Non-small cell lung cancer is a disease of the elderly. In this large matched-pair study, age was not associated with either postoperative survival or the occurrence of postoperative complications after surgery for resectable NSCLC. We demonstrate that the postoperative 30-day mortality was consistent at 1.8% across both age groups, whereas the 90-day mortality exhibited a non-significant difference of 2.2% in patients < 75 years compared to 4.0% in the older age group. Similar findings were reported in the studies by Rivera et al. and Park et al., who indicated an increase in 90-day mortality compared to 30-day mortality in older study populations, while the increase was relatively smaller in younger patients (9, 10). Several publications have reported varied 30-day mortality rates following lung resection in patients with NSCLC, ranging from 0–9% (11–19). The reported 90-day mortality rates ranged from 1.5–10% (5, 9, 13, 19–21). These variations in mortality data may be attributed to the heterogeneity of the study populations. Notably, some studies highlighted that older patients exhibited higher perioperative mortality compared to younger cohorts, emphasizing the impact of age on surgical outcomes (10, 18, 22). Rivera et al. demonstrated a significantly higher 30- and 90-day mortality in elderly patients in a matched study including 1969 patients above the age of 70 (10). However, only patients with NSCLC stage I and II were included in the analysis and tumor stage as a confounder for perioperative mortality and overall survival was not chosen as a matching variable.
In this study 42% of patients ≥ 75 years developed at least one postoperative complication, compared to 36% in the younger patients. Notably, the literature reports varied rates of postoperative complications after lung resection in NSCLC, ranging from 8.4–66.7% (11–19, 21, 23). The wide range of postoperative complication rates probably results from differences in the definition of complications, follow-up period and patient selection. Some publications indicated no significant difference in postoperative complication rates between younger and elderly patient populations, while others emphasized a substantial disparity in complication rates depending on patient age (5, 12, 13, 24, 25). Park et al. compared the postoperative complication rate of 285 patients ≥ 70 years versus 1055 patients < 70 years with stage I NSCLC and demonstrated a significantly higher rate in the elderly (9). However, differences in patient populations, particularly with respect to T category, sex and surgical approach, may have biased the conclusions. We demonstrate an association between male sex, loss of FEV1, extended surgical resections and the development of postoperative complications. Similarly, Saji et al. demonstrated a 2.7-fold increased risk of developing postoperative complications in men (26). In line with the works of Kutluk et al. and Sezen et al., arrhythmias were the leading postoperative complication, regardless of age (21, 27). Others such as Amer et al. described a significantly higher rate of postoperative arrhythmias and need for intensive care in patients over 80 years of age (12). In this work there was no association between the surgical approach and the occurrence of postoperative complications. However, only 9% of patients underwent video-assisted thoracoscopic surgery, and these had a lower tumor stage and were older which reflects our clinical practice and that of others (28, 29).
There was no difference in the length of hospital stay between the groups. In the literature reported hospital stay after lung resection surgery varies between 4 and 21 days (29). However, differences across healthcare systems, where treatments and follow-ups are shifted to outpatient and post-hospital settings make a comparison difficult.
Nearly twice as many patients in the younger cohort received adjuvant chemotherapy compared to the older cohort. Overall, the majority of patients aged ≥ 75 years did not receive guideline-recommended adjuvant therapy, a practice that has been described by others and is certainly due to the lack of data on adjuvant chemotherapy in patients over 75 years of age. Baldvinsson et al. described that of all patients who underwent lung resection with curative intend in Iceland between 1991 and 2014, only 5% in the group of patients ≥ 75 years received adjuvant chemotherapy (13). Similarly, in a study of 337 patients ≥ 80 years with stage I to IIIA NSCLC, only 5 patients received adjuvant chemotherapy and 1 patient received adjuvant radiotherapy following lung cancer surgery (19). Yamanashi et al. compared 246 patients ≥ 75 years with stage IB - IIIA NSCLC who received chemotherapy (n = 102) or best supportive care (n = 144) in a retrospective observational study. After controlling for baseline characteristics using propensity score matching, they found adjuvant chemotherapy being significantly associated with a reduced disease-free survival (30). In contrast, a meta-analysis including 4584 patients with stage I – IIIA NSCLC demonstrated a positive effect of adjuvant chemotherapy on disease-related survival in patients ≥ 70 years of age compared to younger cohorts (31). Similarly, Yano et at. showed that adjuvant postoperative chemotherapy did not adversely affect the clinical performance status 2 years after treatment in a prospective study of 272 patients with NSCLC aged ≥ 75 years (32). While randomized controlled trials of adjuvant chemotherapy in elderly patients are lacking, the importance of isolated adjuvant chemotherapy is decreasing due to the perioperative use of molecular targeted therapy and immunotherapy.
Overall, the median OS was 64.2 months (95% CI 55.7–72.5). Both overall survival and DFS were higher in the younger patients. Median OS was 63.3 months and DFS 43.5 months in the elderly, while median OS was 73.4 months and DFS 69.3 months in the younger patients. In both groups, a significant association was found between OS and ECOG status, sex, and postoperative tumor stage. The influence of tumor stage on OS after lung resection procedures in NSCLC has been demonstrated in many studies. The Kaplan-Meier graphs (Fig. 2) show a difference in OS in patients aged ≥ 75 years probably because of a higher likelihood of dying from non-tumor-related causes as we do not demonstrate lung cancer specific survival. In elderly patients, a more pronounced association between existing comorbidities and survival probability was demonstrated compared to the younger patient group. Patients with diabetes showed a significantly reduced OS and DFS in the univariate analysis and patients with a preoperative serum creatinine ≥ 1.1 mg/dl had reduced OS in the univariate and multivariate regression analysis. Inflammation-based scores showed an association between age and survival. In patients aged ≥ 75 years, SCS > median, CAR ≥ 0.3, GPS ≥ 1, and mGPS ≥ 1 were all significantly associated with reduced OS in univariate analyses. The association between CAR and OS remained in the multivariate regression analysis. In the younger patient there was no significant association between inflammation-based scores and survival. In line with these results Hino et al. demonstrated in octogenarians that GPS ≥ 1 and CCS ≥ 2 was associated with reduced OS (19). Miyazaki et al. also reported an association of GPS and OS in a study population of 97 patients aged ≥ 80 with NSCLC stage I (33). Miura et al. demonstrated that sarcopenia that progresses in the postoperative course is associated with reduced OS in elderly patients with NSCLC (34). Whether this association in the older patients is an expression of a reduced nutritional status or an increased catabolic situation due to the tumor disease remains controversial. On examining the histological subtypes, we demonstrated that adenocarcinomas were significantly less frequent in elderly patients, whereas there were no age-related differences in the frequency of large cell, adenosquamous, and sarcomatoid carcinomas. This finding is consistent with a study by Mery et al., who demonstrated that squamous cell carcinomas were more frequently observed in patients aged ≥ 75 years compared to younger populations (35). Only in the younger cohort, a significant association between adenocarcinoma subtypes and overall survival was observed with lepidic adenocarcinoma being associated with the most favorable outcome, which is consistent with findings from the literature (36–38). It is likely that the generally poorer OS in the elderly does not reflect the beneficial effect of the subtype.
This study has some limitations. Firstly, the retrospective design, leading to incomplete data sets. Secondly, patients receiving VATS had a lower tumor stage and were older, which constitutes a selection bias for the comparison of surgical approaches and postoperative complications. The aim of matching is to reduce bias between study cohorts. However, the number of possible matching variables is limited by the number of patients. A limitation of all matching methods is that the influence of confounders cannot be ruled out.