Significant improvement in overall survival with postoperative traditional Chinese medicine in stage I lung adenocarcinoma: a real-world retrospective cohort study Authors: Boyue Pang, Jianing Chen, Xiang Lv, Mengfan Wang, Li Wang, Chunxia Su, Sufang Zhang Published: 2026 ABSTRACT BackgroundThe clinical efficacy of traditional Chinese medicine (TCM) in the postoperative management of early-stage lung adenocarcinoma (LUAD) remains unclear, particularly in patients who do not receive adjuvant therapy. This real-world retrospective cohort study evaluated the association between postoperative TCM and long-term prognosis in patients with pathological stage I LUAD.MethodsWe retrospectively analyzed pathological stage I LUAD who underwent radical R0 resection and did not receive postoperative adjuvant chemotherapy, radiotherapy, targeted therapy, or immunotherapy. Patients were classified into the TCM group (n=405) and the follow-up group (n=371) according to documented postoperative use of individualized TCM decoctions. Postoperative TCM exposure was defined as continuous individualized herbal decoction treatment for at least 6 months after surgery, prescribed according to syndrome differentiation. Baseline variables included gender, age, smoking status, family history, and T stage. Follow-up was performed through outpatient visits, imaging review, medical record review, and telephone contact until February 16, 2026. The primary endpoint was overall survival (OS), and the secondary endpoint was recurrence-free survival (RFS). The primary analysis used unweighted multivariable Cox regression, and stabilized inverse probability of treatment weighting (sIPTW) analysis was performed as a sensitivity analysis.ResultsA total of 776 patients were analyzed. Over the median follow-up of 92.5 months, 50 death events and 145 recurrence events were documented. After sIPTW adjustment, baseline covariates were well balanced (absolute standardized mean differences <0.01). TCM was associated with improved OS [adjusted hazard ratio (aHR): 0.387, 95% confidence interval (CI): 0.214–0.698; P=0.002], corresponding to a 61.3% reduction in mortality risk. In the TCM group, RFS showed a favorable but non-significant trend (aHR: 0.811, 95% CI: 0.582–1.130; P=0.22). Subgroup analyses demonstrated that the protective effect of TCM on OS was generally consistent across most clinical subgroups, particularly in females (aHR =0.215, 95% CI: 0.061–0.750) and patients aged ≥60 years (aHR =0.413, 95% CI: 0.206–0.828).ConclusionsPostoperative TCM is associated with significantly improved OS in stage I LUAD patients without adjuvant therapy, suggesting its potential role as a complementary maintenance approach requiring prospective validation. Highlight box Key findings • Postoperative traditional Chinese medicine (TCM) was associated with a significant improvement in overall survival (OS) in patients with stage I lung adenocarcinoma (LUAD) who did not receive adjuvant therapy, reducing mortality risk by 61.3%. • The survival benefit was consistent across most subgroups and was particularly pronounced in female and elderly (≥60 years) patients. What is known and what is new? • Patients with stage I LUAD who do not receive adjuvant therapy have limited postoperative management options, and the role of TCM remains controversial due to insufficient high-quality evidence. • This study provides real-world evidence using robust causal inference methods (stabilized inverse probability of treatment weighting and doubly robust Cox models), demonstrating that TCM can improve OS, supporting its role beyond symptomatic management as a survival-enhancing strategy. What is the implication, and what should change now? • TCM may serve as a low-toxicity, feasible maintenance strategy for early-stage LUAD patients ineligible for or not receiving adjuvant therapy. • Future prospective, randomized controlled trials are warranted to validate these findings and to clarify the biological mechanisms underlying the survival benefit, facilitating integration of TCM into standardized postoperative management pathways. Introduction Non-small cell lung cancer (NSCLC) remains the leading cause of cancer-related mortality worldwide, with lung adenocarcinoma (LUAD) representing the most prevalent histological subtype with aggressive biological behaviors. Driven by the widespread clinical implementation of low-dose computed tomography (CT) screening and heightened public health awareness, an increasing proportion of early-stage lung cancers is now being incidentally detected and pathologically diagnosed as stage I LUAD (1,2). While early detection provides a critical window for therapeutic intervention, and radical surgical resection remains the definitive cornerstone of curative intent for these early-stage patients, long-term prognosis remains far from optimal. The underlying risk of occult recurrence and subsequent long-term mortality stubbornly persists, with clinical data indicating that post-surgical 5-year survival rates vary dramatically from 70% to 90%, depending heavily on specific pathological tumor (T) stage and the presence of aggressive high-risk histological or molecular features (3,4). Currently, the optimal postoperative management paradigm for stage I LUAD remains a subject of intense and unresolved clinical debate within the global thoracic oncology community. Under current major international clinical guidelines, such as the National Comprehensive Cancer Network (NCCN) and the Chinese Society of Clinical Oncology (CSCO) guidelines, routine radiological surveillance without any adjuvant systemic intervention is established as the standard of care for stage IA disease, and active observation is mandated for the vast majority of stage IB cases that lack high-risk criteria (5-7). Consequently, a prominent majority of these post-resection patients are relegated to a passive “wait-and-watch” surveillance protocol, receiving only routine, intermittent radiological follow-up (8). However, a non-negligible and clinically significant proportion of patients within this “low-risk” tier still inevitably experiences disease relapse or distant organ mortality driven by microscopic residual disease or dormant occult micro-metastases, highlighting a critical, long-standing unmet medical need for effective, low-toxicity maintenance strategies to optimize long-term prognosis and eliminate subclinical lesions during this prolonged therapeutic hiatus (9). This clinical dilemma is further intensified by ongoing controversies regarding whether patients with stage IB disease harboring high-risk features (such as visceral pleural invasion, lymphovascular space invasion, or high-grade solid/micropapillary components) truly derive a definitive overall survival (OS) benefit from aggressive adjuvant interventions versus observation alone, as the potential systemic toxicities often outweigh the marginal therapeutic gains (7,9). To bridge this therapeutic vacuum, traditional Chinese medicine (TCM), characterized by highly individualized herbal prescriptions, dynamic syndrome differentiation (Bian Zheng Lun Zhi), and a holistic framework aimed at “strengthening body resistance” (Fuzheng), has been historically and widely utilized in China as a crucial complementary and integrative therapy for lung cancer management (10). TCM treatment, including herbal decoctions and/or Chinese patent medicines, has been reported to exert anti-tumor recurrence and metastasis effects, while also improving patients’ quality of life and immune function. The major therapeutic category consists of Fuzheng Quxie (strengthening the body resistance and eliminating pathogenic factors) formulas, with commonly used herbs including Huangqi (Astragali Radix), Yiyiren (Coicis Semen), Dangshen (Codonopsis Radix), Shanyao (Dioscoreae Rhizoma), Shijianchuan (Salvia chinensis Benth.), Baihuasheshecao (Hedyotis diffusa Herba), and Banzhilian (Scutellaria barbata Herba) (11). Cumulative translational evidence suggests that TCM formulas do not merely act as palliative agents, but can actively modulate the complex tumor microenvironment, enhance host systemic immune surveillance, and potentially inhibit the breakout of micro-metastasis (12,13). In the practical landscape of early-stage thoracic oncology, TCM is frequently administered during the critical postoperative recovery phase to accelerate physiological healing. In the context of the present investigation, “TCM” specifically refers to individualized oral herbal decoctions sequentially adjusted and prescribed by senior oncologists based on the core philosophical framework of “FuzhengQuxie” (strengthening body resistance and eliminating pathogenic factors). This therapeutic strategy relies on tailored formulas predominantly rich in Qi-boosting botanicals (such as Radix Astragali) and Yin-nourishing agents (such as Radix Ophiopogonis). Recent high-impact pharmacological investigations have further elucidated that the proposed biological mechanisms of these compound formulas involve systematically restoring T-cell mediated adaptive immune responses, directly regulating the polarization and cytokine secretion profiles of tumor-associated macrophages to reverse immunosuppression (12), and inducing a complex functional crosstalk between autophagy and ferroptosis to suppress the proliferation of residual non-small cell lung cancer cells (13), thereby effectively locking disseminated tumor cells into a state of prolonged microenvironmental dormancy. Nevertheless, previous clinical research has overwhelmingly focused on evaluating TCM as a concurrent adjunct to cytotoxic chemotherapy or radiation, making it exceedingly difficult to methodologically isolate the independent, standalone maintenance effect of TCM (14). Furthermore, existing retrospective studies in this integrative field frequently suffer from inherent selection bias and confounding by indication, where real-world treatment assignment is heavily distorted by baseline prognostic expectations, familial socioeconomic status, and individual health-seeking behaviors (15). To responsibly address these historical methodological limitations and provide a complementary clinical perspective within the controversial post-surgical observation window, we conducted a retrospective cohort study focusing on a sample of 776 patients with pathological stage I LUAD who received no postoperative adjuvant systemic therapies. This study aims to explore the potential clinical association between long-term postoperative TCM maintenance intervention and long-term prognostic trends, specifically OS and recurrence-free survival (RFS). By restricting our evaluation to this “adjuvant-free” cohort tracked within a well-curated, single-center clinical registry, we seek to provide tentative, descriptive real-world evidence regarding the feasible value of TCM as a low-toxicity, supportive maintenance option during the standard observation period, thereby offering a preliminary reference to guide prevention pathways in early-stage LUAD management. We present this article in accordance with the STROBE reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0472/rc). Methods Patient selection and data collection This retrospective cohort study included patients diagnosed with pathological stage I LUAD at Shanghai Pulmonary Hospital from September 1, 2008 to March 1, 2018. To investigate the therapeutic effect of TCM, we applied the following inclusion criteria: (I) patients with complete baseline characteristics and pathological records; (II) patients with pathologically confirmed stage I LUAD according to the 9th edition of the tumor-node-metastasis (TNM) classification for NSCLC; (III) patients with no history of other primary malignancies, and no diagnosis of other active or synchronous malignant tumors within the previous 5 years or currently; (IV) patients who did not receive adjuvant chemotherapy, radiotherapy, targeted therapy, or immunotherapy following surgery; (V) patients who were not enrolled in other interventional clinical trials or receiving any investigational treatments during the study period. (VI) patients in the TCM group who received continuous, individualized postoperative TCM decoctions for a minimum duration of 6 months; (VII) patients with a follow-up duration of more than 3 months and documented integrity of survival information. Exclusion criteria were: (I) patients with incomplete baseline characteristics or pathological records; (II) patients with pathologically confirmed non-LUAD histology or those whose disease stage was not classified as stage I according to the 9th edition of the TNM classification; (III) patients with a history of another primary malignancy, or those diagnosed with other malignant tumors within the past 5 years or currently; (IV) patients who received any form of adjuvant therapy following surgery, including chemotherapy, radiotherapy, targeted therapy, or immunotherapy; (V) patients currently enrolled in other interventional clinical trials or receiving investigational treatments; (VI) patients who did not receive continuous postoperative individualized TCM decoctions (for TCM group); (VII) patients with a follow-up period of less than 3 months, or with undocumented survival information. Baseline variables included gender, age, smoking status, family history, and T stage. This was a total-population consecutive registry cohort, with no sampling applied, resulting in a fixed sample size. For historical cases resected prior to the enforcement of the current TNM system, two senior pathologists independently reviewed the chest CT images and microscopic pathological descriptions to manually restage all cases, thereby ensuring objective criteria and cross-cohort comparability strictly according to the 9th edition of the TNM classification for NSCLC. The flowchart for this study is illustrated in Figure 1. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of Shanghai Pulmonary Hospital (No. L22-401) and individual consent for this retrospective analysis was waived. Flowchart of patient screening and enrollment. LUAD, lung adenocarcinoma; N, number; TCM, Traditional Chinese Medicine. Treatment and follow-up Patients were divided into two groups: the TCM group (TCM combined with routine follow-up) and the follow-up group (routine follow-up only). TCM treatment consisted of individualized herbal formulas prescribed by senior oncologists based on syndrome differentiation. Treatment decisions were made based on patient self-selection after consultation with their thoracic oncologists and TCM physicians, introducing potential clinical and socioeconomic selection mechanisms partly. But TCM treatment in this study was not standardized but guided by the principle of syndrome differentiation (“Bian Zheng Lun Zhi”). Prescriptions were individually tailored and dynamically adjusted according to patients’ symptoms, tongue and pulse findings, and overall clinical status during follow-up. As a result, even for the same patient under the same physician, herbal compositions may vary over time. Therefore, no uniform fixed formula was used in this cohort, and TCM exposure in this study should be interpreted as a longitudinal treatment strategy rather than a standardized intervention. Survival outcomes were determined based on standardized follow-up records. Follow-up was performed through outpatient visits, imaging review, medical record review, and telephone contact until February 16, 2026. Postoperative follow-up was conducted according to a standardized clinical surveillance protocol. During the first 3 years after surgery, patients were followed every 3–6 months with chest CT scans and routine blood tests. A comprehensive systemic examination was performed annually. After 3 years, follow-up was conducted annually with similar imaging and laboratory assessments. A more extensive whole-body evaluation was performed at the 5-year follow-up visit. The primary endpoint was OS, defined as the time from surgery to death from any cause or the last follow-up. The secondary endpoint was RFS, defined as the time from surgery to disease recurrence, death, or the last follow-up. Propensity score (PS) weighting To minimize selection bias and potential confounding factors between the two groups, sIPTW was employed. A logistic regression model was used to calculate the PS for each patient, incorporating the following baseline covariates: gender, age (<60 vs. ≥60 years), smoking status, family history of malignancy, and pathological T stage (T1a, T1b, T1c, T2a). To mitigate the influence of extreme weights and preserve the original sample size, stabilized weights (SW) were derived as follows: ❖ For the TCM group: SW = P/PS; ❖ For the TCM group: SW = P/PS; ❖ For the follow-up group: SW = (1 − P)/(1 − PS). ❖ For the follow-up group: SW = (1 − P)/(1 − PS). P is the marginal probability of receiving TCM. The balance of covariates was assessed using absolute standardized mean differences (ASMDs), with a threshold of <0.1 indicating a negligible imbalance. Statistical analysis To mitigate potential immortal time bias, a 6-month landmark analysis design was applied, whereby patients experiencing death or recurrence within 6 months post-surgery were excluded from the analysis cohort. Primary analysis was performed on the unweighted sample using univariable and standard multivariable Cox proportional hazards regression to estimate unadjusted and adjusted hazard ratio (aHR). sIPTW adjustments and doubly robust estimations were subsequently conducted as sensitivity analyses to assess the robustness of the primary findings. Weighted Kaplan-Meier curves and the weighted log-rank test were utilized to estimate and compare survival distributions between the two groups. To enhance the robustness of our results, a doubly robust estimation was implemented by incorporating potential confounders or clinically significant variables, specifically T stage and tumor differentiation, into a weighted Cox proportional hazards model. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated accordingly. Furthermore, subgroup analyses were performed across clinical strata to assess the consistency of the treatment effect. Missingness in key covariates was low and was addressed using complete-case analysis without imputation. All statistical computations were conducted in R (version 4.5.2) using the survival, survey, survminer and cobalt packages. A two-sided P<0.05 was defined as the threshold for statistical significance. Results Patient characteristics and PS weighting A total of 776 patients with stage I LUAD managed without adjuvant therapy were enrolled in this study, comprising 405 patients in the TCM group (TCM combined with routine follow-up) and 371 patients in the follow-up group (routine follow-up only). Over the follow-up period, 145 recurrence events (81 in the follow-up group, 64 in the TCM group) and 50 death events (34 in the follow-up group, 16 in the TCM group) were recorded. The median follow-up duration for the entire analyzed cohort was 92.5 months (range, 3.2–206.0 months). The baseline confounders adjusted for via sIPTW and subsequent doubly robust Cox regression included gender, age, smoking status, family history of malignancy, pathological T stage, and tumor differentiation. As delineated in Table 1 and the Love plot (Figure 2), the original cohort exhibited several baseline imbalances, most notably in pathological T stage (SMD =0.091). Following sIPTW adjustment, ASMDs for all covariates (including gender, age, smoking status, family history, T stage, and tumor differentiation) were reduced to <0.01. This indicates that the two groups were highly comparable within the weighted pseudo-population. Furthermore, the distribution of SW was appropriately centered around 1.0 without extreme outliers (Figure S1), ensuring the numerical stability and reliability of the subsequent causal estimates. Table 1Baseline characteristics between the TCM and follow-up groups in the original and sIPTW-weighted cohortsCharacteristicCategoryFollow-up (n=371)TCM (n=405)Original SMDWeighted SMDGenderMale152 (41.0)181 (44.7)0.075<0.01Female219 (59.0)224 (55.3)Age group<60 years181 (48.8)181 (44.7)0.081<0.01≥60 years190 (51.2)224 (55.3)SmokingNon-smoker328 (88.4)353 (87.2)0.037<0.01Smoker43 (11.6)52 (12.8)Family historyNo362 (97.6)398 (98.3)0.052<0.01Yes9 (2.4)7 (1.7)T stageT1a108 (29.1)99 (24.5)0.091<0.01T1b175 (47.2)203 (50.1)T1c62 (16.7)67 (16.5)T2a26 (7.0)36 (8.9)Data are presented as n (%). sIPTW, stabilized inverse probability of treatment weighting; SMD, standardized mean differences; T, tumor; TCM, traditional Chinese medicine. Baseline characteristics between the TCM and follow-up groups in the original and sIPTW-weighted cohorts Data are presented as n (%). sIPTW, stabilized inverse probability of treatment weighting; SMD, standardized mean differences; T, tumor; TCM, traditional Chinese medicine. Covariate balance assessment by stabilized inverse probability of treatment weighting. T, tumor. Impact of TCM on RFS During the study period, 145 recurrence events were documented. In the unweighted cohort, there was no statistically significant difference in RFS between the TCM and follow-up groups (log-rank P=0.28). The Kaplan-Meier curves showed a numerically favorable but non-significant trend in the TCM group (Figure S2). For RFS, univariate analysis indicated that smoking history (HR =1.71, 95% CI: 1.12–2.62, P=0.01) and higher T stages, including T1b (HR =1.95, 95% CI: 1.19–3.22, P=0.009), T1c (HR =3.77, 95% CI: 2.21–6.44, P<0.001), and T2a (HR =2.90, 95% CI: 1.50–5.59, P=0.002), were significantly correlated with a higher risk of recurrence. No statistically significant difference in RFS was observed between the study groups in the univariate analysis (HR =0.83, 95% CI: 0.60–1.16, P=0.28). In the multivariable Cox analysis, T stage was identified as the sole prognostic factor for RFS. Specifically, compared with T1a, patients with T1b (HR =1.86, 95% CI: 1.12–3.09, P=0.02), T1c (HR =3.44, 95% CI: 1.99–5.94, P<0.001), and T2a (HR =2.66, 95% CI: 1.36–5.20, P=0.004) disease exhibited significantly shorter RFS (Table S1). Although the TCM group exhibited a favorable trend toward improved RFS compared to the Follow-up cohort (Figure 3), this difference did not reach statistical significance after sIPTW adjustment. The aHR for RFS was 0.811 (95% CI: 0.582–1.130; P=0.22). These results suggest that while TCM intervention offers a profound long-term survival advantage, its efficacy in extending the time to initial recurrence within this specific stage I cohort requires further validation through larger or prospective studies. Kaplan-Meier curves for recurrence-free survival in the sIPTW-adjusted cohort. sIPTW, stabilized inverse probability of treatment weighting; TCM, Traditional Chinese Medicine. Impact of TCM on OS Over the follow-up duration, 50 death events were recorded. In the unweighted cohort, OS was significantly better in the TCM group compared with the follow-up group (log-rank P=0.005). Kaplan-Meier curves demonstrated a consistent survival advantage in the TCM group over long-term follow-up (Figure S3). Regarding OS, both univariate and multivariable Cox regression analyses (Table S1) demonstrated that TCM therapy was a protective factor compared with follow-up alone (multivariable HR =0.40, 95% CI: 0.22–0.73, P=0.003), significantly reducing the risk of death. Additionally, female sex was identified as another protective factor for OS (multivariable HR =0.43, 95% CI: 0.22–0.81, P=0.01). Conversely, advanced T stages (with T1a as the reference) were shown to be independent risk factors for poorer OS in the multivariable model, where T1c (HR =3.03, P=0.02) and T2a (HR =3.35, P=0.02) stages were associated with a significantly increased risk of death. Although advanced age (≥60 years) and a history of smoking were significantly correlated with unfavorable OS in univariate analyse, neither retained independent prognostic significance after adjusting for confounding factors (both P>0.05). The sIPTW-adjusted Kaplan-Meier analysis demonstrated a significant OS benefit for patients in the TCM group compared to the follow-up alone cohort (Figure 4). In the doubly robust Cox proportional hazards model, postoperative TCM intervention was identified as a significant protective factor for OS. The aHR was 0.387 (95% CI: 0.214–0.698; P=0.002), translating to a 61.3% reduction in the risk of mortality for patients in the TCM group. Kaplan-Meier curves for overall survival in the sIPTW-adjusted cohort. sIPTW, stabilized inverse probability of treatment weighting; TCM, Traditional Chinese Medicine. Subgroup analysis Subgroup analyses were performed to evaluate the consistency of the TCM treatment effect on OS across various clinical strata (Figure 5). The results demonstrated a remarkably consistent trend favoring TCM intervention, as the point estimates of HR for OS remained consistently below 1.0 across all analyzed subgroups. A statistically significant survival benefit was particularly pronounced among female patients (aHR =0.215; 95% CI: 0.061–0.750) and patients aged ≥60 years (aHR =0.413; 95% CI: 0.206–0.828). Although the aHR for male patients (0.496) and younger patients (<60 years, 0.350) showed a strong protective trend, these did not reach statistical significance, likely due to the distribution of death events within these strata. The therapeutic advantage of TCM was maintained across all pathological T stages (T1a, T1b, T1c, and T2a). Notably, the most substantial risk reduction was documented in the T1a subgroup (aHR =0.178; 95% CI: 0.022–1.450). While the wide CIs in these subgroups (T1a to T2a) crossed the unity line, which was primarily attributed to the limited number of mortality events in early-stage disease, the consistently low aHR values (all <0.45) underscore the robust and broad clinical utility of TCM as a maintenance strategy for stage I LUAD. Subgroup analysis of overall survival comparing postoperative TCM with follow-up in the sIPTW-adjusted cohort. CI, confidence interval; sIPTW, stabilized inverse probability of treatment weighting; T, tumor; TCM, traditional Chinese medicine. Discussion In this retrospective cohort study of 776 stage I LUAD patients managed without adjuvant therapy, postoperative TCM intervention was associated with a significant improvement in OS. By employing sIPTW to emulate an RCT setting, we demonstrated that TCM yielded a 61.3% reduction in the risk of mortality (aHR =0.387). Notably, while RFS exhibited a favorable trend, the difference did not reach statistical significance. Stratified analyses demonstrated consistent survival improvements of TCM across all evaluated subgroups. Notably, female patients and those with T1a tumors derived the most substantial risk reduction (aHR of 0.215 and 0.178, respectively), underscoring the broad clinical utility of TCM as a maintenance strategy for stage I LUAD. These findings provide robust real-world evidence supporting TCM as a viable, maintenance strategy to bridge the current ‘therapeutic vacuum’ in early-stage lung cancer management. One of the most striking findings of our study is the discrepancy between OS and RFS. While TCM significantly prolonged OS, its impact on preventing early recurrence (RFS) was less pronounced. This phenomenon, often termed the “OS-RFS Paradox”, suggests that the therapeutic value of TCM may extend beyond mere tumor shrinkage or immediate recurrence prevention (16,17). We hypothesize that TCM primarily improves OS through two mechanisms. One plausible explanation lies in the concept of tumor-host equilibrium. Unlike cytotoxic chemotherapy, which aims for the immediate eradication of micro-metastatic tumor cells, TCM characterized by its “FuzhengQuxie” (strengthening the body to eliminate pathogenic factors) philosophy may function as a biological response modifier (18,19). It potentially shifts the balance of the tumor microenvironment toward a state of equilibrium, effectively maintaining residual tumor cells in a dormant phase. This discrepancy suggests that TCM may induce a state of tumor dormancy in residual micro-metastases, effectively shifting the disease toward a ’stable coexistence’ model (19). In this framework, while TCM may not entirely prevent radiological recurrence, it significantly attenuates the biological aggressiveness of recurring clones, thereby slowing disease progression and extending post-recurrence survival (20). Moreover, the significant OS benefit likely reflects an improvement in PRS. By modulating systemic immunity and preserving organ function, TCM may enhance the host’s physiological resilience and “chemo-fitness” (19,21). This pre-conditioning could render patients more tolerant of, or responsive to, subsequent salvage therapies, such as targeted therapy or immunotherapy, upon recurrence (22,23). Some studies have indicated that certain TCM compounds can reverse multidrug resistance or synergize with immune checkpoint inhibitors by remodeling the immune landscape, such as inhibiting the polarization of tumor-associated macrophages (24-26). Consequently, even if recurrence occurs, the disease trajectory in TCM-treated patients may be characterized by an indolent progression rather than a rapid decline, ultimately translating into a superior long-term survival advantage (26,27). This holistic modulation explains why TCM can profoundly reduce mortality risk by 61.3% without a commensurate delay in initial recurrence, suggesting that for stage I LUAD, OS may be a more comprehensive and clinically meaningful endpoint than RFS for evaluating the true value of herbal maintenance therapy. The present study demonstrated a significant OS benefit associated with TCM use, while no statistically significant improvement was observed in RFS. While speculative biological mechanisms such as the induction of tumor cell dormancy and ‘tumor-host equilibrium’ have been proposed, alternative non-biological explanations must be seriously considered. First, this disconnect may stem from unmeasured post-recurrence salvage therapies. Patients choosing long-term TCM often demonstrate higher compliance and stronger health-seeking behaviors, rendering them more likely to receive advanced, costly subsequent lines of targeted or immunotherapies upon disease relapse. Second, a robust healthy-user bias and superior general supportive care within the TCM cohort could substantially prolong post-progression survival without directly delaying initial recurrence. Therefore, the observed 61.3% mortality risk reduction must not be viewed as a pure standalone anti-tumor therapeutic effect. Whether some patients with stage I LUAD should receive adjuvant chemotherapy remains controversial, especially for those without high-risk features (5,7,28). This clinical paradigm creates a significant “therapeutic vacuum”, a prolonged “wait-and-watch” period characterized by intensive radiological surveillance but a lack of proactive intervention to mitigate recurrence risk. Our study addresses this critical unmet need by demonstrating that TCM can serve as an effective maintenance therapy. Unlike cytotoxic agents, which are often poorly tolerated in early-stage patients, TCM may offer a low-toxicity alternative that transforms the passive observation period into a proactive phase of secondary prevention and host-environment optimization. Our subgroup analysis provides further insights into the optimal candidates for TCM intervention. We observed that female patients and those with T1a stage disease derived the most substantial survival benefits. The heightened sensitivity of female patients (aHR =0.215) may be attributed to a distinct biological landscape. LUAD in women is frequently associated with specific driver mutations (such as EGFR) and may be influenced by estrogen-mediated immune modulation (29-31). TCM compounds, often containing phytoestrogens or immunomodulatory polysaccharides, may synergize with these endogenous profiles to enhance anti-tumor immunity (32,33). Furthermore, the exceptional outcome in the T1a subgroup (aHR =0.178) underscores the importance of early intervention. At this earliest invasive stage, the initial tumor burden and clonal heterogeneity are at their nadir, providing a highly favorable “window of opportunity” for TCM to recalibrate the host microenvironment (33,34). By intervening when the “pathogenic Qi” is weak and the systemic immune surveillance is still relatively intact, TCM may more effectively prevent the transition of dormant micro-metastases into lethal recurrences (33,35). Elderly patients aged ≥60 years derived a statistically significant survival benefit from TCM intervention. Elderly patients with stage I LUAD represent a clinically vulnerable population. Due to age-related physiological decline, a higher prevalence of comorbidities, and diminished chemo-fitness, these individuals are often suboptimal candidates for aggressive postoperative interventions (36-38). In this context, individualized TCM decoctions offer a compelling alternative. Unlike cytotoxic chemotherapy, which may impose treatment toxicity in elderly patients, TCM exerts its anti-tumor effects through subtle immunomodulation and host-environment optimization (19,39). Our results provide refined insights into the clinical management of elderly patients with stage I LUAD, suggesting a strategic shift from passive surveillance to active TCM maintenance therapy. These findings suggest that the current “wait-and-watch” strategy may inadvertently miss the most responsive period for immune-modulatory intervention, advocating for a shift toward a risk-stratified maintenance approach in early-stage LUAD management. The reliability of our conclusions is significantly bolstered by the implementation of sIPTW and a doubly robust statistical framework. Retrospective oncology studies investigating TCM are frequently scrutinized for “selection bias” or the “healthy user effect”, the hypothesis that patients with better performance status or more favorable prognostic expectations are more inclined to seek integrative therapies. To rigorously address this potential for confounding by indication, we employed a PS-based weighting strategy to create a balanced pseudo-population. By assigning weights to each individual, we effectively decoupled the treatment assignment from baseline prognostic factors, such as pathological T stage and tumor differentiation, thereby emulating the structural equilibrium of a RCT setting. The achievement of excellent balance (all ASMDs <0.01) ensures that the observed survival divergence is not a spurious association driven by initial clinical imbalances. In addition, the integration of a doubly robust estimation adds a secondary layer of protection against model misspecification. By simultaneously adjusting for covariates in both the weighting and the outcome regression models, we ensured that as long as one of the models is correctly specified, the treatment effect estimates remain unbiased. The fact that the OS benefit (aHR =0.387) remained significant and stable even after such stringent adjustment underscores the intrinsic potency of the TCM intervention in this cohort. This methodology transcends simple correlation analysis, moving toward causal inference by isolating the contribution of TCM to long-term survival. Such a robust statistical foundation provides a high level of real-world evidence, offering a credible basis for clinicians to consider TCM as a validated component of the postoperative management spectrum for stage I LUAD, especially when prospective RCT data are still emerging. Despite the promising survival benefits observed, this study also has several limitations. First, TCM treatment in this cohort was not standardized but guided by syndrome differentiation (“Bian Zheng Lun Zhi”), with prescriptions individually tailored and dynamically adjusted over time according to patients’ symptoms and clinical status. As a result, TCM exposure reflects a longitudinal treatment strategy rather than a uniform intervention. the heterogeneity of TCM prescriptions, while reflective of real-world “syndrome differentiation” (Bian Zheng Lun Zhi), complicates the identification of specific bioactive compounds responsible for the observed OS advantage. Future research incorporating standardized herbal formulas or high-throughput metabolomics is required to isolate the core pharmacological drivers of this effect. Second, treatment allocation was based on patient preference and physician consultation in a retrospective real-world setting, which may introduce selection bias and confounding by indication. Although we used sIPTW to adjust for measured baseline covariates, residual confounding from unmeasured factors, such as socioeconomic status, health-seeking behavior, and nutritional status, cannot be excluded. Notably, although many TCM treatments are now partially reimbursed under the national medical insurance system in China, thereby reducing out-of-pocket costs, the potential influence of socioeconomic factors may still persist. Therefore, causal inference should be interpreted with caution, and our findings require validation in prospective multicenter studies. Third, the molecular landscape, particularly the EGFR mutation status and other actionable drivers, was not systematically analyzed for the entire cohort. In the era of precision oncology, molecular subtypes significantly influence PRS through the differential efficacy of targeted therapies. While our study focused on the role of TCM in an “adjuvant-free” setting, we cannot definitively rule out whether a higher prevalence of sensitive mutations in the TCM group contributed to the superior OS. Finally, as a single-center study, our findings require validation in multi-center prospective RCT to ensure external validity. The remarkable survival advantage, particularly in the T1a subgroup, underscores the need for further research to elucidate the mechanisms of TCM-induced immune remodeling and tumor dormancy in stage I LUAD. Conclusions In conclusion, postoperative TCM intervention was associated with significantly improved OS in patients with stage I LUAD who did not receive adjuvant therapy. TCM may represent a promising low-toxicity maintenance strategy, shifting the clinical focus from merely preventing recurrence to prolonging survival and broadening the management perspective for early-stage lung cancer. While herbal prescription heterogeneity limits definitive causal conclusions, these findings suggest that TCM may represent a promising complementary management option to bridge the observation period. Supplementary The article’s supplementary files as Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0472/prf