Introduction to Locally Advanced NSCLC
About the only thing that lung cancer experts agree on in the management of stage IIIA NSCLC with N2 nodes involved is that this setting is the most controversial one in the field of lung cancer. First, it's worth backing up to clarify what we mean when we talk about stage III, also referred to as locally advanced, NSCLC. As outlined in another summary chapter, staging is based on the 3 components of Tumor (the primary cancer from which everything originated), Nodal status (which lymph node areas has the cancer spread to), and Metastases (has it gone through the bloodstream to develop new lesions in more distant locations, most common the liver, adrenal glands, bones, other parts of the lungs, or the brain). Below are schematic drawings from a not quite current staging system after a new revision in 2010, but this is the staging upon which eligibility for our prior studies are based. Nodes on the same (ipsilateral, in medical jargon) side of the mid-chest (mediastinum, between the lungs), are called N2 nodes, which are considered more advanced than N1 nodes inside the same lung lobe as the main cancer. Nodes on the opposite side (contralateral) of the mediastinum, or above the collarbone/clavicle (supraclavicular nodes) are termed N3 nodes, which are associated with a higher stage and less favorable prognosis than N2 or N1 nodes. The other potential determinant of locally advanced NSCLC is a higher tumor stage: a tumor that involves chest wall or is close to the carina, which is the split of the main airway (trachea) into right and left main bronchi, can be resected but is complicated and designated T3. A tumor that involves classically unresectable structures (though potentially with exceptions, depending on circumstances and the surgeon) that are shown below are called T4. The division between N2 and N3, and between T3 and T4, separates stages IIIA and IIIB, as shown below:
For lung cancer in general, but particularly in the setting of stage III NSCLC, it is worth bearing in mind that our efforts need to be directed at controlling local disease (essentially, the chest-based disease we can see on scans) and distant disease (the potential for spread to other parts of the body by micrometastases that may travel in the bloodstream). We are concerned about both disease that we can see, and presumed/possible disease that we can't see:
In general, tumor stage is correlated with a higher risk of local recurrence, while nodal stage is correlated with an escalating risk of more distant spread. Because there is both risk of local recurrence and distant spread, it is customary to treat with a combination of local therapy (such as surgery or radiation, or both) and systemic therapy (such as chemotherapy, most commonly). Our goal is typically to treat with curative intent, which means that we are often willing to accept the potential for challenging side effects if more aggressive treatment may translate to higher cure rates.
We'll focus on stage IIIA, N2 node-positive NSCLC for the remainder of this summary, as other settings will be discussed separately. Please note that the following discussion includes survival statistics that may be discouraging to some, so only read on if you prefer to see these.
The Heterogeneity of Stage IIIA N2 NSCLC
Stage III non-small cell lung cancer (NSCLC) is a widely heterogeneous disease that includes patients with disease involvement ranging from a microscopic single lymph node involved, which can be associated with a comparatively favorable long-term survival of 30-40% even without treatment beyond surgery, to multi-station, clinically enlarged (on scans) and sometimes bulky lymph nodes with a considerably worse prognosis. This heterogeneity in extent of disease and prognosis, all within the same stage, is one of the factors contributing to the controversy in how best to manage patients with stage IIIA N2 NSCLC. In fact, most experts would advocate treating patients very differently if they have a single node involved with microscopic disease, versus multiple enlarged N2 lymph nodes and areas within the mediastinum, even though they are technically the same stage.
Stage IIIA NSCLC, specifically with N2 lymph node involvement, is generally considered to be the highest stage for which surgery is routinely considered, and it is in this setting that the role of pre-operative, also referred to as induction or neoadjuvant, therapy has been most studied. A couple of small, randomized trials published in the early 1990s demonstrated striking improvements in overall and disease-free survival in the group of patients that received chemotherapy prior to surgery. Each trial enrolled only 60 patients, and both closed early after showing remarkable and statistically significant benefits for the patients who received chemotherapy before surgery:
The curve on the upper right is progression-free survival (PFS), and the one on the lower right is overall survival (OS) from a Spanish trial that was published in the New England Journal of Medicine. Though both this trial and another one from MD Anderson Cancer Center published in the Journal of the National Cancer Institute had many flaws, the fact that these were published from highly respected investigator groups, just a few months apart, with both showing striking differences favoring pre-operative chemotherapy made this approach the common practice for patients who have stage IIIA disease identified prior to planned surgery. The chemotherapy regimens used were older and largely replaced by clinical judgment about the best chemotherapy combinations to offer for NSCLC, but some form of pre-operative therapy became the prevailing standard of care.
Another feasible approach has been the use of pre-operative chemotherapy combined with radiation to an "induction" dose that is about 2/3 the dose of "definitive" radiation given with intent to cure and without a plan for subsequent surgery. The Southwest Oncology Group study 8805 was a phase II trial of 126 patients with stage IIIA or IIIB NSCLC who received two cycles of cisplatin and etoposide (PE) along with concurrent thoracic radiation therapy (RT) to 45 Gray (Gy, the unit of radiation dosing), followed by surgery in cases of non-progression. While this trial included many patients who would widely be considered to be unresectable, 85% of stage IIIA patients and 80% of IIIB patients were able to undergo surgery, and an encouraging three-year survival of 26% was achieved overall (27% and 24% for IIIA and IIIB, respectively). This trial, as well as many others since then, demonstrated the highly significant predictive value of achieving mediastinal lymph node sterilization after induction therapy, with a three-year survival of 44% vs. 18% (p = 0.0005). In other words, the patients for whom chemo/radiation eradicated all of the previously viable cancer within their mid-chest did far better than the patients in whom chemo/radiation did not eliminate all viable nodal disease in the mediastinum. Other studies, including those that have evaluated chemotherapy alone as an induction strategy (example here), have also demonstrated that patients who had no evidence of residual mediastinal disease after neoadjuvant therapy demonstrate a far superior long-term survival than those who do not achieve mediastinal sterilization after induction therapy.
Surgical vs. Non-Surgical Management of Stage IIIA N2 NSCLC
The report that long-term survival is highly associated with response to treatment prior to surgery raises the question of whether surgery may not add considerably to the curability of the overall approach. This is a critical question in light of the added morbidity and potential mortality (10% in SWOG 8805) of a trimodality approach of chemo/radiation followed by surgery. North American Intergroup trial 0139 (INT 0139) was conducted in order to specifically address this issue. In this trial, 439 medically fit patients with T1-3N2M0 NSCLC, some with bulky mediastinal disease, enrolled over more than seven years were first randomized to a surgical or non-surgical arm, then all patients received chemotherapy with PE for 2 cycles and concurrent RT to 45 Gy. All patients were evaluated for response by a CT, and those without progression then received either surgery and two cycles of post-operative (adjuvant) PE chemotherapy or continuation of RT to 61 Gy without a treatment break, followed by 2 cycles of the same chemotherapy. The schema for this trial is shown below:
While PFS favored the arm that received trimodality therapy (median 12.8 vs. 10.5 months, 5 year 22.5% vs. 11.1%), there were more treatment-related deaths on the surgical arm. Overall survival was not significantly different between the two groups (median 23.6 vs. 22.2 months; 3-year 27.2% vs. 20.3% for surgical and non-surgical arms, respectively).
These findings suggested that while more patients after surgery remained cancer-free after a trimodality approach of induction chemoradiotherapy followed by surgery compared with chemo/radiotherapy alone, this benefit was partially offset by a greater frequency of treatment-related deaths in the surgical arm. In the OS curve on the right above, note the chemo/radiation arm on top shortly after surgery, followed by cross-0ver and later superior survival for the surgical arm. This is compatible with the idea that there were more deaths early on during and after treatment with a trimodality approach, but that those who got through the very significant challenge of the treatment may have done better later.
The investigators noted that a significant proportion of the deaths were noted to be in patients who underwent a pneumonectomy (removal of an entire lung) (particularly a right pneumonectomy), an exploratory analysis matched patients who received either a lobectomy or a pneumonectomy with patients who underwent chemo/radiation, controlling for age, sex, performance status, and T stage. This post-hoc (after the fact) analysis suggested that patients who underwent lobectomy had a superior survival compared with their matched counterparts who underwent chemoradiotherapy, while the opposite was true for patients who required a pneumonectomy:
Based on this work, some cancer care specialists try to distinguish between patients who would be projected to require a pneumonectomy to remove all of the tumor surgically and those who would be expected to require only a more limited surgery of a lobectomy. They would be more likely to reserve consideration of surgery for those patients who would not likely require a pneumonectomy, in light of the limited evidence that such patients may not be well served by surgery.
Further Insights from Europe
Another trial that compared a surgical to a non-surgical approach is a European trial by van Meerbeeck and colleagues that enrolled a total of 579 eligible patients with histologically proven stage IIIA N2 NSCLC, again with a range of minimal to bulky disease but all felt to be "unresectable" for one reason or another initially, to receive initial platinum-based doublet chemotherapy for three cycles. Following this, patients underwent repeat imaging, with the 332 responders (including minor responses) randomized to either definitive RT (60-62.5 Gy) or surgery.
Importantly, 47% of patients assigned to surgery underwent a pneumonectomy, a remarkably high proportion (typically associated with higher risk), but the treatment-related mortality with surgery was only 4%. Also, 40% of the patients assigned to surgery underwent post-operative radiation therapy (unplanned on the trial). The two sequential strategies did not differ in median overall survival (16.4 vs. 17.5 months, respectively), 5-year survival (15.7% vs. 14%, respectively), or median progression-free survival (11.3 vs. 9 months, respectively), and the efficacy curves were superimposable:
In light of the fact that nearly half of the patients on the surgical arm underwent a pneumonectomy, many received radiation after surgery, and the nonsurgical arm received an inferior sequential chemo/radiotherapy approach instead of what is widely considered to be a more curative concurrent definitive chemoradiation strategy, it is hard to draw any firm conclusions from this unorthodox study aside from there being no clear advantage to surgery for these patients.
Optimal Pre-Operative Therapy?
Following the Intergroup 0139 trial that suggested a potential for superior survival for a strategy of induction therapy followed by surgery among patients who had non-bulky disease that would not require a pneumonectomy, a key question emerged of whether radiation may add more toxicity than benefit to the pre-operative component of therapy. While induction chemo/radiation is associated with higher rates of a pathologic complete response (pCR) (no evidence of any living cancer cells remaining at the time of surgery) than induction chemotherapy alone, it also adds toxicity, which may be detrimental in a setting such as the Intergroup trial in which the benefit in terms of enhanced disease control was largely counterbalanced by treatment-related mortality. A follow-up Intergroup study run by the RTOG along with SWOG attempted to directly test induction cisplatin-based chemotherapy vs. chemo/radiotherapy (to 50.4 Gy) in patients with non-bulky stage IIIA N2 NSCLC, with surgery planned for all non-progressing patients following induction therapy. Adjuvant Taxotere (docetaxel) for three cycles was planned as post-operative therapy, based on the very encouraging results from SWOG that are discussed elsewhere.
Unfortunately, this trial later closed due to poor accrual rates, leaving this question still unanswered.
Surgery as Initial Therapy?
A final consideration is primary surgery followed by chemotherapy with or without radiation. This approach has not been studied comprehensively, but a survival benefit has been seen in patients with stage III NSCLC who received cisplatin-based chemotherapy in recent adjuvant therapy trials that are discussed elsewhere. Post-operative radiation therapy, while shown to be detrimental for patients with stage I and II NSCLC, has recently been shown to be associated with some potential benefit for patients with N2 disease, based on a retrospective data review of SEER data and an analysis of the ANITA trial. Whether this is best given in a sequential or concurrent therapy with chemotherapy is not known and has not been directly tested. However, if resection is feasible and planned, an induction therapy approach is generally favored for patients identified pre-operatively as having stage IIIA N2 NSCLC.
In summary, patients with stage III NSCLC can be treated with a wide range of approaches, but the data suggest that we may do a better job by individualizing our patient populations in terms of tumor bulk, number of nodal stations involved, patient tolerance of aggressive therapy, sterilization of mediastinal lymph nodes post-induction therapy, and potential need for pneumonectomy. There is no evidence at this time that demonstrates a clear survival benefit for surgery for a broad range of patients with stage IIIA N2, potentially resectable disease compared with concurrent chemoradiotherapy. Despite this, some analyses provisionally suggest that patients with non-bulky N2 nodal disease, especially in a single nodal station and without clinical lymphadenopathy on scans, may be well served by a surgical approach. On the other hand, patients with more marginal functional status, those with bulky mediastinal disease, and arguably those with residual viable mediastinal disease after induction therapy are widely considered to be better served by a non-surgical approach. Meanwhile, it remains unclear whether induction therapy should consist of chemotherapy alone or chemoradiotherapy, which agents are the best for this setting, whether and how to administer post-operative therapy for patients with viable N2 disease at surgery.
It is probably most appropriate to say that there is no "optimal" treatment approach for stage III NSCLC because it encompasses a very wide range of patients in terms of overall health, tumor burden, and other important factors, so it is very appropriate to tailor treatment recommendations based on a range of prognostic factors in the locally advanced NSCLC setting. In this setting perhaps more than any other, it is also critical to develop treatment strategies in a multidisciplinary fashion from many members of the treatment team.
We thank Pfizer Oncology for the educational grant that makes the Reference Library for Lung Cancer possible.