ESTRO 2024 - Abstract Book

S1681

Clinical - Lung

ESTRO 2024

Purpose/Objective:

Efforts to improve local control in locally advanced non-small cell lung cancer (LA-NSCLC) through radiotherapy (RT) dose escalation are frequently hampered by the risk of toxicity, including oesophageal toxicity (ET). Previously, we have reported high freedom from local failure rates (>90% at 1-year), but also high rates of severe ET in LA-NSCLC patients treated in the international, phase II ARTFORCE PET-Boost trial (NCT01024829) on personalised dose escalation (Cooke et al. Radiother Oncol. 2023). The purpose of this study is to assess clinical factors and dose metrics associated with ET in patients treated in PET-Boost.

Material/Methods:

Eligible patients with stage II-III NSCLC, a primary tumour >4 cm and good performance status, were randomised to receive a boosted isotoxic dose to either the primary tumour as a whole, or to an 18 F-FDG-PET defined subvolume within the primary tumour. Treatment consisted of 24 x 3.0-5.4 Gy to the boost target volume, and 24 x 2.75 Gy to involved (mediastinal) lymph nodes. Concurrent or sequential chemotherapy was given if indicated. Concurrent chemotherapy consisted of 3-weekly cycles of a platinum-based doublet regime (cCT-doublet) or daily low-dose cisplatin (cCT-daily), and was institutional choice. Initially, the dose constraint for oesophagus was V36 < 80%, and a max. dose of 115% of 66 Gy was accepted for mediastinal structures. However, this was subsequently lowered to 107% due to high toxicity rates after a planned interim analysis. Additionally, a dose constraint added for the oesophagus + 5mm margin required the maximum dose to be ≤ 66 Gy physical dose (70 Gy EQD2 α/β=10Gy). Acute (< 90 days from start RT) and late (>90 days) ET were graded according to CTCAE v3. Doses were converted to EQD2 (α/β=10Gy) before dose metrics were extracted. Univariate and multivariate analysis was performed using the logistic regression model to determine the odds ratio (OR) and 95% confidence interval (CI) of clinical variables and dose metrics associated with grade ≥3 (G≥3) ET. Tested variables included age, ECOG performance status, smoking status (current vs ex-smoker), chemotherapy regimen (cCT-daily, cCT-doublet, or sequential chemotherapy/RT only), oesophagus mean dose and near maximum dose (D2). In total, 107 patients were randomised: 54 to the whole tumour boost and 53 to the PET-subvolume boosted dose groups respectively. Acute ET grade 2 occurred in 37.4% (40/107), and grade 3 in 12.1% (13/107). Late ET grade 2 occurred in 16% (16/100), grade 3 in 9% (9/100), grade 4 in 2% (2/100) and grade 5 in 3% (3/100). Grade 5 events consisted of oesophageal fistula, occurring at 9.8, 13.4 and 69.9 months, respectively, after start of RT. Of all patients who suffered G≥3 ET at any time (24/107 = 22.4%), 19 were treated before protocol amendment (19/54 = 35.2%) versus five (5/53 = 9.4%) after (Chi2: p = 0.001). Grade 4 (n=2) and grade 5 (n=3) events occurred only prior to the protocol amendment. Of 107 patients, 96 had data available for dose volume histogram (DVH) analysis (table 1). The median follow- up time was 74.5 months. Incidence of G≥3 ET was similar in patients with available DVH data compared to all randomised patients: 12.5% (12/96) acute, and 14.4% (13/96) late. The univariable analysis showed that current smoking (OR 2.75, p = 0.03), treatment with cCT-doublet (compared to no/seq chemotherapy, OR 4.2, p = 0.037), oesophagus mean dose (OR 1.074, p = 0.006) and D2 (OR 1.067, p = 0.017) were statistically significantly associated with G≥3 ET occurring at any time. The multivariable analysis showed oesophagus D2 (OR 1.062; CI 1.02 – 1.13, p = 0.014) was significantly associated with G≥3 ET. When compared to no/seq chemotherapy, cCT -daily had an OR of 2.8 (CI 0.6 – 21.0, p = 0.24), while the use of cCT-doublet was associated with an OR of 6.7 (CI 1.5 – 48.2, p =0.026). No significant difference was found when cCT-daily was compared with cCT-doublet (OR 0.42, p = 0.15). Smoking (OR 2.0, CI 0.7-6.0, p = 0.21) was no longer significant on multivariable analysis. Results: