Real-world outcomes and toxicity of adjuvant chemotherapy in NSCLC: a single-center experience

Lung cancer is one of the most common malignancies, accounting for an estimated 2.2 million cases per year globally. Despite treatment advances, lung cancer remains the leading cause of cancer death worldwide with an estimated 1.8 million deaths in 2020 comprising 18% of total cancer mortality [1]. National Cancer Registry Ireland (NCRI) data reported an annual incidence of 2671 lung cancer cases in 2017, projected to increase to 5450 annual cases by 2045 [2,3].

NSCLC accounts for approximately 80% of lung cancers. For patients with potentially resectable NSCLC, surgical resection of the tumor is the cornerstone of disease management [4]. Despite this, long term outcomes remain poor for patients with resectable disease with 5-year survival rates between 38 and 60% depending on tumor stage [5]. European Society of Medical Oncology (ESMO) guidelines recommend adjuvant chemotherapy for patients with resected stage II and III NSCLC, as well as resected stage IB disease with a primary tumor >4 cm [6]. Meta-analyses assessing systemic treatment in this population have demonstrated a modest benefit, with an approximately 4–5% improvement in 5-year overall survival (OS) [7,8].

In patients undergoing adjuvant chemotherapy, four cycles of cisplatin-based doublet chemotherapy are considered the standard of care. The most frequently studied regimen is cisplatin in combination with vinorelbine, though phase III data have also demonstrated the efficacy of cisplatin in combination with pemetrexed in non-squamous NSCLC [9–11]. Adjuvant chemotherapy has been associated with significant morbidity in clinical trial patient populations, particularly in regimens combining cisplatin and vinorelbine in which respective rates of grade 4 toxicity and treatment-related hospitalisation of up to 57% and 19% have been reported. Treatment toxicity is associated with significant rates of dose reduction and treatment discontinuation.

There are limited published studies providing real-world patient data on the safety and patient outcomes associated with adjuvant chemotherapy in NSCLC [12,13]. With this study, we aimed to characterise patients undergoing adjuvant chemotherapy for resected NSCLC in a tertiary cancer center, and to describe disease-specific outcomes and treatment toxicity in a real-world patient population.

Methods

We performed a retrospective analysis of patients undergoing adjuvant chemotherapy for resected NSCLC in Cork University Hospital (CUH), Ireland and affiliated hospitals in the South/Southwest Hospital Group including University Hospital, Kerry and Mercy University Hospital, Cork over a seven-year period from February 2011 to October 2018 permitting satisfactory patient follow-up for cancer recurrence. CUH is a large, tertiary cancer center with all relevant thoracic oncology specialties on site including pulmonary medicine, cardiothoracic surgery, histocytopathology, medical oncology and radiation oncology.

Records from thoracic multidisciplinary team (MDT) meetings over the above period were reviewed. Our study population included patients with histologically confirmed NSCLC who underwent surgery with curative intent and who subsequently commenced adjuvant chemotherapy. We included patients with microscopic tumor involvement at resection margins (R1 resection) as well as patients who underwent post-operative radiotherapy, either concurrently with adjuvant chemotherapy or sequentially following completion of systemic treatment. All patients underwent appropriate pre-operative staging investigations in line with local guidelines, and relevant imaging and pathology records for all patients were reviewed at a regional thoracic MDT following surgical resection. Exclusion criteria included metastatic disease, a primary tumor not amenable to macroscopic complete resection, and any non-NSCLC histology.

From MDT records we extracted a list of patients who proceeded to adjuvant chemotherapy and data were collected for this population. Data describing patient demographics, disease staging, and tumor characteristics were collected from patient charts and electronic medical records. Data on treatment received and associated toxicities, as well as disease recurrence and survival were also collected. We separately recorded the proportion of patients undergoing thoracic surgery for NSCLC for which adjuvant chemotherapy may have been considered based on tumor staging. Approval for this retrospective study was granted by the Quality and Patient Safety Department of Cork University Hospital.

Information was assimilated using Microsoft Excel. Statistical analysis was carried out using R software. Descriptive statistics were applied to patient characteristics and treatment-related outcomes. Subgroup analysis of outcomes including chemotherapy toxicity and rates of hospitalisation were estimated using logistic regression. Estimates of OS and recurrence free survival (RFS) were generated using the Kaplan-Meier method. Patients with no recorded events of interest at the time of data collection were censored. Key covariates associated with relapse were analysed using Pearson's Chi-squared test, Wilcoxon rank sum test and Fisher's exact test. Predictors of relapse-free survival including chemotherapy backbone (carboplatin vs cisplatin), squamous versus non-squamous histology, age, nodal status, and stage were analysed using Cox proportional hazards model.

Results

Patient demographics & disease characteristics

Records were reviewed for 576 patients who underwent thoracic surgery for NSCLC in CUH from February 2011 to October 2018. We assessed the pathological stage of these patients to determine the proportion of patients which met criteria for consideration of adjuvant chemotherapy based on ESMO guidelines [6]. Of the 576 patients reviewed 255 (45%) met criteria for adjuvant treatment. The breakdown of these patients by stage grouping is described in Figure 1 . Adjuvant chemotherapy was commenced for 62 patients, comprising 11% of all patients, and 23% of patients in whom adjuvant chemotherapy may have been considered by pathological staging criteria.

The median age of patients undergoing adjuvant treatment was 69 years (range: 53–79), with an approximately equal gender distribution. Most patients (94%) had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, and 57 patients (92%) were current smokers at the time of diagnosis. No patients had undertaken neoadjuvant treatment in line with guideline-based standard of the care during the study period. Sixteen tumors (26%) demonstrated squamous cell histology, and forty-six (74%) were non-squamous. By tumor stage, 28 patients (45%) were IIB, 24 (39%) were IIIA, and 10 (16%) were IIIB. Forty-eight patients (77%) were node-positive, and surgical margins were positive for 10 patients (16%). The median tumor size was 4 cm (range: 0.6-9.6 cm). Molecular characterisation and assessment for oncogenic driver mutations was not routinely performed on our study population given the lack of EMA-approved targeted therapies in the adjuvant/non-metastatic setting during the study timeframe. Baseline characteristics are summarised in Table 1.

Adjuvant therapy

Almost all patients (98%) who received adjuvant treatment received doublet, platinum-based chemotherapy. Thirty-one patients (50%) received cisplatin as part of their treatment, and thirty (48%) received carboplatin. A platinum agent was combined with vinorelbine for 26 patients (42%). The median number of days from surgery to commencement of adjuvant chemotherapy was 55. Thirteen patients (21%) received adjuvant radiotherapy sequentially or concurrently with adjuvant chemotherapy. Most patients with R1 surgical resections received post-operative radiotherapy (8/10; 80%). Thirty-seven patients (60%) completed scheduled adjuvant chemotherapy without dose reduction or treatment delay. Seventeen patients (27%) required a dose reduction, and 11 patients (18%) discontinued treatment prior to completion of the planned number of cycles. Eighteen patients (29%) were admitted to hospital due to treatment-related toxicities and 24 patients (39%) developed Common Terminology Criteria for Adverse Events (CTCAE) grade 3 or 4 haematological toxicity during treatment. Grade 3 or 4 neutropenia was recorded in 22 patients (35%). No grade 5 treatment-related deaths were recorded. Details of adjuvant chemotherapy regimens used, and associated toxicities are summarised in Table 2.

Rates of grade 3-4 haematological toxicity were similar for patients receiving cisplatin and carboplatin as their chemotherapy backbone (42% vs 40%). Higher rates of hospital admission were recorded in patients receiving cisplatin (35% vs 20%) but this was not statistically significant. Higher rates of grade 3-4 haematological toxicity were noted in squamous versus non-squamous histology (56% vs 37%) though the difference did not reach statistical significance. Subgroup analysis comparing outcomes by treatment and histological subtype are described in Tables 3 & 4.

Relapse & survival

Median duration of follow-up for patients who underwent chemotherapy was 26 months (range: 7–107). Relapse was recorded in 35 patients (56%). Median RFS was 27 months (2.28 years). Nineteen deaths were recorded during the follow-up period and the median OS was not reached. Kaplan-Meier curves for RFS and OS are shown in Figures 2 & 3, respectively.

Similar rates of relapse were observed in patients receiving cisplatin-based chemotherapy (17/31; 55%) and carboplatin-based chemotherapy (17/30; 57%). More relapses occurred in patients with non-squamous histology than squamous histology (30 of 46 (65%) versus 5 of 16 (31%); p = 0.036) and in patients with more advanced disease stage (p = 0.029). Key co-variates by relapse status are described in Table 5. Disease stage was predictive of shorter RFS using cox proportional hazards model (stage IIIB vs stage IIB: HR = 3.55; 95% CI: 1.12 – 11.29; p = 0.032). Other variables analysed were not predictive of a statistically significant change in RFS. Comparison of relapse rates and RFS between subgroups is limited by the observational nature of our study and imbalances in baseline characteristics of subgroups may have influenced observed differences in disease-specific outcomes. Results of cox proportional hazards model for RFS are summarised in Table 6.

Discussion

This study demonstrates the clinically significant morbidity associated with the use of adjuvant chemotherapy in resected NSCLC. Approximately one in five patients required early cessation of their planned treatment, and 27% required a dose reduction. Moreover, 29% of patients were admitted to hospital due to treatment-associated toxicity, and 39% of patients experienced grade 3 or 4 haematological toxicity. Similar rates of toxicity were identified across treatment regimens. The median age of patients undergoing adjuvant treatment in this study (68) is higher than that of the populations enrolled in phase III trials assessing adjuvant chemotherapy in NSCLC, and high rates of carboplatin use in our patient cohort may be reflective of the co-morbidities and borderline performance status of many real-world patients following thoracic surgery. Of note, no significant differences in haematological toxicity or rates of hospitalisation were identified in patients receiving cisplatin versus carboplatin chemotherapy.

A 56% rate of disease relapse was noted in this study, which is higher than those reported in the chemotherapy arms of both the phase III ANITA trial [10] (46%), and the phase III data presented by Kenmotsu et al. (39%) [10,11]. Similarly, we report a median RFS of 27 months, compared with a median disease-free survival of 36 months in the ANITA trial, and a median RFS of 37 months reported by Kenmotsu et al. Poorer disease-specific outcomes may be attributable to the advanced age and associated co-morbidities of our cohort, as well as the significant proportion of our patients in our study with stage III disease (55% vs 41% in ANITA trial). Furthermore, 16% of patients included in our analysis had a positive (R1) resection margin on operative histology – a patient subgroup which was excluded from pivotal Phase III randomised control trials assessing adjuvant chemotherapy in NSCLC. Inclusion of this patient group is likely to have negatively impacted disease-specific outcomes including RFS but was intended to reflect our real-world experience of adjuvant treatment in a high-risk cohort. Relapse data reported in this limited retrospective study are observational in nature in should be interpreted with this limitation in mind.

Given the careful selection of patients enrolled in phase III trials of adjuvant chemotherapy in NSCLC, the generalisability of resultant outcomes to a real-world patient population must be considered. Studies addressing real-world patient outcomes and treatment safety in this setting are limited [12]. Harada et al. conducted a retrospective cohort study assessing the safety and efficacy of adjuvant chemotherapy for NSCLC in a real-world patient population [13]. This study demonstrated a significant improvement in overall survival and recurrence free survival in patients receiving adjuvant chemotherapy compared with observation only. However, concerning rates of toxicity were also reported, particularly for patients receiving cisplatin in combination with vinorelbine, 9% of whom died due to treatment toxicity. Grade 3 or 4 toxicity was reported in 89% of cases. The higher rates of clinically significant toxicity reported by Harada et al. compared with our data may reflect the prospective nature of the former study. Moreover, the differences may relate to more frequent use of cisplatin-doublet chemotherapy (75% vs 50%), or system-based differences in delivery of cytotoxic treatment and associated supportive care.

There remains an unmet clinical need for strategies to improve outcomes in early-stage NSCLC given the significant rates of disease recurrence post-surgical resection. Prospective studies have sought to establish the role of therapies guided by tumor-derived and circulating biomarkers in resectable NSCLC. The third-generation EGFR inhibitor Osimertinib has been granted approval in the adjuvant setting for resected NSCLC harbouring a sensitising EGFR mutation based on data from the ADAURA trial [14]. In this selected population, adjuvant EGFR-targeting therapy was associated with a median disease-free survival (DFS) of 65.8 months compared with 21.9 months in the placebo control arm. Early phase studies have also assessed the utility of circulating tumor DNA (ctDNA) as a surrogate biomarker for minimal residual disease. Chaudhuri et al. found detectable ctDNA post-surgical resection to be an independent prognostic factor for disease recurrence in early NSCLC [15]. Prospective studies assessing the utility of ctDNA to guide adjuvant and neoadjuvant therapy in NSCLC are ongoing [16].

Multiple phase III trials have also assessed the utility of immune-checkpoint inhibitors in the adjuvant and neoadjuvant settings for early NSCLC. The phase III IMpower010 trial demonstrated a disease-free survival benefit with the use of Atezolizumab compared with supportive care when delivered after standard of care adjuvant chemotherapy for resected NSCLC. A more pronounced benefit was observed in patients whose tumors demonstrated over-expression of programmed death-ligand 1 (PD-L1) [17]. The phase III KEYNOTE-091 trial compared adjuvant pembrolizumab with placebo for resected stage IB (tumors ≥4 cm) to IIIA NSCLC who had previously received adjuvant chemotherapy [18]. An interim analysis of the overall trial population reported a median DFS of 53.6 months with pembrolizumab versus 42 months in the control group. Again, a more marked benefit was noted in the subgroup of patients demonstrating PD-L1 overexpression. The phase III CheckMate 816 trial demonstrated improved event-free survival with the use of neoadjuvant nivolumab plus chemotherapy in patients with stage IB to IIIA resectable NSCLC compared with chemotherapy alone (31.6 vs 20.8 months) [19]. Pathological complete response – a potential early predictor of survival in resected NSCLC [20] - was achieved in 24% of patients receiving neoadjuvant nivolumab plus chemotherapy, compared with 2.2% of patients receiving neoadjuvant chemotherapy alone [20]. Longer term follow-up from a phase I/II study of neoadjuvant nivolumab in resectable NSCLC is suggestive of a sustained benefit compared with historical outcomes [21].

Conclusion

This study adds to the global literature on the use of adjuvant chemotherapy in a high risk, real-world setting, highlighting the clinically significant toxicity in this patient population. Significant rates of hospitalisation and grade 3 or 4 toxicity reflect the clinically meaningful toxicity and morbidity associated with adjuvant treatment in resected NSCLC. Allowing for the inherent limitations of this retrospective study, high rates of disease recurrence were observed. The optimal utilisation of evolving therapeutic strategies as part of the multidisciplinary management of early NSCLC remains unclear. As the landscape of NSCLC management evolves, integrated treatment pathways incorporating molecular characterisation, tumor-derived and circulating biomarkers, as well as pathological response assessment following neoadjuvant therapy may limit the role of cytotoxic chemotherapy for early NSCLC in the future.