Prognostic implications of dominant lesion misidentification in multiple primary lung cancer.

Introduction
Lung cancer is the most commonly diagnosed cancer and the leading cause of cancer deaths worldwide (1). The widespread implementation of computed tomography (CT) screening for lung cancer has triggered a paradigm shift in clinical practice, as the detection of multiple pulmonary nodules has increased, with 0.2–8% ultimately confirmed as multiple primary lung cancers (MPLCs) (2-4).
Clinicians regard surgery as the most effective approach for the treatment strategy of MPLCs (5), However, the core dilemma lies in the complexity of treatment decisions caused by the coexistence of multiple lesions, which creates a conflict between achieving complete oncological resection and preserving pulmonary function (6). In the context of MPLCs, dominant lesions (DLs) are characterized by the highest level of malignancy and represent the primary focus for intervention (5,7,8). Previous studies have consistently demonstrated that the prognosis of MPLCs largely relies on the characteristics of DLs (9-11). Furthermore, current guidelines recommend that therapeutic strategies should prioritize the management of DLs while concurrently addressing secondary lesions (7,8,12).
Preoperative CT serves as a critical basis for surgical strategy formulation. The current clinical decision-making regarding the preoperative identification of the DLs primarily relies on CT characteristics and surgeon experience, whereas the postoperative pathology results are the gold standard to accurately determine DLs (13). This discrepancy may significantly impede accurate assessment of the DLs for clinicians. Misidentification of the DLs inevitably results in erroneous preoperative planning, including insufficient DLs resection, in the staged surgery, prioritization of secondary lesions over DLs resection. These errors may ultimately compromise surgical outcomes and long-term prognosis.
Despite the growing clinical recognition of MPLCs, the prognostic implications of DLs misidentification—a factor that directly induces alterations in surgical strategy—have not been thoroughly investigated to date. This study was designed to systematically examine the effects of such misidentification on the long-term clinical outcomes in patients with MPLC. We present this article in accordance with the STROBE reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-aw-1316/rc).

Methods

Patients
We conducted a single-center retrospective cohort study. The study had registered in the China Research Registration System (No. MR-42-24-038825) (https://www.medicalresearch.org.cn/).
From January 2014 to December 2021, a total of 3,150 consecutive patients undergoing complete surgical resection at Union Hospital, Tongji Medical College, Huazhong University of Science and Technology was initially evaluated. The inclusion criteria were defined as: (I) clinical tumor-node-metastasis (TNM) stage I multiple primary lung adenocarcinoma; (II) at least one tumor was invasive adenocarcinoma; (III) resected tumors were located in different lobes. The exclusion criteria were established as follows: (I) history of other cancer within 5 years; (II) with neoadjuvant treatment; (III) histopathological confirmation of any resected tumor as metastatic disease. Finally, a total of 159 patients with 367 nodules were enrolled (Figure 1). TNM staging was determined according to the 8th edition of the TNM staging system (14).
Demographic, clinicopathological, surgical and survival variables of patients were obtained from our institutional databases. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of the Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (No. UHCT240738). Informed consent was waived in this retrospective study.

CT acquisition and preoperative DLs selection
With regard to the institutional equipment from 2014 to 2021, CT scanning was performed using the SOMATOM Definition AS + multi-slice spiral CT scanner (Siemens Healthineers, Erlangen, Germany) and the Philips Brilliance 64-slice spiral CT scanner. The CT scanners were operated with the following parameters: detector collimation widths of 64×0.6 and 128×0.6 mm, matrix of 512×512, tube voltages of 120 kV, tube current was regulated by an automatic exposure control system, a layer thickness of 1.0–1.5 mm and the bone reconstruction algorithm. The photographed conditions were as follows: a window level of −500 to −700 Hounsfield units (HU) and a window depth of 1,000–2,000 HU as the lung window, and a window level of 30 to 60 HU and a window depth of 350–600 HU as the mediastinal window.
Based on our prior research (15), the preoperative DLs were selected according to the largest mean diameter of pulmonary nodules. Preoperative thin-section CT scans for all enrolled patients underwent meticulous independent review by two physicians (a radiologist and a thoracic surgeon, both with 10 years of experience). Both reviewers systematically measured the mean diameter (average of the long-axis and perpendicular short-axis diameters) of each pulmonary nodule to ensure reliability and objectivity. In cases where the two reviewers initially differed in their identification of the largest nodule (e.g., due to measurement variance or complex morphology), a third senior investigator (a professor in thoracic surgery) served as an adjudicator to review the images and measurements, ensuring a final consensus was reached for each patient.

Surgical strategy
The surgical approach was determined through comprehensive preoperative assessment of lesion characteristics, anatomical location, cardiopulmonary function status, and comorbidity severity.
The following standardized protocols were implemented for MPLCs management: (I) based on a series of studies by the Japan Clinical Oncology Group (JCOG) (16,17), the surgical procedure for the tumor was determined by tumor size and consolidation-to-tumor ratio (CTR). Sublobar resection was performed for tumors ≤2 or ≤3 cm with a CTR ≤0.5, while lobectomy was conducted for tumor exceeding these parameters; (II) for ipsilateral tumors, all tumors were resected completely through single-stage surgery. For contralateral tumors, cardiopulmonary function was comprehensively evaluated: patients with adequate cardiopulmonary reserve underwent single-stage resection, while those with poor cardiopulmonary reserve received staged surgery with an interval; (III) when tumors located in different ipsilateral lobes, bilobectomy was avoided due to potential prognostic compromise. The second lesions consequently underwent sublobar resection followed by adjuvant therapy, while DLs received standard lobectomy.

Determination of the postoperative DLs
Resected lesions were fixed in formalin and sectioned at 5–10 mm. Microscopic assessment of nodules was conducted using hematoxylin and eosin staining. Histopathological assessment was conducted in accordance with the fourth edition of the World Health Organization Lung Tumor Classification (18). Key variables were documented included total tumor size, degree of invasiveness, proportion of each pathological subtype (if present), main pathological type and high-risk pathological factors including spread through air space, visceral pleural invasion, lymphovascular invasion, and lymph node metastasis.
In MPLCs, the postoperative DLs were identified based on the grading of malignancy of the pathology results. The malignancy order was established from low to high as follows: atypical hyperplasia of adenomatous, adenocarcinoma in situ, minimally invasive adenocarcinoma, and invasive adenocarcinoma. For tumors were all diagnosed as invasive adenocarcinoma, criteria of malignancy grading was applied as following criteria: (I) the grading system of invasive pulmonary adenocarcinoma (19): Grade 1, Grade 2, Grade 3. Grade 1: lepidic predominant with no or less than 20% of high-grade patterns; Grade 2: acinar or papillary predominant with no or less than 20% of high-grade patterns; Grade 3: any tumor with 20% or more of high-grade patterns; (II) tumors exhibiting a higher proportion of relative high-grade patterns, despite having the same pathological invasive grading, were classified as postoperative DLs.

Grouping and follow-up
Patients were included in the Inconsistent group (IG) if preoperative misidentification of the DL led to a discrepancy between the surgical procedure performed and the approach that would have been intended based on postoperative retrospective assessment, or prioritized resection of the secondary lesions followed by staged resection of the DLs. Patients not meeting the IG criteria were included in the Consistent group (CG).
All patients received standardized postoperative surveillance. The follow-up intervals were scheduled every 6 months for the first 3 years and annually thereafter. The follow-up period was from the date of the last surgical date until December 2024. Evaluations included physical examinations, hematological testing, thoracic CT and abdomen ultrasound. Additional assessments including abdominal CT, brain magnetic resonance imaging, positron emission tomography, and bone scintigraphy were performed when recurrence was clinically suspected.
Tumor recurrence was defined as: (I) new lesions in pulmonary, osseous, cerebral, or other organ systems confirmed by cytopathological analysis. When biopsy was contraindicated, recurrence was determined through multidisciplinary team discussion based on serial imaging demonstrating interval growth; (II) new nodular/mass lesions within the surgical field exhibiting metabolic activity elevation [elevated maximum standardized uptake value (SUVmax)]; (III) ipsilateral hilar/mediastinal lymphadenopathy [short-axis diameter ≥1 cm with increased fluorodeoxyglucose (FDG) uptake].
The primary endpoints of this study were overall survival (OS) and recurrence-free survival (RFS). OS was defined as the time interval from the last surgery to death from any cause. RFS was calculated from the date of the final surgery to the occurrence of either recurrence or mortality.

Statistical analysis
Categorical variables were represented in numerical and percentage formats, were analyzed using the Chi-squared test. Continuous variables, described as mean and standard deviation, were analyzed using the Student’s t-test. Propensity score matching (PSM) was implemented to mitigate selection bias between two groups. Age, sex, smoking history, chronic obstructive pulmonary disease (COPD) or emphysema, family history of cancer, mean tumor size, location distribution, number of resected nodules, the highest TNM stage and adjuvant therapy of patients were incorporated in the propensity score model. PSM was performing using the nearest-neighbor algorithm with a caliper width of 0.2 standard deviations and the match ratio was 1:2 between two groups. OS and RFS curves were plotted using the Kaplan-Meier method. Log-rank tests were applied for survival comparisons. Univariate and multivariate cox proportional hazards models were conducted in the matched cohort to identify significant prognostic factors. Variables with the univariate analysis P value <0.2 were further assessed for the proportional hazards assumption through examination of the correlation between Schoenfeld residuals and time rank. Variables that did not violate the proportional hazards assumption were ultimately included in the multivariate Cox regression analysis. All statistical tests were two-sided and P values of 0.05 or less were considered statistically significant. All statistical analyses were performed by using SPSS software (version 26.0, IBM, Armonk, NY) and R 4.2.3 software (R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient characteristics
The clinical characteristics of patients are summarized in Table 1. Among 159 patients were analyzed (mean age 57.99±9.68 years), the majority were female (64.78%, 103/159) with no smoking history (84.28%, 134/159) or history of COPD/emphysema (80.50%, 128/159). The mean tumor diameter was 12.58±5.83 mm. Most tumors were distributed ipsilaterally in different lobes (59.12%, 94/159). A total of 148 patients (93.09%, 148/159) underwent resection of 2–3 lesions, with 24 patients (15.09%, 24/159) receiving postoperative adjuvant therapy. Synchronous surgical approaches predominated (76.73%, 122/159), primarily consisting of lobectomy combined with sublobar resection (45.28%, 72/159) or sublobar resection combined with sublobar resection (45.28%, 72/159).

PSM results
Based on the ratio of PSM, the initial cohort comprised 26 patients in IG and 133 patients in CG. After matching, 26 patients in IG and 52 patients in CG were matched. Covariates differences before and after matching are detailed in Tables 2,3. Before PSM, mean tumor size, tumor location, distribution of the highest TNM stage, and adjuvant therapy between two groups were significant differences (P<0.05). After PSM, the mean tumor size in IG was 16.78±5.98 mm and that in CG was 15.36±5.45 mm (P=0.11). In IG, tumors of 10 patients were located in unilateral lobes and 16 patients were located in bilateral lobes. In CG, tumors of 32 patients were located in unilateral lobe and 20 patients were located in bilateral lobes (P=0.09). The difference of the highest TNM stage between two group was not significant (P=0.21). The difference of adjuvant therapy between two group also was not significant (P=0.14).

Postoperative surveillance results
In the pre-matched cohort, IG had comparable 5-year OS to CG (96.15% vs. 98.50%, P=0.11) but significantly worse 5-year RFS (62.86% vs. 92.13%, P=0.001) (Figure 2). After PSM (1:2 matching), the survival outcomes remained consistent: IG maintained significantly worse 5-year RFS (62.86% vs. 91.54%, P=0.004) while showing comparable OS outcomes (96.15% vs. 98.08%, P=0.29) (Figure 3). Figure 4 illustrates a case in which preoperative misidentification of the DL resulted in reduction of surgical scope of the DL, leading to the development of brain metastasis at 4 years after surgery. Figure 5 depicts a patient in whom preoperative misidentification led to prioritized resection of the second lesion followed by DL resection; this surgical sequence was associated with brain metastasis development within 2 years after surgery.

Analysis of recurrence patterns
In the matched cohort (n=78), a total of 9 recurrence events were recorded: IG (n=26) accounted for 7 of these recurrences (34.62%), comprising 4 local recurrences (15.38%: 3 residual lung recurrences and 1 hilar/mediastinal lymph node metastasis) and 3 distant metastases (11.53%: 2 brain and 1 bone metastasis). In contrast, the CG (n=52) experienced only 2 recurrences (3.85%), both of which were local (residual lung recurrence), with no distant metastases observed. Regarding the prognosis analysis of different surgical strategies within the IG: among the 26 patients in the IG, incomplete resection of the true DL occurred in 13 patients. Of these, 2 patients subsequently developed intrapulmonary recurrence and another 2 developed distant metastasis. The remaining 13 patients underwent surgery with an inconsistent sequence (prioritized resection of secondary lesions), among whom 2 experienced intrapulmonary recurrence and 1 developed distant metastasis.
Analysis of the 13 patients with surgical sequence inconsistency revealed that in 8 patients (61.54%), the true DL was radiologically smaller than the secondary lesions and was considered less suspicious for malignancy. In 2 patients (15.38%), the secondary lesions were located in different lobes on the same side, while the DL was in a lobe on the contralateral side; therefore, the side of secondary lesions with the greater tumor burden was resected first. In another 2 patients (15.38%), multiple secondary lesions were clustered within the same lobe on one side, while the DL was located in a lobe on the contralateral side, leading to the prioritized resection of the side of secondary lesions with the greater tumor burden.

Prognostic analysis
In OS analysis, univariate analysis identified COPD or emphysema [hazard ratio (HR) 7.090, 95% confidence interval (CI): 0.642–78.284] and larger mean tumor diameter (HR 1.361, 95% CI: 0.981–1.888) as significant predictors. However, these associations were not significant in multivariate Cox regression after adjustment for confounders (P>0.05) (Table 4).
For RFS, univariate analysis identified four prognostic factors (mean tumor diameter, surgical approach, high-risk pathological factors and adjuvant therapy) were identified as statistically significant factors. Multivariable analysis confirmed inconsistent surgery (HR 5.341, 95% CI: 1.085–26.282, P=0.04) and postoperative adjuvant therapy (HR 5.613, 95% CI: 1.359–23.174, P=0.02) as independent risk factors (Table 5).

Discussion
In our study, we evaluated the prognostic impact of DLs misidentification. No significant differences in OS were observed between the two groups before or after PSM. However, patients in IG exhibited a significantly higher recurrence rate than those in CG. Furthermore, inconsistent surgery and adjuvant therapy were significantly associated with poor RFS and were identified as independent risk factors for RFS. In contrast, no independent risk factors for OS were identified.
Surgical resection represents a crucial therapeutic strategy for MPLCs, as extensively validated in clinical practice (5,7,8). While previous studies suggested that the biological behavior of DLs determined the prognosis of MPLC (9,10,20,21), few researches have specifically focused on the prognostic impact of misidentified DLs leading to alterations in surgical strategy. In this study, we conducted a retrospective analysis following pathological confirmation of postoperative DLs. The results demonstrated no significant difference in OS between two groups, although IG exhibited a higher tumor recurrence rate. The absence of significant OS differences may be attributed to: (I) the inclusion of clinical TNM I stage patients who typically demonstrate favorable 5-year survival rates compared to stages II–IV (22), potentially obscuring the impact of surgical approaches; (II) a mean follow-up duration of 1,541 days that may have been insufficient to capture late recurrence events, particularly in indolent lesions with ground-glass opacity (GGO)-predominant features (23). Notably, the higher recurrence risk in IG aligns with findings from the JCOG0802/WJOG4607L trial, which demonstrated significantly higher recurrence rates following segmentectomy versus lobectomy in peripheral non-small cell lung cancer ≤2 cm (17). This indicates that in our study, misidentification of the DLs resulted in sublobar resection, while this approach may preserve pulmonary parenchyma and improve short-term quality of life, the risk of incomplete resection could serve as a potential trigger for long-term recurrence. Regarding cases where secondary lesions were prioritized for resection due to misidentification of the DL, previous studies have demonstrated a potential association between the surgical interval and patient prognosis in MPLC, suggesting that an interval of 90 to 180 days between procedures may be optimal (24). In the present study, the mean interval between the two surgical procedures for patients undergoing staged surgery was 424.75 days, which is substantially longer than the recommended timeframe. This further suggests that altered surgical sequencing resulting from DL misidentification may serve as a potential risk factor for postoperative recurrence, attribute to the delayed resection of the true DL with higher malignant potential. Furthermore, the overrepresentation of bilateral tumor distribution in the IG underscores how tumor location inherently influences surgical staging decisions. The critical error in these cases was not the staged approach itself, but the flawed hierarchy within that sequence, stemming from the preoperative DL misjudgment.
The optimal surgical strategy for MPLC remains controversial due to the absence of standardized criteria and remains a subject of ongoing debate. While some studies advocate sublobar resection as the primary approach to minimize pulmonary functional impairment from pneumonectomy (25). Current guidelines, however, propose differentiated strategies: for lesions distributed in ipsilateral lobes, lobectomy of the DLs combined with sublobar resection for secondary lesions is recommended, whereas for lesions distributed in bilateral lobes, whereas bilateral lesions necessitate prioritized management of the DLs while strategically addressing secondary lesions8. In our study, prognostic analysis based on the surgical classification criteria established by the JCOG series of trials yielded two critical insights: first, the imperative of accurately identifying the postoperative DLs before surgery, and second, the requirement for adjusted surgical strategies aligned with the biological characteristics of DLs to ensure radical resection of tumor. Furthermore, our findings indirectly support the recommendation for lobectomy for the DLs >2 cm combined with sublobar resection for secondary lesions.
In our survival analysis, univariate analysis identified COPD or emphysema and larger mean tumor diameter as significant risk factors for OS. However multivariate analysis failed to identify any independent risk factors for OS. This discrepancy aligns with the aforementioned reasons, primarily attributable to the inclusion of early-stage lung cancer patients with relatively short follow-up duration and limited survival events. Regarding RFS, both univariate and multivariate analyses identified inconsistent surgery and postoperative adjuvant therapy as independent risk factors. Notably, while some studies supported survival benefits from adjuvant therapy (26,27), our analysis demonstrated a significant association with reduced RFS, findings consistent with another studies focusing on MPLCs (28). We acknowledge that this association likely reflects substantial treatment selection bias rather than a harmful effect of adjuvant therapy. Firstly, our cohort consisted of TNM stage I patients and substantial evidence indicates no survival benefit from adjuvant therapy in stage IA NSCLC (29,30). Notably, patients receiving adjuvant therapy in our study exhibited significantly high-risk features: larger mean tumor diameter and higher proportion of solid/subsolid nodules, both indicative of aggressive tumor biology1. Moreover, clinical decisions regarding adjuvant therapy primarily consider surgery scope and nodule characteristics (size and quantity), so all patients selected for postoperative adjuvant therapy were considered to be at high risk of recurrence. Critically, the limited number of adjuvant therapy recipients (n=16) in the match cohort may exist selection bias. Therefore, we propose that the observed risk association likely reflects inherent high-risk characteristics in patients receiving adjuvant therapy (e.g., large solid nodules, genetic heterogeneity) rather than detrimental effects of the treatment itself.
There are several limitations in this study. First, as a single-center study, the limited sample size (particularly the IG accounting for only 20.16% of the total cohort) compromise the statistical power of the results and introduce potential selection bias, we plan to enroll additional patients for stratified analysis in the future. Therefore, multicenter studies with larger sample sizes are required to validate the generalizability of our findings. Second, MPLC in this study was solely based on pathological examination results. The lack of whole-genome sequencing analysis might have led to the unintended inclusion of intrapulmonary metastatic lesions (which are distinct from MPLCs). Future studies should integrate molecular profiling (e.g., genomic alterations) with pathological features to more accurately distinguish between intrapulmonary metastases and MPLCs. Third, the number of recurrence events, particularly in the matched cohort, was limited. This results in relatively wide confidence intervals in the multivariable analysis and necessitates cautious interpretation of the effect size. Although the association between inconsistent surgery and reduced RFS was significant and consistent across analyses, future studies with larger samples are needed to obtain more precise estimates. Finally, the relatively restricted follow-up duration in the current study may have limited the capture of late-onset survival events (e.g., delayed recurrence or death). In future research, extending the follow-up period would facilitate more comprehensive and robust prognostic assessments.

Conclusions
Accurate preoperative identification of the DL—coupled with the selection of an appropriate DL-guided surgical strategy—could effectively reduce the risk of tumor recurrence. Future investigations should prioritize the development and validation of a standardized, clinically applicable model for the accurate preoperative identification of DLs in this patient population.

Supplementary
The article’s supplementary files as