Mediastinal staging and restaging techniques in locally advanced non-small cell lung cancer in the era of neoadjuvant and perioperative chemoimmunotherapy-a narrative review.

Introduction
Lung cancer (LC) is the leading cause of cancer-related death worldwide (1). Approximately 85% of LC cases are non-small cell lung cancer (NSCLC) (2). The cornerstone of curative treatment is complete surgical resection (3,4), however, nearly 70% of patients are diagnosed with advanced or locally advanced disease (LA-NSCLC) (5,6), which is unresectable and has an unfavourable prognosis. The 5-year overall survival (OS) rate for LA-NSCLC is 37%, compared with 65% for resected localized disease (7). Significant advances have been made over the past decade, including the incorporation of neoadjuvant and/or perioperative chemoimmunotherapy into the therapeutic landscape of LA-NSCLC, establishing it as the new standard of treatment (8-21).
LA-NSCLC is a heterogeneous disease in which mediastinal involvement determines the therapeutic approach. In this scenario, mediastinal staging and restaging techniques play a key role. Regardless of the available mediastinal staging and restaging methods, unforeseen pN2 disease is detected in a substantial proportion of cases. Historically, data on mediastinal staging and restaging techniques have primarily come from chemotherapy treatment settings. The purpose of this review is to identify the scientific evidence regarding mediastinal staging and restaging techniques in patients with LA-NSCLC undergoing neoadjuvant and perioperative chemoimmunotherapy. We present this article in accordance with the Narrative Review reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-1-1352/rc).

Methods
We searched the PubMed/MEDLINE and Google Scholar databases for publications. The search terms used were: NSCLC, stage III, neoadjuvant, mediastinal lymph node staging, and restaging. Articles were included if they were English-language publications from March 1987 to September 2025 and the following types of articles were eligible: systematic reviews, meta-analyses, observational studies, randomized clinical trials, and narrative reviews. Table 1 shows the methods and specifications of the underlying database research.

Results

Chemoimmunotherapy for LA-NSCLC
Historically, neoadjuvant and adjuvant chemotherapy for stages IB–IIIA NSCLC has been associated with a 5% increase in 5-year OS, providing a modest survival benefit (22-24). In view of the positive results in the context of metastatic disease (25-32), chemoimmunotherapy has been investigated as neoadjuvant and perioperative treatment for stages IB–IIIA NSCLC, with compelling evidence of its benefits on tumor response rates and survival (8-21,33-39). Consequently, different neoadjuvant and perioperative chemoimmunotherapy regimens have been approved by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) (40-44) and included in the main oncology management guidelines, which are considered the current standard of care for stages IB–IIIA NSCLC, without epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) alterations (45,46).

Chemoimmunotherapy utility in the neoadjuvant setting
Initial studies explored immunotherapy regimens (monotherapy or combination therapy). The LCMC3 study, a phase II, multicenter, open-label trial, evaluated neoadjuvant atezolizumab (two cycles) followed by surgery in 181 patients with IB-IIIB NSCLC [tumor, node, metastasis (TNM) 8th Edition]. Patients without disease progression could receive optional adjuvant atezolizumab for 12 months. The proportion of cN0, cN1, and cN2 disease was 41%, 27%, and 33%, respectively. After a median follow-up of three years, three-year progression-free survival (PFS) and OS rates were 72% and 80%, respectively. Surgery was performed in 88% of patients, with R0 resection in 76% and pathological complete response (pCR) in 6%. Grade ≥3 (G ≥3) treatment-related adverse events (TRAEs) occurred in 11% of patients. Postoperative complications were rare (9,47).
On the other hand, the NEOSTAR study, a phase II, single-center, open-label, randomized trial (1:1), compared neoadjuvant therapy with nivolumab (3 doses) plus ipilimumab (1 dose) or nivolumab alone (3 doses) in 44 patients with stage IA–IIIA NSCLC (TNM 7th Edition). In the combination arm, the distribution of cN0, cN1, and cN2 disease was 70%, 9%, and 22%, respectively, compared with 67%, 24%, and 10%, respectively, in the monotherapy arm. After a median follow-up of 62.1 months, median OS was not reached. 92% of patients in the combination group and 77% in the monotherapy group underwent surgery, with R0 resection in 100%, and pCR in 29% and 9%, respectively. TRAEs ≥G3 were reported in 22% and 19% of patients in the combination and monotherapy arms, respectively, with treatment discontinuation in 9% and 15%, respectively. Postoperative complications were infrequent (21).
The NEOCOAST study, a phase II, multicenter, open-label, randomized trial (1:1:1:1), explored neoadjuvant immunotherapy regimens in 84 patients with stage IA3–IIIA NSCLC (TNM 8th Edition). Patients received one cycle of durvalumab either as monotherapy or in combination with oleclumab, monalizumab, or danvatirsen. Surgery was performed in 92% of patients, with pCR of 3.7%, 9.5%, 10%, and 12.5%, for monotherapy, combination with oleclumab, monalizumab, and danvatirsen, respectively. TRAEs ≥G3 were documented in less than 5% of patients (11).
Subsequent studies explored neoadjuvant chemoimmunotherapy. The NADIM study, a phase II, multicenter, open-label trial, enrolled 46 patients with stage IIIA NSCLC (TNM 7th Edition) to receive neoadjuvant nivolumab, carboplatin, and paclitaxel (three cycles), followed by surgery and adjuvant nivolumab for 1 year. Tumors with EGFR mutations and/or ALK translocations were excluded. The proportion of cN0, cN1 and cN2 was 20%, 7% and 74%, respectively (10). After a median follow-up of 60 months, the 5-year PFS and OS rates were 65% and 69.3%, respectively (20). Surgery was performed in 89% of patients, achieving R0 resection in 100% and pCR in 63.4%. The rate of TRAEs ≥G3 during neoadjuvant and adjuvant therapy was 30% and 19%, respectively, with treatment discontinuation in 0% and 14% of patients, respectively. Postoperative complications occurred in 29% of patients (10).
On the other side, the CheckMate 816 study, a phase III, multicenter, open-label, randomized trial (1:1), compared neoadjuvant nivolumab plus platinum-based chemotherapy (3 cycles) or chemotherapy alone in 505 patients with stage IB–IIIA NSCLC (TNM 7th Edition). Tumors with EGFR mutations and/or ALK translocations were excluded. After a median follow-up of 68.4 months, the 5-year event-free survival (EFS) and OS rates in the experimental group were 49.2% and 65.4%, respectively, compared with 34.4% and 55%, respectively, in the control group (48). Eighty-four percent of patients in the experimental arm and 76% in the control arm underwent surgery, achieving R0 resection in 84% and 78%, respectively, and pCR in 24% and 2.2%, respectively. TRAEs ≥G3 were reported in 34% and 6% of patients in the chemoimmunotherapy and chemotherapy groups, respectively, with treatment discontinuation in 37% and 4%, respectively (8). Postoperative complications were documented in 45% and 48% of patients in the experimental and control groups, respectively (48).
Additionally, the NADIM II study, a phase II, multicenter, open-label, randomized trial (2:1), enrolled 86 patients with stage IIIA–IIIB NSCLC (TNM 8th Edition) to receive neoadjuvant nivolumab, carboplatin, and paclitaxel (three cycles) or chemotherapy, followed by surgery, and adjuvant nivolumab for six months in the experimental arm. Tumors with EGFR mutations and/or ALK translocations were excluded. In the experimental group, the distribution of cN0, cN1, and cN2 disease was 11%, 18%, and 72%, respectively, compared with 31%, 14%, and 55%, respectively, in the control group. After a median follow-up of 24 months, 2-year PFS and OS rates in the experimental arm were 67.2% and 85%, respectively, compared with 40.9% and 63.6%, respectively, in the control arm. 93% of patients in the experimental group and 69% in the control group underwent surgery, achieving R0 resection in 94% and 85%, respectively, nodal downstaging in 72% and 40%, respectively, and pCR in 37% and 7%, respectively. The rate of TRAEs ≥G3 during neoadjuvant therapy was 22% and 10% in the experimental and control groups, respectively, with treatment discontinuation in 1%, and 5% during adjuvant therapy. Postoperative complications were reported in 42% and 35% of patients in the experimental and control groups, respectively (39).

Chemoimmunotherapy utility in the perioperative setting
Moreover, chemoimmunotherapy has been explored in the perioperative context (49,50). The KEYNOTE 671 study, a phase III, multicenter, double-blind, randomized trial (1:1), compared neoadjuvant pembrolizumab plus platinum-based chemotherapy (four cycles) or chemotherapy plus placebo, followed by surgery, and adjuvant pembrolizumab (13 cycles) or placebo in 797 patients with stage II–IIIB NSCLC (TNM 8th Edition). In the experimental arm, the distribution of cN0, cN1, and cN2 disease was 37%, 20%, and 42%, respectively, compared with 35%, 18%, and 47%, respectively, in the placebo arm (13). After a median follow-up of 41.1 months, the 48-month EFS and OS rates in the pembrolizumab arm were 51.9% and 68%, respectively, compared with 28.1% and 56.7%, respectively, in the placebo arm (34). Surgery was performed in 83% of patients in the experimental group and in 80% of the placebo group, with R0 resection in 92% and 85%, respectively, and pCR in 18.1% and 4%, respectively. TRAEs ≥G3 were documented in 41% and 37% of patients in the chemoimmunotherapy and chemotherapy groups, respectively, with treatment discontinuation in 13% and 6%, respectively. Postoperative complications occurred in 72% of patients (13).
On the other hand, the AEGEAN study, a phase III, multicenter, double-blind, randomized trial (1:1), evaluated neoadjuvant durvalumab plus platinum-based chemotherapy (four cycles) or chemotherapy plus placebo, followed by surgery and adjuvant durvalumab (12 cycles) or placebo in 802 patients with stage IIA–IIIB NSCLC (TNM 8th Edition). Tumors with EGFR mutations and/or ALK translocations were excluded. In the experimental arm, the distribution of cN0, cN1, and cN2 disease was 30%, 20%, and 50%, compared with 27%, 23%, and 50%, respectively, in the placebo arm. After a median follow-up of 25.9 months, the median EFS and OS had not been reached in the durvalumab arm, and were 30 and 53.2 months, respectively, in the placebo arm (17). Seventy-eight percent of patients in the experimental group and 77% in the control group underwent surgery, achieving R0 resection in 95% and 92%, respectively, nodal downstaging from N2 to N0 in 49% and 41%, respectively, and pCR in 17.2% and 4.3%, respectively (17,36). The rate of TRAEs G3–4 was 33% in both groups, with treatment discontinuation in 12% of patients in the durvalumab arm and 6% in the placebo arm. Postoperative complications were reported in 41% and 40% of patients in the experimental and control groups, respectively (17).
On the other side, the CheckMate 77T study, a phase III, multicenter, double-blind, randomized trial (1:1), compared neoadjuvant nivolumab plus platinum-based chemotherapy (four cycles) or chemotherapy plus placebo, followed by surgery and adjuvant nivolumab (one year) or placebo, in 461 patients with stage IIA–IIIB NSCLC (TNM 8th Edition). Tumors with EGFR mutations and/or ALK translocations were excluded. In the experimental arm, the proportion of cN0, cN1, and cN2 disease was 35%, 25%, and 40%, respectively, compared with 38%, 23%, and 40%, respectively, in the placebo arm. After a median follow-up of 41 months, the median EFS was 46.6 months in the experimental group compared with 16.9 months in the placebo group, the median OS was not reached in either arm (16). Surgery was performed in 78% of patients in the nivolumab arm and in 77% of the placebo arm, with R0 resection in 90% and 91%, respectively, and pCR in 25.3% and 4.7%, respectively. TRAEs ≥G3 were reported in 33% and 26% of patients in the chemoimmunotherapy and chemotherapy groups, respectively, with treatment discontinuation in 20% and 26%, respectively. Postoperative complications occurred in 41% and 39% of patients in the experimental and placebo arms, respectively (35).
The RATIONALE-315 study, a phase III, multicenter, double-blind, randomized trial (1:1), compared neoadjuvant tislelizumab plus platinum-based chemotherapy (3–4 cycles) or chemotherapy plus placebo, followed by surgery and adjuvant tislelizumab (8 cycles) or placebo, in 453 patients with stage II–IIIA NSCLC (TNM 8th Edition). Tumors with EGFR mutations and/or ALK translocations were excluded. In the experimental arm, the distribution of cN0, cN1, and cN2 disease was 27%, 37%, and 36%, respectively, compared with 24%, 41%, and 35%, respectively, in the placebo arm. After a median follow-up of 22 months, the 24-month EFS and OS rates in the tislelizumab arm were 68% and 89%, respectively, compared with 52% and 79%, respectively, in the placebo arm. 84% of patients in the experimental group and 76% in the placebo group underwent surgery, achieving R0 resection in 95% and 93%, respectively, and pCR in 41% and 6%, respectively. The rate of TRAEs ≥G3 was 72% in the chemoimmunotherapy arm and 66% in the chemotherapy arm, with treatment discontinuation in 8% and 9%, respectively. Postoperative complications were documented in 64% and 61% of patients in the experimental and placebo groups, respectively (19).
Recently, the NEOTORCH study, a phase III, multicenter, double-blind, randomized trial (1:1), evaluated neoadjuvant toripalimab plus platinum-based chemotherapy (four cycles) or chemotherapy plus placebo, followed by surgery and adjuvant toripalimab (13 cycles) or placebo, in 501 patients with stage II–IIIB NSCLC (TNM 8th Edition). Tumors with EGFR mutations and/or ALK translocations were excluded. In the experimental arm, the proportion of cN0, cN1, and cN2 disease was 8%, 23%, and 69%, respectively, compared with 9%, 19%, and 72%, respectively, in the placebo arm (14). After a median follow-up of 18.3 months, the two-year EFS rates in the toripalimab and placebo arms were 64.7% and 38.7%, respectivately, the median OS had not been reached in the toripalimab group and was 30 months in the placebo group (37). 83% of patients in the experimental arm and 74% in the placebo arm underwent surgery, achieving R0 resection in 96% and 93%, respectively, and pCR in 24.8% and 1%, respectively. The rate TRAEs ≥G3 was 64% in the experimental arm and 54% in the control arm, with treatment discontinuation in 10% and 8%, respectively (14).
In light of advances in the metastatic NSCLC landscape, antibody-drug conjugates (ADCs) have been evaluated in LA-NSCLC. The NEOCOAST-2 study, a phase II, multicenter, open-label, randomized trial, enrolled 202 patients with stage IIA–IIIB NSCLC (TNM 8th Edition) to receive perioperative chemoimmunotherapy plus novel agents, including the TROP-2 ADC datopotamab deruxtecan. Surgery was performed in 95% of patients in the datopotamab deruxtecan arm, with pCR of 35.2%. TRAEs ≥G3 were documented in 21% of patients treated with the TROP-2 ADC (51).
Table 2 summarizes the studies described above.

Evidence of mediastinal staging and restaging techniques in LA-NSCLC in the neoadjuvant and/or perioperative chemoimmunotherapy setting
The response to neoadjuvant chemoimmunotherapy can be assessed by pathological response (PR) and downstaging. The evaluation of these parameters has recently generated considerable interest due to their proven prognostic value. A pathologic complete response [0% residual viable tumor (RVT)] and a major pathologic response (MPR, ≤10% RVT) have been associated with a greater benefit from chemoimmunotherapy (52-54) and have been proposed as potential surrogates for OS (55,56).
Mediastinal staging and restaging techniques are used to determine both the indication for and the response to neoadjuvant treatment. LA-NSCLC is a heterogeneous disease in which mediastinal involvement determines the therapeutic approach, distinguishing between: (I) absence of mediastinal involvement (N0/N1), in which case initial surgery may be considered, (II) ipsilateral mediastinal involvement (N2), after neoadjuvant therapy may be possible, and (III) unresectable mediastinal involvement (N2/N3), definitive chemoradiotherapy is indicated. Despite current mediastinal staging techniques, unforeseen pN2 disease is detected in 25% of patients (1,57).
Furthermore, despite the beneficial improvements associated with neoadjuvant chemoimmunotherapy, ypN2 residual disease has been reported in a considerable proportion of patients in recent clinical trials (9,10,17,39,51). Considering the adverse prognoses associated with ypN2 and its impact on treatment decisions (58,59), restaging to detect residual nodal disease is essential.

Mediastinal staging
Mediastinal staging methods include, non-invasive imaged-based techniques such as computed tomography (CT) and positron emission tomography-computed tomography (PET-CT), minimally invasive endosonographic techniques, like endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and/or endoscopic ultrasound fine-needle aspiration (EUS-FNA), and surgical procedures such as standard cervical mediastinoscopy (CM), extended cervical mediastinoscopy (ECM), anterior mediastinotomy (AM), video-assisted mediastinoscopy (VAM), video-assisted thoracoscopic surgery (VATS), video-assisted mediastinal lymphadenectomy (VAMLA), and transcervical extended mediastinal lymphadenectomy (TEMLA) (60). The evidence in the literature regarding these techniques is derived mainly from the chemotherapy treatment setting.

Non-invasive mediastinal techniques
Chest CT and PET-CT are commonly used together as an initial diagnostic approach for NSCLC, as they allow for an initial evaluation of the primary tumor and potential mediastinal and distant involvement. However, they cannot provide tissue samples and have limited sensitivity and specificity (61-63).
The sensitivity and specificity of chest CT are 57% and 82%, respectively. These considered pathological by size when its short-axis diameter exceeds 1 cm (64,65). Nevertheless, micrometastases are detected in up to 20% of lymph nodes that are considered non-pathologic based on size. Therefore, it is recommended to evaluate additional parameters suggestive of malignancy (66).
Greater accuracy is achieved with PET-CT than with CT, due to its anatomical-functional integration, with sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) of 84%, 89%, 93%, and 79%, respectively (67). PET-CT is influenced by different parameters that reduce accuracy: tumor size ≤3 cm (67), lymph node size >1 cm (64), primary tumor location in the upper lobe (UL) or central region (63,66,68), N1 involvement (65,67), visceral pleural invasion, adenocarcinoma histology, low-grade differentiation, and/or primary tumor with high 18-fludeoxyglucose (FDG) uptake (1). In this regard, the accuracy of PET-CT in detecting N2 involvement decreases as tumor stage increases. False-negative (FN) N2 rates are 3.5% in cT1N0M0, 21–22% in cT2N0M0, and 17–25% in cT1/T2N1M0 (63). The lymph node stations with the highest rate of occult N2 disease are right paratracheal (4R) and subcarinal (7).
Furthermore, non-tumoral pathology, such as inflammatory or infectious diseases, associated with increased metabolic activity, can lead to false positives (FP) N2 on PET-CT (1,67).
The limited accuracy of CT and PET-CT can lead to incorrect staging, which negatively impacts treatment and prognosis. Overstaging may result in withholding potentially curative treatment, while understaging can lead to a futile thoracotomy with incidental pN2 disease (68). To avoid inaccurate staging, scientific societies recommend mediastinal staging based on minimally invasive and/or invasive procedures (61,63,64,66,67), as these techniques provide greater accuracy and enable tissue sampling (61-63). Exclusive radiological mediastinal staging is only accepted in two situations: (I) stage IA NSCLC (tumour size ≤3 cm, without hilar and/or mediastinal lymphadenopathy on CT/PET-CT) of peripheral location (outer third of the lung), in which surgery may initially be considered, and (II) stage IIIA bulky N2 NSCLC (defined as a lymph node cluster with a short axis >25 mm and/or mediastinal invasion preventing measurement or lymph node distinction), as it is considered unresectable (61,64,66).

Minimally invasive mediastinal techniques
Historically, mediastinoscopy has been considered the gold standard for mediastinal staging due to its high diagnostic accuracy (67,69,70). However, its potential risks have relegated this technique to a secondary role. Current guidelines position EBUS-TBNA as the initial test in minimally invasive/invasive mediastinal staging (63,68,70,71).
EBUS-TBNA and EUS-FNA are endoscopic ultrasound techniques that allow for assessment of hilar and mediastinal lymph node stations. EBUS-TBNA enables exploration of the hilar (10R/L), interlobar (11R/L), lobar (12R/L), upper paratracheal (2R/L), lower paratracheal (4R/L), and subcarinal (7) stations (70,72), but not the paraesophageal (8R/L) and pulmonary ligament (9R/L) stations. EUS-FNA allows evaluation of stations 2L, 4L, 7, 8, and 9 (61,67), as well as the celiac lymph nodes, left adrenal gland, and liver. Considering EBUS-TBNA’s limited access to stations 8 and 9, a combined approach, known as mediastinal endosonography, is recommended. The isolated use of EBUS-TBNA is accepted, due to the infrequent involvement of stations 8 and/or 9 and the questionable impact of their evaluation on survival (61). Mediastinal endosonography does not allow assessment of the aorticopulmonary window (5) and para-aortic (6) lymph nodes. However, transesophageal access may, in some cases, permit examination of the aorticopulmonary window (66).
EBUS-TBNA and EUS-FNA have a specificity and PPV close to 100%. The sensitivity and NPV of EBUS-TBNA are 46–97% and 90%, respectively (57,70,73), while those of EUS-FNA are 78% and 90%, respectively (63). Mediastinal endosonography and transesophageal evaluation of the aortopulmonary window are associated with increased sensitivity, 83–94% and 91%, respectively (67).
The accuracy of EBUS-TBNA and EUS-FNA is influenced by factors such as tumor location (higher FN rate in left hemithorax), lymph node characteristics, and operator expertise, among others (70,71,74).
EBUS-TBNA and EUS-FNA have similar overall and serious complication rates of 1% and 0.04% (64,68,71), respectively (67,72,75).
In some cases, tissue sampling by EBUS-TBNA and/or EUS-FNA is not feasible, or non-diagnostic. In these situations, when there is a high clinical suspicion of lymph node involvement, individualized assessment by a multidisciplinary committee is recommended to determine the most appropriate technique for new tissue sampling (76-83). Similarly, when EBUS-TBNA and/or EUS-FNA results are negative, an evaluation by a multidisciplinary committee is suggested to determine the need for a new biopsy. Conversely, a positive EBUS-TBNA result does not require additional confirmatory testing due to its high specificity and PPV.
Historically, VAM was the standard procedure after a negative EBUS-TBNA. However, there is currently no consensus on the necessity of additional VAM. While many contemporary guidelines recommend VAM after a negative EBUS-TBNA, the European Society of Thoracic Surgeons (ESTS) guidelines disagree, stating that in centers with EBUS-TBNA expertise, a negative result does not require additional VAM (65,71).
EBUS-TBNA and VAM provide access to the same nodal stations, some authors argue that VAM has an advantage due to its video-camera guidance, whereas others maintain that EBUS-TBNA offsets this advantage through elastography (71,84).
Moreover, VAM has traditionally been associated with higher sensitivity, lower FN rates, and increased morbidity (70,85). Although VAM sensitivity is slightly superior to that of EBUS-TBNA, 86% and 83%, respectively (71,85), it decreases to 67% when is performed after a negative EBUS-TBNA. VAM sensitivity is also influenced by PET-CT results, with reduced sensitivity and a higher number needed to treat (NNT) when PET-CT is negative (57,71,86-88). In light of these considerations, in order to determine the need for additional VAM after a negative EBUS-TBNA, it is recommended to evaluate the probability of occult N2 disease, the NPV, and PET-CT results. If the probability of N2 involvement is <2.5% (71,74), the NPV >90% (64), and PET-CT negative, invasive staging is not considered cost-effective and is not indicated (71,74). Conversely, if the probability of N2 is >57% (71,74), the NPV <90% (64), and/or PET-CT positive, invasive staging is recommended (71,74).

Invasive mediastinal techniques
Similar to minimally invasive procedures, invasive techniques have a specificity and PPV close to 100%, although their sensitivity and NPV are higher. Table 3 summarizes the sensitivity, specificity, NPV and PPV of minimally invasive and invasive techniques. Invasive procedures are associated with a higher number and greater severity of complications, 1.7–2.5% (62,71).
CM has long been considered to be the gold standard for NSCLC mediastinal staging. Performed through the insertion of a mediastinoscope following a transverse cervicotomy, it allows exploration of stations 2, 4, prevascular-retrotracheal (3), 7, and 10. Unlike other invasive procedures, it does not permit lymph node dissection (67). The sensitivity and NPV of CM are 86% and 94.5%, respectively (89). Most FN are due to insufficient sampling, therefore, biopsy of at least 5 nodal stations is recommended (66), including stations 4 and 7, and, when identifiable, station 2 (65,96). If pN3 is detected during the procedure, the exploration should cease. The complication rate is low, with virtually no mortality (64,65,96).
VAM is performed through the same incision as CM and allows exploration of similar nodal stations (66). Nevertheless, VAM permits lymph node dissection (67). It shows a sensitivity and NPV of 81–89% and 90–92%, respectively. Combining VAM with EBUS-TBNA and EUS-FNA increases sensitivity to 96.7%, however, the rate of unforeseen pN2 is 7–9.3% (61).
In contrast to CM and VAM, stations 5, 6, 8, and 9 can be accessed using ECM, AM, or VATS. These techniques are recommended in case of lymph node enlargement and/or 18-FDG uptake in stations 5 and/or 6, a situation commonly seen in upper-lobe tumors, especially in the left upper lobe (LUL) (90-92).
ECM employs the same incision and approach as CM but requires perforation of the fascia anterior to the aortic arch to reach lymph node stations 5 and 6. ECM entails a higher surgical risk. It allows evaluation of stations 2, 4, 5, 6, and 7 (64,66), with a sensitivity of 61–71% and NPV of 91–94% (66).
In left AM, stations 5 and 6 are explored through a second incision in the second left parasternal intercostal space. Sometimes partial resection of the third costal cartilage is required. Compared with ECM, it demonstrates higher sensitivity and NPV, at 86% and 89%, respectively (91).
VATS is performed through 1–3 intercostal incisions and requires single-lung ventilation. It enables exploration of stations 5, 6, 8, and 9. In contrast to ECM and AM, VATS exhibits higher sensitivity and NPV, close to 100% (92). However, it is more invasive than VAM (93,94), and is usually restricted to the assessment of stations 5 and 6 (66).
Mediastinopleuroscopy allows for simultaneous evaluation of lymph node and pleural spread. This technique should be performed with single-lung ventilation (64).
VAMLA and TEMLA are mediastinal staging techniques that are associated with higher accuracy but greater risk of complications. VAMLA is a transcervical videomediastinoscopic procedure that permits bilateral assessment and lymph node dissection of stations 2, 4, 7, and 8. TEMLA is a surgical technique that, through a cervical incision (5–8 cm) and sternal elevation, enables a more extensive lymphadenectomy of the supraclavicular (1), 2, 3, 4, 5, 6, 7, 8, and 10 stations (60,66). The sensitivity and NPV of VAMLA are 96% and 97%, respectively (66,93,94), and for TEMLA 98% and 99%, respectively (66,95). Both procedures are associated with the development of adhesions and increased technical complexity in subsequent interventions (67). Indications for VAMLA or TEMLA include centrally located tumors without known cN2–3 involvement but with high suspicion of it, tumors in the left hemithorax with cN1 involvement, and bilateral synchronous tumors (66,94).
Table 4 shows lymph node station accessibility of minimally invasive and invasive mediastinal staging techniques.

Minimum requirements for adequate mediastinal staging
Current guidelines recommend that, to be considered adequate, mediastinal staging should include exploration and biopsy of lymph node stations 4 and 7, with assessment and biopsy of station 2 also recommended. In left-hemithorax tumors, stations 5, 6, 8, and 9 should be evaluated and sampled. Furthermore, any lymph node stations with findings suggestive of malignancy should be assessed and biopsied (67).

Mediastinal staging algorithm proposal
PET-CT is recommended as the initial approach for mediastinal and distant staging. Based on CT and/or PET-CT findings:
Peripheral stage IA NSCLC: no additional studies are required, initial surgery may be considered.

Stage IIIA bulky N2 NSCLC, considered unresectable: no further studies are indicated, and a multimodal approach is indicated (61,64,66).

NSCLC with tumor size >3 cm, central location, and/or suspected hilar–mediastinal involvement: minimally invasive mediastinal staging (EBUS-TBNA and/or EUS-FNA) is recommended (61,63,64,66,67):
❖ If EBUS-TBNA and/or EUS-FNA is positive: no further studies are required, and a multimodal approach is indicated (61,63-66,97).

❖ If EBUS-TBNA and/or EUS-FNA is negative: a multidisciplinary evaluation and consideration of the probability of occult N2 disease, the NPV, and PET-CT results, are recommended:
If PET-CT is negative, the probability of occult N2 is <2.5% (71,74) and the NPV is >90% (64): no additional studies are required, and initial surgery may be considered (71,74).

If PET-CT is positive, the probability of occult N2 is >57% (71,74) or the NPV is <90% (64): invasive mediastinal staging could be considered (67,70), with VAM as a possible option (64,71,89):
If VAM excludes N2, surgery may be considered.

If VAM confirms N2, a multimodal approach should be performed (64-66).

In addition to VAM, if station 5 involvement is suspected (UL tumors), VATS, ECM, and/or AM are recommended (91-93).

Figure 1 summarizes the mediastinal staging algorithm.

Intraoperative nodal staging: sampling vs. nodal dissection
In patients undergoing upfront surgery, the definitive pathological pN2 stage is determined based on the examination of resected tissue. Sufficient material must be obtained to ensure accurate staging (67).
The main intraoperative nodal staging methods are: systemic mediastinal lymphadenectomy, lobe-specific lymphadenectomy, and lymph node sampling, with no current consensus on the preferred method (98).
Systemic mediastinal lymphadenectomy, or systemic nodal dissection, aims at the en bloc removal of mediastinal lymph nodes within anatomical landmarks, including the surrounding fatty tissue as well as hilar and intrapulmonary nodes (99,100). Compared with other methods, it provides greater accuracy and is considered the gold standard.
Lobe-specific lymphadenectomy or lobe-specific lymph node dissection entails the resection of the lymphatic tissue associated with the affected lung lobe.
Lymph node sampling involves the selective removal of one or more nodes according to preoperative or intraoperative findings, with no consensus on the optimal extent.
Traditionally, systemic mediastinal lymphadenectomy has been associated with higher morbidity and mortality. Nevertheless, recent studies suggest that, although it entails a higher rate of intraoperative complications, there are no significant differences in postoperative morbidity and mortality (67).
In addition, these methods are part of surgical treatment, therefore, the selected technique is determined by tumor stage.
Several studies have evaluated the survival outcomes of mediastinal staging methods. Some report similar results across nodal resection strategies, with comparable identification of affected N1 and N2 nodes (101) and 5-year survival rates of approximately 45% for complete mediastinal resection and 43% for nodal sampling (102). However, other studies demonstrate better OS with complete mediastinal nodal dissection than with nodal sampling, in patients with pN1 and pN2 disease involving only one nodal station (103), right-sided carcinoma and/or pN2 (104), and in the overall population (105).
Furthermore, outcomes for nodal dissection techniques regarding tumor size and location have been evaluated. In peripheral stage I NSCLC with tumors <2 cm, lobe-specific lymphadenectomy has shown no inferiority to systematic mediastinal lymphadenectomy for disease control, staging and survival. Conversely, nodal sampling is considered inferior for local control and survival (105-107).
In contrast, for stage I NSCLC with tumors ≥2 cm or a central location, and for stage II pN0, the choice between lobe-specific and systemic mediastinal lymphadenectomy remains debated, as no significant differences in DFS or OS have been demonstrated. For stage II pN1/pN2 and stage III NSCLC, systemic mediastinal lymphadenectomy is preferred due to inferior outcomes with other methods for local disease control, disease-free survival (DFS), and OS (98).
Finally, according to the American College of Surgeons Commission, any curative-intent resection of primary pulmonary malignancy should include dissection of at least one hilar station and three distinct mediastinal nodal stations (108,109), one of which must be the subcarinal station (110). A minimum of six lymph nodes must be examined (111,112). This nodal assessment is accepted to define complete resection (113,114).

R0
resection
The principal aim of surgery for NSCLC is complete R0 resection. Historically, the Union for International Cancer Control (UICC) defined R0 as absence of tumor at the primary site, lymph nodes, and distant sites (111,115). However, some patients later experience recurrence, suggesting that this definition may be insufficient. Consequently, different authors have proposed heterogeneous complete resection definitions (108,113,114,116-118). The International Association for the Study of Lung Cancer (IASLC) finally established a complete resection definition as lobe-specific or systematic nodal dissection, the absence of extracapsular tumor extension, and removal of the highest mediastinal node, which must be negative (107,112).

Restaging
Mediastinal restaging allows both assessment of the response to neoadjuvant therapy and determination of the optimal therapeutic strategy (76,119,120).

Non-invasive techniques
CT and PET-CT present higher FP N2 rates of 33% (120) and 20–33% (65,120), respectively, and FN N2 rates of 33% (120) and 25% (65,120), respectively, due to post-treatment changes. PET-CT shows greater accuracy than CT, with sensitivity, specificity, PPV, and NPV, of 83%, 84%, 74%, and 91%, respectively.

Minimally invasive mediastinal techniques
Due to the limited accuracy of imaging techniques, minimally invasive staging methods (EBUS-TBNA, EUS-FNA) are also recommended to complement the assessment (66).
The use of EBUS-TBNA and/or EUS-FNA in restaging shows a sensitivity of 65–80% (65) and 73% (120), respectively. The NPV associated with EBUS-TBNA ranges from 20–78%, reflecting the interhospital heterogeneity in ypN2 (65,66,120). Mediastinal endosonography does not improve overall accuracy (65). According to published data, EBUS-TBNA and/or EUS-FNA achieve a correct diagnosis in 71% of patients and avoid other invasive procedures in 35% of cases (64). The complication rate is 3% (120).

Invasive mediastinal techniques
The performance of invasive procedures (CM, VATS, VAMLA, TEMLA) is reserved for cases of insufficient material or a high suspicion of N2 despite negative results (64,65,119). The decision to perform mediastinoscopy or VATS is based on the previous staging procedure. Due to adhesion formation, re-mediastinoscopy and re-VATS are associated with lower accuracy and higher complication rates (65,120,121). The sensitivity, specificity, PPV, and NPV of re-mediastinoscopy are 60–74%, 100%, 100%, and 73–86%, respectively (64,67,120,122-124), compared with mediastinoscopy at 81%, 100%, 100%, and 90%, respectively (120), and VATS (VATS and re-VATS) at 83-67%, 100%, 100%, and 64–73%, respectively (64,66). The complication rate for mediastinoscopy and VATS is below 4% (121). VATS has been suggested as an alternative to re-mediastinoscopy, however, experience with this approach remains limited (66).
TEMLA has historically been the technique used for restaging. In this context, VAMLA has emerged as an alternative to mediastinoscopy and TEMLA. Sensitivity, specificity, PPV, and NPV for VAMLA are 72–100%, 89–100%, 72–100%, and 89–100%, respectively (120), compared with TEMLA at 95–97%, 100%, 100%, and 97–99%, respectively (64,65). The complication rates for VAMLA and TEMLA are 9.7% (7.3% transient) and 9.7%, respectively. For left-sided tumors undergoing VAMLA, the addition of ECM is recommended (120).
Biopsy of lymph node stations with known disease at staging is recommended (regardless of adenopathic appearance), along with other stations to assess for progression (121).
Following mediastinoscopy for initial staging, VAMLA and TEMLA are discouraged for restaging due to a higher risk of complications, re-mediastinoscopy is therefore recommended instead (65).

Mediastinal restaging in the chemoimmunotherapy setting
Evidence on mediastinal restaging in the neoadjuvant or perioperative chemoimmunotherapy context is very limited, with available data exclusively from PET-CT and overall recommendations from expert consensus. The optimal method for restaging in this setting remains under debate.
With respect to expert consensus, despite the absence of homogeneity among them, it is generally accepted that, since no phase III trials on neoadjuvant therapies mandated mediastinal restaging, data on its clinical value remain limited. Consequently, in clinical practice, contrast-enhanced CT is used to assess if patients, after neoadjuvant therapy, can proceed to surgical resection, without the need for invasive mediastinal restaging (125,126).
Experts against invasive mediastinal restaging in the absence of radiologic progression highlight logistical difficulties, complication risks, and potential surgical delays (126,127). Experts in favor maintain the difference in management and prognosis based on downstaging from N2 to N0/N1 (21,59,125,128). While others recommend it only in specific situations, such as nodal immune flare (21). Overall, invasive mediastinal restaging after chemoimmunotherapy remains an area in which data from larger patient cohorts are needed, particularly from real-world studies that include surgical involvement. The general recommendation is that, in cases with radiologic findings that do not guarantee the feasibility of an R0 resection, an individualized evaluation by a multidisciplinary tumor board should be considered (126).
Regarding PET-CT, in light of the discordance between CT-assessed radiologic responses and histopathologic responses, reported at 41–45%, PET-CT has been suggested as a possible restaging technique due to its potentially greater accuracy compared with CT (129-131). Multiple studies have evaluated the accuracy of PET-CT after neoadjuvant chemoimmunotherapy and report results similar to those observed after neoadjuvant chemotherapy, with sensitivity, specificity, NPV, and PPV of 66–73%, 84–92%, 84%, and 67%, respectively (132).

Conclusions
N2 mediastinal nodal involvement is the most significant prognostic factor for patients with LA-NSCLC. Despite available mediastinal staging and restaging techniques, occult pN2 disease is detected in 25% of LA-NSCLC cases. Given the recent changes in the standard of care in the neoadjuvant setting, it is necessary to define the utility of these procedures in the current therapeutic scenario. This article reviews the role of mediastinal staging and restaging techniques in the neoadjuvant and/or perioperative approach to LA-NSCLC based on published literature. However, it should be noted that there is a significant limitation of scarce data on their performance in the era of chemoimmunotherapy, and most of the papers are based on observational studies.
Considering our findings and the new standard of treatment for patients with LA-NSCLC, further research is required to determine the utility of mediastinal staging techniques within the framework of current therapeutic strategies that include immunotherapy.

Supplementary
The article’s supplementary files as