Trends in mediastinal nodal staging and its impact on unforeseen N2 and survival in lung cancer

Discussion

A significant increase in rates of invasive mediastinal nodal staging in patients with resectable NSCLC was found between 2011 and 2017 in the Netherlands. Increasing use of less invasive endosonography prior to or substituting surgical staging did not lead to an increase in uN2 disease. Performance of invasive mediastinal staging led to a clinically relevant OS benefit in patients with clinical N1–3 disease.

After introduction of registration of endosonography in the Netherlands Cancer Registry in 2011, a significant increase in invasive mediastinal staging in patients with potentially resectable NSCLC in the Netherlands was found. Between 2005 and 2010, only the use of mediastinoscopy was registered, which could have possibly induced overestimation of the increase in use of endosonography from 2011 onward. However, it could be expected that endosonography was not used on a large scale in the Netherlands before 2011. The 2007 European Society of Thoracic Surgeons (ESTS) guideline described endosonography as an optional new technique with high specificity but low negative predictive value, requiring confirmatory invasive surgical technique in case of negative endosonography. After publication of the 2007 ESTS guideline recommending mediastinoscopy, the availability and experience with endosonography has tremendously increased. In the ASTER-1 trial, comparable sensitivity for mediastinal nodal metastases detection was found by endosonography (85%) and surgical staging alone (79%). When adding confirmatory mediastinoscopy to endosonography a significant increase in sensitivity to 94% was found (p=0.02 compared to 79% with surgical staging alone) [4]. Largely based on these facts, the 2015 conjoint European Society of Gastrointestinal Endoscopy (ESGE), European Respiratory Society (ERS) and ESTS guideline recommended EBUS, preferably added by EUS, as initial staging technique followed by confirmatory mediastinoscopy in case no metastases were proven by pathology [1]. The increase in endosonography that we demonstrated in this study was probably based on these publications.

The increase in invasive staging over the years did not result in a decrease in uN2 disease. Adequate selection of patients who might benefit from invasive staging seems therefore important. A nationwide study including 3263 Dutch patients who underwent NSCLC resection in 2017–2018 showed that 69% of these patients had an indication for invasive staging according to the ESGE/ERS/ESTS guideline [7]. With only 32% patients undergoing invasive staging in our analysis, it appears that not all patients with an indication actually underwent invasive staging. Additionally, only 11% of patients underwent combined endosonography and confirmatory mediastinoscopy, suggesting significant nonadherence to the guidelines. Although not deducible from our dataset, possible reasons for this nonadherence could be doctors’ or historical preferences, limited experience with endosonography or limited availability of equipment and endosonography suites. In addition, it may be possible that increasing experience with endosonography led to higher confidence about its negative predictive value, resulting in the omission of confirmatory mediastinoscopy. However, information on medical decision making and detailed data (except the clinical nodal stage) to determine if patients had an indication for invasive staging were lacking in the Netherlands Cancer Registry.

Obviously, higher clinical nodal stages were associated with an increased risk of uN2 and worse OS, underlining the importance of invasive staging in patients with cN1–3. The survival difference among cN1–3 subgroups with or without invasive staging was 5%. Interview-based studies indicated that survival was the most important attribute in lung cancer treatment [8, 9]. Discrete-choice experiments showed that lung cancer patients accepted 2% mortality of lung cancer treatment (surgery or radiotherapy) for one additional year of life or would trade survival for short- or long-term side-effects of therapies [10, 11]. Therefore, with limited morbidity and mortality of invasive mediastinal staging, a 5% increase in OS in this population appears to be defined as clinically relevant by patients. Survival analyses of an observational cohort study of 11 North American hospitals showed significant survival benefit of performance of invasive nodal staging in patients with cN1–3 disease (only Kaplan–Meier figure provided, no absolute data). However, selection bias in this study has to be taken into account [12].

Squamous cell histology was found to increase the use of invasive staging compared to adenocarcinomas. This could be influenced by clinical features such as fast growth, cavitation with necrosis possibly inducing reactivity in lymph nodes and compromised prognosis of squamous cell carcinomas [13, 14]. Next to the histology, tumour location also determined whether invasive staging was used and affected uN2 outcomes. We found patients with left-sided lung tumours to be less likely to undergo invasive staging (in the entire population), while left-sided lung tumours were associated with increased risk of uN2. It is known that ∼25% of all N2 metastases are located in the aortopulmonary stations, which cannot be reached by either EBUS, EUS or cervical mediastinoscopy [5, 15, 16]. The challenging anatomical position as well as the uncertain clinical relevance of aortopulmonary N2 metastases in patients with left upper lobe tumours might have influenced the decision whether to perform invasive staging. Survival of these patients after all seems to be significantly better compared to patients with metastases in the subcarinal station [17]. However, no information of the affected lymph node stations was available in the Netherlands Cancer Registry, making it impossible to interpret and analyse reasons for less adherence to the guideline and the effect of nodal metastatic distribution on survival.

Although detection of unforeseen N2 after definite surgery seems undesirable, the question remains whether upfront detection of N2 leads to improved survival. Garelli et al. [18] also demonstrated a significant survival difference between patients with microscopic (<2 mm) and macroscopic (≥2 mm) uN2, and Yoo et al. [19] showed significant OS differences between patients with one, two to four, and five or more malignant N2 lymph nodes. These results correspond with several retrospective studies reporting better OS in patients with minimal N2 disease [20–22]. Since details on the affected lymph node stations and size of metastases were lacking in the Netherlands Cancer Registry, we were not able to describe details on nodal spread, other than number of affected nodes. Constrained by the available data we were forced to use a very strict cut-off between minimal and extensive uN2 disease. Based on the aforementioned studies the proportion of patients with minimal uN2 disease in our analysis may therefore be underestimated, as more than one affected lymph node might all have been micrometastases and/or located in a single lymph node station as well as distribution of affected nodes among hilar and mediastinal lymph node stations with only minimal spread in N2 stations.

In patients with stage III NSCLC, the choice and timing of treatment (neoadjuvant or adjuvant chemotherapy with or without surgery) may influence survival. Analysis of the American National Cancer Database comprising ∼65% of all lung cancer patients in the United States showed 34% 5-year OS in patients with stage III NSCLC who underwent primary surgical resection and adjuvant treatment (2004–2012, n=3721, all pN2) [23]. In the Netherlands Cancer Registry, analysis of patients with clinical stage IIIA NSCLC (2010–2013, n=4816, 67% cN2, 23% cT4) we found 4-year OS of 39% in patients primarily treated by surgical lung tumour resection, while 4-year OS was 51% in patients receiving neoadjuvant therapy and subsequent surgical lung tumour resection [24]. The ESPATUE trial showed 5-year-OS of 44% in patients with cytologically proven stage IIIA or IIIB (n=81, 70% N2–3, 30% T4) treated by induction chemotherapy and subsequent surgical lung tumour resection [25]. Only patients with sufficient response to neoadjuvant therapy and good clinical condition will generally proceed to surgical lung tumour resection, and thus selection bias should be taken into account when assessing these outcomes. Based on these results, adequate mediastinal nodal staging of patients with resectable NSCLC remains important.

In our study, performance of invasive nodal staging even indicated possible improvement in OS with 5% of patients with cN1–3 disease undergoing primary surgical lung tumour resection. Future research should determine whether this survival benefit persists and should evaluate which subgroups especially benefit from the different invasive mediastinal staging strategies. Patients with extensive N2 disease might benefit from neoadjuvant therapy instead of primary surgical lung tumour resection, whereas minimal N2 disease may accurately be treated by surgery and adjuvant systemic therapy.

The results of this study should be interpreted with the limitation that the Netherlands Cancer Registry lacks detailed information. No information was available on quality of staging techniques (e.g. number of lymph node stations visualised or sampled; use of conventional or video mediastinoscopy; combined use of EBUS and EUS), precluding the assessment of impact of quality on uN2 rates or survival. Except the pathological nodal stage and number of affected lymph nodes, no details on lymph node level and extensiveness of spread with a lymph node or level were available. In addition, this precluded us from dividing uN2 cases in detection errors (lymph node metastasis not detected by FDG-positron emission tomography/computed tomography nor endosonography and mediastinoscopy if performed) or sampling errors (metastasis missed despite lymph node sampling during endosonography and/or mediastinoscopy of a suspicious station on imaging). Additionally, during follow-up, no details on recurrence of the disease or causes of death were reported, limiting the survival analysis to overall survival only. Despite these limitations, this is the first study showing long-term nationwide trends in invasive mediastinal nodal staging of NSCLC and its effect on uN2 disease and OS.