The impact of oncogenic driver mutations on neoadjuvant immunotherapy outcomes in patients with resectable non-small cell lung cancer

Abstract

Background

Neoadjuvant immunotherapy has been demonstrated to be effective and safe in resectable non-small cell lung cancer (NSCLC) patients. However, the presence of different oncogenic driver mutations may affect the tumor microenvironment and consequently influence the clinical benefit from immunotherapy.

Methods

This retrospective study included consecutive NSCLC patients (stage IIA to IIIB) who underwent radical surgery after receiving neoadjuvant immunotherapy at a single high-volume center between December 2019 and August 2022. Pathological response and long-term outcomes were compared based on the driver oncogene status, and RNA sequencing analysis was conducted to investigate the transcriptomic characteristics before and after treatment.

Results

Of the 167 patients included in this study, 47 had oncogenic driver mutations. KRAS driver mutations were identified in 28 patients, representing 59.6% of oncogenic driver mutations. Of these, 17 patients had a major pathological response, which was significantly higher than in the non-KRAS driver mutation group (60.7% vs. 31.6%, P = 0.049). Multivariate Cox regression analysis further revealed that the KRAS driver mutation group was an independent prognostic factor for prolonged disease-free survival (hazard ratio: 0.10, P = 0.032). The median proportion of CD8⁺ T cells was significantly higher in the KRAS driver mutation NSCLCs than in the non-driver mutation group (18% vs. 13%, P = 0.030). Furthermore, immune-related pathways were enriched in the KRAS driver mutation NSCLCs and activated after immunotherapy.

Conclusion

Our study suggests that NSCLC patients with KRAS driver mutations have a superior response to neoadjuvant immunotherapy, possibly due to their higher immunogenicity. The findings highlight the importance of considering oncogenic driver mutations in selecting neoadjuvant treatment strategies for NSCLC patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-023-03560-x.

Introduction

Lung cancer remains the foremost cause of cancer-related deaths worldwide, with non-small cell lung cancer (NSCLC) accounting for approximately 85% of all lung cancer cases [1]. NSCLC has emerged as a paradigm of precision medicine in recent decades due to the identification of various disease subtypes that are defined by specific oncogenic driver mutations, leading to the development of a range of molecularly targeted therapies [2]. Simultaneously, significant strides have been made in the development of immune-checkpoint inhibitors (ICIs), particularly antagonistic antibodies that target the PD-1/PD-L1 axis, for treatment of numerous types of cancers including NSCLC [3, 4].

For decades, chemotherapy has remained the standard neoadjuvant treatment for resectable NSCLCs [5]. Recently, neoadjuvant therapy with nivolumab in combination with platinum-based chemotherapy was approved by the U.S. Food and Drug Administration as the first-line treatment for NSCLC patients, with pembrolizumab also being added to preoperative neoadjuvant indications in the latest National Comprehensive Cancer Network (NCCN) guideline [6–8]. Although the CheckMate 816 study showed encouraging results for the combination of immunotherapy with chemotherapy in patients with resectable stage IB to IIIA NSCLC [6], it remains unclear whether different driver mutation subtypes influence the response to immunotherapy [9]. Some recent studies found that oncogene-positive NSCLC respond less than oncogene-negative NSCLC following neoadjuvant immunotherapy [10, 11], while other study demonstrated the opposite results [12].

The efficacy of ICIs in NSCLC is based on the immunogenicity of tumors and corresponds with the expression of PD-L1 on tumor cells, tumor mutational burden (TMB), and the immune infiltration within the tumor microenvironment [13]. Different oncogenic driver mutations in NSCLC are associated with distinct tumor immune microenvironments, which are closely linked to the efficacy of ICIs [14]. Although patients with specific oncogenic driver mutations in advanced NSCLC have been recognized as poorly responsive to immunotherapy, evidence in the resectable stage or locally advanced setting is still lacking [15, 16]. Furthermore, the efficacy of neoadjuvant immunotherapy on long-term survival in patients with different oncogenic driver alterations remains to be explored.

Therefore, the aim of this study is to investigate the impact of oncogenic driver mutations in patients with resectable NSCLC who underwent radical surgery after the completion of neoadjuvant immunotherapy. Through further RNA sequencing analysis, we aim to provide insights into the potential mechanisms that lead to differential response of neoadjuvant immunotherapy.

Methods

Study design

This study was a retrospective, single-center study. Eligible patients were defined as: 1. treated with neoadjuvant immunotherapy (including PD-1 inhibitors and PD-L1 inhibitors); 2. diagnosed with IIA to IIIB NSCLC, as determined by the American Joint Committee on Cancer (AJCC, 8th edition); 3. received radical surgery at Shanghai Pulmonary Hospital after the completion of neoadjuvant immunotherapy between December 2019 and August 2022. The exclusion criteria of patients were as follows: 1. incomplete treatment modality or clinical information; 2. M1 metastases or N3 lymph node metastases; 3. pathologically confirmed locoregional relapses; 4. histologically confirmed small cell lung cancer (SCLC) or SCLC combined with NSCLC; 5. targeted therapy prior to the radical surgery; 6. without gene testing. The flowchart of this study is shown in Fig. 1. The enrolled patients were divided into driver mutation group and non-driver mutation group. Due to KRAS mutations being the most prevalent oncogenic driver mutations in the patients who received immunotherapy in this study, driver mutation group was further subdivided into KRAS driver mutations and non-KRAS driver mutations to analyze short-term outcomes and long-term survival among the three groups.

Fig. 1.

Study flowchart. SCLC, small cell lung cancer

This study was conducted in accordance with the Declaration of Helsinki and was approved by the institutional review board of Shanghai Pulmonary Hospital (K23-203). Informed consent was waived due to the retrospective and observational nature of the study. Only patients with telephone follow-up were provided with informed consent in this study.

Data collection and follow-up

The demographic and clinical data of the included patients, including clinical characteristics, molecular profiles, treatment details, surgical parameters, pathological data, and survival information, were collected retrospectively. The KRAS, EGFR, BRAF, ROS1, and ALK status of all enrolled patients were tested, while other oncogene (RET, HER-2 and c-MET) status was tested in part of the NSCLC patients through next-generation sequencing, immunohistochemistry (IHC), or polymerase chain reaction (PCR) by using tumor samples. In accordance with the manufacturer’s instructions, the majority of patients (94.6%) in this study underwent Amplification Refractory Mutation System-qPCR method for hotspot mutation detection using the genetic mutation assay kit from Beijing ACCB Biotech Ltd. The panel of PCR mainly consisted of two types: (1) KRAS, EGFR, BRAF, ROS1, ALK; (2) KRAS, EGFR, BRAF, ROS1, ALK, RET, NRAS, PIK3CA, HER-2 and C-MET. In addition, IHC was conducted in some patients to detect ALK fusions using the VENTANA D5F3 antibody clone. The oncogenic driver mutations were defined as any of the following events: KRAS mutations, EGFR mutations, HER-2 amplifications or mutations, BRAF fusions or mutations, RET fusions, ALK translocations or mutations, ROS1 rearrangements or mutations, or C-MET amplifications or mutations.

PD-L1 expression was quantified as the proportion of PD-L1 positive tumor cells through PD-L1 immunohistochemistry assay. Cases with < 100 total tumor cells for scoring were defined as not applicable (NA). The short-term outcomes of interest were rates of major pathologic response (MPR) and pathologic complete response (pCR). MPR was defined as the presence of 10% or fewer residual viable tumor cells, while pCR was defined as no viable tumor cells in both primary and lymph nodes.

Follow-up information is updated periodically through clinics visit, and telephone follow-up was conducted if outpatient follow-up was not feasible through medical records. The end point of follow-up for this study was January 31, 2023.

The long-term outcomes were also analyzed according to the driver mutation status, including overall survival (OS) and disease-free survival (DFS). OS of patients was calculated as the duration between radical surgery and death from any cause. DFS was calculated as the period between the initial surgery and the first occurrence of either tumor recurrence, metastasis, or death from any cause. Survival data were censored at the date of last follow-up.

RNA sequencing analysis

The RNA-seq data of 12 baseline tumor samples and 26 surgical tumor samples were partially obtained from 30 NSCLC patients who were recruited in two published phase II studies of neoadjuvant anti-PD-1 therapy at Shanghai Pulmonary Hospital (LungMate 001 and LungMate 002) [17, 18]. The protocol of sample preparation, library preparation and sequencing were conducted as described previously [18].

The pathway analyses were performed through gene set enrichment analysis (GSEA) with the downloaded GSEA software (www.broadinstitute.org/gsea). The GSEA results based on the criterion of P value < 0.05 and false discovery rate (FDR) ≤ 0.25 were filtered. Then, candidate pathways were visualized based on the normalized enrichment score from the filtered list.

The proportion of the immune cell infiltration of each sample was evaluated by cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT) algorithm based on the LM22 gene set [19]. The relevant procedure was carried out using CIBERSORT package in R.

Statistical analysis

The analysis of categorical variables was carried out through either the chi-square test or Fisher's exact test, as appropriate. For continuous variables with normal distribution, the mean and standard deviation (SD) were calculated and analyzed using Student's t test. Conversely, for continuous variables with non-normal distribution, the median and interquartile range (IQR) were computed and analyzed using the Kruskal–Wallis test. The survival analysis of OS and DFS across mutation groups was conducted using the log-rank test and was visually represented through the Kaplan–Meier curve. To assess the association between each variable and survival outcomes, hazard ratios (HR) and their confidence intervals (CI) were calculated through Cox proportional hazards regression models.

The statistical analyses were conducted using R version 4.2.2. All tests were two-tailed, and P < 0.05 was considered statistically significant.

Results

Clinical characteristics of patients

This study involved a retrospective screening of 241 patients who underwent radical surgery subsequent to receiving neoadjuvant immunotherapy, spanning from December 2019 to August 2022. The exclusion criteria were duly taken into account, resulting in a final sample size of 167 patients for analysis (Fig. 1). The mean age of the study population was 62 years, with males comprising 89.8% of the cohort, and 95.2% (159/167) of the patients received immunotherapy combined with chemotherapy. Among the cohort of enrolled patients, 47 (28.1%) patients presented with driver mutations. Comparison of baseline clinical characteristics between the driver mutation group and the non-driver mutation group revealed no statistically significant distinctions (Supplementary table 1).

To investigate the proportion of driver gene mutations in the driver mutation group, we visualize the detailed information of driver mutation group by waterfall plot (Fig. 2A). Among them, there were 59.6% (28/47) were found to have KRAS mutations, while 17.0% (8/47) harbored EGFR mutations (Fig. 2B). Despite NSCLC patients with EGFR mutations often serving as contraindications for immunotherapy, several patients with EGFR mutations received neoadjuvant immunotherapy in this cohort due to insufficient tissue samples or false negative biopsy results, while further genetic testing of surgical specimens revealed the presence of EGFR mutations. Other oncogenemutations, such as RET fusion, ROS1 fusion, BRAF, C-MET, and HER-2 mutations, accounted for 23.4% (11/47) of driver mutations in this cohort. Detailed information of patients with non-KRAS driver mutations was shown in Supplementary table 2. The percentage of residual viable tumor cells displayed heterogeneity among the various oncogenic driver mutations, and accordingly, the MPR and pCR rates also varied (Supplementary Table 3). Given that KRAS driver mutations constituted the majority of driver mutations, and the residual viable tumor cells of KRAS driver mutations were significantly lower than the other driver mutations (2.5% vs. 30.0%, P = 0.014, Fig. 2C), the driver mutation group was divided into two groups for further analysis: the KRAS driver mutations and the non-KRAS driver mutations.

Fig. 2.

Overview of the clinicopathological characteristics of neoadjuvant immunotherapy-treated NSCLCs with oncogenic driver mutations. A Waterfall plot of clinicopathological characteristics of patients with resected NSCLC. B The pie chart shows the distribution and proportion of each oncogenic driver mutations. C Comparison of the residual viable tumor cells between the KRAS driver mutation group and the non-KRAS driver mutation group. LUAD lung adenocarcinoma; LUSC lung squamous cell carcinoma; MPR major pathological response; pCR pathological complete response

Among all patients, the non-KRAS driver mutation group consisted of a higher proportion of females and never smokers compared to the other two groups (Table 1). Furthermore, non-KRAS driver mutation group had the highest ratio of forced expiratory volume in 1 s/forced vital capacity (FEV1/FVC), with a median of 102.30% among the three groups (P = 0.020), this may be related to the fact that smokers have the lowest proportion among this group of patients. Other characteristics including age, body mass index (BMI), Eastern Cooperative Oncology Group (ECOG) score, driver mutation detective method, stage at diagnosis and neoadjuvant regimes were comparable among the three groups.

Table 1.

Baseline clinical characteristics of included patients stratified by driver mutations

Variables	KRAS driver mutations	Non-KRAS driver mutations	Non-driver mutations	P-value
Number	28	19	120
Age, years (mean (SD))	63.89 (8.26)	59.79 (9.03)	61.98 (7.81)	0.227
Gender (%)
Male	27 (96.4)	12 (63.2)	111 (92.5)	< 0.001
Female	1 (3.6)	7 (36.8)	9 (7.5)
BMI, kg/m2 (mean (SD))	22.75 (3.40)	23.65 (0.82)	23.40 (2.45)	0.768
ECOG score (%)
0	17 (60.7)	13 (68.4)	80 (66.7)	0.811
1	11 (39.3)	6 (31.6)	40 (33.3)
Smoking status (%)
Never	6 (21.4)	10 (52.6)	31 (25.8)	0.023
Former smoker	2 (7.1)	3 (15.8)	6 (5.0)
Current smoker	20 (71.4)	6 (31.6)	83 (69.2)
FEV1/FVC, % (median [IQR])	93.85 [84.28, 97.00]	102.30 [94.60, 104.10]	95.65 [90.67, 101.05]	0.020
Detective method (%)
NGS	2 (7.1)	1 (5.3)	6 (5.0)	0.903
PCR	26 (92.9)	18 (94.7)	114 (95.0)
Stage at diagnosis (%)
IIA/IIB	5 (17.9)	1 (5.3)	20 (16.7)	0.415
IIIA/IIIB	23 (82.1)	18 (94.7)	100 (83.3)
Neoadjuvant regimens (%)
ICI alone	1 (3.6)	0 (0.0)	7 (5.8)	0.513
ICI + chemotherapy	27 (96.4)	19 (100.0)	113 (94.2)

SD standard deviation; BMI body mass index; ECOG Eastern Cooperative Oncology Group; FEV1 forced expiratory volume in 1 s; FVC forced vital capacity; IQR interquartile range; PCR polymerase chain reaction; NGS next-generation sequencing; ICI immune checkpoint inhibitor

All patients with driver mutations received PD-1 inhibitors as neoadjuvant settings, while 11.7% of the patients in non-driver mutation group received PD-L1 inhibitors (Supplementary Table 4). Carboplatin-based and cisplatin-based chemotherapy were used in patients with combination therapy. The neoadjuvant cycles were well-balanced among the three groups (P = 0.875), and the rate of postoperative immunotherapy or postoperative chemotherapy showed no significance among the three groups (P = 0.497 and P = 0.551, respectively).

Surgical profile and pathological information

The majority of patients (80.8%) underwent video-assisted thoracoscopic surgery (VATS), with lobectomy being the most commonly used resection method (Table 2). Only one patient had positive margins. The surgical approach, resection method, operative time, estimated blood loss and resection margins were comparable among the three groups. As shown in Table 3, the tumor size and staging parameters showed no significance among the three groups. However, the histology was found to be different among the three groups (P < 0.001), with the KRAS driver mutation group having the highest proportion of lung adenocarcinoma (LUAD) (78.6%), while patients in the non-driver mutation group were more likely to be diagnosed with lung squamous cell carcinoma (LUSC) (53.3%). Furthermore, the KRAS driver mutation group had the lowest residual viable tumor cells (Supplementary Fig. 1A), and highest MPR rate and pCR rate among the three groups (Supplementary Fig. 1B). Specifically, the MPR rate of the KRAS driver mutation group was significantly higher than the non-KRAS driver mutation group (60.7% vs. 31.6%, P = 0.049).

Table 2.

Intraoperative parameters of patients who received immunotherapy stratified by driver mutations

Variables	KRAS driver mutations	Non-KRAS driver mutations	Non-driver mutations	P-value
Number	28	19	120
Surgical approach (%)
VATS	22 (78.6)	16 (84.2)	97 (80.8)	0.890
Open surgery	6 (21.4)	3 (15.8)	23 (19.2)
Resection method (%)
Pneumonectomy	1 (3.6)	3 (15.8)	5 (4.2)	0.102
Lobectomy	27 (96.4)	16 (84.2)	115 (95.8)
Operative time, min (median [IQR])	139.50 [129.75, 199.75]	140.00 [105.50, 191.00]	164.00 [122.50, 199.00]	0.743
Estimated blood loss, ml (median [IQR])	50.00 [50.00, 100.00]	50.00 [50.00, 100.00]	50.00 [50.00, 100.00]	0.973
Resection margins (%)
R0	28 (100.0)	19 (100.0)	119 (99.2)	0.821
R1/2	0 (0.0)	0 (0.0)	1 (0.8)

VATS video-assisted thoracoscopic surgery; IQR interquartile range

Table 3.

Pathological data of patients who received immunotherapy stratified by driver mutations

Variables	KRAS driver mutations	Non-KRAS driver mutations	Non-driver mutations	P-value
Number	28	19	120
Tumor size, mm (median [IQR])	30.50 [20.00, 39.25]	32.00 [23.50, 36.00]	26.50 [19.75, 39.25]	0.695
pT stage (%)
T1	14 (50.0)	8 (42.1)	70 (58.3)	0.185
T2	10 (35.7)	10 (52.6)	38 (31.7)
T3	2 (7.1)	1 (5.3)	11 (9.2)
T4	2 (7.1)	0 (0.0)	1 (0.8)
pN stage (%)
N0	19 (67.9)	10 (52.6)	66 (55.0)	0.587
N1	2 (7.1)	4 (21.1)	16 (13.3)
N2	7 (25.0)	5 (26.3)	38 (31.7)
Histology (%)
LUAD	22 (78.6)	13 (68.4)	43 (35.8)	< 0.001
LUSC	2 (7.1)	3 (15.8)	64 (53.3)
Others	4 (14.3)	3 (15.8)	13 (10.8)
MPR (%)	17 (60.7)	6 (31.6)	58 (48.3)	0.146
pCR (%)	10 (35.7)	2 (10.5)	27 (22.5)	0.123

IQR interquartile range; LUAD lung adenocarcinoma; LUSC lung squamous cell carcinoma; MPR major pathological response; pCR pathological complete response

Long-term outcomes

To investigate whether the driver mutation status has a long-term impact on patients who received neoadjuvant immunotherapy, a survival analysis based on DFS and OS was performed. In the entire cohort, the 1-year recurrence rate and 1-year survival rate were 88.2% and 96.1%, respectively, with a median follow-up time of 13 months (IQR, 7–18 months). The KRAS driver mutation group was associated with the lowest rate of recurrence events and deaths among the three groups (Fig. 3A, B). Given the potential survival benefit of KRAS mutation group compared with non-driver mutation group, a subgroup analysis was performed between the two groups. Patients in the KRAS driver mutation group had a significantly higher DFS rate compared with patients in non-driver mutation group (P = 0.030) and a higher OS rate that approached significance (P = 0.092). On multivariate Cox regression analysis predicting DFS, KRAS driver mutations were found to be an independent favorable prognostic factor (HR: 0.10, CI: 0.01–0.82, P = 0.032) (Supplementary Table 5).

Fig. 3.

Survival analysis of the entire cohort. A Kaplan–Meier curves showing disease-free survival (DFS) of all patients stratified by driver oncogene status. B Kaplan–Meier curves showing overall survival (OS) of all patients stratified by driver oncogene status. C Comparison of DFS between the KRAS driver mutation group and the non-KRAS driver mutation group using Kaplan–Meier analysis. D Comparison of OS between the KRAS driver mutation group and the non-KRAS driver mutation group using Kaplan–Meier analysis

Transcriptomic analysis

Recent studies have emphasized that PD-L1 expression, TMB, immunogenicity, and CD8⁺ T cell tumor-infiltrating lymphocytes (TILs) are associated with the response to PD-1 blockade therapy in NSCLC patients [20]. A large pooled analysis has confirmed that PD-L1 expression and TMB were higher in KRAS driver mutation tumors than in KRAS wild-type tumors [21]. To further explore the differences in pathway enrichment before and after neoadjuvant immunotherapy based on the KRAS driver mutation status and investigate whether KRAS driver mutation NSCLC has a higher rate of CD8⁺ TILs, we performed transcriptomic analysis using RNA-seq data.

Considering the KRAS driver mutation group exhibits a higher survival benefit than the non-driver mutation group, and to eliminate potential confounding factors and account for the heterogenicity and small sample size within the non-KRAS driver mutation group, GSEA was performed only between the KRAS driver mutation group and the non-driver mutation group. Through ranked GSEA, we found that the top gene sets at baseline that showed a significant association with the KRAS mutations were involved in the antigen receptor mediated signaling pathway, adaptive immune response, and T cell activation pathways (Fig. 4A, B). This suggests that KRAS driver mutations were associated with higher immunogenicity, which are more likely to respond to immunotherapy. In contrast, the non-driver mutation group was associated with cell fate specification, DNA binding transcription activator activity, and DNA templated DNA replication. Furthermore, GSEA analysis of post-treatment tumor samples revealed immune-activated pathways including activation of immune response, positive regulation of immune response, regulation of T cell activation, which were enriched in KRAS driver mutation group, while DNA replication was still observed in non-driver mutation group (Fig. 4C, D).

Fig. 4.

Overview of gene set enrichment analysis (GSEA) between the KRAS driver mutation group and non-driver mutation group before and after neoadjuvant immunotherapy. A Bar plot illustrates the significant pathways enriched in the KRAS driver mutation group and non-driver mutation group before immunotherapy. B GSEA of baseline samples comparing KRAS driver mutation group and non-driver mutation group. C Bar plot depicts the significant pathways between KRAS driver mutation group and non-driver mutation group after immunotherapy. D GSEA of post-immunotherapy samples comparing KRAS driver mutation group and non-driver mutation group. The NES and FDR q-value are presented for each comparison. NES, normalized enrichment score; FDR, false discovery rate

To reveal the immune cell infiltration characteristics in the immune tumor microenvironment (TIME) between the KRAS driver mutation group and the non-driver mutation group, we computed 22 types of immune cell infiltration in different subgroups using the CIBERSORT algorithm (Fig. 5A). The median proportion of CD8⁺ T cell was significantly higher in KRAS driver mutation group at baseline (18% vs. 13%, P = 0.030). After treatment, the proportion of M2 macrophage increased in non-driver mutation group, but not in KRAS driver mutation group (16% vs. 26%, P = 0.044, Fig. 5B) while the proportion of M1 macrophage did not show difference both in KRAS driver mutation group and non-driver mutation group (P = 0.360 and P = 0.280, respectively, Fig. 5C). Notably, the M2 macrophages with high expression of CD163, CD204 and CD206 are induced by IL-4 and IL-13, which plays a key role in exerting immunomodulatory effects [22]. M2 macrophages could thus create a favorable immunosuppressive TME for tumor progression, which in turn could influence immunotherapy resistance and be associated with poor survival outcomes [23, 24]. However, the molecular biology mechanisms of macrophages in immunotherapy remain to be further explored.

Fig. 5.

Analysis of immune cell infiltration and characteristics reveal differences in the tumor immune microenvironment between KRAS driver mutation and non-driver mutation groups before and after immunotherapy. A The differences in immune cell infiltration between the KRAS driver mutation group and the non-driver mutation group before immunotherapy using CIBERSORT. B Box plot shows an increase in the proportion of M2 macrophages after immunotherapy in the non-driver mutation group. C Box plot displays the distribution of the proportion of M1 macrophages in the KRAS driver mutation group and the non-driver mutation group before and after immunotherapy

Discussion

The utilization of neoadjuvant chemotherapy in combination with immunotherapy, has become the prevailing standard of care for patients afflicted with resectable or locally advanced non-small cell lung cancer [8]. Nonetheless, the efficacy of neoadjuvant treatment may vary substantially due to the heterogeneity of NSCLC, particularly in terms of oncogenic driver mutations [14]. In this study, we observed that patients with KRAS driver mutations exhibited a higher rate of MPR compared to non-KRAS driver mutation group. Furthermore, the status of KRAS driver mutations has been identified as an independent favorable prognostic factor for DFS among the wider population of resectable NSCLC patients.

In the era of immuno-oncology, emerging evidence suggests that driver oncogenes have different effects on the TIME that influence the potential for clinical benefit from treatment with ICIs [14]. KRAS mutation is one of the major oncogenic driver mutations in NSCLCs, the rate of KRAS mutation in NSCLC patients was 27% in Caucasians [25], but lower in Asians (10%) [26]. The presence of KRAS mutation is prognostic of poor survival for patients with NSCLC when compared to the absence of KRAS mutation in early research [27]. Nevertheless, the efficacy of anti-PD-1/PD-L1 immunotherapy in KRAS-mutated NSCLC is still not well defined [15, 21, 28–31]. Several previous studies found KRAS mutation was associated with superior response in advanced NSCLC patients who received immunotherapy [15, 21], while other studies found KRAS mutation was not associated with prolonged survival even found KRAS was an independent adverse prognostic factor in immunotherapy [28–30]. These heterogeneous results may relate to the different stages of NSCLCs and different regime of ICIs. A recent systematic review demonstrated that the benefits of immunotherapy in early resectable lung cancer appeared to be better than those observed in advanced lung cancer, especially with the regard to the regimen of immunotherapy in combination with chemotherapy [32]. Nowadays, the indication of immunotherapy is becoming more and more widespread, and the combination of ICIs with chemotherapy has also become mainstream in the adjuvant immunotherapy. Thus, this study was designed to focus on the effect of oncogenic driver mutations on resectable NSCLCs.

Several studies have demonstrated that KRAS-mutant NSCLCs are correlated with an increasing proportion of PD-L1 expression, TMB and CD8⁺ TILs [21, 33]. According to further transcriptomic analysis of our dataset, the proportion of CD8⁺ TILs was significantly increased in the KRAS mutated NSCLCs, which is consistent with the previous study [21]. Additionally, KRAS mutation has been reported to promote inflammatory signaling in the early stages of tumor progression, and can cooperate with external factors such as tobacco or pollutants in the lung to create an inflammatory environment that enhances oncogenic signaling [34]. The higher proportion of current smokers in KRAS driver mutation group was also observed in this study. Furthermore, several immune-related pathways were observed in KRAS driver mutation NSCLCs and further activated after immunotherapy. This finding suggests that KRAS driver mutation is associated with higher immunogenicity and may facilitate a more favorable response to immunotherapy in resectable NSCLCs.

In a phase 1 trial involving 59 patients with metastatic non-small cell lung cancer, Sotorasib, a small molecule targeting KRAS^G12C, demonstrated promising anticancer activity and safety [35]. However, for KRAS-mutated NSCLC, chemotherapy with or without immunotherapy remained the standard care of neoadjuvant schema for the moment [8]. In contrast, NSCLCs with other oncogenic driver mutations have a more diverse array of targeted drugs available. Although EGFR mutations are often contraindicated for immunotherapy, eight patients with EGFR mutations in this cohort still received neoadjuvant immunotherapy due to insufficient tissue samples for genetic testing or false negative biopsy results, while further genetic testing of surgical specimens revealed the presence of EGFR mutations. Unsurprisingly, only one patient (12.5%) with EGFR L858R mutation achieved MPR after neoadjuvant immunotherapy, consistent with the research in the early exploratory phase, indicating that patients with EGFR mutations show limited responsiveness to ICIs [16, 36, 37]. Nevertheless, the EGFR-TKIs have exhibited satisfying efficacy and acceptable safety profile in neoadjuvant therapy [38, 39]. While ALK and ROS1 rearrangements are commonly linked with lower TMB [40], their response to ICIs is very limited according to several previous studies [15, 37]. Conversely, for ALK-positive NSCLC, several targeted treatments including crizotinib, alectinib have been approved and showed superior response than chemotherapy [41, 42]. Other driver mutations consisted of 21% oncogenic driver mutations in our dataset, the efficacy of ICIs in patients with those oncogenic driver mutations still require further exploration. The ongoing NAUTIKA1 study (NCT04302025) focuses on the effect of neoadjuvant targeted therapy, including alectinib, entrectinib, vemurafenib, cobimetinib, or pralsetinib, on untreated patients with resectable stage IB-IIIB NSCLC diagnosed with ALK, ROS1, NTRK, BRAF, or RET mutations. Other ongoing trials such as ALNEO (NCT05015010) and NeoADAURA (NCT04351555) trials are evaluating neoadjuvant targeted therapies regarding specific molecular genotyping.

It is crucial to acknowledge that the impact of oncogenic driver mutations in NSCLCs is not solely determined by the oncogene itself, but also by the concurrent genomic mutations that influence the molecular and clinical heterogeneity of NSCLC tumors [43]. In particular, co-occurring mutations in suppressor genes, such as TP53 and STK11, in KRAS-mutant NSCLCs have demonstrated significant implications for treatment modalities including immunotherapy [44]. A recent study revealed that NSCLC patients with KRAS mutations and concomitant STK11 and/or KEAP1 mutations had poorer prognoses. These patients may derive greater benefits from a combination of chemotherapy, atezolizumab, and bevacizumab (ABCP), as opposed to chemotherapy combined with atezolizumab/bevacizumab (ACP or BCP) [45]. However, the lack of available co-mutation information in this study is attributed to the fact that whole exome sequencing was only performed in a minority of patients. Therefore, further investigations focusing on co-mutations associated with driver oncogenes are still required.

Despite the advantages of our study, its retrospective nature has imposed certain limitations. Firstly, the sample size was relatively small and the follow-up period was not long enough. Nonetheless, given that the efficacy of neoadjuvant immunotherapy for resectable and locally advanced NSCLCs was still in its exploratory phase, our use of pathological response information allowed us to demonstrate the efficacy of immunotherapy more precisely and intuitively. Secondly, the exploratory of response of non-KRAS driver mutations to immunotherapy is still insufficient. Owing to the limitations of test kits and sample size, this study focused solely on the seven major oncogenic driver mutations. Other oncogenic driver mutations, including NTRK, NRG1, HRAS, NRAS, MEK1 need further study to provide a more comprehensive analysis of their response to immunotherapy. Thirdly, whole exome sequencing and detailed information regarding KRAS-mutant subtypes were only available in a minority of patients, which constrained our ability to conduct comprehensive investigations in these domains. Finally, less than half of the patients had pre-treatment PD-L1 expression results, limiting our research on the correlation between driver oncogene status and PD-L1 expression.

In summary, oncogenic driver mutations exert various impacts on the TIME of NSCLCs, thereby influencing the clinical benefit of ICIs. Our study found that patients with KRAS mutations were not only associated with a better pathological response to neoadjuvant immunotherapy in terms of oncogenic driver mutations but also correlated with better prognoses among resectable NSCLC patients. Increased tumor immunogenicity and a higher proportion of infiltrating CD8⁺ TILs were associated with KRAS driver mutations, which may explain the better response in immunotherapy. Consequently, molecular testing is necessary for patients scheduled to receive neoadjuvant treatment, and further prospective studies are still needed in the future.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Concept and design were contributed by ZS, LZ and PZ; Acquisition of data was contributed by ZS, MT, DB, JZ, XZ, YQ, SH, YC, WY and HY; Analysis and interpretation of clinical data and RNA sequencing data were contributed by ZS, MT and LH; Drafting of the manuscript was contributed by ZS and LZ. Critical revision was contributed by ZS, MT, LH, LZ and PZ. Funding acquisition was contributed by ZS and PZ. All authors contributed to the article and approved the submitted version.

Funding

This research was supported by the National Natural Science Foundation of China (Grant no. 82125001), the Innovation Program of Shanghai Municipal Education Commission (Grant no. 2023ZKZD33), Clinical Research foundation of Shanghai Pulmonary Hospital (Grant no. FKLY20004), and Science and Technology Commission of Shanghai Municipality (Grant no. 23YF1435200).

Data availability

RNA sequencing data were deposited in the Genome Sequence Archive database under accession number HRA002071. Data and codes utilized in this study are immediately available from the corresponding author upon reasonable request.

Declarations

Conflict of interest

The authors have declared that no conflict of interest exists.