The subdivision of N2 into N2a and N2b implies that the clinical and pathological evaluation of the mediastinal lymph nodes must be as accurate as possible to ensure correct tumour classification. When selecting among the available staging techniques, the following should be kept in mind, especially when the mediastinum is normal (N0 or N1) in the preoperative imaging [19]: targeted (compared with systematic) endobronchial ultrasound with transbronchial needle aspiration (EBUS-TBNA) [20,21], may understage the mediastinum [22]; endosonographic methods are inferior to mediastinoscopy when the mediastinum is normal by imaging [23], and video-assisted mediastinoscopic lymphadenectomy provides the highest staging accuracy [24]. Systematic nodal dissection is the intraoperative evaluation of choice because lobe-specific systematic nodal dissection may leave involved lymph nodes behind [25].

The new TNM groups based on the subdivision of the N2 category that have moved from their 8th edition stage have to be treated according to guidelines. Specifically, T1N2aM0, now in stage IIB, has to be treated as stage IIIA because a taxonomic change does not imply an automatic therapeutic change [26].

The minimal requirement of intraoperative nodal evaluation must include, at least, six lymph nodes: three from the mediastinum (including the subcarinal station), and three from the hilar and intrapulmonary nodal stations. This nodal assessment can be performed preferentially through systematic nodal dissection (SND) or, if SND is not performed, through lobe-specific SND. If this minimal requirement of intraoperative nodal assessment is not achieved and, otherwise, there is no residual tumour left, the IASLC proposed to call the resection uncertain [11,12]. The Union for International Cancer Control (UICC) and the American Joint Committee on Cancer (AJCC) accepted the concept and coded it as R0(un) [99,100]. Recently, some authors have focused specifically on the impact of the extent of nodal assessment and found an incrementally lower survival as the degree of nodal assessment diminishes [101,102].

The impact on survival of systematic nodal dissection and sampling still is controversial. The ACOSOG Z0030 trial compared systematic sampling and complete mediastinal lymphadenectomy for clinical T1-T2 N0 and non-hilar N1. It revealed no differences in 5-year disease-free survival rates, in local, regional and distant recurrence [103], or in morbi-mortality [104], although more patients were found to have pN2 in the complete mediastinal lymphadenectomy group. However, in clinical practice outside clinical trials, lobe-specific SND is associated with increased mediastinal recurrence [105]. A meta-analysis has shown that selective mediastinal lymphadenectomy may be sufficient in early-stage disease, particularly when preoperative imaging and intraoperative findings suggest low risk of nodal involvement [106]. This is especially true for adenocarcinoma in situ (AIS) and minimally invasive adenocarcinoma (MIA), that do not present haematogenous or lymphatic spread, and for which intraoperative nodal evaluation can be minimized or spared [106–108]. However, it is essential to keep in mind that diagnosis of AIS and MIA require the pathologic study of the whole resected specimen and that the final diagnosis may differ from the intraoperative one performed on frozen sections [109,110]. In contrast, a meta-analysis of randomized trials indicates that systematic lymphadenectomy may improve long-term survival compared to sampling [111]. Another important argument supporting the use of systematic nodal dissection is based on anatomical studies showing that lymphatic drainage patterns are not always strictly lobe-specific, and unpredictable nodal spread can occur [25,112,113]. The risk of a lobe-specific nodal evaluation is that when nodal involvement is found beyond the lobe-specific nodal stations prognosis is worse [114].

In 2013, ACCP recommended that systematic mediastinal lymph node sampling or dissection at the time of anatomic resection should be performed over selective or no sampling to ensure accurate staging in patients with clinical stage I and II NSCLC [115,116]. More than a decade later, current evidence continues to support that systematic nodal evaluation – particularly systematic nodal dissection – remains essential to ensure a complete resection and to improve pathological staging accuracy (GRADE 1B), which is critical for determining prognosis and guiding adjuvant therapy decisions.

The risk of lymph node metastasis in peripheral ≤2cm clinical (c)IA NSCLC ranges between 10 and 20% [117–120] and this is especially significant (15%) if consolidation to tumour ratio is over 0.75 [119]. The detection rate correlates with the number of nodes examined [121,122], with the higher number of lymph nodes examined, the better survival [123], suggesting that more than 9–10 lymph nodes examined provides better survival [121,123–125]. The most frequent site of nodal metastasis are peripheral N1 stations [118,119,126] and examination of both N1 and N2 stations correlates with better survival than examination of N2 stations only [127]. Adjacent lobar-to-segmental stations (#12, #13) are most commonly affected [128] and some controversy remains regarding completion lobectomy if these nodes are found positive intraoperatively due to the risk of isolated lobar-to-segmental positive nodes in other segments [126,129], especially in pure-solid tumours [117]. Non-adjacent interlobar metastasis are very uncommon both in part-solid and pure-solid tumours [117]. There is no evidence from randomized controlled trials (RCT) comparing lobe-specific and systematic lymph node dissection, but evidence from systematic reviews, meta-analyses and retrospective studies suggests no survival differences in this specific cohort of patients with stage cIA (≤2cm) tumours [130,131].

Regarding those tumours less than 1cm, the risk of nodal involvement is low, and there are no overall and lung cancer-specific survival differences if just 1–3 lymph nodes or >4 lymph nodes are examined [132]. In addition, there is no direct comparative evidence of performing lobe-specific or systematic lymph node dissection in this subset of stage cIA tumours (≤1cm tumours) undergoing sublobar resection, but retrospective data and database analyses suggest no survival benefit [133,134]. The risk of nodal metastasis in part-solid tumours, especially with solid component less than 6mm is really low, thus lymph node examination provides no survival benefit [117,135]. Pure GGO or GGO-dominant lesions (consolidation/tumour [C/T] ratio <0.5) less than 3cm present a minimal risk of nodal upstaging (<1%), so lymph node dissection can be precluded [136,137] (GRADE 1A).

Robot-assisted thoracic surgery (RATS) is increasingly being incorporated into standard practice as a minimally invasive approach in oncologic thoracic surgery. Robotic lymphadenectomy has been shown to provide a more thorough nodal dissection than video-assisted thoracoscopic surgery (VATS). Multiple studies have consistently reported a higher number of resected lymph nodes and a greater number of nodal stations examined with the robotic approach [138–141]. Furthermore, RATS also seems to offer intraoperative superiority for intersegmental and segment-specific lymph node dissections, particularly in early-stages [142]. However, there is currently insufficient high-level evidence demonstrating that RATS results in higher nodal upstaging rates or improved short and/or long-term oncological outcomes compared to VATS or open surgery [140,142–144]. When applied to cohorts of patients with a higher incidence of N1–N2 involvement, an increase in nodal upstaging with RATS approach is observed. This may reflect a more accurate staging and potentially improved disease-free survival [145]. To date, there are three randomized trials – RAVAL [146], ROMAN [147], and RVlob [148] – that have directly compared RATS and VATS for lung resection. All three consistently reported that RATS cohorts had higher number of lymph nodes and nodal stations harvested (GRADE 1A) and suggested a potential lower postoperative complication rate compared to VATS.

Stage IA, according to the ninth edition of the TNM classification, includes tumours measuring 3cm or less, and is subdivided into stage IA1 (T1a: ≤1cm), stage IA2 (T1b: >1 to ≤2cm), and stage IA3 (T1c: >2 to ≤3cm), showing a progressive decrease in survival as tumour size increases [15]. The prevalence of occult N2 disease in cT1N0 lung cancer ranges from 1.6% to 9.8% [24,155–163].

Historically, centrally located cT1 tumours have been associated with a higher risk of occult N2 disease. Regarding the definition of central tumour, the ESTS considers tumours located within the inner two-thirds of the lung, whereas the ACCP defines them as those confined to the inner one-third [2,3]. Nonetheless, the role of central tumour location as a predictor of occult mediastinal nodal disease is controversial. Several studies including exclusively patients with tumours classified as cT1 have demonstrated that, regardless of the proposed definition of central location, there are no differences between central and peripheral T1 tumours in terms of occult mediastinal nodal disease [164–167]. However, while no differences in pN2 rate have been demonstrated between peripheral and central cT1 tumours, differences between pN0 and pN ≥1 (no lymph node metastases vs. any nodal involvement [N1 or N2]) have been described [167]. In a study of Takamori et al. including 719 patients with cT1 tumours, central tumour location was associated with a higher risk of any occult nodal disease, although these differences were not observed in patients with tumours smaller than 2cm (T1a and T1b) [168]. In this subgroup, the rate of occult N2 disease is very low, ranging from 2% to 6% [169,170]. Given this low prevalence, the benefit of invasive staging in patients with tumours ≤2cm appears limited.

For the subgroup of peripheral tumours without suspicion of hilar or mediastinal lymph nodes metastases, current guidelines recommend that direct operation can be performed [2–5]. However, several risk factors have been identified through multivariable analyses and prediction models as independently associated with increased risk of occult N2. These include higher PET SUVmax, larger tumour size within the T1 category, consolidation/tumour ratio, serum carcinoembryonic antigen (CEA) level and certain histologic subtypes such as invasive adenocarcinoma [19,156,158,160,171–176].

According to present guidelines, in clinical stage IA, invasive mediastinal staging is not recommended for T1a and T1b tumours. For selected T1a and T1b tumours with one or more additional clinical or radiological risk factors mentioned above, an increased risk of mediastinal metastases has been observed; consequently, invasive mediastinal staging is advised. (GRADE 2B) (Fig. 5). It is evident that not all cancers have all these prognostic factors at the time of their preoperative evaluation. In many cases, small tumours do not have a histologic diagnosis, and even less the cell subtype, when their indication for resection is established. The practical idea is that, if these prognostic factors are available, they should count at the time of deciding invasive mediastinal staging. The accuracy of invasive mediastinal staging methods in patients with a normal mediastinum on PET/CT, and the rationale for selecting the most appropriate technique are discussed in the following section, which addresses patients with an intermediate risk of unsuspected N2 disease.

Fig. 5.

Suggested algorithm for mediastinal nodal staging in patients with non-metastatic NSCLC.

The rate of unsuspected mediastinal nodal disease ranges from 20% to 42% [23,24,79,177,178] for cN1 tumours, 6% to 22% [24,91,157,178,179] for those cN0 tumours greater than 3cm, and 7% to 17.1% [35,180,181] for central tumours. For this subgroup of tumours with intermediate risk of N2, current guidelines recommend invasive mediastinal staging over staging imaging alone [2–5]. However, as reflected in previous guidelines (ESTS, ACCP), no single invasive staging technique is uniformly recommended as the initial approach. The choice between endosonographic methods (EBUS/EUS) and surgical staging procedures (VAM with lymph node biopsy or transcervical lymphadenectomies such as VAMLA or TEMLA) is generally guided by local availability and expertise [2,3].

The diagnostic performance of endosonographic methods for mediastinal staging of cN0-1 NSCLC is low. Two meta-analyses of EBUS-TBNA in cN0-1 NSCLC reported a pooled sensitivity of 0.49 (95% CI, 0.41–0.56) and 0.49 (95% CI, 0.36–0.62), respectively. In the results of subgroup analysis conducted by dividing studies into those with only radiological N0 versus N0/1, pooled sensitivities of EBUS-TBNA were similar between the two groups (0.47 vs 0.5) [41,43].

Few studies have specifically evaluated the performance of EBUS-TBNA in a cohort of patients with normal mediastinum and a unique and clearly well-defined indication for invasive staging (tumours >3cm, centrally located T1 tumours, or N1). Dooms and colleagues [23], based on the results from the first prospective multicentre study (Assessment of Surgical Mediastinal Staging Added to Endoscopic Ultrasound [ASTER] II) in patients with cN1 NSCLC patients, reported a sensitivity of 0.38. The same team [79] reported the results from the first prospective multicentre study (ASTER III) to evaluate the performance of surgical mediastinal staging (by VAM or by VAMLA) in patients with cN1 tumours obtaining a global sensitivity of 0.73 and a negative predictive value of 0.92. Based on the results from ASTER II and III, and the high rate of unsuspected N2, the recommended staging technique for cN1 tumours should be surgical methods such as VAM or VAMLA/TEMLA if local expertise is available (GRADE 1B) (Fig. 5). When endosonography (EBUS/EUS) is indicated as the first choice, negative results should be validated by invasive surgical staging methods (GRADE 1B) (Fig. 5).

For the rest of tumours with intermediate risk of N2 (tumours greater than 3cm, and central tumours) the debate continues regarding the optimal invasive staging strategy: whether endosonography methods should be preferred over surgical procedures as the initial test [181] or whether the choice between VAM, presurgical lymphadenectomies, or endoscopic staging by EBUS-TBNA/EUS- FNA should depend on local expertise [2]. Recently, a large prospective study evaluating the performance of VAMLA in a well-defined cohort of patients with cN0/1 NSCLC [24] reported a high accuracy (sensitivity of 0.98 and negative predictive value of 0.99) and the unsuspected (u) N2/N3 rate according to presurgical nodal and tumour categories assessed by PET/TC were: 3.6% in cT1N0; 16.3–32% in cT2-4N0; and 42% in cN1. The high diagnostic accuracy observed in this prospective cohort has been further supported by Greif et al. in a large retrospective series including more than 1600 patients, which has consistently confirmed the high sensitivity (0.98), negative predictive value (0.99), and very low false-negative rates (2.2%) of VAMLA. These data reinforce the reproducibility and technical reliability of transcervical lymphadenectomies when performed in experienced centres [93]. In addition, Herrmann et al. have demonstrated the added clinical value of VAMLA after negative endosonographic staging. In a real-world cohort of patients with potentially resectable NSCLC and negative EBUS-TBNA findings, subsequent VAMLA identified occult mediastinal N2/N3 disease in approximately 17% of cases, resulting in clinically relevant nodal upstaging and changes in therapeutic management. These findings support the role of surgical mediastinal staging after negative EBUS-TBNA, particularly in patients with intermediate-risk features or suspicious findings on PET/CT [94]. Based on this recent evidence [24,93,94], and considering the low accuracy of endosonography methods for staging cN0-1 NSCLC [41,43], it seems reasonable to consider transcervical lymphadenectomies (VAMLA/TEMLA) as the new reference staging procedure for staging cN0/1 NSCLC (GRADE 1B). However, there are few centres with experience in transcervical lymphadenectomies; therefore, if EBUS/EUS is performed for staging tumours greater than 3cm, negative results should be validated, at least, intraoperatively with a systematic nodal dissection at tumour resection as defined in guidelines.

Locally advanced NSCLC (patients with stage IIIA, IIIB and IIIC) is a highly heterogeneous disease that often displays a complex clinical profile and high tumour burden [15]. Such heterogeneity is explained, at least in part, by tumour size and magnitude of mediastinal nodal involvement.

The indication for invasive mediastinal staging is generally based on radiological evidence of N2/N3 nodal metastasis as enlarged mediastinal lymph nodes on CT of the chest or with elevated/pathologic SUV on PET or PET/CT [173,182]. It is especially recommended for patients who are potential candidates for resection (GRADE 1B). Using a SUV cut-off of 2.5, PET/CT has a sensitivity and specificity for identification of mediastinal nodal involvement of 0.8 and 0.88, respectively. However, false-positive results (20–25%) and false-negative (5%) may occur. For this reason, confirmatory tissue sampling is necessary to avoid incorrect staging [2,5,59].

Assessment of N3 nodes is requested, particularly when patients are being considered for radiotherapy to evaluate locally advanced disease [38,183]. However, the prevalence of metastasis in contralateral hilar (N3) nodes is very low when no metastasis is present in the N2 nodes. Therefore, the indication for assessing the contralateral hilar (N3) nodes typically arises only if metastases are detected in the N2 nodes [184].

In patients with a high suspicion of N2-3 involvement, based on mediastinal lymph node enlargement or PET uptake (and absence of distant metastases), needle-based techniques (EBUS-TBNA, EUS-FNA, or combined EBUS/EUS) are recommended over surgical staging as the initial diagnostic approach, particularly in candidates for radical surgical resection [2–5,44,182,185] (GRADE 1A) (Fig. 5). Given a post-test probability after a negative test of >0.10 for endosonography in a context of high index of suspicion for mediastinal nodal disease, and in particular when the endosonographic procedure does not meet the systematic and technical requirements described in the Systematics and requirements for mediastinal nodal endoscopic staging section, a confirmatory surgical exploration of the mediastinum (with VAM, VAMLA, VATS etc.) is warranted, as this can further reduce the post-test probability to <0.05 [2–5,39,186–188].

Patients with T4 NSCLC with extensive mediastinal involvement but without distant metastases [15] may be eligible for curative-intent treatment. The decision for surgical resection will be determined by a combination of anatomical criteria, nodal status, performance status and physiological reserve, and surgeon's expertise, and may be considered in a carefully selected subgroup of patients (N0-1) [189]. In this specific setting, where imaging methods are often inaccurate in delineating the true extent of mediastinal invasion to assess resectability, invasive exploration may be necessary. CT imaging typically identifies loss of fat planes between the primary tumour and adjacent mediastinal structures, suggestive of direct invasion, whereas lymph node involvement typically appears as discrete nodal enlargements within characteristic mediastinal stations [4,190,191]. PET/CT highlights areas of increased metabolic activity but lacks specificity to conclusively differentiate between direct tumour extension and nodal metastases (occult N2 or N3) [192]. For those selected candidates undergoing radical resection or multimodality treatment, invasive mediastinal staging is required to precisely determine the extent of disease and guide appropriate treatment strategies [3]. Endoscopic techniques such as EBUS-TBNA and EUS-FNA are preferred, while surgical staging is reserved for cases where endoscopic procedures are inconclusive, negative or technically unfeasible [3,72].

Another clinical scenario involves patients with lung tumours (without mediastinal invasion) and extensive mediastinal infiltration from bulky or confluent metastatic lymph nodes [2]. In this setting, radiographic evidence is usually sufficient for staging, and invasive confirmation is not routinely required. However, tissue sampling remains essential to confirm malignancy and obtain histologic and molecular information to guide therapy. The selection of the biopsy site – whether from the primary tumour or a mediastinal location – should be guided primarily by accessibility and patient safety rather than by potential differences in diagnostic yield [2,115].

Finally, another relevant issue concerns the role of invasive nodal evaluation in patients with left lung cancer even with normal mediastinum (N0-1). If this evaluation is not performed before tumour resection, it will not be done at the time of resection. Access to mediastinal nodal stations 4L, 2L, 4R and 2R from the left side – either by thoracotomy or minimally invasive approaches – is hindered by the aorto-pulmonary compound, and these stations will not be explored. A recent study of 200 VAMLAs for cancers of the left lung with negative mediastinum on PET-CT has reported a combined N2–N3 rate of 18.5% (37/200). Among these patients 16 were clinically N0 (4 T1, 6 T2, 2 T3, and 4 T4) and 21 were clinically N1 (3 T1, 13 T2, 3 T3, and 2 T4). Of these 37 positive cases, 11 were in nodal station 4L and 6 in nodal station 4R, which would have passed unnoticed for the anatomic reason mentioned above, if a preoperative nodal evaluation had not been performed [24]. Based on these recent findings and the unpredictable nodal spread – since lymphatic drainage patterns are not always lobe-specific – reported from several anatomical cadaveric studies [25,112,113] invasive mediastinal staging is also advised in cancers in left-sided lung cancers, even in the presence of a radiologically and metabolically normal mediastinum, due to the limitations of intraoperative nodal assessment and the risk of occult mediastinal nodal involvement (GRADE 2B).