Functional thoracic imaging
: diagnosis and treatment response in oncology

Student thesis: Doctoral ThesisDoctor of Medicine by Research

Abstract

The aim of the first part of the thesis is to assess if 18F Fluoro-L-Thymidine (FLT) PET-CT molecular imaging (as a marker of proliferation) can predict response to arginine deprivation treatment (ADIPEG20 combined with pemetrexed and cisplatin) earlier than anatomical imaging using CT (RECIST), in mesothelioma (MPM) and non-small cell lung cancer (NSCLC). FLT PET-CT imaging took place in a longitudinal study with scans at baseline (PET1), approximately 24 hours after the first dose ADIPEG20 on day 2 of cycle 1 (PET2); at the end of cycle 1 of ADIPEMCIS (PET3) on day 16; and at the end of treatment (PET4). The baseline and end of treatment scans coincided with CT imaging, however, the interim scans were at different time points in n=10 MPM and n=8 NSCLC. Using end of treatment CT as the gold standard, the response to treatment was greater on PET4 than CT (mean decrease of 36.5% in SUVmax compared to 21.9% decrease in RECIST length). Also, FLT SUVmax treatment response at PET2 predicted end of treatment response on CT results in nearly 2/3 cases, although on ANOVA analysis there is no statistically significant evidence that a decrease in proliferation (SUVmax) precedes a decrease is size (RECIST length).

The aim of the second part of the thesis is to assess tumour heterogeneity changes in arginine deprivation treatment response in MPM using 18F Fluoro-2-deoxy-D Glucose (FDG) PET data from the ADAM trial (scans at baseline and 4 weeks post treatment) to see if texture features of FDG PET predict response better than RECIST. First-order and high-order primary tumour texture features were measured in n=20 patients. PET parameters, overall survival (OS), progression free survival (PFS) and RECIST-based treatment response (CT at 2 months) were tested by Cox and logistic regression analyses. From baseline to 4 weeks post therapy, there was decrease in skewness (mean 0.15 units, p=0.002) and kurtosis (median 0.2 units, p=0.03). None of the parameters at baseline or post therapy were associated with progression on RECIST. In terms of PFS, increase in uniformity was associated with progression (hazard ratio (HR) 2.3, p=0.02); increase in standard deviation (SD) was associated with decreased risk of progression (HR 0.56, p=0.03); and increase in SUVpeak was associated with a decreased risk of progression (HR 0.51, p=0.03). Texture features become more homogeneous after therapy, but this does not translate to improved survival. TLG at baseline was independently prognostic for OS (p=0.006); with additional asoociations with baseline SUVmean (p=0.03), SUVpeak (p=0.04), metabolic tumour volume (MTV) (p=0.01). Texture features are good at predicting the nature of the tumour.
Date of Award1 Nov 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorGary Cook (Supervisor), Vicky Goh (Supervisor) & Peter Szlosarek (Supervisor)

Cite this

'