Non-small cell lung cancer (NSCLC) with small tumor size is now frequently being detected because of the prevalent use of computed tomography (CT) as a screening tool for pulmonary lesions. Sublobar resections have been reported to yield similar survival as with lobectomy for patients with small peripheral NSCLC. In determining the
indications for sublobar resection, prediction of the pathologic nodal status is important.
Tumor size is consistently a strong predictor for lymph node metastases. The
prevalence of mediastinal metastases increases with tumor size. Asamura and colleagues have found that among patients with resected peripheral non-small cell lung cancer, the prevalence of lymph node metastases increased from 19.5% in tumors 2 cm or smaller to 32.5% in tumors 2 to 3 cm in diameter (Asamura et al. 1996). However, with tumor size of 2 cm or less, the predictive power of lymph node metastases needed to be reappraised.
One of the preoperative predictors for lymph node metastases is radiologic
appearance of tumor on computed tomography (CT). The radiologic appearance of the tumor, categorized as pure ground-glass opacity (GGO), part-solid GGO, or solid tumor, has been reported to be the best predictor for the invasiveness and nodal status of
NSCLC (Kodama et al. 2001; Takamochi et al. 2001; Matsuguma et al. 2002; Suzuki et al. 2002; Ohde et al. 2003; Okada et al. 2003). The consensus is that the more GGO component in a tumor, the less possibility of lymph node metastases, and the better the prognosis of the disease. Reviewing the published literature, there is no solid evidence available to prove that GGO percentage is a good criterion to predict lymph node metastases in early NSCLC with tumor size of 2 cm or less.
reported that preoperative serum CEA level was a predictor for mediastinal nodal metastasis in clinical stage IA NSCLC patients (Koike et al. 2012). Inoue et al. reported that compared to patients with normal preoperative serum CEA levels and with NSCLC tumors of 2 cm or less in diameter, the 5-year mortality rate for patients with higher CEA levels (≥ 5 ng/mL) was significantly worse (92.1% vs. 77.6%; p<0.01). In addition, increased CEA level was associated with a much higher rate of lymph node metastasis in small NSCLC (29.2% vs. 10.3%; p=0.02) (Inoue et al. 2006). In our study,
preoperative CEA level was routinely collected in every lung cancer patient undergoing surgical resection. We evaluated the impact of preoperative CEA levels on lymph node metastases in our study.
In our study, we aimed to evaluate the predictive factors of lymph node metastases in NSCLC of tumor size 2 cm or less. Reviewing the previous literature, only a few studies focused on such a small size of lung cancer. Tumor size, preoperative CEA level,
radiological characteristics, and pleural invasion were reported to be associated with lymph node metastases, and those may be used as a surrogate for tumor invasiveness.
However, the evidence at hand is still weak in the category of patients with small NSCLC. Encountering a growing number of early lung cancer, accurate prediction of preoperative lymph node status may guide us in our surgical strategy. With better prediction of lymph node status, mediastinal lymph node dissection could be avoided in selected patients undergoing lung cancer surgery. Sublobar resection, via wedge
resection or segmentectomy, may also be justified in small NSCLC.
Methods and Materials
Cases of non-small cell lung cancer after surgical resection in the National Taiwan University Hospital from January 2011 to December 2015 were retrospectively
reviewed. Inclusion criteria were patients with solitary pulmonary tumor of preoperative
clinical staging T1aN0M0, pathologically-proven lung cancer after surgical resection, and pathological tumor size 2 cm or less. Exclusion criteria were cases of synchronous pulmonary malignancy, metastatic carcinoma of the lung, synchronous lung cancer, and lung cancer after neoadjuvant therapy. Data to be collected for analysis were age, sex, preoperative CEA level, preoperative pulmonary function results, pathological tumor size, tumor histology subtypes (lepidic, acinar, papillary, micropapillary or solid) for adenocarcinoma, pathological lymph node status (N1 and N2 status), visceral pleural invasion (VPI), lymphovascular invasion (LVI), radiological appearance of tumor, GGO ratio of tumor, and surgical method (wedge resection, segmentectomy or lobectomy).
The predictive factors of interest in our study were tumor size, preoperative CEA level, and radiological tumor appearance. The tumor size was recorded from the
pathology reports. The size was evaluated as categorical data and continuous data to see how the tumor size affects the prediction of lymph node metastases. Preoperative serum CEA level was reviewed from medical records. If there was more than one CEA level, the level immediately prior to surgery was recorded. Radiological tumor appearance was defined as GGO percentage. The ratio of GGO will be evaluated as categorical data (e.g. < 25%) to correlate with lymph node metastases.
Logistic regression analyses will be performed to identify predictors for lymph node metastases.
Results
Table 1 showed the demographic data of the two groups. No significant difference between pN0 and pN+ groups were found in relation to age, gender, smoking, and preoperative pulmonary function. Tumor size, GGO% on CT, surgical method (including number of resected lymph nodes), and pathologic results seemed to be associated with a significant difference between the pN and pN groups. Preoperative
serum CEA level was higher in the pN+ group than in the pN0 group, although the p value failed to reach significance (p=0.072). On logistic regression analysis for the prediction of lymph node metastases (Table 2), tumor size, preoperative CEA levels, GGO%, number of resected lymph nodes, non-lepidic subtype, VPI, and LVI were significant predictors for lymph node metastases. We included those preoperative factors that showed significance to predict lymph node metastases into multivariate analysis. Only tumor size, preoperative CEA level, and GGO% on CT remained significant predictors of lymph node metastases. With preoperative variables grouped into categorical data, tumor size ≥ 1.5 cm, CEA ≥ 3 ng/mL, and GGO < 25% were noted to have strong predictive values for lymph node metastases (Table 3).
Conclusions
For NSCLC patients with clinical stage T1aN0M0, tumor size, preoperative serum CEA levels, and GGO percentage on CT were significant predictive factors for lymph node metastases after surgery. For patients with tumor size less than 1.5 cm, serum CEA levels less than 3 ng/mL, and GGO predominant tumors, avoiding lymph node
dissection can be justified. Sublobar resection, instead of standard lobectomy, may be a good alternative for this group of patients.