Evolution is about information

Living in the computer and information age, most people are familiar with the concept of digital information. Computer hard drives and CD’s hold a certain number of bits of information. The stored information can readily be transformed into music, pictures or documents. Information has an objective reality and is deeply connected to fundamental “laws of nature.” Evolution is about the copying and reproduction of information. It is about a very special class of information that specifies and enables its own reproduction and survival. Simple examples of self-replicating information are contained in computer viruses.

DNA and genetic information

The genetic information contained within a human cell consists of two very similar but non-identical sets of information, joined at the time of fertilization and conception. The information specifies details for the production and regulation of the machinery of life. The information is to a large extent redundant as copies are received from each parent.

The information storage system, the “hard drive” of cells that holds the information needed for cell replication and survival, is made of DNA and proteins. The primary storage medium is DNA. Digital information is stored in duplicate in the sequences of bases that make up the famous double stranded helix of DNA.

Each strand of the double helix contains the same information but in a kind of complementary or mirror image. This provides a “back-up” copy of the data. When cells divide the DNA strands separate. The individual strands then serve as templates for the synthesis of a new strand of DNA. In the process the double helical structure of DNA is restored and each cell receives one set.

The DNA of a normal human cell has approximately 6 billion base pairs divided into 46 very large double stranded chains called chromosomes. The DNA strands in chromosomes are folded and coiled around themselves. In normal human cells there are 22 pairs of chromosomes and two sex chromosomes. Half of the chromosomes are inherited from each parent.

DNA is like a long chain or necklace, made of four different colored beads or bases. Different sequences of the beads (bases) specify different information. The information is contained within the sequence. The bases that make up DNA are composed of four different chemical structures, Adenine, (A), Thymine (T), cytidine (C), and guanine (G). These are like the colors of the beads. The double helix of DNA has two strands of DNA that bind together in a complementary fashion. A binds to T and G binds to C.

To understand how the sequence of bases or beads contains information, consider the analogy to letters and words. Individual letters usually do not have meaning. However, sequences of letters that form words have meaning. The information is in the sequence or pattern. (This same concept will be encountered again when we examine the requirements for the specific cure of cancer and critical aspects of cancer drug design.)

By a complicated series of biochemical processes, the information contained in the DNA base sequences is converted into RNA, proteins and the machinery of life.

Epigenetic information

DNA is the major information storage system of cells, but there are others. Information stored outside of the DNA sequences is called epi-genetic. Epi-genetic information can reside in proteins that regulate the expression of the information within DNA. Epi-genetic information can also be encoded by minor chemical modifications of the DNA bases and of such proteins as histones. The addition of methyl groups (base methylation) can turn off or deny access to the information within parts of the DNA.

The importance of epi-genetic information is illustrated by Dolly the cloned sheep. When the nucleus containing the DNA from a skin cell was transplanted into an ovum lacking a cell nucleus, the epi-genetic information was reset. A skin cell was transformed into a healthy sheep. Similar experiments have been done using mouse melanoma cells. [1] Amazingly, the cancer cells grew into mice. Ultimately, the resulting mice developed melanomas and other cancers. Epi-genetic information can also produce differences between genetically identical twins. [2]

Mutations and genetic alterations

From time to time computer files become corrupted. Data is lost or modified. The same occurs to the genetic and epi-genetic information of cells. The causes are innumerable. We live in a chaotic world. Molecules rapidly bounce around. There is background radiation from cosmic rays and natural decay of radioactive elements such as potassium 40. Reactive free radicals are constantly being produced. Toxic and mutagenic chemicals naturally abound in the environment. There are always thermodynamic and statistical fluctuations.

It is inevitable that cellular information becomes damaged or altered during storage and replication. Oxygen alone is estimated to damage about ten thousand DNA bases per cell per day. Normal cell division is accompanied by the formation of about 50 double stranded DNA breaks per cell. [3] Cells have evolved extensive but imperfect repair mechanisms to correct DNA damage. Since DNA is double stranded there is a backup copy of the required information to facilitate repair. Nonetheless, with time irreversible information loss and corruption does occur. This necessarily results in genetic variation between cells, a pre-condition for evolution.

The level of data corruption is enormously increased above the natural background levels by agents that damage DNA. Classic examples include cigarette smoke, radium, sunlight, X-rays and anticancer drugs. One of the worst and most potent examples of DNA damaging agents was found in natural health food products containing Chinese herbs used for weight reduction. [4] The herbs contained aristolochic acids, a class of chemicals that bind tightly to DNA. An epidemic of cancer of the kidney and urinary tact resulted. This should cause one to question the widely held notion about the safety and superiority of “natural remedies.” There are in fact a large number of natural chemicals present in food that very efficiently damage DNA and are carcinogenic , such as aflatoxins and Ochratoxin A. [5] Cooking food, especially meats, generates large numbers of mutagenic and carcinogenic compounds. [6] Common viral infections damage DNA, not just cancer causing viruses like papillomavirus, but also ordinary run of the mill viruses. Viral and non-viral DNA, can become incorporated into the cellular DNA and alter the information content of cells. [7]

Agents that interfere with the proper transfer of information to each cell at the time of division cause profound alterations in the information content of cells. Such agents can cause the chromosomes to be unevenly distributed at the time of cell division. The DNA itself, at least initially, is not damaged. The net result is an abnormal number of chromosomes in the cell, or a condition called aneuploidy. This can trigger genetic instability and lead to the acquisition of other genetic alterations. There is strong evidence that aneuploidy can cause cancer. [8][9] A rare inherited defect in a gene called BUB1B causes aneuploidy and is associated with cancer. A great debate is raging as to which is more important in the evolution of cancer, aneuploidy or mutations to DNA. Curative therapy had better be able to deal with the fact that both occur.

Genetic alterations are random

The genetic alterations seen in cancer cells arise almost exclusively by random processes. There may be an identifiable cause, like exposure to cigarette smoke or radiation, but the actual changes are random. There are fragile or weak spots in the normal cellular DNA that predispose it to localized damage. [10] Certain very specific genetic alterations, for whatever reason, do occur with a higher than expected probability. For example, chronic myelogenous leukemia (CML) is initially caused by the exchange of very specific regions of DNA from different chromosomes, giving rise to the Philadelphia chromosome. However, based on what is known about the mechanisms of DNA damage, and what has been experimentally observed, we know that almost all the genetic changes are random. [a]

Sex, gene shuffling and genetic variation

Darwin recognized that variation was inevitable in populations of organisms. Much of the variation Darwin observed within plants and animal populations was due to the gene shuffling that occurs during sexual reproduction. Sex mixes up genes and causes genetic variation. Similar but generally non-identical pieces of DNA are exchanged between chromosomes during the formation of the sperm and ovum. The normal shuffling of DNA that occurs during sexual reproduction is very efficient at creating variation within populations. Cancer cells do not engage in sex and sexual reproduction. Cancer cells frequently engage in processes of DNA shuffling between chromosomes that efficiently generates genetic variation. The net result is the same.

A technique called SKY enables the visualization of gene shuffling between chromosomes. With this technique each of the 23 chromosomes is pained with a different color. The shuffling of DNA between chromosomes is evidenced by multi-colored chromosomes. Normally each chromosome is a single color. The picture below shows the SKY analysis of the chromosomes from a patient with pancreatic cancer. Extensive mixing up of the DNA between chromosomes is seen. In addition the total number of chromosomes is abnormally increased.

The chromosomal number and pattern of chromosomal abnormalities or karyotype seen in the patient’s pancreatic cancer was incredibly complex as shown below. By contrast, the normal karyotype of a male is described simply by 46 XY and a female is 46 XX.

The karyotype of the patient’s pancreatic cancer is total genetic chaos

The chromosomal abnormalities seen reflect the enormous diversity and genetic chaos of cancer. It should be noted that the exact combination of genetic alterations is expected to be different in every patient and in every cancer cell.

Extreme genetic diversity is observed in cancer cells

An enormous number of genetic and epi-genetic alterations have been observed in cancer cells. The more one looks the more one finds. There is an entire computer database devoted to cataloging the over 20,000 mutations of just one cancer associated gene, P53. The Sanger Institute has a computer database that to date lists 18,478 mutations in 21 different genes in cancer cells. Currently about 300 different genes have been implicated in causing cancer. But this understates the complexity as each gene can have a large number of different mutations and combinations of mutations. The Mitelman Database of Chromosome Aberrations in Cancer currently contains 44,750 chromosomal abnormalities in cancer. The more one looks the more genetic alterations one finds.

In a study of seventy-five cases of breast cancer, every patient’s cancer was found to have a different set of genetic alterations. Even in the same patient huge differences exist between cancer cells. In one study, the genetic composition of 31% of metastatic lesions in breast cancer differed almost completely from that of the paired primary tumor in the same patient. In renal cell cancer “multiple, genetically almost completely different cells exist in various locations of one and the same patient.” Similar results have been seen with many other cancers.

A recent study of large scale gene sequencing in breast and colon cancer cells exposed to plain view the enormous genetic complexity of cancer. Every cancer had a different set of mutated genes. The "average" cancer cell had 93 mutated genes. Enormous genetic complexity was also observed in high resolution genomic profiles of lung cancer.

How many mutations are there per cancer cell?

The exact number of mutations in a cancer cell has never been counted. That would require sequencing all of the billions of bases in a cancer cell and comparing the result to that of a normal cell from the same person. The proposed Human Cancer Genome Project would provide some information along these lines. A number of studies have examined portions of the DNA from cancer cells, counted the number of mutations, and then extrapolated to estimate the total number in the cell. A variety of different biochemical approaches have been used.

Dr. Garth Anderson and Dr. Daniel Stoler (and their colleagues) at Roswell Park Cancer Institute, examined the DNA of human colon cancer cells. On the basis of their data they concluded that the cancer cells examined had on average 11,000 mutations. Dr. Michael Stratton , from the Sanger Institute, presented results at the 94th AACR meeting that involved the sequencing of large portions of the DNA in cancer cells. The number of mutations varied, but was in the range of thousand to tens of thousands per cell. Cancer cells that are defective in the ability to correct errors during DNA replication, (mismatch repair defects) have even larger numbers of random mutations. Mismatch repair deficient cancer cells can have as many as 100,000 random mutations per cell.

The number of possible cancer cell types is essentially unlimited or infinite

Let’s simplify the problem of estimating the number of different types of cancer cells. We will consider just simple mutations: deleting a DNA base or changing to one of three other bases. We will ignore all the other complex mechanism of genetic alterations. For the purposes of our estimate of the number of different types of cancer cells, let’s assume 10,000 random mutations per cell. Let us assume these mutations are randomly distributed among the 6 billion bases of the diploid human genome. If the point mutation can correspond to a deletion or change to one of three new bases, then the number of combinations is about:

That’s the number 1 followed by 68,000 zeroes. This number is bigger than the estimated number of atoms in the entire universe.

Objection

That’s the number of possible types of cells. How do we know these are cancer cells?

Response

We don’t. And most probably are not. But the number 10^68,000 is so large that this just doesn’t matter. If 99.99999999% of the random combinations of mutations were lethal or incompatible with the cells being malignant, then the process of cancer would almost always self-terminate. Cancer would put itself out. Yet even with this erroneously high estimate, there would still be at least 10^67,990 different types of cancer cells! Indeed, given the extreme sensitivity of cancer cell growth to the probability of malignant cell survival even if 60% of the combinations of genetic alterations were dead end then cancer would put itself out.

Our estimate focused just on one limited class of genetic alterations. It ignored more complex DNA rearrangements, massive DNA losses and duplications, gains and losses of chromosomes, and epi-genetic alterations. We also based our estimate on a low number of mutations, just 10,000 per cell.

The message is clear. The genetic and epi-genetic diversity of cancer is for all practical purposes unlimited. The number of different types of cancer cells that could potentially arise in the same patient is for practical purposes, infinite.

Objection

The cell has about 6 billion base pairs in its DNA. You’re counting cells that differ by even a trivial change in a single base as different.

Response

The reality is that a single base change out of 6 billion can be the difference between life and death for a patient. About ten percent of patients with lung cancer respond initially very well to a drug called Iressa (Gefitinib ).[11],[12] The drug works by shutting off a protein called Epidermal Growth Factor Receptor (EGFR). It binds tightly to certain mutant forms of EGFR found in some patient’s lung cancer cells. With time Iressa fails to work and the patients suffer progressive disease. The change of a single base in DNA that encodes the mutant EGFR protein has been shown to cause drug resistance. [13],[14] The second example, is Gleevec. Chronic myelogenous leukemia is initially caused by a particular set of genetic alterations that create a new protein called Bcr-Abl. Gleevec binds to this protein and inhibits its function. Gleevec is extremely effective in treating CML, at least initially. With time resistance develops. Changes in a single base pair in the DNA that encodes the protein Bcr-Abl have been shown to confer drug resistance and lead to disease progression.[15] The conclusion is inescapable, even the most minor change to the DNA can be of great importance in cancer.

Objection: mutations to non-coding DNA

Although the cell has about 6 billion base pairs in its DNA only a small percent of the bases actually encode proteins. Therefore, most of the mutations must be just noise.

Response

Approximately 98.8 % of the DNA in human cells does not code for proteins. However, it is incorrect to conclude that changes in non-coding DNA are of no biological consequence. That’s like saying that parts of your computers hard drive that don’t actually contain data for a word document can be damaged or changed with impunity. Non-coding regions of DNA provide structure to DNA and chromosomes and bind or influence the binding of DNA regulatory proteins. RNA from non-coding DNA can also influence gene regulation. [16], [17] The importance of non-coding DNA is highlighted by the fact that many of the sequences have been evolutionarily conserved over a time period of 450 million years. The sequences are associated with genes that regulate embryonic development. [18][19] The sequences contain key information that evolved during the transition from single cell organisms to multi-cellular organisms. It is this information that is key to keeping individual cells in line and preventing cancer. Indeed, evidence is accumulating for a role of non-coding small RNA (microRNA) molecules in cancer.[ 20] [21] Extensive changes in micoRNA gene copy numbers have been observed with high frequency in cancer cells. We cannot ignore genetic alterations that occur in non-coding DNA.

Objection: Genetic noise

Most mutations and genetic alterations that arise in cancer cells are just noise, random noise. They are of absolutely no importance.

Response 

It is probable that almost all genetic alterations in cancer cells are of no biological or clinical significance. But enough are! Even a small tumor contains tens of billions of cancer cells. With so many cells, and so many different genetic alterations, clinically significant variation is almost inevitable. There is no basis to summarily conclude that the random changes of cancer cells are just noise and of no significance. Short of experimentally characterizing every unique set of genetic alterations, there is no way to know which are significant. There is no such thing as an average cancer cell or a typical cancer cell. Cancer is about statistical extremes. Cancer is about the rare cell out of billions of cells that evolves resistance to therapy and falsifies our favorite theories. There is no escaping the conclusion that the variation available to cancer cells is almost unlimited.

Objection: cancer stem cells

Experimental analysis of tumors does show extensive and diverse genetic alterations. But most tumors cells are dead end. What counts are the tumor stem cells and the genetic alterations in stem cells.

Response 

Stem cells are cells that have the capacity to repeatedly proliferate and give rise to tumors. It has been known for 50 years that only a very small percentage of tumor cells have the capacity to give rise to colonies in tissue culture or tumors upon injection into animals or people. In the 1950’s patients with cancer actually volunteered to have their own tumor cells injected into their thighs. Tumors only formed at the injection site if more than one million cells were injected. Similar results have been observed in animals. In a study of 143 human tumors of 24 different types, the median percentage of colony forming cells was only 0.02 %. [22] In other words only about 2 out of every 10,000 tumor cells were able to grow in tissue culture and give rise to new colonies of cells.

Normal tissues like the bone marrow and the gastrointestinal tract have stem cells that are responsible for tissue growth and regeneration. Most normal cells in the body lack the ability to replicate. Tumors can mimic this type of organizational hierarchy and can have small populations of stem cells that are really responsible for the disease.

The situation is similar to an ant colony. Tumor stem cells are like queen ants. In ant colonies only the queens can reproduce. The worker ants, which make up almost the entire ant population are sterile and cannot form new colonies. Normal human tissues have a similar organizational structure with distinct stem cell populations. In normal tissues most cells are sterile and cannot reproduce. This social order may or may not be preserved during the course of tumor cell evolution. It developed during the transition from single cell organisms to multi-cellular organisms and represents one of the primary mechanisms of tyranny of the multi-cellular organism over its single cellular components. Cancer is the breakdown of this tyranny and the re-activation of evolution. Cancer cell lines exist in which a very high proportions of the cells have the capacity to cause cancer and to replicate seemingly without limit. However, cell lines have also been described that retain in vitro a subset of cells with stem cell like properties. [23]

A recent study by Dr. Michael Clarke and his colleagues at the University of Michigan demonstrated that only a tiny fraction of human breast cancer cells were able to give rise to tumors when injected into mice. [24] Cells with the ability to form tumors could be identified by the presence of a specific pattern of proteins or markers on the cell surfaces. As few as 100 cells with the pattern were sufficient to form a tumor, while the injection of one hundred thousand cells without the pattern failed to generate tumors in mice.[4] Similar results indicative of stem cells have been observed in leukemia and a number of other cancers. [25]

It is the stem cells that evolve in cancer In fact, by definition unless a cancer cell has stem-cell-like properties it cannot contribute to tumor cell evolution.

What is a cancer cell?

The definition of cancer is clear, but it is by no means clear how to determine if a given cell is a cancer cell. In most tumors nearly half of all tumor cells are destined to die, or will fail to replicate. If not, tumor growth would be explosively rapid. In some tumors there may be an organized hierarchy of proliferation, with a restricted population of stem cell. There is no doubt that genetic and epi-genetic alterations cause cancer. This would seem to imply that malignant behavior, the defining feature of cancer, is a property of the cell. But the problem is much deeper. A tumor cell always exists in an environmental context. A tumor cell placed in one environment can remain dormant for prolonged periods of time, perhaps for the life span of the patient. The tumor cell can remain asleep, not replicating and not engaging in invasive behavior. The very same cell, and its offspring, when placed in a different environment can engage in malignant behavior and give rise to a large tumor. [26] [26.2]

In a very elegant series of experiments Dr. George N. Naumov and colleagues, incubated mouse breast cancer cells in tissue culture with fluorescent nanospheres. The tiny fluorescent particles were internalized by tumor cells. When viewed under a microscope with UV light the cancer cells glowed intensely green. However, every time the cells divided the number of nanospheres per cell and the green glow decreased. After three cell divisions the cells were scarcely fluorescent.

When the fluorescent tumor cells were injected into the fat pads on the backs of mice large tumors developed within 3 weeks. However, when tumor cells were injected into the tail veins of mice, large numbers of tumor cells landed in the liver. But tumors didn’t form. The intensely green cells solitary tumor cells could easily be seen. Even after 11 weeks half of the injected tumor cells could still be seen in the liver. They were dormant, alive but not proliferating and not forming tumors. After eleven weeks the cells were isolated from the liver and placed in tissue culture media. The tumor cells that had previously been sleeping awoke and began to divide. The resulting tumor cells were then injected into the fat pads of mice. All the mice developed large tumors in the fat pads. These results clearly demonstrate the crucial role of the tumor cell environment in determining if a cell can engage in malignant behavior and cause the disease of cancer.

Dr. Naumov and his colleagues made another series of very interesting observations. They used the same technique to label a genetically related line of mouse breast cancer cells. This cell line was selected for its ability to develop metastatic tumors in the liver. Sub-clones of cells that have a propensity to grow in particular organs have been observed numerous times and represent excellent examples of tumor cell evolution. When injected into the tail veins of mice these tumors cells found their way into the liver. By day 10, more than 60% of the cells were solitary, isolated cells, alive but not dividing. Approximately 0.6% (6 out of 1000) of the tumor cells injected formed metastatic colonies in the liver at day 10. Four days later the total number of metastatic lesions in the liver had decreased 100 fold. In other words, 99% of the tumor colonies had died off, but the total amount of tumor in the liver had increased about five fold.

By day 21 the tumor was lethal in size, and had almost completely replaced the normal liver. It turned out that only about 0.006% (6 out of 100,000 cells) of the injected cells were ultimately responsible for giving rise to the lethal mass of tumor. Surprisingly at day 21, over twenty percent of the injected tumor cells were still sitting in the liver, alive, not having replicated. About thirty cancer cells that gave rise to large metastatic lesions in the liver defined the fate of the mice.

Where these thirty cells, stem cells, and all the others just non-stem cells? Perhaps yes, perhaps no. Either way, the huge difference between the two genetically related cell lines is clear. One causes fatal metastatic disease in the liver, the other does not. The reason is due to cancer cell evolution, or tumor stem cell evolution. The difference in words is just semantics. The real issue is what can be known about all malignant cells that could evolve.

The concept of a cancer cell is in reality a bit murky

Cancer is not only a property of the cell. Rather, it is a property of the cell, the environment and the future. We can’t know the future. You cannot look at a cell, even with the most powerful techniques in molecular biology and know if the cell can give rise to a tumor. It is indeterminate. You need to wait and see what happens. If the cell divides and its progeny give rise to other cancer cells and a tumor then we know in retrospect that the cell was cancerous. Of course from a practical point of view pathologists routinely and confidently identify cancer cells in biopsies and correctly diagnose malignant disease. But that’s not good enough for our purposes. To consistently cure cancer we need a theory of cancer that is accurate to one part in one thousand trillion. So we better be very careful with our logic. It is better to acknowledge the fuzziness of the concept of a cancer cell and deal with it, rather then to ignore the issue.

Another level of chaos in cancer: natural selection

The problem is not just the sheer diversity of random genetic variation. Natural selection is also an incredibly diverse, unpredictable, stochastic process.

Evolution is not random. Richard Dawkins expresses the non-randomness of evolution with a poetic beauty:

Of all the wolves that might survive, a nonrandom sample — the fleetest of foot, the canniest of wit, the sharpest of sense and tooth are the ones that do survive and pass on their genes. [27]

The same holds true for tumor cell evolution. On the one hand, tumor cell evolution is an enormously diverse, unpredictable, stochastic evolutionary process, while on the other hand, evolution is not random. However, there is no contradiction. The problem is that the word random is too coarse and imprecise a term to accurately describe evolution.

There is an important difference between a random and a stochastic processes. In a sense, a stochastic process has a “memory”. The outcome of a stochastic process depends upon prior events and the history of the system. In the case of evolution, the “memory” is stored in DNA sequences. By contrast, not all random processes have a “memory” — consider the roll of die or the flip of a coin.

Evolution obeys a unique rule: “survival of the fittest”.

If the selective pressures are fixed and unchanging this rule can impose order and a degree of non-randomness to evolution. This is illustrated by Dawkins’s example of wolves in the wild.

However, under differing selective conditions the result can be totally different. Wolves evolved into dogs. Man domesticated wolves by selecting and breeding the tamest of each generation, over and over again. Further selection for desired traits created a wide variety of dogs, ranging in size from the Chihuahua to the 280-pound Irish Wolfhound.

The selective pressures are critical in defining the outcome.

Tumor cell evolution is an enormously diverse, unpredictable, stochastic process.

The selective pressures that influence tumor cell survival and replication depend upon both the properties of the cell and the environment and change with time. Cell survival involves multiple, complex, interacting, stochastic processes and factors including: the genetic make-up of the cell and of all other cells in competition for survival; the location of the cell; the immune response; drugs; nutrients; hormones; oxygen; etc. In addition, tumor cell evolution is iterative. Outputs from one generation become inputs for the next generation and further compound the chaos.

Iterative processes frequently generate chaos. Even a very simple well-defined set of rules can give rise to totally chaotic and unpredictable results when the output from one cycle becomes the input for the next cycle. Minor differences in the initial starting point can become magnified and after multiple cycles give output results that are completely different. The mathematician Stephen Wolfram gives many examples of this type of behavior in his book, A New Kind of Science. [28] The rolling and folding of dough provides a good example. Two particles of flour that are initially next to each other, in almost exactly the same starting position, can become widely separated, and end up in totally different places after multiple cycles of rolling and folding. The same type of chaos-generating effect can occur during tumor cell evolution, but the situation is far more complex. In tumor cell evolution the rules or selective pressures that determine natural selection keep changing. [b] And the changes are stochastic. It’s like randomly changing the procedure used to repeatedly roll and fold dough.

Yet another level of chaos in cancer

There is yet another layer of randomness and complexity. Bulk tumors are complex self-organized structures. They arise from the more general form of natural selection that Richard Dawkins refers to as “survival of the stable.” [29] Tumor growth requires a blood supply and the ability to invade surrounding tissues. Both the establishment of a blood supply and invasiveness are cooperative, social processes, involving the interaction of many cells. Bulk tumors form because the components that have been selected form a stable structure. The genetic and epi-genetic make-up of a cancer cell is important. But equally important are the interactions between cells within the bulk tumor. The elaboration of growth factors that stimulate angiogenesis (new blood vessel formation) by one sub-set of tumors cells can enable the survival of other different cancer cells. The interactions can also be negative. Just as societies collapse, so do tumors. Tumors commonly have massive regions of dead and dying cells. A common cause is oxygen deficiency due to compression of the blood supply by the tumor cells. This type of collapse exerts selective pressure on tumor cells.

The selective pressures that act upon tumor cell populations vary both in time and location. Darwin recognized that different species evolve that are adapted to survival in different ecological niches. The same applies to tumor cells. The body provides many different tissue environments for tumor cell growth. The cellular environment in the liver is quite different from that in the brain or lungs. Cancer cells that spread from a tumor to different tissues have different genetic and epi-genetic alterations. Sub-lines of cancer cells from a tumor that preferentially colonize particular organs such as the liver, lung, or bone marrow have been demonstrated in animal models. The cancer literature is full of examples. [30]

The limitations of observation and empirical data in cancer

The study of tumor cell evolution is the study of history. Our understanding of the pathways of tumor cell evolution in a patient is limited to a historical description of identifiable tumor cells that have evolved. Yet even this historical description is incomplete. It is rarely possible to identify all tumor cells in a patient with disseminated cancer.

Efforts to identify or predict the course of tumor cell evolution in patients with metastatic cancer are futile and destined to fail. Evolution is an unpredictable, stochastic process that is contingent upon an enormous number of unpredictable events.

Bulk tumor and identifiable metastatic lesions are accessible to observation and experimental characterization. However, conclusions based on the analysis of a subset of cancer cells cannot be generalized to all cancer cells in the patient.

There is simply no logical validity in generalizing from a subset to the entire set. Nature doesn’t respect or obey that type of inductive logic. A parable told by Bertrand Russell illustrates the problem:

A farmer feeds a chicken every morning. The chicken becomes habituated to this ritual. It expects the farmer to feed it every morning into perpetuity. Then comes the morning when the farmer comes to wring its neck so that it may be cleaned, dressed, and served for dinner that night. [31]

And so it is with cancer. If we examine 10 million cancer cells and find them all sensitive to a drug, we cannot conclude that all cancer cells in the patient are drug sensitive. They almost certainly are not. One drug resistant cancer cell can become a big problem. Although not overtly stated, the flawed logic of induction is embedded in many of the failures in cancer research.

Oncogenes and tumor suppressor genes

It is possible to identify particular genetic alterations that cause cancer. About 300 different oncogenes and tumor suppressor genes have been identified to date. An oncogene is an abnormal gene or segment of DNA that can cause or contribute to making a cell cancerous. A tumor suppressor gene is a gene that when lost or deleted, or inactivated contributes to a cell becoming cancerous. The discovery that particular acquired genetic abnormalities can cause cancer was a landmark event in the history of science and the work of intellectual giants. It has resulted in tremendous insights into the mechanisms and regulation of cell division. The discovery of oncogenes and tumor suppressor genes corroborated the theory of cancer proposed in 1929 by Dr. Thomas Boveri. [32]

The existence of oncogenes and tumor suppressor genes is direct evidence of tumor cell evolution. In 1989, Dr. Harold E. Varmus and Dr. J. Michael Bishop received the Nobel Prize in Medicine for their work on oncogenes. In his Nobel Lecture, Dr. Bishop quoted the biologist Dobzhansky, “Nothing in biology makes sense except in the light of evolution.” [33]

Turning oncogenes on and off in mice

The development of techniques to insert oncogenes into cells along with a molecular switch that can turn the oncogene on or off has provided a powerful tool for the study of cancer. It is a very elegant well-defined system. The exact cause of the cancer is known, at least initially. Many different oncogenes have been studied using such models. In mice, when the oncogenes are turned on, tumors grow and cause progressive disease. When the oncogenes are then shut off in many cases, the tumors rapidly shrink. In some cases even very large tumors containing several billion cells will melt away and become invisible to the naked eye. Some mice will remain tumor free for prolonged periods of time and perhaps even be cured. However, in a significant fraction of the mice the tumors will recur. Despite the fact that the oncogene is shut off, tumor cells grow. When these cells are examined, new genetic and epi-genetic alterations are frequently found. New mechanisms of malignancy evolve that enable the continued propagation of the tumor cells. [34]

In one study, an oncogene called MYC was inserted in mouse cells. Lymphomas developed when the oncogene was activated. When the MYC gene was shut off the cancers regressed. Some cancers relapsed and the mice developed progressive disease. Lymphoma cells in relapsed mice had acquired new chromosomal alterations. In each of 11 mice the lymphoma cells had acquired different, unique sets of genetic alterations. [35] The resistant lymphoma cells were no longer dependent upon the function of the initiating oncogene, MYC. Needless to say, the course of tumor cell evolution and escape from MYC dependence was different in each mouse. Similar results have been observed in MYC induced breast cancers and with a variety of other oncogenes. [36] [37]

The probability of a cell evolving resistance or escaping oncogene dependence is dependent upon the total number of cancer cells present. People typically have much greater numbers of cancer cells than a mouse and much greater time for tumor cell evolution to occur. A 3-gram tumor in a mouse is huge. By contrast the same sized tumor in a person is considered small. Accordingly, the problem of tumor cell evolution is much more apparent and severe in people. This is one of the reasons it’s so much easier to cure mice of cancer than people.

It is common to see pictures of two mice: one untreated with a huge and ugly tumor; the other mouse looking healthy with no visible tumor after treatment. Although impressive, such pictures can be quite deceiving. What is generally gone in the healthy appearing mouse is just the “tip of the exponential iceberg.” A two-log reduction in tumor cell burden will shrink the tumor size 99% and make it appear gone. But often millions of malignant cells remain that can cause the tumor to reappear.

Studies done in tissue culture can be even more misleading. A drug is studied and kills 95% of the cancer cells, or decreases cancer cell growth by 95%. If the drug is nontoxic to normal cells these results may be headline news. However, such results often have little real significance. It is the remaining 5% of cells that don’t respond that count and that falsify theories and cause treatment failures. In theory and practice, a single drug resistant cancer cell can lead to progressive disease.

Chronic myelogenous leukemia

Chronic myelogenous leukemia (CML ) provides an example in people that is strictly analogous to the previously discussed oncogene experiments in mice. CML is a cancer of bone marrow stem cells that causes patients to have very high white blood cell counts and often enormously enlarged spleens. The disease is almost always caused by a specific genetic alteration that results in the formation of an abnormal chromosome called the Philadelphia chromosome. A piece of chromosome 9 is broken off and switched with a piece of chromosome 22. As a result a gene called ABL from chromosome 9 is fused to a gene called BCR on chromosome 22. The net result is that an oncogene, Bcr-Abl is formed. This oncogene codes for the production of a new protein that causes the cells to be cancerous. The oncogene causes CML when transferred into mice. CML is a very slowly progressive type of leukemia. The chronic phase typically lasts several years, during which the disease is stable and undergoes little evolutionary change. This is followed by an accelerated phase followed by blast crisis. These phases are characterized by the acquisition of new genetic abnormalities and rapid disease progression. Survival after onset of the blast phase is usually only several months.

At the time of diagnosis early stage CML has usually very little genetic heterogeneity. The cancer cells are almost all very similar. This contrasts very sharply with most solid cancers that have substantial genetic differences between cancer cells at diagnosis. The rate of acquisition of new clinically significant genetic alterations is also slow in early stage CML. This is evidenced by the very slow disease progression. In addition, CML has a single well-defined causal lesion. [c] CML represents a best-case scenario in terms of understanding and treating cancer. It starts of with very little evolutionary character and minimal complexity. In other words the rate of mutation initially is low.

Gleevec

Gleevec is a drug that inhibits the function of the Bcr-Abl protein. It specifically targets the cause of the CML. Gleevec is an extraordinary anticancer drug. About 95% of patients with chronic phase CML achieve a complete but temporary response with normalization of the white blood cell count with minimal side effects. The drug is so effective it was on the cover of Time magazine. However, with time tumor cell evolution occurs, drug resistant clones evolve, and the disease progresses. In a study of 144 patients with chronic phase CML treated with Gleevec, 22 (16%) had disease relapse over a 16-month period of time. [38] The acquisition of new genetic alterations was associated with drug failure. In newly diagnosed patients with CML the rate of resistance to Gleevec is approximately 4% per year.

The molecular mechanisms by which CML cells evolve resistance to Gleevec have been examined in detail. In most patients mutations arise in the Bcr-Abl oncogene and result in a slightly altered protein that is no longer inhibited by Gleevec. To date 15 different mutant forms of Bcr-Abl have been detected in patients and 112 variants have been identified or created in tissue culture systems. [39] In some cases, the leukemic cells have acquired the ability to pump the drug out of cells. In others, the oncogene is over-expressed. In yet other instances, totally novel mechanisms of malignancy have evolved that are independent of the Bcr-Abl oncogene. [40] [41] [42] [43] This coincides with the onset of genetic instability. In this situation the leukemic cells no longer require the initiating oncogene, Bcr-Abl. Other drugs such as BMS-354825 and AMN107 have been developed that are active against some mutant forms of the Bcr-Abl protein. [44] [45] However, the development of drugs targeted to novel mechanisms of malignancy that are independent of Bcr-Abl is highly problematic. The onset of genetic instability opens the floodgates to virtually unlimited complexity and genetic diversity. [46] [47]

Gleevec is a great and important medical advance. Thousands of patients have benefited greatly from the drug, but the drug is not curative and does not address the clinical problem of tumor cell evolution. The development of drug resistance even in CML forebodes badly for targeted drug therapies in common solid cancers. The major forms of cancer display much more extensive genetic variation at diagnosis than CML.

Gastrointestinal stromal tumors (GIST )

A rare type of solid abdominal cancer called gastrointestinal stromal tumor or GIST further illustrates the problem. GIST tumors are commonly associated with mutations in a gene called KIT. The oncogene produces a protein called, “kit” that, like Bcr-Abl, is inhibited by Gleevec. Gleevec shrinks many gastrointestinal stromal tumors. However with time resistant cells evolve, and the disease progresses. Examination of the KIT gene in resistant cancer cells from the same patient revealed four different sets of mutations that conferred Gleevec resistance. [48] New mechanisms of malignancy unrelated to KIT have also been observed in Gleevec resistant GIST cancer cells. [49] It is the same story as for CML.

Patients with GIST that develop Gleevec resistance have been treated with another targeted drug, SU11248. Transient responses were observed. The median time to progression was 6.3 months with SU11248 versus 1.5 months with placebo. [50]

Drug resistance

The development of drug resistance to Gleevec in CML and GIST is not unique. Resistance has been described to virtually every anticancer agent examined. We are not aware of a single anticancer agent to which it has not been possible to select drug resistant clones of cancer cells. Multiple drug resistance is also commonly observed to evolve. Cancer cells evolve that can survive exposure to four or five unrelated anticancer drugs. Cancer cells have even been described that grow better, or require, the presence of anticancer drugs for growth. Examples include the prostate cancer drugs bicalutamide and hydroxyflutamide and the breast cancer drug tamoxifen. [51] [52] Imagine that. The very drugs that are supposed to kill the cancer cells can actually improve their growth. Such is the power and problem of tumor cell evolution. Not surprisingly, the mechanisms of drug resistance are incredibly diverse. No one mechanism has been shown to be of overriding clinical importance. The development of drug resistance is direct evidence of tumor cell evolution and a major problem in the treatment of cancer.

Tumor cell escape from immune attack

Tumor cell evolution also enables cancer cells to evade destruction by the immune system. Cells that can be destroyed are destroyed. What is left is resistant to immune destruction. The mechanisms of escape are extremely diverse. [53] Tumor cells can lose the antigens that trigger an immune response. An antigen is a molecule that the immune system reacts against and attacks. In addition, tumor cells can inhibit the immune response, or develop resistance to the killing mechanisms involved. For a tumor to grow and cause disease in the first place it must evade destruction by the immune system. The immune response is therefore a major selective pressure that directs the flow of tumor cell evolution. [54] [55] This is one reason that immunotherapy has had so little success in the cure or chronic control of cancer in patients. Another reason will become clear when we discuss the information requirements for the specific cure or control of cancer.

Tumors recruit immune cells and other normal cells to help in the process of invasiveness. Tumor cells evolve that not only escape destruction by immune attack, but also that subvert normal immune cells to enhance tumor growth. Tumors recruit normal white blood cells to help in the process of tissue invasion. [56][57] Tumors also can release soluble factors that stimulate normal cells in the environment to produce enzymes that digest connective tissue and facilitate invasiveness. [58] It’s like renting bulldozers to clear space for new apartment houses.

Pre-metastatic niches

Malignant cells can release chemicals that travel in the blood stream and recruit bone marrow derived cells to prepare distant sites in the body for new colony formation. [59] This can occur before the cancer cells reach these sites. The new sites called pre-metastatic niches in turn liberate chemical signals that can attract malignant cells and thereby facilitate the process of new colony formation and the spread of cancer. The formation of these pre-metastatic niches is an invasive process that utilizes normal cellular machinery important for wound healing. [60]

The evolution of cancer cells that use normal cells and normal cellular machinery to confront selective pressures and achieve a survival and reproductive advantage is the dominant theme in cancer.

Information loss and cancer cell evolution

Multi-cellular organisms evolved from single cell organisms. Six hundred million years of variation and natural selection has created many layers of protection against cellular evolution within an organism. A yeast cell, which is a single cell organism, has about 6000 genes. Only about one thousand are essential for the yeast to grow and reproduce. [61] A cell inside a multi-cellular organism (i.e., the human body) has similar information and gene requirements for cell replication. The basic tasks that must be completed are essentially the same. The cellular machinery used by yeast and human cells to divide is remarkably similar. The precise number of genes in human cells is unknown but exceeds 30,000. Most of these genes are not required for cell replication. Many encode information that specifies and maintains the normal architecture of the human body. This involves restraining and constraining the tendency of cells to behave as single cell organisms. These genes are tumor suppressor genes.

Many genes also function to minimize and correct the inevitable genetic alterations and information corruption that occurs in cells. Other genes function to detect genetic alterations and kill the cell or halt cell replication. It is therefore not surprising that the loss of genetic information is a major mechanism by which cancer cells evolve from normal cells. Cancer cells very commonly evolve with deletions of large portions of DNA. Even entire chromosomes are frequently lost. The information needs of a single cell organism, a cancer cell, are less than that of a multi-cellular organism. To a large extent cancer involves the loss of information that evolved to make single cell organisms subservient slaves to the needs of the multi-cellular organism. Information that is required for or promotes cell survival and malignant behavior (i.e. proliferation and invasiveness) is generally conserved. [62] [63] Dr. R.A. Gatenby and Dr. B.R. Frieden sum up the problem well:

The nonlinear dynamics of stochastic information loss constrained by somatic evolution indicate that carcinogenesis will not be associated with any predictable, fixed sequence of genomic alterations. Rather, sporadic clinical cancers are emergent structures produced by multiple, fundamentally nondeterministic genetic pathways. [63]

However, cancer is a disease of exceptions. Chromosomes and genetic information can also be gained and amplified in cancer. For example, in the childhood cancer, neuroblastoma the cancer cells commonly will have 3, 4, or even 5 copies of each chromosome per cell. [64] In addition, genes are frequently found to be increased in copy number in cancer cells. [65] Ordinarily a cell will have two copies of a gene, one form each parent. However, cancer cells will frequently evolve with additional copies of genes that enhance tumor cell survival and proliferation. For example, tumor cells frequently will develop resistance to the anticancer drug 5-fluorouracil. This drug causes the irreversible inhibition of the enzyme thymidylate synthase , (TS). One mechanism by which cancer cells evolve resistance to 5-fluorouracil is to increase the number of gene copies for the enzyme TS. [66]

The rate of tumor cell evolution, yet another stochastic process

Evolution in general and tumor cell evolution in particular, does not proceed at a constant rate. There can be periods of stagnation and periods of intense evolutionary change. Cells are complex systems of a large number of interacting components. Changes to one cellular component can trigger an avalanche of effects. A particular genetic or epi-genetic alteration can have profound consequences at distant sites and times. A genetic alteration that disables the repair of DNA damage can markedly accelerate the rate of mutation, tumor cell evolution and disease progression. Processes that interfere with telomere function can also trigger genetic instability. [67] [68]

Some cancer cells shuffle the DNA between chromosomes. One mechanism that cancer cells use to do this involves the same normal cellular machinery that is used for crossing over of chromosomes during sperm and ovum formation. When tumor cells evolve this capacity the rate of tumor cell evolution is markedly accelerated. [69] Since tumor cell evolution improves the ability of cancer cells to survive and replicate, accelerating evolution can increase the rate of disease progression. Defects in the segregation of chromosomes between cells at the time of cell division results in some cells having excess chromosomes and others an incomplete set. This condition called aneuploidy can result in rapid tumor cell evolution. [70] [71] Cancer cells can have bizarre numbers of chromosomes and still grow quite well. However aneuploidy can also impair cell survival. Such contradictions are commonplace in cancer. Like most other features in cancer, the rate of progression of tumor cell evolution is stochastic and unpredictable.

Cancer and self-organizing criticality (SOC)

A common property displayed by systems comprised of large numbers of interacting components is called self-organizing criticality (SOC). The physicist Per Bak, in his book, How Nature Works provides an absolutely fascinating and highly readable overview on the pervasive role of SOC in nature. [72] Evolution displays features of SOC. [73] as does cancer, at many different levels. [74]

Sandpiles, avalanches, and power laws

The classic example of SOC is a sandpile. [75] In 1987 the most frequently cited scientific paper in the entire field of physics was on the topic of sandpiles. It may seem very strange, but some of the most important and interesting science has come from studying what happens when you drop grains of sand on top of each other and form a sandpile. After a sandpile reaches a certain critical size, the addition of a single grain of sand can trigger an avalanche. The size and frequency of the avalanches is described by a power law. [d] Small avalanches are frequent. Massive avalanches also occur, but infrequently. The same mathematical law describes the frequency of both small avalanches and avalanches millions of time larger. Although comprised of individual grains, the sandpile exhibits behavior that can only be described globally within the context of the entire pile of sand. A sandpile cannot be understood by traditional reductionist methods of science. Very deep and important insights into an astonishingly wide range of phenomena in nature have resulted from these studies.

The size of an avalanche that results from adding a grain of sand to the pile is unpredictable. A signature of self-organizing criticality is power law statistics. If you examine a large number of tumors, some tumors will have a small number of readily detectable chromosomal alterations. Others will have large numbers. The relationship between the number of evolved chromosomal alterations and frequency is described by a power law. [76] Systems that display SOC are inherently unpredictable and not amenable to adequate analysis by traditional reductionist methods. A sandpile cannot be understood like a Swiss clock.

Mutation rates and cancer evolution

It is the unpredictability and randomness of it all that has made cancer such a difficult challenge. Cancer cells develop marked genetic instability and increased mutation rates during disease progression. A large number of different biochemical mechanism that have been observed to cause genetic instability in cancer cells. [77] This reflects the complexity of the biochemical processes that must succeed for the accurate replication and transmission of genetic data. Epi-genetic or genetic lesions that compromise any of the critical steps can cause genetic instability, increase the variation within the cancer cell population and potentially accelerate tumor cell evolution.

However, there are strong reasons to believe that an elevated mutation rate is not required for the development of cancer. [78] Nor is the exposure to cancer causing chemicals required. A variety of data indicates that about six to ten genetic alterations in a cell are required to cause cancer. Not just any mutations will work. The genetic alterations must free single cells from the tyranny of the multi-cellular organism. One way to do so is to create oncogenes or inactivate tumor suppressor genes. Mutations that actually damage critical sites of DNA are rare events in “normal” cells. Diseases such as colon cancer can evolve over decades and thousands of cell divisions. Given the large number of stem cells present in the colon, and the large number of cell divisions, mathematical models predict that the normal baseline rate of mutations is sufficient to explain the development of colon cancer. [79] (This does not imply that carcinogens and other preventable environmental factors play no role. They clearly do.) Dr. Peter Calabrese and colleagues sum up the results of their mathematical analysis as follows:

There is a “lottery like accumulation of stochastic mutations”… The process is analogous to number picking in a lottery. Although the odds of winning for any given player may be extremely low (say 1 in 120 million), the players that win the lottery play far fewer than 120 million attempts. Our pre-tumor progression model is similar to a lottery because a large number of stem cells are at risk. Although the probability any given stem cell accumulates a rare combination of mutations in a lifetime is incredibly small, the probability just one of the many stem cells accumulates these mutations is not nearly as small. [80]

In another paper, Dr. Ian Tomlinson and colleagues estimated, based on the mutation rates of normal cells, the number of mutations that will accumulate during normal cell replication in colon stem cells and in colon cancer. [81] By age 65, approximately 125,000 mutations per stem cell will accumulate before the onset of cancer, just during normal cell division. For a polyp the total number of mutations was estimated to be approximately 10¹² (one trillion). The total number of different mutations in a small colon cancer was estimated to be about 10¹⁴ (one hundred trillion) per tumor, not per cell. These numbers don’t include the excess mutation rate that occurs during tumor evolution and progression. [82]

Fields of pre-cancerous cells

Most tumors evolve over long periods of time. There is a continuum in the accumulation of genetic alterations. The cells and tissues can appear normal, and behave normally, while genetic alterations on the road to cancer are silently evolving. A great deal of the time and changes required for tumor cell formation may be completely undetectable. It is totally unpredictable which cells will actually progress to cancer. Normal appearing, non-tumor cells adjacent to breast cancer and at distant sites frequently show evidence of genetic alterations. [83] [84] Many, if not most types of cancer develop from a field of abnormal cells. [85] Often multiple cancers will arise in the same patient. Prostate cancer in a patient typically arises at multiple sites with in the prostate gland.

The vast majority of pre-cancerous cells and lesions do not progress to cancer. Even when clones and small colonies of cancer cells do evolve, most will die and become extinct. This colony extinction can result largely from statistical factors [86] However, the opposite also holds. Colonies can become firmly established and grow into bulk tumors.

Yet another layer of complexity in cancer: the example of EGFR

The problem of cancer is not just the overwhelming stochastic complexity and diversity that arise from tumor cell evolution. There is yet another level of complexity. The machinery of cells, the biochemistry of life is enormously complex. One example will suffice to illustrate the problem. There is a receptor protein called EGFR that is on the cell surface. When activated by binding to EGF, the protein triggers cell division. Many cancers over-express EGFR, or have mutant forms that are always active. The drug Iressa binds to EGFR and blocks its action. A small number of patients with lung cancer will respond dramatically to Iressa, with tumor shrinkage. However with time resistant cancer cells evolve and the cancer progresses.

The biochemical pathways inside cells that are involved with EGFR signaling have been extensively studied. The complexity of the interactions is mind-boggling. The diagram below from The Johns Hopkins Human Protein Reference Database illustrates some of the proteins and circuits involved in EGFR signaling. [87]

The complexity is just staggering. EGFR sits like a grain of sand at the top of a large sandpile. Disturbances at many other points along the way can arise and give the same net result as EGFR activation, cell division. From what we know about tumor cell evolution, we can be confident that virtually all-possible sites of activation downstream of the EGFR will be exploited by some cancer cells, in some patients, at some point in time. The analogy to the sandpile, although metaphorical, was chosen because the interconnections of cellular gene expression networks are described by a power law statistic, typical of systems that exhibit self-organizing criticality. [88]

Hormone independent cancers

The same type of situation occurs in prostate cancer with respect to the androgen receptor and the estrogen receptor in breast cancer. Normal prostate cancer cells require androgens for growth. Normal breast tissue requires estrogens. Most of the time cancers from these sites are hormone dependent and require androgens or estrogens respectively. With disease progression hormone dependency is frequently lost. The number of biochemical mechanisms by which this occurs is staggering, but not surprising given the complexity of the underlying biochemical pathways.

The evolution of androgen independent prostate cancer

It is instructive to consider the case of prostate cancer. Prostate cancer initially is androgen dependent. Like normal prostate cells the cancer cells initially require androgens for growth. Castration, or the use of drugs that block androgens are very effective, at least initially in controlling prostate cancer. However, with time androgen independence can evolve. Currently, androgen independent prostate cancer is incurable. Multiple mechanisms have been shown to result in androgen independence or been strongly implicated in the acquisition of androgen independence. [89] Click here for a partial list.

Evolution and extreme resistance to change

We hope that to have convinced you that evolution is about change, and that in cancer the potential variations and potential complexity that arises from this change is almost infinite. Let’s now consider another important aspect of evolution, stability and resistance to change.

Evolutionary systems (including cancer) can displays features that are so resistant to change as to seem almost eternal. So much for evolution being about change.

Certain DNA sequences and the encoded proteins have remained essentially unchanged for between one to two billion years. During the same time period, the landmass of Pangea fragmented to form the present continents. The Atlantic Ocean was created. Mountain ranges were formed. Tall mountain ranges disappeared by gradual erosion. Yet these genes survived it all. Richard Dawkins gives an excellent example, a protein called histone-H4. [90] This protein has 306 amino acids and is almost identical in peas, cows and other living organisms. The protein differs in peas and cows by only 2 amino acids out of 306. The common ancestor of peas and cows lived about between 1 and 2 billion years ago. That’s how long the gene has survived. Or more precisely, that’s how long the process of natural selection and evolution has conserved the gene. Mutations that arose in the gene were consistently selected against for one to two billion years.

Evolution is much more about conservation and stability than about change

Evolutionary systems in nature are extraordinarily complex and involve millions of interacting parts that function together to promote self-survival and reproduction. The gradual process of variation and natural selection, re-iterated over and over again, generated the complex machinery of life. It took eons of time. There are almost an infinite number of ways that millions of parts can be assembled. Most configurations will be random junk piles that lack the functions required for survival and self-reproduction. The evolution of complex functional machinery is necessarily a very slow, gradual, iterative process. A million years is a blink of the eye on the time scale of evolution, with respect to the generation of complex functional machinery. There is another important factor that causes evolution to be a highly conservative process.

Evolution is a cumulative process. Evolution is about modification of existing life. Extinctions occur. But all known life on earth is related. This gives an evolutionary layering of complexity. New components must fit and interact in a functional manner with existing machinery and provide a selective advantage. The interdependency and interconnectedness of biological systems makes it difficult to change some existing components. It is like the pieces of an interlocking puzzle. To change the shape of one piece requires altering the shape of neighboring pieces. This in turn could require changing the shape of their neighbors, and set off a kind of chain reaction. All of which is improbable when the source of change is random variation. Some components are so important and interact with so many other critical parts that their change is almost impossible, without impairing survival. This is the case not only for histone H4 but also for a large number of evolutionarily conserved genes and proteins.

Constraints to tumor cell evolution

All cancers have a limited period of time to evolve. Even the evolution of colon cancer, which evolves over decades, is severely constrained by time. The evolutionary journey from normal colon stem cell to cancer cell typically involves several thousand cell divisions. [91] That is enough to generate astronomical tumor cell diversity, to enable resistance to virtually any single cancer therapy, and to preclude a reductionist understanding of cancer. But it is not enough time to create extensive new cellular machinery. It is not enough time to revise or replace large numbers of essential, evolutionarily conserved genes and proteins that have survived for hundreds of millions of years or more. This may seem contradictory or paradoxical. It is not.

Cancer is evolutionary, but constrained. Tumor cell evolution has limits. The fundamental cause of these limits is the short time available for tumor cells to evolve.

What can be known about cancer?

It is time to ask a very basic question. Given that cancer is an astronomically diverse, unpredictable, stochastic, evolutionary process:

What can be known about cancer?

Objection

What relevance does this have to the real world and to cancer? It is philosophy and metaphysics, not science. You can’t experimentally determine the answer. Cancer research is an experimental science. An enormous amount is already known. There are already 1.7 million scientific publications on cancer, 300 known oncogenes , hundreds of tumor antigens... The rate of new scientific articles on cancer is about 150,000 per year. So much new data are being generated that it’s become necessary to use arrays of super computers to handle the terabytes [e] of information. The NCI just invested $60 million to CaBIG, a huge computer network to handle all the data. With systems biology and gene arrays we can now look at the expression of virtually every gene in a tumor, at the same time. “As of Dec. 2004, more than 300 studies have been published representing 10,000 human tumors and 200 million data points.” [102]

Isn’t it a bit ridiculous to ask:

What can be known about cancer?

Response 

To the contrary, this is the single most important question to ask, given the overwhelming evidence that cancer is an enormously diverse, unpredictable, stochastic, evolutionary process. We pose this question not for rhetorical purposes, not to criticize the hard work of thousands of brilliant and dedicated scientists. We pose this question because the answer is the key to the specific cure or control of cancer. Only known or knowable properties of cancer can be targeted.

Before we proceed to answer this question it is necessary to take a side excursion to examine the structure of knowledge and the logical basis of scientific knowledge.