75P - Predictive value of CT texture analysis in lung cancer patients undergoing nivolumab

Date 07 May 2017
Event ELCC 2017
Session Poster Display Session
Topics Immunotherapy
Thoracic malignancies
Therapy
Presenter Valerio Nardone
Citation Annals of Oncology (2017) 28 (suppl_2): ii24-ii27. 10.1093/annonc/mdx092
Authors V. Nardone1, P. Tini1, E. Aldi2, P. Pastina1, L. Sebaste1, G. Battaglia1, S.F. Carbone2, V. Ricci2, L. Pirtoli1, P. Correale1
  • 1Radiation Oncology Unit, Azienda Ospedaliera Universitaria Senese - Santa Maria delle Scotte, 53100 - Siena/IT
  • 2Diagnostic Radiology Unit, Azienda Ospedaliera Universitaria Senese - Santa Maria delle Scotte, 53100 - Siena/IT

Abstract

Background

Nivolumab is a human mAb to PD-1 with significant antitumor activity in a number of solid malignancies including NSCLC, kidney cancer and melanoma. At the present, no predictive factor has been identified for this drug, thus its administration is mostly empirical, at price of frequent adverse events and high costs. In this context, we evaluated whether baseline CT texture analysis (TA) could be used to identify advanced NSCLC patients who will benefit by Nivolumab treatment, by taking in consideration that the therapeutic effects PD-1 blockade depends by its ability of reactivating a pre-existing immuneresponse, whose activity is strictly related to the presence of necrosis, hypoxia, and inflammation in the tumor sites, which can be evaluated by imaging assessments.

Methods

A retrospective analysis was performed on a sample of seventeen NSCLC patients who received salvage therapy with Nivolumab 3 mg/kg every 15 days between October 2015 and January 2017 with a median follow up of twelve months. The gross primary tumor volume before treatment (baseline) was contoured on pre and post contrast CT sequences. TA parameters were extrapolated by using LifeX Software ©, tested for reliability and then correlated with patients’ outcome, in particular with early failure defined as a confirmed disease progression or death within 6 months (7 patients), by means of univariate and multivariate analysis.

Results

We found a significant correlation among TA parameters and patients’ outcome at univariate analysis. In fact, early vs longer responders showed differences in term of volume in voxel (p:0.049), entropy (p:0.046), compacity (p:0.033), GLCM-contrast (p: 0.018), GLCM-dissimilarity (p:0.017), LRHGE (p:0.019), coarseness (p:0.036), contrast (0.020), ZP (p:0.025), pre contrast: GLCM-contrast (p:0.031), GLCM-dissimilarity (p:0.035), contrast (p:0.038), SZE (p:0.029), ZP (p:0.040). Multivariate analysis (Logistic regression) confirmed a significant correlation between early failure and post contrast GLCM dissimilarity (p:0.011, OR: 3.30, AUC: 0.800; 95% CI 0.578-1.00).

Conclusions

Our results, if confirmed on a larger series, suggest that TA may predict early failure, and therefore may help the selection of patients who may benefit by Nivolumab treatment.

Clinical trial identification

Legal entity responsible for the study

Azienda Ospedaliera Universitaria Senese

Funding

Azienda Ospedaliera Universitaria Senese

Disclosure

All authors have declared no conflicts of interest.