Medicine and evolution, Part 2: Applying evolutionary principles to cancer

Last week, I inaugurated a new series on this blog entitled Medicine and Evolution. I even wrote what was to be the second post in the series, a post that (I hoped) would illustrate the utility of applying approaches used to study evolution to human disease. That post is essentially complete, other than requiring the addition of some links. That's what I was going to do last night, until Stranger Fruit turned me on to this study:

In a study published online today in Nature Genetics, Carlo Maley, Ph.D., a researcher at The Wistar Institute, and his colleagues report that precancerous tumors containing a population of highly diverse cells were more likely to evolve into cancer than those containing genetically similar cells. The finding suggests that, in at least some forms of cancer, the more genetically diverse a precancerous tumor is, the more likely that tumor is to progress to full-blown cancer. If so, genetic diversity might act as a biomarker for cancer risk among patients with precancerous tissues.

"Although researchers first defined cancer in evolutionary terms in the 1970s, few researchers have actually studied the disease this way," says Maley, lead author on the study and an assistant professor in the molecular and cellular oncogenesis program at Wistar. "We wanted to know: If we measured a precancerous tumor's genetic diversity at baseline, could we predict who would go on to get cancer?"

To find out, the scientists decided to analyze data on a precancerous condition called Barrett's esophagus, in which cells lining the lower esophagus change due to repeated exposure to stomach acid from reflux, a condition often referred to as heartburn. Doctors typically adopt a "wait and watch" approach to treating patients with Barrett's esophagus because the condition only rarely leads to cancer and is difficult to treat surgically.

In the study, Maley and colleagues analyzed precancerous tumor data from 268 patients, including multiple biopsies within each tumor. On average, these patients were followed for 4.4 years, during which time 37 developed cancerous tumors. Overall, the database used in the study represents more than 32,000 distinct genotypes of different cells within the tumors.

Using computational techniques to analyze the data, the researchers calculated measures of diversity inside the tumors. Essentially, they counted cell varieties and measured the genetic difference, or divergence, between those varieties. "Simply put, we took ecology measures of species diversity and translated them into measures of cell diversity within tumors," Maley says. The found a striking correlation between increased diversity of tumor cells and progression to cancer. For every additional cell variety detected in a tumor, the patient was twice as likely to progress to cancer.

Maley suggests that genetically diverse tumors have a high probability of generating mutant cells that will flourish and spread, allowing the tumor to transform and grow. In the future, in addition to serving as a biomarker for cancer risk, he adds, measures of genetic diversity might help doctors assess the success of cancer prevention therapies.

In fact, he speculates, genetic diversity among tumor cells might help explain why therapy sometimes fails. If a tumor contains a diverse population of cells, some of those cells are more likely to resist treatment, Maley says. Adapting to and surviving chemotherapy, these resistant cells could breed, leading to a cancer relapse. He hopes to pursue this hypothesis in the future. "More immediately," he adds, "we intend to validate the new study with other cohorts and other types of tumors."

Woo hoo! Mere days after inaugurating this series, I'm made aware of an example from my area of interest and expertise, cancer! Granted, I haven't taken care of esophageal cancer since my training, but the principles in this paper could be applied to any cancer.

My first reaction, though, was that this concept isn't exactly new news. Although it's long been postulated that cancers derive from a single malignant cell, we've known for a long time that malignant tumors are genetically unstable, and their high mutation rates produce multiple clonal populations of cells in a single tumor. For example, not all cells in a tumor are capable of metastasis; it's usually a small subpopulation that acquires the ability to metastasize. True, there is controversy over the question of whether the ability to metastasize is there from very early on during carcinogenesis or whether it develops later, but it is clear that it is not there from the very beginning. Cancerous cells acquire the ability to metastasize sometime during their progression from early dysplasia to frank cancer. Also, the genetic instability of tumor cells is also one of the driving forces leading to resistance to chemotherapy, as it provides genetic diversity upon which the selective pressure of various therapies can operate. Chemotherapy, even highly effective chemotherapy, does not kill all the cells in a tumor. The problem facing oncologists since chemotherapy was first developed is that, even if an agent kills 99.9% of the cells, in a tumor with billions of cells that will still leave millions of cells still viable. Naturally, those cells tend to be the cells that are either inaccessible to the chemotherapy or the most resistant to the chemotherapeutic agent. These resistant cells can then proliferate adn repopulate, and the next time around chemotherapy will kill fewer of them. It's a great example of natural selection at the cellular level.

So I had to look up the full article, which can be found here. I rapidly changed my tune. What Maley et al at the Wistar Institute have done is something that I've never seen in a medical journal before. They adapted diversity measures from ecology and evolution for use in quantifying the clonal diversity in a premalignant condition using actual tissue samples from human biopsy specimens.

Before I continue, a few words about Barrett's esophagus from a surgeon's perspective are in order. I have to point out that I wouldn't necessarily agree with the characterization that Barrett's esophagus "only rarely leads to cancer." In actuality, the risk is 30-125 times that of the normal population without Barrett's, which translates to a 0.4% to 0.5% chance per year of developing adenocarcinoma of the esophagus, a form of cancer that is much less common than the usual type of esophageal cancer, squamous cell carcinoma. If you develop Barrett's from gastroesophageal reflux disease (GERD) when you're in your 40's, for example, during your expected additional life expectancy of another 30-40 years, you have a 15-20% chance of developing a highly deadly malignancy. True, we can surgically remove the esophagus to prevent the cancer. However, the necessary surgery (esophagectomy) is not generally not recommended for this purpose because it's a big operation with a significant complication rate that produces a permanent change in gastrointestinal function, due to the fact that the replacement for the resected esophagus is crafted by pulling the stomach up into the chest and connecting it with the cervical esophagus in the neck. It's hard to justify doing it on 100 people to prevent maybe 15 cancers. There are times when we will do surgery for Barretts, specifically if high grade dysplasia (precancerous cellular change) is seen, but that's because high grade dysplasia portends a high risk of impending esophageal cancer or is a sign that it's already there and you just missed it through sampling error in your biopsy specimen.

Over the years, doctors have been looking for characteristics that distinguish Barrett's esophagus that will develop into esophageal cancer from Barrett's that won't, so that surgery could be offered to the patients at highest risk for progressing to cancer. Unfortunately, we have yet to find reliable prognostic factors based on tumor or patient characteristics that can be used to distinguish those who would benefit from surgery and those who don't need it. Consequently, we are indeed forced to take a "wait and watch" approach and subject the patient to frequent endoscopies.

Now, using the principles of evolution, Maley et al have found one potential indicator of which patients with Barrett's esophagus will progress to cancer and which will not. Basically, they adapted a diversity measure from ecology and evolution known as the Shannon diversity index. I'm going to have to leave it to my evolutionary biology colleagues to tell me more whether this was appropriately done, but for purposes of this paper the authors treated each sample ot as a single organism but as thousands of cells. Diversity was measured as the number of distinct clones of cells within the specimen. Using various measures of genetic diversity, including loss of heterozygosity (LOH), microsatellite instability, and differences in chromatin content and telomere length, the number of distinct clones in each patient was estimated and the Shannon index calculated. These were then correlated with the occurrence of cancer. The investigators also controlled for established genetic risk factors, such as loss or mutation of the p53 tumor suppressor gene.

Diversity measures were calculated in a baseline cohort of 268 individuals with Barrett's esophagus using a systematic biopsy sampling method and identified 32,000 genotypes. These patients were followed for a median of 4.4 years, during which time 37 of these patients (13.8%) developed invasive esophageal cancer (thus further reinforcing that cancer is not rare in patients with Barrett's). The development of adenocarcinoma correlated strongly with the upper quartile of the number of clones, genetic divergence, and the Shannon index. Few of the patients in the lower three quartiles developed cancer, whereas up to 50% of the patients in the upper quartile did. They also developed a model where various measures from this study could be used to estimate relative risk of developing cancer.

As a scientist, I find this fascinating because this study provides compelling evidence that principles used in evolution can be used to describe tumor cell population. As Evolving Thoughts put it:

The logic of evolution (which in this case means both changes in allele frequency and also diversification, that is, a kind of cell phylogeny) is that the higher the viable variation, the more likely a viable further variant is. Since cell lineages are rather like asexual lineages of, say, bacteria, the same models ought to work.

Apart from the intellectual satisfaction of seeing how selective mechanisms can be applied to describing tumor behavior, the physician in me is itching to see how this could be applied to patient care. For example, could a test be developed in which we could take biopsy specimens from patients with Barrett's esophagus, measure the number of distinct clones of cells in the specimen, and then use that number to identify high risk patients to whom esophagectomy should be offered? Could we use this observation to develop treatment strategies less likely to result in resistance? As the authors themselves concluded:

The generation of cellular genetic variants and clonal diversity on which natural selection acts may be a fundamental evolutionary mechanism of neoplastic progression with profound clinical implications. Our diversity measures are based on one-fourth the number of biopsies normally used for pathological grading. If confirmed in other Barrett's esophagus cohorts and other neoplasms, then neoplasms may be considered as evolving ecosystems in which evolutionary and ecological measures of diversity are widely applicable for assessment of risk of progression to cancer because they quantify genetic heterogeneity within viable, evolving clones; integrate all mechanisms generating genomic instability and are easily generalized to other premalignant neoplasmos, as well as being readily scaled from single cells (as in FISH studies) to entire neoplasms. Clinically, assessment of clonal diversity may be a unified method to identify high-risk patients for early detection as well as warning of possible variants that may be resistant to cancer prevention interventions.

Another question that this study raises is exactly what the selective pressures are that contribute to the development of cancer from precancerous lesions. Considering these questions leads to testable hypotheses about what the mechanisms may be, such as independence from antiproliferative and antiapoptotic signals or the ability to evade the immune system. "Intelligent design" certainly can't say as much with regard to human disease.

Of course ID adherents will point out that there's a difference between applying evolutionary principles to a cluster of premalignant cells and applying them to whole organisms. That may be true, but we should remember here that these investigators took evolutionary principles originally applied to populations of organisms and then applied them to premalignant cell clusters, showing that evolutionary principles can be utilized even in this case. To borrow terms from the ID crowd, this is taking macroevolution and applying it to microevolution, not the other way around.

NEXT UP (later this week): How evolution can lead to the identification of previously unsuspected genes causing disease.