Commentary on “Prognostic heterogeneity of lung cancer associated with cystic airspaces and identification of novel imaging subtypes: a multi-center cohort study”

Dear Editor, We read with great interest the important study by Wang et al, published in the International Journal of Surgery, regarding the prognostic heterogeneity of lung cancer associated with cystic airspaces (LCCA)[1]. This study leverages large-scale, multi-center data to innovatively propose an imaging-based classification of LCCA grounded in solid components, and it clearly elucidates the correlation with long-term patient survival. This work provides an important reference for risk stratification and clinical management of LCCA. However, we believe that three aspects merit further discussion. This letter complies with the TITAN Guidelines 2025 on transparency of artificial intelligence use; not a single AI tool was used in its creation[2]. First, while the study excellently delineated the poor prognostic phenomenon associated with the solid nodule (SN) subtype, the underlying biological drivers remain to be elucidated. In the era of precision medicine, it is imperative to investigate its intrinsic mechanisms. We hypothesize that the highly aggressive imaging features of the SN subtype may be associated with specific genetic mutation profiles and a distinct tumor immune microenvironment (TIME). Indeed, existing studies have confirmed that lung adenocarcinomas harboring KRAS co-mutations (e.g., KRAS/TP53) often exhibit more aggressive behavior[3]. Furthermore, lineage differences within the tumor immune microenvironment, such as the immune-inflamed phenotype or the immune-excluded phenotype, directly determine the tumor’s response to immune checkpoint inhibitors[4]. Future research urgently needs to utilize technologies such as spatial transcriptomics to establish a direct mapping relationship between imaging features and molecular and immune profiles[5]. This will not only elucidate the biological basis for the poor prognosis of the SN subtype but also enable the prospective identification of patient subpopulations likely to respond to specific targeted or immunotherapies. Second, while all patients in the study cohort underwent surgery, the analysis did not stratify based on specific treatment regimens to assess their modulating effect on prognosis. This may obscure the substantial influence that treatment decisions themselves exert on patient outcomes. First of all, for early-stage LCCA, the extent of surgical resection – specifically the choice between lobectomy and sublobectomy – is a key variable influencing long-term survival. This has been confirmed in specific patient populations by existing research[6]. Furthermore, for high-risk SN subtypes, the administration and efficacy of postoperative adjuvant therapy are crucial. The IMpower010 study has established adjuvant immunotherapy as a standard of care in selected patients with non-small cell lung cancer[7]. Therefore, future analyses must incorporate treatment modalities as interaction terms in regression models. This approach will clarify which surgical methods and adjuvant treatment regimens provide the greatest survival benefit for patients with the SN subtype, thereby offering clinically actionable treatment recommendations with genuine guiding significance. Third, the ultimate value of a clinical study lies in its ability to be conveniently applied in practice. The study identified multiple prognostic factors. Integrating these to develop and validate a user-friendly visual prognostic nomogram would significantly enhance its clinical utility[8]. More importantly, this study was based on a single preoperative CT scan, representing a static assessment at a single time point. However, the dynamic evolution of LCCA over time, particularly in lesions containing ground-glass components, may harbor more complex prognostic information. Studies have shown that the volume doubling time (VDT) of pulmonary nodules is a key indicator for predicting their malignancy potential and prognosis[9]. Future research directions should include the development of dynamic risk models. Utilizing sequential CT imaging to quantify the growth rate or morphological changes of solid components, known as delta-radiomics analysis, may enable earlier and more accurate identification of high-risk patients requiring active intervention compared to single timepoint classification[10]. This approach could optimize decision-making regarding the timing of active surveillance versus intervention. In conclusion, the study by Wang et al serves as a crucial cornerstone for understanding the heterogeneity of LCCA. We are confident that by delving deeper into its biological origins, clarifying its complex interactions with treatment modalities, and developing dynamic, practical clinical tools, we can maximize the value of this imaging-based stratification system.