A topic that came up in a recent expert round table case discussion was the issue of how to manage a patient with a lung cancer for which the pathology report says "NSCLC not otherwise specified (NOS)", or "poorly differentiated NSCLC, NOS". What does this actually mean, and what does it mean in terms of treatment options?

Tumors of pretty much all types are categorized by their tumor grade, how "differentiated" they are, which basically means, "how much do the cancer cells look like the cells they started out as?". Different cells of the body start out as stem cells, which means that they're not specialized to be any special kind of cell, like one that detects light in the back of the retina, lines the esophagus, or is optimized for lung function. Most cells of the body are differentiated, so that the appearance under a microscope shows that it's a liver cell, part of the kidney filtering mechanism, heart muscle, etc.

Cancer cells, however, have mutations in them that make them grow and divide faster than other cells (that's why they make a tumor that pushes other tissues aside), and they usually have several. As they grow and divide, they often make sloppy copies of their DNA that leads to more mutations. Cancers therefore are made of cells that may look a lot like the normal cells they originated from (well-differentiated), or they have lots of mutations that make the cells look so chaotic that they don't look at all like the cells they started out as (poorly differentiated).

Today, oncologists want to know whether a non-small cell lung cancer (NSCLC) is an adenocarcinoma, squamous cell carcinoma, large cell neuroendocrine carcinoma, etc. But about 20% of the time on various studies, we get an answer back of "NSCLC not otherwise specified". As explained by Dr. Matt Horton, expert lung cancer pathologist, a lung tumor may be classified as NSCLC NOS because of either of two reasons:

1) there isn't enough tissue, because the biopsy material was very scant, or

2) the tumor is so poorly differentiated that even with all of the material in the world, a good pathologist couldn't identify the underlying NSCLC histology

Dr. Horton says that many of the conclusions of NOS are because there isn't enough pathology material to review, and now there is an ever-growing list of reasons for why we'd want to ensure that this isn't a problem. First, knowing whether a cancer is a squamous cell carcinoma or an adenocarcinoma or another subtype is important for deciding whether it's appropriate for patients to get Avastin (bevacizumab) (since the risk of serious or even fatal bleeding complications is considerably higher in patients with squamous NSCLC who receive an anti-angiogenic agent like Avastin) or Alimta (pemetrexed) (because the evidence shows that Alimta isn't effective in patients with a squamous NSCLC). Second, there's now a growing role for collecting enough tissue to due testing for specific mutations like EGFR, possibly an ALK rearrangement, and arguably K-RAS. With the field moving to molecularly-based treatment plans for specific molecular defects, and more clinical trials and treatments being dependent on molecular testing, it's necessary to have tissue to do these studies on.

But sometimes even with a resected early stage cancer, a situation in which a pathologist has all of the tumor tissue he or she could want, it's not possible to make a good histologic assignment because the cancer cells just look too chaotic under the microscope. A study from ASCO 2009 showed that, as you'd expect, expert pathologists from academic centers tend to have agreement in their interpretation of NSCLC subtype than community-based pathologists, but neither groups really agree consistently on poorly differentiated tumors. Essentially, this is just saying the same thing we'd say about interpreting ambiguous lesions on scans, or treating aggressive and resistant cancers: the hard ones are hard for everyone. In fact, there is also evidence that, stage for stage, patients with poorly differentiated cancers don't tend to do as well as patients with better differentiated cancers. This is probably because they're also more aggressive. The cancers that are so chaotic that experts can't determine the histologic subtype are also likely to grow and divide very quickly, and be far less likely to have a specific mutation like EGFR or an ALK mutation driving them -- instead, they're driven by a combination of many, many mutations.

It's worth noting that there are companies now studying the "molecular signature" of a poorly differentiated cancer that promises to clarify whether a tumor is actually a squamous NSCLC or adenocarcinoma or other NSCLC subtype. Companies like Biotheranostics can do detailed molecular typing, and others exist as well. I don't generally use these, because I think that there may well be something different about a cancer that requires this kind of assessment to know its histology than the ones that a pathologist can assign. I wouldn't presume that a cancer that is so poorly differentiated that you need molecular testing to determine that it's a lung adenocarcinoma will act the same as a better differentiated adenocarcinoma.

So practically speaking, what does it mean if a cancer is so poorly differentiated that you can't assign a histology? First, these tend to be very unlikely to carry an EGFR mutation or ALK rearrangement, so I would favor chemo-based treatment and wouldn't prioritize testing for every molecular marker under the sun early in the workup. As I mentioned above, these tumors have a tendency to be bad actors, but we still treat them basically the same as other NSCLC tumors. I personally am less enthused about using Avastin in patients with a very poorly differentiated tumor, though the ECOG study that showed a survival benefit with Avastin added to chemo included 20% of patients with NSCLC NOS, and they really didn't have increased bleeding at all. Still, I'm a little more concerned that bleeding could be an issue, particularly if the cancer is in the central part of the chest and near any major blood vessels. That said, other experts, including ones included in the case discussion from the beginning of this post, are still comfortable giving Avastin to patients with NSCLC NOS, so it's clearly a debatable question.

The other issue is whether to give Alimta to patients with a poorly differentiated NSCLC that can't be assigned a histology. The approval of Alimta is now for patients with non-squamous NSCLC, and NSCLC still is considered non-squamous if it isn't called squamous. Still, it's not clear that patients with a very poorly differentiated cancer will benefit as much as the patients with a moderately differentiated adenocarcinoma. When other chemo agents commonly used for NSCLC aren't dependent on the histology, I and I believe most oncologists favor an alternative to Alimta as a first line approach and tend to relegate it to a later choice. This isn't the same situation as being potentially more dangerous than other options, but we'd always want to use agents with the best probability of success as our first treatment.

This is certainly a tough area, and it's also a moving target. More and more of the field is becoming dependent on the tissue ("the tissue is the issue"), and we may move toward identifying particular mutations and genetic patterns associated with better results with chemo A vs. chemo B, but right now this is the state of the field.