Filling the Gap: Creating a Prognostic Staging System for Patients Treated With Neoadjuvant Chemotherapy

David Winchester, MD, City of Hope, had an instrumental role in the creation of the breast cancer staging system published in the 8th edition of the American Joint Committee on Cancer (AJCC) Cancer Staging Manual. In this interview, Dr. Winchester discusses a major update to the AJCC breast cancer staging system, addressing a long-standing gap in prognostic staging for patients treated with neoadjuvant chemotherapy. Dr. Winchester explains how incorporating treatment response and tumor biology enhances staging relevance and outlines what this means for clinical decision-making, patient communication, and future editions of the staging manual.

Transcript

What motivated you to pursue the development of a novel prognostic staging system specifically for patients treated with neoadjuvant chemotherapy?

Dr. Winchester: I was instrumental in the creation of the 8th edition of the AJCC breast cancer staging scheme that’s been out for I think close to 7 years. And so that was, I would say, a controversial release from the AJCC, a real shift in paradigm from the standard approach to staging breast cancer patients.

And just to go back a step, the last 8 editions have really focused on the acronym TNM, “T” standing for tumor, “N” standing for node, and “M” standing for metastatic disease. That’s been historically the basis of how every cancer patient’s been staged, according to the American Joint Committee on Cancer. And it’s been important for us to have a perspective, have a reference point to make predictions about survival. And everyone, the laypeople, patients, you know, our country is well versed with stage. You know, people talk about, oh, it’s stage 4, right? Or it’s early stage.

So stage has a lot of relevance for patients to understand their prognosis. “Four” obviously sounds like a death sentence. That’s changed in many respects. So really the 8th edition, which was published about 7 years ago, tried to make staging more relevant. The reason it wasn’t so relevant is because breast cancer is a rapidly advancing field that relies much more on other variables besides those 3 acronyms. So the different subsets of breast cancers include different molecules, different markers to help clinicians understand how they are treated, and prognosis is also affected by those markers.

And so the 8th edition was a breakthrough to incorporate 4 other variables besides those 3, which included the tumor grade, which is a measure of how rapidly cells grow; the inclusion of estrogen receptor, which is hormone receptor-positive breast cancer. Another marker for that is progesterone. And the 4th variable is a cell surface molecule called HER2, which has become a very important marker to help us understand when to invoke monoclonal antibodies to HER2-positive breast cancer. So the 8th edition accomplished our goal of creating relevancy of incorporation of those variables with the TNM. So it then went from 3 to 7 different variables. And the model worked beautifully: it’s been confirmed by several different databases well beyond the United States borders and it’s helped us to make predictions based upon those important variables as well, besides the TNM.

The problem with the 8th edition, though, is that we barely reached a point 7 years ago to have enough data to both that model that I just described to you. And that model created clinical staging, which is a patient’s initial stage when they’re first diagnosed before they’re treated, and then a pathologic prognostic stage, which gives further detail to the patient’s prognosis. And most of those patients then and now are patients that are treated first with surgery. So going from a clinical diagnosis to a prognostic pathologic diagnosis, it’s simple to do. That was accomplished in the 8th edition. But the big missing piece in the 8th edition was we didn’t have the data to stage patients if they had chemotherapy first.

So essentially up until the 8th edition, every patient could be staged both clinically and pathologically after surgery. But the 8th edition specifically excluded patients who were given chemotherapy as their first point of treatment. So there’s been a big gap then for patients the last seven years we’ve not had a clinical stage … I’m sorry, have not had pathologic post-surgical stage assignment. And the problem with this is their life expectancy of their prognosis has been hard to define. So we’re not really sure, and clinicians have known for years that the response to the drugs up front becomes very important to define their survival. So we had to wait, still being a member of the AJCC, we had to wait for enough data to create this model. And that’s what this model is done. So it’s filled in the blanks of an increasing number of patients who get treated with drugs first before they have surgery.

And I’d say that we barely made, I would say the critical number of patients to study to have validated models. And so that’s important for us to create a system that’s going to stage patients that have chemotherapy first. We have to have a model that’s really been tested, validated, and then most importantly, for people to use it, it needs to be published in a peer-reviewed journal. So this is kind of the last missing piece that we finally got over that that barrier. The rest of the 8th edition is now going to essentially stay the same for the 9th edition, the next edition. The one exception is there’s going to be a new stage in scheme for metastatic stage 4 breast cancer as well. But the biggest thing that we’ve been waiting to address are these growing number of patients that have drugs first.

Can you walk us through how you defined and categorized treatment response, and why it was important to distinguish between complete, partial, and no response?

Dr. Winchester: So there are different ways to categorize response, but the tools that we could create with this database, and just to go back a step, the database that was the foundation for the 8th edition was the National Cancer Database and that is the database that is populated by patients treated in the United States and Canada from hospitals that are Commission on Cancer–accredited hospitals. About 1400 hospitals participate in this. Essentially all the patients that are treated, not just breast cancer, but other cancers as well. Their data is collected by cancer registrars in their hospitals. And then it’s submitted to the National Cancer Database, which collects the data and scrubs it. And it’s free to the public to study, which is how I used it. It’s a database now that contains millions of patients. And so that’s our source of data.

And so we had to create a definition, as you asked about, on what constitutes a complete response, partial response, and no response—and really within that category [no response] is progression. So some patients are given their chemotherapy and they progress. That’s obviously a bad situation. Our hope is that we see a complete response. And so standard definitions that are more or less universally accepted would be for breast cancer, a patient who starts with any stage of disease—excluding stage 0—and ends up having no evidence of cancer in the breast and no cancer within lymph nodes … and one can argue that we could include patients with metastatic breast cancer, but that’s sometimes hard to really understand as far as a complete response is concerned, takes years sometimes to know that. So this study excludes patients who have a clinical diagnosis up front of stage 4 breast cancer. It’s stage 1 through 3. And so the definition of complete response then consists of anybody stage 1 to stage 3C could become stage 0. Partial response is going down either in the T or the N category while the other category stays the same or gets better too, but doesn’t get to 0. And the non-responders or the progressive responders are patients that go from the same stage before surgery or chemotherapy to the same stage after chemotherapy or to a more advanced stage.

So roughly those are 3 equal size groups, but they vary in size according to the receptor combinations which make them unique.