Randomized phase II adjuvant trial to compare two treatment durations of icotinib (2 years versus 1 year) for stage II-IIIA EGFR-positive lung adenocarcinoma patients (ICOMPARE study)

Abstract

Background

Despite the prolonged median disease-free survival (DFS) by adjuvant targeted therapy in non-small-cell lung cancer patients with epidermal growth factor receptor (EGFR) mutations, the relationship between the treatment duration and the survival benefits in patients remains unknown.

Patients and methods

In this multicenter, randomized, open-label, phase II trial, eligible patients aged 18-75 years with EGFR-mutant, stage II-IIIA lung adenocarcinoma and who had not received adjuvant chemotherapy after complete tumor resection were enrolled from eight centers in China. Patients were randomly assigned (1 : 1) to receive either 1-year or 2-year icotinib (125 mg thrice daily). The primary endpoint was DFS assessed by investigator. The secondary endpoints were overall survival (OS) and safety. This study was registered at ClinicalTrials.gov (NCT01929200).

Results

Between September 2013 and October 2018, 109 patients were enrolled (1-year group, n = 55; 2-year group, n = 54). Median DFS was 48.9 months [95% confidence interval (CI) 33.1-70.1 months] in the 2-year group and 32.9 months (95% CI 26.6-44.8 months) in the 1-year group [hazard ratio (HR) 0.51; 95% CI 0.28-0.94; P = 0.0290]. Median OS for patients was 75.8 months [95% CI 64.4 months-not evaluable (NE)] in the 2-year group and NE (95% CI 66.3 months-NE) in the 1-year group (HR 0.34; 95% CI 0.13-0.95; P = 0.0317). Treatment-related adverse events (TRAEs) were observed in 41 of 55 (75%) patients in the 1-year group and in 36 of 54 (67%) patients in the 2-year group. Grade 3-4 TRAEs occurred in 4 of 55 (7%) patients in the 1-year group and in 3 of 54 (6%) patients in the 2-year group. No treatment-related deaths or interstitial lung disease was reported.

Conclusions

Two-year adjuvant icotinib was shown to significantly improve DFS and provide an OS benefit in EGFR-mutant, stage II-IIIA lung adenocarcinoma patients compared with 1-year treatment in this exploratory phase II study.

Highlights

• •Two-year adjuvant icotinib had a significantly prolonged DFS compared with 1-year treatment in stage II-IIIA NSCLC.
• •PFS benefit translated into a significant OS benefit, possibly due to the longer duration of adjuvant targeted therapy.
• •Safety outcomes were similar in both groups.

Introduction

Only 20%-25% of non-small-cell lung cancer (NSCLC) patients are suitable for potentially curative resection, which remains its current standard of care.¹ Before progression to a more advanced stage, patients with stage I-IIIA NSCLC are potentially curable by complete surgical resection.² While high recurrence rates prevent patients from achieving long-term survival, the 5-year survival rates of patients with completely resected NSCLC are ∼65% for those in stage IIA and 41% for those in stage IIIA.³ Currently, as the only proven drug regimen shown to improve survival, adjuvant chemotherapy after surgery has improved the survival of patients with early-stage NSCLC, yet chemotherapy usually causes more grade 3 or worse adverse effects (AEs). Statistically, the 5-year survival rates increased by 5% with an 11% reduction in the risk of death for post-operative chemotherapy, and there is still much room for therapeutic improvement.^4,5

Epidermal growth factor receptor (EGFR) mutations, such as exon 19 deletions (ex19del) and exon 21 Leu858Arg (L858R) mutations, are the most prevalent oncogenic drivers in NSCLC, accounting for ∼50% in the Asian population.6, 7, 8, 9 As with post-operative adjuvant chemotherapy, EGFR tyrosine kinase inhibitors (EGFR-TKIs) have also emerged as recommended first-line treatments for EGFR mutation-positive advanced NSCLC.10, 11, 12, 13, 14, 15 Of some prior randomized studies, post-operative adjuvant targeted therapy was documented to be associated with survival benefit in patients with NSCLC. Early attempts, such as the prospective, single-arm, phase II study SELECT, reported that patients with EGFR mutations may benefit from adjuvant EGFR-TKI therapy, showing that the disease-free survival (DFS) of NSCLC patients who received post-operative adjuvant erlotinib therapy was up to 89% in 2 years.¹⁶ These findings provide a basis for future clinical studies. Afterward, two studies (ADJUVANT and EVAN) confirmed that 2-year adjuvant targeted therapy was able to bring superior DFS and safety to EGFR-mutant patients with stage IIIA (N2) disease compared to adjuvant chemotherapy.^17,18 More recently, a phase III randomized controlled trial (ADAURA) identified that once-daily osimertinib significantly improved the median DFS compared with placebo.¹⁹ Another phase III clinical randomized study (EVIDENCE) of treatment for EGFR-mutant stage II-IIIA NSCLC showed that patients treated with adjuvant icotinib had a longer DFS and higher tolerance compared with patients treated with adjuvant chemotherapy.²⁰ In addition to the abovementioned studies, a recent multicenter study (ALCHEMIST) at the National Cancer Institute (NCI) is ongoing in patients with stage IB-IIIA NSCLC to explore whether an overall survival (OS) benefit can be observed in the adjuvant target treatment setting.²¹ To date, post-operative adjuvant targeted therapy in patients with EGFR mutations has become a hot topic in NSCLC.

However, few studies have focused on the duration of adjuvant targeted therapy, and its effect on survival is still unclear. The majority of reported adjuvant trials have a 2-year cyclical treatment except that ADAURA reported improvement in DFS rate with 3 years of adjuvant osimertinib therapy. Remarkably, 6 months of adjuvant gefitinib also yielded prolonged DFS for those with N2-positive disease.²² Concurrently, a previous study demonstrated the therapeutic equivalence between 1 year and 2 years of trastuzumab adjuvant therapy in breast cancer.²³ Therefore, exploring the optimal duration of adjuvant EGFR-TKIs for NSCLC is warranted. In line with the study design of the EVIDENCE study, which was ‘chemo free’ in the adjuvant icotinib arm,²⁰ we report the results of this ICOMPARE study with adjuvant icotinib monotherapy, to compare the efficacy and safety of 1-year adjuvant icotinib with 2-year adjuvant icotinib for patients with stage II-IIIA NSCLC after complete resection. Our results should be considered exploratory as the problem has not been studied.

Patients and methods

The trial was conducted in accordance with the amended Declaration of Helsinki, and the protocol was approved by the institutional review boards of each participating center. All patients provided written informed consent.

Study design and participants

This prospective, multicenter, open-label, phase II randomized trial was conducted in eight hospitals in China. Eligible patients were aged 18-75 years with confirmed stage II-IIIA lung adenocarcinoma [the seventh edition of the American Joint Committee on Cancer (AJCC) TNM (tumor–node–metastasis) staging system] who had undergone complete resection (defined as free resection margins with systematic nodal dissection showing that the highest mediastinal node was negative, with no extracapsular nodal extension of the tumor).^24,25 Only patients with confirmed EGFR-activating mutations (ex19del, or L858R) were eligible. Additional inclusion criteria were Eastern Cooperative Oncology Group performance status ≤2 and adequate hematological tests (absolute neutrophil count ≥1500 per cubic millimeter, platelet count ≥100 000 per cubic millimeter, and hemoglobin level ≥9 g per deciliter) and liver (serum total bilirubin level ≤1.5 times the upper limit of the normal range) and kidney function (creatinine clearance ≥60 ml per minute). We excluded patients if they had previous exposure to erlotinib, gefitinib, cetuximab, trastuzumab, chemotherapy, or systemic antitumor therapy. Other key exclusion criteria were previous radiotherapy, myocardial infarction within 6 months, angina, heart disease, and history of interstitial pneumonia. Pregnant or lactating women were also ineligible.

Procedures

Eligible patients were randomized to receive icotinib at 125 mg thrice daily for either 1 year or 2 years, which was initiated within 4-6 weeks after surgery, and no other previous anticancer therapy was permitted. Treatment was stopped temporally if the patient developed grade 3 or higher AEs until the AE returned to grade 1-2. If grade 3 or higher toxic effects re-appeared after the resumption of drug therapy, the patient was withdrawn from the study. Icotinib dose remained unchanged during the study period unless it was discontinued. Before surgery, whole-body positron emission tomography/computed tomography (PET/CT), chest CT, and brain magnetic resonance imaging (MRI) were carried out to rigorously exclude distant metastases. Patients could also receive abdominal and supraclavicular lymph node ultrasound examination, bone scans, and brain MRI as another staging method. All patients adhered to the same schedule of follow-up visits, including the recording of tumor assessments, symptoms, and AEs (according to the NCI Common Toxicity Criteria version 4.0), which were carried out in the first month after the start of treatment and then every 3 months for the first 2 years. After that, these assessments were scheduled to be carried out every 6 months for the next 3 years. At each follow-up visit, chest CT and abdominal and supraclavicular lymph node ultrasonography were carried out. Furthermore, brain MRI and bone scans were carried out every 12 months following local standard clinical routine or based on patient’s symptoms.

Randomization and masking

The patients were randomly assigned (1 : 1) to receive icotinib for either 1 year or 2 years until tumor recurrence or unacceptable toxic effects occurred. Pocock & simon minimization method was used for randomization with the randomization factors of EGFR mutation type (19del versus L858R) and clinical stage (IIA versus IIB versus IIIA).²⁶ None of the researchers, patients, or analysis teams were blinded to the treatment allocation.

Outcomes

The primary endpoint was investigator-assessed DFS, which was defined as the time from randomization to documented disease relapse or death, whichever occurred first. The secondary endpoints were investigator-assessed OS (time from randomization to death from any cause) and safety. DFS after the end of treatment (EOT), which was defined as the interval between the date that the patients completed prescribed duration of treatment without disease recurrence during therapy and the date recurrence occurred, and the efficacy, including objective response rates (ORRs; the proportion of patients with complete response or partial response), of icotinib retreatment after recurrence were also analyzed as exploratory endpoints.

Statistical analysis

DFS was analyzed in the intention-to-treat (ITT) population (all patients who underwent randomization) and per-protocol (PP) population (excluded ITT patients with major protocol violations and insufficient drug exposure time). Safety dataset was summarized for the patients who received at least one dose of icotinib.

The historical benchmark in this study design used a median DFS of 26 months [95% confidence interval (CI) 23-29 months] from a post-operative adjuvant chemotherapy phase II trial.²⁷ Based on the lower confidence limit of the 95% CI, the median DFS with icotinib for the 1-year group was assumed to be 23 months, with a 49-month median DFS for the 2-year group. The study provided 80% power to demonstrate a DFS hazard ratio (HR) of 0.47 at a two-sided 5% significance level with 56 DFS events. A total of 120 patients were required to be randomized with 2 years of enrollment, and a dropout rate of ∼15%.

We used Kaplan–Meier curves to describe DFS and OS and carried out treatment comparisons with the log-rank test stratified by disease stage and mutational status. We estimated HRs and 95% CIs using a stratified Cox proportional hazards regression model. Proportional hazards assumptions were tested using graphical approach and the Schoenfeld residuals method. Safety variables were analyzed descriptively in terms of the number (percentage) of patients with AEs. Multiple comparisons were not taken into account in analyses of secondary endpoints.

The data cut-off date was 27 September 2020. All statistical analyses were carried out using SAS version 9.4 (SAS Institute Inc., Cary, NC). This study is registered with ClinicalTrials.gov, number NCT01929200.

Results

Patient demographics and characteristics

Between September 2013 and October 2018, 245 patients were screened in eight hospitals across China, of whom 109 were randomly assigned to receive icotinib for 1 year (n = 55) or for 2 years (n = 54); all 109 patients were included in the ITT analysis set. The final analysis in the present study consisted of 109 patients. However, the included patients could still provide a power of 80%. The two groups were well balanced in terms of baseline characteristics (Table 1). Approximately 67% of the patients were female, and the median age was 59 years (range 32-76 years). The ex19del mutation was found in 60 (55%) patients, and the 21L858R mutation was found in 49 (45%) patients. Fifty-seven (52%) patients had stage II disease, and 52 (48%) had stage IIIA disease.

Table 1.

Baseline characteristics

	1-year group (n = 55)	2-year group (n = 54)	P value
Sex
Male	17 (31)	19 (35)	0.6868
Female	38 (69)	35 (65)
Age, years
Median (range)	59 (32-76)	58.5 (41-75)
Smoking history
Yes	12 (22)	11 (20)	1.0000
No	43 (78)	43 (80)
ECOG performance status
0	22 (40)	17 (31)	0.5634
1	32 (58)	35 (65)
2	1 (2)	2 (4)
Lung cancer resection type
Lobectomy	54 (98)	51 (94)	0.3634
Bilobectomy	1 (2)	3 (6)
Diameter of the tumor
Median (range)	2.8 (1-9)	2.5 (1-8.5)
Visceral pleural involvement
Yes	29 (53)	22 (41)	0.1146
No	24 (44)	32 (59)
NK	2 (4)	0 (0)
Vascular invasion
Yes	16 (29)	20 (37)	0.4198
No	39 (71)	34 (63)
Lymph node metastasis
Yes	45 (82)	43 (80)	0.8121
No	10 (18)	11 (20)
Mediastinal lymph node metastasis
Yes	24 (44)	25 (46)	0.8483
No	31 (56)	29 (54)
Tumor stage
IIA	18 (33)	18 (33)	1.0000
IIB	11 (20)	10 (19)
IIIA	26 (47)	26 (48)
EGFR mutation
19del	31 (56)	29 (54)	0.8483
21L858R	24 (44)	25 (46)

Data are presented as n (%), unless otherwise specified.

ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; 19del, deletion in exon 19; L858R, Leu858Arg point mutation in exon 21.

Of the 55 patients who completed the initial scan for staging in the 1-year group, 37 patients (67%) underwent whole-body PET/CT and brain MRI. Of the 54 patients in the 2-year group, 34 patients (63%) underwent whole-body PET/CT and brain MRI. Overall, the two groups were reasonably balanced across staging approaches.

In total, 88 (81%) of 109 patients completed the PP treatment: 47 (85%) of 55 patients in the 1-year group and 41 (76%) of 54 patients in the 2-year group (Figure 1). Four patients discontinued icotinib in the 1-year group, with three patients due to AEs (one patient with grade 3 dyspnea, one patient with grade 3 rash, and one patient with grade 2 rash/diarrhea) and one patient withdrew for personal reasons. Three patients discontinued icotinib in the 2-year group, with two patients due to AEs (one patient with grade 3 rash and one patient with grade 3 nausea and grade 2 fatigue) and one patient withdrew for personal reasons. And 14 patients experienced relapse during the treatment period (4 in the 1-year group and 10 in the 2-year group). The median duration of the 1-year group was 12.0 months [interquartile range (IQR) 12.0-12.0 months] and the mean was 11.8 months (95% CI 10.5-13.1 months); the median duration of the 2-year group was 24.0 months (IQR 24.0-24.0 months) and the mean was 21.6 months (95% CI 20.0-23.2 months).

Figure 1.

Trial diagram.

EGFR, epidermal growth factor receptor.

Efficacy results

At the cut-off date, with a median follow-up of 44.1 months (95% CI 37.1-49.9 months), DFS events occurred in 56 (51%) of 109 patients [31 (56%) in the 1-year group and 25 (46%) in the 2-year group], including 10 local recurrences [5 (16%) patients in the 1-year group and 5 (20%) patients in the 2-year group] and 46 distant metastases [26 (84%) patients in the 1-year group and 20 (80%) patients in the 2-year group; Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2023.101565]. These metastases occurred in the brain [11 (35%) patients versus 8 (32%) patients], lymph nodes [10 (32%) patients versus 6 (24%) patients], lung [8 (26%) patients versus 8 (32%) patients], bone [2 (7%) patients versus 7 (28%) patients], and liver [3 (10%) versus none].

In the ITT population, the median DFS was 48.9 months (95% CI 33.1-70.1 months) in the 2-year group and 32.9 months (95% CI 26.6-44.8 months) in the 1-year group (HR 0.51; 95% CI 0.28-0.94; P = 0.0290; Figure 2A). There was no significant difference in DFS between the 2-year and 1-year groups in patients with stage N0 (HR 0.28; 95% CI 0.03-2.73; P = 0.2734) or N1 disease (HR 1.40; 95% CI 0.65-2.99; P = 0.3882), while the median DFS in patients with stage N2 disease was significantly improved by 2-year icotinib treatment (64.8 versus 25.9 months; HR 0.29; 95% CI 0.12-0.69; P = 0.0053). In the PP population, the median DFS was 50.9 months [38.6 months-not evaluable (NE)] in the 2-year group and 34.7 months (95% CI 27.9-50.9 months) in the 1-year group (HR 0.24; 95% CI 0.10-0.57; P = 0.0006; Supplementary Figure S1A, available at https://doi.org/10.1016/j.esmoop.2023.101565).

Figure 2.

Analysis of survival for the intention-to-treat population.

(A) Kaplan–Meier plots of DFS. (B) Kaplan–Meier plots of DFS after the EOT. (C) Kaplan–Meier plots of OS. CI, confidence interval; DFS, disease-free survival; EOT, end of treatment; HR, hazard ratio; NE, not evaluable; OS, overall survival.

The effect of both treatment groups on DFS was consistent across the majority of prespecified subgroups; however, 2-year treatment was favored over 1-year treatment among N stage (N2 versus N0/N1), overall stage (IIIA versus IIA/B), and vascular invasion (yes versus no) subgroups (Figure 3A).

Figure 3.

Forest plots for the HRs (triangles) and 95% CIs (horizontal line) of disease-free survival and overall survival in subgroups of patients for the intention-to-treat population.

(A) Forest plots of disease-free survival. (B) Forest plots of overall survival.

CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; HR, hazard ratio.

A total of 88 patients (47 in the 1-year group and 41 in the 2-year group) who completed PP medication were included in the analysis for the DFS after EOT, in which 39 experienced relapse (25 in the 1-year group and 14 in the 2-year group). After the EOT, the median DFS was 26.8 months (95% CI 14.5-46.1 months) in the 2-year group and 22.6 months (95% CI 14.9-38.9 months) in the 1-year group (HR 0.71; 95% CI 0.33-1.53; P = 0.3832; Figure 2B).

Overall, disease relapse was observed in 56 patients, and 14 patients experienced relapse during the treatment period (4 in the 1-year group and 10 in the 2-year group). Among the rest of the patients, 32 patients were retreated with icotinib (20 in the 1-year group and 12 in the 2-year group). The median interval duration of re-administration after drug withdrawal was 13.9 months (range 4.3-50.0 months) in the 1-year group and 14.5 months (range 3.7-46.6 months) in the 2-year group, respectively. Among the 30 assessable patients, ORRs in both groups were 66.7%, with 12 of 18 patients in the 1-year group and 8 of 12 patients in the 2-year group achieving a response. The overall median progression-free survival (PFS) was 18.0 months (95% CI 8.1 months-NE), and no difference in PFS was observed between the two groups (P = 0.507).

In the ITT population, 13 of 55 patients (24%) in the 1-year group and 7 of 54 patients (13%) in the 2-year group had OS events. The median OS was NE (95% CI 66.3 months-NE) in the 1-year group and 75.8 months (95% CI 64.4 months-NE) in the 2-year group. A 66% reduction in the risk of death with 2 years of icotinib was observed in the ITT population (HR 0.34; 95% CI 0.13-0.95; P = 0.0317; Figure 2C). Additionally, 2-year icotinib was also associated with a significantly improved 5-year OS rate [2-year versus 1-year: 0.88 (95% CI 0.70-0.96) versus 0.72 (95% CI 0.55-0.84); P = 0.032]. In the PP population, the median OS was 75.8 months (95% CI 61.6 months-NE) in the 2-year group and NE (95% CI 66.3 months-NE) in the 1-year group (HR 0.16; 95% CI 0.03-0.85; P = 0.0183; Supplementary Figure S1B, available at https://doi.org/10.1016/j.esmoop.2023.101565).

The effect of both treatment groups on OS was consistent across the majority of prespecified subgroups; however, 2-year treatment was favored over 1-year treatment among N stage (N2 versus N0/N1) and overall stage (IIIA versus IIA/B) subgroups (Figure 3B).

Safety results

Overall, 109 patients were included in the safety set (55 in the 1-year group and 54 in the 2-year group). Treatment-related adverse events (TRAEs) were reported in 41 patients (75%) in the 1-year group and in 36 patients (67%) in the 2-year group. The majority of TRAEs were grade 1-2 in severity. The most common TRAEs were rash (44% in the 1-year group and 41% in the 2-year group), followed by diarrhea (33% in the 1-year group and 33% in the 2-year group). The most frequent grade 3-4 TRAEs were liver dysfunction [2 (4%) in the 1-year group and 0 in the 2-year group]. The common TRAEs are listed in Table 2. No deaths due to TRAEs were reported.

Table 2.

TRAEs

TRAEs	1-year group (n = 55)		2-year group (n = 54)
	Any grade	Grade 3 or 4	Any grade	Grade 3 or 4
Rash/pruritus	24 (44)	1 (2)	22 (41)	1 (2)
Diarrhea	18 (33)	0	18 (33)	0
Dry skin	0	0	2 (4)	0
Alopecia	14 (25)	0	8 (15)	0
Oral ulcer	6 (11)	0	12 (22)	0
Nausea	2 (4)	0	4 (7)	1 (2)
Liver dysfunction	11 (20)	2 (4)	7 (13)	0
Neutropenia/leucopenia	2 (4)	0	2 (4)	0
Dry mouth	0	0	1 (2)	0
Depression	0	0	1 (2)	0
Dyspnea	1 (2)	1 (2)	0	0
Onychogryphosis	1 (2)	0	0	0
Hoarseness	1 (2)	0	0	0
Myocardial infarction	0	0	1 (2)	1 (2)
Cough	1 (2)	0	0	0
Fatigue	0	0	1 (2)	0

Data are presented as n (%), unless otherwise specified.

TRAEs, treatment-related adverse events.

A total of seven grade 3 or higher adverse events were observed in 109 patients; of these, 3 patients refused further treatment and discontinued and 4 patients continued to receive treatment after temporal interruption, among which 1 patient finally discontinued treatment due to re-appeared grade 3 AE. In addition, one patient continued medication after 2 weeks’ interruption due to brain surgery. The dosages of icotinib remained unchanged in both groups during the study.

Discussion

To the best of our knowledge, this is the first large randomized trial to evaluate the duration of adjuvant targeted therapy in EGFR-mutant NSCLC patients after resection. In previous studies, adjuvant icotinib has been shown to be effective in EGFR-mutant NSCLC patients, with an acceptable tolerability profile.^20,28 Here, clinical outcomes indicated that 2-year icotinib adjuvant therapy significantly prolonged DFS compared with 1-year adjuvant therapy without increasing toxicity in EGFR-mutant NSCLC patients without adjuvant chemotherapy. In addition, favorable safety profile and notable compliance (76% in the 2-year group and 85% in the 1-year group) were also seen in the current study.

Several studies have evaluated the role of EGFR-TKIs in the adjuvant setting. The phase III study ADJUVANT initially reported that adjuvant treatment with 2-year gefitinib prolonged DFS compared with adjuvant chemotherapy in patients with EGFR-mutant NSCLC.¹⁷ Another phase II study, EVAN, also revealed that 2-year erlotinib was associated with a longer DFS than chemotherapy in the adjuvant setting.¹⁸ A similar study EVIDENCE further confirmed the role of icotinib as adjuvant treatment, which showed a significantly longer DFS than chemotherapy in stage II-IIIA NSCLC patients with EGFR mutations.²⁰ In the current study, the median DFS of the 2-year group based on the ITT population was 48.9 months (95% CI 33.1-70.1 months), which was consistent with that of the EVIDENCE study (47.0 months) but longer than that of the EVAN study (42.4 months). Heterogeneity among patients should be noted in cross-trial comparisons. The EVIDENCE study consisted of patients with stage II-IIIA disease, whereas the EVAN study only recruited patients with stage IIIA disease, and most patients (approximately 97%) had stage N2 disease. In addition, as shown in the ADJUVANT study,¹⁷ the subset of patients with N2 disease seemed to benefit more from adjuvant gefitinib, which might explain the numerically longer DFS compared with chemotherapy in the EVAN study (42.4 versus 21.0 months). In our study, we observed similar results in terms of the TKI duration: patients with N2 disease had even more DFS benefits from 2-year adjuvant icotinib than those in the 1-year group (64.8 versus 25.9 months; HR 0.29; 95% CI 0.12-0.69; P = 0.0053). Furthermore, DFS benefit was clear in patients with stage IIIA and vascular invasion, suggesting that longer duration of post-operative EGFR-TKIs was required for the EGFR-mutant patients with locally advanced NSCLC. More recently, the ADAURA study, which evaluated 3-year treatment with osimertinib (a third-generation TKI) in stage IB-IIIA lung cancer, showed superior DFS to placebo. On balance, although some differences exist in study design between ICOMPARE (current study) and ADAURA (e.g. adjuvant chemotherapy was allowed but not mandated in ADAURA, while it was not allowed in ICOMPARE), we noted that an even longer duration (>2 years) of adjuvant TKI therapy might result in more survival benefits, especially in patients with locally advanced disease (N2), and this strategy should be explored in further studies.

Despite the immature OS data, 2-year adjuvant icotinib was associated with longer OS (HR 0.34; 95% CI 0.13-0.95; P = 0.0317) and a significantly higher 5-year OS rate (88% versus 72%; P = 0.032) than 1-year treatment, suggesting that the DFS benefits might potentially transfer into OS benefits. Inconsistent OS benefits following adjuvant EGFR-TKIs have been suggested in some studies. In light of the updated ADJUVANT/CTONG1104 survival analysis after 6-year follow-up, the median OS was 75.5 and 62.8 months with gefitinib and adjuvant chemotherapy, respectively (HR 0.92; 95% CI 0.62-1.36; P = 0.674). Notably, although there was no significant difference between treatments, a median of 75.5 months was still the best survival for patients with completely resected NSCLC (pathological N1/N2), suggesting that patients have received benefit from post-operative adjuvant targeted therapy.²⁹ Another EVAN trial has shown that the DFS benefit (erlotinib versus chemotherapy group: 42.4 versus 21.0 months; HR 0.268; 95% CI 0.136-0.531; P = 0.0003) of post-operative adjuvant targeted therapy eventually translates into OS benefit (84.2 versus 61.1 months; HR 0.318; 95% CI 0.151-0.670).³⁰ This observation suggests that stage IIIA patients harboring EGFR-sensitive mutations who undergo complete resection of tumor could directly benefit from post-operative adjuvant targeted therapy compared with adjuvant chemotherapy. These discrepancies may be due to different population and subsequent treatment after disease recurrence; more patients in the chemotherapy group received subsequent therapy with EGFR-TKIs after relapse in the CTONG1104 trial. In addition, our study also suggested that the survival advantage from adjuvant EGFR-TKIs is further enhanced by longer duration medication, which may be validated by the mature OS analysis in the ADAURA study comparing the 3-year treatment with adjuvant osimertinib versus placebo in patients with NSCLC harboring EGFR mutation.¹⁹

In our study, 14 of 56 patients (4 in the 1-year group and 10 in the 2-year group) relapsed during icotinib treatment. The recurrence rate (25%) appeared to be higher than that in the SELECT study (10.0%).¹⁶ We noted that 45% of patients (45/100) with stage I disease were enrolled in the SELECT study, and most of the patients had received adjuvant chemotherapy before erlotinib, which could explain this difference in the recurrence between the two studies. Short-term recurrence (<6 months) might be associated with primary TKI resistance. It should be noted that the median DFS after EOT was 26.8 months in the 2-year group, which was longer than that of the 1-year group (22.6 months) and consistent with that of erlotinib (25.4 months) in the SELECT study.¹⁶ Interestingly, the duration of adjuvant icotinib therapy in the 2-year group was 12 months longer than that in the 1-year group, whereas the median DFS was prolonged by 16.0 months, and 2-year treatment with adjuvant icotinib prolonged exactly the median DFS after EOT by almost 4 months compared with 1-year treatment. These findings indicated that prolonged duration of adjuvant TKI therapy could bring DFS benefits by improving DFS after EOT. It is possible that long-term EGFR-TKI therapy could further restrain the growth and gene cloning of tumor cells, hence delaying the recurrence of the disease.³¹ Recurrence sites in these 56 patients were predominantly distant. Compared to the 1-year group, patients in the 2-year group had higher incidences of lung (32% versus 26%), bone (28% versus 7%), and pleural (8% versus 3%) metastasis, while the incidences of brain (32% versus 35%), lymph node (24% versus 32%), and liver (0% versus 10%) metastasis were lower. We speculate that different cell phenotypes or gene expressions may exist in different sites, which may lead to better or worse drug efficacy for certain sites. In addition, differences in recurrence rates may also be related to individual differences among patients. Furthermore, the drug concentration in in vivo tissues may differ due to factors including drug metabolism, distribution, and excretion. Thus, more in-depth research is needed to explore the underlying mechanisms. The brain was the most common relapse site in the icotinib group [11 (7%) of 151 patients] from the EVIDENCE study.²⁰ In this study, the brain was also the most common site of recurrence, and patients in the 2-year group had a lower incidence of brain metastasis compared to those in the 1-year group.

We also investigated the role of icotinib retreatment in patients after recurrence. In this cohort of patients, 32 patients were retreated with icotinib. The ORR was 66.7%, which was similar to that in advanced disease with first-line EGFR-TKI treatment (58.0%-71.2%).^13,14,32,33 However, the median PFS seemed to be longer in this study (18.0 versus 9.7-13.1 months). This may be explained by the finding that compared with patients with advanced disease, patients with recurrence had a high proportion of single-site lesions, a lower tumor burden, and less cell heterogeneity, resulting in better disease control by EGFR-TKIs. This finding suggests that it may be an appropriate strategy to use TKIs again in recurrent patients after the end of adjuvant targeted therapy. A low frequency of dose modifications and discontinuations of icotinib and no new safety signals were reported. Furthermore, 7 of 109 (6%) patients experienced grade 3-4 TRAEs, and no interstitial lung disease was observed. Generally, these results demonstrated that adjuvant icotinib was well tolerated, suggesting that it is possible to prolong the duration of adjuvant targeted therapy in the future.

Several limitations should be noted. Firstly, patients did not receive adjuvant chemotherapy before adjuvant icotinib treatment. To date, platinum-based adjuvant chemotherapy is still the standard treatment for stage II-IIIA NSCLC after surgical resection.⁵ However, the main purpose of ICOMPARE is to explore the optimal duration of adjuvant targeted therapy among EGFR-mutant patients, and the results of this study have already provided key evidence for this issue by comparing different TKI exposure times. Another study designed to evaluate the efficacy of adjuvant icotinib following adjuvant chemotherapy is now ongoing (NCT02125240). Our study was also limited by the relatively small sample size, although the study was sufficiently powered for the primary endpoint, and the less potential for generalizability or extrapolation of our results to non-Asian populations with lung adenocarcinoma. These limitations make our study more concept generating than practice defining. However, our study does provide additional evidence for the design of future confirmatory trials, and fulfill the knowledge gap in the lack of robust results from randomized trials.

In conclusion, this is the first large exploratory randomized trial to compare the duration of post-operative EGFR-TKIs in patients with stage II-IIIA EGFR-mutant lung adenocarcinoma, demonstrating that a longer treatment duration of 2 years showed improved survival benefits compared with 1 year. Although the optimum duration of adjuvant EGFR-TKIs remains unknown, our data revealed that it should be at least 2 years or even more. More studies are warranted to continue the assessment of longer treatment durations to achieve further survival improvements.