With OCT as the gold standard, the per-patient specificity and PPV of CCTA + plain scan for diagnosis of ruptured plaques were 89% and 90%, and the sensitivity and NPV reached 71% and 69%. The agreement between CCTA + plain scan and OCT for the diagnosis of ruptured plaques reached a Kappa value of 0.70, and cavity depth and calcification at the rupture site might be the major factors influencing the diagnostic performance of CCTA+ plain for ruptured plaque.
3.1 Diagnostic Value of CCTA + Plain Scan for Ruptured Plaques and Advantages of Our Study
Our study results indicated that CCTA + plain scan has high specificity and PPV for the diagnosis of ruptured plaques, as well as medium sensitivity and NPV. Our findings improve upon the existing data, firstly by using OCT optical biopsy as the gold standard. With a resolution of up to 10 to 15 μm (nearly 10× that of IVUS, OCT allows clear in vivo assessment of the fibrous cap thickness of plaques, slight ruptured plaque, and arterial dissection [6]. Previous studies have evaluated CCTA with ICA or IVUS as the standard, with a limited resolution that often caused misdiagnosis of ruptured plaques and thus wrongly estimated the true diagnostic efficacy of CCTA for these plaques. Recently, Obaid et al [13] found by comparing CCTA with IVUS in diagnosing ruptured plaques, CCTA had specificity of up to 91% but sensitivity of only 33%. We observed a similar specificity but a much greater sensitivity (71%; per patient). Although IVUS has been widely used in the morphological evaluation including plaque load and histological characteristics, it is still very challenging to assess minute changes in the vessel wall and detect ruptured plaques [16]. In a recent study with OCT as the gold standard, Ohashi et al [17] found that the specificity of the latest-generation high-resolution IVUS for ruptured plaques was only 57.1%. With ICA as the gold standard, Madder et al [11] and Bilolikar et al [12] found that CCTA had a low PPV for ruptured plaque (57.8% and 58.8%, respectively) but a high NPV (91.9% and 95%, respectively); this excellent negative predive capability might be because ICA missed some small ruptured plaques. When Obaid et al [13] used IVUS as the gold standard, the sensitivity of ICA for ruptured plaques was only 49%. Secondly, we compared enhanced and plain scan images to differentiate calcifications from signs of rupture, which was helpful to decrease the number of false positive ruptured plaques. Among the 27 OCT confirmed non-rupture patients in this study, 4 (14.8%) had false positive signs of rupture on CCTA alone that were confirmed as calcifications on plain scan. Overall, the specificity and PPV of CCTA + plain scan were improved by more than 10% compared with CCTA alone, and there was good agreement between CCTA + plain scan and the gold-standard OCT. Finally, compared with the 64-row spiral CT and traditional filtering reconstruction used in the previous studies, the third-generation dual-source CT used in this study has a higher temporal resolution (66 ms), and in combination with a photon detector and advanced modeled iterative reconstruction, it can display the signs of ruptured plaque more clearly, thus further improving the diagnostic efficacy of CCTA for ruptured plaques. Besides, ruptured plaques had more high-risk signs than non-ruptured plaques in our study. That this was only significant for spotty calcification may be associated with a small sample size.
3.2 Factors Affecting Diagnostic Performance of CCTA + Plain Scan
In this study, 40 ruptured plaques were diagnosed by OCT, of which 13 (32.5%) were misdiagnosed as non-ruptured plaques (false negative) by CCTA + plain scan. Thus, the sensitivity and NPV of CT for ruptured plaques were unsatisfactory. The following reasons are proposed based on comparison among three image sequences (plain scan, CCTA, and OCT). Firstly, it is difficult to detect slight rupture (i.e., ruptured plaques with a small cavity) on CT with limited spatial resolution. In our study, 11 CCTA false negative ruptured plaques had cavity depths of only 0.46 mm on average at OCT. The cavity depth was evidently greater (0.98 mm) in true positive ruptured plaques, while there was no significant difference in the cavity width. This may indicate that small longitudinal rupture (cavity depth <0.46 mm) is an important factor influencing the accurate diagnosis of ruptured plaque by CCTA, and the diagnostic efficacy of CCTA might be further improved by adding radiomics and deeply mining the internal plaque features in the further study. Secondly, calcifications can block intra-plaque dye penetration in some areas, which is easily mistaken for the partial volume effect of calcifications and thus leads to false-negative diagnosis of ruptured plaque. In our study, the cavities of 2/10 false negative ruptured plaques were located near calcifications. Finally, previous studies have shown that small calcifications could be easily misperceived as intra-plaque dye penetration, thus contributing to the number of false positive results [11-13]. In the study of Obaid et al [13], all 3 false positive results occurred in plaques with small calcifications, and in our study, the comparison between plain and enhanced scans resulted in correction of 4/28 false positive ruptured plaques to true-negative non-ruptured plaques, which mitigated the CCTA alone false-positive rate for ruptured plaques to some degree. Undeniably, our plain scan, which was performed with the routine calcification scoring sequence, other than the newly added sequence, had a bigger slice thickness (3 mm), though the radiation dose was not increased; thus, small calcifications could have been missed, which may be an explanation for the false positive results in this study.