Long-Term Outcomes of Indocyanine Green Versus Inflation-Deflation for Intersegmental Plane Identification in Segmentectomy for Early-Stage Lung Cancer

Abstract

Objectives

This study aimed to evaluate whether the method of intersegmental plane (ISP) identification—indocyanine green (ICG) fluorescence vs the conventional inflation-deflation (ID) technique—affects resection margin adequacy, oncologic outcomes, or postoperative pulmonary function after anatomical segmentectomy for early-stage non–small-cell lung cancer (NSCLC).

Methods

We retrospectively analysed 410 patients who underwent segmentectomy for clinical N0M0 NSCLC between January 2018 and December 2023. The ISP was identified using the ID and ICG methods in 286 and 124 patients, respectively. Perioperative, pathologic, and survival outcomes were compared before and after inverse probability of treatment weighting (IPTW).

Results

Operative time and hospital stay were slightly shorter in the ICG than in the ID group (108.0 vs 115.0 min, P = .031; 4.0 vs 5.0 days, P = .003), whereas postoperative complication rates and pulmonary function changes were comparable. Central resection margins were longer in the ID group (2.5 vs 2.0 cm, P = .014), whereas lateral margin distances did not differ significantly. Recurrence-free and overall survival were similar between the 2 groups (log-rank P = .97 and P = .18). In IPTW-adjusted multivariable Cox analysis, the ISP identification method was not associated with recurrence or survival.

Conclusions

The ISP identification method was not associated with differences in functional or oncologic outcomes after anatomical segmentectomy. These findings suggest that the choice of technique may be guided by surgeon preference and technical considerations.

Lung cancer remains the leading cause of cancer-related mortality worldwide.

Graphical abstract

INTRODUCTION

Lung cancer remains the leading cause of cancer-related mortality worldwide. The widespread use of low-dose computed tomography (CT) screening has enabled the detection of smaller, early-stage tumours, prompting a shift towards parenchyma-sparing surgical strategies.¹ Anatomical segmentectomy has emerged as a curative option for early-stage non–small-cell lung cancer (NSCLC), supported by randomized trials (JCOG0802/WJOG4607L and CALGB 140503) demonstrating comparable relapse-free survival and superior overall survival (OS) compared with lobectomy.^2,3 However, a higher local recurrence rate highlights the need for adequate resection margins, with studies recommending margins ≥2 cm or a tumour-to-margin ratio ≥1.0 to ensure oncologic safety.^4,5

Accurate identification of the intersegmental plane (ISP) is essential for achieving curative resection with adequate parenchymal margins while preserving lung function. The conventional inflation-deflation (ID) method is a ventilation-based technique that visualizes ISPs based on differential lung inflation, whereas indocyanine green (ICG) fluorescence imaging is a perfusion-based technique that delineates segmental boundaries by distinguishing perfused from non-perfused lung parenchyma after intravenous injection.^6–8 While ICG adoption has accelerated globally due to its technical convenience and clearer anatomical delineation, there is a relative lack of formal, long-term comparative data evaluating its oncologic superiority or equivalence over the conventional ID method.^9,10 Furthermore, most existing studies have focused on technical success or short-term perioperative outcomes, leaving a significant evidence gap regarding whether these different ISP identification principles—ventilation-based vs perfusion-based—translate into clinically meaningful differences in margin adequacy and long-term recurrence-free survival.

This study aimed to compare ICG- and ID-guided anatomical segmentectomy in early-stage NSCLC, focusing on resection margin adequacy, local recurrence, and postoperative pulmonary function.

METHODS

Ethical statement

The Institutional Review Board of Seoul National University Hospital approved this study (approval no. 2507-118-1661, 2025-07-25), and the requirement for written informed consent was waived. No biological specimens or identifiable personal data were stored for future research use.

Patients

We retrospectively reviewed the medical records of 846 patients who underwent anatomical segmentectomy for clinical N0M0 NSCLC between January 2018 and December 2023. After excluding patients with prior diagnostic wedge resection, surgery for recurrent lung cancer, non-anatomical procedures, tumour size >4 cm on preoperative CT, thoracotomy, or pathologic T2b tumours, 410 patients were included in the final analysis (Figure 1).

Figure 1.

Flow Diagram of Patient Selection and Study Cohort. Abbreviations: ICG, indocyanine green; ID, inflation-deflation

Operation procedure

The surgical approach and technique were determined according to the operating surgeon’s preference. Procedures were performed via video- or robotic-assisted thoracic surgery. For the ID technique, the ISP was initially estimated using anatomical landmarks—segmental vein, artery, and segmental fissure—and marked.⁷ After transection of the target segmental bronchus, differential lung ventilation was applied to confirm the ISP. The demarcation line was marked on the lung surface while the lung was inflated; the lung was then deflated, and the parenchyma was divided along the marked line using staplers.¹¹ For the ICG technique, the segmental artery, vein, and bronchus were first completely divided. Subsequently, near-infrared thoracoscopy (Karl Storz, Tuttlingen, Germany, or Stryker Corporation, Kalamazoo, MI, USA) was used following intravenous injection of 3-5 mL of ICG solution. The fluorescent intersegmental boundary was clearly visualized and marked on the lung surface, and the lung parenchyma was divided along the demarcated line using staplers.¹²

All procedures were performed by 8 experienced thoracic surgeons proficient in anatomical segmentectomy. Following the introduction of ICG in April 2021, the technique was adopted by multiple surgeons, whereas the ID method was subsequently performed by a smaller number of surgeons.

Pathologic and clinical evaluation

Pathologic staging followed the 8th edition of the tumour-node-metastasis classification.¹³ Central and lateral resection margins were assessed microscopically on deflated specimens after staple removal. The central resection margin was defined as the shortest distance from the tumour to the proximal bronchial stump and the lateral margin as the shortest distance to the nearest peripheral parenchymal staple line. Operative time was defined as the interval from skin incision to completion of skin closure in the operating room.

Postoperative management

Per the institutional protocol, chest tubes were removed when no air leak was present and drainage was <200-250 mL/day. Patients were typically discharged the day after tube removal unless complications occurred. Postoperative death was defined as death within 30 days or during the index hospitalization. Pulmonary function tests were performed preoperatively (approximately 1 month before surgery) and at 6 months postoperatively.

Statistical analysis

Analyses were conducted using IBM SPSS Statistics v25.0 (IBM Corp., Armonk, NY, USA) and R software (RStudio, PBC, Boston, MA, USA). Continuous variables are presented as mean ± standard deviation, and categorical variables as frequencies and percentages.

Patients were grouped by ISP identification method (ID vs ICG). Outcomes were compared for the entire cohort and in a subgroup undergoing surgery after April 2021, when ICG guidance was introduced.

Normality was assessed using the Shapiro-Wilk test. Student’s t-test or nonparametric tests were used for continuous variables, and χ² or Fisher’s exact tests for categorical variables. A 2-sided P < .05 was considered statistically significant.

Inverse probability of treatment weighting (IPTW) based on propensity scores was applied to adjust for confounding. Propensity scores were estimated using logistic regression, including age, sex, CT solid component size, preoperative forced expiratory volume in 1 s (FEV₁), diffusing capacity of the lung for carbon monoxide (DLCO), and clinical stage. IPTW was selected to retain the full cohort, balance covariates between groups, reduce selection bias, and provide greater statistical efficiency than that of matching.

The primary outcome was resection margin distance (central and lateral). Key secondary outcomes included recurrence-free survival (RFS: time from surgery to recurrence or last follow-up) and overall survival (OS: time from surgery to death or last follow-up). Local recurrence was defined as tumour recurrence occurring at the surgical margin or within the ipsilateral lung. Survival was estimated using the Kaplan-Meier (KM) method and compared using the log-rank test. Weighted Cox proportional hazards models were used to calculate hazard ratios (HRs) with 95% CIs after IPTW adjustment.

RESULTS

Baseline characteristics

Table 1 presents baseline characteristics. Of 410 patients, 286 (69.8%) and 124 (30.2%) underwent segmentectomy using the ID and ICG methods, respectively. Preoperative pulmonary function parameters, including forced vital capacity (FVC), FEV₁, and the FEV₁/FVC ratio, were comparable. DLCO was significantly higher in the ICG than in the ID group (ICG: 100.0% [89.0-116.0] vs ID: 94.0% [84.0-108.0], P = .002).

Table 1.

Baseline Characteristics of Patients According to Surgical Method

Characteristics	ID (n = 286)	ICG (n = 124)	P-value
Sex (male, %)	134 (46.9%)	63 (50.8%)	.53
Age (years)	65.0 [59.0-72.0]	64.0 [60.0-70.0]	.69
Smoking history			.67
Never smoker	168 (58.7%)	78 (62.9%)
Ex-smoker	81 (28.3%)	30 (24.2%)
Current smoker	37 (12.9%)	16 (12.9%)
Surgical approach			.14
VATS	272 (95.1%)	113 (91.1%)
Robot	14 (4.9%)	11 (8.9%)
Resected segment			.77
RUL	46 (16.1%)	23 (18.5%)
RLL	88 (30.8%)	32 (25.8%)
LUL	87 (30.4%)	40 (32.3%)
LLL	65 (22.7%)	29 (23.4%)
Resected Segment number			.95
1	119 (41.6%)	53 (42.7%)
2	69 (24.1%)	32 (25.8%)
3	59 (20.6%)	23 (18.5%)
4	39 (13.6%)	16 (12.9%)
Pulmonary function
FVC (%)	104.5 [94.0-113.0]	104.0 [97.0-115.0]	.18
FEV₁ (%)	107.0 [94.2-118.8]	110.0 [99.8-120.0]	.090
FEV₁/FVC	75.0 [69.0-79.0]	75.0 [70.8-80.0]	.47
DLCO (%)	94.0 [84.0-108.0]	100.0 [89.0-116.0]	.002

Continuous variables are expressed as median [IQR] as appropriate, and categorical variables as n (%). Abbreviations: DLCO, diffusing capacity of the lung for carbon monoxide; FEV₁, forced expiratory volume in 1 second; FVC, forced vital capacity; ICG, indocyanine green; ID, inflation-deflation; LLL, left lower lobe; LUL, left upper lobe; RLL, right lower lobe; RUL, right upper lobe; VATS, video-assisted thoracoscopic surgery.Bold values indicate statistical significance (P < 0.05).

Although the clinical stage distribution differed modestly (P = .058), both cohorts were predominantly composed of clinical T1a and T1b disease. Pathologic invasive tumour size was slightly larger in the ICG group (P = .008), whereas pathologic stage and histologic subtype were comparable (Table 2). After IPTW adjustment, covariate balance was well achieved, with standardized mean differences (SMDs) <0.1 for all variables (Table S1).

Table 2.

Tumour Characteristics According to Surgical Method

Characteristics	ID (n = 286)	ICG (n = 124)	P-value
CT total size (cm)	1.6 [1.2-2.1]	1.6 [1.3-2.1]	.94
CT solid size (cm)	1.1 [0.6-1.7]	1.0 [0.6-1.5]	.50
Clinical stage			.058
Tis, T1mi	16 (5.6%)	3 (2.4%)
T1aN0M0	122 (42.7%)	53 (42.7%)
T1bN0M0	96 (33.6%)	56 (45.2%)
T1cN0M0	27 (9.4%)	7 (5.6%)
T2aN0M0	25 (8.7%)	5 (4.0%)
Pathologic total size (cm)	1.6 [1.2-2.0]	1.5 [1.2-2.2]	.64
Pathologic invasive size (cm)	1.1 [0.6-1.6]	1.3 [0.9-1.8]	.008
Pathologic stage			.96
Tis, T1mi	41 (14.3%)	15 (12.1%)
T1aN0M0	76 (26.6%)	31 (25.0%)
T1bN0M0	100 (35.0%)	47 (37.9%)
T1cN0M0	30 (10.5%)	14 (11.3%)
T2aN0M0	39 (13.6%)	17 (13.7%)
Histologic type			.37
Adenocarcinoma	270 (94.4%)	114 (91.9%)
Squamous cell carcinoma	16 (5.6%)	10 (8.1%)
Pathologic features
STAS positive	33 (11.5%)	23 (18.5%)	.084
Lymphovascular invasion	18 (6.3%)	4 (3.2%)	.24

Continuous variables are expressed as median [IQR], and categorical variables as n (%). Abbreviations: ICG, indocyanine green; ID, inflation-deflation; N0M0, node-negative, metastasis-negative disease; STAS, spread through air spaces; Tis, carcinoma in situ; T1mi, minimally invasive adenocarcinoma. Bold values indicate statistical significance (P < 0.05).

Perioperative and pathologic outcomes

Median operative time (115.0 [90.0-140.0] vs 108.0 [85.0-130.0] min, P = .031) and hospital stay (5.0 [4.0-6.0] vs 4.0 [4.0-6.0] days, P = .003) were longer in the ID group than in the ICG group. Postoperative complication rates did not differ significantly (31/286 [10.8%] vs 15/124 [12.1%], P = .57), and prolonged air leak occurred at similar rates (18/286 [6.3%] vs 8/124 [6.5%]) (Table 3).

Table 3.

Perioperative and Pathologic Outcomes According to Surgical Method

Variable	ID (n = 286)	ICG (n = 124)	P-value
Operative time (min)	115.0 [90.0-140.0]	108.0 [85.0-130.0]	.031
Hospital stay (days)	5.0 [4.0-6.0]	4.0 [4.0-6.0]	.003
Complication	31 (10.8%)	15 (12.1%)	.57
Pneumonia	10 (3.5%)	4 (3.2%)
Prolonged air leak	18 (6.3%)	8 (6.5%)
Pneumothorax	3 (1.0%)	2 (1.6%)
Chylothorax, bleeding	0 (0.0%)	1 (0.8%)
Central margin (cm)	2.5 [1.6-3.5]	2.0 [1.5-3.0]	.014
Lateral margin (cm)	2.1 [1.3-3.1]	1.8 [1.2-2.6]	.14
Margin-to-tumour ratio (M/T ratio)	1.80 [1.0-3.0]	1.38 [1.0-2.7]	.19
M/T ratio ≥1	175 (61.2%)	68 (54.8%)	.28
Central margin ≥2 cm	196 (68.5%)	72 (58.1%)	.045
Lateral margin ≥2 cm	148 (51.7%)	47 (37.9%)	.010
ΔFVC (%)	−8.0 [−14.5 to −0.5)]	−7.0 [−13.8 to −1.0)]	.96
ΔFEV₁ (%)	−9.0 [−16.0 to −2.0)]	−7.0 [−15.0 to −2.0)]	.39
ΔFEV₁/FVC (%)	−1.0 [−4.0 to 2.0]	0.0 [−2.0 to 2.0]	.077
ΔDLCO (%)	−10.0 [−17.0 to −2.8)]	−11.0 [−19.0 to −2.0)]	.45

DLCO, diffusing capacity of the lung for carbon monoxide; FEV₁, forced expiratory volume in 1 second; FVC, forced vital capacity; ICG, indocyanine green; ID, inflation-deflation; M/T ratio, margin-to-tumour ratio. Bold values indicate statistical significance (P < 0.05).

Central resection margins were longer in the ID group (2.5 [1.6-3.5] vs 2.0 [1.5-3.0] cm, P = .014), whereas lateral margin distances were similar (2.1 [1.3-3.1] vs 1.8 [1.2-2.6] cm, P = .14). However, the margin-to-tumour ratio and postoperative pulmonary function changes did not differ significantly between groups.

In the subgroup including T1a, T1b, and TIS/MIA tumours (N = 346), the ICG group had smaller resection margins than the ID group. The central margin was significantly shorter in the ICG group (2.0 [1.5-2.9] vs 2.5 [1.6-3.5] cm; P = .011), with a lower proportion achieving ≥2 cm (65/112 [58.0%] vs 161/234 [68.8%]; P = .049). Lateral margins were also numerically smaller (1.8 [1.2-2.6] vs 2.0 [1.3-3.1] cm; P = .27), with fewer patients achieving ≥2 cm (44/112 [39.3%] vs 116/234 [49.6%]; P = .073).

Subgroup analyses according to segmentectomy complexity demonstrated generally comparable perioperative and functional outcomes between the ID and ICG groups in both simple and complex segmentectomy cohorts. In the simple segmentectomy subgroup, the ICG group showed a larger pathologic invasive tumour size (1.5 vs 1.1 cm, P = .019), shorter hospital stay (4.5 vs 5.0 days, P = .019), and a lower rate of achieving a lateral margin ≥2 cm (41.4% vs 58.6%, P = .015). In the complex segmentectomy subgroup, no significant differences were observed in margin outcomes, postoperative complications, or postoperative pulmonary function changes, although preoperative DLCO was higher in the ICG group (104.0% [94.2-116.8] vs 94.0% [87.0-110.0], P = .010) (Tables S2 and S3).

Recurrence outcomes before and after IPTW

Median follow-up was significantly longer in the ID than in the ICG group (54.0 [36.0-64.0] vs 30.0 [24.0-36.0] months, P < .001). During follow-up, 21 (7.3%) and 6 (4.8%) patients in the ID and ICG groups experienced recurrence (P = .45). Local recurrence occurred in 11 (3.8%) and 3 (2.4%) patients in the ID and ICG groups, respectively (P = .56); distant recurrence occurred in 10 (3.5%) and 3 (2.4%) patients in the ID and ICG groups, respectively (P = .77).

Local RFS was similar between the groups on unweighted KM analysis (log-rank P = .78). The 3-year local RFS was 96.5% (95% CI, 94.2-98.8) and 97.1% (95% CI, 93.9-100.0) in the ID and ICG groups, respectively. Median local RFS was not reached in either group. After IPTW adjustment, weighted KM curves for local RFS remained comparable (P = .75) (Figure 2A and B).

Figure 2.

Kaplan-Meier Curves for Local and Overall Recurrence-Free Survival Before and After Inverse Probability of Treatment Weighting (IPTW) Adjustment. Kaplan-Meier survival analysis comparing recurrence-free survival between patients undergoing segmentectomy using the inflation-deflation (ID) and indocyanine green (ICG) methods. (A) Local recurrence-free survival (RFS) before IPTW adjustment (log-rank P = .78). (B) Local RFS after IPTW adjustment (log-rank P = .75). (C) Overall RFS before IPTW adjustment (log-rank P = .97). (D) Overall RFS after IPTW adjustment (log-rank P = .91)

Overall RFS did not differ significantly on unweighted KM analysis (log-rank P = .97). The 3-year total RFS was 94.2% (95% CI, 91.2-97.2) and 95.1% (95% CI, 90.9-99.4) in the ID and ICG groups, respectively. Median total RFS was not reached in either group. After IPTW adjustment, KM analysis continued to show no significant difference (P = .91) (Figure 2C and D). In IPTW-adjusted multivariable Cox regression analysis, the surgical method was not associated with total recurrence (HR, 1.41; P = .56). Advanced clinical stage (≥T1B) remained independently associated with an increased risk of recurrence (HR, 5.50; P = .012), whereas neither central nor lateral resection margin was significantly associated with recurrence (Table 4).

Table 4.

Cox Regression Analysis for Recurrence-Free and Overall Survival Before and After IPTW Adjustment

Outcome	Variable	Univariable HR (95% CI)	P-value	Multivariable HR (95% CI)	P-value
RFS (before)	ICG vs ID (ref: ID)	1.27 (0.47-3.41)	.63	1.19 (0.44-3.21)	.73
	Clinical stage ≥1B	6.40 (1.88-21.86)	.003	6.33 (1.84-21.73)	.003
	C_margin	1.14 (0.81-1.59)	.45	1.15 (0.75-1.76)	.51
	L_margin	1.05 (0.81-1.35)	.72	0.95 (0.66-1.36)	.77
RFS (after)	ICG vs ID (ref: ID)	1.36 (0.45-4.08)	.59	1.41 (0.45-4.36)	.56
	Clinical stage ≥1B	5.71 (1.49-21.84)	.011	5.50 (1.46-20.81)	.012
	C_margin	1.19 (0.88-1.61)	.26	1.15 (0.77-1.72)	.48
	L_margin	1.10 (0.93-1.29)	.25	1.02 (0.77-1.36)	.90
OS (before)	ICG vs ID (ref: ID)	5.41 (0.38-77.64)	.21	3.47 (0.24-51.15)	.37
	Clinical stage ≥1B	2.57 (0.51-12.91)	.25	3.91 (0.74-20.76)	.11
	C_margin	0.55 (0.28-1.08)	.081	0.64 (0.27-1.55)	.32
	L_margin	0.56 (0.29-1.09)	.088	0.64 (0.29-1.44)	.28
OS (after)	ICG vs ID (ref: ID)	4.08 (0.38-44.02)	.25	2.94 (0.21-40.82)	.42
	Clinical stage ≥1B	1.89 (0.40-8.98)	.42	2.81 (0.40-19.54)	.30
	C_margin	0.56 (0.40-0.78)	<.001	0.69 (0.43-1.13)	.14
	L_margin	0.55 (0.39-0.77)	<.001	0.61 (0.37-1.02)	.058

C_margin, central margin; ICG, indocyanine green; ID, inflation-deflation; IPTW, inverse probability of treatment weighting; L_margin, lateral margin; OS, overall survival; RFS, recurrence-free survival.

Survival outcomes before and after IPTW

During follow-up, 9 (3.1%) patients in the ID group and 1 (0.8%) patient in the ICG group died (P = .28). Unweighted KM analysis demonstrated no significant difference in OS between groups (log-rank P = .18). Three-year OS was 99.6% (95% CI, 98.8-100.0) and 97.4% (95% CI, 92.6-100.0) in the ID and ICG groups, respectively (P = .47). Median OS was not reached in either group.

After IPTW adjustment, no significant difference was observed on weighted KM analysis (log-rank P = .26) (Figure 3A and B). In IPTW-adjusted multivariable Cox proportional hazards analysis, the surgical method was not associated with OS (HR, 2.94; P = .42). Although a greater lateral resection margin distance was associated with improved OS in univariable analysis, this association was not maintained after multivariable adjustment (HR, 0.61; P = .058) (Table 4).

Figure 3.

Kaplan-Meier Curves for Overall Survival Before and After Inverse Probability of Treatment Weighting (IPTW) Adjustment. Kaplan-Meier curves comparing overall survival (OS) before (A) and after (B) inverse probability of treatment weighting adjustment between patients undergoing segmentectomy using the inflation-deflation (ID) and indocyanine green (ICG) methods. No significant differences in OS were observed between the ID and ICG groups in either the unweighted analysis (A, P = .18) or the IPTW-adjusted analysis (B, P = .26)

DISCUSSION

We evaluated whether differences in ISP identification techniques translate into clinically meaningful differences in oncologic and functional outcomes after anatomical segmentectomy. After IPTW adjustment, OS and RFS were comparable, indicating that ICG guidance was not associated with improved oncologic outcomes compared with the ID method.

Three key findings emerge. First, once an adequate parenchymal margin is achieved, the precision of ISP identification has minimal impact on oncologic outcomes, emphasizing margin adequacy—rather than visualization technique—as the principal determinant of survival and recurrence. Second, by integrating postoperative pulmonary function with oncologic end-points, the study demonstrates that ICG guidance provides no functional or oncologic benefit over ID. Third, this study represents one of the earliest IPTW-adjusted analyses comparing long-term oncologic outcomes between ICG- and ID-guided segmentectomy.

ICG guidance may shorten operative time, facilitate ISP visualization, reduce blood loss, and decrease postoperative complications.^14,15 However, these findings should be interpreted cautiously. ISP identification does not alter bronchovascular dissection, hilar exposure, or parenchymal division, which determine bleeding risk and operative complexity. Therefore, true physiological differences in morbidity are unlikely, and reported advantages may reflect surgeon experience, procedural heterogeneity across institutions, or selection bias. Although operative time was modestly shorter in the ICG group, this difference was unlikely to be clinically meaningfulbecause postoperative complication rates, prolonged air leak, and perioperative recovery were similar between the groups. The slightly longer hospital stay in the ID group likely reflects temporal differences in enhanced recovery after surgery implementation rather than technical limitations, particularly given the similar rates of prolonged air leak.

Despite similar perioperative outcomes, ICG offers rapid and reliable ISP visualization and technical convenience, particularly in minimally invasive surgery.^9,14,16 Superior anatomic delineation is reported with ICG than with ID.^10,17 Although ICG provided clearer visualization of the ISP, this did not translate into wider margins. In fact, the ID group achieved a ≥2 cm lateral margin more frequently, suggesting that margin adequacy may depend more on resection planning and intentional parenchymal tailoring than on the visualization technique itself.

Lateral margin distance and the margin-to-tumour ratio are established determinants of local control after sublobar resection. A margin ≥2 cm or margin-to-tumour ratio ≥1 is consistently associated with reduced local recurrence.^4,18 Thus, oncologic success in segmentectomy depends primarily on margin adequacy rather than strict adherence to the anatomic segmental plane.¹⁹ Consistent with this concept, neither central nor lateral margins differed significantly between the ICG and ID groups. Although the ID group had a slightly longer central margin, the lateral margin—most relevant for oncologic control—was equivalent. This finding likely reflects bronchovascular anatomy and surgeon-directed resection planning rather than the ISP visualization technique itself. Although a greater lateral margin distance was associated with improved OS in univariable analysis, this association was not maintained after multivariable adjustment. Neither lateral nor central margin independently predicted survival outcomes. Given the low number of death events and the favourable prognosis of this predominantly early-stage cohort, these findings should be interpreted cautiously. This association between margin distance and OS should be interpreted cautiously. The absence of a corresponding reduction in recurrence suggests that this effect may reflect tumour biology rather than a margin-driven mechanism.²⁰ Given the low recurrence rate, limited number of events, non-cancer–related mortality, and variability inherent in weighted analyses, this finding should be considered exploratory.^2,3,21 Residual confounding related to patient frailty or competing risks may also partly explain this association.

Several studies further support the central role of margin adequacy. Ueda et al²² reported that only one-third of tumours ≤ 2 cm achieved guideline-recommended margins using the anatomic segmental plane alone. Cheng et al²³ similarly found adequate margins in only 57.4% of conventional segmentectomies, with extended or image-guided approaches improving adequacy to 75.8%. These findings suggest that many small peripheral NSCLCs, particularly those near intersegmental borders, require intentional deviation from the standard anatomic plane to secure adequate margins, explaining why superior anatomic visualization with ICG does not necessarily translate into improved margins or oncologic outcomes.

Functional preservation is another key objective of segmentectomy. We observed comparable postoperative pulmonary function changes between the groups, indicating that both techniques allow lung-sparing parenchymal tailoring without compromising residual function.

Follow-up in the ICG group was shorter due to its introduction in April 2021; nevertheless, survival outcomes remained excellent. Three-year OS was 99.6% and 97.4% in the ID and ICG groups, with corresponding 3-year RFS rates of 94.2% and 95.1%, respectively. These results are consistent with those of randomized trials of segmentectomy for early-stage NSCLC. The JCOG0802/WJOG4607L trial reported 5-year OS and RFS rates of 94.3% and 88.0%, respectively. However, the CALGB/Alliance trial reported a 5-year OS of 80.3% and disease-free survival of 63.6%.^2,3

This study has several limitations. First, its retrospective, single-centre design may involve selection bias and unmeasured confounders. Notably, surgeon- and time-related clustering associated with the 2021 introduction of ICG may have influenced outcomes, as the adoption of fluorescence coincided with evolving surgical and perioperative practices. In addition, robotic-assisted surgery was more frequently performed in the ICG group, and the enhanced 3D visualization provided by the robotic platform may have influenced ISP identification independent of the ICG technique itself. Second, the follow-up duration was disparate; the shorter median follow-up in the ICG group (30 vs 54 months) might underestimate late recurrences. Third, margin assessment on deflated, fixed specimens may not fully reflect the 3D complexity of the ISP due to tissue shrinkage. Finally, the low event rate in this early-stage cohort limited the statistical power to detect subtle survival differences, necessitating larger multicentre studies for long-term confirmation.

CONCLUSION

ICG and ID-based ISP identification yielded similar functional and oncologic outcomes after segmentectomy in early-stage NSCLC. Long-term prognosis after segmentectomy appeared to depend more on clinical stage than on the ISP visualization method.

Glossary

Abbreviations

DLCO

diffusing capacity of the lung for carbon monoxide

FEV₁

forced expiratory volume in one second

FVC

forced vital capacity

HR

hazard ratio

ICG

indocyanine green

ID

inflation-deflation

IPTW

inverse probability of treatment weighting

ISP

intersegmental plane

KM

Kaplan-Meier

NSCLC

non–small-cell lung cancer

OS

overall survival

RFS

recurrence-free survival

AUTHOR CONTRIBUTIONS

Seon Yong Bae (Conceptualization, Methodology, Formal analysis, Writing—original draft), Dae Hyeon Kim, Taeyoung Yun, Ji Hyeon Park, Bubse Na, Kwon Joong Na, Samina Park, Hyun Joo Lee, Chang Hyun Kang, and Young Tae Kim (Resources), and In Kyu Park (Resources, Conceptualization, Supervision, Writing—review & editing, Validation)

SUPPLEMENTARY MATERIAL

Supplementary material is available at ICVTS online.

FUNDING

None declared.

CONFLICTS OF INTEREST

None declared.

DATA AVAILABILITY

The data underlying this article cannot be shared publicly to protect the privacy of the individuals who participated in the study. However, the data will be shared on reasonable request to the corresponding author.