaluate the low vascular wall signal, so we described the vascular wall contour and the presence or absence of an irregular wall. In this study, the diagnostic performance of lumen stenosis was poor with a specificity of 43.26%, and 59 vessels with lumen stenosis were incorrectly diagnosed as venous invasion. As we all know, in addition to the vein, HAE can also invade the artery. According to a study based on MRI, the incidence of arterial invasion by HAE lesion was 26.87% [34]. We focused on veins in this study, the MDCT criteria for evaluating venous invasion included lesion-vessel contact angle > 180°, irregular wall, lumen stenosis and occlusion. Whether these criteria can be applied directly to the artery need further confirmation. Several studies have shown that arterial invasion and venous invasion should be treated separately due to different imaging signs. Veins are more susceptible to invasion mainly because the venous wall is thinner and the venous lumen is larger than the arteries [43].

There were several limitations in this study. First, this was a single-center study, and potential selection bias may have existed. Second, each vein may have different anatomical trends or properties, and this study did not assess the portal vein, hepatic vein and inferior vena cava separately. Third, in this study, due to the characteristics of HAE invading veins, the hepatic arteries were not included in the study. However, in clinical work, some hepatic arteries will also be affected. Finally, the criteria and CT signs were subjective and varied from observer to observer, and some potential features of HAE lesions could not be recognized by the naked eye. Computer-aided diagnostic systems are expected to solve this disadvantage in the future. In subsequent studies, we will combine with other centers, include hepatic arteries and classify the veins to explore differences in imaging signs of arterial and venous invasion. In addition, artificial intelligence will be used to study vascular invasion in order to show its internal characteristics. Although there are many limitations or deficiencies in this study, the observed discrepancies between imaging and intraoperative findings or pathological reports have major impact on clinical decision making and selection of appropriate treatment strategies.

留言 (0)

沒有登入
gif