Subpleural fibrotic interstitial lung abnormalities are implicated in non-small cell lung cancer radiotherapy outcomes

We retrospectively reviewed the medical records of patients with NSCLC treated with radiotherapy between January 2010 and November 2021. We included patients treated with radical-intent radiotherapy for locally advanced NSCLC or with salvage radiotherapy for locoregional recurrence postoperatively. Of the 188 consecutive patients who met these criteria, 9 who received stereotactic body radiation therapy for salvage and 4 who had a short follow-up duration (< 6 months) were excluded. We collected information from the remaining 175 patients’ medical records, including age, sex, Eastern Cooperative Oncology Group performance status, smoking history, blood sampling, pulmonary function tests, pathology, and imaging data. Patients treated with radical-intent radiotherapy were staged according to the 8^th edition of the Union for International Cancer Control tumour–node–metastasis classification. This study was approved by the Ethics Committee of Aichi Medical University (approval no. 2021-545) with an opt-out approach regarding the analysis before this study. Al l procedures involved in the study adhered to the principles of the Declaration of Helsinki. The re quirement for the acquisition of informed consent from patients was waived owing to the retrospective nature of this study.

Patients were immobilised in the supine position using an external vacuum-type body mould and/or thermoplastic body mask, and a CT scan with a 2-mm slice thickness was conducted for treatment planning. Respiratory motion was confirmed by obtaining CT images in both the expiratory and inspiratory phases or during shallow breathing. Both the target volume and normal organ structures were contoured using treatment planning systems (CMS XiO, Elekta, St Louis, MO, USA; Eclipse, Varian Medical Systems, Palo Alto, CA, USA; or Maestro, MIM Software, Inc., Cleveland, OH, USA). The clinical target volume (CTV) was defined as a 0–5-mm expansion of the primary tumour and metastatic lymph nodes. To account for respiratory migration, a 5–15-mm margin was added to the CTV to define the planning target volume (PTV). Radiotherapy was delivered by 6- or 10-MV x-rays from linear accelerators (Clinac iX or TrueBeam STx, Varian Medical Systems, Palo Alto, CA, USA). Overall, 60 Gy in 30 fractions was delivered to the PTV, and 119 (68%) patients received prophylactic elective nodal irradiation of 40 Gy in 20 fractions to the lymph node area. In exceptional cases, 11 (6%) patients received dose escalation up to 64–70 Gy.

Additionally, six (3%) patients underwent hypofractionated radiotherapy of up to 2.7–3 Gy per fraction to shorten the duration of treatment. The irradiation methods used were three-dimensional conformal or intensity-modulated radiotherapy (IMRT). Two-step IMRT was used for the entire period or only as boost irradiation of 20 Gy in 10 fractions.

The dose constraints were as follows: the goals for the global and spinal maximum doses were < 107% (< 125% allowed) and < 50 Gy (< 52 Gy allowed), respectively, whereas the goals for the lung volumes were 5 Gy (V5) < 60%, V20 < 25%, and mean lung dose < 12 Gy. Vx refers to the percentage of the lung volume receiving > x Gy. The goal for the mean heart dose was < 20 Gy. Allowance values for lung and heart doses were not set; however, every effort was made to reduce them as much as possible.

Patient age, general condition, and organ function determined the appropriateness of concurrent chemotherapy and its regimen. The most commonly used regimen was weekly carboplatin/paclitaxel, and cisplatin/docetaxel or cisplatin/vinorelbine was also administered. Durvalumab has been available to patients at our institution since October 2018. We performed CT after concurrent chemoradiation therapy was completed to ensure that there was no disease progression or grade ≥ 2 RP before administration. Durvalumab was intravenously administered at a dose of 10 mg/kg every 2 weeks for 1 year. If grade 2 RP appeared, the administration was suspended until the patient recovered to grade 1. We stopped administration in cases of serious adverse events, such as grade ≥ 3 RP or confirmed disease progression.

ILAs were defined according to the Fleischner Society classification.¹² The grouping was primarily performed by two physicians: a chest physician engaged in ILDs and a diagnostic radiologist engaged in chest radiology. They reviewed medical records including past history, family history, and reasons for imaging studies before reviewing CT images. Next, they used diagnostic chest CT images taken before treatment for grouping. All patients were classified into the following three groups: normal (no abnormalities), ILA, and ILD. The ILA group was further subdivided into NS, SNF, and SF groups. This process was performed independently by a chest physician and a diagnostic radiologist without discussion and was later collated. Grouping of any discrepant cases was finalized by consultation among the two, with another radiologist acting as an intermediary. Details of the CT protocols used for grouping are as follows. SOMATOM Definition AS/AS+/Flash (Siemens Healthcare, München, Germany) and Light Speed VCT VISION (General Electric Healthcare, Milwaukee, Wisconsin, USA) CT scanners were used. Slice thickness was 2–2.5-mm. Tube voltage was 120 kVp. Tube current was auto exposure control. Scan mode was helical acquisition. Pitch factor was 0.8–1.5. Rotation time was 0.3–0.5.

We measured the time to the event from the date of radiotherapy commencement. Toxicity was graded according to the CTCAE version 5.0. Grade 2 RP was considered an event with steroid administration for chest symptoms; administration of general cough suppressants was not included. Survival analysis was performed for patients under radical treatment.

All statistical analyses were performed using EZR version 1.55 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) based on R and R commanders.¹³ Patient characteristics were compared using Fisher's exact test for categorical variables and Student's t-test or the Mann–Whitney U test for continuous variables. Additionally, the Kaplan–Meier method was employed to estimate the cumulative incidence of RP and survival rates, and comparisons were performed using Gray's or log-rank test with post-hoc Bonferroni analyses for multiple comparisons. Finally, a Cox proportional hazards model was used for univariate and multivariate (stepwise elimination) analyses to determine factors contributing to RP and survival rates. Statistical significance was set at p < 0.05. Factors demonstrating p < 0.2 in the univariate analysis were included in the multivariate analysis. Receiver operating characteristic (ROC) curves were used to evaluate the relationship between RP and lung V20. The cut-off point was determined based on the Youden index, and well-balanced sensitivity and specificity values were obtained.¹⁴

The final analysis included 175 patients, with the normal, ILA, and ILD groups comprising 105, 64, and 6 patients, respectively. Table 1 summarises patient characteristics.

The p-values represent comparisons between the normal and ILA groups. The normal group was younger and had more patients with adenocarcinomas than the ILA group. The median follow-up period of the 70 surviving patients was 2.5 (range, 0.6–12.1) years. The 64 patients with ILA were subcategorised into NS, SNF, and SF subtypes comprising 5, 28, and 31 patients, respectively. Examples of the ILA subcategories are shown in Figure 1. The two specialists (a chest physician and a diagnostic radiologist) matched on 125 of the 175 patient classifications (72%), and the remaining cases were grouped by discussion among the three physicians.

FIGURE 1.

Grade ≥ 2 RP was identified in 71 (41%) patients. Of the 105 patients in the normal group, 21 (20%), 8 (8%), and 1 (1%) exhibited grades 2, 3, and 4, respectively. Of the 64 patients in the ILA group, grades 2, 3, 4, and 5 were observed in 19 (30%), 8 (13%), 2 (3%), and 8 (13%) individuals, respectively. Of the six patients in the ILD group, 2 (33%) and 3 (50%) showed grades 2 and 3, respectively. Seven of the eight patients with grade 5 RP were in the ILA-SF group, and one was in the ILA-SNF group (details shown in Table 2).

Figure 2 shows the cumulative incidence of grade ≥ 2 RP by group or subgroup. The incidence of RP was significantly higher in the ILA group than in the normal group (p < 0.001), particularly between the SNF (p = 0.014) and SF groups (p < 0.001) and the normal group. Table 3 presents the results of the univariate and multivariate analyses.

FIGURE 2.

ILAs (hazard ratio [HR]: 2.33, p = 0.01), IMRT (HR: 0.38, p = 0.03), and lung V20 (HR: 54.8, p = 0.03) were significantly associated with the cumulative incidence of grade ≥ 2 RP in the multivariate analyses.

ROC analysis using V20 showed intermediate grade ≥ 2 RP predictive accuracy in the normal group; the area under the curve (AUC) was 0.72 (95% confidence interval [CI]: 0.60–0.83). The sensitivity and specificity for a cut-off value of 21% for V20 were 76% and 67%, respectively. Conversely, the predictive accuracy in the ILA group was low, with an AUC of 0.59 (95% CI: 0.45–0.75); furthermore, the sensitivity and specificity for a cut-off value of 19% were 76% and 56%, respectively. During follow-up after radiotherapy, five patients received molecular-targeted therapies, and two had grade ≥ 2 RP. However, it was not until 2 years and 10 years after radiotherapy that the two patients, respectively, began receiving molecular-targeted therapies, which had already cured their RP.

The 2-year OS and PFS rates for the 134 radically treated patients were 47.8% (95% CI: 38.5–56.4%) and 21.2% (95% CI: 14.4–28.9%), respectively. Patients in the ILA group had worse 2-year OS than those in the normal group (35.3% vs. 54.6%, p = 0.005); however, the multivariate analysis showed no significant difference. No significant difference was found in the 2-year PFS between the ILA and normal groups (17.3% vs 24.1%, p = 0.26).

The characteristics of the patients treated with radical-intent radiotherapy are summarised in Table 4. The parameters to be entered into the multivariate analysis are listed separately for the normal and ILA-SF groups.

No significant bias was found between the two groups; however, only one patient in the ILA-SF group received durvalumab. Patients in the ILA-SF group had significantly worse 2-year OS (29.2% vs 54.6%, p < 0.001) and PFS (9.7% vs 24.1%, p = 0.009) rates than those in the normal group. Table 5 shows the results of the univariate and multivariate analyses for the OS and PFS rates.

The multivariate analysis revealed that ILA-SF was the only independent adverse factor for OS (HR: 3.07, 95% CI: 1.17–8.10, p = 0.02) and that it also tended to influence PFS (HR: 1.95, 95% CI: 0.91–4.14, p = 0.08). In contrast, younger age, adenocarcinoma, concurrent chemotherapy, and durvalumab treatment contributed to the prolongation of OS and/or PFS.