Investigators in Boston Re-Invent the Wheel

A report published in Cell from Dana-Farber Cancer Institute describes a technique to measure drug cov150hinduced cell death in cell lines and human cancer cells. The method “Dynamic BH3 profiling” uses an oligopeptidic BIM to gauge the degree to which cancer cells are “primed” to die following exposure to drugs and signal transduction inhibitors. The results are provocative and suggest that in cell lines and some human primary tissues, the method may select for sensitivity and resistance.

We applaud these investigators’ recognition of the importance of phenotypic measures in drug discovery and drug selection and agree with the points that they raise regarding the superiority of functional platforms over static (omic) measures performed on paraffin fixed tissues. It is heartening that scientists from so august an institution as Dana-Farber should come to the same understanding of human cancer biology that many dedicated researchers had pioneered over the preceding five decades.

Several points bear consideration. The first, as these investigators so correctly point out: “DBP should only be predictive if the mitochondrial apoptosis pathway is being engaged.” This underscores the limitation of this methodology in that it only measures one form of programmed cell death – apoptosis. It well known that apoptosis is but one of many pathways of programmed cell death, which include necroptosis, autophagy and others.

While leukemias are highly apoptosis driven, the same cannot so easily be said of many solid tumors like colon, pancreas and lung. That is, apoptosis may be a great predictor of response except when it is not. The limited results with ovarian cancers (also apoptosis driven) are far from definitive and may better reflect unique features of epithelial ovarian cancers among solid tumors than the broad generalizability of the technique.

A second point is that these “single cell suspensions” do not recreate the microenvironment of human tumors replete with stroma, vasculature, effector immune cells and cytokines. As Rakesh Jain, a member of the same faculty, and others have so eloquently shown, cancer is not a cell but a system. Gauging the system by only one component may grossly underestimate the systems’ complexity, bringing to mind the allegory of elephant and the blind man. Continuing this line of reasoning, how might these investigators apply their technique to newer classes of drugs that influence vasculature, fibroblasts or stroma as their principal modes of action? It is now recognized that micro environmental factors may contribute greatly to cell survival in many solid tumors. Assay systems must be capable of capturing human tumors in their “native state” to accurately measure these complex contributions.

Thirdly, the ROC analyses consistently show that this 16-hour endpoint highly correlates with 72- and 96-hour measures of cell death. The authors state, “that there is a significant correlation between ∆% priming and ∆% cell death” and return to this finding repeatedly. Given that existing short term (72 – 96 hour) assays that measure global drug induced cell death (apoptotic and non-apoptotic) in human tumor primary cultures have already established high degrees of predictive validity with an ROC of 0.89, a 2.04 fold higher objective response rate (p =0.0015) and a 1.44 fold higher one-year survival (p = 0.02) are we to assume that the key contribution of this technique is 56 hour time advantage? If so, is this of any clinical relevance? The report further notes that 7/24 (29%) of ovarian cancer and 5/30 (16%) CML samples could not be evaluated, rendering the efficiency of this platform demonstrably lower than that of many existing techniques that provide actionable results in over 90% of samples.

Most concerning however, is the authors’ lack of recognition of the seminal influence of previous investigators in this arena. One is left with the impression that this entire field of investigation began in 2008. It may be instructive for these researchers to read the first paper of this type in the literature published in in the JNCI in 1954 by Black and Spear. They might also benefit by examining the contributions of dedicated scientists like Larry Weisenthal, Andrew Bosanquet and Ian Cree, all of whom published similar studies with similar predictive validities many years earlier.

If this paper serves to finally alert the academic community of the importance of human tumor primary culture analyses for drug discovery and individual patient drug selection then it will have served an important purpose for a field that has been grossly underappreciated and underutilized for decades. Mankind’s earliest use of the wheel dates to Mesopotamia in 3500 BC. No one today would argue with the utility of this tool. Claiming to have invented it anew however is something quite different.

Why Oncologists Don’t Like In Vitro Chemosensitivity Tests

In human experience, the level of disappointment is directly proportional to the level of expectation. When, for example, the world was apprised of the successful development of cold fusion, a breakthrough of historic proportions, the expectations could not have been greater. Cold fusion, the capacity to harness the sun’s power without the heat and radiation, was so appealing that people rushed into a field about which they understood little. Those who remember this episode during the 1990s will recall the shock and dismay of the scientists and investors who rushed to sponsor and support this venture only to be left out in the cold when the data came in.

Since the earliest introduction of chemotherapy, the ability to select active treatments before having to administer them to patients has been the holy grail of oncologic investigation. During the 1950s and 60s, chemotherapy treatments were punishing. Drugs like nitrogen mustard were administered without the benefit of modern anti-emetics and cancer patients suffered every minute. The nausea was extreme, the bone marrow suppression dramatic and the benefits – marginal at best. With the introduction of cisplatin in the pre Zofran/Kytril era, patients experienced a heretofore unimaginable level of nausea and vomiting. Each passing day medical oncologists wondered why they couldn’t use the same techniques that had proven so useful in microbiology (bacterial culture and sensitivity) to select chemotherapy.

And then it happened. In June of 1978, the New England Journal of Medicine (NEJM) published a study involving a small series of patients whose tumors responded to drugs selected by in vitro (laboratory) chemosensitivity. Eureka! Everyone, everywhere wanted to do clonogenic (human tumor stem cell) assays. Scientists traveled to Tucson to learn the methodology. Commercial laboratories were established to offer the service. It was a new era of cancer medicine. Finally, cancer patients could benefit from effective drugs and avoid ineffective ones. At least, it appeared that way in 1978.

Five years later, the NEJM published an update of more than 8,000 patients who had been studied by clonogenic assay. It seemed that with all the hype and hoopla, this teeny, tiny little detail had been overlooked: the clonogenic assay didn’t work. Like air rushing out of a punctured tire, the field collapsed on itself. No one ever wanted to hear about using human tumor cancer cells to predict response to chemotherapy – not ever!

In the midst of this, a seminal paper was published in the British Journal of Cancer in 1972 that described the phenomenon of apoptosis, a form of programmed cell death.  All at once it became evident exactly why the clonogenic assay didn’t work. By re-examining the basic tenets of cancer chemosensitivity testing, a new generation of assays were developed that used drug induced programmed cell death, not growth inhibition. Cancer didn’t grow too much, it died too little. And these tests proved it.

Immediately, the predictive validity improved. Every time the assays were put to the test, they met the challenge. From leukemia and lymphoma to lung, breast, ovarian, and even melanoma, cancer patients who received drugs found active in the test tube did better than cancer patients who received drugs that looked inactive. Eureka! A new era of cancer therapy was born. Or so it seemed.

I was one of those naive investigators who believed that because these tests worked, they would be embraced by the oncology community. I presented my first observations in the 1980s, using the test to develop a curative therapy for a rare form of leukemia. Then we used this laboratory platform to pioneer drug combinations that, today, are used all over the world. We brought the work to the national cooperative groups, conducted studies and published the observations. It didn’t matter. Because the clonogenic assay hadn’t worked, regardless of its evident deficiencies, no one wanted to talk about the field ever again.

In 1600, Giordano Bruno was burned at the stake for suggesting that the universe contained other planetary systems. In 1634, Galileo Galilei was excommunicated for promoting the heliocentric model of the solar system. Centuries later, Ignaz Semmelweis, MD, was committed to an insane asylum after he (correctly) suggested that puerperal sepsis was caused by bacterial contamination. A century later, the discoverers of helicobacter (the cause of peptic ulcer disease) were forced to suffer the slings and arrows of ignoble academic fortune until they were vindicated through the efforts of a small coterie of enlightened colleagues.

Innovations are not suffered lightly by those who prosper under established norms. To disrupt the standard of care is to invite the wrath of academia. The 2004 Technology Assessment published by Blue Cross/Blue Shield and ASCO in the Journal of Oncology and ASCO’s update seven years later, reflect little more than an established paradigm attempting to escape irrelevance.

Cancer chemosensitivity tests work exactly according to their well-established performance characteristics of sensitivity and specificity. They consistently provide superior response and, in many cases, time to progression and even survival. They can improve outcomes, reduce costs, accelerate research and eliminate futile care. If the academic community is so intent to put these assays to the test, then why have they repeatedly failed to support the innumerable efforts that our colleagues have made over the past two decades to fairly evaluate them in prospective randomized trials? It is time for patients to ask exactly why it is that their physicians do not use them and to demand that these physicians provide data, not hearsay, to support their arguments.

The Frustrating Reality – When a Tumor Sample isn’t Sufficient for Testing

A dying leukemia cell

A dying leukemia cell

The principles underlying the Rational Therapeutics EVA-PCD platform reflect many years of development. Recognizing the importance of cell death measures — apoptotic and non-apoptotic — our laboratory dismissed growth-based assays. The closure of Oncotech, the principal purveyor of proliferation-based assays, illustrates the demise of a failed paradigm in the study and testing of human tumor biology. A second principal of our work is the need to examine all of the operative mechanisms of cell death (autophagic, necrotic, etc.). Laboratories that measure only one mechanism of cell death (e.g. caspase activation as a measure of apoptosis) miss important cell responses that are critical to the accurate prediction of clinical response. The third principle of our work is the maintenance of cells in their native state.

These fundamentals provide the basis of our many successes, but also a constraint. Because we do not propagate, subculture or expand tissues, we can only work with the amounts of tissue provided to us by our surgeons. While some labs propagate small biopsy samples into larger populations by growth to confluence, this introduces irreconcilable artifacts, which diminish the quality of sensitivity profiles. Avoiding this pitfall, however, demands that a tissue sample be large enough (typically 1cm3) to provide an adequate number of cells for study without growth or propagation.

This is the reason our laboratory must request biopsies of adequate size. The old computer dictum of “garbage in, garbage out” is doubly true for small tissue samples. Those that contain too few tumor cells, are contaminated, fibrotic or inadequately processed will not serve the patients who are so desperately in need of therapy selection guidance. As a medical oncologist, I am deeply disappointed by every failed assay and I am more familiar than most with the implications of a patient requiring treatment predicated on little more than intuition or randomization.

We do everything within our power to provide results to our patients. This sometimes requires low yield samples be repeatedly processed. It may also set limitations on the size of the study or, in some circumstances, forces us to report a “no go” (characterized as an assay with insufficient cells or insufficient viability). Of course, it goes without saying that we would never charge a patient for a “no-go” assay beyond a minimal set up fee (if applicable). But, more to the point, we suffer the loss of an opportunity to aid a patient in need.

Cancer patients never undergo therapy without a tissue biopsy. Many have large-volume disease at presentation, so it is virtually always possible to obtain tissue for study if a dedicated team of physicians makes the effort to get it processed and submitted to our laboratory. The time and energy required to conduct an excisional biopsy pales in comparison to the time, energy and lost opportunities associated with months of ineffective, toxic therapy.

Forms of Cell Death

Following the description of apoptosis in the British Journal of Cancer in 1972, scientists around the world incorporated the concept of programmed cell death into their cancer research. What is less understood is the fact that apoptosis is not synonymous with programmed cell death. Programmed cell death is a fundamental feature of multicellular organism biology. Mutated cells incapable of performing their normal functions self-destruct in service of the multicellular organism as a whole. While apoptosis represents an important mechanism of programmed cell death, it is only one of several cell death pathways. Apoptotic cell death occurs with certain mutational events, DNA damage, oxidative stress and withdrawal of some growth factors particularly within the immune system. Non-apoptotic programmed cell death includes: programmed necrosis, para-apoptosis, autophagic cell death, nutrient withdrawal, and subtypes associated with mis-folded protein response, and PARP mediated cell death. While apoptotic cell death follows a recognized cascade of caspase mediated enzymatic events, non-apoptotic cell death occurs in the absence of caspase activation.

With the recognition of programmed cell death as a principal factor in carcinogenesis and cancer response to therapy, there has been a growing belief that the measurement of apoptosis alone will provide the insights needed in cancer biology. This oversimplification underestimates the complexity of cell biology and suggests that cancer cells have but one mechanisms of response to injury. It has previously been shown that cancer cells that suffer lethal injury and initiate the process of apoptosis can be treated with caspase inhibitors to prevent caspase-mediated apoptosis. Of interest, these cells are not rescued from death. Instead, these cells committed to death, undergo a form of non-apoptotic programmed cell death more consistent with necrosis. Thus, commitment to death overrides mechanism of death.

Labs that focus on measurements of caspase activation can only measure apoptotic cell death. While apoptotic cell death is of importance in hematologic cancers and some solid tumors, it does not represent the mechanism of cell death in all tumors. This is why we measure all cell death events by characterizing metabolic viability at the level of cell membrane integrity, ATP content, or mitochondrial function. While caspase activation is of interest, comparably easy to measure and useful in many leukemias and lymphomas, it does not represent cancer cell death in all circumstances and can be an unreliable parameter in many solid tumors.