Virtual Library
Start Your Search
M. Hawkins
Author of
-
+
MINI 33 - Radiotherapy and Complications (ID 164)
- Event: WCLC 2015
- Type: Mini Oral
- Track: Treatment of Locoregional Disease – NSCLC
- Presentations: 1
-
+
MINI33.02 - Dose-Escalated Radiotherapy for NSCLC: Heart Doses versus Survival in IDEAL-CRT (ID 454)
18:40 - 18:45 | Author(s): M. Hawkins
- Abstract
Background:
Radiotherapy dose-escalation trials have achieved inconsistent levels of overall survival (OS) for non-small cell lung cancer (NSCLC). For stage III NSCLC, RTOG-0617 reported poorer OS for 74Gy than for 60Gy delivered in 2Gy fractions with concurrent chemotherapy (median OS 20·3 vs 28·7 months). The phase I/II IDEAL-CRT CRUK funded trial (Sponsor: University College London (C13530/A10424)) of isotoxic dose-escalation delivered mean and maximum doses of 67.5 and 73Gy in 30 fractions over 40 days, equivalent to 69 and 75.6Gy in 2Gy fractions (EQD2), and reported 1- and 2-year OS rates 87.8% (95% CI:80.7-94.9) and 67.1% (95% CI:56.3-78.0) respectively. Here we investigate associations between OS and dose-distributions delivered to heart, left ventricle (LV), lung minus gross tumour volume (lung-GTV) and oesophagus.
Methods:
For 80 of 82 IDEAL-CRT patients, heart, LV, lung-GTV and oesophagus differential dose volume histograms (DVHs) were extracted from planning data using CERR. Since prescribed doses-per-fraction varied, physical DVHs were converted to EQD2s using the linear-quadratic equation with an α/β of 3Gy for heart, LV and lung-GTV, and 1.7Gy for oesophagus. Patient-to-patient DVH variability was efficiently represented using a small number of Varimax-rotated principal components (PCs) which explained 95% of the total variance: four for oesophagus, five each for heart and LV, and fourteen for lung-GTV. OS was modelled from the start of treatment using Cox regression.
Results:
On univariate analysis there is evidence that larger planning target volumes (PTVs) and greater Heart-PC3 and LV-PC4 coefficients are associated with worse OS (see table). Heart-PC3 represented heart volumes receiving doses of 65-75Gy, and LV-PC4 represented LV volumes receiving 1-5Gy. No lung-GTV or oesophageal dosimetric parameters were significantly associated with OS (p-values>0.1). Although OS improved with increasing prescribed dose, this trend did not reach statistical significance. The estimates of these effects did not change markedly after adjusting for other covariates in the multivariate analysis, and the relationship between greater Heart-PC3 and worse OS remained highly significant.Results (n=80)
Mean values Parameters Mean Minimum Maximum EQD2 lung-GTV dose 12.1Gy 6.2Gy 18.5Gy PTV 448.6cm[3] 138.7cm[3] 1262.2cm[3] Heart volume receiving >5Gy (V5) 33.9% 0% 99.9% Heart volume receiving >30Gy (V30) 11.3% 0% 47% Univariate Cox model for OS Covariate Hazard Ratio 95% Confidence Interval p-value EQD2 prescribed dose 0.951 0.883-1.024 0.185 PTV 1.002 1.000–1.003 0.026 Heart-PC3 1.257 1.070-1.478 0.005 LV-PC4 1.257 0.998–1.584 0.052 Multivariate Cox model for OS Covariate Hazard Ratio 95% Confidence Interval p-value EQD2 prescribed dose 0.977 0.900-1.060 0.570 PTV 1.001 0.999-1.003 0.179 Heart-PC3 1.228 1.039-1.452 0.016 LV-PC4 1.203 0.943-1.536 0.137
Conclusion:
We found strong evidence of an association between lower OS in IDEAL-CRT patients and heart volumes receiving 65-75Gy. There was some evidence of a negative association between OS and LV volumes receiving 1-5Gy. RTOG0617 has reported a negative relationship between OS and heart volumes receiving greater than 5Gy. Thus OS gains potentially achievable through tumour dose-escalation may be offset by associated heart dose increases. Further studies are required to improve understanding of radiation effects on heart substructures, aiming to identify avoidance structures for possible cardiac-sparing dose-escalated NSCLC treatments.
