Introduction

As the twentieth century draws to a close, the epidemic of lung cancer continues its lethal course throughout the world, and will lead to death in 10% of all persons now living. In the United States, about 180,000 new cases of lung cancer were expected in 1996, with an estimated overall mortality rate of 85%.[1] Roughly 75% of new cases will be non-small-cell lung cancer (NSCLC). While in past years lung cancer was predominantly a disease afflicting men, with squamous cell the most common histology, recent years have seen a trend toward equality in the numbers of cases in men and women and a shift to adenocarcinoma as the most common histology.

During the 1970s, standard therapy for patients with "locally advanced" NSCLC was treatment with radiation therapy alone, generally in doses ranging from 50 to 60 Gy over 5 or 6 weeks. This treatment gave reasonable palliation, with about 60% to 80% of patients having marked symptomatic improvement in cough and hemoptysis. A few patients were cured, however, and overall 5-year survival averaged about 5%. Five-year survival rose to 10% in carefully selected patients who had excellent performance status and few, if any, symptoms, (ie, those whose tumors were discovered incidentally, perhaps on a chest x-ray before elective surgery). At the time, local disease control was considered reasonably good, in the range of 60% at these doses. Many of these reports cited freedom from local progression, however, not durable local control.[2] With the short survival of many patients and the difficulty of distinguishing between local tumor progression and postradiation pulmonary fibrosis, such estimates appear to have been significantly in error. Local control as assessed by bronchoscopic examination and biopsy following radiation therapy is more in the range of 20% with these or similar dose/fractionation regimens.[3]

Spurred by the high incidence of lung cancer and its great lethality, clinical investigators in single institutions and cooperative research groups have made great efforts over the past two decades to improve these dismal figures, and we have seen modest but real progress. This review will consider three areas of development: improved imaging and stage classification of patients with so-called locally advanced non-small-cell lung cancer, improvements in radiation therapy dose delivery (physics) and fractionation (radiation biology), and integration with many different rationales of radiation and cytotoxic chemotherapy.

Patient Staging and Selection for Clinical Trials

Patients with stage III NSCLC are heterogeneous in overall clinical status and stage. Performance status and weight loss are highly significant prognostic factors both for the ability to tolerate treatment and for survival.[4] The otherwise healthy patient with an incidental diagnosis of stage IIIA disease based on a chest x-ray done before elective surgery will likely fare far better than the symptomatic patient who has significant weight loss.

The importance of substaging patients with stage III NSCLC has been increasingly recognized in recent years. Until the New International Staging Classification was adopted in 1986, stage III included patients with locally advanced primary and/or nodal disease and those with extrathoracic metastases. The 1986 system established stage IV for patients with extrathoracic visceral disease, and subdivided stage III. Stage IIIA includes patients with potentially resectable disease, and IIIB encompasses those with direct mediastinal invasion, and/or malignant pleural effusions, and/or contralateral mediastinal nodal metastases, and/or supraclavicular metastasis, who are clearly not surgical candidates.[5]

While this distinction is valid when comparing IIIA patients treated surgically with IIIB patients treated nonsurgically, how the prognosis of these groups is affected following either radiation alone or radiation combined with chemotherapy is controversial.[6-8] Disease bulk is generally an important determinant of the ability of radiation to achieve local control. Omitting bulk from the staging system likely obscures differences among patients treated nonsurgically. While T4 disease due to invasion of unresectable mediastinal structures and T4 disease due to the presence of a malignant pleural effusion preclude curative surgery, they are not equivalent when definitive radiation therapy is being considered. Radiation may be appropriate for mediastinal but not for pleural T4 disease. Data also suggest that the number of involved mediastinal nodal sites and/or the total number of involved nodes is prognostically important at least for surgically-treated patients.[9]

The greatest heterogeneity within stage IIIA concerns patients with T3,N0 and T3,N1 disease. Patients with T3,N0 disease, treated surgically, can have 5-year survivals in the range of 30% to 50%, and those with T3,N1 disease will have a 5-year survival of about 25% compared with a 5-year survival of 15% or less for patients with T3,N2 disease. Current recommendations to revise the staging classification take into account some of this heterogeneity.[10] It may be, however, that surgically- and nonsurgically-treated patients require somewhat different staging systems.

The techniques used to evaluate mediastinal nodal involvement are important in determining prognosis, particularly the percentage of long-term survivors. Computed tomography (CT) has been widely used in the past decade and is generally equal or superior to magnetic resonance imaging for assessing mediastinal node enlargement. Nonetheless, it is essential to recognize that nodal enlargement is not synonymous with nodal involvement by tumor and that a cutoff of 1.5 cm will result in about a 20% false-positive and false-negative CT scan rate.[11] Of concern is a series of clinically-staged patients that show a tail on the survival curve suggesting that many of these long-term survivors may have been N2 radiographically but N1 or N0 histologically. Since the survival of patients with T2,N0 disease is significantly better when they are treated with resection as opposed to nonsurgical therapy, patients who are otherwise good candidates for resection deserve histologic confirmation of suspected N2 disease. Reports of clinical trials of N2 patients should also indicate the proportion with bulky disease (visible on chest x-ray), CT detectable disease, or only mediastinoscopically detectable disease. Table 1 gives a suggested revision of the staging system that considers some of these factors and suggests treatments by substage.

Because of the wide range of prognoses for patients with stage IIIA and IIIB NSCLC, determining at the outset whether an individual should be treated aggressively with some expectation of cure or whether palliation of symptoms with minimal treatment duration and toxicity is more appropriate is essential. Such initial triage must consider stage-based prognosis, co-morbid conditions, and the preferences of patients, which may well differ from those of their physicians. These considerations will lead to treatment that is more appropriate for the individual patient (and his or her family) and to more rational use of costly medical resources as well.

Historical Results of "Standard" Radiation Therapy

In the 1970s the Radiation Therapy Oncology Group (RTOG) conducted a series of trials that attempted to define the appropriate radiation dose, its fractionation, and volume parameters. While these trials were well conceived and implemented for their time, they were conducted without technology now considered standard (like CT-based treatment planning), routine use of custom-shaped field blocking, and selection of patients with favorable prognostic features (good performance status and minimal weight loss.) Many of the early RTOG trials also allowed the use of posterior spinal cord shields, which resulted in underdosing of the anterior mediastinal nodes.

RTOG 73-01 randomized 376 patients among 4 fractionation schema, 40 Gy given in 2 sessions of 20 Gy in 5 fractions separated by a 2-week break, and continuous course treatment (2 Gy per fraction, 5 days a week to total doses of 40, 50, or 60 Gy).[2] Survival at 2 years—but not at 5 years—was better for the higher-dose radiation groups. Freedom from local progression at 3 years was reported as 67%, 58%, and 48% for patients receiving 60 Gy, 50 Gy, and 40 Gy, respectively, in the continuous radiotherapy groups and as 49% for patients in the 40-Gy split-course radiation group. Unfortunately, freedom from progression is not synonymous with local control, and when disease relapses systemically, patients may not be assessed carefully for local control. Thus, these figures for absence of local progression greatly overestimated the actual probability of local control.

Curran et al looked critically at the relapse patterns of patients with stage IIIA and IIIB non-small-cell lung cancer who had been treated for cure with radiation therapy as a single modality during the 1980s at the Fox Chase Cancer Center (Philadelphia).[6] While distant failure is often considered the most common mode of failure in such patients, these data show that local and distant failure occur equally (Figure 1). Local control (or, rather, failure to observe local progression) was achieved in, at best, 50% of patients, with no differences in either survival or patterns of failure for patients with stage IIIA and IIIB disease.

A more critical assessment of local tumor control was provided by Le Chevalier et al, who randomized patients either between radiotherapy alone or preceded and followed by chemotherapy and assessed all patients bronchoscopically.[12] Rates of tumor clearance were poor—17% for the combined modality arm and 15% for the radiotherapy arm (65 Gy). Radiation therapy, delivered in daily fractions of 1.8 to 2.5 Gy to total doses of 60 to 65 Gy over 6 or 7 weeks, is clearly not adequate to provide good local control for patients with stage IIIA/B non-small-cell lung cancer. Considering that such doses are rarely able to control bulky tumors at other sites (eg, head and neck, cervix, breast), this should come as no great surprise.

The poor results with conventional radiation therapy in patients with non-small-cell lung cancer have led to several new approaches for improving local control. These fall broadly into two categories—those that seek to better define and target volume to allow higher radiation doses to the tumor while sparing normal tissues and those that alter radiation fractionation to exploit biologic differences between tumor and normal tissues.