PT  - JOURNAL ARTICLE
AU  - Christine Cung
AU  - Wenchang Wong
AU  - Rowena Martin
AU  - Ivan Ho Shon
TI  - Radiomic analysis of pre-treatment FDG PET prognosticates survival in non-small cell lung cancer
DP  - 2020 May 01
TA  - Journal of Nuclear Medicine
PG  - 277--277
VI  - 61
IP  - supplement 1
4099  - http://jnm.snmjournals.org/content/61/supplement_1/277.short
4100  - http://jnm.snmjournals.org/content/61/supplement_1/277.full
SO  - J Nucl Med2020 May 01; 61
AB  - 277Objectives: Intratumour heterogeneity, quantified by radiomic analysis of 18F-fluorodeoxyglucose positron emission tomography (FDG PET), has shown associations with patient outcomes in non-small cell lung cancer (NSCLC). We aimed to determine the prognostic value of intratumour heterogeneity measured by radiomic analysis of pre-treatment FDG PET in NSCLC patients treated with curative intent. Methods: Two-hundred and forty-one NSCLC patients reviewed by the lung cancer multidisciplinary team at a single institution between 2012 and 2017 were considered. A final cohort of sixty-eight histologically-confirmed NSCLC patients (Stage I to III) treated with curative intent were retrospectively reviewed. All patients underwent pre-treatment FDG PET integrated with computed tomography (FDG PET/CT) within 12 weeks of starting treatment. For radiomic analysis, semi-automatic, three-dimensional segmentation of primary tumours was performed on FDG PET using a threshold of SUV 2.5. Fifty-seven radiomic features were extracted using 3D Slicer software with the Heterogeneity CAD extension, including first-, second- and higher-order textural features and shape features. Univariate and multivariate Cox regression models were used to assess the prognostic value of these features against overall survival (OS), disease-free survival (DFS), local recurrence-free survival (LRFS) and distant metastasis-free survival (DMFS). Multivariate analysis adjusted for clinical stage, performance status and age. For those features that were significant on multivariate Cox regression, threshold values were determined by receiver operating characteristic analysis. Kaplan-Meier survival curves were constructed to compare survivals above and below these thresholds. Results: Thirty-seven OS, thirty-one DFS, twenty LRFS and twenty-one DMFS events occurred within a median follow-up period of 25.9 months (range: 2.2 - 75.9 months). On multivariate Cox analysis, several radiomic features were significantly associated with outcome. Most notably, kurtosis, a first-order textural feature, was significantly prognostic of all endpoints. Higher kurtosis was associated with poorer prognosis for OS (p = 0.013; HR = 1.507; 95% CI = 1.092-2.079), DFS (p = 0.009; HR = 1.593; 95% CI = 1.126-2.253), LRFS (p = 0.033; HR = 1.506; 95% CI = 1.033-2.195) and DMFS (p = 0.016; HR = 1.611; 95% CI = 1.091-2.379). Using the Kaplan-Meier method, higher kurtosis was a statistically significant prognosticator of poorer median DFS (13.4 versus 23.7 months, p = 0.0017), LRFS (14.5 versus 30.1 months, p = 0.023) and DMFS (14.5 versus 30.1 months, p = 0.023). Conclusions: Our study demonstrated that intratumour heterogeneity, measured by radiomic analysis of pre-treatment FDG PET, can provide independent prognostic information for NSCLC patients treated with curative intent.