For patients with locally advanced NSCLC, the question of whether to pursue a surgical or a non-surgical approach has a great deal to do with the extent of mediastinal (middle of the chest) lymph node involvement. The mediastinal nodes are shown here:

(click to enlarge)

First, at the time of initial staging, patients with bulky (>3 cm) disease in the mediastinum, or those with disease involvement more than one nodal station, are less appropriate candidates for surgery than those with non-bulky and single-station disease. In fact, a French retrospective review of over 700 patients with N2 disease who underwent surgery at any of six centers (Andre abstract here) demonstrated that there are quite varied long-term outcomes for different patients that all fall under the same stage of IIIA with N2 disease, and that the patients with a single-station and microscopic involvement (as opposed to clinical enlargement that is visible as abnormal on CT (greater than 1 cm in diameter):

That was in a group of patients who underwent surgery, and just a view of how patients did after the fact.

What has also been quite interesting, and often noted, is how important the eradication of disease from the mediastinal lymph nodes (also known as lymph node sterilization) after induction chemo or chemo and radiation is as a predictive factor for future outcomes. More than a decade ago, Kathy Albain and colleagues from the Southwest Oncology Group reported on SWOG 8805 (abstract here) , an interesting trial in which radiation to 45 Gy and concurrent cisplatin-based chemo were given to patients with locally advanced NSCLC before planned surgery (including stage IIIB disease, interestingly, although that has never emerged as a standard approach). Several intriguing results came from this trial, but one of the most striking findings was that patients who had no evidence of residual active cancer in their mediastinal lymph nodes after induction therapy did remarkably better than those who had residual disease in their mediastinal nodes, even though both groups underwent surgery if they didn't have progression of disease after induction. In fact, the five-year survival was three times better (!!) in patients who cleared their mediastinum after induction therapy:

So in light of the fact that survival was largely predicted by how patients did based on treatment BEFORE they ever went to surgery, this raised the question of whether the surgery was really necessary for potentially resectable locally advanced NSCLC, an issue still debated now (see post on this subject).

Several other trials have demonstrated similar results. For instance, a more recent trial of chemotherapy alone with cisplatin and taxotere before surgery for stage IIIA N2 NSCLC conducted in Switzerland (abstract here) demonstrated that the survival for patients who had no evidence of residual N2 disease had a remarkably better survival than patients who did not, a difference that was similar in magnitude to the difference in survival seen between patients who had all disease removed at surgery and those who had incomplete resections:

In the past few years, there have been some reports that counter the idea that survival is destined to be far worse in patients with residual disease after induction therapy and suggest that patients may feasibly still undergo surgery(abstract here), but in most cases we are less inclined to recommend surgery or patients with viable N2 disease. Surgery can still address all detectable disease, but the fact that the cancer survived induction therapy suggests that this may be a more resistant, aggressive cancer that is more likely to recur after treatment. Knowing this, and that definitive chemo and radiation can also potentially cure locally advanced NSCLC without the challenge of recovering from major lung surgery, most lung cancer teams recommend a non-surgical approach in this situation. But having a resistant cancer that survives planned pre-operative treatment is suggestive that this cancer is going to be particualrly challenging to cure no matter how it is managed.