Introduction

Treatment for lung cancer is mainly determined according to the cancer stage. The standard treatment for stage I lung cancer is anatomical pulmonary resection with lymph node dissection [1], while multimodal treatment is preferred for higher-stage tumors. Therefore, exact determination of cancer stage is very important for successful treatment of lung cancer.

Chest computed tomography (CT) and positron emission tomography (PET)/CT are the main techniques for the clinical staging of lung cancer. Invasive preoperative lymph node staging using endobronchial ultrasonography-guided transbronchial needle aspiration might be necessary in selected cases, but it is unreasonable for patients with no sign lymph node metastases at preoperative imaging. According to the guideline of the European Society of Thoracic Surgeons (ESTS), invasive mediastinal lymph node staging can be skipped if there are no suspect lymph nodes on CT or PET/CT, the tumor size on imaging is ≤3 cm, and the tumor is confined to the periphery of the lung [2]. However, even patients who meet the three criteria for clinical N0 (cN0) tumors occasionally have postoperative lymph node upstaging, despite the cN0 diagnosis under the ESTS guideline.

We have reported that the maximum standardized uptake value (SUVmax) of the main tumor on PET/CT scan is a predictor of occult lymph node metastasis in cN0 lung tumors [3], and other studies have confirmed that the tumor SUVmax is a predictor of postoperative nodal upstaging [4,5,6,7]. However, lymph node metastasis seldom occurs in lung cancers presenting as ground glass opacity nodules [8, 9], which are characterized by low SUVmax and are well known as less invasive, mainly lepidic adenocarcinomas. Therefore, the effect of SUVmax may not be important when the lepidic component of the tumor is excluded.

Previous studies of nodal upstaging in clinical stage I lung cancer have been based on the seventh edition TNM classification, in which the tumor (T) stage is based on total tumor size, including the lepidic component. In the eighth edition TNM classification, the T stage is determined by the size of the invasive component only, excluding the lepidic component. As a result, we expect that the role of SUVmax as a predictive factor for nodal upstaging may be diminished under the eighth edition TNM staging system.

The purpose of this study was to identify the risk factors for nodal upstaging after surgery in cN0 peripheral non-small cell lung cancer (NSCLC) with invasive component size ≤3 cm, which is classified as clinical stage T1 under the eighth edition TNM classification, and to determine whether SUVmax is a predictor of postoperative nodal upstaging in cN0 lung cancers under the new classification.

Patients and methods

Patients

From January 2010 to December 2017, 1630 consecutive patients were diagnosed with lung cancer and underwent therapeutic surgical resection at a tertiary hospital in Korea. Among these, 553 patients satisfied the following inclusion criteria: (1) anatomical lobectomy with systematic lymph node dissection; (2) tumor stage T1 (invasive component size ≤3 cm) under the eighth edition TNM staging system; (3) cN0; 4) peripheral location; 5) no neoadjuvant treatment.

The patients were classified into two groups according to surgical pathology results: those with cN0 tumors preoperatively and pathologic N0 (pN0) tumors postoperatively (pN0 group) and those with cN0 tumors that were upstaged postoperatively to pN1 or pN2 tumors (nodal upstaging group). Clinicopathological characteristics were compared between the two groups. Operative procedures included lobectomy and bilobectomy, and dissection of more than three mediastinal lymph node stations was performed in every patient by en bloc resection of the lymph nodes and adjacent fat tissues. For patients with right-sided tumors, the paratracheal and subcarinal lymph nodes were routinely dissected and for those with left-sided tumors, para-aortic, subaortic, and subcarinal lymph nodes were routinely dissected. Other visible mediastinal lymph nodes were also resected. N1 lymph nodes, including hilar, interlobar, and lobar nodes, were also routinely dissected. This study was approved by the institutional review board of Seoul St. Mary’s Hospital at the Catholic University of Korea.

Preoperative lymph node staging

Preoperative lymph node staging is based on contrast-enhanced chest CT and F-18 fluorodeoxyglucose (FDG)-PET/CT scanning. Lymph nodes with short-axis diameter >10 mm on the CT scan and high SUVmax relative to the surrounding mediastinal structures on PET-CT are considered metastatic. However, lymph nodes with high SUVmax are considered benign if the suspect node contains benign calcifications or if it shows high attenuation with a distinct margin on unenhanced CT [3, 10], and general, symmetric, and equivocal FDG uptake in the mediastinal lymph nodes is interpreted as a reactive change due to inflammation. When tumors are clearly diagnosed as cN0 at preoperative imaging and complete resection appears possible, surgery is performed without invasive preoperative lymph node staging.

Histopathology and re-staging

All surgical specimens were examined by pathologists whose observations were recorded. Each pathology report was reviewed for tumor size, location, lymph node status, pleural invasion, lymphatic invasion, and vascular invasion.

Tumors were restaged according to the eighth edition of TNM classification [11] by measuring the greatest dimension of the invasive component on the pathology specimen [12], and tumors with invasive component size ≤3 cm were classified as stage T1 in accordance with the eighth edition of the TNM classification. In the case of adenocarcinoma, total tumor size was defined as the greatest dimension of the tumor including lepidic component. On the other hand, invasive component size was defined as greatest dimension of the invasive component excluding lepidic component of the tumor.

Invasion into the visceral pleura was defined as tumor extension beyond the elastic layer of the visceral pleura. Lymphatic invasion was defined as tumor cells observed in the lymphatic lumen. Vascular invasion was defined as tumor cells observed in the vascular lumen. Histologic features, as well as the presence of pleural, lymphatic, and vascular invasion, were determined using hematoxylin–eosin staining. If the findings could not be determined using hematoxylin–eosin staining alone, special staining, such as Verhoeff–Van Gieson elastic stain, was performed as necessary. In particular, Verhoeff–Van Gieson elastic stain was performed for detailed evaluation of visceral pleural invasion where present.

Statistical analysis

Student’s t test or the Wilcoxon rank-sum test was used for continuous variables, and Chi-squared or Fisher’s exact test was used for categorical variables. Univariate and multivariate logistic regression was used to identify predictors of postoperative nodal upstaging. All variables with p < 0.1 in the univariate analysis were entered into the multivariate analysis, and p < 0.05 was considered statistically significant. All statistical analyses were performed with SPSS version 24.0 software (IBM Corp, Armonk, NY).

Results

Comparison of clinicopathological characteristics

Among 553 study patients with peripheral cN0 NSCLC, 54 (9.8%) had postoperative nodal upstaging (26 [48.1%] to pN1, and 28 [51.9%] to pN2: 10 [18.5%] with strictly N1 metastases and 18 [33.3%] with N1 + N2 metastases). A comparison of the clinicopathological characteristics of patients with and without nodal upstaging is presented in Table 1. Patients with upstaged tumors had higher serum carcinoembryonic antigen (CEA) levels (mean 5.1 vs 2.3, p = 0.004), tumor SUVmax (6.2 vs 4.1, p < 0.001), total tumor size (2.4 cm vs 2.1 cm, p = 0.003), and tumor invasive component size (2.1 cm vs 1.4 cm, p < 0.001). Patients in the nodal upstaging group also had significantly higher incidence of visceral pleural invasion (p < 0.001), lymphatic invasion (p < 0.001), and vascular invasion (p < 0.001).

Table 1 Clinicopathological characteristics in patients with non-upstaged (pN0) and upstaged lymph nodes in peripheral stage I non-small cell lung cancer
Full size table

Nodal upstaging was more frequent for tumors with higher pathologic T stage (Table 2). There was no nodal upstaging in T1a tumors.

Table 2 Pathologic T and N stages in the nodal upstaging group according to the eighth edition TNM classification
Full size table

Table 3 shows the incidence and distribution of metastatic mediastinal lymph nodes based on the location of the main tumor. Lymph node metastases from upper lobe tumors were most frequently found in the upper mediastinal lymph nodes (87.5% from right upper lobe tumors and 100% from left upper lobe tumors), while the subcarinal or lower mediastinal nodes were the main sites for metastasis of right lower lobe (70%) and left lower lobe tumors (100%).

Table 3 Incidence and distribution of mediastinal lymph node metastasis
Full size table

Univariate and multivariate analyses for predictors of lymph node upstaging

Table 4 shows the main risk factors for lymph node upstaging according to the logistic regression analyses. In univariate analysis (Table 4A), CEA level, tumor SUVmax, involved lobe, total tumor size, invasive component size, visceral pleural invasion, lymphatic invasion, and vascular invasion had p values of <0.1. These variables were entered into the multivariate model, and we conducted two multivariate analyses according to the method of measuring tumor size: multivariate analysis T, which adopted total tumor size, including the lepidic component (Table 4B), and multivariate analysis I, which adopted invasive component size (Table 4C). In multivariate analysis T, serum CEA level, visceral pleural invasion, and lymphatic invasion were significant risk factors for nodal upstaging and SUVmax and total tumor size were not. In the multivariate analysis I, serum CEA level (hazard ratio [HR] = 1.113, p = 0.002), invasive component size (HR = 2.398, p = 0.004), visceral pleural invasion (HR = 2.901, p = 0.005), and lymphatic invasion (HR = 9.336, p < 0.001) were significant risk factors for nodal upstaging and SUVmax was not. When the postoperative findings of visceral pleural invasion, lymphatic invasion, and vascular invasion were omitted and the multivariate analysis for predictors for nodal upstaging was repeated (Table 5), the serum CEA level (HR = 1.106, p = 0.001) and the invasive component size (HR = 3.115, p < 0.001) were significant predictors for nodal upstaging, and SUVmax was not.

Table 4 Univariate and multivariate analyses for risk factors of nodal upstaging (logistic regression model)
Full size table
Table 5 Multivariate analysis for predictors of nodal upstaging (logistic regression model)
Full size table

Distribution of invasive component sizes in non-upstaging and upstaging groups

We created a scatter plot to examine the distribution of invasive component sizes in the two groups (Fig. 1). All tumors in the nodal upstaging group measured at least 1.2 cm. Lymph node metastasis was not detected in cases in which the invasive component size was <1.2 cm, and this result is consistent with the finding that nodal upstaging did not occur in patients with stage T1a tumors (Table 2).

Fig. 1
figure 1

Lymph node status according to the size of the invasive component. The scatter plot shows that all tumors in the nodal upstaging group (pN1 or pN2) were 1.2 cm or larger

Full size image

Discussion

The accuracy of clinical staging for lung cancer has improved with the evolution of chest CT and PET/CT scanning. However, despite these technical improvements, nodal upstaging is still reported in approximately 10 to 15% of patients with clinical stage I lung cancer [3,4,5, 13,14,15]. In this study, nearly 10% of patients (54 of 553, [9.8%]) with peripheral cN0 NSCLC with invasive component ≤3 cm were upstaged to pN1 or N2 disease after anatomic resection.

Accurate prediction of lymph node status before surgery facilities more accurate treatment planning. Patients with N2 disease may have a better prognosis with neoadjuvant chemotherapy, and sublobar resection is not sufficient for clinical N0 tumors with known risk of occult lymph node metastasis. Therefore, the potential for nodal upstaging must be carefully considered in every case.

Under the seventh edition TNM classification, the tumor SUVmax on preoperative PET/CT imaging was a predictor of nodal upstaging [3,4,5,6,7]. In the eighth edition TNM classification, which was published in 2017, the tumor (T) stage is determined by the size of the invasive component only, and the lepidic component is not measured for T staging [12, 16]. Therefore, the importance of SUVmax is relatively reduced because predominantly lepidic tumors, which are usually characterized by a low SUVmax at PET/CT, are re-classified as very small tumors. In this study, we examined risk factors for nodal upstaging under the eighth edition TNM staging system, which classifies tumors with invasive component size ≤3 cm as stage T1 and found that SUVmax is no longer predictive of nodal upstaging in clinical T1N0M0 NSCLC. Instead, the size of the invasive component is a significant risk factor for nodal upstaging. Therefore, we propose that invasive component size can replace SUVmax as a predictor of occult lymph node metastasis. Although the invasive component size is measured definitively only at histology, it can be predicted preoperatively by determining the extent of consolidation in the tumor at CT [16], and this measurement can be used to predict the risk of nodal upstaging. We found that all tumors in the nodal upstaging group were 1.2 cm or larger, while there was no occult lymph node metastasis in tumors smaller than 1.2 cm, which indicates that lymph node metastasis may not be a concern in clinical stage T1aN0M0 lung cancer. This also suggests that extensive lymph node dissection might be safely avoided in tumors with small invasive components.

Our finding that the preoperative serum CEA level was associated with the risk of nodal upstaging was consistent with the results of Yamazaki et al. [17]. This suggests that preoperative serum CEA levels, taken together with the preoperative estimate of the tumor invasive component size, should be useful for noninvasive preoperative assessment of the risk of occult lymph node metastasis in patients with peripherally located cN0 NSCLC of ≤3 cm. Visceral pleural and lymphatic invasion were also associated with the risk of nodal upstaging in our patients. When either of these is diagnosed after surgery without adequate lymph node dissection, a treatment plan considering the possibility of occult lymph node metastasis should be established.

Complete resection of lung cancer requires systematic lymph node dissection [18], which is the complete dissection of all mediastinal tissues or dissection of at least three mediastinal lymph node stations, including the subcarinal nodes [18, 19]. Our study population included only patients who underwent lobectomy and systematic lymph node dissection. Therefore, the pathologic lymph node staging is thought to be very accurate. Furthermore, the mean number of dissected lymph nodes was about 15 in this study. Data from the American College of Surgery Oncology Group (ACOSOG) Z0030 trial suggest that resection of 10 nodes might be considered adequate [20], and Dai et al. have reported that lymph node evaluation is important for accurate staging and that removal of 8 to 11 nodes can significantly reduce the risk of undetected nodal positivity [21]. Therefore, we believe that the lymph node evaluation in this study was appropriate in terms of the number of dissected lymph nodes. Systematic lymph node dissection may not be necessary if the possibility of lymph node metastasis is convincingly low. In such cases, local lymph node dissection-like lobe-specific nodal dissection may be appropriate [22, 23]. If the systematic lymph node dissection is omitted, the operation time can be reduced, thereby significantly reducing the general anesthesia time. Our analysis confirmed that upper lobe tumors mainly metastasized to the upper mediastinal lymph nodes and lower lobe tumors metastasized mainly to the subcarinal or lower mediastinal lymph nodes. This could imply that accurate staging and satisfactory outcomes are possible even after local lymph node dissection in patients with a low risk of nodal upstaging. Of course, this must be proven by large-scale studies.

This study had several limitations that should be considered. First, we used a retrospective study design. Second, this study was conducted in patients treated at a single institution. Nonetheless, we were able to collect very detailed clinical information from the electronic medical record, along with data from surgical specimens and pathology reports. In this regard, it was advantageous to have a single institution study. We believe that our data can be used as the basis for future investigations and that our results might be further clarified and refined by future studies with larger patient populations.

In conclusion, under the eighth edition TNM classification, tumor SUVmax is no longer a predictor of nodal upstaging in peripheral cN0 stage I NSCLC, but tumor invasive component size and serum CEA level are. Visceral pleural invasion and lymphatic invasion were also risk factors of nodal upstaging. Future studies that include multi-center data and larger sample sizes are necessary to confirm and supplement these results.