Cancer Explained – The Role of Cell Death

Following a recent blog, I received an inquiry from one of our readers. The individual asked whether I could better explain my oft repeated statement that “cancer doesn’t grow too much, it dies too little.” The questioner was puzzled by my assertion that chemotherapy drugs acted to stop cells from growing, while she had come to believe that this was synonymous with killing them. This dichotomy is at the crux of our modern understanding of cancer.

In response, I would like to examine the very basis of what is known as carcinogenesis, the process by which cancer comes to exist.

For more than a century, scientists believed that cancer cells were growing more rapidly than normal cells. They based this on serial measurements of patient’s tumors, which revealed that tumor dimensions increased. A small lump in the breast measuring one-half inch in diameter would be found six months later to be one inch in diameter. And six months after that it was two inches in diameter. This was growth, plain and simple, and so it was reasoned that cancer cells must be growing too much. As such, cancer therapies, per force of necessity, would need to stop cancer cells from growing if they were to work at all.

And then, in 1972, a paper was published in the British Journal of Cancer that described the phenomenon of apoptosis, a form of programmed cell death. Although it would be almost a decade before cancer researchers fully grasped the implications of this paper, it represented a sea change in our understanding of human tumor biology.

Let’s use the example of a simple mathematical equation. Every child would recognize the principles of the following formula:
Tumor mass = growth rate – death rate
This simple equation represents the principle of modern cancer biology. Where cancer researchers went wrong was that they mistakenly posited that the only way a tumor mass could increase was through an increase in the growth rate. However, as any child will tell you, a negative of a negative is a positive. That is, at a given growth rate, the tumor mass can also increase if you reduce the death rate. Thus, the “growth” so obvious to earlier investigators did not reflect an increase in proliferation but instead a decrease in cell attrition. Cancer didn’t grow too much it died too little, but the end result was exactly the same.

It should now be abundantly clear exactly why chemotherapy drugs, designed to stop cells from growing, didn’t work. Yes, the drugs stopped cells from growing, and yes any population of “growing cells” would suffer the effect. But they didn’t cure cancers because the cancers weren’t growing particularly fast. Indeed, the fact that chemotherapy works at all is almost an accident. Contrary to our long held belief that we were inhibiting cell proliferation, chemotherapy drugs designed to damage DNA and disrupt mitosis, were actually working (when they did at all) by forcing the cells to take inventory and decide whether they could continue to survive. If the injury were too extreme, the cells would commit suicide through the process of cell death. If the cells were not severely damaged or could repair the damage, then they carried on to fight another day. None of this, however, had anything to do with cell growth.

Chemosensitivity Testing: Lessons Learned

Like all physicians and scientists engaged in the study of cancer biology and cancer treatment, I had accepted that cancer was a disease of abnormal cell growth. I remember reading the lead article in the New England Journal of Medicine (NEJM) that described the clonogenic assay (Salmon, S. E., Hamburger, A. W., Soehnlen, B. S., et al. 1978. Quantitation of differential sensitivity of human tumor stem cells to anticancer drugs. N Engl J Med 298:1321–1327).

I sat in a laboratory at Georgetown University reading about a lab test that could accurately predict the outcome of cancer patients, without first having to give patients toxic drugs. It seemed so logical, so elegant, so inherently attractive. Sitting there as a medical student, far removed from my formal cancer training, I thought to myself, this is a direction that I would like to pursue.

But I was only a first year student and there were miles to go before I would treat cancer patients. Nonetheless, selecting drugs based on a laboratory assay was something I definitely wanted to do. At the time I had no idea just how difficult that could prove to be.

After medical school I found myself in California. There I met an investigator from the National Cancer Institute who had recently joined the faculty at the University of California, Irvine. He too had read the NEJM paper. Being several years ahead of me in training he had applied the clonogenic technique at his laboratory at the National Cancer Institute. Upon his arrival in California, he had continued his work with the clonogenic assay.

All was going along swimmingly until the NEJM published their report documenting the results of five years experience with the clonogenic assay.  It wasn’t a good report card. In fact the clonogenic assay got an “F.”

Despite the enthusiastic reception that the assay had previously enjoyed, the hundreds of investigators around the world who had adopted it and the indefatigable defense of its merits by leading scientists, it seemed that something was very wrong with the clonogenic assay and I desperately needed to know what that was.

It so happens that in parallel to clonogenic assays, my colleague was working on a simpler, faster way to measure drug effects. Using the appearance of cells under the microscope and their staining characteristics, one could skip the weeks of growth in tissue culture and jump right to the finish line. The simple question to be answered was: Did the drugs and combinations kill cancer cells in the test tube? And if they did kill cancer cells in the test tube, would those drugs work in the patient? The answer was, “YES!”

Despite the clonogenic assay’s supporters, it turned out that killing cancer cells outright in the test tube was a much, much better way to predict patient’s outcomes. It would be years before I understood the depth of this seemingly simple observation and the historical implications it would have for cancer therapy.

In Chapter 7 of my soon-to-be-released book, Outliving Cancer I examine the impact of programmed cell death on human biology.

What is Cancer?

This is a question that has vexed scientific investigators for  centuries, and for the last century, our belief was predicated upon physical observation that cancer reflected altered  cell growth. After all, to the untrained eye, or even to the rather sophisticated eye, the mass in the pelvis or the lymph node under the arm, or the abnormality on a chest x-ray, continued to expand upon serial observation. This was “growth” (at least since the time of Rudolph Virchow); and growth it was reasoned represented cell division.

Based upon the cell growth model, cancer therapists devised drugs and treatments that would stanch cellular proliferation. If cells were growing, then cells needed to reproduce the genetic elements found in chromosomes leading to the duplication of the cell through mitosis. If chromosomes were made of DNA, then DNA would be the target of therapy. From radiation to cytotoxic chemotherapy, one mantra rang through the halls of academia, “Stop cancer cells from dividing and you stop cancer.”

As in many scientific disciplines, nothing spoils a lovely theory more than a little fact. And, the fact turned out to be that cancer does not grow too much, it dies too little. Cancer doesn’t “grow” its way into becoming a measurable tumor, it “accumulates” its way to that end.

In 1972, we realized that the most basic understanding of cancer biology up to that point was absolutely, positively wrong.

Working in a laboratory during my fellowships, I began to realize that something was wrong with the principles that guided cancer therapeutics. My first inkling came from the rather poor outcomes that many of my patients experienced despite high-dose, aggressive drug combinations.

Then, it was the failure of the clonogenic assay to predict clinical outcomes that further raised my suspicions. I began to ponder cell growth – cell death, cell growth – cell death. With each passing day the laboratory analysis that I conducted identified active treatments that worked.  Using short-term measures of cell death (not cell growth),. I could predict which of my patients would get better.  All of the complicated and inefficient clonogenic assay investigations could not. Cell growth – cell death – what was I missing?

It would be years before I would attend a special symposium on the topic of cell death that it all became abundantly clear.

My “eureka” moment is captured in Chapter 6 of my soon-to-be-released book, Outliving Cancer.

Cancer and the Great Divide

There are two types of cancer patients: those we can treat and those we can’t. As I reflect on this year and the years past during which we have applied the process of laboratory-guided treatment, I am reminded of this fact.

The EVA-PCD functional profile enables us to choose active treatments for patients, but I have sometimes wondered whether we are, in fact, choosing patients for the available drugs.  While the end result may not be all that different, e.g. superior clinical outcomes over randomly administered (standard) therapies, the path to that outcome, leaves room for interesting discussion.

I first pondered this issue at the time of completion of our earliest study. That study was conducted in childhood acute lymphoblastic leukemia (ALL). Recognizing that the corticosteroids were among the most important drugs for ALL, we exposed freshly isolated lymphoblasts from ALL patients to dexamethasone (ex vivo). At the fourth day we measured the degree of cell death and separated the patients in “sensitive” and “resistant “ subgroups. Strikingly, those children whose lymphoblasts died in the laboratory following exposure to dexamethasone (ex-vivo), virtually all survived without relapse, while those children whose lymphoblasts did not die in the laboratory following dexamethasone exposure (ex-vivo) relapsed at an alarming rate with only 25 percent still alive at the sixth year of follow up (p=0.009).

What we had succeeded in doing by Day 4 of diagnosis was something that all the known prognostic factors, like age, WBC and male vs. female could not do, namely accurately identify the responders and survivors.

Today, when I test patients in our laboratory, I consistently double or even triple the response rates over standard protocols, yet a subset of patients are not found sensitive to the available therapies. Patients who do not respond to chemotherapy are today known, in the oncologic vernacular, as “failing therapy.” If we view these “non-responders” as a biologically distinct group (not unlike the dexamethasone-resistant ALL patients above) then our role, in the field of functional profiling, is to quickly segregate the responders (to available drugs) from the non-responders and move those “non-responders” immediately to something that will work for them. In this light, patients no longer “fail therapies” but instead “therapies fail patients.” It is then our mandate to use the ex-vivo platforms to find (and yes, discover) novel therapies and combinations that will meet their unmet need.

As the New Year is upon us I am filled with the expectation that 2013 will be one of discovery and innovation. Never before have so many interesting compounds been available for study. If we are fortunate enough to succeed in our efforts to collaborate with members of the drug development community and have the opportunity to intelligently apply functional profiling, for drug discovery, 2013 could be a very good year indeed.

Phar Lap and the Treatment of Leukemia

Phar Lap (1926-1932) was a thoroughbred horse bred in New Zealand. After winning the Melbourne Cup and 37 other races, his victory at the Agua Caliente racecourse in Tijuana, Mexico, established the track record in 1932.

With each victory, his detractors became more strident. He was even the target of an assassination attempt. To prevent him from winning (and thereby disrupting the betting odds) officials would add lead bricks to his saddle. On the occasion of the Melbourne cup of 1930 he carried 138 pounds of lead, yet won the race. A quote from the Sydney Morning Herald dated Wednesday, November 5, 1930, read, “The question was not which horse could win, but could Phar Lap carry the weight. Could he do what no other horse before him had done?”

It appeared that the one thing that race officialdom feared above all else, was a horse that could consistently beat the field and win the race.

The tale of Phar Lap was brought to mind after a colleague forwarded a paper published in the journal Leukemia on August 10, 2012: “The use of individualized tumor response testing in treatment selection: second randomization results from the LRF CLL4 trial and the predictive value of the test at trial entry.” (E Matutes, AG Bosanquet et al, Leukemia, Letter to the Editor.)

Published as a letter to the editor, the paper describes correlations between the TRAC (tumor response to antineoplastic compounds) assay, a short-term suspension culture cell death laboratory assay (very similar to our work) and clinical response, time to progression and overall survival in patients with chronic lymphocytic leukemia (CLL) who received chemotherapy as part of the LRF CLL4 trial conducted in England between 1999 and 2004.

The initial trial was a blinded correlation between laboratory assay results and patient response to one of three treatment regimens. An examination of the data reveals a clear and statistically significant correlation between drug sensitivity and overall survival (p = .0001). The 10-year survival of drug sensitive patients was 28 percent, while the 10-year survival for drug resistant patients was 12 percent.

Significant correlations with survival were observed for known prognostic factors like 17p and 11q deletion, as well as IGHV mutational status. Correlations were also observed between the TRAC assay results and these prognostic factors.

The report goes on to describe a second randomization that took place at the time of disease progression, either failure of first-line therapy or reoccurrence within 12 months. In this part of the study, 84 relapsed patients were allocated to standard therapy and their outcomes were compared with 84 patients allocated to treatment guided by the TRAC assay. The drugs tested in the assay-directed arm included chlorambucil, cytoxan, methylprednisolone, prednisolone, vincristine, doxorubicin, mitoxantrone, 2CDA, fludarabine and pentostatin. In vitro resistance for combinations was defined as resistance to all constituent drugs in the combination, while drug sensitivity was defined as TRAC-assay sensitivity for any of the drugs used in combination. No discussion of synergy analysis was included.

In examining this study, I cannot help but be reminded of Phar Lap. First, marshaling a study of 777 CLL patients, and conducting 544 TRAC analyses, is a phenomenal undertaking for which these authors should be commended.

Second, the observation of a significant correlation between laboratory assay results and overall survival, as well as the biological implications of this platform’s capacity to correlate with molecular markers is a demonstrable and noteworthy success, however unheralded.

Where the analogy with poor Phar Lap’s struggles, weighted down with lead, becomes most poignant is the final portion of the study wherein 84 patients received assay-directed therapy. To wit, we must remember that in 2012, drug refractory CLL remains an incurable malignancy (with the exception of a small subset of successfully transplanted patients) and that no chemotherapy-alone trial has provided a survival advantage in this group. But this only begins to explain this trial’s results.

Among the virtually insurmountable hurdles that these investigators were forced to confront was the fact that fully 52 percent of the standard treatment arm group were destined to receive fludarabine. This drug, the current gold standard for previously treated patients who fail chlorambucil (constituting 73 percent of the patients in this part of the trial), has an objective response rate of 48 – 52 percent in this population. As the drug would likely be identified as active in vitro as well, this had the impact of pitting the assay arm and the standard arm against one another, frequently using exactly the same treatment.

While this does not mean that the assay arm could not succeed, it does have an enormous impact upon the sample size calculations used to determine the number of patients required to achieve significance.  No pharmaceutical company would ever allow a registration trial to be conducted against an “unknown” control arm, particularly one using the same therapy as the study arm – not ever! Despite these burdens, the assay-directed arm had a superior one-year survival, while virtually all other trends favored the group who received assay-selected therapy. The results of this study are worthy of recognition and further support the clinical relevance, predictive validity and importance of functional analyses. Yet, this interesting study in CLL is unceremoniously relegated to the status of a Letter to the Editor in Leukemia. Perhaps, like Phar Lap, no one really wants to upset the odds.

The Tumor Micro Environment

As I was reading the October 1 issue of the Journal of Clinical Oncology, past the pages of advertisement by gene profiling companies, I came upon an article of very real interest.

While most scientists continue to focus on cancer-gene analyses, a report in this issue from a collaboration between American and European investigators provided compelling evidence for the role of tumor associated inflammatory cells in metastatic human cancer. (Asgharzadeh, S J Clin Oncol 30 (28)3525–3532 Oct 1, 2012) Through the analysis of children with metastatic neuroblastoma, they found that the degree of infiltration into the tumor environment by macrophages had a profound effect upon clinical outcome. This study confirmed earlier reports that macrophage infiltration is an integral part and potential driver of the malignant process.

Using immunohistochemistry and light microscopy the investigators scored patients for the number of CD163(+) macrophages, representing the alternatively activated (M2) subset within the tumor tissue. They then examined inflammation related gene expressions to develop a “high” risk, “low” risk algorithm and applied it to the progression free survival in these children.

Highly significant differences were observed between the two groups. This report adds to a growing body of literature that describes the interplay between cancer cells and their microenvironment. Similar studies in breast cancer, melanoma and multiple myeloma have shown that tumor cells “co-opt” their non-malignant counterparts as they drive transformation from benign to malignant, from in-situ to invasive and from localized disease to metastatic. These same forces have the potential to strongly influence cellular responses to stressors like chemotherapy and growth factor withdrawal. While we may now be on the verge of identifying these tumor attributes and characterizing their impact upon survival, these analyses represent little more than increasingly sophisticated prognostics.

The task at hand remains the elucidation of those attributes and features that characterize each patient’s tumor response to injury toward ultimate therapeutic response. To address this level of complexity, we need the guidance of more global measures of human tumor biology, measures that incorporate the dynamic interplay between tumors cells, their stroma, vasculature and the inflammatory environment.  These are the “real-time” insights that can only be achieved using human tissue in its native state. Ex vivo analyses offer these insights. Their information moves us from the realm of prognostics to one of predictives, and it is after all predictive measures that our patients are most desperately in need of today.

Systems Biology Comes of Age: Metastatic Lung Cancer in the Crosshairs

Cancer therapists have long sought mechanisms to match patients to available therapies. Current fashion revolves around DNA mutations, gene copy and rearrangements to select drugs. While every cancer patient may be as unique as their fingerprints, all of the fingerprints on file with the federal AFIS (automated fingerprint identification system) database don’t add up to a hill of genes (pun intended), if you can’t connect them to the criminal.

To continue the analogy, it doesn’t matter why the individual chose a life of crime, his upbringing, childhood traumas or personal tragedies. What matters is that you capture him in the flesh and incarcerate him (or her, to be politically correct).

The term we apply to the study of cancer, as a biological phenomenon is “systems biology.” This discipline strikes fear into the heart of molecular biologists, for it complicates their tidy algorithms and undermines the artificial linearity of their cancer pathways. We frequently allude to the catchphrase, genotype ≠ phenotype, yet it is the cancer phenotype that we must confront if we are to cure this disease.

Using a systems biology approach, we applied the ex-vivo analysis of programmed cell death (EVA-PCD®) to the study of previously untreated patients with non-small cell lung cancer. Tissue aggregates isolated from their surgical specimens were studied in their native state against drugs and signal transduction inhibitors. This methodology captures all of the interacting “systems,” as they respond to cytotoxic agents and growth factor withdrawal. The trial was powered to achieve a two-fold improvement in response.

At interim analysis, we had more than accomplished our goal. The results speak for themselves.

First: a two-fold improvement in clinical response – from the national average of 30 percent we achieved 64.5 percent (p – 0.00015).

Second: The median time to progression was improved from 6.4 to 8.5 months.

Third: And most importantly the median overall survival was improved from an average of 10 – 12 months to 21.3 months, a near doubling.

These results, from a prospective clinical trial in which previously untreated lung cancer patients were provided assay directed therapy, reflects the first real time application of systems biology to chemotherapeutics. The closest comparison for improved clinical outcome with chemotherapeutic drugs chosen from among all active agents by a molecular platform in a prospective clinical trial is . . .

Oh, that’s right there isn’t any.

Type I Error

Scientific proof is rarely proof, but instead our best approximation. Beyond death and taxes, there are few certainties in life. That is why investigators rely so heavily on statistics.

Statistical analyses enable researchers to establish “levels” of certainty. Reported as “p-values,” these metrics offer the reader levels of statistical significance indicating that a given finding is not simply the result of chance. To wit, a p-value equal to 0.1 (1 in 10) means that the findings are 90 percent likely to be true with a 10 percent error. A p-value of 0.05 (1 in 20) tells the reader that the findings are 95 percent likely to be true. While a p-value equal to 0.01 (1 in 100) tells the reader that the results are 99 percent likely to be true. For an example in real time, we are just reporting a paper in the lung cancer literature that doubled the response rate for metastatic disease compared with the national standard. The results achieved statistical significance where p = 0.00015.  That is to say, that there is only 15 chances out of 100,000 that this finding is the result of chance.

Today, many laboratories offer tests that claim to select candidates for treatment. Almost all of these laboratories are conducting gene-based analysis. While there are no good prospective studies that prove that these genomic analyses accurately predict response, this has not prevented these companies from marketing their tests aggressively. Indeed, many insurers are covering these services despite the lack of proof.

So let’s examine why these tests may encounter difficulties now and in the future. The answer to put it succinctly is Type I errors. In the statistical literature, a Type I error occurs when a premise cannot be rejected.  The statistical term for this is to reject the “null” hypothesis. Type II errors occur when the null hypothesis is falsely rejected.

Example: The scientific community is asked to test the hypothesis that Up is Down. Dedicated investigators conduct exhaustive analyses to test this provocative hypothesis but cannot refute the premise that Up is Down. They are left with no alternative but to report according to their carefully conducted studies that Up is Down.

The unsuspecting recipient of this report takes it to their physician and demands to be treated based on the finding. The physician explains that, to his best recollection, Up is not Down.  Unfazed the patient, armed with this august laboratory’s result, demands to be treated accordingly. What is wrong with this scenario? Type I error.

The human genome is comprised of more than 23,000 genes: Splice variants, duplications, mutations, SNPs, non-coding DNA, small interfering RNAs and a wealth of downstream events, which make the interpretation of genomic data highly problematic. The fact that a laboratory can identify a gene does not confer a certainty that the gene or mutation or splice variant will confer an outcome. To put it simply, the input of possibilities overwhelms the capacity of the test to rule in or out, the answer.

Yes, we can measure the gene finding, and yes we have found some interesting mutations. But no we can’t reject the null hypothesis. Thus, other than a small number of discreet events for which the performance characteristics of these genomic analyses have been established and rigorously tested, Type I errors undermine and corrupt the predictions of even the best laboratories. You would think with all of the brainpower dedicated to contemporary genomic analyses that these smart guys would remember some basic statistics.

Beyond Our Borders

I recently returned from Brazil where I participated in a cancer symposium. During my visit I encountered many highly skilled physicians with expertise in breast, thoracic, gastrointestinal and orthopedic oncology. The degree of collegiality and enthusiasm was palpable. The most exciting aspect of my visit was the warm reception and extremely high level of interest in the clinical application of our laboratory platform. It was a refreshing reminder that the parochial thinking of the American oncology community is not the norm throughout the world.

Upon my return, I had the pleasure of meeting a charming 61-year-old woman from New Delhi, India. In review of her chart I recognized her name as a patient for whom we had conducted a study in February of 2012. Her husband, an accomplished businessman, had learned of our laboratory and worked diligently to obtain, process and transport a portion of his wife’s tumor from the surgical suite to our lab. Despite multiply recurrent disease and numerous prior treatments, this patient’s ovarian cancer cells revealed exquisite sensitivity to a drug combination in the laboratory. Her physicians at the Apollo Hospital of New Delhi delivered the treatment exactly as outlined by our lab, and here sitting across from me was the patient in complete remission six months later. The family had traveled from India to meet me and express their thanks.

Each of these experiences speaks volumes for the globalization of cancer care. Cancer patients, whether from Brazil, India or China are more alike than different. Each confronts a seemingly insurmountable adversary. Each in their own way seeks out the best information and advice. And each can be best managed with those treatments found uniquely effective for their tumor. Perhaps once we have conquered cancer in India and Brazil, the EVA-PCD® assay will be ultimately accepted in the United States of America.

The Human Micro Biome

There is a growing recognition that we as a species, humans that is, are not a single organism but a community of organisms living in synchrony. As scientists have recognized for many years, the human gut, skin, and digestive tract are colonized by trillions of bacteria, fungi and other microbes. What we did not realize until recently, was how important these organisms are to our health and well-being

The microenvironment of the human gastrointestinal tract reflects the interplay between bacteria, our diet, intestinal digestive enzymes, lipids, polysaccharides, amino acids, and the by-products of metabolism. The specific make-up of each individual reflects their environment, diet, and family heritage. Indeed, our bacterial flora are transmitted to us by our mothers, who prior to the advent of pasteurized baby-foods, pre-chewed their infant’s food.

More to the point, we now realize that bacterial infections and exposures to foreign antigens early in life protect and prepare us for a healthy adult life. Many modern maladies, such as asthma, diabetes, hypertension, even possibly autism and schizophrenia, may reflect infections, immune responses and the timing thereof. It has been suggested that infections with parasites modulate our immune response. In our increasingly clean environment, devoid of hookworms, tapeworms, and the like, our overactive immune system creates autoimmunity in the form of rheumatoid arthritis, systemic lupus and other maladies.

This reflects the growing recognition that human biology is in fact human ecology. The importance of this cannot be overstated when we examine human tumor biology. We are continually bombarded by the teachings of a cadre of scientists who believe whole heartedly that they can answer the puzzle of human cancer by examining the intricacies of individual human cancer cells, primarily at the level of DNA.  Nothing could be further from the truth.

Take for example just one of the myriad of signaling pathways. Beta catenin is among the most potent tumor promoters. The deranged function of beta-catenin has been identified in several human tumors including prostate, lung and colon. Its closest association being that with colon cancer, wherein the loss of the APC protein (adenomatous polyposis coli), results in a particularly aggressive form of the disease.  The APC protein normally combines with axin and glycogen synthase kinase 3 beta (GSK3B) which all together function to regulate beta-catenin. It is the loss of APC that releases Beta-catenin and drives polyps to become cancerous.

However, upstream of this triumvirate of regulatory proteins are the integrin-ca cadherin proteins that communicate across the cell membrane. By changing the environment of the colon itself, we can influence the integrins, which regulate the cadherins. This in turn regulates beta catenin.  Thus, colon cancer may not arise from changes in our genetic makeup but instead may be driven by micro-environmental changes in the colonic milieu that alter cellular behavior and drive malignant transformation.

Again and again, we are forced to recognize the complexity of human biology. Now we realize that it is not just the genome to the transcriptome to the proteome, but indeed the micro biome.