Introduction
Lung cancer is the leading cause of cancer death in males and the second leading cause of cancer death in females worldwide (Torre et al.
2016). The 5-year survivals of patients with pathological stage IA after surgery are 92%, 86%, and 81% for stages IA1, IA2, and IA3, respectively (Nowak et al.
2016). Among early-stage patients, 23–29.1% develop recurrence despite curative resection (Kelsey et al.
2013; Taylor et al.
2012). According to the American Society of Clinical Oncology adjuvant therapy guideline for resected non-small-cell lung cancers (NSCLCs), adjuvant chemotherapy is recommended for patients with stage IIA, IIB, or IIIA disease who have undergone complete surgical resection (Kris et al.
2017). However, the indications for postoperative chemotherapy for stage I patients are still controversial (Bradbury et al.
2017). The decision of which stage IB patients to treat with adjuvant chemotherapy is not as clear as in other stages. Additional prognostic markers beyond stage are needed to determine who may be in need of adjuvant chemotherapy or more aggressive treatment approach.
Previous studies have considered various assessment methods, including grading systems based on certain pathological, architectural, or pathological characteristics, and genomic profiling, for investigating stage I NSCLC patients with a high chance of early relapse (Zhao et al.
2015; Kadota et al.
2012; Ooki et al.
2017; Kratz and Jablons
2009). In the present study, we constructed a novel but concise prognostic model based on conventional clinical and pathological characteristics to stratify patients who underwent complete anatomical resection into different risk groups for developing early recurrence. Using the model, we were able to identify a subset of stage I patients with a higher risk of recurrence and poor survival who may be in need of more aggressive adjuvant treatments or closer follow-up strategies.
Discussion
Patients with early-stage NSCLC after complete surgical resection are at substantial risk for recurrence. The role of adjuvant chemotherapy for stage I patients is still controversial because previous randomized trials have not reported consistent results (Kris et al.
2017; Bradbury et al.
2017). In daily practice, the tumor–node–metastasis (TNM) system is used to differentiate the prognoses of patients with NSCLC. However, patients with the same TNM stage may have completely different outcomes (Chansky et al.
2009). Because of the heterogeneous nature of NSCLC, it would be imprecise to predict survival using the TNM staging system alone. For this reason, various studies have identified early-stage NSCLCs with poor survival that could potentially benefit from adjuvant treatment.
In our study, the presence of necrosis had significant prognostic implications, in line with previous studies (Makinen et al.
2017; Qian et al.
2018; Yi et al.
2018; Li et al.
2017; Yoshizawa et al.
2011). One of the possible reasons for this may be that the process of tumor necrosis may release proinflammatory intracellular contents into the tumor microenvironment, inducing an inflammatory response involving a diverse set of immune cells such as neutrophils. Tumor-associated neutrophils are associated with poor prognoses in a variety of cancer types, and a previous study reported consistently increased levels of necrosis and infiltration of neutrophils (Li et al.
2017). Another explanation is that most SOL and MIP adenocarcinomas are correlated with the presence of tumor necrosis, with SOL adenocarcinomas showing the highest indices (Makinen et al.
2017). Similar results were found in our study; in 44 adenocarcinomas with necrosis, 25 had SOL and MIP patterns (
p = 0.009).
Yoshizawa et al. (
2011) proposed three architectural grades: low (LEP), intermediate (ACN and PAP), and high grade (SOL and MIP) based on the predominant growth patterns of invasive carcinomas. In our study, there are no patients with pure lepidic growth pattern. However, most lung adenocarcinomas have mixed growth patterns. A combination of high-grade parts could result in more aggressive biological behavior. Sica et al. (
2010) proposed a grading system to stratify prognostic differences in early-stage adenocarcinomas by integrating the two most representative grades of a tumor, which provided a more comprehensive description of tumor aggressiveness. In a previous study (Zhao et al.
2015), we validated the prognostic effects of the grading system for stage I adenocarcinomas in an Asian population. In addition, patients with high-grade components were at higher risk for local–regional recurrence following sublobar resection, so adjuvant therapy may be needed for these patients, even in the early stages (Zhao et al.
2018). Kadota et al. (
2012) proposed a grading system that combined the predominant growth pattern and the mitotic count in ten high-power fields (HPFs). Adenocarcinomas with an intermediate architectural grade (ACN or PAP) with a low (< 3/10 HPFs) mitotic count were classified as low grade. Intermediate architectural grade with intermediate–high mitotic counts was considered intermediate grade and high architectural grade with any mitotic count was classified as high grade. The combination of the architectural grade and the mitotic count also stratified stage I lung adenocarcinomas into different risk groups of recurrence. However, Barletta et al. (
2010) did not recognize mitotic count as a significant prognostic factor.
Previous studies of prognostic signatures of genetic expressions (Shedden et al.
2008; Chen et al.
2007; Kratz et al.
2012; Wistuba et al.
2013) have identified two molecular prognostic markers for lung adenocarcinoma (Kratz et al.
2012; Wistuba et al.
2013), but their accuracy in predicting survival is limited. Li et al. (
2017) proposed an individualized prognostic signature for non-squamous NSCLC based on immune-related gene pairs along with clinical factors, but its clinical utility also needs to be further tested and validated. Martinez-Terroba et al. (
2018) proposed a prognostic signature based on three proteins (BRCA1, QKI, and SLC2A1) to stratify early-stage lung adenocarcinoma patients. Taken together, these studies have illustrated the importance of genetic aspects in predicting prognoses of lung adenocarcinomas. However, considering the additional cost of various genes selected from different studies, which may not be generalized in different populations, the actual clinical utilization of these genetic profiling methods still remains limited in daily clinical practice.
In this study, we identified a novel and practical prognostic model based on morphological features and the TNM stage of patients who underwent anatomical resection of stage I adenocarcinomas. The predictive ability of this model is higher than that of TNM staging and pathological architectural score, with a greater AUC. Stratifying pathological stage I patients into high- and low-risk subgroups for predicting early relapses might have an important impact on individualized treatment strategies.
The limitations of our study included its retrospective nature and the small size of the cohort, which also prevented us from developing an external validation cohort. The model also failed to incorporate other clinical and pathological prognostic factors (e.g., tumor markers, PET–CT value, and central or peripheral tumor) and some important recognized prognostic molecular factors (e.g., KRAS mutations, EGFR mutations, and ALK rearrangements). Incorporation of other prognostic factors may improve this model. Further studies are also needed to validate this prognostic model to assess its real efficacy and to better analyze its utility in clinical practice and trials.
In conclusion, we developed a prognostic model based on morphological features and TNM stage for stage I lung adenocarcinoma patients. We were able to identify low- and high-risk subgroups of stage I patients after surgery, which may help clinicians select individual treatments and management strategies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.