Can perfusion heterogeneity in CT perfusion maps of NSCLC flaw clinical considerations based on global mean blood flow values?

Abstract: Purpose: Tumour heterogeneity is one of the key features regarding progression disease, diagnosis, and therapeutic response. CT perfusion (CTp) studies are usually based on mean perfusion values, computed on a single slice or the whole tumour. The aim of this work is to determine to what extent the heterogeneity pouring out from single slices can flaw studies using global mean values only. Methods and Materials: 10 patients with single NSCLC lesions (26 examinations) were enrolled and underwent CTp. ROIs were drawn on aorta and on five central slices of each lesion. Mean global BF values were calculated on each slice (mSBF) and for the whole tumour (mWBF). The entropy E (i.e., a direct measure of data information content) was computed on BF values to assess the hemodynamic heterogeneity. For each lesion, the equivalence between each mSBF and mWBF was checked to assess whether mWBF of a lesion could represent the perfusion characteristics of a slice. Results: Five slices only, all pertaining to different lesions, were represented by the mWBF, none of them bringing the highest information content in its tumour (i.e., the maximum E). 11 couples of lesions having the same mWBF were found, but with slices with very different perfusion patterns, mSBF and E values. Conclusion: While global perfusion values computed on whole tumours may improve reproducibility, they cannot allow for tumour haemodynamic heterogeneity. Tumours with same mWBF could have very different response to therapies. Accordingly, additional parameters like BF entropy are needed to support mWBF in clinical considerations.