Perioperative recovery and survival after lobectomy versus sublobar resection in adults aged ≥ 70 years with NSCLC ≤ 3 cm: complementary SEER and real-world cohort analyses

Abstract

Background

For older adults with non-small cell lung cancer (NSCLC), the choice of resection must balance durable cancer control against physiological reserve and postoperative recovery. We combined population-level long-term survival data with granular real-world perioperative data to examine this trade-off.

Methods

The Surveillance, Epidemiology, and End Results (SEER) cohort included 2,630 patients aged ≥ 70 years with NSCLC ≤ 3 cm diagnosed in 2010–2015 (855 lobectomy, 1,421 wedge resection, and 354 segmentectomy). Optimal 1:1 propensity-score matching retained 809 patients per group. The two-center cohort included 223 patients treated by video-assisted thoracoscopic surgery in 2017–2020 (145 lobectomies and 78 sublobar resections: 73 wedge resections and five segmentectomies). Overlap weighting used age, sex, tumor size, clinical stage, consolidation-to-tumor ratio, smoking, and recorded cardiopulmonary comorbidities. The primary real-world survival analysis was restricted to tumors ≤ 2 cm and 36 months.

Results

In SEER, matched sublobar resection was not associated with a statistically significant OS difference for tumors ≤ 2 cm (hazard ratio [HR] 1.18, 95% confidence interval [CI] 0.93–1.51; P = 0.177), whereas OS was lower after sublobar resection for tumors > 2–3 cm (HR 1.31, 95% CI 1.03–1.66; P = 0.028). Procedure-specific models showed no significant difference between segmentectomy and lobectomy for tumors ≤ 2 cm (HR 1.03, 95% CI 0.83–1.26), while wedge resection was associated with lower OS (HR 1.32, 95% CI 1.14–1.53). In the real-world cohort, patients selected for wedge-dominant sublobar resection were older and had more recorded comorbidities. After overlap weighting, wedge-dominant sublobar resection was associated with shorter chest-tube duration (mean difference − 2.49 days, 95% CI − 3.56 to − 1.37) and less hypoproteinemia (risk difference − 14.4% points, 95% CI − 27.6 to − 1.7). The composite in-hospital complication risk was 16.4% points lower (95% CI − 34.7 to 1.3), whereas total hospital stay did not differ significantly (mean difference − 1.60 days, 95% CI − 4.62 to 2.04). For tumors ≤ 2 cm, adjusted 3-year OS was 93.0% after lobectomy and 88.5% after wedge-dominant sublobar resection (HR 1.73, 95% CI 0.19–15.63; P = 0.558).

Conclusions

In routine practice, wedge-dominant sublobar resection was selected for older adults with greater comorbidity burden and was associated with shorter chest drainage and less hypoproteinemia, while total hospital stay was similar after adjustment. Among patients with tumors ≤ 2 cm, the real-world cohort showed no statistically significant 3-year OS difference. In the longer-term SEER analysis, segmentectomy showed no significant adjusted OS difference from lobectomy for tumors ≤ 2 cm, whereas wedge resection warrants greater oncologic caution.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12877-026-08089-z.

Background

Population ageing and wider use of thoracic computed tomography have increased the number of older adults undergoing assessment for lung cancer [1, 2]. Chronological age alone, however, does not describe operative fitness. Cardiopulmonary disease, multimorbidity, nutritional reserve, physical function, and the ability to recover independence vary substantially among patients of the same age [3–7]. For an older adult, the value of an operation therefore depends not only on cancer control, but also on how much physiological reserve is consumed during treatment and how quickly the patient can return to usual life.

Lobectomy has long provided a sound oncological operation through anatomical resection and systematic nodal evaluation. Sublobar resection preserves more functional lung and may lessen the early treatment burden, making it particularly relevant when comorbidity or limited reserve narrows the margin for postoperative recovery [3, 4, 8–13]. Modern randomized trials have also changed the treatment landscape for carefully selected peripheral NSCLC. CALGB/Alliance 140,503 demonstrated non-inferior disease-free survival after sublobar resection for peripheral, pathologically node-negative tumors ≤ 2 cm [14], while JCOG0802/WJOG4607L showed favorable overall survival after anatomical segmentectomy for peripheral tumors, with a higher local-recurrence rate [15]. These results support lung-sparing surgery in selected patients, while underscoring the importance of tumor size, operative anatomy, margins, and nodal assessment.

Older adults treated in routine practice are less uniform than trial populations, and wedge resection is often used when physiological vulnerability influences surgical choice. Population registries can describe long-term survival across large groups, whereas institutional data can show the comorbidity pattern behind treatment selection and the details of postoperative recovery. We therefore combined complementary SEER and two-center real-world analyses to examine tumor-size- and procedure-specific overall survival, perioperative recovery, and the clinical trade-off between lobectomy and sublobar resection in adults aged 70 years or older with NSCLC ≤ 3 cm.

Methods

Study design

This retrospective dual-cohort study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [16]. The SEER and institutional cohorts were analyzed separately. SEER was used to examine long-term population-level overall survival, whereas the institutional cohort was used to examine preoperative selection, in-hospital recovery, and 36-month overall survival. The institutional cohort was not treated as a formal validation set because the populations, treatment eras, covariates, procedure composition, and follow-up differed.

SEER cohort

Patient-level records were derived from the National Cancer Institute SEER program. The supplied extract contained patients diagnosed in 2010–2015 with follow-up through 2022. Eligible records were age ≥ 70 years, NSCLC recorded in the SEER*Stat selection, primary tumor diameter 1–30 mm, M0 disease, registry-derived AJCC seventh-edition stage IA1–IIIA, a primary site in an upper, middle, or lower lung lobe, and surgery codes for wedge resection, segmentectomy, or lobectomy. The complete-case analysis included 2,630 patients: 855 underwent lobectomy, 1,421 wedge resection, and 354 segmentectomy.

Real-world institutional cohort

We reviewed patients aged ≥ 70 years who underwent video-assisted thoracoscopic lobectomy, wedge resection, or segmentectomy at Tianjin Chest Hospital or Tianjin Medical University General Hospital from January 2017 through December 2020. Inclusion was based on preoperative clinical assessment of a primary, resectable NSCLC with maximum diameter ≤ 3 cm. Pathological upstaging after resection was permitted because treatment selection preceded pathological staging. Exclusion criteria were tumor diameter > 3 cm, neoadjuvant treatment, pneumonectomy, sleeve resection, non-primary or non-malignant lesions, incomplete survival follow-up, and conversion to thoracotomy for an intraoperative technical event. The verified dataset comprised 223 patients: 145 lobectomies and 78 sublobar resections (73 wedge resections and five segmentectomies). Surgical selection reflected tumor location and radiological appearance, age, cardiopulmonary reserve, comorbidity, and patient preference. Formal ASA class, ECOG performance status, frailty score, FEV1, and DLCO were not consistently stored in the analytic dataset.

All institutional procedures were thoracoscopic. In routine practice, lobectomy was generally accompanied by evaluation of six lymph-node stations, including three mediastinal stations; sublobar resection was generally accompanied by evaluation of three stations, including one mediastinal station. Absolute lymph-node counts and patient-level station counts were not consistently available. The observed pathological-stage distribution may therefore reflect both tumor biology and differences in nodal detection. Margin distance and margin-to-tumor ratio were also not available as analyzable fields.

Outcomes and follow-up

SEER overall survival (OS) was measured from diagnosis to death from any cause or last recorded follow-up. Institutional OS was measured from surgery to death from any cause or the administrative vital-status censoring date and was updated through January 2024. Median follow-up was estimated by reverse Kaplan–Meier. The primary real-world survival analysis used a common 36-month horizon in patients with tumors ≤ 2 cm. A ≤ 3 cm analysis was secondary. Cause of death was documented when available, but uniform cause adjudication was not possible; recurrence date, disease-free survival, local recurrence, and recurrence pattern were not consistently recorded.

Institutional perioperative outcomes were chest-tube duration, total hospital length of stay, hypoproteinemia, cardiac complications, pulmonary complications, surgical complications, and a composite of any recorded complication. The observation window was the index hospitalization, not the first three postoperative days. Surgical complications comprised the database-recorded categories of prolonged air leak, postoperative hemorrhage, or chylothorax. Uniform prospective severity grades and detailed diagnostic thresholds for every component were unavailable.

The postoperative systemic-treatment field only recorded whether patients received postoperative chemotherapy, targeted therapy, and/or immunotherapy. Treatment regimens were formulated based on clinical assessment and molecular testing results, without records of specific treatment modalities, medication regimens, timing of treatment, molecular targets, or treatment duration. No patient received neoadjuvant therapy. Postoperative treatment was reported descriptively and was not used as a baseline propensity covariate.

Statistical analysis

Continuous variables are summarized as mean ± standard deviation and were compared with the Mann–Whitney U test. Categorical variables are summarized as number (percentage) and were compared with Pearson’s chi-square test or Fisher’s exact test when expected counts were sparse. No analytic variable in the verified institutional dataset was missing.

For SEER, the propensity score for sublobar resection was estimated from age group, sex, diagnosis year, race, primary lobe, continuous tumor size, and registry-derived AJCC stage. Optimal 1:1 matching without replacement used a caliper of 0.20 pooled standard deviations of the propensity-score logit. Standardized mean differences (SMDs) assessed balance; an absolute SMD ≤ 0.10 was prespecified as acceptable. Kaplan–Meier methods estimated 5-year OS. Cox models compared sublobar resection with lobectomy using robust standard errors clustered by matched pair. Follow-up was estimated by reverse Kaplan–Meier. Sensitivity analyses used propensity-score overlap weighting and full-cohort adjusted Cox models that separated wedge resection and segmentectomy.

For the institutional cohort, the propensity score included age, sex, continuous tumor size, preoperative clinical stage, consolidation-to-tumor ratio (CTR > 50% versus ≤ 50%), smoking history, hypertension, diabetes, arrhythmia, coronary heart disease, cerebrovascular disease, and COPD/asthma. Overlap weights were PS for lobectomy and 1 − PS for sublobar resection [17]. This estimand emphasizes patients with clinical equipoise and retains all observed patients while down-weighting those with extreme treatment probabilities. Covariate balance was assessed with SMDs. Adjusted means, risks, mean differences (MDs), risk differences (RDs), weighted Kaplan–Meier estimates, 36-month restricted mean survival time (RMST), and weighted Cox HRs were calculated. Percentile 95% CIs and P values were obtained from 2,500 nonparametric bootstrap samples. Up-to-2:1 propensity-score matching was retained as a sensitivity analysis. All tests were two-sided; P < 0.05 denoted statistical significance.

Results

SEER cohort

Before matching, 855 patients had undergone lobectomy and 1,775 sublobar resection. Optimal matching retained 809 patients in each group; the matched sublobar group included 636 wedge resections and 173 segmentectomies. Maximum absolute SMD decreased from 0.716 before matching to 0.075 after matching. Median follow-up was 120 months in the matched overall cohort.

Across all tumors ≤ 3 cm, matched 5-year OS was 57.6% after lobectomy and 50.7% after sublobar resection (HR 1.23, 95% CI 1.04–1.46; P = 0.019). For tumors ≤ 1 cm, the HR was 1.03 (95% CI 0.51–2.08; P = 0.942). For tumors > 1–2 cm, the HR was 1.21 (95% CI 0.93–1.57; P = 0.154), and the combined ≤ 2 cm comparison yielded HR 1.18 (95% CI 0.93–1.51; P = 0.177). For tumors > 2–3 cm, sublobar resection was associated with lower OS (HR 1.31, 95% CI 1.03–1.66; P = 0.028) (Fig. 1; Supplementary Table S3).

Fig. 1.

Surveillance, Epidemiology, and End Results (SEER) propensity-score-matched overall survival and covariate balance. Panels (A–E) show 5-year Kaplan–Meier curves for tumors ≤ 1 cm, > 1–2 cm, ≤ 2 cm, > 2–3 cm, and overall ≤ 3 cm. Hazard ratios (HRs) compare sublobar resection with lobectomy; 95% confidence intervals (CIs) and P values use robust standard errors clustered by matched pair. Shaded bands in panels (A–E) represent 95% CIs. Panel (F) shows the maximum absolute standardized mean difference (SMD) within each covariate domain before and after matching. Detailed level-specific balance is reported in Supplementary Table S2. AJCC, American Joint Committee on Cancer; OS, overall survival

Procedure-specific adjusted analyses clarified the combined result. For tumors ≤ 2 cm, segmentectomy was not associated with a significant OS difference from lobectomy (HR 1.03, 95% CI 0.83–1.26; P = 0.804), whereas wedge resection was associated with lower OS (HR 1.32, 95% CI 1.14–1.53; P < 0.001). For tumors > 2–3 cm, both wedge resection (HR 1.30, 95% CI 1.10–1.53; P = 0.002) and segmentectomy (HR 1.31, 95% CI 1.04–1.66; P = 0.024) were associated with lower OS than lobectomy. The overall overlap-weighted result was consistent with the primary matched analysis (HR 1.25, 95% CI 1.10–1.42; P < 0.001) (Supplementary Tables S4–S5).

Real-world treatment selection

The institutional cohort included 145 older adults who underwent lobectomy and 78 who underwent sublobar resection. Those selected for sublobar resection were older (mean 74.79 versus 72.79 years; P < 0.001) and had more recorded comorbidities (mean 1.53 versus 0.99; P < 0.001). Hypertension (51.3% versus 41.4%), diabetes (29.5% versus 20.7%), and coronary heart disease (43.6% versus 24.1%) were all more frequent in the sublobar group; the between-group difference was statistically significant for coronary heart disease (P = 0.004). In contrast, sublobar patients more often had clinical stage IA1 disease and CTR ≤ 50%, reflecting favorable tumor characteristics alongside greater medical vulnerability (Table 1). Of the 78 sublobar procedures, 73 were wedge resections and five were segmentectomies.

Table 1.

Preoperative characteristics of the complete real-world cohort

Characteristic	Category	Lobectomy	Sublobar resection	P value	Test
Number		145	78
Age, years	Mean ± SD	72.79 ± 2.90	74.79 ± 3.74	< 0.001	Mann–Whitney U
Recorded comorbidity count	Mean ± SD	0.99 ± 1.00	1.53 ± 1.16	< 0.001	Mann–Whitney U
Sex	Male	75 (51.7%)	42 (53.8%)	0.871	Pearson chi-square
	Female	70 (48.3%)	36 (46.2%)
Smoking history	Yes	72 (49.7%)	31 (39.7%)	0.202	Pearson chi-square
	No	73 (50.3%)	47 (60.3%)
Maximum tumor diameter	≤ 1 cm	12 (8.3%)	10 (12.8%)	0.554	Pearson chi-square
	> 1–2 cm	67 (46.2%)	34 (43.6%)
	> 2–3 cm	66 (45.5%)	34 (43.6%)
Clinical stage	IA1	12 (8.3%)	26 (33.3%)	< 0.001	Pearson chi-square
	IA2	70 (48.3%)	42 (53.8%)
	IA3	62 (42.8%)	10 (12.8%)
	IIB	1 (0.7%)	0 (0.0%)
CTR	≤ 50%	66 (45.5%)	63 (80.8%)	< 0.001	Pearson chi-square
	> 50%	79 (54.5%)	15 (19.2%)
Hypertension	Yes	60 (41.4%)	40 (51.3%)	0.202	Pearson chi-square
	No	85 (58.6%)	38 (48.7%)
Diabetes	Yes	30 (20.7%)	23 (29.5%)	0.191	Pearson chi-square
	No	115 (79.3%)	55 (70.5%)
Arrhythmia	Yes	9 (6.2%)	8 (10.3%)	0.411	Pearson chi-square
	No	136 (93.8%)	70 (89.7%)
Coronary heart disease	Yes	35 (24.1%)	34 (43.6%)	0.004	Pearson chi-square
	No	110 (75.9%)	44 (56.4%)
Cerebrovascular disease	Yes	6 (4.1%)	9 (11.5%)	0.068	Pearson chi-square
	No	139 (95.9%)	69 (88.5%)
COPD/asthma	Yes	3 (2.1%)	5 (6.4%)	0.132	Fisher exact
	No	142 (97.9%)	73 (93.6%)

Values are n (%) unless stated otherwise

CTR Consolidation-to-tumor ratio, COPD Chronic obstructive pulmonary disease

Clinical stage was determined preoperatively. The recorded comorbidity count is the sum of hypertension, diabetes, arrhythmia, coronary heart disease, cerebrovascular disease, and COPD/asthma

Adenocarcinoma was the predominant histological type in both groups. Pathological nodal metastasis was detected in 13 of 145 lobectomy patients and one of 78 sublobar patients (P = 0.022). Postoperative systemic therapy was recorded for 19 lobectomy patients and five sublobar patients (Table 2).

Table 2.

Pathological findings and postoperative systemic treatment in the real-world cohort

Characteristic	Category	Lobectomy	Sublobar resection	P value	Test
Histological type	Adenocarcinoma	121 (83.4%)	66 (84.6%)	0.615	Pearson chi-square
	Squamous-cell carcinoma	18 (12.4%)	8 (10.3%)
	Adenosquamous carcinoma	2 (1.4%)	1 (1.3%)
	Pulmonary sarcomatoid carcinoma	2 (1.4%)	0 (0.0%)
	Other	2 (1.4%)	3 (3.8%)
Pathological stage	IA1	12 (8.3%)	24 (30.8%)	< 0.001	Pearson chi-square
	IA2	50 (34.5%)	36 (46.2%)
	IA3	34 (23.4%)	8 (10.3%)
	IB	36 (24.8%)	9 (11.5%)
	IIA	4 (2.8%)	0 (0.0%)
	IIB	8 (5.5%)	0 (0.0%)
	IIIA	1 (0.7%)	1 (1.3%)
Lymph-node metastasis	Yes	13 (9.0%)	1 (1.3%)	0.022	Fisher exact
	No	132 (91.0%)	77 (98.7%)
Postoperative systemic therapy	Yes	19 (13.1%)	5 (6.4%)	0.190	Pearson chi-square
	No	126 (86.9%)	73 (93.6%)

The postoperative systemic-treatment field did not distinguish chemotherapy, targeted therapy, or immunotherapy. The two pulmonary sarcomatoid carcinomas were retained as NSCLC

Adjusted postoperative recovery

Overlap weighting achieved excellent balance: maximum absolute SMD was < 0.002 in both the complete ≤ 3 cm cohort and the ≤ 2 cm subset (Supplementary Table S1; Fig. 2A). In the complete cohort, adjusted mean chest-tube duration was 6.90 days after lobectomy and 4.41 days after sublobar resection (MD − 2.49 days, 95% CI − 3.56 to − 1.37; P < 0.001). Adjusted total hospital stay was 18.27 and 16.67 days, respectively; the difference was not statistically significant (MD − 1.60 days, 95% CI − 4.62 to 2.04; P = 0.375). Hypoproteinemia was less frequent after sublobar resection (RD − 14.4% points, 95% CI − 27.6 to − 1.7; P = 0.022). The adjusted risk of any recorded in-hospital complication was 59.5% after lobectomy and 43.1% after sublobar resection (RD − 16.4% points, 95% CI − 34.7 to 1.3; P = 0.065). Other individual complication categories did not reach statistical significance (Table 3; Fig. 3A–B).

Fig. 2.

Real-world cohort: patient selection and 3-year overall survival. Panel (A) shows covariate balance before and after overlap weighting for tumors ≤ 2 cm. Panel (B) shows overlap-weighted 3-year overall survival for tumors ≤ 2 cm, with the weighted Cox HR and unweighted numbers at risk. Shaded bands in panel B represent 95% CIs. CI, confidence interval; CTR, consolidation-to-tumor ratio; HR, hazard ratio; Lob., lobectomy; OS, overall survival; SMD, standardized mean difference; Sub., sublobar resection

Table 3.

Unadjusted and overlap-weighted perioperative outcomes in the complete real-world cohort

Outcome	Unadjusted lobectomy	Unadjusted sublobar	Unadjusted P	Adjusted lobectomy	Adjusted sublobar	Adjusted effect (95% CI)	Adjusted P
Chest-tube duration, days	6.47 ± 3.32	3.82 ± 1.74	< 0.001	6.90	4.41	MD − 2.49 days (− 3.56 to − 1.37)	< 0.001
Hospital stay, days	17.92 ± 6.20	14.88 ± 7.04	< 0.001	18.27	16.67	MD − 1.60 days (− 4.62 to 2.04)	0.375
Hypoproteinemia	31 (21.4%)	10 (12.8%)	0.147	25.4%	11.0%	RD − 14.4% points (− 27.6 to − 1.7)	0.022
Cardiac complications	24 (16.6%)	8 (10.3%)	0.234	17.1%	10.0%	RD − 7.1% points (− 18.0 to 4.8)	0.228
Pulmonary complications	33 (22.8%)	13 (16.7%)	0.304	23.1%	26.4%	RD 3.3% points (− 12.4 to 19.6)	0.673
Surgical complications	12 (8.3%)	5 (6.4%)	0.793	8.3%	13.5%	RD 5.2% points (− 7.3 to 17.4)	0.404
Any recorded in-hospital complication	82 (56.6%)	26 (33.3%)	0.001	59.5%	43.1%	RD − 16.4% points (− 34.7 to 1.3)	0.065

Overlap weights incorporated age, sex, tumor size, clinical stage, CTR, smoking history, hypertension, diabetes, arrhythmia, coronary heart disease, cerebrovascular disease, and COPD/asthma

CI Confidence interval, MD Mean difference, RD Risk difference

Negative effects favor sublobar resection

Fig. 3.

Real-world cohort: adjusted recovery measures and in-hospital complications. Panel (A) shows adjusted mean differences (MDs) in chest-tube duration and total hospital stay in the complete ≤ 3 cm cohort. Panel (B) shows adjusted risk differences (RDs) for in-hospital complications in the complete cohort. Horizontal bars represent 95% confidence intervals (CIs). Negative MDs and RDs favor wedge-dominant sublobar resection. CTR, consolidation-to-tumor ratio

Real-world 3-year survival

The clinically central ≤ 2 cm analysis included 79 lobectomy and 44 sublobar patients, with five deaths in each group by 36 months. Reverse Kaplan–Meier median follow-up was 45 months. Adjusted 3-year OS was 93.0% after lobectomy and 88.5% after sublobar resection; the difference was not statistically significant (HR 1.73, 95% CI 0.19–15.63; P = 0.558). The adjusted OS difference was − 4.5% points, and the 36-month RMST difference was − 1.42 months (95% CI − 3.51 to 0.48) (Table 4; Fig. 2B).

Table 4.

Adjusted 36-month overall survival in the real-world cohort

Analysis	Lobectomy/Sublobar n	Deaths by 36 months	Adjusted 3-year OS	HR (95% CI)	P value	Additional estimate
≤ 2 cm: overlap weighting (primary real-world analysis)	79/44	5/5	93.0%/88.5%	1.73 (0.19–15.63)	0.558	36-month RMST difference − 1.42 months (− 3.51 to 0.48)
≤ 3 cm: overlap weighting (secondary)	145/78	11/15	91.8%/78.0%	2.89 (1.20–8.02)	0.026	36-month RMST difference − 1.77 months (− 3.45 to − 0.13)
≤ 2 cm: matched sensitivity analysis	35/20	3/2	92.5%/90.0%	1.40 (0.17–11.63)	0.755	Propensity-score matching sensitivity analysis
≤ 3 cm: matched sensitivity analysis	74/40	7/9	90.0%/77.5%	2.42 (0.75–7.87)	0.141	Propensity-score matching sensitivity analysis

HR compares sublobar resection with lobectomy. RMST difference is sublobar minus lobectomy

CI Confidence interval, OS Overall survival, RMST Restricted mean survival time

In the secondary complete ≤ 3 cm analysis, adjusted 3-year OS was 91.8% after lobectomy and 78.0% after sublobar resection (HR 2.89, 95% CI 1.20–8.02; P = 0.026). The up-to-2:1 matched sensitivity estimate was directionally similar (HR 2.42, 95% CI 0.75–7.87; P = 0.141) (Table 4; Supplementary Table S6).

Discussion

The present study defines a clinically useful trade-off for lung cancer surgery in older adults. In the two-center cohort, wedge-dominant sublobar resection was selected for patients who were approximately two years older and carried more recorded comorbidities, particularly coronary heart disease. After balancing measured differences, sublobar resection was associated with chest drainage that was approximately 2.5 days shorter and a 14.4-percentage-point lower risk of hypoproteinemia. The composite in-hospital complication risk was 16.4% points lower as a point estimate, while adjusted total hospital stay was similar. Among patients with tumors ≤ 2 cm, adjusted 3-year OS showed no statistically significant difference between the operations. These findings place sublobar resection as a meaningful option when an older adult’s recovery needs and physiological reserve are central to treatment planning.

This clinical pattern reflects the heterogeneity of ageing. Patients of similar chronological age may differ greatly in cardiopulmonary reserve, multimorbidity, nutrition, mobility, and capacity to recover from a major resection. Earlier studies of lung-cancer surgery in older adults likewise emphasized that comorbidity and physiological condition are more informative than age alone [4, 6, 7, 10–13]. Mery and colleagues reported no clear long-term survival advantage for lobectomy over wedge resection among patients aged 71 years or older in an earlier SEER analysis [8]. Our real-world cohort adds a contemporary perioperative dimension: the patients directed toward sublobar resection were medically more complex, yet their adjusted early recovery profile was favorable in several measurable domains.

The shorter chest-tube duration is especially relevant in older adults because earlier removal can facilitate mobility, respiratory exercise, sleep, and transition out of hospital-based care. The lower adjusted rate of hypoproteinemia points in the same direction, suggesting a smaller immediate physiological burden. These results complement previous reports that minimally invasive thoracic surgery can reduce complications in older adults and the CALGB 140,503 perioperative analysis showing that sublobar procedures can be performed safely in appropriately selected patients [3, 9, 18]. The similar total hospital stay in our cohort does not contradict faster postoperative recovery: the sublobar group entered treatment with more hypertension, diabetes, coronary heart disease, and overall comorbidity, a profile that commonly requires more extensive preoperative optimization.

The survival results are most informative when tumor size and procedure type are considered together. In both the matched SEER cohort and the adjusted real-world cohort, tumors ≤ 2 cm showed no statistically significant OS difference between lobectomy and sublobar resection at the prespecified follow-up horizons. This observation is clinically aligned with CALGB/Alliance 140,503 and JCOG0802/WJOG4607L, which established a role for lung-sparing surgery in carefully selected peripheral tumors [14, 15]. Our study extends that discussion to adults aged 70 years or older, including patients with a comorbidity burden and wedge-dominant practice pattern that are common in routine care.

At the same time, the SEER procedure-specific analysis shows why the term ‘sublobar resection’ should not be treated as a single oncological entity. For tumors ≤ 2 cm, segmentectomy and lobectomy had similar adjusted OS, whereas wedge resection was associated with lower OS. For tumors > 2–3 cm, lobectomy was associated with better OS than either limited procedure. Anatomical segmentectomy preserves parenchyma while retaining segmental anatomy and systematic nodal assessment; wedge resection may be the most proportionate operation for a medically vulnerable patient, but its quality depends strongly on tumor position, margin adequacy, nodal evaluation, and radiological features. Thus, the benefit of preserving lung tissue should be matched to the oncological demands of the individual tumor.

The complementary design helps translate population evidence into geriatric surgical decision-making. SEER supplies the longer survival horizon and distinguishes tumor-size and procedure patterns; the two-center cohort shows who actually receives a lesser resection and how recovery unfolds during the index hospitalization. Together, the findings support a structured preoperative conversation that integrates tumor size and radiology, feasibility of anatomical resection, expected margin and nodal quality, cardiopulmonary comorbidity, physiological reserve, and the patient’s priorities for recovery and independence. This approach is consistent with the broader geriatric principle that relevant health domains should be assessed together rather than allowing chronological age to determine treatment by itself [5–7].

Strengths and limitations

The main strength of this study is the complementary design. Patient-level SEER reanalysis provided long follow-up, complete balance reporting, tumor-size strata, robust matched inference, and procedure-specific estimates. The institutional cohort added clinical stage, CTR, smoking, six cardiopulmonary comorbidities, and detailed in-hospital outcomes that were unavailable in SEER. Based on verified source data, preoperative clinical stage was used for adjustment related to treatment selection. Overlap weighting was applied to retain all participants. Effect estimates together with 95% confidence intervals were presented instead of relying solely on P values for statistical inference.

The study also has limitations, which are consolidated here. Both cohorts were retrospective and treatment selection may have been influenced by factors not captured in the propensity models. The supplied SEER extract did not contain histology, grade, detailed comorbidity, pulmonary function, performance status, frailty, absolute lymph-node counts, margin measurements, recurrence endpoints, or cause-specific death; its stage field was registry-derived rather than explicitly labelled clinical TNM. The institutional cohort was modest in size, and only five participants underwent segmentectomy; accordingly, the real-world sublobar findings mainly describe wedge-dominant practice. The ≤ 2 cm survival analysis included only 10 deaths and produced a very wide 95% CI (HR 1.73, 95% CI 0.19–15.63), indicating limited statistical power. It should therefore be interpreted as hypothesis-generating rather than confirmatory and does not establish statistical non-inferiority. Patient-level lymph-node counts and station counts, margin distance, recurrence, FEV1, DLCO, ASA/ECOG status, frailty, baseline body mass index, serum albumin, formal nutritional screening, sarcopenia, pain, analgesic use, patient-reported outcomes, and post-discharge functional recovery were not consistently available. Because nutritional status can influence both treatment selection and postoperative recovery in older adults [19], its absence represents an additional source of residual confounding. Total hospitalization could not be separated into preoperative preparation and postoperative stay; consequently, the proposed effect of preoperative optimization on length of stay is a clinical explanation rather than a directly measured result. Complications were retrospectively recorded during the index hospitalization without uniform prospective severity grading, and postoperative systemic treatment was available only as a binary variable. These considerations define the scope of the conclusions while preserving the value of the observed recovery and survival patterns for hypothesis-informed, individualized care.

Conclusions

For adults aged 70 years or older with NSCLC ≤ 3 cm, the choice between lobectomy and sublobar resection is a balance between oncologic scope and the burden of recovery. In routine two-center practice, wedge-dominant sublobar resection was selected for older adults with more comorbidity and was associated, after adjustment, with shorter chest drainage and less hypoproteinemia; total hospital stay was not significantly different. Among tumors ≤ 2 cm, adjusted 3-year OS did not differ significantly from lobectomy; these short-term institutional findings primarily reflect wedge-resection practice because 73 of 78 sublobar procedures were wedge resections. The longer-term SEER analysis suggests that the most favorable oncologic trade-off is seen with segmentectomy for tumors ≤ 2 cm, whereas wedge resection and tumors > 2 cm warrant greater caution. Surgical planning should integrate tumor size and radiology, anatomical feasibility, nodal and margin quality, physiological reserve, comorbidity, and the patient’s recovery priorities.

Abbreviations

CI

Confidence interval

COPD

Chronic obstructive pulmonary disease

CTR

Consolidation-to-tumor ratio

HR

Hazard ratio

MD

Mean difference

NSCLC

Non-small cell lung cancer

OS

Overall survival

PS

Propensity score

PSM

Propensity-score matching

RD

Risk difference

RMST

Restricted mean survival time

SEER

Surveillance, Epidemiology, and End Results

SMD

Standardized mean difference

VATS

Video-assisted thoracoscopic surgery

Authors’ contributions

FC and HCY collected and curated the institutional data. FC and WCX drafted the manuscript. FR prepared the figures. JC and YX supervised the study and critically revised the manuscript. XL provided financial support for this research. All authors read and approved the final manuscript.

Funding

This study was funded by the Noncommunicable Chronic Diseases–National Science and Technology Major Project (2025ZD0548504) and the Tianjin Key Medical Discipline Construction Project (TJYXZDXK-3–031 C).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The two-center retrospective study was reviewed and approved by the Medical Ethics Committee of Tianjin Chest Hospital (No. 2024YS-031-01) and the Ethics Committee of Tianjin Medical University General Hospital (No. IRB2024-YX-045-01). Both committees waived the requirement for informed consent because the study used existing clinical data. The SEER component used de-identified registry data under the applicable data-use agreement.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.