The dataset consists of 203 patients characterized by various demographic and clinical attributes. After excluding implausible values, 189 patients remained for analysis, as detailed in Table 1. The gender distribution was 65% male (123 patients) and 35% female (66 patients). In the durvalumab subgroup (n = 86), the distribution was similar, with 62% male and 38% female. The median age at diagnosis was 65.5 years overall and 64 years for the durvalumab group, with an interquartile range (IQR) of 59–71 years for both.

Regarding tobacco use, 18% were non-smokers, 36% were smokers, and 32% were ex-smokers; 13% had unknown smoking status. The durvalumab group had similar smoking proportions.

Cancer staging according to the Union for International Cancer Control (UICC) showed 49% in stage IIIB, 37% in stage IIIA, and 11% in stage IIIC, with similar distributions in both groups. Squamous cell carcinoma (SCC) was the most common histological subtype, present in 55% of all patients and in 51% of those treated with durvalumab. Adenocarcinoma (AC) was diagnosed in 40% of patients. The histological subtype was unknown for 5% overall and for 6% in the durvalumab subgroup.

Regarding PD-L1 status, 20% had less than 1% expression, 54% had 1% or greater, and 25% were unknown. In the durvalumab group, 90% had PD-L1 expression of ≥ 1%. Four patients (5%) with PD-L1 levels below 1% were treated with durvalumab.

Overall, 86 patients (45%) were treated with durvalumab, while 79 patients (42%) did not receive durvalumab therapy. The majority of tumors were classified as grade G2 (34% or 64 patients) and grade G3 (33% or 62 patients), with similar proportions in the durvalumab-treated group. Grades G1 and G4 were less common, each accounting for 1% of the patients (2 patients each). Tumor grading was unknown for a substantial portion (59 patients; 31%).

Regarding T status, T4 stage tumors were most prevalent, found in 104 patients (55%) treated with or without durvalumab and in 44 patients (51%) receiving durvalumab. Other T stages ranged widely from T1 to T3. The most common N stage was N2 with 77 patients (41%), followed by N3 with 54 patients (29%). An N0 status was observed in 27 patients (14%) and N1 in 21 patients (11%).

Pulmonary comorbidities were present in 82 patients (43%) considering all included cases and in 49 patients (56%) in the durvalumab-treated group, with slightly fewer patients without pulmonary comorbidities in this group.

Radiation planning was FDG-PET/CT based in the majority of cases for both treatment groups (all patients 95%; durvalumab group 96%). The most frequently used radiation technique was VMAT (all patients: n = 120, 63%; durvalumab group: n = 51, 59%), followed by IMRT. Regarding radiation dose, the median dose was 66 Gy, with 25th and 75th percentiles of 60 Gy and 66 Gy considering all patients (durvalumab group: median dose 66 Gy; 25th and 75th percentiles 62.25 Gy and 66.60 Gy, respectively).

Chemotherapy regimens showed that cisplatin was the most commonly used chemotherapeutic drug in combination with vinorelbine (all patients: n = 111, 60%; durvalumab group: n = 55, 64%). Paclitaxel, etoposide, and carboplatin were also used less frequently.

The median OS for the entire cohort was 13.7 months (Table 2). At the time of analysis, 52 patients had died and 82 patients had experienced a progression.

The mean GTV1 before RT was 145.29 ml, with 25th, 50th, and 75th percentiles of 61.36 ml, 145.29 ml, and 204.93 ml, respectively. Before initiation of the radiation boost, the mean GTV2 was 99.58 ml, with the 25th, 50th, and 75th percentiles at 32.93 ml, 70.45 ml, and 126.85 ml, respectively.

The mean absolute difference between the two volumes was 45.70 ml, with the 25th, 50th, and 75th percentiles at 5.59 ml, 25.71 ml, and 66.34 ml, respectively, for all patients and 42.08 ml with an IQR of 4.00–59.8 ml for patients treated with durvalumab after CRT. Figure 2 shows a scatter plot of the relationship between the absolute difference in GTV1 and GTV2 and Martingale residuals. The Cox proportional hazards model generally fits the data well for most patients, as indicated by the concentration of residuals around 0. Figures S1 und S2 show the Kaplan–Maier curves of high, intermediate, and low absolute GTV1 and GTV2 volumes. Figure S3 illustrates the Kaplan–Meier plot of high, intermediate, and low absolute changes between the two volumes.

Martingale residual plot of the absolute differences between GTV1 and GTV2

Median OS was 19.77 months, 12.95 months, and 13.37 months for patients with a low, intermediate, and high baseline GTV1 before RT, respectively. Moreover, the median OS was 14.47 months, 13.42 months, and 14.37 months for patients with a high, intermediate, and low GTV2, respectively, before the initiation of boost RT. The respective HR values can be found in Tables 3 and 4. For patients treated with durvalumab, the pretreatment GTV had no significant association with OS (HR = 1.01, 95% CI 0.97–1.04; p = 0.46). In contrast, the absolute volume change between GTV1 and GTV2 was significantly correlated with OS, with an HR of 0.96 per millimeter (95% CI 0.91–1.04; p = 0.03) for patients treated with durvalumab. To adjust for a potential immortal time bias, we set 42 days as the landmark based on the latest administration of durvalumab in the PACIFIC trial [15]; the resulting HR was 0.96 (0.92–0.99; p = 0.031).

Figure 3 shows the overall survival of patients grouped into patients with a high, intermediate, and low absolute tumor volume reduction. The groups were formed according to the first and third quantile. Figures 4 and 5 display the Kaplan–Meier curves for high, intermediate, and low absolute volumes of GTV1 and GTV2 of the durvalumab subgroup. Figure 6 presents the Kaplan–Meier plot for the absolute changes between the two volumes, categorized as high, intermediate, and low for patients treated with durvalumab.

Kaplan–Meier plot of OS according to absolute GTV difference quantiles before radiotherapy of the durvalumab subgroup. High, intermediate, and low GTV differences referring to the 25% and 75% quantiles

Kaplan–Meier plot of OS according to absolute GTV1 before radiotherapy of the durvalumab subgroup. High, intermediate, and low GTV1 referring to the 25% and 75% quantiles

Kaplan–Meier plot of OS according to absolute GTV2 before radiotherapy boost of the durvalumab subgroup. High, intermediate, and low GTV2 referring to the 25% and 75% quantiles

Kaplan–Meier plot of OS according to relative GTV change during radiotherapy (from GTV1 to GTV 2) of the durvalumab subgroup. Low, intermediate, and high GTV decrease referring to the 25 and 75% quantiles

The HRs regarding administered radiation dose, histology, and grading can be found in Tables 3 and 4, respectively.

In short, UICC stage, age at onset, pulmonary comorbidities, and smoking status were not found to be prognostic factors in terms of survival. Only durvalumab treatment was significantly associated with improved OS, with an HR of 0.454 (95% CI 0.209–0.990; p = 0.047). All results of our Cox regression analysis can be found in Tables 3 and 4.

We performed an additional patterns-of-failure analysis to examine the distribution of recurrences in all patients and specifically in those treated with durvalumab. Among all patients, 25 experienced local recurrence, 3 had regional recurrence, and 29 developed distant metastases (Table 5). In the durvalumab-treated group, 9 local recurrences, 1 regional recurrence, and 17 distant metastases were observed (Fig. 6). To illustrate the relationship between intrathoracic progression and distant progression with OS, we computed Kaplan–Maier curves. These Kaplan–Maier curves can be found in the supplement section (figures S4 and S5).