Management of medically inoperable and tyrosine kinase inhibitor-naïve early-stage lung adenocarcinoma with epidermal growth factor receptor mutations: a retrospective multi-institutional analysis

After approval by the Sichuan Academy of Medical Sciences and Sichuan Provincial People’s Hospital Investigation Committee, patient information was gathered from five academic centers. Patients who had medically inoperable stage I to III EGFR mutant lung adenocarcinoma between January 1, 2010, and December 31, 2011, were identified. Diagnosis and staging of primary tumors were performed according to AJCC version 8. The inclusion criteria were TKI-naïve patients with newly diagnosed stage I to III disease who refused either surgery or chemotherapy for clinical node-positive disease, or patients who could not tolerate surgery but had resectable N disease. Patients who were treated with radiation followed by EGFR-TKI or with EGFR-TKI followed by radiation at primary site progression (named R + TKI) and patients who received only EGFR-TKI therapy (named as TKI alone) were included. The exclusion criteria were as follows: patients who had prior EGFR-TKI use patient who had EGFR-TKI resistance mutations patients for whom EGFR-TKI treatment was not performed after radiotherapy patients who received chemotherapy or immunotherapy, or received third-generation TKIs such as osimertinib for T790M mutation during TKI treatment patients with brain, visceral or bony metastases or patients who were missing covariable data or had an insufficient follow up time. To lessen a potentially confounding variable, patients who received surgical resection or neoadjuvant chemo- or immunetherapy at the time of initial treatment were also excluded. Radiation included stereotactic body radiation therapy (SBRT) or conventional external beam radiation therapy (EBRT). The SBRT dose ranged from 10 to 18 Gy in 3 to 5 fractions, while conventional EBRT ranged from 50 to 74 Gy in 25 to 35 fractions. The site of radiotherapy was the primary lesion. Tumor response was assessed using RECIST1.1, an evaluation criterion for the efficacy of solid tumors. Follow-up after treatment ocuurred once every 4 months in the first year, once every 6 months in the second and third years, and once every year in the fourth and fifth years.

The following variables were collected for subsequent analysis: age, gender, clinical stage, smoking history, EGFR mutation, clinical stages, type of RT delivered, name of the EGFR-TKI, and type of systemic therapy after progressing on EGFR-TKI treatment. Systemic disease status was assessed by the presence or absence of brain, or visceral or bone metastases at the time of initial treatment. The site of first progression after primary site radiation (SBRT or conventional) was identified. The date of initial cancer diagnosis; clinical stage; RT treatments; systemic therapy treatments; distant metastases including intracranial; visceral or bony disease; most recent follow-up; and death were recorded.

Positron emission tomography-computed tomography (CT) and CT scans of the chest, abdomen, pelvis, and bone scan were reviewed to ascertain the clinical stage, and any uncertain lesions required biopsies to rule out metastases. Pulmonary function tests (PFT) were performed before and after chest radiation to monitor the changes in lung function for all SBRT patients. Mediastinal node disease was evaluated by combinating PET and contract CT, and suspicious nodes were biopsied by endobronchial ultrasound (EBUS). EGFR mutations were evaluated by polymerase chain reaction amplification through next-generation sequencing (NGS) techniques. Exons 18 to 21 were analyzed for the following mutation; a deletion on exon 19 (E746-A750), or a point mutation on exon 21 (L858R). The stduy excluded ALK rearrangements, Rose1 mutations and rare mutations. Tumor response was assessed using RECIST1.1, an evaluation criterion for the efficacy of solid tumors. Follow-up after treatment was once every 4 months in the first year, once every 6 months in the second and third years, and once every year in the fourth and fifth years.

Statistical analyses were conducted using SPSS 25.0 software (SPSS, Chicago, IL, USA) and GraphPad Prism 7.04 (GraphPad Inc., La Jolla, CA). Characteristics of patients (categorical variables) in the two groups were analyzed by the χ² test with two-sided Fisher’s exact test. Kaplan-Meier OS curves and PFS curves were generated. Log-rank testing was used to assess the differences between the curves. OS was defined from the date of initial diagnosis until the date of death. PFS was defined from the date of initial diagnosis until the date of recurrence of a prior irradiated primary site(s) or the development of a new lesion. Using Cox proportional hazards analysis, univariable and multivariable variables were examined for the factors associated with OS and PFS. A value of p < 0.05 was considered statistically significant.

For the R + TKI group, the median OS after primary therapy was 42.6 months, while that of the TKI alone group was 29.4 months (log-rank p < 0.001; Fig. 1a). The median PFS in these two treatment groups was 24 months and 14.7 months respectively (log-rank p < 0.001; Fig. 1b). In the univariate analysis, advanced stages, EBRT and TKI alone were associated with significantly shorter OS. Following the multivariate analysis R + TKI was independently associated with improved OS relative to TKI alone (adjusted HR 0.420; 95% CI 0.287 to 0.614; p < 0.001; Table 2), after controlling for other significant covariables. In addition, multivariate analysis also showed that R + TKI was independently associated with improved PFS, compared to TKI alone (adjusted HR 0.420; 95% CI 0.291 to 0.605; p < 0.001; Table 3), after controlling for the significant covariables. Controlled covariables included age gender, nodal status, stages, RT strategy.

To explore the potential variations of the survival benefits in patients with different clinicopathological parameters, we subdivided patients according to their pathological stages, T stages and nodal status. Regardless of the therapeutic strategy, patients with higher pathological stages had a significantly shorter OS and PFS (log-rank p < 0.001; Fig. 2a-b). In comparison, nodal positive cases had inferior OS at the margin level of significance (log-rank p = 0.064, Fig. 2c) and significantly shorter PFS (log-rank p = 0.006, Fig. 2d).

Those stage I patients who had the best survival outcomes did not have improved OS or PFS when they received combined radiation and TKI therapy (log-rank p = 0.38 and 0.50 respectively, Fig. 3a and c). In stage II patients, although R + TKI did not improve OS (log-rank p = 0.14, Fig. 3b), it substantially prolonged PFS (log-rank p = 0.022, Fig. 3e). In stage III patients who had the worst prognosis, R + TKI significantly improved both OS and PFS, compared to TKI alone (log-rank p < 0.001, Fig. 3c and f).

The median OS of the stage III R + TKI group was 30 months, which was similar to that of stage I and II patients who received TKI alone (30.5 months and 30.1 months respectively). In contrast, the median OS of stage III TKI alone was 27.8 months. The median PFS of the stage III R + TKI group was 21.5 months, which was longer than that of stage I and II patients who had TKI alone (14.85 months and 15.35 months respectively). In comparison, the median PFS of stage III TKI alone was only 14 months.

When dividing the patients according to their T stages, R + TKI significantly improved both OS and RFS in the early T stage (T1/T2) (log-rank p = 0.017 and 0.004 respectively, Fig. 4a-b) and the late T stage (T3/T4) (log-rank p < 0.001, Fig. 4c-d) cases. In the subgroups divided by nodal status, R + TKI also significantly improved OS and PFS in nodal negative cases (log-rank p = 0.007 and 0.017 respectively, Fig. 5a-b) and nodal positive cases (log-rank p = 0.007 and < 0.001 respectively, Fig. 5c-d) cases.