Lung cancer is one of the leading causes of cancer-related death worldwide (Bray et al., 2018). Non-small cell lung cancer (NSCLC) represents about 80% of all lung cancer cases. Over 30% of NSCLC cases are detected in a locally-advanced stage which is often unresectable (Siegel et al., 2012). Around 25% of NSCLC patients are diagnosed as stage III disease. This heterogeneous group of patients is primarily treated with curative platinum-based chemoradiotherapy (CRT) regardless of histology and molecular subtype (Xiao and Hong, 2021). While concurrent CRT (cCRT) improves survival rates, the prognosis of stage III NSCLC remains poor, with a median progression-free survival (PFS) of 8 months and a 5-year overall survival (OS) of 15%–30% (Yoon et al., 2017).

The PACIFIC trial (Antonia et al., 2017) that aimed to explore better therapeutic options for NSCLC, has provided a new standard of management of unresectable stage III NSCLC patients. This phase 3 randomized controlled trial (RCT) examined the efficacy of consolidation durvalumab, a programmed death-ligand 1 (PD-L1) immune checkpoint inhibitor (ICI), following curative cCRT for stage III disease. Recently released 5-year results of the trial indicate an OS of 42.9% and PFS of 33.1% with consolidation durvalumab therapy (Faivre-Finn et al., 2021). These outcomes have validated the survival benefits offered by durvalumab as compared to CRT alone thereby establishing a new standard-of-care. Durvalumab has now been approved by healthcare authorities around the globe for the management of unresectable stage III disease (Ettinger et al., 2019).

It is well-known that patients enrolled in clinical trials are not entirely representative of patients seen in clinical practice. The rigorous inclusion criteria of RCTs often exclude the elderly, those with comorbidities, and with poor performance status leading to the inclusion of a younger and healthier patient population who may endure treatment-related adverse events with clinical benefits (Blonde et al., 2018). Therefore, the results of pivotal RCTs often lack external validity and need to be complemented by real-world data that include heterogeneous patient population (Pasello et al., 2020), and may provide more detailed information on the actual efficacy and toxicity of the new therapeutic agent. Since the results of the PACIFIC trial (Antonia et al., 2017), several studies have reported the real-world safety and efficacy of consolidation durvalumab for stage III NSCLC after curative CRT (Faehling et al., 2020; Jung et al., 2020; Miura et al., 2020; Desilets et al., 2021). Recently, a systematic review by Wang et al. (2022) aimed to collate such evidence. However, this review included just 13 studies. Furthermore, the survival analysis of the review had only five studies and a meta-regression analysis was not performed due to insufficient data. In view of these limitations, the aim of our current study was to conduct an updated systematic review and meta-analysis assessing the real-world efficacy and safety of consolidation durvalumab after CRT for stage III NSCLC.

Real-world studies are an integral part of the evidence on the efficacy and safety of new anticancer drugs. They are highly valuable as complementing RCTs due to their high external validity (Blonde et al., 2018). Since the PACIFIC trial (Antonia et al., 2017), several authors have reported real-life data of treating unresectable stage III NSCLC with durvalumab after CRT. Our study presents the most updated compiled evidence on the subject.

An important advantage of real-life studies compared to their baseline RCTs is their broader inclusion criteria. This is apparent when the baseline characteristics of the patients in the PACIFIC trial (Antonia et al., 2017) and the included real-world studies are compared. Firstly, most lung cancer patients are elderly with a median age of 70 years (Bray et al., 2018). However, such patients are frequently excluded from RCTs. In the PACIFIC trial (Antonia et al., 2017), the median age of patients was 64 years while in most of the real-world included studies the median age was higher than 64. In the studies of Sankar et al. (2022), Tsukita et al. (2021) Nishimura et al. (2022), and LeClair et al. (2022), the median age the highest, at 69, 70, 69.5, and 69.8 years, respectively. This age range is much more reflective of the actual age of NSCLC patients in actual clinical practice. Secondly, in the PACIFIC trial (Antonia et al., 2017) only patients with ECOG PS of 0–1 were eligible for inclusion, which is consistent with clinical trials for other ICIs (Dall’Olio et al., 2020). Individuals with ECOG >2 represent a fragile set of patients who have a poor survival rate and are more prone to complications (Albain et al., 1991). While the US Food and Drug Administration and European Medicines Agency have approved the use of ICIs in this cohort, evidence for such patients is usually extrapolated from RCTs. It can be noted that many of the real-world studies in this review included patients with ECOG >2. Thirdly, only patients undergoing cCRT were included in the PACIFIC trial (Antonia et al., 2017). While cCRT achieves better survival as compared to sCRT, sCRT is still used in elderly cohorts and in patients with large tumor volumes (Xiao and Hong, 2021). Lastly, the time from CRT to durvalumab was set as <42 days in the PACIFIC trial (Antonia et al., 2017). This duration was exceeded in many studies, included in our review.

Despite these differences in baseline characteristics, our real-world analysis demonstrated that the 1-year OS and PFS were in line with the PACIFIC trial (Antonia et al., 2017). In fact, the OS and PFS in our analysis were only marginally higher (our review vs. PACIFIC; OS: 85% vs. 83.1% and PFS: 60% vs. 55.9%) (Antonia et al., 2017; Antonia et al., 2018). In comparison, the previous real-world review of Wang et al. (2022), which included only five studies in the survival analysis, reported a 1-year OS (379 patients) and PFS (330 patients) of 90% and 62%, respectively. In the current review, the OS and PFS analysis included 14 and 15 studies with 1969 and 1,379 patients, respectively, potentially providing better estimate of the real-world data of the actual survival rates. The higher survival rates in real-world studies could be due to the heterogeneity in the use of Response Evaluation Criteria for Solid Tumors amongst different centers leading to overestimation of treatment efficacy (Wang et al., 2022). Secondly, patient follow-up and screening may not be as rigorous in clinical practice as compared to clinical trials which too could raise the survival figures.

The results of our meta-regression analysis showed that an increased proportion of patients with an ECOG score of 0 had better PFS. ECOG represents the ability of patients to care for themselves and a physical ability in terms of walking, working, or time spent confined to a bed or chair during waking hours (Dall’Olio et al., 2020). A lower score indicates a healthier patient which may have influenced the PFS rates. However, we noted that median age did not influence OS and PFS. Treatment of older cancer patients is challenging as the clinician needs to balance between toxicities of combined multiple treatments and a potential curative effect. Furthermore, due to poor ECOG status and multiple comorbidities such patients are frequently undertreated (Socinski et al., 2021). In line with our results, Lau et al. (2021) have reported that elderly patients receiving consolidation durvalumab after CRT for stage III NSCLC have similar overall response rates, OS, and PFS as compared to their younger counterparts. Post-hoc analysis of the PACIFIC trial also suggests similar efficacy of durvalumab in elderly patients (Socinski et al., 2021). However, these results must be interpreted with caution due to the selection bias in real-world studies and the small sample size of elderly patients in RCTs.

We reported that median time from CRT to durvalumab did not influence the survival rates. Unlike the PACIFIC trial (Antonia et al., 2017) that only included patients receiving durvalumab within 42 days, real-world studies included patients beyond this time range. Indeed, patients in the PACIFIC trial (Antonia et al., 2017) were healthier (as demonstrated by their age and ECOG status) and could recover quickly from the adverse events of CRT. The prolonged time of durvalumab initiation in the real world points out to the heterogeneous population requiring longer time to recover from the toxicities of CRT. Additionally, logistic delays may occur in the real-world setting due to off-protocol systemic treatment planning and administration. In the PACIFIC trial (Antonia et al., 2018), OS was reduced in patients initiating durvalumab after 14 days. However, Desilets et al. (2021) have found no impact of durvalumab timing on OS in a real-world setting. Further studies are needed to assess the association of the time lag between CRT and durvalumab with the survival rates.

The association between PD-L1 expression and survival with ICI is unclear. Some studies suggest that higher PD-L1 expression results in better treatment outcomes with ICIs (Herbst et al., 2014; Rizvi et al., 2015), while others suggest otherwise (Brody et al., 2017). Due to heterogeneity of reporting in the included studies, we used two moderators for the meta-regression analysis, namely, the proportion of patients with PD-L1<1% and ≥50%. However, we noted no relationship between PD-L1 expression and survival. Similarly, the PACIFIC trial investigators noted no difference in OS and PFS in different subgroups based on PD-L1 status, although the trial was not initially designed to interpret such difference (Antonia et al., 2018). Similar results were reported by the real-world studies of Offin et al. (2020), Desilets et al. (2021) and Faehling et al. (2020). However, Jazieh et al. (2021) demonstrated that patients with >50% PD-L1 expression had significantly longer PFS and OS than those with lower PD-L1 expression. This variability in the results may be due to the variability in testing for PD-L1 status and the sample size. Future RCTs stratifying patients based on PD-L1 expression are needed for stronger evidence.

The rates of endocrinal, cutaneous, and musculoskeletal toxicities in our analysis were 11%, 8%, and 5%, respectively, while in the PACIFIC trial (Antonia et al., 2017) they were 11%, 7%, and 5%, respectively. The incidence of all-grade pneumonitis was slightly lower in our analysis as compared to the PACIFIC trial (Antonia et al., 2017) (27% vs. 33.9%). However, we reported higher incidence of grade ≥3 pneumonitis (8% vs. 4%) and more frequent discontinuation of therapy due to pneumonitis (17% vs. 4.8%) compared to the PACIFIC trial data. A higher incidence of severe pneumonitis and more frequent rate of therapy discontinuation in real-world studies could be due to baseline differences in the study population. Sankar et al. (2022) in their retrospective analysis of the US veterans cohort noted a higher proportion of smokers and baseline lung disease in the veteran’s cohort as compared to the PACIFIC trial (Antonia et al., 2017). Both these factors are associated with severe disease (Nishino et al., 2016). We noted higher rates of pneumonitis in Asian population as compared to Western studies, suggesting a role of ethnicity in the incidence of pneumonitis. Furthermore, studies with a higher proportion of elderly patients had increased pneumonitis rates. Similar results were reported by the previous review (Wang et al., 2022). We showed that delays in durvalumab administration reduced the incidence of pneumonitis. It is known that radiation pneumonitis is associated with thoracic radiotherapy. While significant advances have been made to limit the incidence of radiation pneumonitis using 3D conformational techniques and intensity-modulated radiotherapy, the use of radiotherapy combined with immunotherapy is still linked to higher incidence of pneumonitis. In most cases, it is difficult to distinguish between radiation and immune-mediated pneumonitis (LeClair et al., 2022). Thus, a delay in durvalumab could have decreased the burden of radiation pneumonitis in the studies by allowing recovery time and reducing the overall incidence of pneumonitis. Lastly, our analysis indicated a reduced incidence of pneumonitis with an increased proportion of patients with PD-L1 expression of ≥50%. However, previous studies showed that high PD-1 expression correlates with an increased risk of pneumonitis. Increased PD-L1 leads to increased T-cell activation, which can cause collateral damage to normal lung tissue manifesting as pneumonitis (Chao et al., 2022). Such contradictive results are difficult to explain since we noted no such relationship when PD-L1 expression <1% was used as a moderator for the same analysis. More research is needed to assess the role of PD-L1 expression and the risk of pneumonitis associated with durvalumab treatment.

Our review has some limitations. First, real-world studies are observational and are prone to bias due to errors in record-keeping or data entry. The role of selection bias in influencing results cannot be excluded. Secondly, there was high interstudy heterogeneity in most of our results, which was expected due to variability in study populations, patient characteristics, and treatment protocols. We attempted to explore the source of such heterogeneity by subgroup and meta-regression analyses. Thirdly, many of the included studies had a small number of patients with limited follow-up. All outcome data were not universally reported, which reduced the number of studies in each meta-analysis. Fourthly, our analysis consisted of only single-arm studies. Future real-world studies comparing outcomes of unresectable stage III NSCLC with and without durvalumab can provide optimal complementary evidence to the PACIFIC trial (Antonia et al., 2017). Lastly, since follow-up of most included studies was short, we could assess only 1-year survival rates.

Nevertheless, our review presents the most comprehensive and up-to-date real-world evidence on the efficacy of durvalumab after CRT for unresectable stage III NSCLC. The pooled figures generated by our review may be used as a guide to physicians treating the heterogeneous cohort of stage III NSCLC patients in clinical practice that are usually not eligible or healthy enough for participation in clinical trials. These figures also provide a real-world estimation of the efficacy and safety of the drug and would help in patient-doctor interactions. The congruence of the results of our review and the PACIFIC trial (Antonia et al., 2017) further supports the use of durvalumab in improving outcomes in unresectable stage III NSCLC.

Real-world evidence suggests that the short-term efficacy and safety of durvalumab are consistent with that of the PACIFIC trial. A higher proportion of severe pneumonitis is seen in clinical practice leading to treatment discontinuation. There is a need for further research analyzing the impact of various confounders like age, ethnicity, time to durvalumab, and PD-L1 status on the overall outcomes.