The value of multiple diffusion metrics based on whole-lesion histogram analysis in evaluating the subtypes and proliferation status of non-small cell lung cancer

Yao Chen,Hong Yang,Yuan Qin,Chuanjiang Guan,Wenbing Zeng,Yong Luo
DOI: https://doi.org/10.3389/fonc.2024.1434326
IF: 4.7
2024-10-30
Frontiers in Oncology
Abstract:Objective: Limited studies have explored the utility of whole-lesion histogram analysis in discerning the subtypes and proliferation status of non-small cell lung cancer (NSCLC), despite its potential to provide comprehensive tissue assessment through the computation of additional quantitative metrics. This study sought to assess the significance of intravoxel incoherent motion (IVIM) and diffusion kurtosis imaging (DKI) histogram parameters in discriminating between squamous cell carcinoma (SCC) and adenocarcinoma (AC), and to examine the correlation of each parameter with the proliferative marker Ki-67. Materials and methods: Patients with space-occupying lesions detected by chest CT examination and with further routine MRI, DKI and IVIM functional sequence scans were enrolled. Based on the pathological results, seventy patients with NSCLC were selected and divided into AC and SCC groups. Histogram parameters of IVIM (D, D*, f) and DKI (D app , K app ) were calculated, and the Mann–Whitney U test or independent samples t test was used to analyze the differences in each histogram parameter of the SCC and AC groups. Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic performance of the histogram parameters. The correlation coefficient between histogram parameters and Ki-67 was calculated using Spearman's or Pearson's methods. Results: The D 10th percentile , D 90th percentile , D mean , D median , D app 10th percentile , D app 90th percentile , D app mean , D app median , D app skewness , D app SD of the AC groups were significantly higher than those of the SCC groups, while the K app entropy and K app SD of the SCC groups were significantly higher than those of the AC groups. All the above differences were statistically significant (all P < 0.05). ROC curve analysis revealed that D app mean showed the best performance for differentiating AC from SCC lesions, with an area under the ROC curve of 0.832 (95% confidence interval [CI]: 0.707-0.919). But there was no statistically significant difference in diagnostic efficacy compared to other histogram parameters (all P>0.05). D app 90thpercentile , D app mean , K app skewnes showed a slight negative correlation with Ki-67 expression (r value -0.340, -0.287, -0.344, respectively; P 0.05). Conclusions: Our study demonstrates the utility of IVIM and DKI histogram analyses in differentiating NSCLC subtypes, particularly AC and SCC. Correlations with the Ki-67 index suggest that D app mean , D app 90th percentile , and K app skewness may serve as markers of tumor aggressiveness, supporting their use in NSCLC diagnosis and treatment planning.
oncology
What problem does this paper attempt to address?