Introduction

Among all cancers, lung cancer remains the leading cause of death [1]. For patients with clinical stage I non-small cell lung cancer (NSCLC), the standard treatment is anatomical resection and systemic nodal dissection. However, about 25 % of patients are not amenable to lobectomy or pneumonectomy for various reasons. Limited surgery (wedge resection or segmentectomy) and new stereotactic ablative radiotherapy (SABR) are other options for such patients.

If clinical N0 tumours could be reliably identified, it would be possible to avoid anatomical resection and systemic nodal dissection. To achieve successful outcomes for SABR or limited surgery, it is necessary to perform accurate staging, especially of nodal metastasis. Being able to identify occult lymph node metastasis could improve outcomes and may permit individualized adjuvant systemic treatment beyond SABR or limited surgery. Important prognostic information can be obtained through accurate staging of NSCLC, and the best treatment approach can be determined.

Lymph nodes have been staged in patients with NSCLC using chest computed tomography (CT). Contrast-enhanced CT has become the most prevalent imaging modality for tumour node metastasis (TNM) staging, but it is limited in terms of evaluating lymph node status because positivity is predicted based only on node size [2]. Lymph nodes are being staged more frequently in patients with NSCLC using 18F-fluorodeoxyglucose positron emission tomography (FDG-PET), which is based on the fact that glucose metabolism is typically increased in malignant cells [3]. New imaging systems using integrated FDG-PET/CT have recently been able to overcome the inherent disadvantages of FDG-PET, such as poor-quality anatomical information [4]. This integrated approach is more sensitive than CT alone in terms of lymph node staging for NSCLC [5]. As a practical approach, many centres schedule patients with negative mediastinal uptake on integrated FDG-PET/CT for resection [6].

The sensitivity and specificity of the ability of contrast-enhanced CT to predict mediastinal lymph node metastasis are 57–68 and 76–82 %, respectively [7, 8], whereas integrated FDG-PET/CT can identify nodal disease with 79–85 % sensitivity and 87–92 % specificity [7, 8]. Surgeons have widely applied integrated FDG-PET/CT to evaluate nodal metastasis. Currently, a practical approach adopted in many centres for SABR is to schedule patients with negative hilar and mediastinal lymph node uptake on integrated FDG-PET/CT for resection, preoperative mediastinoscopy or endobronchial ultrasound-guided transbronchial needle aspiration. The present study aimed to detect occult lymph node metastasis in patients with clinical stage I NSCLC diagnosed using preoperative integrated FDG-PET/CT. We also aimed to identify potential risk factors for nodal involvement, since it is expected that accurate prediction of occult lymph node involvement based on associated risk factors would facilitate the selection of appropriate candidates for preoperative mediastinoscopy or endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA).

Patients and methods

Patient eligibility

A total of 444 consecutive patients underwent pulmonary resection for lung cancer at Sagamihara Kyodo Hospital (Sagamihara, Kanagawa, Japan) between April 2007 and May 2015. Among these, we reviewed the records of 246 who had clinical stage I NSCLC diagnosed by integrated FDG-PET/CT according to the 7th edition TNM classification and who underwent complete surgical resection by lobectomy or greater and systematic lymph node dissection. Complete resection was defined as the macroscopic and microscopic absence of residual cancer (R0). Patients were excluded if they underwent limited resection (segmentectomy or wedge resection), incomplete resection or neoadjuvant chemotherapy/radiotherapy. None of the patients underwent preoperative mediastinoscopy or EBUS-TBNA during this period.

We reviewed the following clinicopathological information: age, sex, smoking habit (never smoked or have smoked), history of lung disease, concurrent diabetes, preoperative serum carcinoembryonic antigen (CEA) level (cut-off at normal upper limit of 5.0 ng/mL), tumour laterality (right or left side), lobar distribution of tumour, tumour location, maximum standardized uptake value (SUVmax) of primary tumour, procedure, histological type, tumour size (cm), grade, pleural invasion (negative or positive) and lymph node metastasis (n0, n1 or n2). All clinical, intraoperative radiological and pathological findings were reviewed at two hospitals (Sagamihara Kyodo Hospital and Yuai Clinic, Kanagawa, Japan). The patient characteristics and preoperative tumour evaluations are shown in Table 1. Non-small cell lung carcinoma was diagnosed based on the World Health Organization (WHO) classification [9]. Pre- and postoperative TNM staging was applied [10]. The institutional review board at Sagamihara Kyodo Hospital approved the data collection and analyses and waived the need to obtain written informed consent from each patient during May 2015.

Table 1 Characteristics of patients and tumours (n = 246)
Full size table

Computed tomography

All patients were assessed within four weeks of surgery using diagnostic-quality, contrast-enhanced CT of the chest with a slice thickness of 1 mm. A tumour was considered to be central if its centre was located in the inner one-third of the lung parenchyma (adjacent to the mediastinum) on transverse CT images. Tumours centred in the outer two-thirds of the lung parenchyma on transverse CT images were considered to be peripherally located tumours. The maximal diameter of lung nodules was measured on contrast-enhanced CT images of the chest.

Integrated 18F-fluorodeoxyglucose positron emission tomography imaging

All patients were assessed by integrated FDG-PET/CT within four weeks before undergoing surgical resection. After fasting for six hours, FDG (3.5 MBq/kg body weight) was intravenously injected if the patient’s blood sugar level was <200 mg/dL. Images were acquired from a single Eminence-SOPHIA PET/CT combined scanner (Shimadzu, Kyoto, Japan) starting 60 min later [11]. After attenuation correction, acquired image data were reconstructed using a dynamic row-action expectation maximization algorithm [12]. The reconstructed sectional images were then visually evaluated and quantified using the SUVmax inside a volume of interest (VOI) placed on lesions. The SUVmax was calculated as:

$$ [({\text{maximum}}\;{\text{activity}}\;{\text{in}}\;{\text{VOI}})/({\text{volume}}\;{\text{of}}\;{\text{VOI}})]/[({\text{injected}}\;{\text{FDG}}\;{\text{dose}})/({\text{patient}}\;{\text{weight}})]. $$

The quality of the radiation measurements was assured by calibration of the PET/CT scanner in accordance with the National Electrical Manufacturers Association (NEMA) NU-2 2001 standard [13].

Nodal uptake with SUVmax > 2.5 was considered positive [14–17]. A cylindrical region of interest (ROI) was manually placed over the transaxial site with the highest signal intensity to determine the SUV, which was normalized to the body weight of each patient. The SUVmax within a ROI served as our reference value [18].

Three experienced radiologists individually analysed integrated FDG-PET/CT images. If initial assessments differed, the final conclusion was reached by consensus.

Surgical resection

All patients underwent anatomical lung resection and radical lymphadenectomy conducted by thoracic surgeons at Sagamihara Kyodo Hospital. Nodal dissection techniques were standardized at the time of surgical resection. All systematic lymph node dissection proceeded according to the criteria of the American Thoracic Society, and at least three hilar and mediastinal stations each were removed.

Pathological examination

Experienced pulmonary pathologists assessed all resected tumour specimens, and NSCLC was histologically assessed based on the WHO classification. All dissected lymph nodes were stained with haematoxylin and eosin and then histologically assessed.

Statistical analysis

Data were statistically analysed using SPSS version 21.0 software (SPSS, Chicago, IL, USA). The optimal discriminative cut-off for SUVmax was assessed from receiver operating characteristic (ROC) curves. Predictors of lymph node metastasis in clinical stage I NSCLC using integrated FDG-PET/CT were identified by univariate analysis using Fisher’s exact test or Pearson’s Chi-square test. Independent predictors of occult lymph node metastasis of clinical stage I NSCLC were also assessed by multivariate analysis using logistic regression. Values of P < 0.05 were considered statistically significant.

Results

Patient characteristics

Table 1 shows the characteristics of the patient and pathological features of nodal involvement for 246 patients [female, n = 146; male, n = 100; mean age 69 (range 35–86) years]. The median tumour size was 2.8 cm, and the median SUVmax of the primary tumour was 3.0. One and three patients underwent total pneumonectomy and bilobectomy, respectively, and all others were treated by lobectomy.

Incidence of occult nodal metastasis

N1 and N2 disease involvement was evident in 13 (5.3 %) and in 18 (7.3 %) of the 246 patients, respectively. The negative predictive value (NPV) for occult lymph node metastasis on integrated FDG-PET/CT was 87.4 %.

Risk factors for occult lymph node metastasis

Figure 1 shows the ROC curve for predicting lymph node metastasis based on the SUVmax of the primary tumour, which had an area under the curve (AUC) of 0.665 (P = 0.003). The optimal cut-off for SUVmax of the primary tumour for occult lymph node metastasis determined from the ROC curve was 3.0. Table 2 shows predictors of lymph node metastasis in clinical stage I NSCLC selected by univariate analysis. The SUVmax of the primary tumour, histological type (adenocarcinoma) and tumour size represented independent predictors of occult lymph node metastasis for patients with clinical stage I NSCLC determined by integrated FDG-PET/CT according to multivariate analysis (Table 3). Table 4 shows predictors of mediastinal lymph node metastasis in clinical stage I NSCLC diagnosed by integrated FDG-PET/CT. The SUVmax of the primary tumour, histological type (adenocarcinoma) and tumour size were independent predictors of occult mediastinal lymph node metastasis of clinical stage I NSCLC by integrated FDG-PET/CT.

Fig. 1
figure 1

Area under receiver operating characteristics curve for SUVmax of primary tumour to predict occult lymph node metastasis. AUC, 0.665 (95 % CI 0.578–0.751), P = 0.003. Hypothetical threshold SUVmax 3.0 (arrow) yielded 83.9 % sensitivity and 46.5 % specificity. AUC, area under the receiver operating characteristics curve; SUVmax, maximum standardized uptake value

Full size image
Table 2 Univariate analysis of factors associated with occult lymph node metastasis
Full size table
Table 3 Multivariate analysis of factors associated with occult lymph node metastasis
Full size table
Table 4 Univariate and multivariate analysis of factors associated with occult mediastinal lymph node metastasis
Full size table

Follow-up

The median postoperative follow-up period was 4.5 years. Cancer recurred in 43 (17.5 %) of the 246 patients [intrathoracic recurrence, n = 21 (8.6 %); distant metastasis, n = 22 (8.9 %)]. Distribution of number of the patients with recurrence was as follows: N0 disease, n = 18 (7.3 %); N1 disease, n = 10 (4.1 %); N2 disease, n = 15 (6.1 %). And 30 (12.2 %) of the 246 patients had died of lung cancer by the end of the follow-up. Distribution of number of the patients with death was as follows: N0 disease, n = 10 (4.1 %); N1 disease, n = 7 (2.8 %); N2 disease, n = 13 (5.3 %).

Discussion

FDG-PET is widely used in the staging of NSCLC [14], and it should become a promising optimization tool for lymph nodal staging. Although integrated FDG-PET/CT provides useful information about hilar and mediastinal nodal metastases in lung cancer, microscopic lymph node metastases are barely detectable on integrated FDG-PET/CT.

We applied the traditional SUVmax cut-off of 2.5, and thus hilar and mediastinal lymph nodes with SUVmax ≥ 2.5 were considered positive.

Occult lymph node metastasis was detected in 31 (12.6 %) of 246 patients, and the NPV was 87.4 %, which is in agreement with previous studies [19–24]. Not all patients underwent systematic lymph node dissection in some studies [19, 20], which might have resulted in underestimation of the true prevalence of occult nodal disease. In the present study, all patients underwent anatomical lung resection and radical lymphadenectomy. All systematic lymph node dissection proceeded according to the criteria of the American Thoracic Society, and at least three hilar and mediastinal stations each were removed. It seemed to be very important to perform anatomical lung resection and radical lymphadenectomy for accurate diagnosis of nodal status.

The use of routine mediastinoscopy is not cost-effective for patients with peripheral stage I disease because the incidence of mediastinal disease among more central lesions is a known observation by many. And stage I NSCLC represents a heterogeneous group of tumours, so it is important to examine the risk factors for occult lymph node metastasis.

Univariate analysis significantly associated CEA, SUVmax of the primary tumour, adenocarcinoma, tumour size and pleural invasion with occult lymph node metastasis. Among patients with clinical stage I NSCLC determined by integrated FDG-PET/CT, multivariate analysis identified SUVmax of the primary tumour, adenocarcinoma and tumour size to be independent predictors of occult lymph node metastasis.

The incidence of lymph node metastasis increases as tumour size increases [25, 26]. Asamura et al. reported that the prevalence of lymph node metastasis increased from 19.5 % in tumours ≤2.0 cm in diameter to 32.5 % in tumours 2.0–3.0 cm in diameter in patients with resected peripheral NSCLC [26]. Large clinical tumours were significantly associated in the present study with an increased prevalence of occult lymph node metastases.

The clinical importance and relevance of SUVmax for primary NSCLC have been demonstrated. Several previous studies have reported that a higher SUVmax for primary NSCLC was significantly associated with nodal involvement, although the definition of the SUV threshold considerably varies among studies [4, 17, 23, 27, 28]. SUVmax provides information about biological aggressiveness, key pathological features and the potential for tumour spread. SUVmax of the primary tumour should be provided by radiologists in the text of every integrated FDG-PET/CT report.

Adenocarcinoma was also a risk factor for occult lymph node metastasis in our patients. Kanzaki et al. [29] considered that occult lymph node metastasis was more likely to occur in patients with lung adenocarcinoma. In contrast, Al-Sarraf et al. [6] reported that primary tumour cell type did not affect the incidence of occult MLN metastasis.

Lymph node metastases more frequently have a short-axis diameter ≤1 cm when they arise from adenocarcinoma than from squamous cell carcinoma [2]. Therefore, we speculate that patients with lymph node metastasis from squamous cell carcinoma might have been excluded from the series by Kanzaki et al., but would have been included in our series of patients.

From our results, we speculate that the SUVmax of the primary tumour, adenocarcinoma and tumour size reflect biological aggressiveness of the tumour. Integrated FDG-PET/CT will remain an important tool in the management of NSCLC, but is not a good stand-alone diagnostic tool and must be used with other patient characteristics in determining whether to refer a patient for tissue procurement.

The retrospective design is the main limitation of the present study. Prospective or randomized trials are warranted to clarify the true incidence of occult lymph node metastasis among patients with negative lymph node uptake on integrated FDG-PET/CT. We did not stage mediastinal invasion, which might have changed the denominator of truly occult disease. We excluded patients who had been treated by neoadjuvant chemotherapy and radiotherapy and those who underwent segmentectomy, to avoid considerable inaccuracies. Although our criteria might have caused some degree of selection bias, the overall number of patients was probably large enough to overcome it.

The SUVmax of the primary tumour, adenocarcinoma and tumour size were risk factors for occult lymph node metastasis in patients with NSCLC diagnosed as clinical stage I by preoperative integrated FDG-PET/CT. False-negative results are inevitable, but may be somewhat predictable using some risk factors. These findings would be helpful in selecting patients who would benefit from preoperative mediastinoscopy or EBUS-TBNA. Our findings might facilitate pretherapy evaluation and decision-making.