The introduction of immune checkpoint inhibitors (ICIs) has paved the way for radical changes in the treatment of advanced/metastatic non-small cell lung cancer (NSCLC). The superiority of ICIs over standard chemotherapy (CT) has been widely demonstrated; however, there is now a rising need to identify which patients are most likely to benefit from immunotherapy (IT). In this challenging scenario, ICIs-based therapy in elderly/frail patients is still a relevant point of discussion that requires further investigation. In clinical practice, more than half of all patients with NSCLC are aged over 70 years, and nearly 10% are 80 years or older (1). Due to immune-senescence, there is a hypothetical risk of reduced efficacy and increased toxicity with ICIs. However, some data from clinical trials suggest that older patients might benefit from IT similarly to younger patients, with an acceptable safety profile (2). However, clinical trials mostly include patients with a performance status (PS) of 0–1 and median age at trial enrollment was about 10 years younger than the median age of NSCLC diagnosis. For this reason, data on ≥75 years or those with PS 2 patients are mostly derived from post-hoc analyses of small subgroups with limited statistical power and high risk of selection bias. Unlike CT, ICIs treatment is often given until disease progression or unacceptable toxicity, and the impact of this long-term treatment remains unclear. Additionally, the combination of IT and CT has become the standard of care in first-line NSCLC improving efficacy as compared to CT alone but also leading to a higher rate of adverse events. Therefore, there is a rising medical need to identify the most appropriated treatment strategy for frail and elderly populations to avoid over- or under-treatment and preventing useless toxicity. Indeed, this requires the design of pragmatic clinical trials that enroll populations as more similar to those observed in the real-life setting.

IT with ICIs has become a cornerstone in the treatment of advanced/metastatic NSCLC. There is growing interest in defining the efficacy and safety of this approach in older and frail patients, identifying the most effective treatments in this setting. Due to the poorly characterized immune-senescence phenomenon, it remains unclear whether ICIs-based treatments are less effective in older patients. Aging-related changes in the immune system, collectively known as immune-senescence, may contribute to resistance to IT (37). Thymic involution and chronic antigenic stimulation cause naive T cells to convert into virtual memory cells, potentially impairing immune responses to pathogens and tumors. Additionally, the reduced number of naive CD8+ T cells, along with an increase in antigen-experienced CD4+ and CD8+ T cells, leads to a diminished capacity to respond to newly encountered antigens. Furthermore, decrease of costimulatory signals (like CD28 and CD27 on T cell) or the upregulation of Tim-3 and CD57 were described in the elderly and have been linked to reduced response to ICIs (38). Immune-senescence phenomenon involves also B cells and innate immune response (39).

Despite this, individual clinical trials have shown a similar survival benefit of ICIs compared to CT across younger and older patients with an acceptable safety profile, based however on underpowered post-hoc analyses unable to definitely address this crucial point. At this aim, this systematic review and meta-analysis were carried out to summarize and rank the efficacy of first line ICIs-based treatments in advanced/metastatic NSCLC, considering age and PS. Our findings confirmed that ICIs-based regimens significantly improved OS and PFS compared to CT in all subgroups, with a greater impact on PFS. A previous meta-analysis by Zhou et al. found that PFS benefits were more pronounced than OS benefit in the first-line setting, whereas the opposite was observed in later treatments lines (40). For patients over 75 years, our analysis did not demonstrate a statistically significant OS benefit of ICIs over CT. This finding may be due to limited data and small sample size of this subgroup, thereby limiting the confidence in the observed outcome. PFS could not be analyzed due to insufficient data (only 3 studies). Notably, only IM-131 (25) reported a statistical significant PFS (77 patients), whereas CM-9LA (70 patients) and NIPPON (47 patients) showed negative outcomes (31, 35). Caution is required in interpreting these results, and validation in larger cohorts of elderly patients is needed.

In older patients, ICI monotherapy seems to perform better in the pairwise as well as NMA, while ICI+CT was the best strategy in younger patients. In patients aged over 65 years ICI+CT (pembrolizumab+CT, cemiplimab+CT and camrelizumab+CT) ranked highest in OS, though without statistical significance in all comparisons. In patients over 65 years cemiplimab monotherapy ranked first in OS with statistical significance for most comparisons. For PFS, camrelizumab+CT and pembrolizumab+CT ranked highest in <65 years (with a statistical significance only over nivolumab, CT, durvalumab+tremelimumab, avelumab, and atezolizumab+CT) whereas pembrolizumab monotherapy had the highest probability of reducing disease progression in ≥65 years (non-statistically significant comparisons: pembrolizumab+ipilimumab, camrelizumab+CT, cemiplimab, toripalimab+CT, serplulimab+CT). When stratified by PS, OS differences between PS 0 and PS 1 were minimal whereas for PFS ICI/CT performed slightly better in both PS 0 and PS 1. In PS 0 patients, cemiplimab+CT ranked first in OS (statistically significant vs. all treatments) with pembrolizumab+CT ranked second. In PS 1, cemiplimab monotherapy performed best in OS, while cemiplimab+CT and pembrolizumab+CT performed worse than in PS 0.

These rankings led us to conclude that: i) combination therapies appear more effective in younger and fit patients, potentially due to a stronger immune response and/or better treatment tolerance; ii) mono immune-therapy appeared more effective in older and PS 1 patients, likely reflecting a different benefit-risk balance and reduced treatment tolerance. Furthermore, our analysis of ICIs type demonstrated that anti-PD1 therapy (alone or in combination), outperformed anti-PD-L1 and anti-CTLA-4 regimens in OS.

Unlike previous meta-analyses investigating this field, our study provides a more comprehensive comparison of available treatment strategies in this setting, including the largest number of RCTs and the most recent data updates. Moreover, at the best of our knowledge this is the first NMA considering both age and PS to investigate the frailty scenario.

Our results should be evaluated in the current landscape of studies focused on the relevance of age and PS. Landre et al. performed a NMA considering only patients over 65 years, demonstrating OS and PFS benefit of the anti-PD-1/PD-L1 plus CT, but with no consistent evidence in patients aged over 75 years (41). In a NMA focused only on older patients with PDL-1 ≥50%, cemiplimab monotherapy emerged as the preferred treatment strategy, consistent with our previous findings (42). Instead, Sun et al. found similar OS efficacy across age groups but no PFS benefit in either young and older patients (though their analysis included only 8 trials) (43). Regarding PS, an interesting meta-analysis in real-world advanced NSCLC patients with PS ≥2 reported detrimental effects of ICIs therapy on OS, PFS and ORR, raising concerns about treatment suitability in frail patients (42). All together these studies support our finding of relevant role of age and PS in predicting effectiveness of immune-oncology strategies. However, it should be emphasized the under-representation of frail patients in clinical trials, limiting the applicability of trials data in real-world clinical practice. Chronological age and PS alone do not fully capture a patient’s condition; a comprehensive, multidisciplinary evaluation is necessary to assess overall PS and health. PS deterioration may result from various factors, including disease burden, comorbidities, age, and the overall frailty, making it critical to differentiate between cancer-related PS decline and other underlying conditions. Comprehensive Geriatric Assessment (CGA) represents a well-established, multidimensional approach for evaluating older patients. Initially developed within geriatrics, CGA has been increasingly applied in oncology to guide treatment decisions in elderly cancer patients and to better characterize frailty (44). Recent prospective data in elderly NSCLC patients undergoing CGA suggest that frailty, comorbidities, and low albumin levels are associated with worse survival outcomes and higher toxicity, highlighting the importance of CGA-guided treatment decisions (45). Among elderly patients, safety concerns remain a key consideration due to impaired renal and/or cardiac function, increased comorbidities, declining organ function and cognitive impairment (46). Studies suggest that ICIs safety profile is comparable between older/frail and fit patients, though CT-based combinations inevitably increase toxicity (47–49). Therefore, adequate geriatric screening is essential to prevent both over- or under-treatment in this population.

Several limitations of this meta-analysis should be acknowledged. Firstly, data were extrapolated from published RCTs rather than individual patient data. Heterogeneity was evident when pooling data across different ICIs or CT backbone, trial design, histology and PD-L1 expression. Formal analyses of safety and toxicity stratified by age/PS could not be performed because data were not reported in clinical trials. Additionally, several data points relied on post-hoc analyses and ongoing trials have yet to report survival outcomes, introducing potential bias. Longer follow-up is needed to fully assess the long-term impact of ICIs on OS. Furthermore, in older patients (≥75 years), non-cancer-related mortality may significantly interfere with OS outcomes in phase III trials.

This systematic review and meta-analysis confirmed that ICI-based therapy significantly improved OS and PFS compared to CT across all subgroups, except for patients with ≥75 years. The best treatment strategy seems to vary by age and PS, with ICI monotherapy being most effective in older/PS 1 patients, while ICI+CT combinations performed better in younger/PS 0 patients. In conclusion, there is an urgent need to design future RCTs focusing on the use of IT in frail populations as a whole, improving patients stratification using geriatric tools. Considering that the rate of end-of-life IT is increasing, this could guide the clinicians in discriminating situations of over- or under-treatment, providing recommendations for clinical practice in selecting optimal strategies in these patients.