FormalPara Key Summary Points

Management of early-stage resected non-small cell lung cancer (NSCLC) is quickly evolving to include new neoadjuvant and adjuvant treatment options.

It is therefore of interest to understand treatment patterns and clinical outcomes such as survival among patients with resected stage I–III NSCLC who may be eligible for such therapies.

In this real-world study conducted in Ontario, Canada, use of adjuvant chemotherapy was low between 2010 and 2019, even though such treatment was observed to significantly prolong survival among patients with stage II/III disease.

Among patients whose disease was assumed to have recurred/relapsed, median overall survival was markedly reduced compared with that of patients whose disease was assumed to not have recurred/relapsed, regardless of subsequent therapy.

These findings underscore a high need for the newly available and anticipated peri-adjuvant treatment options to improve clinical outcomes in early-stage resected NSCLC.

Introduction

Lung cancer is the most commonly diagnosed malignancy in Canada, accounting for an estimated 13% of all new cancer cases and 24% of all cancer-related deaths in 2023 [1]. Approximately 31,000 new lung cancer diagnoses occurred across the country in 2023 [1], with 10,639 new cases in the province of Ontario in 2022 [2]. Of these diagnoses, non-small cell lung cancer (NSCLC) was the most common disease subtype, representing an estimated 88% of all cases [3]. In Canada, approximately 45–66% of patients with NSCLC present with early-stage disease (stage I–III) at diagnosis [3], in part because of the recent introduction of screening programs that support early detection [4,5,6].

Among eligible individuals, guideline-recommended treatment of early-stage NSCLC conventionally involves surgical resection (e.g., lobectomy, segmentectomy, or pneumonectomy) followed by adjuvant chemotherapy depending on patient and tumor characteristics [7,8,9]. For some patients, such treatment is associated with greatly prolonged survival that may approach cure [7, 10]. For others, however, the risk of recurrence remains high, ranging from 45 to 76% at 5 years among patients with stage IB through stage III disease [10]. Upon progression to advanced disease stages, the goals of therapy shift from cure to prolongation of survival and maintenance of health-related quality of life [11,12,13]. In this setting, recommended treatment options include systemic therapies such as single-agent or doublet platinum-based chemotherapy, molecularly targeted options, and immunotherapy [14, 15]. Numerous molecular biomarkers are now clinically available, such as gene alterations of the epidermal growth factor receptor (EGFR), the Kristen rat sarcoma viral oncogene homolog gene (KRAS), and anaplastic lymphoma kinase (ALK), as well as the programmed cell death 1 receptor (PD-1) and its ligand (PD-L1) [16, 17].

Given the high rates of disease recurrence in NSCLC and the benefits of novel treatments used in advanced NSCLC settings, the efficacy of these agents is being evaluated in earlier disease stages. One such drug category is epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs), which are targeted to patients with certain EGFR mutations (EGFRm). These mutations are observed in 14–22% of Canadian NSCLC cases [18,19,20,21,22,23], with a similar prevalence reported between patients with early-stage versus advanced disease [24, 25]. Initial evaluations of early-generation EGFR-TKIs (i.e., erlotinib and gefinitib) in the adjuvant NSCLC setting showed a limited survival impact and did not lead to changes in clinical practice [26,27,28,29,30,31,32,33]. However, more recently, the third-generation EGFR-TKI osimertinib was associated with favorable outcomes for disease-free survival (DFS) and overall survival (OS) in the phase III ADAURA trial of resected stage IB–IIIA NSCLC [34,35,36]. These results led to osimertinib becoming the first targeted agent approved for adjuvant use in early-stage NSCLC after complete resection, with approval received from Health Canada [37] and multiple other regulatory bodies in 2020 and 2021 [38, 39]. Several other trials of agents used in neoadjuvant and adjuvant settings (e.g., immunotherapies) are ongoing or have now shown similarly positive results, further transforming the management and outcomes of early-stage NSCLC [40,41,42].

With approximately 15 million individuals, the province of Ontario represents 39% of the Canadian population [43]. Recent real-world studies of NSCLC in this region have focused on patients with advanced or metastatic disease, reporting on treatment patterns and survival results [44,45,46,47,48]; others have analyzed much older datasets [49,50,51]. Therefore, there is a paucity of contemporary, real-world, population-level outcomes data for patients with resected early-stage NSCLC. Such data would provide context for the true unmet needs of this patient population in light of growing evidence for the use of novel therapeutics in early-stage patients. Therefore, the objective of the current study was to understand the baseline characteristics and clinical outcomes of patients with resected stage I–III NSCLC living in Ontario, Canada, with a specific focus on those with an EGFRm.

Methods

Study Design

In collaboration with the HOPE Research Centre and the Institute of Clinical Evaluative Sciences (ICES) Data & Analytic Services (DAS), a retrospective, longitudinal evaluation was conducted of real-world, population-level data. Eligible patients included those diagnosed with NSCLC in Ontario, Canada, between April 1, 2010, and March 31, 2019, who had at least 6 months of follow-up after diagnosis. Included patients had a valid Ontario Health Insurance Plan (OHIP) card and met the following inclusion criteria: International Classification of Disease for Oncology, third edition (ICDO3) diagnosis code C34 (locally advanced lung cancer) and relevant morphology codes in the Ontario Cancer Registry (OCR); non-squamous histology; stage I, II, or III NSCLC; and underwent ≥ 1 surgery for lung resection within 180 days of NSCLC diagnosis, as indicated by Canadian Classification of Health Interventions (CCI) codes. Patients were excluded from the study if they did not have a valid ICES (IKN) number, did not have a lung cancer diagnosis in the OCR, had a second primary cancer, died less than 2 weeks after diagnosis, or had erroneous data entry. Patients could have received neoadjuvant or adjuvant therapy, which were defined as therapy received for a maximum of 90 days before or within 6 months after surgical resection, respectively.

Data Sources

Patient data were accessed via the ICES administrative databases, which were used to collect real-world data through OHIP and other linked population-level provincial or national health databases. Individual patients who met the study inclusion criteria were linked to 10 different databases to retrieve treatment and outcomes data. As described previously [44, 45, 56], these databases provide information on various types of clinical and resource utilization; patient diagnoses, demographics, medication use, and vital statistics were of interest for the current study. Given the nature of administrative databases such as the OCR, patients were not recruited at individual hospitals or office-based clinics; rather, index cancer diagnoses were reported and captured in the OCR and then linked with other administrative databases to determine the study outcomes.

Patient Selection

As the ICES databases do not capture EGFRm status, receipt of EGFR-TKI therapy in the setting of disease recurrence/relapse (i.e., > 6 months after resection) was used as a proxy to presume the presence of an EGFRm. This assumption was made on the basis that EGFR-TKI therapy is the standard of care for patients with sensitizing EGFRm (e.g., exon 19 deletion or exon 21 L858R substitution), which represents the vast majority of EGFRm [15, 52]. Moreover, EGFR-TKI therapies are specifically indicated for these individuals in Canada [15, 37, 53,54,55]. Additionally, as EGFRm occur primarily in non-squamous histology [24, 52], patients with squamous disease were excluded from the overall cohort. Notably, the etiology of non-squamous and squamous disease also differs, providing further rationale for focusing on one histological subtype. The ICES databases are also unable to track disease relapse; therefore, receipt of certain therapies (EGFR-TKIs and platinum doublet chemotherapy) more than 6 months after lung resection was used as a proxy to presume recurrence/relapse of NSCLC.

Using the above criteria, four patient cohorts were defined for analysis: (1) all patients meeting the study inclusion and exclusion criteria; (2) an EGFR-TKI cohort of patients who received an approved EGFR-TKI (gefitinib, erlotinib, afatinib, or osimertinib) via the Ontario Drug Benefit (ODB) plan (which includes patients prescribed oral medication who are either aged ≥ 65 years or < 65 years and on disability or social assistance) > 6 months after resection and were assumed to have experienced disease recurrence/relapse and to have an EGFRm; (3) a non-EGFR-TKI cohort of patients who received platinum doublet chemotherapy > 6 months after resection and were assumed to have experienced disease relapse but were unlikely to have an EGFRm; and (4) a non-relapse cohort of patients who did not receive systemic anti-cancer treatment.

Study Outcomes

All study outcomes were reported at the aggregate level. The proportion of patients who underwent resection for non-squamous stage I–III NSCLC within 180 days of diagnosis in Ontario between April 1, 2010, and March 31, 2019 (the overall cohort), was calculated. In this group, baseline patient characteristics were evaluated (by disease stage at diagnosis where available) and included patient age, sex, geographic location (i.e., whether rural), Charlson comorbidity scores, and the time from diagnosis to resection. Treatment patterns were documented in the overall cohort and by disease stage at diagnosis (where available) and included receipt of neoadjuvant or adjuvant therapy, EGFR-TKIs/chemotherapy for recurrence/relapsed disease, and/or curative or palliative radiation. Overall survival was calculated from the date of resection (as well as the time of therapy initiation for patients with recurrence/relapse) until the end of follow-up or death for each of the four abovementioned cohorts.

Statistical Analysis

Statistical analyses were conducted at ICES and were performed using SAS Enterprise Guide software (version 7.1; SAS Institute, Cary, NC, USA). Patient demographics and baseline characteristics were summarized by number and percentage for categorical variables or by mean and standard deviation (SD) for continuous variables. Survival analyses were stratified to examine the time to loss to follow-up or death from any cause from the date of lung resection or metastasis. Median overall survival (mOS) and 95% confidence intervals (CI) were evaluated using Kaplan–Meier methods for censored data. Survival results were compared in a Kaplan–Meier analysis based on stratification and log-rank test.

Ethical Considerations

The study was approved by the Sunnybrook Health Sciences Centre Research Ethics Board (REB #2450–2017) and performed in compliance with Good Clinical Practice (GCP) and Good Pharmacoepidemiology Practice (GPP), including the archiving of essential documents. No patient recruitment or consent was required. Data are retained for 10 years after study completion.

Results

Patient and Clinical Characteristics

Between April 1, 2010, and March 31, 2019, a total of 83,252 patients were diagnosed with lung cancer in Ontario, Canada (Fig. 1). Of these patients, 21,752 had non-squamous stage I–III NSCLC (stage I: 8721; stage II: 3921; stage III: 9,110). In total, 8255 (38.0%) of these patients (the overall cohort) underwent resection for early-stage lung cancer within 180 days of diagnosis, at rates of 56.0% (n = 4881), 54.2% (n = 2124), and 13.7% (n = 1250) for stage I, II, and III disease, respectively (Table 1). Among those who underwent resection within 180 days of diagnosis, almost two-thirds (63.5%) of patients underwent resection within 60 days of diagnosis. The mean age of patients was 68.0 years (SD 9.1), with slightly more female (53.5%) than male patients (Table 1). The mean Charlson comorbidity scores were 0.61, 0.62, and 0.67 for stage I, II, and III disease, respectively. Approximately 15% of patients lived in a rural setting across all groups.

Fig. 1
figure 1

Flow chart of patients considered and included in the analysis

Full size image
Table 1 Characteristics of patients with early-stage (I–III) resected non-squamous NSCLC
Full size table

Treatment

In the overall cohort, adjuvant chemotherapy was received by 1617 patients (19.6%). Use of such therapy was highest among patients with stage III disease (50.0% vs. 3.5% in stage I and 38.7% in stage II; p < 0.001 across disease stages) (Table 1). A total of 195 patients (2.4% of the overall cohort; 19.1% of patients who received EGFR-TKIs/chemotherapy) received an EGFR-TKI via ODB; 825 patients (~ 10%) received chemotherapy in this setting. In the overall cohort, curative radiation was received by 1215 (14.7%) patients and palliative radiation was received by 1276 (15.5%). The proportion of patients receiving these treatments increased with advancing disease.

Overall Survival

In the overall cohort, mOS after the time of resection was 8.2 years (95% CI: 7.8, 8.7) (Fig. 2). When stratified by disease stage at diagnosis and receipt of adjuvant chemotherapy, patients with stage I disease who did not receive adjuvant therapy had prolonged mOS compared with those who did receive such therapy (not estimable [NE] vs. 7.5 years [95% CI: 6.3, NE]) (Fig. 3). Conversely, patients with stage II or III disease who did not receive adjuvant treatment had shorter mOS than those who did receive it (stage II: 4.4 years [95% CI: 4.0, 4.9] vs. 7.6 years [95% CI: 6.5, 8.5]; stage III: 2.7 years [95% CI: 2.3, 3.3] vs. 4.4 years [95% CI: 3.6, 5.1]). The duration of mOS was similar between patients with stage I or stage II disease who received adjuvant chemotherapy and between those with stage II or stage III disease who did not receive adjuvant chemotherapy.

Fig. 2
figure 2

Kaplan–Meier OS after resection among patients with early-stage (I–III), non-squamous NSCLC: overall cohort. CI confidence interval; NSCLC non-small cell lung cancer; OS overall survival

Full size image
Fig. 3
figure 3

Kaplan–Meier OS after resection in the overall cohort, with stratification by disease stage at diagnosis and receipt of adjuvant chemotherapy. Adj. adjuvant; CI confidence interval; Chemo chemotherapy; NE not estimable; OS overall survival

Full size image

Among patients who received therapy in the recurrence/relapse setting, mOS after resection was approximately 1.5 years longer in the EGFR-TKI cohort than in the non–EGFR-TKI cohort (5.1 years [95% CI: 4.4, 5.8] vs. 3.6 years [95% CI: 3.2, 3.9]) (Fig. 4). Similarly, mOS after disease metastasis was 1.2 years longer in the EGFR-TKI cohort (2.3 years [95% CI: 1.8, 3.0] vs. 1.1 years [95% CI: 1.0, 1.3] in the non-EGFR-TKI cohort (Fig. 5). In the non-relapse cohort of patients who did not receive therapy, mOS was not reached after a median follow-up of 9 years.

Fig. 4
figure 4

Kaplan–Meier OS from the date of resection in the non–EGFR-TKI and EGFR-TKI cohorts of patients who received therapy for recurrence/relapse. CI confidence interval; EGFR-TKI epidermal growth factor receptor-tyrosine kinase inhibitor; OS overall survival

Full size image
Fig. 5
figure 5

Kaplan–Meier OS from the date of treatment initiation in the non–EGFR-TKI and EGFR-TKI cohorts. CI confidence interval; EGFR-TKI epidermal growth factor receptor-tyrosine kinase inhibitor; OS overall survival

Full size image

Discussion

Numerous treatment options have been developed for locally advanced and metastatic NSCLC over the last 15 years; however, until recently, progress in the management of early-stage disease has remained limited to the introduction of adjuvant cisplatin-based doublet chemotherapy. Although studies of the adjuvant use of early-generation EGFR-TKIs (e.g., gefinitib and erlotinib in the SELECT, RADIANT, ADJUVANT-CTONG, EVAN, and NCIC CTG BR19 trials) have shown some improvements in DFS, no statistically significant benefit in OS was reported. It should be noted that these findings have been challenged because of inappropriate patient selection, poor study design, inadequate treatment duration, and/or limited efficacy [26,27,28,29,30,31,32,33, 57]; still, the early EGFR-TKIs did not receive regulatory approval or the opportunity for reimbursement in Ontario for adjuvant use. In contrast, given recent positive results for osimertinib, atezolizumab, pembrolizumab, durvalumab, and nivolumab in neoadjuvant, adjuvant, and perioperative settings [36, 40,41,42, 58], additional practice-changing therapeutic options are finally becoming available for early-stage NSCLC. Therefore, it is important to understand the real-world characteristics, treatment, and survival of patients who may be eligible for these therapies.

This retrospective analysis represents the most recent evaluation to address this need for real-world data, providing perspective on treatment patterns in Ontario, Canada, and including a study period sufficient to capture survival outcomes across multiple patient subpopulations. Previous Canadian real-world studies have focused on advanced or metastatic NSCLC [44,45,46,47,48] or analyzed considerably older datasets of patients with early-stage resected disease [49,50,51]. In the latter group of studies, patient data were derived from 2001 to 2006, with treatment patterns and survival outcomes examined in relation to patient age (i.e., elderly patients) and timing of adjuvant chemotherapy. These studies also included patients with de novo stage IV disease and were conducted before the availability of certain treatments used in the metastatic setting (e.g., some EGFR-TKIs), which would have impacted their survival results. The current study is therefore unique, having included patients diagnosed with early-stage NSCLC who underwent resection and following them through the long-term progression of their disease. As adjuvant osimertinib is now approved and reimbursed across most of Canada [37], a key goal was to understand the characteristics of patients with an EGFRm.

The current study found that 38% of patients with non-squamous stage I–III NSCLC underwent surgical lung resection, with rates of 56% and 54% observed among those with stage I and II disease, respectively. These relatively low rates are surprising, given that lung resection has been the standard of care for eligible patients with early-stage disease [8, 9]. It is possible, however, that some patients who did not undergo resection received radiotherapy, and specifically stereotactic body radiotherapy, an approach that has increased in use in recent years but was not captured in this study. Regardless, the resection rates for stage I and II disease were generally similar to those reported in an earlier Ontario-based study of stage I—II lung cancer (58.9%; 2010–2012 dataset) [59], as well as those published for other regions [60, 61]. In the previous study of data from Ontario, the authors noted a potential relationship between rates of resection and surgical consultation [59]; however, the latter outcome was not evaluated in the current analysis.

This study also found that in the overall cohort, only 38.7% of stage II and 50.0% of stage III patients received adjuvant chemotherapy after surgical lung resection. In previous real-world investigations of NSCLC conducted in Ontario, utilization rates for adjuvant therapy ranged from 31 [49] (stages I–IV) to 46% [25] (stages IB–IIIA) of patients who underwent resection. The low utilization observed in the current study is concerning, given current guideline recommendations [7,8,9] based on results of the LACE meta-analysis and others: the evidence clearly indicates that adjuvant chemotherapy is associated with a survival benefit among patients with stage II and III NSCLC [10, 62, 63]. Indeed, in the current study, mOS was up to 1.7 times longer among stage II/III patients who had received adjuvant therapy. An alternative, partial explanation for such limited use may relate to patient fitness (e.g., age, frailty), physician preferences, and/or patient concerns related to the potential toxicity of chemotherapy. This finding may also underscore a need for novel adjuvant treatment options during the study period.

Not surprisingly, the overall proportion of patients receiving neoadjuvant chemotherapy was also very low (0.9%) in the current analysis. However, given recent US approval of nivolumab with platinum-doublet chemotherapy (based on CheckMate-816) [64] in this setting, positive interim results for perioperative durvalumab with chemotherapy (AEGEAN trial) [58] and pembrolizumab with chemotherapy (KEYNOTE-671 trial) [65], and ongoing evaluations of other treatments (e.g., osimertinib in NeoADAURA), it will be of interest to monitor the use of neoadjuvant/perioperative therapy and surgery in the near future. Changes in treatment patterns among patients with stage III disease will be of particular interest, should they become eligible for resection after neoadjuvant therapy.

The current study additionally showed that unlike the results for stage II/III disease, patients with stage I disease who received adjuvant chemotherapy had significantly worse mOS than those who went without adjuvant chemotherapy. Although disease substaging details were not available for analysis, this finding is likely explained by the fact that stage IA disease is not typically treated with adjuvant therapy and is associated with better survival outcomes than stage IB [10]. In the overall cohort, 59% of patients had stage I disease, and only 3.5% of these individuals received adjuvant therapy. Additionally, a higher proportion of stage I patients who received adjuvant therapy also received curative or palliative radiation therapy (data not shown), suggesting the presence of larger tumors at baseline. Collectively, these factors may explain the nonintuitive survival results for adjuvant therapy in the stage I population. Regardless, it is recognized that in general, the presence of certain molecular alterations that were not evaluated in this study may have impacted clinical outcomes [66]. For example, a meta-analysis of 41 publications found that patients with KRAS mutations had significantly worse survival than those with wild-type tumors [67].

Another notable observation was that among patients who received therapy in the recurrence/relapse setting, 19.1% (195 of 1020) received an EGFR-TKI via ODB. Although receipt of EGFR-TKI therapy was used as a proxy for EGFRm positivity in this study, this frequency appears to be reasonable, as other studies of EGFRm prevalence have reported frequency of 20–24% in Canadian and North and South American populations [21, 24, 25]. Considering that mOS after resection was longer in the EGFR-TKI cohort than in the non-EGFR-TKI cohort (who received platinum doublet chemotherapy), it becomes evident that patients with EGFRm must be appropriately identified and receive timely therapy with an EGFR-TKI. It should be noted that most EGFR-TKI-treated patients received gefitinib (n = 130), an early-generation therapy that had an mOS of approximately 1.8 years in the phase II IPASS trial (advanced NSCLC EGFRm population) [68]. Although the current real-world study’s results are generally aligned with this finding, increased use of the third-generation EGFR-TKI osimertinib may further extend the survival benefit (mOS in the FLAURA trial was ~ 3.2 years) [69]. Additionally, in the non-relapse cohort, mOS was not reached over 9 years of follow-up. This finding indicates that early-stage patients who do not experience disease relapse may survive to an age similar to that in the general population. Thus, cure is a reasonable expectation and should be the goal for all early-stage patients with resected NSCLC.

This study has both strengths and limitations. The study included a large number of patients, which may permit generalizability of some results to the broader North American population. Most of the study’s limitations are common to retrospective analyses of secondary data. Firstly, the ICES database did not provide information on NSCLC disease substaging (e.g., IA, IIB) at diagnosis, disease stage at treatment initiation, or any patient performance status (e.g., Eastern Cooperative Oncology Group status). These factors can impact treatment selection and subsequent clinical outcomes, including survival. Secondly, as the ICES databases did not capture information on the EGFRm status of patients, EGFR-TKI prescribing in the setting of disease recurrence/relapse was used as a proxy for the presence of the mutation. Therefore, it is possible that some patients with EGFRm were included in the non-EGFR-TKI and non-relapse cohorts. Thirdly, patient status in terms of disease progression was based on the prescribing of medications used to treat disease recurrence/relapse rather than other health record information. As a result, some patients who experienced disease relapse but did not receive subsequent therapy were likely included in the non-relapse cohort. Although prolonged OS (potentially reflecting cure) was observed in this group, the presumably reduced survival among untreated progressed patients likely reduced the estimate. Furthermore, as mentioned previously, the survival results may have been impacted by the presence of unidentified genetic mutations. Fourthly, in Ontario, public funding of oral cancer agents is limited to individuals aged ≥ 65 years or who are on social assistance (regardless of age); accordingly, information for the utilization of these agents among patients aged < 65 years is limited, and given this, analyses were conducted using data from ODB. Finally, there was a need for patient censoring in some cases (e.g., due to loss to follow-up, loss of OHIP coverage), data were potentially missing for some outcomes, and there was a paucity of data on the cause of death.

Conclusions

The results of this study show that among patients with resected, early-stage, non-squamous NSCLC living in Ontario, Canada, adjuvant chemotherapy was received by a relatively small proportion of patients between 2010 and 2019 despite guideline recommendations. Patients who received treatment in the recurrence/relapse setting generally had poor survival, although outcomes varied on the basis of treatment type (and thus by molecular cancer type). Among patients who were assumed to not have experienced relapse, mOS was not reached over 9 years of follow-up. Overall, these findings suggest that the adjuvant treatment options available at the time of the study were underused, and that although cure is possible after resection of NSCLC, many patients still experience disease progression. Accordingly, emerging neoadjuvant and adjuvant treatment options are greatly needed to prevent disease recurrence and improve long-term survival. The study results further indicate that given the fast-moving clinical landscape and evolving options, timely identification of driver mutations will be important for patients with resectable disease as well as those with recurrent/relapsed NSCLC to determine treatment decisions. Once more widely available, the use of liquid biopsy to detect disease recurrence is therefore likely to play a key role in this setting [70]. Collectively, the study findings improve our understanding of the characteristics, treatment, prognosis, and unmet needs of patients with resected, early-stage, non-squamous NSCLC, in turn providing context to the results of recent and ongoing adjuvant clinical studies. As such, the results may be helpful to clinicians and health technology assessment committees making decisions for this population.