Harold Varmus left his studies of English literature to become a doctor. But fresh out of medical school he began doing research. His career took an unexpected turn during a backpacking trip to California when he happened to meet J. Michael Bishop, a biologist at the University of California, San Francisco. Soon after, in 1970, Varmus joined Bishop’s lab. The two began studying viruses that can trigger tumors, to better understand the genetic roots of cancer. That collaboration would earn them the 1989 Nobel Prize in Physiology or Medicine for their discovery of the origin of oncogenes, or genes that can cause cancer. Varmus would go on to lead the National Institutes of Health and the National Cancer Institute.

Over his almost 50-year career, Varmus has seen the field of cancer genetics move from the stuff of laboratory experiments to become the cornerstone of cancer treatment. But there’s much left to learn and, as Varmus points out, the field’s most transformative discoveries still mostly benefit only the fraction of patients treated at academic medical centers.

Speaking with Knowable Magazine this fall, Varmus discussed his essay published in the 2017 Annual Review of Cancer Biology. This interview has been edited for length and clarity. (Varmus's remarks are in regular type. Transitional material by Laura Beil is in boldface.)

Rather than reflect on his own career — a typical “intellectual autobiography,” as he called it, an area he had already explored at length in his 2009 book “The Art and Politics of Science” — he chose to look more broadly at the evolution of cancer research.

I had already written a personal account — a memoir. I went into a lot of [personal] stuff there: What I saw when I was a trainee at the NIH, and doing science for the first time at the advanced age of 28 or 29, was that some new technology, maybe the technology of molecular biology, was going to be applicable to try to understand the genetic basis of cancer.

I began to study cancer, not through traditional ways, but by using animal viruses that cause cancers, trying to understand the genetic origins of virus-induced cancer. That morphed into a different field as additional technologies came along that allowed us to examine the genetics of the cell with the same kind of scrutiny and detail that was possible for working with genetically much simpler organisms, namely viruses.

The essay that I wrote for Annual Reviews explored how tumor virology became cancer biology and became much more focused on cells and animals, with a diminished role for cancer viruses themselves.

What I’ve tried to show is that there has been remarkable convergence of the goals of oncologists who were trying to understand and treat cancer with the objectives of people who are trying to understand fundamental mechanisms by which normal cells become cancers.

He spoke of one major barrier to modern cancer treatment, which is cost. One new leukemia drug approved in late August has a price tag of $475,000 per patient.

I think there are more fundamental things that are much more widespread than these situations in which extraordinary costs are being levied for a few patients. I mean, it is an issue, I’m not denying that, but one of the things that I am very concerned about is the lack of access on many, many occasions to even the most fundamental products of cancer treatment.

Right now only a distinct minority of patients have any genetic evaluation of their tumors. I certainly can’t imagine any scientist contracting cancer and not wanting to know immediately which genes have mutated and what that might portend for the use of either targeted therapy or immunotherapy.

I’m not talking about the half-million-dollar therapeutics, which a very small number of patients are currently asking for. What concerns me is that we don’t have routine reimbursement for a test as simple as either a panel test or a whole exome analysis of a primary tumor. [Panel tests and exome analyses are ways of doing targeted searches for cancer-causing mutations in a tumor, rather than trying to sequence its entire genome, saving time and expense.]

Insurers, especially government insurers, need to be appealed to, to try to get them on board to reimburse for the cost of a several-hundred or a thousand-dollar test. They’re all very eager and willing to support numerous radiological assessments that frequently are not even that useful and done too often. But reimbursement for genetic testing of cancers, which is something that really ought to be a routine procedure, is impeding the ability of someone who’s not at an academic health center — as 85 percent of cancer patients are not — to get access to what is a starting point in using the modern tools of cancer therapy.

Moreover, we’re generating a huge database of information that can be correlated with clinical outcomes in a way that will be instructive for all oncologists. [Creating a national database ecosystem was one of the priorities identified for the “Cancer Moonshot” at the National Cancer Institute.]

He envisions one possible way to increase access.

One solution, something I’ve just written about and am hoping to get published soon, is to get the Centers for Medicare and Medicaid Services, CMS, to provide payment for these tests under an existing mechanism that involves coverage for evidence determination. [This allows for coverage for the purpose of generating data to better evaluate a test or therapy.]

More than 59 million Americans live in rural areas, yet more than 94 percent of oncologists are concentrated in cities, according to a 2015 report by the American Society of Clinical Oncology.

The whole notion here is that if Medicare and Medicaid covered these services, the insurers will come along and they’ll have enough experience to ascertain the value of these tests that have been done before. If there is coverage for evidence development, there will be a huge amount of data accumulated and we will be able to make some judgments about appropriate testing. Oncologists will become accustomed to doing this routinely.

Only a very small fraction of patients are getting their tumors tested for the genetic lesions that are, in a sense, one of the great products of the last 30 or 40 years of cancer research. We’re just not taking advantage of that.

Varmus spoke about metastatic cancer, which remains one of the greatest challenges in cancer treatment.

We’re much more effective at removing primary lesions and just getting rid of the cancer that way. But I wouldn’t take a totally pessimistic view, because there are cancers that have spread that we do treat effectively.

There are many patients who have now responded to some of the new immunotherapies very effectively. And in a sense, all hematological cancers, all leukemias, are metastatic. We treat chronic myeloid leukemia very effectively and many lymphomas that are basically metastatic.

I think what has been particularly difficult is understanding the rules of metastasis. Why do certain cancers metastasize and others do not? What’s the relationship between the metastasis itself — that is the cancer cells — and the place in which it lands? I think all of us have recognized that by focusing on primary cancers we’re doing something that’s smart scientifically because it’s a manageable place to start to understand how cancer arises. But we also are failing to grapple with the part of the cancerous process that’s most likely to kill people.

[Another question is:] What are the steps in the metastatic process that require a special force … getting out of the primary site or in the blood vessels or landing in foreign territory and prospering there? The work that my own lab has done in this area has mainly shown that the cells can do many of these things without having any discernibly special properties. It really is not so clear what makes a cell that has lodged in a foreign place able to prosper and cause the death of the organism.

In part, [the emphasis on primary tumors has arisen] because the models for studying metastasis have not been that good. I think they’ve been improved recently, but you can’t really study metastasis unless you’re working in animals or with samples from human beings.

On the other end of the spectrum, Varmus talked about how embryonic cells may provide more insight into the origins of cancer, including a greater ability to look at the interplay of genetics, chance and environmental factors.

I do think — and my own recent lab work manifests this belief — that we need to learn more about how cancer initiates. We focus so much of our work on fibroblast and certain hematopoietic [blood] cells. [Embryonic cells] give us an opportunity to study the initiation of cancer in a wide variety of cells that wasn’t possible years ago. I think this is a new era which is very exciting with respect to cancer initiation. One can now look at the complete repertoire of genetic changes in the cell, begin to track evolution of cancers, begin to study the initiation of cancers by growing cell types of defined character in culture.

My own lab now is working on a number of things, for example the ability to take human embryonic stem cells and differentiate them into a wide variety of cell types. Others are taking advantage of technologies that allow growth of human tissues and the development of tissues characteristic of adult organs, so-called organoids.

All of these new methodologies are allowing us to study cancer in a much more sophisticated way than was possible say 20 or 30 or 40 years ago when almost all cancer research was done with fibroblasts or with bone marrow. Now you can grow colonic epithelium or lung epithelium — human cells — and introduce genes or control them with a variety of technologies that have evolved over the last 30 or 40 years to do things that were simply inconceivable in the early days.