How an evening conversation on the Acela train to Baltimore and almost a decade of research are reshaping the science of prostate cancer.
On an otherwise nondescript evening in 2005, a team of three investigators huddled in the quiet car of an Acela train headed back to Baltimore. They were on the trail of an elusive foe, an important clue in the deaths of countless men across the world. These were not police detectives, however, but prostate cancer researchers from Johns Hopkins.
On this night, their muffled discussion veered in a novel direction that seemed particularly promising. Soon, whispers rose to an excited chatter about a new hypothesis. Because the quiet car is an inviolable sanctum for business travelers, it wasn’t long before a conductor emerged in the half-light to squelch the spirited discussion.
Eight years later the outburst on the Acela would lead to a notable discovery in prostate cancer research: a prototype test to predict aggressiveness of the disease.
The resulting study, published in October 2013 in the journal Cancer Discovery, would implicate otherwise-harmless DNA fillers known as telomeres as key factors in predicting prostate cancer mortality in men with this cancer.
A Tale of Two Outcomes
From a public health perspective, the costs of prostate cancer are profound.
Aside from nonmelanoma skin cancer, prostate cancer is the most common cancer among men in the U.S., according to the CDC, and one of the leading causes of cancer death among men. It accounted for 28,560 deaths nationwide in 2010. One in seven American men will be diagnosed with prostate cancer in their lifetimes.
Every one of those diagnoses will hit like a sledgehammer.
A typical course of action involves surgery. In the past, prostate cancer surgery was bloody and complex. Pioneering work by Patrick C. Walsh, MD, at Johns Hopkins revolutionized this surgery, substantially reducing complications. However despite these improvements, and the advent of less invasive laparoscopic and robotically assisted techniques, potential side effects still include incontinence and impotence. A small number of surgeries result in patient death.
“These are serious quality-of-life issues. Prostatectomy is not something you want to go through if you don’t have to,” says Alan Meeker, PhD, an associate professor of Pathology at the Johns Hopkins School of Medicine.
Meeker and Elizabeth Platz, ScD, MPH, a professor of Epidemiology at the Bloomberg School, were the lead investigators on a perplexing cold case that had stumped medical professionals for years: Two men can have prostate cancers that look identical even to expert pathologists, but result in wildly divergent outcomes.
Pathologists review prostate cancers for their characteristics including size (volume) and the Gleason sum, which is based on the degree to which the cancer cells differ in appearance and growth pattern from normal cells. A treating physician can then consider these indicators in combination with the man’s age and the stage of his cancer along with the PSA test, which measures prostate-specific antigen (PSA) in the bloodstream, to estimate the man’s prognosis.
“Take two patients who are the same on paper. Their Gleason score is the same. The volume of the tumor is the same. They’re the same age. They have the same PSA level. And yet, one patient will be dead in seven years, and the other will live two decades and die of old age,” Meeker says flatly.
Balanced against this is the reality that most prostate cancers grow slowly. Current research shows that many men live with it for years and later die of something else.
“Most patients, understandably, want the cancer out. Recent data suggest, however, that for some men surgery can wait or is not needed at all. The trick is to be able to know which men,” says Angelo De Marzo, MD, PhD, a professor and prostate cancer pathologist at the Johns Hopkins School of Medicine and the third investigator on the Acela that night.
From the outset, the trio was looking to address shortcomings of the currently used prognostic tools by finding a biological marker for the severity of this cancer. Like a fingerprint to a homicide detective, such a biomarker would help identify prostate cancer patients with aggressive cancers who require immediate removal of the prostate and those suffering less threatening forms of the disease who might forgo treatment in favor of disease surveillance.
Eventually, the team hopes to apply the test they are developing to men whose cancers are detected at a very early stage by screening. “Some of these guys may have a cancer that probably shouldn’t even be picked up, but if it is, now they face the difficult decision about whether to be treated, and if so, the type of treatment,” Meeker says.
“Tissue-based markers like this have the potential to greatly improve treatment and decision-making in men with prostate cancer. This study is an important advance,” says Jerry W. Shay, PhD, an expert in telomeres and cancer at the University of Texas Southwestern Medical Center. Dr. Shay was not involved in the research but penned a commentary accompanying the study in Cancer Discovery.
Telomeres cap the ends of human chromosomes, like the plastic aglets at the end of a pair of shoelaces. They are part of the DNA structure, but their role is protective; they prevent the loss of genes during cell replication and keep the ends of chromosomes from fraying.
In healthy cells, each time a cell divides, small bits of DNA naturally get lost from the ends of the chromosomes. Fortunately, those ends are the telomeres, which do not contain genes. As cells replicate over the course of their lives, telomeres naturally grow shorter, bit by bit. Oxidation from smoking and other causes can also lead to shortened telomeres.
Whether it’s due to aging or oxidation, telomere shortening plays a key role in genetic instability—whole chromosomes spliced end-to-end, cells missing entire chromosomes and others with chromosomes severed at odd places where mother and daughter cells engaged in a tug-of-war over a fused chromosome. Such abnormalities are at the heart of many cancers, not just in the prostate.
Human cells have built-in mechanisms to fight cancer. Some shut down their own reproductive ability as they age, a process called senescence. Others kill themselves when they grow too old, a form of genetic suicide known as apoptosis. Malignancies somehow bypass these protective mechanisms, allowing genetically damaged cells to replicate until things go awry.
In prostate cancer, extensive telomere shortening had been noticed in comparisons with nearby healthy cells. Meeker, De Marzo and Platz were aware of this back in 2005, but the prostate cancer field as a whole had largely ignored telomeres as a potential biomarker. The team’s hypothesis, hatched on the train, was simple: If shortened telomeres play a role in cancer, tumors with the shortest telomeres would be the most chromosomally unstable, and thus the most aggressive cancers.
“We had been working on telomeres for years for different reasons and we thought we should be testing whether telomere length could tell us more about the prognosis of prostate cancer,” Platz explains.
At the time, the team lacked a way to measure the length of telomeres in individual prostate cancer cells, in nearby noncancerous cells including stromal cells, and in healthy appearing epithelial and stromal cells far removed from the tumor.
These three cell compartments make up what is now understood as the cancer ecosystem. Once controversial in the cancer research community because it incorporated otherwise normal-appearing cells outside the tumor, this view has gained wide acceptance. The ecosystem approach—a fundamentally new way of looking at cancer—has opened intriguing lines of scientific inquiry.
Meeker, a telomere biologist by training, developed the method for telomere measuring with De Marzo. Adapting a well-known fluorescent staining technique known as FISH, he tags specific sections of DNA—in this case the telomeres—with fluorescent labels. With proper lighting, the stained telomeres fluoresce bright red when viewed under a microscope. The ember-like glow is an unmistakable indication of their presence and, more importantly, their length. Longer telomeres naturally attract more fluorescent label. Measuring telomeres, then, is as “easy” as measuring the intensity of the glow—brighter glow, longer telomere and vice versa.
The real beauty of the method, however, is not that Meeker can stain telomeres, nor even that the stain could be quantified; these features were well known to science. The big advance is being able to measure telomere length, cell by cell, in patient tissue samples. Earlier systems were able only to resolve broad average telomere length from a mash of ground up cancer and non-cancer cells—somewhat helpful but hardly revolutionary.
Cell-specific resolution allows the team to measure telomere length by type of cell. They can measure telomeres not only in cancer cells, but also in stromal and epithelial cells, the nearby, noncancerous cells—the all-important ecosystem. This capability was vital to the study’s success.
At this point, scientifically speaking, the research team was halfway home. They had a way to measure the telomeres. Now they needed telomeres to measure. Fortunately, Platz knew of the perfect resource that would allow them to test their hypothesis.
Harvard University, where Platz trained years ago, maintains a cohort known as the Health Professionals Follow-up Study, which was designed to study diet, lifestyle and chronic disease risk. Harvard identifies and stores tissues from participants in that cohort diagnosed with prostate cancer and meticulously tracks each man’s medical history over time, noting how and when they die, among many other health-related factors. The study reaches back to the mid-1980s and includes some 50,000 men. Not all of them have prostate cancer, of course, but enough do for the team to conduct a meaningful study.
Using this cohort, the team—now expanded to include their Harvard colleagues—in effect traveled back in time, studying real tissues from real patients to learn which men subsequently saw their cancers return, how many died of those cancers, exactly how long after recurrence these men died and how many died of causes other than cancer.
Out of this rich vein, their Harvard colleagues were able to identify a group of 596 men who had undergone prostatectomies, some dating back 20 years or more. The men averaged 65.3 years of age at the time of surgery.
In addition to Meeker’s cell-by-cell measuring technique, these other pieces—the population, study design and tissue—were paramount to success. The researcher’s enemies are bias and chance. Bias can take many forms, Platz says. For example, “observer bias” occurs when the data collector, aware of the hypothesis, consciously or unconsciously skews the data in favor of the objective. Chance, on the other hand, is a different sort of beast.
“Sometimes you pick the right study cohort and what look like promising results just don’t hold up,” Platz says. “My job as an epidemiologist is to preempt and root out the bias and chance.”
Vital as the cohort’s tissue samples and measuring technique were, they hardly ensured the team’s success. The tissue samples weren’t yet in a format suitable for efficient measurement, nor had the measurement technique ever been used before on this scale and for this purpose. The going proved slow. In 2005, each step of the telomere imaging and data collection had to be done manually.
“We would stain the samples, and the pathology research fellow would identify the cell types and circle each one by hand on an image displayed on a touch screen. Then the computer could count the pixels in each sample to get the intensity of the fluorescence,” Meeker explains. “It was meticulous and painstaking work.”
After a year of measuring telomeres in 40,000 or so cells, the team completed the exhausting data collection—only to discover their hypothesis, apparently, was dead wrong.
“We thought shorter telomeres in the cancer cells would be associated with poor outcome, but that’s not what the data showed,” Platz says. “There was no association! None. Zero.”
All that time, all that work, seemingly down the drain. “We were very disappointed,” she recalls.
Still, the scientists persevered. The cell-by-cell resolution of Meeker’s technique allowed them to look beyond the cancer cells into other parts of the tumor microenvironment.
When they scrutinized the non-cancerous stroma, they were surprised to find that shortness in telomeres was strongly associated with risk of progression and death from prostate cancer.
During a follow-up brainstorming session in Platz’s office, Christopher Heaphy, PhD, (then a postdoc and soon-to-be first author of the paper published from this work), remarked about how genomic variability in nature is critical to species adaptation, helping species to endure disease, famine, drought and more.
“It was an ‘A-ha moment,’” recalls Platz. “The genetic variability might possibly help a population of cancer cells in the same way it helps a population of plants and animals in the wild. When a new threat arises, some cells die, but others survive and are stronger for it, helping the line carry on.”
Having changed tracks, they were suddenly back on track. When Platz segmented the men based on variability in telomere length from cancer cell to cancer cell, a clear pattern emerged. Those whose prostate cancers had the greatest variability in telomere length among the cancer cells were the most likely to progress and die of their disease.
All of the men had abnormally short cancer cell telomeres, but those with the worst outcomes were those with more cell-to-cell variability in telomere length among the cancer cells. Perhaps in cancer, like in other biological systems, genetic variability increases adaptability and the likelihood that the population survives and evolves new abilities. These can include aggressive behaviors like the capability to invade surrounding blood vessels and spread to other areas of the body—the lethal process known as metastasis.
During their next meeting, De Marzo suggested looking at telomere length variability in the cancer together with telomere length in the stroma.
“That’s when this phenomenal result appeared,” Platz says.
Greater telomere length variability in the cancer, when paired with shorter telomeres in the stroma, is the very worst combination to have.
At the other end of the spectrum, less variable telomeres in the cancer combined with longer telomeres in the stroma is the best type of cancer. “If one can call any cancer good,” she hedges.
Men with the more-variable/shorter combination had eight times the risk of progressing and 14 times the risk of dying of their prostate cancer. Of the men with the less-variable/longer combination, one died of prostate cancer when, statistically, about six deaths would be expected. That patient lived more than 16 years after first diagnosis—twice as long as men with the more variable/shorter combination.
The numbers were dramatic for what the team now refers to as the “telomere biomarker,” showing clear evidence that it has the potential to distinguish between men in need of aggressive treatment, and men who can forgo treatment altogether.
“Among all of their findings, the most stunning outcome of the team’s research was the clue as to how the cancer microenvironment predicts the cancer’s behavior,” says William Nelson, MD, PhD, director of the Sidney Kimmel Comprehensive Cancer Center at Hopkins, who did not take part in the research. “The cancer-associated stroma is a co-conspirator in the crime. Shortened telomeres are simply an indication of damage. We now have a new place to look to intercept the process of cancer,” Nelson says.
Down the Line
Though the Platz-De Marzo-Meeker team has ascertained the suspect, more sleuthing remains.
Their first priority: validating the findings. Eventually, an important goal will be to establish thresholds for the telomere biomarker by which to evaluate various tumors to recommend surveillance over treatment. The team is currently developing technologies to automate the evaluation process—necessary if a widely used clinical test is to materialize.
“This is a provocative study that adds to our predictive capability and could result in significantly less-aggressive therapy for men at low risk,” says cancer researcher Shay.
Platz reflects on the sheer scale of team effort necessary to make this discovery. The study included investigators from two major institutions working in several disciplines—biological, clinical and population sciences. It also took essential resources in the form of many federal grants, available participant data, and tissue and other existing infrastructure.
Not least, it took time. Almost a decade passed from the day on the Acela to final publication.
The team now has momentum, and is emboldened by broader implications of their discovery. Their findings about telomeres and prostate cancer prognosis may also apply to other cancers.
The economics of their discovery is striking. Fewer unneeded prostatectomies could result in big health care savings. At an average cost of $19,214, prostatectomies cost some $2.6 billion annually, according to the CDC.
“From the perspective of optimizing population health and best using limited health care dollars toward that goal, the development of better prognostic tools for the most common cancers in men is imperative,” Platz says.