The potential of ADC values derived from DWI in predicting pathological subtypes and stages (TNM and Mosaoka) of TETs has been previously reported [6, 10, 13,14,15]. Considering the possible role of ADC values in differential diagnosis and predicting pathological subtypes and stages of TETs [10, 13,14,15], the purpose of this retrospective study was to evaluate the potential of ADC for predicting pathological T stages or pathological Masaoka stage in patients with TETs. The present study demonstrated that pathological T stages were negatively correlated with mean ADC values in TETs. Based on the ADC cut-off values derived from ROC analyses, ADC values seemed to more accurately distinguish pathological T stages of TETs compared with Masaoka stages. Moreover, two distinct sets of optimal cut-off values of mean ADC for differentiation between pathological T stages of TETs were identified.

Dynamic water diffusion within tissues is affected depending on membrane permeability, cellular volume fraction, and tissue microstructure [10, 19], and is quantitatively expressed as ADC at varied b values [20]. Since ADC changes reflect the pathophysiological deficit, ADC was used to discriminate malignant from benign mediastinal tumors [21, 22], and to monitor long-term therapeutic responses in mediastinal lymphadenopathy and breast cancer [23, 24]. Notably, the b value indicates the strength and timing of the gradients in DWI, which influences the sensitivity to diffusion-based contrast [25] and the subsequent calculation of ADC values [13, 25]. A range of b values (0 to 800 s/mm2) were commonly used in studies exploring the diagnostic ability of ADC in distinguishing subtypes and stages of TETs [6, 13, 18, 26]. Worth noting, no contrast agent (eg, gadolinium) was used for DWI in the current study, as contrast medium can have a significant impact on ADC value in a time and lesion-type dependent manner, as well as the quality of images [27, 28].

A study using a 1.5 T MR scanner found that mean ADC values were 1.43 ± 0.26 and 1.31 ± 0.31 × 10–3 mm2/sec for early and advanced Masaoka stages of thymomas, respectively [6]. The present study with a 3 T MR scanner observed that TETs with more advanced pathological Masaoka stages had lower mean ADC values [6]. Another study utilizing a 3 T MR scanner reported that the optimal cut-off value in discriminating between advanced and early Masaoka stages may be the 10th percentile of ADC values [14]. Thus, the variability between the current results with those of others may be in part due to distinct MR scanners, b values and/or ADC measurements used [6, 14].

Masaoka stages I and II of TETs are resectable, but Masaoka stage III overlaps with stages T2 to T4, including unresectable and resectable TETs [8]. T2 stage TETs with pericardium invasion may be managed by thoracoscopy. T3 stage TETs often require full sternotomy. T4 stage TETs with intrapericardial pulmonary artery invasion and aortic invasion require cardiopulmonary bypass during surgery and are considered inoperable. In contrast, the degree of invasion to different organs is classified into various T stages in the TNM staging system [8]. Masaoka stage IV reflects only M1 stage of the TNM staging system and cannot differentiate resectable TETs from unresectable TETs. For example, T3N0M1a is equivalent to Masaoka stage IV and it is recommended to actively remove all visible tumors according to the current guideline for TNM staging. TNM stages are significantly correlated with WHO histological classification in TETs, and its prognostic significance is suggested [11]. Thus, compared with pathological Masaoka stages, preoperatively predicting pathological T stage may provide more valuable information to facilitate the selection of optimal therapeutic strategies.

Clinical T staging is an important component of assessing risk and managing patients with TET. Inconsistencies in clinical T staging and pathological T staging can lead to unwanted variability in the selection of surgical methods thus impacting the extent of resection that may be required. Currently, clinical T staging largely depends on preoperative CT or conventional MRI; however, considerable inconsistencies still exist in pathological T staging. The addition of preoperative ADC values as part of clinical T staging may improve consistency, which warrants further investigation. Regarding ADC and WHO pathological staging, a study of thymic epithelial tumors has previously confirmed the correlation between ADC value and WHO pathological classification [13]. However, the WHO pathological classification, compared with the TNM staging system or the Masaoka staging system, is poorly correlated with prognosis and is of little clinical significance [29, 30].

To our knowledge, this study is the first to demonstrate a negative correlation between ADC values and the pathological T stages in patients with TETs. A previously published database study of 907 TET patients revealed that the T stage was a predictive factor for recurrence of TETs [31], so predicting the pathological T stage preoperatively is of clinical significance. In this study, only the pathological T stages were assessed because all four T stages were found in the study population. Given that the distribution of different WHO histological types in different T stages was uneven in the current study, pathological T1 and T3 were assessed to determine whether differences existed with WHO histological classification. A preliminary post-hoc analysis revealed no significant difference in ADC value of different pathological histological types under the same pathological T stage, although the ADC value of the rare pathological T1 stage thymic carcinoma was lower. TETs with the same pathological T stage, but different pathological WHO histological types may have different thresholds; however, in clinical practice, the pathological T1 or T2 thymic carcinoma and the pathological T3 or T4 type A thymoma are very rare. Larger scale multicenter studies are warranted to further address this question considering any limitations associated with output from a single institution study with a small number of patients.

Future experiments should investigate whether the addition of preoperative ADC values for clinical T staging can improve the consistency between clinical T staging and pathological T staging. In addition, optimal cut-off values of mean ADC were identified. The AUC values for discriminating pathological T stages were between 0.821 and 0.908 in this single institution retrospective study, suggesting preoperative ADC has promising diagnostic value in predicting pathological T stages in patients with TET.

The predictive accuracy of diagnostic imaging is critical for differential diagnosis of TETs, and has been improved via multiparametric chest MR, such as the combination of DWI and perfusion-weighted imaging [10], and the radiomics nomogram including conventional MR imaging parameters, ADC value, and radiomic signatures [32]. Moreover, ADC and texture parameters derived from DWI were used together to predict pathological WHO types and Masaoka stages of TETs [3].

Limitations in the current study need to be addressed. First, this research is a single institution study with a small number of patients, so larger-scale multicenter studies are warranted to confirm current findings. Second, T4 invasion usually is unresectable, and only one patient with clinically suspicious T4 TET underwent surgery; therefore, selection bias cannot be excluded. Third, additional imaging parameters will be included in the future to precisely discriminate pathological T3 from T4 TETs. Fourth, small tumors may have been overlooked because it is difficult to identify and measure tumors less than 2 cm due to the potential imaging artifacts. Finally, it should be recognized that the optimal ADC cut-off values may vary, depending on the MRI vendor and software.