Staging in lung cancer, as well has two categories, clinical and pathologic. The clinical staging is based on what appears on scans like the CT and PET scan that are now pretty routine parts of the staging workup. Our scans are better than ever before, but some lymph nodes with cancer involvement are not enlarged and have no visible abnormalities, and no scan can pick up lesions that are only visible as a small collection of cancer cells under a microscope. Because of that, pathologic staging, which is the stage that a patient ends up with after the tissue from surgery has been reviewed under a microscope, is more precise, and it's associated with a "better" prognosis stage for stage vs. clinical staging, because clinical stage includes a fraction of people who actually have higher stage cancer than is detected on scans. It's also possible for scans to over-estimate a stage, such as when lymph nodes in the middle of the chest are enlarged due to inflammation or infection (such as in the setting of a pneumonia in a lung lobe blocked by tumor -- known as a post-obstructive pneumonia). But understaging from scans is more common.

Investigators from NYU raised the question of what clinical features were associated with having "occult" mediastinal (mid-chest, between the lungs) lymph node involvement and actually having pathologic N2 disease if the clinical staging indicated stage I NSCLC, or no lymph node involvement with cancer. This issue is particularly relevant with regard to the question of whether mediastinoscopy, a moderately invasive procedure described here, should be done in all patients or more selectively, for the patients with a higher risk of having mediastinal node involvement and being higher stage. This is particularly important because most experts would recommend chemo alone or with radiation before surgery, or chemo and radiation without surgery, for patients with mediastinal node involvement. In fact, even among expert thoracic surgeons, this issue of whether the invasive mediastinoscopy procedure should be done on everyone or more selectively, and if so whom, is a very controversial one.

The report by Lee and colleagues in the Annals of Thoracic Surgery (abstract here) reported their findings from reviewing records on 224 patients treated there from 2000 through 2006 who had no evidence of nodal involvement on CT or PET scans and then had surgical staging done in a uniform way by two experienced thoracic surgeons at NYU. The investigators then reviewed the likelihood of having mediastinal N2 lymph node involvement with factors like location of the tumor (central or peripheral in the chest, peripheral being in the outer 1/3 of the chest on films), tumor size, pathologic subtype (adeno vs. squamous, etc.), and PET standard uptake value, which is correlated with metabolic activity of the cancer (higher cell division rates lead to more uptake of labeled sugar molecules).

They found that 16 of 224 patients (7%) with clinical stage I NSCLC had mediastinal nodes involved. When they looked at the clinical factors, they saw that central tumors were more likely to have N2 nodal involvement and that larger tumors were increasingly likely to also be upstaged by mediastinoscopy or full surgery:

(Click to enlarge)

In looking at the tumor subtype of these resected cancers, the breakdown was very skewed toward adenocarcinomas. Among the 224 resections of clinical stage I cancers, 178 were adenocarcinomas, among which 100 were adeno/BAC and 9 more "pure BAC" (wow, that's a lot of BAC). Most NSCLC series in the US aren't that weighted toward adenocarcinoma (just 34 squamous, 5 large cell, and 7 "other"), and I think this may be because NYU is one of the key sites in the I-ELCAP screening trial, so they may see a lot of surgical candidates who are detected from screening, which has been controversial in part because some people think that many of the cancers that are detected are really small, slow-growing BAC lesions that may not be as threatening to survival as lung cancers picked up outside of screening efforts). Regardless, the investigators found that all 16 cases of occult N2 disease were in patients who had adenocarcinoma (9%, vs 0% for squamous/large cell/other), but the difference in rate wasn't statistically significant (p = 0.082), meaning that there was enough of a likelihood that this difference was just random chance alone that the association of higher risk with adenocarcinoma histology may not be true, even if it's a strong trend.

Finally, as has been noted in some other studies, the patients who had mediastinal involvement had a higher median SUV (6.0), than the patients who didn't have mediastinal nodes (median 3.6). Looking at it another way, they also found that the likelihood of occult N2 disease was significantly higher in patients who had a primary tumor with a maximum SUV over 4.0 compared with 4.0 or lower (10.5% vs. 1.9%, p = 0.007).

These results aren't entirely conclusive, but they're quite compelling, since this is a large collection of patients from a center that is very good at thoracic surgery and has a pretty uniform approach between their two highly trained thoracic surgeons. It would be particularly helpful if these variables were validated by looking at these associations in another collection of patients with clinical stage I NSCLC from another center. Still, some of these factors have been identified in prior studies, so these results jibe with much of the literature out there. The authors suggest that in light of these findings, it's appropriate to use mediastinoscopy selectively, foregoing it in patients who have very few or none of the risk factors described above. In particular, the smaller, peripheral, low SUV tumors (especially if not adenocarcinoma, but that was only a minority of their series) are quite unlikely to have mediastinal involvement, so it may be very appropriate to minimize the cost and the risk of another invasive procedure and play the odds that the scans are telling the whole story and there is no cancer in the mediastinal lymph nodes.