Wednesday, June 7, 2017

A Big Advance in Cancer Care

One May 23 a very significant event took place in the history of cancer treatment. The FDA approved use of an existing drug for first line treatment of cancers based purely on specific DNA changes in the tumor, regardless of the body site of the tumor.

This is a first for the FDA. All previous cancer drug approvals have been for cancer originating in specific body site such as colon, breast, lung, etc.

This is a long post, but I want to describe the context so you can understand the significance of this event.

After the human genome was sequenced at the beginning of the 21st century, cancer researchers began to discover that some DNA changes were common in cancers originating in the same site: for example, a particular DNA mutation in the gene called BRAF is common in colon cancers. Drugs - called targeted therapies - were developed to kill cells with that specific mutation. Similar targeted therapies were developed for other mutations, which had common occurrence in other tumor sites.

Targeted therapies were a great advance and they are still being developed. But most of them produce drug resistance. The drug kills the cells at which it is aimed, but the tumor evolves. Yes, just like Darwin described: other cells in the tumor do not carry the targeted mutation and are thus more adapted to their environment and outcompete the targeted cells. The drug stops working and the tumor grows again: cancer recurrence.

In a separate thread, cancer researchers have wondered since the 1960s why the immune system doesn't attack tumors. One core answer to that question is provided by Siddhartha Mukherjee in "The Emperor of All Maladies": cancer is us. It is not a foreign substance or organism in the body, it is our own cells that have sustained DNA damage and lost control over their growth and division. Thus the immune system does not generally recognize cancer cells as foreign.

The immune system is incredibly complex. But knowing the human genome and being able to use that knowledge as a scaffold has allowed researchers to gain much more understanding of adaptive immune cell function; specifically the proteins on the surface of T cells and B cells, the core components of the adaptive immune system.

The most significant discovery in this realm so far was that some tumors could "hide" from the immune system by manufacturing a specific protein (PD-L1). Drug companies were able to make drugs that target that protein, unmasking the tumor cells and "releasing the brakes" on the immune system so that it would attack the tumor.

Results have been spectacular. It is not an exaggeration to say that this has been the single most significant impact on cancer care in decades. Cancers that were death sentences before these drugs, such as the first approved cancer type, metastatic melanoma, suddenly were treatable and some patients cancers disappeared completely (e.g. Jimmy Carter).

This class of drug is called immunotherapy for obvious reasons.

Unfortunately, these drugs don't work for all tumors. They currently produce a partial or complete response in about 30-40% of the cancers in which they are most used. So the obvious question is: why? And can we tell ahead of time which patients will benefit?

One of the drugs (pembrolizumab, marketed by Merck in the US as Keytruda) uses a biomarker to try to determine a patient's likely response. A biomarker is just a biological factor you can measure. Simple examples are weight or blood type. A more complex one (to measure) is amount of PD-L1 produced by a tumor. Pembrolizumab was believed to work in patients with "high" PD-L1 expression and was approved by the FDA for patients with that biomarker - in specific cancers.

[Full disclosure: I used to be employed by Merck but I have no current financial interest]

But more recent research has shown that what might really tell us which patients will respond to immunotherapies is how many mutations they have in their tumor cells' DNA. DNA is the template for proteins; DNA mutations can result in changes in the proteins produced. The theory is that the more mutations a cell has, the more changes are likely to occur in the proteins the cell has on its surface. And the more protein changes on the cell surface, the more likely the immune system will recognize the cell as "bad" and kill it.

As I've described in previous posts and above, cancer is a disease of the DNA. Accumulated mutations eventually activate an oncogene, causing cells to lose control over growth and division, or mutations disable a tumor suppressor gene, leading to the same problem.

Cells have multiple mechanisms to repair DNA damage and do so on a regular basis. But if you end up with mutations in the repair mechanisms, you're in trouble. One of these mechanism is mismatch repair. Loss of the mismatch repair mechanism can lead to one specific type of hypermutation called microsatellite instability.

These two characteristics of tumor DNA lead to high tumor mutational load. Remember that a high number of mutations seems to increase the likelihood that a tumor will respond to immunotherapy.

And now we're back at the the recent FDA approval. It says:

Keytruda (pembrolizumab) is indicated for the treatment of adult and pediatric patients with unresectable or metastatic solid tumors that have been identified as having a biomarker referred to as microsatellite instability-high (MSI-H) or mismatch repair deficient (dMMR). This indication covers patients with solid tumors that have progressed following prior treatment and who have no satisfactory alternative treatment options and patients with colorectal cancer that has progressed following treatment with certain chemotherapy drugs.

The most important part is that it is approved for any solid tumor that has these DNA characteristics, regardless of the body site in which the tumor arose.

[You may not be used to the term solid tumor. It means what it sounds like: a cohesive mass of cancer cells in a mass. Liquid tumors includes such cancers as leukemias and lymphomas in which the cancer cells are distributed through the body in blood, bone marrow or the lymphatic system.]

People like myself who work in cancer genomics have long "known" that a day would come when we would treat cancers by their molecular (DNA) characteristics instead of their site of origin.

This FDA approval represents the first milestone in that transition.

Friday, May 5, 2017

Four Years

May 6, 2013 was the day I first visited my doctor about a sensation in my throat. She noticed a lump on my neck.

Thus, today is the fourth anniversary of my cancer diagnosis.

I'm doing just fine, thank you. I go in for surveillance follow-ups every six months now with my next appointment in July.

Cancer survival statistics are based on "5-year survival", which is pretty simple: is the patient alive or dead five years from diagnosis? If a particular cancer is going to kill you, it usually does so by then.

So, a year from today I'll (hopefully) be an official survival statistic!

Since we're on the topic of cancer survival, I thought I'd take the opportunity to give you the big picture.

One thing I've learned in my new job is where cancer statistics come from: cancer registries.

Every hospital that treats cancer must report cancer incidence to their state registry. The process is designed so that a particular patient is only reported once (even though they may be treated at multiple hospitals) to avoid double counting.

These registries are staffed by Certified Tumor Registrars. These are specially trained people, many with medical backgrounds, who carefully curate each cancer case to accurately record data such as patient demographics (age, gender, zip code, etc.) date of diagnosis (important for survival statistics), actual diagnosis, site of the tumor, initial treatment and response to initial treatment, and a whole bunch of other data.

That data is collected at the state or regional level and then reported onward to the federal government through two programs: the National Cancer Institute's SEER program (Surveillance, Epidemiology, and End Results, covering about 28% of the US population in 15 regions) or the CDC's National Program of Cancer Registries (covering 45 states and D.C.).

Here's a diagram I made for a talk I'm giving soon that illustrates the process (click to embiggen):

All that careful data collection enables us to see trends in cancer over long periods of time.

Here are the stats on new cancer incidence and death rates as of 2013:

The long-term trends are encouraging. Here are graphs showing changes in cancer survival for the most common forms of cancer, comparing survival in the 1975-1977 period to the 2006-2012 period:

Some highlights:

  • In the 1970s you had 75% chance of surviving breast cancer. Now you're chances are above 90%.
  • Prostate cancer 5-year survival was 67.8% and is now above 99%.

There are still too many cancer types in which we haven't made much progress: lung and pancreatic cancer are still very bad news, for example.

Many of the cancer types with poor survival, especially those two, are due to the fact they tend to be asymptomatic and therefore are not discovered until they are at an advanced stage.

Researchers are hard at work on that. Tumors, including nascent ones, shed both complete cells and also contents of dead cells into the bloodstream. There is hope that "liquid biopsies", which aim to detect cancer cells or cancer cell DNA in the bloodstream, will be able to detect these and other cancers much earlier. That approach is still a ways from routine use. As you can imagine, the number of tumor cells or amount of tumor DNA in the bloodstream is very low; identifying them reliably is like trying to tune in a very weak radio signal.

One advance that doesn't yet show up in the survival statistics above, since they currently cover the period through 2013, are the "immunotherapy" drugs Opdivo (nivolumab) and Keytruda (pembrolizumab). These are among the first approved drugs that leverage the immune system to attack cancer - they essentially "decloak" the cancer so the immune system can see it. These drugs were approved by the FDA in 2014 and have had a dramatic effect in some cancers, especially metastatic melanoma (melanoma that has spread). About a third of these patients, for whom the prognosis was previously less than one year of survival, have experienced complete remission.

Anyway, I'm glad to still be here four years after my own diagnosis. And I look forward to becoming a statistical survivor a year from now.

Monday, April 3, 2017

The End of PAP Smears?

I've written several times about the importance of HPV vaccination for all young teens, and commented that if all teens were vaccinated HPV would vanish, and so would cervical cancer and my type of oral cancer. PAP smears would no longer be necessary.

Now there's news that PAP smears may already be on their way out.

DO NOT take this as advice to stop getting PAP smears as recommended!

When the PAP smear was developed (by Georgios Papanikolaou) it was a breakthrough in women's health. It has saved millions of lives by this point. But we subsequently discovered that most cervical cancers are caused by infection with Human Papillomavirus (HPV). Thus, detection of persistent HPV infection is a much earlier indicator of risk for cervical cancer than PAP smears.

Here is a STAT News article that covers the topic well. One highlight:

There are signs it’s catching on. Last year, the Netherlands wholesale switched from Pap tests to HPV tests, and Australia is set to follow in its footsteps this year. The journal Preventive Medicine devoted an entire issue to HPV testing in February. Clinical trials of at-home HPV testing are underway across the US, Europe, and Canada.

Sunday, April 2, 2017

Tumor Board

I wrote back in the fall that I had changed jobs and now work at the cancer center where I was treated: the Dana-Farber Cancer Institute. I work in Informatics, which is at the boundary between Computer Science and research - we try to use information technology to enable and accelerate scientific progress.

DFCI has a very strong research program and most doctors treating patients in the clinic are also researchers. A key part of my job is to understand the landscape of cancer treatment and research. To further that knowledge, this week I attended my first Tumor Board.

A Tumor Board is a regular meeting of a group of oncologists, surgeons, radiologists and related providers at which they discuss challenging cancer cases. Cases are put on the agenda by a doctor treating the patient and the Tumor Board is an opportunity to seek advice from peers regarding options for the patient.

The meeting looked very much like the photo above, but that is a random photo I found on the internet.

This first board I attended was for the Head and Neck Oncology Center, which is the "disease center" in which I was treated. In fact, one of my oncologists was present and the other was mentioned several times.

I came away from the meeting with two observations:

One: these people are amazingly skilled. I was especially impressed by the radiologist at the podium who would bring up the medical imagery (CT, MRI, PET or ultrasound scans) for the patient being discussed. He was impressively fast at finding just the right image to illustrate the aspect of the case that was under discussion, in real time as doctors were describing the case.

Two: I am damn lucky. My case was simple. I was given the standard of care for my cancer type and I experienced a "complete response" (i.e., my tumor disappeared). The patients discussed at Tumor Boards are not lucky. They either did not receive the correct treatment (elsewhere), or they did not respond to that treatment (or to subsequent treatments). The options discussed for these patients were poor, including very disfiguring surgery.

I mentioned standard of care. That means the currently accepted best treatment for each specific cancer type. These are based on clinical evidence and are published in several forms. One is the National Comprehensive Cancer Network (NCCN) Guidelines.

Believe it or not, not all cancer patients receive the standard of care, typically due to oncologists who have not kept up with the current state of the field. Academic cancer centers like DFCI see many referred patients whose initial treatment was suboptimal, leaving patients with poorer options.

As I've written before, if you are diagnosed with cancer and offered a treatment plan: get a second opinion! My bias is that you get that second opinion at an academic cancer center. One way to to find one of those is to look at National Cancer Institute Comprehensive Cancer Centers.

If the treatment plan proposed by the second opinion team (and your plan should be proposed after you have been seen by a multi-disciplinary team) agrees with the plan proposed by your local oncologist then by all means get treated locally. If the first and second opinions do not agree, either assess them for yourself using the guidelines above or seek a third opinion and go with the majority.

The choice you make for primary treatment will limit subsequent choices you will have - don't be a passive patient and just follow what the first oncologist recommends.