Lung cancer is the leading cause of cancer deaths in the United States and non-small cell lung cancer (NSCLC) accounts for approximately 80% to 85% of all cases.¹ The 1997 revision of the lung cancer staging system² classifies stage IIIA as T1-3N2 and T3N1 disease, whereas stage IIIB includes either N3 or T4 disease. In the Surveillance, Epidemiology, and End Results (SEER) 2004 registry, 29.7% of new NSCLC cases presented with stage III disease, of which 12.1% were stage IIIA and 17.6% were stage IIIB.³ The International Association for the Study of Lung Cancer (IASLC) Lung Cancer Staging Project database (see Chapter 30) of 67,725 cases of NSCLC revealed 5-year overall survival figures for clinically staged IIIA and IIIB disease of only 18% and 8%, respectively, and for pathologically staged IIIA and IIIB disease, it was 25% and 19%, respectively.⁴ The poor outcomes observed are caused by locoregional failure rates between 30% and 55%, with distant failure rates in the range of 40% to 60%.⁵

Staging Subsets of Stage III NSCLC Strategies to improve outcomes have centered on improved staging and tailoring aggressive treatment regimens to subcategories of stage III disease that are most likely to benefit. This population is heterogeneous in terms of presentation and outcomes—an issue that is being addressed in the forthcoming seventh edition of the TNM stage groupings.⁴ This chapter will not address the treatment of three categories of patients with stage III disease, namely, (a) stage T3N1 that are generally resectable, (b) tumors of the superior sulcus where the addition of surgery is considered useful,⁶,⁷ and (c) stage IIIB patients with malignant pleural effusions.

Improvements in staging modalities, such as positron emission tomography with ¹⁸F-fluorodeoxyglucose (FDG-PET), multislice computed tomography (CT), and endoscopic techniques, have had a profound effect on the pretreatment evaluation of NSCLC. The routine use of PET is now recommended by the American College of Chest Physicians (ACCP) as a standard staging investigation for stages IB to IIIB prior to a curative treatment⁸ (see Chapter 27). The false-positive and false-negative rates with CT scans alone are unacceptably high.⁹,¹⁰,¹¹ FDG-PET has a higher sensitivity of 84% and specificity of 89%, and the sensitivity improves to 93% and specificity to 95% when using combined PET-CT.¹²,¹³,¹⁴ As false-positive rates range from 10% to 15%, it has been recommended that any positive PET findings should be confirmed by cytopathology in patients who are candidates for surgery.¹⁴,¹⁵,¹⁶ However, the same approach can be taken for patients who are candidates for high-dose concurrent chemoradiotherapy (CRT).¹⁷ Up to 30% of conventionally staged patients with stage III cancer referred for radical radiotherapy are excluded after a PET scan, mainly because of the detection of occult metastatic disease.¹⁸ At least part of the incremental survival benefit seen in stage III NSCLC over the last decade may be a result of the migration of patients with otherwise occult metastases to stage IV. Although mediastinoscopy is still considered, the gold standard for confirming N2 disease,¹⁹ only a minority of patients treated in nonsurgical trials have undergone this invasive procedure. In recent years, minimally invasive staging by techniques such as biopsy during endoscopic ultrasound (EUSFNA) and transbronchial needle aspiration without (TBNA) or with endobronchial ultrasound guidance (EBUS-TBNA) are increasingly used to minimize invasive staging procedures²⁰,²¹,²² (see Chapters 28 and 29). Radiation oncologists with access to endoscopic staging can use this approach to optimally define nodal target volumes in patients with stage III disease.

Factoring in Tumor and Patient Characteristics An appropriate treatment strategy for stage III NSCLC can only be formulated once the extent of disease, any comorbidity, and general fitness of a patient has been fully characterized. The ACCP guidelines have chosen to classify N2 tumors into
four subsets for the purpose of generating rational treatment guidelines.²³ Nodal metastases found either after or during a surgical resection are denoted as IIIA₁ and IIIA₂, respectively. Nodal disease identified preoperatively is IIIA₃, whereas bulky or fixed multistation N2 disease is classified as IIIA₄. Although there is no uniform definition of what constitutes “bulky” nodal disease, the ACCP recommendation includes nodes of 2 cm or larger in short-axis diameter as measured by chest CT, multistation nodal disease, and/or groupings of multiple positive smaller lymph nodes.²³

Patients with lung cancer generally tend to have significant smoking-related comorbidities, suboptimal pulmonary function and impaired performance score. Each of these factors may impact on their ability to tolerate potentially curative treatment, and each is exacerbated with increasing age.²⁴ The median age of presentation for NSCLC cases from a SEER database analysis was 67,³ and this is expected to rise in the future.

Primary Treatment The choice of optimal local treatment has been the topic of active research in the last 2 decades. Current guidelines recommend that platinum-based combination CRT should be the primary treatment in patients with stage IIIA NSCLC and ipsilateral mediastinal disease identified preoperatively.²³ The same strategy is also recommended for patients with stage IIIB NSCLC and a good performance score with minimal weight loss (≤5%).²⁵

The role of surgery in stage III disease was evaluated in two large randomized trials, which concluded that the addition of surgery did not improve survival in comparison to either sequential chemoradiotherapy (CT-RT)⁵ or concurrent CRT²⁶ alone (see Chapter 55). Subgroup analysis has suggested that surgical outcomes were more favorable in patients with “downstaged” N2 disease, and in whom a radical resection was achieved with a lobectomy rather than a pneumonectomy.²⁶ However, conditions that favor postsurgical survival are generally established only after a resection. The majority of patients with N2 disease will still have persistent disease despite induction therapy, and “gaps” in CRT, arising during mediastinal restaging risk exposing such patients to a poorer survival caused by longer overall treatment time.²⁷

Radiotherapy alone is an inappropriate treatment for fit patients as an updated metaanalysis by the NSCLC Collaborative Group revealed a significant survival benefit for sequential CTRT versus radiotherapy alone.²⁸ When concurrent CRT was compared to radiotherapy alone, a survival advantage was seen for the former (HR = 0.88; 95% CI, 0.81 to 0.95), with an absolute benefit of 3.2% (from 13.4% to 16.6%) at 3 years. 28 Concurrent CRT also improved the progression-free survival, and there was no evidence that any patient subgroup benefited more or less from concurrent CRT.

A second metaanalysis based on 6 trials and 1199 patients evaluated overall and progression-free survival after the concurrent administration of chemotherapy and radiation versus the sequential administration of both modalities.²⁹ At a median follow-up of 5 years, a significant survival benefit was observed for concurrent CRT with an absolute benefit of 6.6% (from 18.2% with sequential CT-RT to 24.8% with concurrent CRT) at 3 years. Concurrent CRT decreased locoregional progression (HR = 0.76; 95% CI, 0.62 to 0.94) but distant progression rates were similar. Concurrent CRT increased acute grade 3 and 4 esophageal toxicity from 3% to 18%, but no significant difference in acute pulmonary toxicity between both approaches was observed.

Choice of Systemic Agents with Concurrent Radiotherapy Platinum-based concurrent CRT regimens improve outcomes in several malignancies, including lung cancer.²⁹ A scheme widely used in recent phase III trials consists of cisplatin 50 mg/m² on days 1, 8, 29, and 36, and etoposide 50 mg/m² on days 1 through 5 and 29 to 33.⁶,²⁶ Cisplatin given at full dose provides both a radiosensitization effect within the irradiated volume, and an overall survival benefit related to reduced distant metastases, as demonstrated in the lung adjuvant cisplatin evaluation (LACE) metaanalysis.³⁰ Significant pathological response rates were observed using an induction scheme of cisplatin-etoposide for stage III NSCLC with only 45 Gy of concurrent radiation in INT 0139, with an 18% pathologic complete response (pT0N0) and 46% complete nodal response (T_any N0).²⁶

Immunohistochemical analysis in resected NSCLC for excision repair cross-complementation group 1 (ERCC1) protein expression suggested that adjuvant cisplatin-based chemotherapy was only beneficial in ERCC1 negative tumors.³¹ Studies are currently underway to prospectively validate this finding in the adjuvant setting, and the likely role of ERCC1 expression in selecting chemotherapy schemes in primary CRT remains to be elucidated.

Although the use of concurrent low-dose carboplatin-paclitaxel is widespread in the United States, the outcomes of some recent trials with this combination have been disappointing.³² This finding may be explained by the fact that three randomized clinical trials of carboplatin plus radiotherapy have not been shown to be superior to radiotherapy alone.³³,³⁴,³⁵

Adjuvant Management of Completely Resected Stage III NSCLC The adjuvant management of stage III disease consists of cisplatin-based chemotherapy³⁰,³⁶,³⁷; however, locoregional failure rates of 20% to 40% persist in this setting.³⁶,³⁷ In the Adjuvant Navelbine International Trialist Association (ANITA) trial, a nonrandomized subanalysis comparing 5-year overall survival in N2 patients who did or did not receive postoperative radiotherapy (PORT) found higher survival rates in patients receiving radiotherapy in both the observation and chemotherapy arms (21% vs. 17% and 47% vs. 34%, respectively).³⁶ A retrospective SEER study also reported superior survival rates associated with PORT in N2 disease.³⁸ This has renewed interest in adjuvant radiotherapy for resected patients—a strategy that fell out of favor after the PORT metaanalysis of 2128 patients with stage I to III disease reported a significant adverse effect of PORT on survival in 1998.³⁹ This metaanalysis has been criticized as many of the included trials used radiotherapy techniques that are now considered suboptimal, leading to higher morbidity
and mortality rates than more recent studies.⁴⁰,⁴¹ A phase III European Intergroup trial (LungART), designed to address this issue, opened in 2007 and will compare three-dimensional (3D) conformal PORT to no PORT.⁴² Present guidelines do not recommend adjuvant radiotherapy for completely resected stage IIIA disease²³,⁴³ but do suggest PORT when the risk of nodal recurrence is elevated because of extranodal spread, close or microscopically positive resection margins, or involvement of multiple nodal stations.

The last 2 decades has witnessed a dramatic shift from two-dimensional (2D) radiotherapy, where bony landmarks and planar images were used to guide field setup, to four-dimensional (4D) radiotherapy, where CT-contoured target volumes are modified to account for intrafraction motion. The use of 2D radiotherapy for lung cancer carries a risk for target miss in up to 15% of patients,⁴⁸ and may partly account for the poor local control rates seen in NSCLC trials performed to date. Target volumes drawn on CT images now represent the minimum standard of care.

Defining the Gross Tumor Volume for the Primary Lung Tumor Precise definition of the gross tumor volume (GTV) is important, as contouring errors are compounded when volumetric expansions are made to account for microscopic tumor spread, motion, and setup errors. The boundary between macroscopic tumor and the lung parenchyma is best defined when viewing CT images with proper window width and level settings. Distinguishing the tumor border is more difficult when atelectasis or pulmonary effusions are present, potentially leading to an overestimation of the GTV. Reviewing diagnostic scans in which intravenous contrast has been given, or using contrast for CT simulation, may be helpful. PET scans have been used to avoid atelectatic lung in the target volume,⁴⁹ but data to validate this approach are awaited.⁵⁰

PET and PET-CT images (where functional and anatomical data are coregistered as a result of being obtained during the same procedure) may serve a far greater purpose in NSCLC delineation than crude tumor identification. Using PET for radiotherapy planning results in a reduction in interobserver contouring variability, and more consistent delineation of GTV.⁵¹ Some groups have proposed using PET-CT for generating GTVs automatically by using a standardized uptake value (SUV) threshold,⁵⁰ an approach that may be confounded by factors, including (a) heterogeneous uptake of FDG within the tumor as a result of necrosis, hypoxia, or degree of tumor differentiation, (b) elevated SUV levels as a result of inflammatory processes, and (c) low SUVs from small tumors caused by partial volume effects.⁵² It has been recommended that radiation oncologists should work closely with their nuclear medicine counterparts to interpret PET scans, and caution has been advised when using automatic delineation.⁵⁰