Differentiation of brain metastases originating from lung and breast cancers using apparent diffusion coefficient histogram analysis and the relation of histogram parameters with Ki-67

Abstract

Purpose: A fast, reliable and non-invasive method is required in differentiating brain metastases (BMs) originating from lung cancer (LC) and breast cancer (BC). The aims of this study were to assess the role of histogram analysis of apparent diffusion coefficient (ADC) maps in differentiating BMs originated from LC and BC, and then to investigate further the association of ADC histogram parameters with Ki-67 index in BMs.

Methods: A total of 55 patients (LC, N = 40; BC, N = 15) with BMs histopathologically confirmed were enrolled in the study. The LC group was divided into small-cell lung cancer (SCLC; N = 15) and non-small-cell lung cancer (NSCLC; N = 25) groups. ADC histogram parameters (ADC_max, ADC_mean, ADC_min, ADC_median, ADC₁₀, ADC₂₅, ADC₇₅ and ADC₉₀, skewness, kurtosis and entropy) were derived from ADC maps. Mann-Whitney U-test, independent samples t-test, receiver operating characteristic (ROC) analysis and Spearman correlation analysis were used for statistical assessment.

Results: ADC histogram parameters did not show significant differences between LC and BC groups (p > 0.05). Subgroup analysis showed that various ADC histogram parameters were found to be statistically lower in the SCLC group compared to the NSCLC and BC groups (p < 0.05). ROC analysis showed that ADC_mean and ADC₁₀ for differentiating SCLC BMs from NSCLC, and ADC₂₅ for differentiating SCLC BMs from BC achieved optimal diagnostic performances. Various histogram parameters were found to be significantly correlated with Ki-67 (p < 0.05).

Conclusion: Histogram analysis of ADC maps may reflect tumoural proliferation potential in BMs and can be useful in differentiating SCLC BMs from NSCLC and BC BMs.

Keywords: Brain metastases; apparent diffusion coefficient; breast cancer; histogram analysis; lung cancer.

Figure 1.

Representative case of brain metastasis (BM) from small-cell lung cancer (SCLC). Axial post-contrast T1-weighted (T1W) image demonstrates a BM comprising non-enhanced necrotic and peripheral mild enhanced solid areas located in the left frontal parasagittal region (a). Regions of interest (ROIs) were manually drawn, encompassing all lesion voxels across multiple slices on the apparent diffusion coefficient (ADC) map (b). Note that the necrotic area within the tumour with the lack of enhancement on post-contrast T1W image (a) and fast diffusion on the ADC map (b) is not included in the ROI. ADC histogram map shows low ADC value for the metastatic mass (c). The photomicrograph (immunohistochemistry (IHC), ×400) shows 95% immunohistochemical staining level for the Ki-67 proliferation index (d).

Figure 2.

Representative case of BM from breast cancer (BC). Axial post-contrast T1W image demonstrates a heterogeneously enhanced BM located in the right anterior region of the cerebellum (a). ROIs were manually drawn, encompassing all lesion voxels across multiple slices on the ADC map (b). ADC histogram map shows high ADC values for the metastatic mass (c). The photomicrograph (IHC, ×400) shows 30% immunohistochemical staining level for the Ki-67 proliferation index (d).

Figure 3.

Receiver operating characteristic curves of significant ADC histogram parameters for differentiating brain metastasis of SCLC from non-small-cell lung cancer (a) and SCLC from BC (b).