3.1 Clinicopathologic Features of Patients
As shown in table 1, a total of 316 patients were enrolled, among which 56 (17.7%) were in the AIS group, 98 (31.0%) in the MIA and 162 (51.3%) in the IAC group. Females were dominant but no statistical difference was found in the sex composition (p=0.331). Patients in IAC group were significantly elder than those in AIS or MIA groups (p<0.001) while the AIS and MIA groups had a similar age range, indicating the positive correlation between age and invasiveness of lung adenocarcinoma. The majority (83.3%) of the IAC patients received the lobar resection, while the overwhelming percentage (87.5%)of AIS patients underwent wedge resection. However, the resection type in MIA patients showed no obvious predilection. Noteworthily, 29% of the MIA group had a smoke history, compared with the 10% in the IAC group and 7% in the AIS group. This result suggested that smoke is positively correlated with the onset of lung adenocarcinoma but show no obvious effect on invasive evolution during tumor progression.
3.2 HRCT Morphological Features of Different Groups
As shown in table 2, the HRCT morphological features exhibited significant differences among the three groups. In the AIS group, the frequency of PGGN pattern was dominantly high (91.07%). However, the PGGN frequency dramatically dropped to 60.20% and 4.94% in MIA and IAC group respectively (p<0.001, with a highest χ2 value among all characteristics).
The AIS group had 87.50% round/oval nodules, which decreased to 85.7% and 61.7% in MIA and IAC group, suggesting that the invasive lung adenocarcinoma tend to be shaped irregularly (p<0.001). Besides, IAC lesions showed higher frequency of pleural invasion compared other two groups (p<0.001), while no statistical differences in the vascular invasion and lymphatic invasion and tumor necrosis status were found among three groups. The majority of patients (82.1%) in IAC group showed ≥ 50% solid proportion. In sharp contrast, only 9.2% in MIA group and none in AIS group exhibited ≥ 50% solid proportion. Further, MIA group showed more cases with undefined border than AIS and IAC(p<0.05). We also observed that smooth margin took a dominant proportion In the AIS group while most cases in the IAC group had a lobulate margin (p<0.001). It is notable that the vascular convergence was rare in the AIS(26.8%) group, and it exhibited gradient increase from IAC (41.9%) to MIA(71.4%) group.
In addition, patients in MIA and IAC group showed higher level of air bronchograms and vacuole signs than AIS group (p<0.001, respectively). Finally, another important characteristic with a high χ2 was whether there exists the pleural indentation, which showed a high probability in IAC (77.8%) cases but hardly in AIS (1.8%) (p<0.001).
3.3 HRCT Quantitative Features of Different Groups
As shown in table 3, we found that the diameter and solid diameter of the lesions were notably larger in the IAC group compared to AIS and MIA (p<0.001), and no statistical difference between AIS and MIA. For MGGNs, the CT values of ground-glass opacity (GGO) and ground-glass opacity solid portion (GGO-solid) were both higher in the IAC group than AIS and MIA (p<0.001). In contrast, we observed no statistical different of CT radiodensity values of PGGNs among the three groups.
3.4 ROC curve analysis of IAC group
The diagnostic performances of above variables that exhibited unique features in IAC group was analyzed by receiver operating characteristic (ROC) analyses (Fig. 2). The CT value of MGGN (solid component) exhibited the largest area (0.917). Important cutoffs were as follow: CT value-MGGN (solid) = -107 HU, sensitivity = 77%, specificity = 93%; Diameter = 14.7 mm, sensitivity = 90%, specificity = 81% (table 4).
Further, we developeda linear regression model and acquired a novel factor (named Combined Marker, or C-Maker) with 13 variables: C-marker =0.068*shape - 0.063*pleural invasion + 0.326*solid proportion - 0.056*border - 0.214*margin + 0.077*vascular convergence +0.181*air bronchograms - 0.124*vacuole sign + 0.027*pleural indentation + 0.001 * CT value of MGGN (solid) + 4.352e-005* CT value of MGGN (GGO) + 0.01*diameter - 0.004 * solid diameter. Here, CT value of MGGN (GGO), CT value of MGGN (solid), diameter and solid diameter were numerical variables and other variables were categorical data (encoded by 0 or 1 and 1 to 3, as shown in table 2). Again, we draw a ROC curve using C-marker, which exhibited an AUS as high as 0.982 (Figure 2). The cutoff value for IAC was 0.287, with a sensitivity of 0.935 and specificity of 0.977, which could be strongly recommended in determination of lung carcinoma classification(table 4).
3.5 The recurrence rate of patients
We analyzed the recurrence rate after two years’ follow up and found significant difference among the three groups (χ2 = 6.805, p = 0.033). No recurrence happened in the AIS and MIA groups, while 7 cases (4.3%) suffered recurrence in the IAC group. This is consistent with previous knowledge that IAC cancer type had a higher malignancy and poorer prognosis.