The predictive value of energy spectral CT parameters for assessing Ki-67 expression of lung cancer

To investigate the predictive value of energy spectral CT parameters for Ki-67 expression in lung cancer.

In this retrospective analysis, 27 primary lung cancer patients confirmed by pathological examination were enrolled between December 2018 and February 2019. All patients underwent baseline arterial phase (AP) and venous phase (VP) energy spectral CT scanning followed by surgery in our institution. The iodine concentration (IC), normalized iodine concentration (NIC) and the slope of 40-80 keV energy spectrum curve (λ_HU) were derived from dual-energy virtual imaging on a Siemens postprocessed workstation. Immunohistochemical examination was performed to analyze Ki-67 expression. The ROC curves were used for predicting the performance of energy spectral parameters for Ki-67 expression.

The tumors appeared larger in Ki-67 high expression group than the low expression group (P=0.046). The energy spectral parameters were higher in venous phase when compared to arterial phase, but only the venous phase NIC (vpNIC) was significantly different from that of the arterial phase NIC (apNIC) (P<0.01). There are significant differences in high and low Ki-67 expression groups for vpNIC and venous λ_HU (vpλ_HU), (P=0.033 and 0.037 for vpNIC and vpλ_HU, respectively). vpNIC ROC analysis showed borderline P value (P=0.056) with an AUC, sensitivity (SE), specificity (SP) and cut-off value (0.717, 92.86, 61.54 and ≤0.347), respectively. The AUC, SE, SP and cut-off value of vpλ_HU were 0.698, 92.86, 53.85 and ≤2.407, respectively.

The energy spectral parameters (NIC and λ_HU) of venous phase might be used for predicting Ki-67 stratification. The venous phase energy spectral parameters were higher than the arterial phase. Furthermore, low expression Ki-67 group showed association with higher IC, NIC and λ_HU than high expression group.