Three-year adjuvant treatment of icotinib in stage II-IIIA EGFR-mutated lung adenocarcinoma (ICWIP): a randomized, double-blind, placebo-controlled phase 3 trial

Abstract

This multi-center, double-blind, placebo-controlled, randomized phase 3 trial (ICWIP) assessed the efficacy and safety of 3-year icotinib adjuvant treatment versus placebo in patients with completely resected stage II-IIIA epidermal growth factor receptor (EGFR) mutated lung adenocarcinoma who had completed four cycles of adjuvant chemotherapy (NCT02125240). Eligible patients were randomly assigned 1:1 to receive either icotinib (125 mg, three times daily) or placebo for 3 years. The primary endpoint was independent review committee (IRC)-assessed disease-free survival (DFS). Between June 3, 2014, and November 15, 2018, 66 patients received icotinib, and 69 patients received placebo. At the data cutoff date of October 31, 2024, the median follow-up for DFS was 78.3 months (interquartile range [IQR] 71.3–95.6) for the icotinib group and 69.9 months (IQR 35.2–96.4) for the placebo group. IRC-assessed median DFS was 48.2 months (95% confidence interval [CI] 33.0–63.4) in the icotinib group and 20.0 months (95% CI 14.7–32.9) in the placebo group (hazard ratio [HR] 0.54, 95% CI 0.34–0.87; p = 0.011). The 3-year DFS rates were 58.5% (95% CI 39.9–73.1%) for the icotinib group and 36.1% (95% CI 19.9–52.6%) for the placebo group, and the 5-year DFS rates were 39.4% (95% CI 22.2–56.2%) for the icotinib group and 25.8% (95% CI 8.7–47.1%) for the placebo group. The median overall survival (OS) was not reached in either group (HR 0.84, 95% CI 0.39–1.81; p = 0.65). The estimated 5-year OS rates were 88.6% (95% CI 57.9–97.5%) for the icotinib group and 83.4% (95% CI 57.7–94.2%) for the placebo group. Treatment-related adverse events occurred in 48.5% (32/66) and 23.2% (16/69) of the patients in the icotinib group and the placebo group, respectively. Three-year adjuvant treatment with icotinib significantly improved DFS compared to placebo in patients with completely resected stage II-IIIA EGFR-mutated lung adenocarcinoma who had completed four cycles of adjuvant chemotherapy, with a manageable safety profile.

Introduction

Lung cancer remained the leading cause of cancer mortality, accounting for an estimated 2.6 million newly diagnosed cases and over 1.9 million cases of death globally.¹ The morbidity and mortality of lung cancer in China ranked the first in the global cancer epidemic.² Non-small cell lung cancer (NSCLC) is the most common pathological subtype, accounting for 85% of all lung cancers.³ Although surgical resection remains the cornerstone of treatment for early-stage NSCLC, most patients still relapse and metastasis with an unsatisfactory 5-year overall survival (OS) rate of 36–60%.⁴ Over the past decades, there has been a paradigm shift founded on large clinical trials; that is, platinum-based adjuvant chemotherapy constitutes the standard-of-care for patients with completely resected stage II–IIIA NSCLC.⁵ However, the 5-year disease-free survival (DFS) rate of 39.4% for patients with completely resected NSCLC after receiving cisplatin-based adjuvant chemotherapy.⁶ The 5-year OS rate for patients with stage I-IIIA resected NSCLC who received cisplatin-based adjuvant chemotherapy increased 5% compared to those who did not receive cisplatin-based adjuvant chemotherapy.⁷ Attempts to improve outcomes by combining chemotherapy with bevacizumab have failed to demonstrate OS benefits in this setting.⁸ Thus, any adjuvant therapy likely to provide efficacy and safety benefit would be a valuable asset.

Epidermal growth factor receptor (EGFR), a member of receptor tyrosine kinases, is involved in the pathogenesis and progression of different carcinoma types.⁹ It is worth noting that the EGFR mutation causes a constant signal, leading to tumor progression.⁹ Unfortunately, approximately 10–30% of patients with NSCLC carried EGFR mutation, particularly as high as 50% in East Asia.¹⁰ Reassuringly, several studies have demonstrated that first-generation EGFR tyrosine kinase inhibitors (TKIs) prolonged DFS in resected EGFR-mutated NSCLC. The median DFS was 50.5 months with erlotinib and 48.2 months with placebo in the RADIANT study (hazard ratio [HR] 0.90, 95% confidence interval [CI] 0.74–1.10; p = 0.324), was 28.7 months with gefitinib and 18.0 months with chemotherapy in the ADJUVANT/CTONG1104 study (HR 0.60, 95% CI 0.42–0.87; p = 0.0054), was 35.9 months with gefitinib and 25.1 months with chemotherapy in the IMPACT study (HR 0.92, 95% CI 0.67–1.28; p = 0.92), was 42.4 months with erlotinib and 21.0 months with chemotherapy in the EVAN study (HR 0.268, 95% CI 0.136–0.531; p < 0.0001), was 47.0 months with icotinib and 22.1 months with chemotherapy in the EVIDENCE study (HR 0.36, 95% CI 0.24–0.55; p < 0.0001), was 48.9 months with 2-year icotinib and 32.9 months with 1-year icotinib in the ICOMPARE study (HR 0.51, 95% CI 0.28–0.94), was not reached (NR) with icotinib or observation in the CORIN study (3-year DFS rate 96.1% vs. 84.0%; HR 0.23, 95% CI 0.07–0.81; p = 0.013), was 41.4 months with icotinib in the ICAPE study, and was 61.8 months with 12-month icotinib, 63.2 months with 6-month icotinib, and 23.7 months with observation in the ICTAN (GASTO1002) study (HR 0.40, 95% CI 0.27–0.61; p < 0.001).^11–19 Thus, EGFR-TKIs were introduced into the adjuvant therapy of resected EGFR-mutant NSCLC. However, the optimal duration of adjuvant EGFR-TKI treatment remains uncertain, with treatment durations mostly ranging from one to two years across previous studies.^11–19 Two-year treatment with icotinib was approved by the China National Medical Products Administration (NMPA) for adjuvant therapy of patients with EGFR-mutated stage II-IIIA NSCLC on June 1, 2021, following the positive results of the EVIDENCE study.¹⁵ Notably, the EVIDENCE study preceded the ADAURA study, with the first patient enrolled in June 8, 2015, compared to November 2015 for the ADAURA study.²⁰ The ICWIP study (NCT02125240) was the first clinical study designed to evaluate the efficacy and safety of three-year adjuvant EGFR-TKI treatment. In the ICWIP study, we evaluated the efficacy and safety of three-year adjuvant icotinib treatment compared with placebo in patients with completely resected stage II-IIIA EGFR-mutated lung adenocarcinoma who had completed four cycles of adjuvant chemotherapy.²¹

Icotinib is a first-generation EGFR-TKI developed by Betta Pharmaceuticals Co., Ltd, Hangzhou, China. The ICOGEN study demonstrated that icotinib (4.6 months; 95% CI 3.5–6.3) was noninferior to gefitinib (3.4 months; 95% CI 2.3–3.8) in terms of progression-free survival (PFS; HR 0.84, 95% CI 0.67–1.05; p = 0.13) among patients with previously treated advanced NSCLC.²² The CONVINCE study demonstrated that the median PFS was 11.2 months (95% CI 9.2–12.6) in the icotinib group and 7.9 months (95% CI 6.5–10.2) in the chemotherapy group (HR 0.61, 95% CI 0.43–0.87; p = 0.006) in the first-line treatment among patients with advanced EGFR-mutated lung adenocarcinoma.²³ The EVIDENCE study demonstrated that the median DFS was 47.0 months (95% CI 36.4–not estimable [NE]) in the icotinib group and 22.1 months (95% CI 16.8–30.4) in the chemotherapy group (HR 0.36, 95% CI 0.24–0.55; p < 0.0001) for patients with completely resected stage II-IIIA EGFR-mutated NSCLC in the two-year adjuvant icotinib treatment setting.¹⁵

Here, we reported the results of the efficacy and safety of three-year adjuvant icotinib treatment compared with placebo for Chinese patients with completely resected stage II-IIIA EGFR-mutated lung adenocarcinoma following standard adjuvant chemotherapy.

Results

Patients and treatment

From June 3, 2014, to November 15, 2018, 157 patients were screened for eligibility at 10 hospitals across China (List of participating hospitals is provided in Supplementary Material 1). Of these, 141 patients were enrolled and randomly assigned in a 1:1 ratio to receive either icotinib (n = 68) or placebo (n = 73). Following protocol amendment to version 2.0, 60 patients were enrolled (29 in the icotinib group and 31 in the placebo group). Additionally, 81 patients were enrolled prior to protocol version 2.0, among whom 58 patients (32 in the icotinib group and 26 in the placebo group) had not yet reached 2 years of adjuvant therapy at the time of the amendment and subsequently continued treatment under protocol version 2.0. Among the remaining 23 patients, 15 patients (icotinib group: n = 4; placebo group: n = 11) experienced disease recurrence. Three patients (icotinib group: n = 1; placebo group: n = 2) completed 2 years of adjuvant therapy under protocol version 1.3, discontinued the study treatment, and continued to be followed up for DFS and OS; Two patients in the icotinib group withdrew informed consent, and three patients in the placebo group refused further study treatment, all of whom remained in follow-up for DFS and OS. The full analysis set (FAS) and the safety set (SS) comprised 135 patients, 66 in the icotinib group and 69 in the placebo group, and six patients were excluded from both sets for not receiving any study treatment received. A total of 9.6% (13/135) of patients were excluded from the per-protocol set (PPS). Eight patients in the icotinib group and five in the placebo group discontinued treatment owing to withdrawal of consent, refusal of therapy, or other reasons. (Fig. 1).

Fig. 1.

Study flow chart. FAS full analysis set, PPS per-protocol set; SS safety set

The baseline characteristics were well balanced between the two groups (Table 1). A total of 55.6% (75/135) of patients had stage IIIA disease. A total of 85.2% (115/135) of patients underwent lobectomies alone. The median interval from resection to randomization was 4.5 months (interquartile range [IQR] 4.0–5.7) in the icotinib group and 4.5 months (IQR 3.9–5.1) in the placebo group.

Table 1.

Baseline characteristics of patients in the FAS^a

Characteristics	Icotinib group (n = 66)	Placebo group (n = 69)	P value
Median age (IQR), years	66 (57–72)	67 (57–72)	0.94
Sex			0.24
Male	24 (36.4%)	32 (46.4%)
Female	42 (63.6%)	37 (53.6%)
ECOG PS			0.99
0	46 (69.7%)	48 (69.6%)
1	20 (30.3%)	21 (30.4%)
Clinical stage			0.65
IIA	22 (33.3%)	24 (34.8%)
IIB	5 (7.6%)	8 (11.6%)
IIIA	39 (59.1%)	36 (52.2%)
Unknown	0	1 (1.4%)
T stage			0.62
T0	1 (1.5%)	0
T1	20 (30.3%)	16 (23.2%)
T2	34 (51.5%)	40 (58.0%)
T3	9 (13.6%)	7 (10.1%)
T4	2 (3.0%)	4 (5.8%)
Unknown	0	2 (2.9%)
N stageb			0.82
N0	5 (7.6%)	7 (10.1%)
N1	26 (39.4%)	27 (39.1%)
N2	35 (53.0%)	33 (47.8%)
Unknown	0	2 (2.9%)
M stage			NE
M0	66 (100.0%)	69 (100.0%)
Smoking history			0.64
Former or current smoker	14 (21.2%)	17 (24.6%)
Never smoker	52 (78.8%)	52 (75.4%)
EGFR mutation	66 (100.0%)	69 (100.0%)	0.33
Ex19del	28 (42.4%)	35 (50.7%)
L858R	38 (57.6%)	34 (49.3%)
Type of resection			0.77
Lobectomy	55 (83.3%)	60 (87.0%)
Othersc	9 (13.6%)	8 (11.6%)
Unknown	2 (3.0%)	1 (1.4%)
Completion of 4-cycle platinum-based adjuvant chemotherapy	63 (95.5%)	67 (97.1%)	NE
Unknown	3 (4.5%)	2 (2.9%)
Time from resection to randomization, months	4.5 (4.0-5.7)	4.5 (3.9-5.1)

Data are expressed as the median (IQR) or n (%)

ECOG, Eastern Cooperative Oncology Group; PS, performance status; EGFR, epidermal growth factor receptor; FAS, full analysis set; IQR, interquartile range; Ex19del, exon 19 deletion; L858R, exon 21 codon p.Leu858Arg; NE, not estimable

^aFAS included all randomly assigned patients who had received at least one study treatment

^bChi-square tests revealed no statistically significant differences in the distribution of N1 (p = 0.98), N2 (p = 0.55), or the overall N-stage (N1 and N2 combined, p = 0.30) between the icotinib group and placebo group

^cOthers included pneumonectomy, wedge resection, and bilobectomy

Efficacy

At the data cutoff date of October 31, 2024, the median follow-up for DFS in the FAS was 78.3 months (IQR 71.3–95.6) for the icotinib group and 69.9 months (IQR 35.2–96.4) for the placebo group. The median duration of treatment was 32.9 months (IQR 20.6–33.1) for icotinib and 20.0 months (IQR 10.7–33.1) for placebo. Among 135 patients assessed by the Independent Review Committee (IRC), 95 DFS events (recurrence or death) occurred: 46 in the icotinib group and 49 in the placebo group. The IRC-assessed median DFS was 48.2 months (95% CI 33.0–63.4) in the icotinib group and 20.0 months (95% CI 14.7–32.9) in the placebo group (HR 0.54, 95% CI 0.34–0.87; p = 0.011; Fig. 2a). The 3-year DFS rates were 58.5% (95% CI 39.9–73.1%) and 36.1% (95% CI 19.9–52.6%), and the 5-year DFS rates were 39.4% (95% CI 22.2–56.2%) and 25.8% (95% CI 8.7–47.1%) in the icotinib group and the placebo group, respectively. In the PPS, the IRC-assessed median DFS was 51.4 months (95% CI 33.9–65.6) in the icotinib group and 19.6 months (95% CI 13.2–32.9) in the placebo group (HR 0.50, 95% CI 0.30–0.82; p = 0.0061; Fig. 2b).

Fig. 2.

DFS by IRC and OS. a Kaplan‒Meier curves of DFS assessed by IRC in the FAS. b Kaplan‒Meier curves of DFS assessed by IRC in the PPS. c Kaplan‒Meier curves of OS in the FAS. d Kaplan‒Meier curves of OS in the PPS. The tick marks indicate censored data. Between-group differences were evaluated with the stratified log-rank test. P values are two-sided. HR and 95% CI were estimated with a stratified Cox proportional hazards model. DFS, disease-free survival; IRC, Independent Review Committee; FAS, full analysis set; PPS, per-protocol set; OS, overall survival; HR, hazard ratio; CI, confidence interval; NR, not reached; NE, not estimable

Subgroup analysis in the FAS showed a consistent DFS benefit in the placebo group across several predefined subgroups, including never-smokers, females, and patients with T3 disease, and those with an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 (Fig. 3a and Supplementary Fig. S1a–h). However, no significant difference in the IRC-assessed median DFS was observed between groups for stage II disease (median, 51.4 months [95% CI 39.7–NE] vs. 32.9 months [95% CI 15.0–65.1] (HR 0.57, 95% CI 0.29–1.11; p = 0.10; Supplementary Fig. S1a) or stage IIIA disease (median, 35.0 months [95% CI 22.3–64.7] vs. 16.2 months [95% CI 10.8–30.9]; HR 0.62, 95% CI 0.37–1.04; p = 0.071; Supplementary Fig. S1d).

Fig. 3.

Forest plots of DFS by IRC and OS. a Forest plot of the IRC-assessed DFS subgroup analysis in the FAS. b Forest plot of OS subgroup analysis in the FAS. HR and 95% CI were estimated with a stratified Cox proportional hazards model. The circles indicate HR among subgroups of patients, and the horizontal lines indicate corresponding 95% CIs. DFS, disease-free survival; IRC, Independent Review Committee; FAS, full analysis set, HR hazard ratio; CI, confidence interval; EGFR, epidermal growth factor receptor; Ex19del, exon 19 deletion; L858R, exon 21 codon p.Leu858Arg; ECOG, Eastern Cooperative Oncology Group; PS, performance status; OS, overall survival; NE, not estimable; NA, not assessed

For nodal status, the IRC-assessed median DFS did not differ significantly between the N0 and N1 subgroups (median, NR [95% CI 48.2–NE] vs. 51.4 months [95% CI 33.0–83.3]; HR 4.48, 95% CI 0.60–33.71; p = 0.15), but was significantly longer in N0 versus N2 disease (median, NR [95% CI 48.2–NE] vs. 33.9 months [95% CI 22.3–63.4]; HR 7.48, 95% CI 1.01–55.22; p = 0.048; Supplementary Fig. S1e). Compared to the placebo group, the icotinib group did not significantly improve IRC-assessed DFS in patients with N2 disease (HR 0.61, 95% CI 0.36–1.03; p = 0.066), N1 disease (HR 0.70, 95% CI 0.35–1.37; p = 0.30) or N0 disease (HR 0.18, 95% CI 0.02–1.90; p = 0.15; Supplementary Fig. S1f–h).

Investigator-assessed DFS in the FAS and PPS is shown in Supplementary Fig. S2a–b, with subgroup analysis presented in Supplementary Figs. S2c and S3a–h.

A post hoc analysis of IRC-assessed DFS from the date of surgical resection was conducted. In the FAS, the median DFS was 51.4 months (95% CI 38.3–69.2) in the icotinib group and 23.0 months (95% CI 19.9–39.3) in the placebo group (HR 0.53, 95% CI 0.33–0.84; p = 0.0073). In the PPS, the median DFS was 55.4 months (95% CI 38.9–70.7) in the icotinib group and 22.9 months (95% CI 18.6–39.3) in the placebo group (HR 0.48, 95% CI 0.29–0.80; p = 0.0044; Supplementary Fig. S4a–d).

According to the IRC assessment, the lung and brain were the most common sites of recurrence: lung recurrence occurred in 16.7% (11/66) of patients in the icotinib group and 23.2% (16/69) in the placebo group. Brain recurrence occurred in 21.2% (14/66) of patients in the icotinib group and 11.6% (8/69) in the placebo group (Supplementary Table S1). Among patients with brain recurrence, the investigator-assessed median DFS was 28.3 months (95% CI 14.1–43.3) in the icotinib group and 14.9 months (95% CI 0.4–21.3) in the placebo group (HR 0.31, 95% CI 0.11–0.85; p = 0.02).

The median follow-up for OS was 79.1 months (IQR 70.0–87.0) for the icotinib group and 78.7 months (IQR 64.4–85.8) for the placebo group. In the FAS, death occurred in 24.2% (16/66) and 23.2% (16/69) of patients in the icotinib group and the placebo group, respectively. The median OS was NR in both groups (the icotinib group: 95% CI 94.4–NE; the placebo group: 95% CI 103.3–NE; HR 0.84, 95% CI 0.39–1.81; p = 0.65; Fig. 2c). The estimated 5-year OS rates were 88.6% (95% CI 57.9–97.5%) and 83.4% (95% CI 57.7–94.2%) in the icotinib group and the placebo group, respectively. OS in the PPS was consistent to that in the FAS (Fig. 2d) and across subgroups (Fig. 3b).

When the OS results were adjusted by censoring at the initiation of subsequent anti-cancer therapy, the median OS remained NR in both groups (the icotinib group: 95% CI 94.4–NE; the placebo group: 95% CI NE-NE; HR 1.06, 95% CI 0.28–3.97; p = 0.93). The adjusted 5-year OS rates were 89.9% (95% CI 52.6–98.4%) in the icotinib group and 90.8% (95% CI 57.1–98.4%) in the placebo group.

Subgroup analysis of OS in the FAS was presented in Supplementary Fig. S5a–h. For patients with stage II disease, the median OS was NR (95% CI 94.4–NE) in the icotinib group and NR (95% CI NE-NE) in the placebo group (HR 1.46, 95% CI 0.46–4.64; p = 0.52; Supplementary Fig. S5a), with 5-year OS rates of 88.0% (95% CI 38.3–98.7%) and 86.3% (95% CI 55.1–96.4%), respectively. For stage IIIA disease, the median OS was NR (95% CI 78.7-NE) in the icotinib group and 103.3 months (95% CI 63.3–NE) in the placebo group (HR 0.68, 95% CI 0.28–1.67; p = 0.40; Supplementary Fig. S5d), with corresponding 5-year OS rates of 89.2% (95% CI 51.4–98.3%) and 80.2% (95% CI 43.3–94.3%), respectively.

Subsequent treatments

At the data cutoff date on October 31, 2024, 53.0% (35/66) of patients in the icotinib group and 50.7% (35/69) of patients in the placebo group received at least one subsequent anti-cancer treatment (Supplementary Table S2). The most frequent post-recurrence therapy was EGFR-TKIs, with osimertinib being the most commonly used agent in the icotinib group (39.4% [26/66]) and icotinib being the most commonly used in the placebo group (44.9% [31/69]). A total of 53.0% (35/66) of patients in the icotinib group and 31.9% (22/69) of patients in the placebo group received at least three lines of subsequent therapy. Among 69.7% (46/66) of patients who had DFS events in the icotinib group, 18 patients received osimertinib as the first post-study treatment. Among 71.0% (49/69) of patients who had DFS events in the placebo group, 31 patients received icotinib as the first post-study treatment. The median duration of subsequent icotinib treatment was 11.3 months (IQR 10.1–21.0).

Safety

In the SS, any-grade treatment-emergent adverse events (TEAEs) occurred in 54.5% (36/66) of patients in the icotinib group versus 31.9% (22/69) of those in the placebo group. The most common TEAEs were rash (30.3% [20/66] in the icotinib group vs. 11.6% [8/69] in the placebo group) and diarrhea (21.2% [14/66] in the icotinib group vs. 11.6% [8/69] in the placebo group). Any-grade treatment-related adverse events (TRAEs) occurred in 48.5% (32/66) of patients in the icotinib group and 23.2% (16/69) in the placebo group (Table 2). Only 1.5% (1/66) of patients in the icotinib group experienced grade 3 elevated alanine aminotransferase and aspartate aminotransferase, which were deemed treatment-related. No serious adverse events (AEs), dose adjustments, treatment discontinuation, or deaths were attributed to the treatment in either group.

Table 2.

AEs in the SS^a

AEsb	TEAEs				TRAEsc
	Icotinib group (n = 66)		Placebo group (n = 69)		Icotinib group (n = 66)		Placebo group (n = 69)
	Any grade	≥Grade 3	Any grade	≥Grade 3	Any grade	≥Grade 3	Any grade	≥Grade 3
Any events	36 (54.5%)	1 (1.5%)	22 (31.9%)	0	32 (48.5%)	1 (1.5%)	16 (23.2%)	0
Rash	20 (30.3%)	0	8 (11.6%)	0	20 (30.3%)	0	8 (11.6%)	0
Diarrhea	14 (21.2%)	0	8 (11.6%)	0	14 (21.2%)	0	8 (11.6%)	0
Pruritus	6 (9.1%)	0	3 (4.3%)	0	6 (9.1%)	0	2 (2.9%)	0
Liver dysfunction	4 (6.1%)	0	0	0	4 (6.1%)	0	0	0
Mouth ulceration	3 (4.5%)	0	1 (1.4%)	0	3 (4.5%)	0	1 (1.4%)	0
Alopecia	3 (4.5%)	0	0	0	3 (4.5%)	0	0	0
ALT increased	2 (3.0%)	1 (1.5%)	2 (2.9%)	0	2 (3.0%)	1 (1.5%)	1 (1.4%)	0
AST increased	2 (3.0%)	1 (1.5%)	1 (1.4%)	0	2 (3.0%)	1 (1.5%)	0	0
Dyspnea	2 (3.0%)	0	1 (1.4%)	0	1 (1.5%)	0	0	0
Blood bilirubin increased	2 (3.0%)	0	1 (1.4%)	0	2 (3.0%)	0	1 (1.4%)	0
Transaminase increased	2 (3.0%)	0	0	0	2 (3.0%)	0	0	0
Insomnia	1 (1.5%)	0	1 (1.4%)	0	0	0	0	0
Asthenia	1 (1.5%)	0	1 (1.4%)	0	1 (1.5%)	0	0	0
White blood cell count decreased	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Abdominal discomfort	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Abdominal pain	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Abdominal distension	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Hypertension	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Paronychia	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Cough	1 (1.5%)	0	0	0	0	0	0	0
Skin exfoliation	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Skin fissure	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Vocal cord hemorrhage	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Dizziness	1 (1.5%)	0	0	0	0	0	0	0
Blood creatinine increased	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Eye pain	1 (1.5%)	0	0	0	0	0	0	0
Eyelid erythema	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
Hiccups	1 (1.5%)	0	0	0	1 (1.5%)	0	0	0
γ-glutamyltransferase increased	0	0	1 (1.4%)	0	0	0	0	0
Nausea	0	0	1 (1.4%)	0	0	0	1 (1.4%)	0
Radiation pneumonitis	0	0	1 (1.4%)	0	0	0	0	0
Hypoesthesia	0	0	1 (1.4%)	0	0	0	1 (1.4%)	0
Dry skin	0	0	1 (1.4%)	0	0	0	1 (1.4%)	0
Basophil percentage increased	0	0	1 (1.4%)	0	0	0	0	0
Chest pain	0	0	1 (1.4%)	0	0	0	0	0
Blood triglyceride level increased	0	0	1 (1.4%)	0	0	0	0	0
Blood creatine phosphokinase increased	0	0	1 (1.4%)	0	0	0	0	0
Foreign body sensation	0	0	1 (1.4%)	0	0	0	1 (1.4%)	0
Swelling	0	0	1 (1.4%)	0	0	0	1 (1.4%)	0

Data are expressed as n (%)

AEs, adverse events; ALT, alanine aminotransferase; AST, aspartate aminotransferase; SS, safety set; TEAEs, treatment-emergent adverse events; TRAEs, treatment-related adverse events

^aSS included all randomly assigned patients who received at least one study treatment and one safety assessment

^bListed are all AEs in either group. AEs were coded using the Medical Dictionary for Regulatory Activities version 21.1 and graded based on the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0. Events are listed in descending order of frequency in the icotinib group

^cThe causality between AE and study treatment was assessed by investigators

Health-related quality of life (HRQoL)

A total of 104 patients completed at least one valid HRQoL assessment: 78.8% (52/66) of patients in the icotinib group and 75.4% (52/69) of patients in the placebo group. The baseline mean Functional Assessment of Cancer Therapy–Lung (FACT-L) scores were similar between the two groups. Throughout the study, no significant difference was observed in the mean change from baseline in the total FACT-L scores and subscale scores between groups (Supplementary Fig. S6a–f). A clinically significant decline in the total FACT-L score, defined as a reduction of ≥6 points, was observed in the icotinib group at visit 12 (Supplementary Fig. S6a).

Biomarker analysis

In total, 102 patients had tumor tissue samples at baseline to profile the genetic alterations, including 46 patients from the icotinib group and 56 patients from the placebo group. In the icotinib group, the most commonly mutated genes were TP53 (58.7% [27/46]), followed by ATM (10.9% [5/46]), BRCA1 (10.9% [5/46]), FGFR1 (10.9% [5/46]), and KMT2C (10.9% [5/46]; Fig. 4a). In the placebo group, the most commonly mutated genes were TP53 (50.0% [28/56]), followed by ATR (12.5% [7/56]), BRCA1 (10.7% [6/56]), BRCA2 (10.7% [6/56]), RB1 (10.7% [6/56]), and ROS1 (10.7% [6/56]; Fig. 4b). In 75 patients with a positive circulating tumor DNA (ctDNA)-minimal residual disease (MRD), the IRC-assessed DFS was significantly prolonged in the icotinib group compared to the placebo group (median, 50.9 months [95% CI 33.9–83.3] in the icotinib group vs. 20.1 months [95% CI 15.0–49.3] in the placebo group; HR 0.48, 95% CI 0.25–0.95; p = 0.013; Fig. 4c). Among 30 baseline ctDNA-MRD-negative patients, there was no between-group difference in DFS (median, 31.8 months [95% CI 20.0–NE] in the icotinib group vs. 30.9 months [95% CI 11.3–NE] in the placebo group; HR 0.73, 95% CI 0.29–1.83; p = 0.48; Fig. 4d).

Fig. 4.

Biomarker analysis. Genetic alteration spectrum in tumor tissue samples at baseline in the icotinib group (a) and the placebo group (b). c Kaplan‒Meier curves of IRC-assessed DFS between the icotinib and placebo groups in ctDNA-MRD-positive patients in the FAS. d Kaplan‒Meier curves of IRC-assessed DFS between the icotinib and placebo groups in ctDNA-MRD-negative patients in the FAS. e Kaplan‒Meier curves of IRC-assessed DFS between the icotinib and placebo groups in patients with EGFR Ex19del in the FAS. f Kaplan‒Meier curves of IRC-assessed DFS between the icotinib and placebo groups in patients with EGFR L858R in the FAS. g Kaplan‒Meier curves of OS between the icotinib and placebo groups in patients with EGFR Ex19del in the FAS. h Kaplan‒Meier curves of OS between the icotinib and placebo groups in patients with EGFR L858R in the FAS. HR and 95% CI were estimated with a stratified Cox proportional hazards model. EGFR, epidermal growth factor receptor; Ex19del, exon 19 deletion; L858R, exon 21 codon p.Leu858Arg; ECOG, Eastern Cooperative Oncology Group; PS, performance status; DFS, disease-free survival; IRC, independent review committee; CI, confidence interval; HR, hazard ratio; NE, not estimable; ctDNA, circulating tumor DNA; MRD, minimal residual disease; FAS, full analysis set; OS, overall survival; NR, not reached

At disease recurrence, blood samples from 53 patients were analyzed, including 24 patients from the icotinib group and 29 patients from the placebo group. In the icotinib group, common mutated genes included DNMT3A (16.7% [4/24]), EGFR (12.5% [3/24]), and TP53 (12.5% [3/24]) at disease recurrence (Supplementary Fig. S7a). In the placebo group, common mutated genes included DNMT3A (27.6% [8/29]), EGFR (17.2% [5/29]), and TP53 (13.8% [4/29]) at disease recurrence (Supplementary Fig. S7b). The EGFR T790M mutation was detected in 2 relapsed patients in the icotinib group.

In the FAS, in patients who had EGFR exon 19 deletion (Ex19del), the median DFS assessed by IRC was 51.4 months (95% CI 27.9–83.3) in the icotinib group and 24.5 months (95% CI 16.2–45.7) in the placebo group (HR 0.59, 95% CI 0.32–1.10; p = 0.097; Fig. 4e). Among patients harboring the EGFR exon 21 codon p.Leu858Arg (L858R) point mutation, the median DFS was 48.2 months (95% CI 27.6–64.7) in the icotinib group and 18.2 months (95% CI 10.9–33.7) in the placebo group (HR 0.62, 95% CI 0.36–1.07; p = 0.088; Fig. 4f). In the patient with Ex19del, the median OS was NR (95% CI 74.7–NE) in the icotinib group and 103.3 months (95% CI 103.3-NE) in the placebo group (HR 1.17, 95% CI 0.44–3.11; p = 0.76; Fig. 4g). There was also no significant between-group difference in the median OS (NR [95% CI NE-NE] in the icotinib group vs. NR [95% CI NE-NE] in the placebo group; HR 0.84, 95% CI 0.30–2.32; p = 0.74) among patients who harbored L858R (Fig. 4h).

In the icotinib group, IRC-assessed DFS (51.4 months [95% CI 27.9–83.3] in patients with Ex19del and 48.2 months [95% CI 27.6–64.7] in patients with L858R; HR 0.85, 95% CI 0.47–1.53; p = 0.58), investigator-assessed DFS (54.8 months [95% CI 27.9–NE] in patients with Ex19del and 44.5 months [95% CI 28.0–66.5] in patients with L858R; HR 0.79, 95% CI 0.43–1.43; p = 0.43) between patients with Ex19del and those with L858R. In the icotinib group, the median OS was NR (95% CI 74.7–NE) in patients with Ex19del and NR (95% CI NE–NE) in patients with L858R (HR 1.34, 95% CI 0.50–3.57; p = 0.56); the estimated 5-year OS rate was 81.0% (95% CI 38.3–95.5%) in patients with Ex19del and 94.2% (95% CI 53.3–99.8%) in patients with L858R. All supplementary figures and tables are provided in the Supplementary Material 1.

Discussion

Currently, osimertinib is the standard of care in patients with resected stage II–IIIA EGFR-mutated NSCLC globally. Meanwhile, both osimertinib and icotinib are the standard of care in this patient population in China. Osimertinib was approved by the China NMPA on April 7, 2021, for the adjuvant treatment of stage IB–IIIA EGFR-mutated NSCLC, while icotinib was approved on June 3, 2021, for stage II–IIIA EGFR-mutated NSCLC.

With a median follow-up of 78.3 months (IQR 71.3–95.6) for the icotinib group and 69.9 months (IQR 35.2–96.4) for the placebo group, the ICWIP study demonstrated that 3-year adjuvant icotinib treatment significantly prolonged IRC-assessed median DFS versus placebo (48.2 months [95% CI 33.0–63.4] in the icotinib group vs. 20.0 months [95% CI 14.7–32.9] in the placebo group; HR 0.54, 95% CI 0.34–0.87; p = 0.011) after completing four cycles of adjuvant chemotherapy. To our knowledge, this is the longest follow-up among adjuvant therapies of EGFR-TKIs in NSCLC and offers a more comprehensive view of 3-year treatment outcomes with first-generation EGFR-TKIs. The median follow-up was 47 months in the RADIANT study,¹¹ 36.5 months (IQR 23.8–44.8) in the ADJUVANT/CTONG1104 study,¹² 70 months in the IMPACT study,¹³ 33.0 months (IQR 17.8–43.1) in the EVAN study,¹⁴ 24.9 months (IQR 16.6–36.4) in the EVIDENCE study,¹⁵ 59.9 months (range 0–82) for the osimertinib group and 56.2 months (range 1–86) for the placebo group in the ADAURA study.²⁴ Notably, in the ICWIP study, the 5-year DFS rate was 39.4% (95% CI 22.2–56.2%) in the icotinib group and 25.8% (95% CI 8.7–47.1%) in the placebo group assessed by IRC, supporting the durable clinical benefit in delaying disease recurrence of adjuvant icotinib therapy. The treatment effect was more pronounced in the PPS (median DFS 51.4 months in the icotinib group vs. 19.6 months in the placebo group; HR 0.50, 95% CI 0.30–0.82; p = 0.0061). Icotinib exhibited a favorable safety profile and good compliance without negatively affecting HRQoL.

The ICWIP study is the first to extend a 3-year adjuvant EGFR-TKI treatment regimen, yielding a median DFS of 48.2 months (95% CI 33.0–63.4). The median DFS was 50.5 months in the RADIANT study,¹¹ 28.7 months (95% CI 24.9–32.5) in the ADJUVANT/CTONG1104 study,¹² 35.9 (95% CI 30.0–47.7) in the IMPACT study,¹³ 42.4 months (95% CI 31.7–NR) in the EVAN study,¹⁴ and 47.0 months (95% CI 36.4–NR) in the EVIDENCE study.¹⁵ Importantly, unlike previous studies, the ICWIP study mandated four cycles of adjuvant chemotherapy prior to study entry, potentially enhancing DFS. The early separation of Kaplan‒Meier curves was sustained for approximately 7 years, similar to the ADAURA study,²⁵ highlighting the durability of benefits not observed in the 2-year EGFR-TKI adjuvant therapy.^11–14 This long-term efficacy is likely attributable to the extended icotinib treatment duration, as supported by the survival benefits observed in the ICOMPARE study.¹⁶ Nevertheless, the optimal duration of adjuvant EGFR-TKI therapy in EGFR-mutated NSCLC remains uncertain. Prior studies have established the efficacy of adjuvant icotinib administered for 1, 1.5, and 2 years.^16–18 The ICWIP study extended this duration to 3 years. Ongoing studies, including the TARGET study (NCT05526755), are still evaluating the potential benefit of 5-year adjuvant osimertinib treatment. We observed that the ICTAN (GASTO1002) study recently reported shorter durations of adjuvant icotinib therapy and provided important evidence supporting the feasibility of treatment de-escalation in selected patients.¹⁹ However, the ICTAN and ICWIP studies were independent studies initiated during a similar period but designed to address different clinical questions. The ICWIP study initiated first-patient enrollment on June 3, 2014, earlier than the ICTAN study in July 2014. ICWIP was a randomized, double-blind, placebo-controlled phase III study evaluating three-year adjuvant icotinib treatment, whereas ICTAN was an open-label, randomized controlled phase III study investigating shorter durations of adjuvant icotinib treatment (1 year and 6 months) against an observation-only control group. Therefore, the two studies are not directly comparable. Taken together, these findings suggest that both prolonged EGFR-TKI exposure and treatment de-escalation strategies warrant further investigation to optimize the balance between efficacy, toxicity, treatment burden, and long-term disease control.

A key feature of the ICWIP study was sequential chemotherapy followed by EGFR-TKI therapy. Prior evidence has demonstrated an additive survival benefit of adjuvant chemotherapy in stage II-III NSCLC.⁷ The ADAURA study also supported this sequential strategy, showing that chemotherapy contributes to DFS improvement.²⁰ Furthermore, the tolerability and adherence to 3-year icotinib treatment were favorable in the ICWIP study.

Central nervous system control is critical in EGFR-mutated NSCLC. Although icotinib demonstrated superior efficacy for brain metastases versus whole-brain irradiation in the BRAIN study,²⁶ the most common recurrence sites in the ICWIP study were the brain (21.2% [14/66] with icotinib and 11.6% [8/69] with placebo) and lungs (16.7% [11/66] and 23.2% [16/69]).

Identifying suitable patients for adjuvant EGFR-TKI therapy remains a major issue that needs to be addressed. In the ICWIP study, patients with more advanced clinical or nodal stage appeared to have poorer DFS, likely related to the presence of occult metastatic disease in this patient population.²⁷ This finding suggested that patients with lymph node-positive disease or more advanced disease should be considered at high risk of recurrence and should intensify treatment. This underscores the need to identify patients with MRD undetectable by standard imaging. In this context, ctDNA-based next-generation sequencing (NGS) analysis at baseline could serve as a valuable tool, as supported by a previous study.²⁸ In the ICWIP study, patients with a positive ctDNA-MRD at baseline derived a significant DFS benefit in the icotinib group compared with the placebo group, whereas among ctDNA-MRD-negative patients, DFS was not significantly different between the two groups. Therefore, icotinib following chemotherapy represents a potential adjuvant strategy for baseline ctDNA-MRD-positive patients.

In the ICWIP study, most patients in the placebo group crossed over to icotinib treatment upon disease recurrence, and both study treatment groups received extensive subsequent anti-cancer therapies. Consequently, the DFS benefit did not translate to an OS benefit. The treatment landscape for this patient population has evolved significantly in recent years in China. The third-generation EGFR-TKI osimertinib was approved by the China NMPA on March 22, 2017. Since then, several other third-generation EGFR-TKIs have also been approved by the China NMPA.^29–33 Patients in both groups could receive third-generation EGFR-TKI treatment after disease recurrence or progression on first-generation EGFR-TKI treatment. In the ICWIP study, both groups received extensive and relatively balanced postrecurrence therapies. Among patients who experienced recurrence, 53.0% in the icotinib group and 50.7% in the placebo group received subsequent anti-cancer treatment. In the placebo group, 44.9% of recurrent patients crossed over to icotinib treatment, and 26.1% subsequently received third-generation EGFR-TKI treatment. Meanwhile, among recurrent patients in the icotinib group, 42.4% received third-generation EGFR-TKIs as subsequent therapy. In the ADAURA study, only approximately 22% of patients in the osimertinib group received subsequent therapy, compared with approximately 54% in the placebo group, of whom approximately 43% received third-generation EGFR-TKIs as salvage treatment.²⁴ Despite this substantial use of third-generation EGFR-TKIs in the placebo group after recurrence, a significant OS benefit in favor of adjuvant osimertinib was nonetheless demonstrated.²⁴ This indicates that there was more benefit for OS of osimertinib treatment in the adjuvant setting than in the first-line setting. Both groups of the ICWIP study had access to effective salvage therapies, particularly sequential osimertinib-based treatment strategies and third-generation EGFR-TKIs, which may have attenuated potential OS differences despite the DFS benefit observed with adjuvant icotinib treatment.

Emerging treatment strategies for NSCLC patients progressing after EGFR-TKIs have demonstrated improved outcomes. Sintilimab plus bevacizumab plus chemotherapy (ORIENT-31)³⁴ and ivonescimab plus chemotherapy (HARMONi-A)³⁵ showed superior PFS versus chemotherapy alone in the second-line setting. The OptiTROP-Lung04 study demonstrated significantly improved PFS for sacituzumab tirumotecan compared to pemetrexed plus platinum-based chemotherapy in patients with EGFR-mutated locally advanced or metastatic nonsquamous NSCLC that had progressed after EGFR-TKI treatment.³⁶ The SACHI study demonstrated significantly improved PFS for savolitinib plus osimertinib compared to chemotherapy for EGFR mutation-positive, MET-amplified advanced NSCLC patients after EGFR-TKI failure.³⁷ The ALTER 0303 study demonstrated that OS was significantly longer in the anlotinib group than in the placebo group in patients with advanced NSCLC progressing after second-line or further treatment.³⁸ The OptiTROP-Lung03 study demonstrated significantly improved PFS for sacituzumab tirumotecan compared to docetaxel in patients with locally advanced or metastatic EGFR-mutated NSCLC after previous treatment failure with EGFR-TKIs and platinum-based chemotherapy.³⁹ More studies on the postline treatment of EGFR-mutated NSCLC patients are ongoing.

In contrast, the ADAURA study was an international, multi-center clinical trial, and patients in the ADAURA study had few subsequent treatment options, which may explain the observed OS benefit with osimertinib. In advanced NSCLC patients, third-generation EGFR-TKIs consistently demonstrated superior PFS versus first-generation EGFR-TKIs in several phase 3 trials.^40–45 The FLAURA study demonstrated a significantly prolonged median OS of osimertinib compared to gefitinib or erlotinib, with 38.6 months (95% CI 34.5–41.8) versus 31.8 months (95% CI 26.6–36.0; HR 0.80, 95.05% CI 0.64–1.00; p = 0.046), respectively.⁴⁶ Subsequent Chinese pharmaceutical companies developed third-generation EGFR-TKIs that did not demonstrate or report significant OS benefits compared to first-generation EGFR-TKIs, mainly due to the evolving postprogression treatment landscape in China.^40–45,47

Notably, the 5-year OS outcomes (88.6% [95% CI 57.9–97.5%]) for the icotinib group in the ICWIP study were comparable to those reported for the osimertinib group in the ADAURA study (85% [95% CI 79–89%]).²⁴ To date, no head-to-head clinical study has directly compared first- and third-generation EGFR-TKIs in the adjuvant setting. Therefore, current evidence is insufficient to definitively establish the superiority of third-generation EGFR-TKIs over first-generation EGFR-TKIs in the adjuvant treatment setting. Importantly, both icotinib and osimertinib have been established as the China NMPA-approved adjuvant options for EGFR-mutated NSCLC since June 1, 2021, and April 14, 2021, respectively. At present, icotinib is a standard adjuvant treatment option only in China, while osimertinib is approved not only in China but also in the United States of America and other countries. While the findings of the ICWIP study are specific to Chinese patients rather than global patients, the study provides important evidence supporting prolonged adjuvant icotinib treatment in Chinese patients within the context of local clinical practice.

The use of first-generation EGFR-TKIs in the adjuvant treatment setting enables a sequential treatment strategy, allowing patients to receive osimertinib or other third-generation EGFR-TKIs upon disease recurrence. Given that resistance mechanisms and subsequent treatment approaches following first-generation EGFR-TKIs are relatively well characterized, this comprehensive management strategy may facilitate a more continuous and manageable EGFR-targeted treatment paradigm.⁴⁸

In the ICWIP study, TP53 mutations were frequent in recurrent patients, which may represent a mechanism of resistance to icotinib, consistent with previous studies.^49,50 DNMT3A mutations may be associated with resistance and brain metastasis.^51,52 Furthermore, EGFR T790M mutations were identified in 2 recurrent patients in the icotinib group, consistent with the known resistance mechanisms to first-generation EGFR-TKIs.⁴⁸ In the ICWIP study, there was no significant difference in DFS between patients harboring Ex19del and those with L858R mutations, demonstrating consistent efficacy for these patients, similar to the CONVINCE study.²³

The safety profile of icotinib in the ICWIP study was consistent with previous findings,^15,22,23 and no new safety concerns emerged. Despite the 3-year icotinib treatment duration in the ICWIP study being longer than the 2-year treatment duration in the EVIDENCE study,¹⁵ no cumulative toxicities were observed. Compared with 3-year osimertinib treatment (91.4% of any grade; 10.7% of ≥grade 3), 3-year icotinib treatment (48.5% of any grade; 1.5% of ≥grade 3) was associated with a lower incidence of TRAEs.²⁰ Notably, no AEs led to dose adjustment, treatment discontinuation, or death, and no serious AEs were reported in the ICWIP study. In contrast, the incidence of AEs leading to dose adjustment, treatment discontinuation, or death and the incidence of serious AEs for osimertinib were 12% (42/337), 12.3% (43/337), <1% (1/337), and 20% (68/337), respectively, in the ADAURA study.²⁵ Furthermore, the percentage of patients who completed 3 years of planned adjuvant treatment with icotinib (87.9% [58/66]) in the ICWIP study was higher than that with osimertinib (66% [224/339]) in the ADAURA study,²⁵ supporting the favorable tolerability of icotinib. HRQoL was not adversely affected by the 3-year icotinib treatment, reinforcing its suitability for long-term use in adjuvant settings.

Several limitations of the ICWIP study are acknowledged. This study enrolled exclusively Chinese patients, which may limit the generalizability of these findings to other ethnic patient populations. The biomarker analysis was exploratory, its prognostic and predictive value requires future validation. Additionally, the cost-effectiveness of available adjuvant EGFR-TKIs should be considered in clinical decision-making. Finally, OS data remain immature and are awaited with longer follow-up.

In conclusion, the ICWIP study met its primary endpoint, demonstrating that the 3-year adjuvant treatment of icotinib significantly improved DFS compared to placebo in patients with completely resected stage II-IIIA EGFR-mutated lung adenocarcinoma who had completed adjuvant chemotherapy, with a manageable safety profile.

Materials and methods

Study design and patients

The ICWIP study was a multi-center, randomized, double-blind, placebo-controlled, phase 3 trial conducted at 13 hospitals across China. The study design has been described previously.²¹ Eligible patients were 18–75 years old, with pathologically confirmed stage II-IIIA lung adenocarcinoma as defined by the 7th edition of the American Joint Committee on Cancer (AJCC 7th) TNM staging system. All patients had sensitizing EGFR mutations (Ex19del or L858R) and had undergone complete surgical resection (R0) followed by four cycles of platinum-based adjuvant chemotherapy.

The patients were required to initiate the study treatment within 4–8 weeks of the last dose of chemotherapy. Additional eligibility criteria included an ECOG PS of 0 or 1, a life expectancy of at least 6 months, and adequate bone marrow, cardiac, hepatic, and renal function. Key exclusion criteria included prior systemic therapy or radiotherapy, presence of the EGFR exon 20 T790M mutation, or a history of other malignancies within the past 5 years, except for adequately treated basal cell carcinoma or carcinoma in situ. A full list of eligibility criteria is provided in the Supplementary Material 2.

The study was conducted in accordance with the Declaration of Helsinki and the International Council for Harmonization guidelines for Good Clinical Practice. The protocol and all amendments were reviewed and approved by the institutional review board or independent ethics committee at each participating hospital (approval number of Cancer Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College: 14-001/775). All patients provided written informed consent prior to enrollment.

Randomization and masking

Eligible patients were randomly assigned 1:1 to receive either icotinib or placebo. Randomization was performed using the Pocock and Simon minimization method with a computerized interactive web-based system. Stratification factors included disease stage (II vs. IIIA), EGFR mutation type (Ex19del vs. L858R), ECOG PS (0 vs. 1), and smoking history (yes vs. no). Treatment assignments were blinded to patients, clinicians, and study investigators, including those responsible for the outcome assessment.

Procedures

Patients received icotinib 125 mg orally three times daily or a matching placebo until disease recurrence, unacceptable toxicity, death, or completion of the study treatment. Unmasking was performed when disease recurrence occurred. After unmasking, patients in the icotinib group stopped icotinib treatment, and patients in the placebo group were permitted to crossover to icotinib treatment until disease progression. Dose modifications were made based on the severity of toxicity. If ≥grade 3 AEs could not be managed with supportive care, icotinib was interrupted until recovery to ≤grade 2. The maximum duration of study treatment interruption per episode was 14 days, and treatment must be immediately suspended upon suspicion of interstitial lung disease (ILD). The study drug is permanently withheld following the ILD confirmation but may be resumed after definitive radiological exclusion.

Disease recurrence was assessed by IRC and investigators using the Response Evaluation Criteria in Solid Tumors version 1.1. Imaging examinations, including contrast-enhanced computed tomography, magnetic resonance imaging, and bone scans, were performed within 28 days of treatment initiation, every 12 weeks during the first 3 years, and every 24 weeks thereafter. During the 24-week intervals, patient survival status was also evaluated by telephone every 12 weeks. AEs were monitored through physical examinations, vital signs, laboratory tests, and a 12-lead electrocardiogram at each 12-week visit. HRQoL was assessed using the FACT-L questionnaire at baseline and every 12 weeks. Blood samples were collected at baseline and every 12 weeks during the first 96 weeks of study treatment. Thereafter, optional sampling was conducted every 24 weeks and at disease recurrence, contingent upon patient consent for NGS testing of genetic alterations. Tumor tissues were collected at baseline using NGS to test genetic alterations (Supplementary Material 3). The ctDNA positive means detectable MRD.

Outcomes

The primary endpoint was IRC-assessed DFS, defined as the time from randomization to the first radiological evidence of recurrence or death from any cause. Secondary endpoints included 3-year DFS rates and 5-year DFS rates, OS, 5-year OS rate, safety, and HRQoL. OS was defined as the time from randomization to death from any cause. Safety was evaluated using the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0. Biomarkers were analyzed as exploratory endpoints.

Statistical analysis

The sample size was calculated based on the primary endpoint DFS. According to the results of the ADJUVANT/CTONG1104 study,¹² we assumed a median DFS of 40 months and 22 months in the icotinib group and the placebo group, respectively. Assuming a 45-month enrollment period and a 5-year follow-up, 112 patients (89 DFS events) were required to detect an HR of 0.55 in favor of the icotinib group, with a two-sided α of 0.05 and 80% power. Accounting for an expected dropout rate of 10%, the final target sample size was set at 124 patients (62 per group; protocol version 2.0). The study protocol was amended from version 1.3 to version 2.0 on July 4, 2017; the amendment included the extension of the enrollment period from 12 to 45 months and prolongation of the treatment duration from 2 to 3 years. Following the protocol version 2.0, no further protocol amendment was made to extend enrollment beyond the planned 45-month period. However, 141 patients were ultimately randomized. This was due to a subset of patients who had already provided informed consent form and completed screening and who were permitted to enroll as the study approached its target accrual, while patient demand for participation remained steady during this period. This operational adjustment did not compromise the study design, randomization integrity, blinding procedures, or primary endpoint assumptions.

Efficacy analyses were performed in the FAS, which included all randomized patients who received at least one dose of study treatment. DFS and OS were also analyzed in the PPS, serving as supportive results. The PPS was consisted of all randomized patients who received the study treatment according to the protocol, did not receive prohibited concomitant medications, and had no major protocol deviations, including patients who discontinued the study due to the occurrence of endpoint events. DFS and OS were analyzed using the Kaplan‒Meier method and compared using stratified log-rank tests. Censoring rules were applied for patients without DFS events at the date of the last efficacy assessment or OS events at the date last known to be alive. HRs and 95% CIs were calculated using stratified Cox proportional hazards models. OS results were adjusted by censoring at the initiation of subsequent anti-cancer therapy. The SS was defined as all patients who received at least one dose of study treatment and one safety assessment. FACT-L scores were analyzed using Analysis of Covariance. All statistical analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, NY, USA). The full statistical analysis plan is provided in the Supplementary Material 4.

Supplementary information

41392_2026_2994_MOESM1_ESM.docx ^{(15.4MB, docx)}

Supplementary Material 1 Supplementary figures, supplementary tables, and list of participating hospitals

Supplementary Material 2 Study protocol ^{(260KB, docx)}

Supplementary Material 3 Methods ^{(16.3KB, docx)}

41392_2026_2994_MOESM4_ESM.docx ^{(49.8KB, docx)}

Supplementary Material 4 Statistical Analysis Plan

CONSORT_2025_editable_checklist ^{(33.4KB, docx)}

Author_Checklist_for_Research_Article ^{(13.8MB, pdf)}

Author contributions

Yuan-Kai Shi was the principal investigator, contributing to funding acquisition from the Beijing Municipal Science and Technology Commission Major Projects (Grant Nos. D141100000214005; D14110700020000), study design, supervision, coordination, patient enrolment, patient management, manuscript writing, review, and editing. Yu-Tao Liu contributed to study design, patient enrolment, and patient management. Le Tang, Zhi-Yi Song, De-Ruo Liu, Gang Cheng, Wen-Hua Xiao, Shu-Cai Zhang, Yi Hu, Li Zhang, Jun-Feng Liu, Jun Wang, and Gui-Lan Dong contributed to patient enrollment and patient management. Xiao-Hong Han contributed to EGFR gene testing at the central laboratory. All authors had access to all data in this study and were ultimately responsible for the decision to submit the manuscript for publication. The authors were responsible for the accuracy and completeness of the data and the fidelity of the study to the protocol. All authors have read and approved the article. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Data availability

The data utilized in this study are presented within the article or its supplementary materials. The raw data have been archived in the Genome Sequence Archive, accessible under the accession number HRA019228 (https://ngdc.cncb.ac.cn/gsa-human/). These data are under controlled access by human privacy regulations and are only available for research purposes. Additional data in a deidentified form could contact the corresponding author professor Yuan-Kai Shi (E-mail: syuankai@cicams.ac.cn). The study protocol (Supplementary Material 2) and statistical analysis plan (Supplementary Material 4) are available as supplementary materials to this publication.

Competing interests

All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/disclosure-of-interest/and declare no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous three years; and no other relationships or activities that could appear to have influenced the submitted work.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41392-026-02994-w.