Any suspected metastatic site identified on CT should be confirmed pathologically with biopsy. In general, biopsy sites should be selected to confirm highest possible disease stage and to maximize tissue yield.

Although there are no randomized controlled trials directly supporting this recommendation, the authors emphasize the importance of accurate staging when considering a treatment plan for patients with suspected stage III lung cancer. Evidence of metastasis, if present on history, physical examination, routine cross-sectional imaging (CT scan of the chest and upper abdomen), or FDG PET-CT scan, should prompt efforts to unequivocally confirm the stage with tissue biopsy if technically feasible, especially for suspected oligometastatic disease.

Studies examining the detection of occult metastases demonstrate that patients with clinical stage III disease and a negative clinical examination (defined here as history, physical examination, and CT imaging) have a clinically meaningful rate of detection of metastases on PET or PET-CT scan. For example, studies have demonstrated that 17%-24% of patients with clinical stage III lung cancer had unexpected stage IV disease detected by FDG-PET scan.¹³²

A MA of 56 studies including 8,699 patients demonstrated improved detection of occult metastasis with the use of integrated PET-CT scan compared with clinical staging without PET scan.¹³³ PET-CT scan, however, is not adequate to exclude CNS metastases, given the high obligate utilization of glucose by the brain. Studies of CNS imaging in asymptomatic patients with stage III lung cancer find a wide range (2%-21%) of patients with clinically occult CNS metastases.^134-136 Dedicated brain imaging, preferably with contrast-enhanced MRI scan (a contrast-enhanced head CT scan may be substituted if MRI is contraindicated) is, therefore, necessary to exclude clinically silent CNS metastases in patients with clinical stage III lung cancer.

The accuracy of imaging alone (CT or integrated PET-CT scan) is inadequate for confidently staging the mediastinum for most situations. A 2003 MA of 39 studies by Gould et al⁷ demonstrated point estimates from a pooled sensitivity of 0.61 (95% CI, 0.50 to 0.71) and a specificity of 0.79 (95% CI, 0.66 to 0.89) for CT scan alone. The same authors examined metabolic imaging with PET-CT scan and found a sensitivity of 0.85 (95% CI, 0.67 to 0.91) and a specificity of 0.90 (95% CI, 0.82 to 0.96).⁷ A 2014 Cochrane Review of the evidence for PET-CT–based staging of mediastinal nodes included 45 studies using different criteria to define a positive scan. The analysis showed point estimates of sensitivity and specificity of 81.3% (95% CI, 70.2 to 88.9) and 79.4% (95% CI, 70 to 86.5), respectively, using maximum standardized uptake values ≥ 2.5 on PET‐CT as the criterion. The same Cochrane review found that studies defining a positive node on PET-CT scan as standardized uptake value greater than mediastinal background had similar sensitivity (77.4%; 95% CI, 65.3 to 86.1) and slightly higher specificity (90.1%; 95% CI, 85.3 to 93.5), but still too low to substitute for pathologic confirmation. This study also found very high between‐study heterogeneity and wide 95% CIs around point estimates of sensitivity and specificity.¹³⁷ This inaccuracy dictates that mediastinal node metastasis, even for nodes with increased FDG activity on imaging, must be confirmed pathologically whenever feasible and that this is particularly important when the pretest probability of N2 nodal involvement is intermediate.

When pathologic confirmation of mediastinal stage is indicated, the choice of invasive staging should be based on the accuracy of the technique used, the availability of timely access to accurate staging, the cost(s), safety, and availability of experienced clinicians to perform the procedure. Pooled point estimates of the sensitivity of EBUS (0.81; 95% CI, 0.75 to 0.86) compared with mediastinoscopy (0.81; 95% CI, 0.75 to 0.86) for diagnosing mediastinal lymph node involvement in patients with potentially operable NSCLC suggest that the two approaches are equivalent. The specificity of both methods is similar.⁴ A 2019 MA found that patients with clinical N0 or N1 disease by imaging have a prevalence of N2 nodal disease of 0.15 (95% CI, 0.06 to 0.24) and that the mean negative predictive value for EBUS in these patients is 91% (95% CI, 82 to 100).⁵ This and other studies conclude that the rate of detection of N2 or N3 disease may be meaningfully increased when EUS is added to EBUS, such that the number of patients needed to undergo endoscopic staging to detect occult N2 or N3 involvement is 14 (95% CI, 10.8 to 16.3) when performing EBUS alone and decreases to approximately seven when also performing EUS.^5,6 A small prospective randomized trial did not find any additive value of performing EUS in patients who first underwent EBUS.²⁹ Randomized trials^30,138 and aggregate data from these MAs demonstrate that endoscopic-based staging is at least equivalent to mediastinoscopy and is more cost-effective,^139,140 at least when used as the first staging modality. A 2017 decision analysis suggests that the cost-effectiveness of EBUS-based mediastinal staging varies by the pretest probability of N2 disease with mediastinal staging not cost-effective in clinical N0 patients if the probability of N2 involvement is very low (< 2.5%). In patients with a probability of mediastinal metastasis from 2.5% to 57%, EBUS-transbronchial fine needle aspiration was considered cost-effective (using the threshold of $80,000 US dollars per quality adjusted life year) as the only staging modality. The same study suggested that confirmatory mediastinoscopy is cost-effective when the risk of N2 metastases was > 57% after a negative or nondiagnostic EBUS-transbronchial fine needle aspiration.¹⁴¹ A useful model (Help with Oncologic Mediastinal Evaluation for Radiation; HOMER) has been published online for predicting the probability of nodal metastasis.^142,143

Direct comparisons of treatment outcomes for patients with stage III NSCLC with and without multidisciplinary team care have not been performed. However, a study using propensity score matching of patients taken from a national (Taiwan) cancer registry suggests the advantage of treatment in the context of a multidisciplinary team. Outcomes in patients with stage III and IV NSCLC treated with multidisciplinary care were significantly better than those in patients treated outside the context of a multidisciplinary team (adjusted hazard ratio [HR] for mortality = 0.87; 95% CI, 0.84 to 0.90).¹⁴⁴ A study from a single center found that adherence to multidisciplinary recommendations was associated with improved overall and PFS compared with patients for whom care deviated from those recommendations.¹⁴⁵ Given the complexity, and heterogeneity of patients with stage III lung cancer, and the varying impact of comorbidities on suitability for systemic chemotherapy, thoracic radiation, and lung resection, the authors of this guideline emphasize the importance of the input of a multidisciplinary team, consisting of at least a medical oncologist, a radiation oncologist, and a thoracic surgeon, all of whom devote a significant portion of their clinical care to thoracic oncology. Outside of emergent, life-threatening situations, this discussion should take place before initiating any plan of treatment to allow for the coordination and proper timing of multimodality treatment.

Biopsy should generally be performed from the site that would establish the highest stage when feasible. Potential tissue yield for pathologic analysis and molecular sequencing should also be considered.

Curative-intent treatment involves multimodal treatment; surgery, radiation, or systemic therapy alone is not optimal. Both chemoradiotherapy without surgery and chemotherapy (with or without radiation) with surgery are treatment options. RCTs comparing multimodal treatment with or without surgery have demonstrated equivalent OS (ie, for overlapping CI, see evidence in Table 1).^31,33,35,147 This result has been consistent among the RCTs although the trials involved differences in treatment regimens and inclusion criteria (Table 2). Three of these trials accrued patients primarily from 1990 to 2000, whereas the other two primarily accrued from 2000 to 2010.^31,147 Surgical multimodality treatment was compared against definitive concurrent chemoradiotherapy in two trials and against sequential chemotherapy and then radiotherapy in the other three trials. Some trials permitted patients with bulky mediastinal involvement or unresectable disease; some trials randomly assigned patients only after assessment of response to induction therapy.

TABLE 1. Evidence Table Regarding Multimodality Treatment With or Without Surgery for Stage III NSCLC

View larger version (988K)

TABLE 2. Characteristics of Randomized Clinical Trials Involving Multimodality Treatment With or Without Surgery for Stage III Non–Small-Cell Lung Cancer

View larger version (402K)

Although avoiding surgery if not proven to be beneficial is a reasonable recommendation, there are factors that justify consideration of surgery in a multimodality approach. Currently, surgery for NSCLC, even with stage III disease, in most experienced centers primarily involves minimally invasive approaches. Although not significant, the HR for OS generally favors the surgery arms in the RCTs (with one exception)³³ and one of the five RCTs demonstrated a statistically significantly better PFS.³² Patient preferences are also an important consideration.

Given the evidence, it is reasonable to ask whether there are specific patient or tumor characteristics that would favor definitive chemoradiotherapy or a multimodality approach including surgery. It is important to note that there are no data from RCTs that clearly establish the superiority of one or the other approach in any specific subgroup. Thus, selection of the approach is based on extrapolation of data and expert consensus.

As a first step, evaluate for factors that argue against surgery. This includes the inability to achieve a complete resection, as an incomplete resection does not improve outcomes relative to no resection at all. This also includes the presence of N3 node involvement, as surgery generally does not include removal of contralateral or supraclavicular nodes, resulting in an incomplete resection. Finally, this may include extensive mediastinal involvement (ie, that obscures identification of discrete nodes and American College of Chest Physicians type D N2,3 involvement).¹⁴⁸

A high perioperative mortality can offset any potential long-term survival advantage to a surgical multimodality approach. This was demonstrated in a retrospective matching analysis of the Intergroup 0139 study, which compared neoadjuvant chemoradiation followed by surgery versus definitive chemoradiation.³² This study had high perioperative mortality after pneumonectomy (26%) compared with lobectomy (1%). Matching features of patients in the nonsurgical arm to those in the surgical arm (ie, those who underwent pneumonectomy or lobectomy) demonstrated that the high perioperative mortality in the pneumonectomy arm resulted in persistently worse long-term OS in the surgical arm, whereas those patients who underwent lobectomy had better long-term OS in the surgical arm. It should be noted that perioperative mortality involves many factors besides the extent of resection (eg, open v video-assisted thoracoscopic surgery approach, pulmonary reserve, and comorbidities). Many series report low perioperative mortality after neoadjuvant treatment and pneumonectomy (average 7% in a MA),¹⁴⁹ suggesting that pneumonectomy alone is not an adequate marker of a perioperative risk that overshadows any potential long-term benefit. Although arbitrary, we suggest that perioperative mortality above 5% would argue against a surgical multimodality treatment approach.

It is important to note that favorable prognostic factors (better outcomes regardless of treatment approach) should not be misinterpreted as predictive that the inclusion of surgery in the treatment approach is beneficial. Such prognostic factors include a low disease burden (primary tumor or extent of N2 involvement or both), young age, few comorbidities, and low PET activity. Similarly, treatment response to neoadjuvant therapy (tumor shrinkage, mediastinal downstaging, and decreased PET activity) is prognostic, but has not been demonstrated as predictive for improved outcomes with surgery.

The decision whether to include or exclude surgery as part of a multimodal treatment approach for patients with discrete N2 involvement should be made during a multidisciplinary discussion that includes a surgeon. This consensus reflects that the ability to achieve complete resection and the degree of perioperative risk are key factors; assessment of these requires surgical input. Tumors with overt mediastinal infiltration or clear N3 involvement may be reasonably excluded from surgery without surgical input.

T4N0 disease alone should not be viewed as a contraindication to the consideration of surgery.¹⁵⁰ Surgical feasibility and extent of resection must be considered when offering surgical resection. These decisions should be made by surgical physicians, rather than nonsurgical physicians. A review of surgical resection (with or without other modalities) demonstrates 5-year OS rates of approximately 30%.¹⁵¹ Limited data suggest possible benefit for preoperative chemo- or chemoradiotherapy^152-154 On the other hand, in database studies of T4N0M0 tumors on the basis of size alone, the use of preoperative therapy was not associated with significantly better survival.¹⁵⁵ The best results for large T4N1M0 tumors appears to be resection and adjuvant chemotherapy.¹⁵⁶

• Patients with stage III NSCLC generally should not be excluded from consideration for surgery by nonsurgical physicians.

• Presence of oncogenic driver alterations, available therapies, and patient characteristics should be taken into account.

• Patients and providers should consider enrollment on clinical trials when appropriate.

A MA in patients with resectable stage IB-IIIA NSCLC involving 15 randomized controlled trials demonstrated significantly improved OS, time to distant recurrence, and recurrence-free survival when preoperative chemotherapy was used compared with surgery alone.¹⁵⁷ This analysis included 2,385 patients and demonstrated a 13% reduction in the relative risk of death with the use of preoperative chemotherapy. There is no published randomized phase III trial in patients with stage III NSCLC comparing neoadjuvant versus adjuvant chemotherapy. The NATCH trial was conducted in earlier-stage disease (stages IA > 2 cm to II) and demonstrated identical disease survival rates for neoadjuvant versus adjuvant chemotherapy.¹⁵⁸ Possible downsides of a neoadjuvant versus adjuvant chemotherapy approach are a delay in surgical resection because of early progression or toxicity. Benefits of neoadjuvant chemotherapy include a high compliance to systemic platinum-based treatment (90% v 61% in the NATCH trial¹⁵⁸), reduction in tumor size with increased operability, early eradication or prevention of micrometastasis, and the opportunity to assess the efficacy of the neoadjuvant treatment. Multiple trials have demonstrated improved survival to be associated with nodal downstaging and pathologic complete response (CR) rates and R0 resection.^27,41

A MA of three randomized controlled trials in stage IIIA N2 NSCLC indicated higher rates of mediastinal downstaging, pathologic CR of mediastinal lymph nodes, and higher rates of R0 resections with the addition of radiotherapy to neoadjuvant chemotherapy. However, induction chemoradiation did not improve overall or PFS rates as compared with chemotherapy alone.¹¹ In addition, a large randomized controlled trial, which included a considerable proportion of patients with stage IIIB NSCLC (> 60%), did not also demonstrate improved survival with neoadjuvant chemoradiotherapy as compared with neoadjuvant chemotherapy despite improved pathologic response rates and mediastinal downstaging.^37,42

Intergroup 0139 compared neoadjuvant chemoradiation with definitive chemoradiation in patients with operable stage IIIA N2 NSCLC and identified a benefit in PFS with surgery but no OS benefit. However, in an unplanned subgroup analysis, an OS benefit was observed in the subgroup of patients treated with lobectomy, suggesting that neoadjuvant chemoradiation followed by surgery may represent a possible treatment option for well-selected patients who could undergo a lobectomy. However, as no individual phase III trial has demonstrated a significant survival benefit using neoadjuvant chemoradiotherapy compared with neoadjuvant chemotherapy alone, neoadjuvant chemotherapy alone is the preferred treatment approach.¹¹

Although local control in patients with locally advanced NSCLC has consistently been shown to correlate with OS,^18,159 local control is particularly important for superior sulcus tumors. Like for other patients with NSCLC, those with superior sulcus tumors who achieve local control have better survival than those who do not.¹⁶⁰ In addition, patients with superior sulcus tumors can uniquely present with brachial plexus invasion and Horner's syndrome, in addition to generalized edema, chest wall invasion, and shortness of breath,¹⁶¹ making local control critically important. However, control of locoregional disease is the major challenge when treating superior sulcus lung cancers,¹⁶² yet it remains challenging since these tumors have a predilection for early invasion into bone, vascular, and nervous structures at the apex of the chest.¹⁶³

An attractive option for optimizing local control in the lung apex is the use of concurrent chemoradiation followed by surgery. Induction therapy for patients with superior sulcus tumors may have an effect on micrometastatic disease and can reduce tumor bulk and increase resectability, and the combination of two local modalities with radiotherapy and surgery can help to maximize local control.¹⁶⁴ This trimodality approach with chemoradiation followed by surgery has been demonstrated to achieve excellent local control and prolonged OS in this patient population.^165,166

Intergroup 0160-SWOG 9416 assessed the use of two cycles of cisplatin and etoposide concurrently with radiation therapy (45 Gy in 1.8 Gy daily fractions) among 110 patients with T3-4, N0-1 superior sulcus NSCLC. Those with stable or responsive disease underwent resection, followed by two additional cycles of chemotherapy. In total, 88 patients underwent this trimodality approach, 83 of whom achieved a complete resection. Pathologic CR or minimal microscopic disease was achieved in 61 patients, with pathologic CR associated with improved OS (P = .02). The 5-year OS in patients who underwent this trimodality treatment was 54% after complete resection versus 44% for the whole cohort, and the median survival was 33 months for all eligible patients versus 94 months for patients who had an R0 resection.¹⁵³

In SWOG S0220, 44 eligible patients with T3-4, N0-1 superior sulcus NSCLC underwent induction therapy with cisplatin-etoposide concurrently with thoracic radiotherapy to 45 Gy, followed by surgical resection in nonprogressing patients and then consolidation docetaxel for three cycles. Twenty-nine patients underwent surgery, with all but one undergoing an R0 resection and 21 achieving a pathologic CR or near CR to neoadjuvant chemoradiation. This trimodality regimen achieved excellent local control, with the isolated site of first recurrence local in only two patients, and the 3-year OS was 61%.¹⁶⁷

There have been multiple randomized controlled trials to evaluate the benefit of adjuvant chemotherapy after surgical resection; some compared different chemotherapy regimens or chemotherapy versus best supportive care. Most of these studies included stage I-III NSCLC, and inclusion criteria were based on older versions of the American Joint Committee on Cancer staging system. Therefore, data must be extrapolated to focus on stage III patients. Overall, the results have been mixed, with several studies showing a benefit to adjuvant chemotherapy,^56,57,59,61 whereas others show little or mixed survival benefit.^44,48,58,60 The ANITA study presented by Douillard et al⁵⁷ is one of the largest and most recent RCTs conducted in this setting in which adjuvant cisplatin-based chemotherapy was compared with best supportive care in patients with completely resected IB-IIIA NSCLC. This study showed a 5-year OS benefit of 8.6% with adjuvant chemotherapy. Another frequently cited RCT, the International Adjuvant Lung Cancer Trial (IALT) also showed a 5-year OS benefit with cisplatin-based adjuvant chemotherapy although the effect was more muted at 4%.⁵⁹ However, the Adjuvant Lung Project Italy (ALPI) study, a RCT conducted in Italy,⁴⁵ did not demonstrate an OS benefit with adjuvant chemotherapy (HR, 0.96; 95% CI, 0.81 to 1.13; P = .589). This difference may be explained by the toxicity of the chemotherapy regimen used in ALPI (mitomycin, cisplatin, and vindesine) compared with what was used in ANITA (cisplatin and vinorelbine) and IALT (cisplatin doublets with etoposide, vinorelbine, vinblastine, or vindesine). Toxicity (both from the chemotherapy itself and the lack of optimal supportive care medications) likely played an important role in the negative results seen in other earlier trials.⁵⁸ A MA by the Lung Adjuvant Cisplatin Evaluation (LACE) Collaborative Group, which pooled and analyzed data from the five largest trials testing adjuvant cisplatin versus observation in surgically resected NSCLC, demonstrated a 5.4% 5-year improvement in OS, and this benefit was most pronounced in stage III patients (HR, 0.83; 95% CI, 0.72 to 0.94).¹⁶⁸

Cisplatin also has significant toxicities including ototoxicity, nephrotoxicity, and neurotoxicity, which may limit its use, especially in older patients. Carboplatin-based adjuvant chemotherapy has also been evaluated. Notably, Ou et al⁵⁶ randomly assigned surgical patients with resected stage III NSCLC to adjuvant carboplatin-based chemotherapy or observation and found a statistically significant improvement in OS with adjuvant chemotherapy (33 months v 24 months, P = .037). Therefore, although the bulk of adjuvant studies used cisplatin-based chemotherapy, carboplatin-based chemotherapy is a reasonable alternative for patients with contraindications to cisplatin. Although direct comparisons have not been made, it is reasonable to infer that there is no benefit from triplet chemotherapy. Therefore, for patients with completely resected stage III NSCLC who did not receive neoadjuvant treatment, adjuvant platinum-doublet chemotherapy should be offered.

The ADAURA trial evaluated 3 years of adjuvant osimertinib versus placebo in resected stage IB-IIIA NSCLC.⁴⁹ The primary end point of the study was DFS in patients with stage II-IIIA disease (unlike the OS end point in the adjuvant chemotherapy regimens discussed), which was shown to be significantly prolonged with the use of osimertinib, HR 0.17 (P < .001), with median DFS not reached in the osimertinib arm and 19.6 months in the placebo arm. In a subgroup analysis of patients with stage IIIA NSCLC, the DFS HR was 0.12 (95% CI, 0.07 to 0.20). OS data are still immature at the time of this guideline, and neither cure rate nor outcomes upon progression are yet known, but this approach is now approved by the US Food and Drug Administration (FDA) and should be discussed with eligible patients.

Although adjuvant chemotherapy was allowed on the ADAURA trial, it was not required before osimertinib initiation. Given the known OS benefit of chemotherapy in this population, the panel recommends adjuvant platinum chemotherapy for patients with chemotherapy eligible resected NSCLC with EGFR exon 19 deletion or exon 21 L858R mutations before consideration of osimertinib.

The use of postoperative radiotherapy (PORT) after resection of NSCLC has been controversial for several decades. Before the routine use of adjuvant platinum-based systemic therapy, PORT was evaluated as a means to reduce high rates of regional and distant failure after primary resection of NSCLC. Nine randomized trials conducted from 1966 to 1995 examined the utility of PORT compared with surgery alone and have been summarized in the PORT Meta-Analysis (930, 929, and 599). Overall, PORT was found to reduce local regional relapse (HR, 1.16; P = .005); however, OS was also reduced (5-year OS 58% v 53%, P = .001), driven by an increase in noncancer death in the PORT arm (18% v 11%) felt to stem from cardiopulmonary toxicity. This detriment seemed to be confined primarily to N0 and N1 disease, with a potential benefit for N2 disease. Notable criticisms of these trials relate to the time period that they were run, the use of outdated radiotherapy equipment and techniques, and larger doses and larger treatment volumes that are known to increase the risk of toxicity compared with modern radiotherapy. However, a subgroup analysis of a randomized trial and multiple population-based analyses in more modern eras have suggested that PORT may be associated with a survival benefit for patients with pN2 disease.^62,169,170

To address the criticisms of the PORT Meta-Analysis and the conflicting more modern literature, the Lung ART study was initiated to explore the impact of modern PORT specifically in patients with completely resected N2 NSCLC receiving standard-of-care neoadjuvant or adjuvant cytotoxic chemotherapy and was recently reported at ESMO 2020.^171(abstr) A total of 501 patients treated primarily in France (85%) and the United Kingdom (10%) were randomly assigned 1:1 to PORT (54 Gy over 5 weeks) or no PORT, stratified by center, administration of chemotherapy, histology, extent of mediastinal nodal involvement, and use of PET-CT. R0 resection and absence of nodal extracapsular extension were required for eligibility. Central review was used for radiotherapy, and most patients (89%) received three-dimensional conformal radiation therapy (3DCRT; v 11% intensity-modulated radiation therapy [IMRT]). The primary end point was DFS. PORT was found to reduce mediastinal relapse (46.1% v 25%), but not DFS (47.1% v 43.8%, P = .16), driven in large part by an increase in death (14.6% v 5.3%). Further investigation into causes of death revealed numerical reductions in progression or recurrence (69.4% v 86.1%) with PORT, at the cost of increased cardiopulmonary toxicity (16.2% v 2.0%). OS was likewise not improved with PORT (66.5% v 68.5%, P = not significant).

There is much yet to learn from the Lung ART trial in terms of which subgroups, if any (such as high % lymph node–positive, high number of positive nodal stations, close margins, etc), might benefit from PORT and whether additional technological innovations such as IMRT and/or proton therapy could improve the therapeutic ratio in light of the clear reductions in relapse with PORT. Similarly, since patients with positive margins and extracapsular extension were excluded, findings from this study do not directly apply to these patients, and it is reasonable to consider PORT in these cases. However, given the overall lack of clear benefit from older trials, coupled with new data from Lung ART, there are currently no data to support the routine use of PORT for patients with completely resected N2 who have received neoadjuvant and/or adjuvant platinum-based chemotherapy.

There have been prospective randomized trials comparing concurrent chemotherapy and radiation with sequential chemotherapy and radiation (ie, chemotherapy before or after radiation). In a SR of 2,043 patients from 11 randomized control trials (six phase III and five phase II) spanning from 1992 to 2005, concurrent chemoradiation resulted in a statistically significant increase in median survival time, response rate, and tumor relapse control.¹⁵ Of the 10 trials included in the survival analysis, three individual trials showed a statistically significantly improved survival with concurrent chemoradiation, whereas the majority of the remaining trials showed a trend favoring the concurrent strategy. With a median follow-up of 3.3 years, the pooled analysis demonstrated a median survival of 16.3 months with concurrent chemoradiation versus 13.9 months with sequential chemoradiation (median ratio 1.17; 95% CI, 1.09 to 1.26), which remained significant when excluding trials that did not have survival as the primary end point. The greatest impact of the concurrent strategy was in decreasing locoregional relapses (odds ratio, 0.68; 95% CI, 0.52 to 0.87). Although the details of the platinum regimens and toxicity data were reported variably among the trials, an increase in toxicity with the concurrent strategy was noted, including both hematologic such as neutropenia and thrombocytopenia and nonhematologic such as nausea and/or vomiting, stomatitis, and esophagitis. In an individual-patient MA, with a subset of trials from the SR examined, six trials (1,205 patients) with a median follow-up of 6 years noted significant survival benefit of concurrent chemoradiation over sequential chemoradiation (HR, 0.84; 95% CI, 0.74 to 0.95; P = .004).¹⁸ This translated into absolute survival benefits of 5.7% at 3 years and 4.5% at 5 years. The greatest impact of chemoradiation was again observed in decreasing locoregional progression (HR, 0.77; 95% CI, 0.62 to 0.95; P = .01). In this MA, there was no difference in results whether sequential chemotherapy was given as induction or consolidation although only one of the six trials used a consolidation approach. There was also an increased relative risk of grade 3-4 esophageal toxicity noted. In a pooled study of five completed RTOG trials (461 patients), combined sequential and concurrent chemoradiation resulted in similar severe acute nonhematologic toxicities as the sequential approach; however, there were more severe late nonhematologic toxicities (26% v 14%; P = .046) observed, including severe late lung toxicity.²⁸ One of the largest trials included in the aforementioned SR and individual patient MA was RTOG 9410, a phase III study in 610 patients with stage II-IIIB NSCLC randomly assigned to three arms: sequential chemoradiation, concurrent chemoradiation with standard fractionation, and concurrent chemoradiation with hyperfractionation. The median survival and 5-year survival rate were higher in patients treated with concurrent chemoradiation with standard fractionation compared with sequential treatment (concurrent median 17.0 months, 5-year OS 16% and sequential median 14.6 months, 5-year OS 10%; P = .046).⁶⁹ Patients with a good performance status should receive concurrent chemoradiation over sequential chemoradiation,⁷⁶ especially as the previously conducted trials demonstrating increased toxicity compared with the concurrent approach that used older radiation techniques such as two-dimensional and 3D conformal and/or more antiquated chemotherapy regimens. However, before initiation of concurrent therapy, multidisplinary tumor board review is recommended to assess patient fitness and risk of morbidities with concurrent as opposed to sequential therapy.

There has been much discussion regarding the most effective chemotherapy regimen among patients with locally advanced NSCLC who receive concurrent chemoradiation. The most widely accepted regimen and hence most studied involves platinum doublet.^16,73,74,172 Two randomized phase III controlled trials have evaluated patients on platinum doublets.^73,74 In the first, 200 patients were allocated to etoposide 50 mg/m² once daily on days 1-5 with cisplatin 50 mg/m² once daily on days 1 and 8 (EP) every 4 weeks for two cycles concurrent with radiation v paclitaxel 45 mg/m² and carboplatin (area under curve [AUC] 2; paclitaxel and carboplatin [PC]) once daily on day 1 weekly concurrent with radiation. Consolidation chemotherapy was given to 50.5% of patients in the EP arm and 35.4% in the PC arm (P = .035), and the most common regimen was PC. The median survival times were 23.3 months in the EP arm v 20.7 months in the PC arm (log-rank test P = .095; HR, 0.76; 95% CI, 0.55 to 1.05). Chemotherapy interruptions were more common in the PC arm compared with the EP arm during concurrent treatment (37.5% v 13.6%; P < .001). Two- and five-year OS rates in the EP arm were 48.4% and 28% and were 42.7% and 19.7% in the PC arm, respectively. A significant factor contributing to the improved PFS and OS in the EP arm after longer follow-up appeared to be a higher rate of completion of treatment compared with the PC arm.⁷⁴ This represents the largest phase III study to date to compare the two most historically common platinum regimens with the longest median follow-up of patients. The second notable phase III trial was the PROCLAIM study of 598 patients with nonsquamous NSCLC enrolled between 2008 and 2012 from 125 institutions in 21 countries comparing pemetrexed-cisplatin with etoposide-cisplatin. Patients received pemetrexed 500 mg/m² intravenously (IV) plus cisplatin 75 mg/m² IV once every 3 weeks for three cycles with concurrent radiation followed by consolidation pemetrexed 500 mg/m² IV once every 3 weeks for four cycles vs etoposide 50 mg/m² once daily on days 1-5 with cisplatin 50 mg/m² once daily on days 1 and 8 (EP) once every 4 weeks for two cycles concurrent with radiation followed by consolidation chemotherapy of two cycles of platinum doublet (EP, vinorelbine plus cisplatin, and PC). Median survival was not significantly longer for pemetrexed-cisplatin versus etoposide-cisplatin (median OS, 26.8 v 25 months), and 1-, 2-, and 3-year survival rates did not differ in each arm (76% v 77%, 52% v 52%, and 40% v 37%, respectively). Although PROCLAIM was not designed as a noninferiority trial, given numerically similar outcomes, pemetrexed-cisplatin has been adopted as an option for the treatment of patients with nonsquamous NSCLC.⁷⁴

In a SR and network MA in 2019, 14 randomized controlled trials (12 two-arm studies and two three-arm studies) published before 2018 reviewed 12 categories of platinum doublets (EP, PC, pemetrexed plus cisplatin or carboplatin, S-1 plus cisplatin, uracil plus tegafur-cisplatin, vinorelbine plus cisplatin, gemcitabine plus cisplatin, docetaxel plus cisplatin, irinotecan plus carboplatin, mitomycin plus vindesine plus cisplatin, PC plus cetuximab, and pemetrexed plus cisplatin or carboplatin plus cetuximab) among 2,975 patients with the primary outcome of OS. Both the aforementioned phase three trials were included. EP and PC were the only regimens that could be compared directly favoring EP in OS. Other comparisons did not offer statistically significant differences regarding survival.¹⁶ Although S1-cisplatin showed tolerability and prolonged survival, to date, it is not approved by the FDA. Attention to both tolerability and OS is key to treatment selection and is of significant importance among older patients.¹⁷³

There have been randomized control trials examining concomitant chemotherapy and radiation versus radiation alone in patients with locally advanced NSCLC.¹⁹ In a recent MA and SR, which included a total of 13 randomized control trials (1,936 patients), patients receiving some form of chemotherapy with radiation had a higher OS (pooled HR, 0.72; 95% CI, 0.62 to 0.84; P < .001) and a higher PFS (pooled HR, 0.73; 95% CI, 0.60 to 0.89; P = .002) compared with patients receiving radiation alone. This MA and SR examined heterogeneous treatment regimens. However, there are randomized trials that specifically examined sequential chemotherapy followed by radiation versus radiation alone. Dasgupta et al⁷⁰ reported on 103 patients randomly assigned to three treatments, including radiation alone (65 Gy in 30 fractions) versus neoadjuvant cisplatin-etoposide chemotherapy for three cycles followed by radiation (60 Gy in 30 fractions) and three more cycles of chemotherapy. In the chemotherapy-containing radiation arm, there were a statistically significant improvement in survival and a decrease in distant metastases (48.6% v 62.5%). Kim et al⁸⁵ reported a phase III randomized trial examining induction platinum-based chemotherapy for three cycles followed by radiation versus radiation alone (60-65 Gy in 1.8-2 Gy fractions) in 101 patients. A trend for improved OS with the incorporation of chemotherapy was observed (median 13.8 v 8.5 months, respectively), with significance found in a subgroup of patients with nonsquamous NSCLC and stage IIIB disease. When only examining this subgroup (stage IIIB nonsquamous), the median survival was 11.6 months with sequential chemoradiotherapy versus 8 months with radiotherapy alone (P = .046). In a study by Komaki et al,⁸² 490 patients with NSCLC were randomly assigned to three arms: radiation alone (60 Gy in 30 fractions), weekly cisplatin-vinblastine chemotherapy followed by radiation (60 Gy in 30 fractions), and hyperfractionated radiation. There were significant differences in median OS observed, with longest survival noted in the sequential chemoradiation arm: 11.4, 13.6, and 12.3 months, respectively. Despite no differences in local tumor control rates, there were fewer distant metastases observed in the sequential chemoradiation arm. In the CALGB8433 trial, 78 patients were assigned to cisplatin-vinblastine weekly for 5 weeks followed by radiation on day 50 (60 Gy in 30 fractions) versus radiation alone (60 Gy over 6-7 weeks).⁸⁰ After a 7-year follow-up, the median survival was greater in the sequential chemoradiation arm compared with the radiation-alone arm (13.7 months v 9.6 months, P = .012; 3-year OS, 24% v 10%, respectively). Finally, there were 66 patients randomly assigned to cisplatin-based chemotherapy for three cycles followed by radiotherapy versus radiotherapy alone, with a higher response rate (53% v 32%, respectively) and a trend for higher median survival noted (52 weeks v 30 weeks, respectively).⁹⁵ Because of the improved survival observed in prospective randomized studies of sequential chemoradiation versus radiation alone, patients not fit to undergo concurrent chemotherapy radiation with locally advanced NSCLC should be offered a sequential chemoradiation approach.

In one of the earliest cooperative group radiation dose-escalation randomized trials, RTOG 7301 established 60 Gy as the early standard-of-care dose for radiotherapy in patients with locally advanced NSCLC by showing that 60 Gy led to longer 3-year OS rates than 50 Gy, 40 Gy, or 40 Gy delivered as a split course with a 2-week break after 20 Gy (15% v 10% vs. 6% v 6%) and fewer infield recurrences (35% v 49% v 58% v 53%).^174,175 The delivery of 60 Gy in 30 fractions was subsequently similarly found to improve outcomes of hyperfractionated radiotherapy in a phase III North Central Cancer Treatment Group and Mayo Clinic trial.¹⁰² On the basis of these early data from the RTOG 7301 study, the investigators concluded that higher doses of radiation are necessary to improve intrathoracic tumor control.¹⁷⁴

Given the high rates of local and regional failures after concurrent chemotherapy when radiation therapy is delivered to 60 Gy in 30 fractions,¹⁰⁰ as well as the known association between local control and OS in patients with locally advanced NSCLC,^18,159 many groups have investigated further radiation dose escalation beyond the 60 Gy RTOG 7301 standard. Such dose-escalation strategies have been trialed in many forms, such as increasing the absolute dose of radiotherapy while maintaining 2 Gy daily fractions,¹⁰⁰ increasing the absolute dose of radiotherapy with mild hypofractionation (increasing the dose per fraction),¹⁷⁶ delivering higher total doses of radiotherapy by delivering boosts to a smaller target volume,^104,177 maintaining the same absolute radiotherapy dose while increasing the biologically effective dose through more appreciable hypofractionation (larger doses per fraction),^127,178 delivering an isotoxic integrated boost (increased dose per fraction to a limited portion of the target volume, typically gross tumor, limited by predicted risk of toxicity),¹⁷⁹ boosting high-uptake areas of PET avidity within the primary tumor,¹⁸⁰ or increasing the absolute dose using hyperfractionated (small fractions delivered more than once daily) radiotherapy.¹⁰⁸

Although many of these earlier phase trials showed promising results, a definitive phase III trial of dose escalation was needed to assess if the standard-of-care dose should change from 60 Gy in 2 Gy fractions. RTOG 0617 was a randomized phase III trial with a two-by-two factorial design testing both the addition of cetuximab to concurrent chemoradiation and the use of 74 Gy compared with 60 Gy thoracic radiation in 2 Gy daily fractions. Concurrent chemotherapy was paclitaxel 45 mg/m² and carboplatin AUC 2 once a week, followed by two cycles of consolidation chemotherapy with paclitaxel 200 mg/m² and carboplatin AUC 6 once a week. Radiation therapy was delivered with either IMRT or 3DCRT. The median OS was 28.7 months for patients treated to 60 Gy versus 20.3 months for patients treated to 74 Gy (P = .0072). Although there were no statistical differences in all grade ≥ 3 toxic effects between radiotherapy arms, grade ≥ 3 dysphagia (12.1% v 3.2%; P = .0005) and esophagitis (17.4% v 5.0%, P < .0001) occurred more commonly among patients receiving 74 Gy, and there were more treatment-related deaths in the 74-Gy arm (9 v 3).^100,181

Despite expectations for an improvement in OS with dose escalation above 60 Gy, RTOG 0617 showed that 74 Gy radiation given in 2 Gy fractions with concurrent chemotherapy was not better than 60 Gy, was associated with inferior outcomes, and was potentially harmful.^100,181 Patients receiving 74 Gy were also found to have a greater decline in quality-of-life scores at 3 months (45% v 30%; P = .02).¹⁸² A long-term report of this trial showed that 5-year OS (32.1% v 23.0%; P = .007) and PFS (18.3% v 13.0%; P = .055) were better with the standard dose arm, and this further established 60 Gy as the standard-of-care radiation dose to be delivered with concurrent chemotherapy, with the OS rate being among the highest reported in the literature for stage III NSCLC.¹⁸¹

Although RTOG 0617 used 60 Gy as the standard dose arm that was compared with the dose escalation of 74 Gy, numerous previous cooperative group and other large studies have used radiation doses of > 60 Gy as their standard dose arm. As an example, 64 Gy in 2 Gy daily fractions was used in the control arm of the phase III ECOG 2597 trial assessing induction chemotherapy followed by standard versus hyperfractionated radiotherapy.¹⁰¹ Recent SRs and MAs of randomized trials assessing radiation doses have been reported and demonstrate that common doses delivered for locally advanced NSCLC include 64 Gy and 66 Gy in 2 Gy fractions, with 1.8 Gy daily fractions to total 64.8 Gy or 66.6 Gy also used.^23,24 Furthermore, the ongoing phase III NRG Oncology RTOG 1308 trial uses 70 Gy in 2 Gy daily fractions as its standard dose.¹⁸³

Comparisons of 60 Gy in 2 Gy fractions with doses above 60 Gy but < 74 Gy specifically delivered in once-daily radiotherapy using 1.8-2.0 Gy fractions are lacking to date. With well-established once-daily dose fractionation regimens from 60 to 70 Gy used in previous cooperative group studies and no direct comparisons between such doses, it remains to be defined what is the optimal radiation total dose. As such, higher doses above 60 Gy and up to 70 Gy can be considered for well-selected patients. Results of RTOG 1308, in which patients are being treated to 70 Gy with more strict normal tissue dose constraints than have previously been used in cooperative group studies, may in the future question the choice of 60 Gy as the standard dose of radiotherapy delivered concurrent with chemotherapy for locally advanced NSCLC. Doses above 70 Gy, however, should not be delivered concurrently with chemotherapy outside the context of a clinical trial.

In cases where doses above 60 Gy are being delivered, careful attention to irradiation doses to the heart, lungs, and esophagus is required to avoid excessive morbidity and even mortality from treatment. Dose volume parameters that should be carefully assessed in patients considered for doses up to 70 Gy include the mean lung dose and lung V20, cardiac mean and volumetric dose, esophageal mean dose, and spinal cord maximum dose, among others. Advanced radiation modalities may better spare these critical organs at risk and help to facilitate dose escalation more safely. In RTOG 0617, although patients treated with IMRT had more advanced stage grouping (P = .056) and larger planning target volumes (P = .005) than those treated with 3DCRT, IMRT was associated with lower lung irradiation doses (P = .08), lower heart doses (P < .05), fewer grade ≥ 3 pneumonitis events (P = .0653), and higher compliance with full-dose consolidative chemotherapy, with no difference in OS despite larger and less favorable tumors treated with IMRT.¹⁸⁴ The use of IMRT in population-based analyses has been reported to be associated with a survival benefit relative to 3DCRT in locally advanced NSCLC.¹⁸⁵ Similarly, proton therapy in a population-based analysis has been reported to be associated with a survival benefit relative to photon radiotherapy in locally advanced NSCLC¹⁸⁶ and may improve outcomes by more safely allowing for dose escalation, reducing toxicities, and reducing radiation-induced lymphopenia.¹⁸⁷

In less common situations where patients have locally advanced NSCLC and either are not candidates for (eg, comorbidities or pulmonary function) or refuse chemotherapy, radiation alone is typically used. This recommendation pertains to the radiation dose and fractionation that is used in these circumstances. There are prospective studies to support radiation dose and fractionation in this setting. Some are older and were largely carried out before the refinement of concurrent chemoradiation therapy, 3DCRT, IMRT, and/or proton therapy techniques. More recently, a phase III randomized trial was published that addresses this specific question.

Older trials have also evaluated dose-escalated approaches when radiation is delivered as monotherapy. The most notable of the older trials was termed CHART (Continuous Hyperfractionated Accelerated Radiation Therapy). This phase III trial published by Saunders et al¹⁸⁸ in 1997 enrolled 563 patients across 13 centers in a 3:2 ratio to either CHART (1.5 Gy three times a day for 12 consecutive days) or 60 Gy in 30 fractions of 2 Gy each excluding weekends. Two-year OS favored CHART (29% v 20%; P = .004). A subsequent trial modified the CHART regimen to CHARTWEL (CHART WeekEnd Less) in an effort to make the regimen more widely accepted.^189,190 CHARTWEL randomly assigned 406 patients to either 60 Gy in 40 fractions over 2.5 weeks or 66 Gy in 33 fractions over 6.5 weeks. There were no differences in 2-, 3-, or 5-year OSs between regimens (31%, 22%, and 11% v 32%, 18%, and 7%, respectively; HR, 0.92; P = .43).

The first study to use 3DCRT was RTOG 9311, a phase I/II dose-escalation study, either alone or after two cycles of systemic chemotherapy.¹⁷⁶ Doses were 2.15 Gy/fraction to 70.9 Gy, 77.4 Gy, 83.8 Gy, or 90.3 Gy. The safe total radiation dose was determined to be 83.8 Gy in patients with a lung V20 < 25% and 77.4 Gy for lung V20 of 25%-36%.

None of these regimens are routinely used today, as CHART and CHARTWEL required three times daily radiation therapy, and the RTOG 9311 trial safe dose of 83.8 Gy was administered over 39 total fractions. Most recent studies have investigated accelerating treatment with hypofractionation with larger radiation doses delivered once per day. Concerns about serious toxicity have been better addressed with the emergence of IMRT techniques and greater sparing of normal tissues. The preference has become hypofractionated radiation therapy using fraction sizes of 2.5-4 Gy. The studies supporting this preference consist of phase I prospective trials testing the use of more sophisticated techniques such as simultaneous integrated boosts where the gross tumor volume receives a higher daily fraction than the surrounding microscopic tumor volume and/or lymph nodes. Examples of these regimens include 60 Gy in 30 fractions to the larger planning target volume and administering up to 78 Gy simultaneously to the gross tumor volume.¹⁹¹

Gomez et al¹⁹² published a prospective phase I dose-escalation trial using proton beam that escalated radiation dose in a 15-fraction course using 45 Gy (relative biologic effectiveness [RBE]), 52.5 Gy (RBE), and 60 Gy (RBE) in a 3 + 3 study design. With a median follow-up of 13 months (range 8-28 months), two of the 25 patients experienced dose-limiting toxicities. A similar, more recent multicenter prospective trial demonstrated that radiation doses in 15-24 fractions can be safely delivered with proton therapy concurrently with chemotherapy.¹⁷⁸ The authors of these trials determined that hypofractionation was well-tolerated and encouraged phase II and III trials. As protons are not available at most centers, a phase I dose-escalation study was later reported using photons at the University of Texas Southwestern in patients who were not chemotherapy candidates. The trial enrolled 55 patients among three doses, 50, 55, and 60 Gy, all delivered in 15 fractions. One patient developed grade 3 esophagitis, and two cases of grade 3 dyspnea were felt related to therapy. There was no association between fraction size and toxicity (P = .24), and the median OS was 6 months at all dose levels (P = .59).¹⁹³ The same investigator group subsequently conducted a randomized, multi-institution phase III comparison of 60 Gy in 15 versus 30 fractions of image-guided photon radiation therapy in patients ineligible for concurrent chemoradiation or with a Zubrod performance status of 2 or greater.¹²⁷ The primary end point was 1-year OS. The trial was closed to accrual when an interim analysis suggested futility at meeting the primary end point. With 103 randomly assigned patients, there was no significant difference in 1-year OS between groups (37.7% for the hypofractionated regimen and 44.6% for the conventional regimen [P = .29]), and no differences in grade 3 or higher adverse events were noted. Although the trial was not powered to test equivalence, these results do reassure that a regimen of 60 Gy in 15 fractions of radiation alone may yield similar results without additional toxicity, and the authors note that the convenience of the shorter treatment regimen may warrant its consideration for well-selected patients.¹²⁷

The phase III randomized PACIFIC trial evaluated consolidation durvalumab for 1 year versus placebo in 713 patients with unresectable stage III NSCLC after definitive platinum-based chemoradiotherapy with at least two cycles.¹²² Patients were enrolled irrespective of the PD-L1 status of their tumors, had no progression after definitive chemoradiotherapy, and had a good performance status (Eastern Cooperative Oncology Group 0 or 1), and were randomly assigned within 42 days of completing thoracic radiation. The primary end points of the trial included PFS and OS. The median PFS was 16.8 months (95% CI, 13.0 to 18.1) with durvalumab versus 5.6 months (95% CI, 4.6 to 7.8) with placebo (HR, 0.52; 95% CI, 0.42 to 0.65; P < .001). Grade 3 or 4 pneumonitis occurred in only 4.4% in the durvalumab arm and was the main higher-grade toxicity. An updated analysis with a median follow-up of 34.2 months continued to show a significant benefit in OS (HR, 0.72; 95% CI, 0.59 to 0.89) and PFS (HR, 0.55; 95% CI, 0.45 to 0.68). The median OS with durvalumab was 47.5 months versus 29.1 months in the placebo arm. The estimated 5-year OS rates were 42.9% versus 33.4% in favor of durvalumab, and 5-year PFS rates were 33.1% versus 19.0%, respectively.^126,194 Therefore, this trial provides level 1 evidence demonstrating a survival benefit from the administration of consolidation durvalumab. A post hoc analysis of the limited number of patients with available PD-L1 status (63% of the trial population) suggested a lack of benefit of durvalumab in PD-L1–negative tumors (20% of trial population; HR, 1.05; 95% CI, 0.69 to 1.62) and EGFR-mutated patients (6% of the study population, n = 43, HR, 0.97; 95% CI, 0.40 to 2.33). This analysis was performed in a small proportion of the study population and needs further evaluation before definitive conclusions can be made although certain authorities outside of the United States have restricted consolidation durvalumab to PD-L1–positive patients only.

Figures 3 and 4 provide visual interpretations of these recommendations in the management algorithm.

FIG 3. Algorithm for evaluation and staging in stage III NSCLC. CT, computed tomography; FDG, fluorodeoxyglucose; NSCLC, non–small-cell lung cancer; PET, positron emission tomography.

FIG 4. Management of stage III NSCLC algorithm. Arrows with dotted lines indicate that the level of obligation is a May (moderate) recommendation. ^aPatients with stage III NSCLC generally should not be excluded from consideration for surgery by nonsurgical physicians. ^bCarboplatin may be substituted for cisplatin in patients with contraindications to or deemed ineligible for cisplatin. CT, computed tomography; MRI, magnetic resonance imaging; NSCLC, non–small-cell lung cancer; PET, positron emission tomography; PS, performance status; RT, radiation therapy.