Lymph node (LN) metastasis is the primary and most common route of metastatic spread in patients with lung cancer, occurring in 30% to 40% of cases [1]. Mediastinal LN dissection (MLND) plays a crucial role in accurate LN staging, which has prognostic and therapeutic implications for patients diagnosed with non-small cell lung cancer (NSCLC) [2,3]. However, the optimal extent of MLND remains unclear [4-7]. Guidelines from the European Society of Thoracic Surgeons [8] and the International Association for the Study of Lung Cancer (IASLC) [9] recommend including at least 3 mediastinal LN stations in the dissection. However, the IASLC has also indicated that station 4L (S4L) LN dissection (S4L-LND) is not necessary for patients with left-sided tumors [9]. Furthermore, the American College of Surgeons Oncology Group Z0030 trial concluded that although comprehensive MLND for left-sided NSCLC should involve the excision of stations 5–9, the removal of S4L-LN is not obligatory [5].

In addition to unclear guidelines, risks associated with S4L-LND have been reported. These include potential damage to surrounding structures, such as the left recurrent laryngeal nerve or the thoracic duct [10]. Consequently, the decision to perform S4L-LND is influenced in part by the preference and expertise of the surgeon. The present study was designed to evaluate the clinical impact of S4L-LND on survival and to explore the potential benefits of incorporating S4L-LND into the treatment of NSCLC.

Patients

A prospectively maintained lung cancer database at our institution in Seoul, South Korea, was searched to identify all patients who underwent surgery for primary left lung cancer between January 2007 and December 2021. Their clinical records were then retrospectively reviewed. During this period, a total of 3,852 patients were identified as having undergone surgery with curative intent for primary left-sided lung cancer, with procedures specifically involving lobectomy or segmentectomy.

Patients were excluded from the study if they (1) were diagnosed with small-cell lung cancer (n=70), (2) had received neoadjuvant therapy (n=89), (3) underwent surgical procedures other than video-assisted thoracoscopic surgery (n=798), (4) had received treatment for lung cancer in the previous 2 years (n=24), (5) were diagnosed with double primary lung cancers (n=247), (6) had distant metastases (n=3), (7) exhibited adenocarcinoma in situ (n=6), or (8) underwent resection other than R0 (n=45). Ultimately, 2,601 patients were enrolled in the study (Fig. 1). The study protocol received approval from the institutional review board of Asan Medical Center (2023-1118). Due to the retrospective nature of the research, the requirement for informed consent was waived.

Figure 1. Patient flow chart. SCLC, small-cell lung cancer; VATS, video-assisted thoracoscopic surgery.

Preoperative and postoperative management

The diagnostic, staging, and surgical resection procedures adhered to well-established and widely accepted protocols, as detailed in the Supplementary Material (Supplementary Fig. 1) [11].

For LNs with suspicious characteristics, if computed tomography (CT) or positron emission tomography results indicated clinical N2 (cN2) disease, mediastinal LN biopsy was performed. In patients with biopsy-confirmed cN2 disease, treatment strategy was determined by a multidisciplinary team including specialists from medical oncology, radiology, and thoracic surgery.

Although cN2 disease typically does not require immediate surgical intervention, our institution has developed surgical approaches for this condition and evaluated their outcomes in various studies. The criteria for considering upfront surgical intervention in patients with cN2 disease are as follows: (1) the primary tumor can be resected without the need for pneumonectomy; (2) mediastinal LN metastasis is confined to a single zone; (3) the involved LN is clearly demarcated from surrounding tissues and is less than 3.0 cm in diameter; and (4) CT or positron emission tomography reveals no evidence of extranodal tumor invasion, such as full-thickness LN capsular invasion or extension of tumor cells beyond the LN capsule, which would suggest external tumor infiltration into the LN. Regarding (2), the specific zones for mediastinal LN metastasis are defined as follows: the upper zone includes LNs at levels 2–4, the aortopulmonary zone includes LNs at levels 5 and 6, the subcarinal zone involves level 7 LNs, and the lower zone comprises LNs at levels 8 and 9 [11].

Although our institution has not yet established a protocol detailing indications for S4L-LND, the procedure was performed when S4L-LN metastasis was suspected based on radiological evidence. In other cases, the decision to perform S4L-LND was left to the discretion of the surgeon. One surgeon, who was proficient in mediastinoscopic procedures, adapted the video-assisted mediastinoscopic lymphadenectomy technique for use in patients with left-sided NSCLC [12]. Other surgeons generally performed S4L-LND when the cancer was primarily located in the left upper lobe (LUL) and had a stage above T2 or N1.

Follow-up

All patients adhered to a standardized follow-up regimen. Comprehensive data were systematically collected through clinical assessments every 6 months for the first 2 years postoperatively and annually thereafter. Chest CT scans were performed at each clinical visit or when disease recurrence was suspected. Treatment modalities and chemotherapy regimens for patients experiencing relapse were determined at the discretion of the attending physician.

Definitions

Overall survival (OS) was defined as the time from the date of surgical intervention to the date of death from any cause, as determined by an exhaustive review of records in the Korean National Security Death Index Database. Recurrence-free survival (RFS) was defined as the period from the date of surgery to the first evidence of recurrence. In the absence of recurrence, patient data were censored at the last confirmed recurrence-free time point. Postoperative complications that occurred within 1 month after surgery included pneumonia, air leak lasting more than 7 days, chylothorax, vocal cord paralysis, and postoperative hemorrhage. The rates of postoperative complications were systematically ascertained through a retrospective review of medical records.

Statistical analysis

Continuous variables were reported as means with standard deviations, while categorical variables were reported as frequencies and percentages. The normality of the distribution for individual parameters was assessed using the Shapiro-Wilk test. For continuous variables that were normally distributed, the Student t-test was used for comparisons. In contrast, the Wilcoxon rank-sum test was employed for continuous variables that were not normally distributed. Categorical variables were compared using the chi-square test or the Fisher exact test, as appropriate. OS and RFS were estimated using the Kaplan-Meier method and compared using the log-rank test.

To mitigate the potential impacts of confounding variables, 1:1 propensity score matching (PSM) was performed between patients who underwent S4L-LND and those who did not [13]. PSM was conducted based on patient age, sex, smoking history, comorbidities (hypertension, diabetes, coronary artery disease, chronic obstructive pulmonary disease, hepatic disease, and renal failure), history of pulmonary tuberculosis, pulmonary function, tumor location, surgical extent, histologic type, tumor size, American Joint Committee on Cancer (AJCC) eighth edition clinical tumor (cT) and node (cN) stages, lymphovascular invasion, and receipt of adjuvant chemotherapy. The matching process utilized a greedy nearest-neighbor matching algorithm with the incorporation of caliper restrictions.

All statistical computations were performed using R ver. 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria). p-values less than 0.05 were considered to indicate statistical significance.

Baseline characteristics

This study enrolled 2,601 patients, of whom 1,126 underwent S4L-LND and 1,475 did not. Analysis of their baseline demographic and tumor characteristics revealed significant differences between groups in terms of tumor location (p<0.001), surgical extent (p<0.001), AJCC eighth edition clinical (p<0.001) and pathological (p<0.001) tumor-node- metastasis stages, adjuvant chemotherapy (p=0.008), cT stage (p=0.014), cN stage (p<0.001), and pathological N stage (p=0.006) (Table 1). To address these disparities in baseline characteristics, the patients underwent 1:1 PSM, producing 1,036 matched pairs. The baseline characteristics of these pairs were found to be comparable.

Table 1 . Baseline patient data before and after propensity score matching.

Characteristic	Overall cohort (N=2,601)				Propensity score matching (N=2,072)			SMD
	4LND- (n=1,475)	4LND+ (n=1,126)	p-value		4LND- (n=1,036)	4LND+ (n=1,036)	p-value
Age (yr)			0.909				0.724	0.017
<65	797 (54.0)	605 (53.7)			554 (53.5)	563 (54.3)
≥65	678 (46.0)	521 (46.3)			482 (46.5)	473 (45.7)
Sex			0.976				0.628	0.023
Male	788 (53.4)	600 (53.3)			541 (52.2)	553 (53.4)
Female	687 (46.6)	526 (46.7)			495 (47.8)	483 (46.6)
History of smoking	684 (46.4)	519 (46.1)	0.918		458 (44.2)	475 (45.8)	0.480	0.033
No. of comorbidities per patient			0.344				0.976	0.018
0	620 (42.0)	493 (43.8)			456 (44.0)	457 (44.1)
1	517 (35.1)	406 (36.1)			375 (36.2)	374 (36.1)
2	263 (17.8)	171 (15.2)			151 (14.6)	155 (15.0)
≥3	75 (5.1)	56 (5.0)			54 (5.2)	50 (4.8)
History of pulmonary tuberculosis	130 (8.8)	108 (9.6)	0.540		98 (9.5)	95 (9.2)	0.880	0.010
Pulmonary function
FEV1 (%)	89.6±14.2	90.0±14.6	0.498		89.9±14.0	90.0±14.6	0.874	0.007
DLCO (%)	86.2±16.0	87.1±15.8	0.154		87.2±15.6	87.1±15.8	0.910	0.005
Tumor location			<0.001				0.855	0.010
Left upper lobe	822 (55.7)	730 (64.8)			663 (64.0)	658 (63.5)
Left lower lobe	653 (44.3)	396 (35.2)			373 (36.0)	378 (36.5)
Surgical extent			<0.001				0.826	0.012
Segmentectomy	382 (25.9)	219 (19.4)			205 (19.8)	210 (20.3)
Lobectomy	1,093 (74.1)	907 (80.6)			831 (80.2)	826 (79.7)
Histology			0.589				0.803	0.018
Adenocarcinoma	1,246 (84.5)	956 (84.9)			883 (85.2)	888 (85.7)
Squamous	178 (12.1)	139 (12.3)			127 (12.3)	119 (11.5)
Others	51 (3.5)	31 (2.8)			26 (2.5)	29 (2.8)
Clinical stage (eighth AJCC)			<0.001				0.331	0.039
I	1,236 (83.8)	892 (79.2)			873 (84.3)	863 (83.3)
II	179 (12.1)	149 (13.2)			123 (11.9)	119 (11.5)
III	60 (4.1)	85 (7.5)			40 (3.9)	54 (5.2)
cT stage			0.014				1.000	0.002
T1	1,021 (69.2)	727 (64.6)			695 (67.1)	696 (67.2)
T2–4	454 (30.8)	399 (35.4)			341 (32.9)	340 (32.8)
cN stage			<0.001				0.529	0.022
N0	1,418 (96.1)	1,007 (89.4)			992 (95.8)	985 (95.1)
N1–2	57 (3.9)	119 (10.6)			44 (4.2)	51 (4.9)
Adjuvant chemotherapy	244 (16.5)	233 (20.7)	0.008		183 (17.7)	198 (19.1)	0.427	0.036
Adjuvant radiotherapy	55 (3.7)	59 (5.2)	0.077		50 (4.8)	42 (4.1)	0.455	0.035
LVI	383 (26.0)	304 (27.0)	0.585		265 (25.6)	264 (25.5)	1.000	0.002
Pathological stage (eighth AJCC)			<0.001				0.843	0.017
I	1,093 (74.1)	760 (67.5)			747 (72.1)	736 (71.0)
II	221 (15.0)	195 (17.3)			162 (15.6)	171 (16.5)
III	161 (10.9)	171 (15.2)			127 (12.3)	129 (12.5)
Pathological tumor size (mm)	26.3±13.4	27.2±13.5	0.090		26.6±12.4	26.8±13.4	0.747	0.014
pN stage			0.006				0.647	0.021
N0	1,233 (83.6)	990 (79.0)			855 (82.5)	846 (81.7)
N1–2	242 (16.4)	236 (21.0)			181 (17.5)	190 (18.3)

Values are presented as number (%) or mean±standard deviation unless otherwise stated..

4LND-, patients without left lower paratracheal node resection; 4LND+, patients with left lower paratracheal node resection; SMD, standardized mean difference; FEV1, forced expiratory volume during the first second; DLCO, diffusing capacity of carbon monoxide, AJCC, American Joint Committee on Cancer; cT stage, clinical T stage; cN stage, clinical N stage; LVI, lymphovascular invasion; pN stage, pathological N stage..

Perioperative information associated with harvested LNs

Supplementary Fig. 2 illustrates the probability of LN metastasis at each LN station in patients with lung cancer in the LUL and the left lower lobe (LLL). The incidence of metastasis to the S4L-LN was 9.3% for LUL lung cancer—representing the highest rate of mediastinal LN metastasis for the LUL—compared to 4.8% for LLL lung cancer. Station 5 exhibited the second-highest rate of mediastinal LN metastasis for the LUL at 7.2%, while station 7 displayed the highest rate for the LLL at 7.7%.

Risk factors for S4L-LN metastasis

Of the 1,126 patients who underwent S4L-LND, 87 (7.7%) tested positive for S4L-LN involvement. Logistic regression analyses were employed to determine the associations between clinicopathological features and S4L-LN involvement (Table 2). Univariate or multivariate analyses were performed using baseline characteristics from Table 1, excluding postoperative variables, to identify factors that could be determined prior to surgery. After excluding correlated variables, variables with a p-value of less than 0.10 in univariate analysis were included in the initial multivariable risk model. The final multivariable model was chosen through stepwise selection, with the Akaike information criterion used for comparison; the lowest Akaike information criterion was considered to indicate the best fit.

Table 2 . Logistic regression analysis of risk factors associated with station 4L nodal involvement.

Variable	Univariate			Multivariate
	OR (95% CI)	p-value		OR (95% CI)	p-value
Age (yr)		0.078
<65	-
≥65	0.672 (0.431–1.046)	0.078
Sex		0.152
Female	-
Male	0.726 (0.469–1.125)	0.152
History of smoking		0.209
No	-
Yes	1.319 (0.858–2.019)	0.209
Tuberculosis		0.695
No	-
Yes	1.151 (0.570–2.326)	0.695
cT stage		<0.001
T1	-
T2–4	2.373 (1.545–3.645)	<0.001
cN stage		<0.001			<0.001
N0	-			-
N1–2	14.097 (8.933–22.248)	<0.001		12.404 (7.556–20.364)	<0.001
Tumor location		<0.001			<0.001
Left lower lobe	-			-
Left upper lobe	2.484 (1.485–4.156)	<0.001		2.837 (1.659–4.852)	<0.001
Surgical extent		0.021
Segmentectomy	-
Lobectomy	0.472 (0.249–0.894)	0.021
Histology
Others	-	-
Adenocarcinoma	1.450 (0.350–6.000)	0.609
Squamous	1.036 (0.216–4.972)	0.965

OR, odds ratio; CI, confidence interval; cT stage, clinical T stage; cN stage, clinical N stage..

Univariate analyses indicated that advanced cT stage (p<0.001), advanced cN stage (p<0.001), tumor location (p<0.001), and extent of surgery (p=0.021) were predictive of station 4L LN involvement. Multivariate logistic regression analysis revealed that cN stage (p<0.001) and tumor location (p<0.001) were independently associated with the risk of S4L-LN involvement.

Table 3 summarizes the associations of S4L-LN metastasis with tumor location and LN stage in the 2 matched groups. In the S4L-LND group, regarding tumor location, the LUL was associated with a higher rate of 4L LN metastasis than the LLL, although this difference was not statistically significant (3.6% versus 2.0%, p=0.061). In contrast, the rate of 4L LN metastasis increased significantly with advancing cN stage (cN0, 2.3%; cN1, 5.8%; cN2, 32.6%; p<0.001).

Table 3 . Distribution of 4LN metastasis according to tumor location and cN stage.

	Tumor location				cN stage				Overall
	LUL (n=1,321)	LLL (n=751)	p-value		cN0 (n=1,904)	cN1 (n=77)	cN2 (n=56)	p-value
S4L-LN examined	658 (49.8)	378 (50.3)	0.855		985 (49.8)	22 (42.3)	29 (67.4)	0.039	1,036 (50.0)
% positivea) (metastasis)	47 (3.6)	15 (2.0)	0.061		45 (2.3)	3 (5.8)	14 (32.6)	<0.001	62 (3.0)

Values are presented as number (%)..

LUL, left upper lobe; LLL, left lower lobe; S4L-LN, station 4L lymph node..

^a)% positive if lymph node station was examined..

Survival analyses

At the completion of this study, 376 patients (14.4%) had died, and 495 patients (19.0%) had experienced recurrence during the follow-up period. Prior to PSM, OS (p=0.0022) (Fig. 2A) and RFS (p=0.033) (Fig. 2B) rates were significantly higher in patients who did not undergo S4L-LND compared to those who did. However, after PSM, no significant differences were found in OS (p=0.12) (Fig. 2C) or RFS (p=0.24) (Fig. 2D) between these groups.

Figure 2. Kaplan-Meier curves in the patients without left lower paratracheal node resection (4LD-) and patients with left lower paratracheal node resection (4LD+) groups for overall survival (OS) before matching (A); recurrence-free survival (RFS) before matching (B); OS after matching (C); and RFS after matching (D).

Subgroup analysis after PSM, stratified by tumor location, revealed that OS rates in patients with LUL cancer did not differ significantly based on whether patients underwent S4L-LND (p=0.7) (Fig. 3A). However, OS rates were significantly lower in patients with LLL cancer who underwent S4L-LND compared to those who did not (p=0.0026) (Fig. 3B). The use of S4L-LND was not significantly associated with OS in patients stratified by cN stage (cN0, p=0.28; cN1–2, p=0.13) (Fig. 3C, D). Furthermore, RFS rates did not differ significantly between patients who did and did not undergo S4L-LND for either LUL (p=0.26) (Supplementary Fig. 3A) or LLL (p=0.63) (Supplementary Fig. 3B) cancers. Similarly, no significant differences in RFS were observed upon stratification by cN stage (cN0, p=0.39; cN1–2, p=0.19) (Supplementary Fig. 3C, D). Additionally, OS rates did not differ significantly in patients stratified by cT stage (cT1, p=0.55; cT2–4, p=0.13) (Supplementary Fig. 4A, B) or by the extent of surgery (lobectomy, p=0.17; segmentectomy, p=0.5) (Supplementary Fig. 4C, D). All figures for RFS are presented within the Supplementary Material.

Figure 3. Overall survival (OS) curves following the matching of patients without left lower paratracheal node resection (4LD-) and patients with left lower paratracheal node resection (4LD+) groups for left upper lobe (LUL) (A), left lower lobe (LLL) (B), clinical N0 (C), and clinical N1–2 cancer (D).

The associations between S4L-LN metastasis and various risk factors were assessed with univariate and multivariate logistic regression analyses.

Postoperative complications in the matched patient population

Table 4 summarizes the postoperative complications experienced by patients in the 2 matched groups. During the follow-up period, the rate of postoperative complications was significantly higher in patients who underwent S4L-LND compared to those who did not (7.0% versus 3.9%, p=0.002). Furthermore, the specific rates of prolonged air leak (1.9% versus 0.8%, p=0.036) and vocal cord palsy (3.4% versus 1.1%, p=0.001) were also significantly higher in the S4L-LND group.

Table 4 . Comparison of postoperative complications.

	No S4L-LND (n=1,036)	S4L-LND (n=1,036)	p-value
Patients with ≥1 complication	40 (3.9)	73 (7.0)	0.002
Pneumonia	8 (0.8)	9 (0.9)	1.000
Air leak >7 days	8 (0.8)	20 (1.9)	0.036
Chylothorax	6 (0.6)	8 (0.8)	0.789
Vocal cord palsy	11 (1.1)	35 (3.4)	0.001
Hemorrhage	8 (0.8)	6 (0.6)	0.789

Values are presented as number (%)..

S4L-LND, station 4L lymph node dissection..

Determining the optimal extent of MLND requires a careful balance between its surgical risks and oncologic benefits. Station 4L is a critical part of the left mediastinal LN drainage system, indicating that it may play a pivotal role in left mediastinal LN metastasis. Previous studies have assessed the clinical relevance of S4L-LND in patients with left-sided NSCLC by comparing their survival rates to those of patients who did not undergo the procedure [14-16]. Furthermore, research has demonstrated the importance of S4L-LND in patients with left-sided NSCLC [17]. S4L-LND not only facilitates the therapeutic removal of potentially metastatic or micrometastatic nodes but also enables more accurate LN upstaging, which can be followed by adjuvant chemoradiotherapy [17]. Among patients with N2 lesions, the 5-year OS rate has been reported to be lower in patients with left-sided lung cancer compared to those with right-sided lesions, possibly due to anatomical factors that result in inadequate LN dissection [18]. The therapeutic index for the expected benefit of S4L-LND has been calculated as 1.93 in patients with primary left upper division NSCLC [19]. In contrast, a separate study reported that routine S4L-LND did not improve survival; furthermore, complication rates were significantly higher in patients who underwent S4L-LND compared to those who did not, likely due to the anatomical complexity of the dissection [20]. From a surgical perspective, performing comprehensive MLND in patients with left-sided lung cancer is technically demanding due to anatomical constraints, such as the proximity of the aorta, the pulmonary artery, and the recurrent laryngeal nerve. In part, the risk of damaging these surrounding structures explains why S4L-LNs may not be the appropriate approach for all patients.

The American College of Chest Physicians recommends invasive mediastinal staging for certain patients, including those with cN1 disease [21,22]. In the present study, multivariate logistic regression analysis revealed tumor location and cN stage as significant risk factors for 4L LN metastasis. Moreover, the rate of this metastasis increased significantly with more advanced cN stage (cN0, 2.3%; cN1, 5.8%; cN2, 32.6%; p<0.001). The rate of 4L LN metastasis was also higher in the LUL than in the LLL group (3.6% versus 2.0%, p=0.061). When evaluating the rate of metastasis by LN station, the S4L-LN was found to face the highest risk of mediastinal metastasis in patients with LUL lung cancer (9.3%). Accordingly, selective S4L-LND should be considered in patients with cancer affecting the LUL or at an advanced cN stage.

Following PSM, OS and RFS rates did not differ significantly between patients who underwent S4L-LND and those who did not. However, prior to PSM, the OS rate was lower among patients who underwent S4L-LND. The involvement of the S4L-LN may indicate an advanced stage of cancer, and these survival results may thus reflect the more frequent application of S4L-LND in patients with clinically suspected advanced-stage cancer. Subgroup analyses of PSM patients, stratified by the risk of S4L-LN involvement and including factors such as tumor location and cN stage, revealed that the inclusion of S4L-LND was not associated with a survival benefit. Although S4L-LND has been recommended for patients with larger tumors (>3 cm) [21], the present study indicates that S4L-LND does not significantly impact OS in patients with tumors at an advanced cT stage (p=0.13).

Complete dissection of the S4L-LN may lead to an increased rate of surgical complications, particularly hoarseness resulting from injury to the left recurrent laryngeal nerve and chylothorax due to damage to the lymphatic system [10]. Although the frequency of chylothorax did not differ significantly between patients who underwent S4L-LND and those who did not (p=0.789), the former group exhibited a significantly higher risk of vocal cord palsy (3.4% versus 1.1%, p=0.001).

The present study had several limitations. First, it was a retrospective analysis of patients treated at a single center, which introduces inherent constraints even with the prospective collection of data. Furthermore, the decision to conduct S4L-LND was left to the surgeon’s discretion, potentially introducing selection bias even after PSM. Additionally, the proportion of patients who underwent preoperative mediastinal lymph node biopsy was relatively low. To validate these results, randomized multicenter clinical trials with larger samples are necessary.

In conclusion, the present study evaluated factors associated with the need for S4L-LND, including clinical stage and tumor location. The rate of 4L LN metastasis was found to be significantly associated with both cN stage and tumor location, suggesting that patients with advanced cN disease or LUL cancer could benefit from S4L-LND. However, S4L-LND did not significantly influence OS. These results imply that while routine S4L-LND may not be warranted for all patients with left-sided NSCLC, it could be advantageous for a subset of this population, particularly those with advanced cN stage or LUL lung cancer.

Author contributions

Conceptualization: JKY. Data curation: JKY, HKP. Formal analysis: JKY, HKP. Investigation: JKY, HKP, YLK. Methodology: JKY. Project administration: JKY. Visualization: JKY, HKP. Writing–original draft: JKY, HKP, YLK Writing–review and editing: all authors. Final approval of the manuscript: all authors.

Conflict of interest

Activities not related to this article: K.J.N. is a cofounder and chief medical officer of Portrai. Except for that, no potential conflict of interest relevant to this article was reported.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Supplementary materials

Supplementary materials can be found via https://doi.org/10.5090/24.022. Supplementary Fig. 1. Patient flow chart. Supplementary Fig. 2. Metastasis rate of each lymph node station according to tumor location. Supplementary Fig. 3. Recurrence-free survival (RFS) curves after matching of 4LD- and 4LD+ groups in the left upper lobe (LUL) cancer (A), left lower lobe (LLL) cancer (B), clinical N0 (C), and clinical N1–2 (D). Supplementary Fig. 4. Overall survival (OS) curves after matching of 4LD- and 4LD+ groups in the cT1 (A), cT2–4 (B), lobectomy (C), and segmentectomy (D).