I've been involved in a wide range of discussions, both here and in my own clinical, about the fairly common situation of how to approach a situation in which the story on paper and what you see actually happening are incompatible. For instance, last week I and several of my colleagues participated in a journal club (a group discussion of a new and/or controversial journal article or two), in which the topic was the potential utility of doing surgery for unusually early small cell lung cancer tumors. We've also had several recent questions about patients in whom the diagnosis of bronchioloalveolar carcinoma (BAC) is being considered, and it's not clear whether to treat this sometimes very indolent cancer as a full-fledged NSCLC, a non-entity that might sometimes be ignored, or as a separate category worthy of being managed differently from the standard approaches for other NSCLC subtypes.

It's important to highlight that the discrepancy between the expected outcome based on a pathology report and the clinical picture in front of you can cut both ways. In some cases, you may have a biopsy of a lung nodule that shows no cancer, but if it's growing and continues to grow, that's not very reassuring, and you'd suspect that the biopsy missed the diagnostic part of the tumor that would confirm viable cancer. In other settings, a biopsy of a lung nodule might diagnose cancer, leading down a path toward the typical management with surgery, etc., but if you happened to have old films that showed that the nodule was actually minimally changed over 3 years or more, it might be reason to take a step back and wonder whether you haven't already been furnished with some valuable information that might lead you to individualize and change your treatment plan.

To be more specific, I'll turn to an ASCO presentation I reviewed there that had also actually just been published in the journal Cancer (abstract here), based on work from a group from Beth Israel Deaconess Medical Center in Boston. This research compared outcomes in the SEER database, a registry of cancer results from large sections of the US that provide a picture of the overall cancer outcomes nationwide, for the 684 patients with resected node-positive BAC tumors to the results for the 9809 patients with node-positive non-BAC tumors. The investigators attempted to match the BAC patients with non-BAC patients who shared similar characteristics for age, sex, type of surgery, number of lymph nodes collected, number of positive nodes, and a few other factors. They found some modest differences: cancer-specific survival was better among BAC patients with N2 node involvement compared with N2 node-positive non-BAC patients, for example, but there wasn't a difference in overall survival for N2 patients when comparing BAC to non-BAC NSCLC, and there were no significant differences at all for N1-node positive patients.

The main point is that the curves, BAC vs. non-BAC, pretty much track together.

But you've probably read some of my prior writings on BAC that suggest it can be a slower growing, indolent cancer that merits being considered as a separate category. The problem that I see is that the data came from the SEER database, which is just a garbage bag of all of the results from all sorts of places -- some very good, and some where the physicians probably made mistakes in classifying cancers as BAC that shouldn't have been. In fact, I attended a conference on BAC in NYC in 2004 at which 6 expert pathologists reviewed tumor tissue from 131 patients who one of them felt had BAC, and the 6 experts only unanimously agreed on the diagnosis of BAC in a single patient! The diagnosis of BAC is tough, it's controversial, and the SEER database is kind of amateur hour. It's a great place to get the view from 20,000 feet, to see the big picture of what's happening in the real world. It's not the way to judge finesse work. And in this study, 27% of patients with "BAC" were noted to have poorly differentiated cancer, which is exactly what you shouldn't see for BAC. In fact, BAC is technically a non-invasive cancer that shouldn't be able to travel to the lymph nodes. If there's cancer in the lymph nodes, that's from an invasive component that isn't BAC (you can have a mix of BAC and invasive cancer, but that isn't necessarily anything like the indolent BAC we sometimes see). And the curves above show that the behavior of the cancer is far more important than what it's called. If it's in the nodes, it's invasive, and it can spread. It's not acting like an indolent cancer.

On the other hand, if someone has a SCLC that doesn't spread to the nodes and is just a solitary 2 cm tumor, that's very unusual for SCLC, so you can question whether the pathologic diagnosis is really correct, but I'd still argue that it's reasonable to modify your treatment plans based on what's happening rather than what you'd expect from the textbook. Sometimes the cancer doesn't read the textbook.

And although we strongly suspect that a never-smoking Asian woman with a lung adenocarcinoma is going to respond well to tarceva, if we see that this person is getting worse, we have to accept that there are differences between what we expect from a population and how individuals do. Just as we can't use population results to say anything definitive about an individual patient's prognosis, the general guidelines give us a sense of what to expect, but we have to modify that heavily based on our experience. Believe what you see.

The basic premise is that everything we write here about treatment standards are general principles, not hard and fast rules. Medicine is full of situations every day of patients who throw you a curve, and the best ones know when to follow the rules and when to bend or break them. That also leads to style differences in what we might consider to be the best approach, and that's OK.