Peter Duesberg, chromosomal chaos, and cancer: An intriguing hypothesis argued poorly

A lot of readers (well, a couple, anyway) have been asking me about the recent article by Peter Duesberg in the most recent issue of Scientific American entitled Chromosomal Chaos and Cancer. I suppose it’s because I’m not only a cancer surgeon (which in and of itself is not enough to qualify me to comment on this topic) but rather because I’m also a cancer researcher and a molecular biologist (which, I submit, does make me qualified to comment on this topic). Peter Duesberg, as you may know, is the controversial scientist who is perhaps the foremost advocate of the discredited hypothesis that HIV does not cause AIDS. I had been tempted to comment on Dueseberg’s hypothesis based on the orgasmic reaction of Duesberg’s sycophants in the HIV/AIDS “dissident” community to the recent publication of an article on it by Duesberg in Scientific American. One such booster, even went so far as to say:

The one that first comes to mind as particularly relevant to Peter and AIDS is that it does seem impossible that a man who might just be correct concerning something as complicated as the genetic basis of malignancy could be so totally wrong about something as straightforward as whether HIV kills T-cells.

This is perhaps the dumbest thing I’ve heard said about evaluating Duesberg’s aneuploidy hypothesis of cancer. To demonstrate why, let me recast the question a bit to: “The one that first comes to mind as particularly relevant to Peter and AIDS is that it does seem impossible that a man who is clearly so utterly wrong concerning something as now scientifically straightforward as whether AIDS is caused by HIV could actually be correct about something as complex as the genetic basis of malignancy.”

HIV/AIDS denialists would be screaming bloody murder about such a question were I to pose it as anything other than a rhetorical device, and they would have a point to some extent. Yet they think nothing conflating the scientific validity of Duesberg’s ideas concerning cancer, which might indeed be partially or mostly correct, with his discredited hypothesis that HIV does not cause AIDS, implying that because he might be correct about cancer implies that he is correct about AIDS. It doesn’t. Sorry, but the two issues are at best peripherally and weakly related and at most not related at all. That is why I do not plan on discussing why Duesberg’s ideas regarding HIV are wrong, even though the results of such HIV/AIDS denialism leads to quackery and has serious real-world consequences for real people. Tara and Nick have done far more than my minor efforts to critically examine such wingnuttery. Instead, I decided simply to ask the question about Duesberg’s chromosomal chaos hypothesis as the cause of cancer and ask: Is there any “there” there? It’s a question I’ve asked myself before but never written about, and this seemed like an opportune time to discuss the issue.

The first thing that struck me about the Scientific American article is that it looked very much like a popular version of two very similar recent review/opinion articles that Duesberg published in 2005 and 2006. I’m mainly going to discuss the Scientific American article because it’s basically the same message in a form more palatable to the educated lay reader. But, before I begin, I’d like to point out a couple of things. First, the concept that chromosomal abnormalities cause cancer dates back at least to 1914, when the German zoologist Theodor Boveri, based on studies of sea urchin development, first suggested it. Indeed, this featured prominently as a milestone in cancer research in a display in at the recent 100th anniversary meeting of the American Association for Cancer Research. Thus, the basis of Duesberg’s idea is quite old. Indeed, the concept that chromosomal derangements caused cancer predominated for 40-50 years, until the solution to the structure of DNA, the elucidation of the genetic code, and study of genetics led to an emphasis on genetic causes of cancer. Combined with the observation that tumor cells are genetically unstable, leading to many mutations, the genetic hypothesis led to the discovery of oncogenes and tumor suppressors. Now, potential chromosomal causes are again being looked at, and for whatever part Duesberg’s advocacy had in spurring this he is to be acknowledged, even if his boosters do have an annoying tendency to make it sound as though scientists would have zero interest in studying chromosomal causes of cancer were it not for Duesberg, which, given the attention shown to this topic at recent meetings that I’ve attended, is ridiculous.

It has been known for many many decades that most cancer cells are aneuploid. This means that, instead of having the correct number of chromosomes (in the human, 46 chromosomes, two matched sets of 22, plus the sex chromosomes, either XX or XY). In some diseases, such as Down syndrome (known as trisomy-21), there are either missing or extra chromosomes. In the case of Down syndrome, there is a third copy of chromosome 21. Such abnormal chromosomal numbers come about when, during meisosis (cell division that produce germ cells), the newly copied chromosomes don’t segregate properly to the two daughter cells, one to each. Instead, both go to one or the other cell. In cancer cells, the situation is much worse; such missegregation during mitosis can lead to aneuploidy that is much more severe, to the point where some cancer cells can have 70 chromosomes or more. Because certain genetic mutations, for example in DNA damage repair genes, can lead to chromosomal instability that can in turn lead to aneuploidy, the basic argument has been over the relative importance of the roles of aneuploidy and the accumulation of genetic mutations as leading to cancer. On the one extreme, there is the argument that aneuploidy is the primary cause of cancer, causing the accumulation of genetic mutations through breaks in chromosomes. On the other extreme is the argument that aneuploidy is a consequence, not a cause, of cancer. Duesberg, as you may guess, takes the extreme version of the former view. These days, most other scientists studying this question tend to consider both important to varying degrees in the development of cancer, the present pressing scientific question being: Which causes which and how? It’s very much a chicken-or-the-egg problem. Does mutation lead to aneuploidy or aneuploidy lead to large numbers of genetic derangements that lead to cancer? Or are both aneuploidy and mutation responsible in differing proportions depending on the cancer?

Unfortunately, although he describes how tumors evolve under the selective pressures of the organs in which they arise, acquiring the ability to proliferate, evade apoptosis, become insensitive to normal growth arrest signals, and metastasize, Duesberg seems unable to restrain himself from overselling his case and lapsing into straw men arguments about rival hypotheses. He’s very frustrating that way. The concept that aneuploidy may play a major and, in some cases, primary role in carcinogenesis is a legitimate scientific idea with scientific evidence to support it. It doesn’t need to be defended or argued using such bad arguments, but Duesberg can’t seem to help himself. The problem is, Duesberg’s thinking is black-and-white, all-or-nothing. He can’t seem to fathom the concept that both aneuploidy and genetic mutations might feed upon each other. In addition, he also almost totally neglects other evidence implicating other causes or important factors in the progression of cancer, such as cancer stem cells, tumor angiogenesis, or even the metabolic hypothesis ( i.e., the Warburg effect), which, like the chromosomal hypothesis, is also enjoying a resurgence. Cancer is a complex set of diseases, likely with multiple causes contributing to the development of different cancers in different proportions, which is why my skeptical antennae start twitching whenever I hear someone like Duesberg (or anyone else, for that matter) postulate in essence a single cause for all cancer.

Let’s look at what Duesberg argues. In essence, he argues that aneuploidy comes first and is the prime inciting event that starts the cascade of genetic changes that lead to malignancy. DNA is damaged, either through mutagens or other causes, and then, through what becomes a self-catalyzing process, aneuploidy leads to progressive chromosomal alterations that lead to increasingly widespread genetic alterations in a process that feeds on itself, leading to chromosomal instability and cancer. Indeed, Duesberg postulates that carcinogens work as “aneuploidogens” rather than as mutagens.

i-5f42bcb19bb3d952ab2279b12c019eb1-464314423_92c662218e.jpg

i-492d73aca52eb3c05b23bebeb14e92ba-464314427_d8d6570628.jpg

Grrlscientist has posted a concise summary of Duesberg’s arguments here, and Hank Barnes has a copy here. (I’m guessing that Duesberg or Harvey Bialy gave him the PDF.)

So, here we have a scientific hypothesis with a moderate degree of plausibility based on what we know. What arguments can Duesberg marshal in its favor? Disappointingly, most of his arguments leave much to be desired. Let’s start with Duesberg’s first argument, which is so bad that it needs to be quoted extensively to be appreciated:

Cancer risk grows with age. Lamentably common, cancer afflicts about one in three people at some point in their lives, but mostly after the age of 50, which is when chances for malignancy soar. Thus, cancer is, by and large, a disease of old age. The gene mutation theory of cancer’s origins, however, predicts that the disease should be quite common in newborns. If, as that hypothesis holds, about half a dozen mutations to critical genes were necessary to ignite malignancy, certainly some of those mutations would accumulate like SNPs over the course of generations in the genomes of many individuals. A baby could thus inherit three of six hypothetical colon cancer mutations from her mother, for example, and two from her father and be at extremely high risk of cancer from picking up the missing sixth mutation in any one of her billions of colon cells. Some babies might even be born with colon cancer from inheriting all six hypothetical colon cancer mutations from their parents. But colon cancer is never seen in children. Indeed, even laboratory mice intentionally engineered to carry an assortment of ostensibly carcinogenic mutations from birth can live and propagate happily, with no higher risk of developing tumors than normal lab mice.

[…]

Interestingly, among the rare exceptions to cancer’s age bias are children with congenital aneuploidy, as in Down syndrome, or with inherited chromosome instability syndromes, such as the disease known as mosaic variegated aneuploidy (MVA), which also causes severe mental retardation. Defects of mitotic spindle assembly in the cells of children with MVA produce random aneuploidies throughout their bodies, and nearly one third develop leukemia or unusual solid cancers.

Being born aneuploid, or prone to aneuploidy, clearly accelerates processes that lead to cancer. Indeed, the inherent instability of aneuploid cells would explain why most aneuploid embryos, as Boveri observed 100 years ago, would not be viable at all and thus why newborns are cancer-free and cancer is not heritable.

.

This argument is a strawman and neglects other factors, to boot. For one thing, contrary to what Duesberg states, the “gene mutation theory of cancer” does not necessarily predict that cancer should be quite common in newborns. Quite frankly, I don’t know what Duesberg was smoking when he wrote that and don’t see his rationale for arguing that. For one thing, it’s not true that colon cancer is never seen in children. It’s quite rare, to be sure, but in families with genetic mutations that predispose to colon cancer, the disease can appear in children as young as 5 or 6. I presume that Duesberg picked colon cancer because the genetic sequence of mutations that occur as the lining of the colon goes from dysplasia, to polyps, to noninvasive cancer, to invasive cancer was worked out so elegantly by Burt Vogelstein back in the early 1990’s. In addition, it is true that patients with Down syndrome are more susceptible to certain cancers, specifically hematopoietic (blood) cancers, but it also turns out that they are less susceptible to solid tumors, thanks to high levels of the antiangiogenic factor endostatin, the gene for the precursor of which happens to reside on chromosome 21 and is expressed at a higher level because there are three copies rather than two. Second, Duesberg neglects to notice that oncogenes and tumor suppressors cause dysregulated growth and morphogenesis, the sine qua non of cancer. Although it is true that deleting some tumor suppressors results in viable mice who appear normal other than a predisposition to cancer, such potent growth dysregulation could cause embryonic lethality just as easily as aneuploidy could, depending on the specific gene.

Indeed, although it’s not as obvious, Duesberg’s making in essence the same mistake as Dr. Egnor: confusing selective mechanisms in germ line cells with those in somatic cells. A basic consideration of evolutionary theory makes it obvious that there would be major selective pressure against mutations that result in cancer in babies and children, because the cancer would kill the child before reproductive age. In contrast, cancer-predisposing genes that are either neutral before reproductive age or confer some advantage (as has been postulated, for example, for some mutations in the BRCA1 gene, which predisposes to breast cancer), may not be selected against and could remain in the germline at a high frequency in the population. Thus, from a strictly evolutionary perspective, it is not at all surprising that inherited cancers are rare in newborns and young children, even if we postulate, for the sake of this argument, that aneuploidy has nothing to do with cancer and the “genetic mutation theory” of cancer is 100% correct. Duesberg should know this, given his invocation of selective pressures leading to increased aneuploidy in cancer.

Finally, Duesberg tries to argue that mutations don’t occur at a sufficient frequency under normal conditions to account for the increasing rate of cancer with age, but does not address the shortening of telomeres or that it may only take one or two mutations in key genes, such as those involved in DNA damage repair, to make the a single cell deficient in repairing its own DNA, which can then lead to an increased mutation rate, leading both to mutations and increased aneuploidy. Indeed, it has been postulated that the prerequisite for some cancers is what is termed a “mutator phenotype,” in which a much higher rate of mutation is observed. In any case, aneuploidy is easier to detect than mutations, assays for which can only look at a small fraction of the genome at one time and are not sensitive enough to detect mutated cells in a background of normal cells. In contrast, small numbers of aneuploid cells can be detected in a normal background. The bottom line is that chromosomal instability is a feature of virtually all cancers but the evidence that it is driven by primarily aneuploidy is conflicting and nowhere near the slam dunk that Duesberg seems to think; in essence, it’s unclear whether it is primarily aneuploidy that drives mutation or primarily mutation that leads to aneuploidy.

Here’s the next bad argument:

Carcinogens take a very long time to cause cancer. Numerous chemicals and forms of radiation have been shown to be carcinogenic in animals or established as the source of occupational or accidental cancers in humans. But even the strongest carcinogens at the highest survivable doses never cause cancer right away. Instead the disease emerges only after delays lasting years or even decades. In contrast, when substances known to cause gene mutations are administered to bacteria, the cells begin displaying new phenotypes within hours; in larger organisms such as flies, the effect is seen within days. A gene mutation scenario therefore does not explain why cells exposed to carcinogenic agents become cancer cells…

Does anyone see the flaw in an argument comparing humans to bacteria or flies in this manner? Let’s look at flies, because they are eukaryotes. The average lifespan of, for example, Drosophila is much shorter than a human’s, on the order of 30 days or so. Carcinogens generally require cellular replication before cancer can develop. So, let’s see, a latency period for cancer after exposure to carcinogens of few days in the life of a fruit fly like Drosophila is not unlike a latency period of a couple of decades in a human, if you compare it to the organism’s overall life span. Bacteria reproduce amazingly rapidly; so it is not surprising that they respond to chemicals even faster. As for strong carcinogens not causing cancer right away, nothing in the genetic mutation theory of cancer demands that they must, particularly given that strong doses may result in more deleterious mutations and that the ability of a normal cell to repair its own DNA is quite prodigious. It may ultimately be shown experimentally to be true that aneuploidy is a better explanation for the long latency period of human cancers, but there is nothing in the mutation theory that demands that there must be a short latency period after a tumor cell is exposed to a carcinogenic agent, especially since it is now understood that “multiple hits” are usually required occurring in a single cell to result in cancer.

Here’s the next Duesberg argument:

Carcinogens, whether or not they cause gene mutations, induce aneuploidy. Scientists have looked for the immediate genetic effects of carcinogens on cells, expecting to see mutations in many crucial genes, but instead have found that some of the most potent carcinogens known induce no mutations at all. Examples include asbestos, tar, aromatic hydrocarbons, nickel, arsenic, lead, plastic and metallic prosthetic implants, certain dyes, urethane and dioxin. Moreover, the dose of carcinogen needed to initiate the process that forms malignant tumors years later was found to be less than one-thousandth the dose required to mutate any specific gene. In all cases, however, the chromosomes of cells treated with cancer-causing doses of carcinogens were unstable–that is, displaying higher than usual rates of breakage and disruption.

Sure, carcinogens induce aneuploidy, but just because some carcinogens do not directly damage DNA does not necessarily mean that the induction of aneuploidy must be the mechanism by which they cause cancer. It might be, but it doesn’t necessarily have to be. For example, part of the mechanism by which asbestos is thought to cause cancer is by causing chronic inflammation, which chronically stimulates nearby cells to divide, while at the same time exposing them to a milieu containing numerous inflammatory cytokines secreted by white blood cells that invade the area. In the case of dioxin, for example, there appears to be a receptor to which dioxin binds that stimulates cell survival pathways. Either of these mechanisms, and others, could be at play in the carcinogenesis due compounds that do not cause extensive DNA damage and could account for the long latency period between exposure to carcinogen and cancer.

The rest of Duesberg’s arguments range from the “so what?” to the more intriguing. For example, he points out that different patterns of aneuploidy are seen in different tumors. Given the known genetic heterogeneity of cancers, this is not at all surprising and doesn’t necessarily mean that aneuploidy is the primary cause of cancer. For example, the Philadelphia chromosome, the result of a reciprocal translocation, an exchange of genetic material, between chromosomes 9 and 22. It is very common, virtually pathognomonic for chronic myelogenous leukemia. This is associated with the excessive production of an oncogene (Bcr-Abl), which activates cell cycle genes and induces genomic instability. Certainly, there must be some sort of aspect to the structure of these two chromosomes that make this particular translocation as common as it is, but the end result that drives CML is a mutation that results in the excessive production of an oncogene. Indeed, molecularly targeting this oncogene has resulted in a very effective drug against CML called Gleevec. In fact, targeting single gene abnormalities, although it hasn’t resulted in a cure for cancer, has resulted in several very effective treatments, such as Herceptin, which is directed against the Her-2/neu oncogene. Finally, lots of single gene mutations in transcription factors, DNA repair enzymes, or cell signaling molecules can alter the expression of dozens or hundreds of downstream (or even thousands) of genes in predictable ways. Large scale chromosomal breaks are not necessary to account for such globally deranged gene expression.

Duesberg next argues that “gratuitous traits do not contribute to cancer survival,” referring specifically to genes for metastasis and drug resistance, oddly enough. The problem with this argument is that a subset of the genes for carcinogenesis are not infrequently also genes involved in metastasis. Similarly, genes involved in drug resistance often have other functions than just drug resistance. For example, the mdr1 multidrug resistance gene product is an ATP-dependent channel that extrudes a variety of substances, not just chemotherapeutic agents, from a cell, an excellent example of how evolution can coopt the function of an existing protein to do another function. Finally, Duesberg appeals to the ability of cancer cells to change their phenotype rapidly, supposedly much faster than genes can mutate. This may indeed be true, and it may indeed be possible that aneuploidy contributes to this, given its ability to cause wholesale rearrangements of chromosomes and hundreds (or even thousands) of genes in one fell swoop. However, all it takes is selection by drug for tumor cells that express ever-increasing amounts of mdr1; chromosomal rearrangements, although they certainly may contribute to drug resistance, do not appear to be strictly necessary for it. Moreover, once again, Duesberg cannot resist overselling his case:

Once this vicious cycle is under way, the fact that every cell would be randomly generating its own new phenotypes could explain an observation made decades ago by Leslie Foulds of the Royal Cancer Hospital in London that “no two tumors are exactly alike … even when they originate from the same tissue … and have been induced experimentally in the same way.” Such individuality is yet another hallmark of cancer that cannot be explained by the activity or inactivity of specific genes, which would be expected to have consistent effects each time and in each cell.

The problem with this argument is that it’s only partially correct. Using the “gene chip,” which now allows scientists to assay the levels of every gene in the human genome at the same time, we now know that tumors often have a surprisingly similar pattern of expression of thousands of genes. Indeed, one of the most startling findings early in the use of such gene chips was that nearly all breast cancer could be divided on the basis of gene chip experiments into a small number of distinct subtypes, the main ones of which include the “basal” (more aggressive) or “luminal” (less aggressive) phenotype. Moreover, the expression of the Her-2/neu oncogene leads to a distinct, identifiable, reproducible gene expression pattern, in direct contradiction to Duesberg’s claims above. Indeed, tests based on such gene chips are already making their way into the clinic to estimate a patient’s risk of recurrence and guide chemotherapy decisions.

Lest one think that I’m hostile to Duesberg’s hypothesis, let me disabuse you of the notion right now. Although I think Duesberg’s an utter crank and pseudoscientist when it comes to his HIV/AIDS denialism, I find some of his work in cancer intriguing, and I disagree with Mark and Larry that it was such a horrible thing to feature him in an article in Scientific American, especially given the disclaimer. It is clear to me that epigenetics (cellular factors other than genes that regulate gene activity) and chromosome structure are very important in carcinogenesis, more so than had been appreciated before. However, contrary to how Duesberg’s sycophants like to portray the chromosomal hypothesis of cancer as an epic battle that’s all about Duesberg, who is portrayed as the lone voice arguing for the hypothesis that aneuploidy is the cause of cancer, in reality it’s nothing more than yet another scientific controversy that is, fortunately, no more nasty than a lot of other controversies in science, such as, for example, the hypothesis that changes in cell metabolism are the cause of, not a consequence of, carcinogenesis. It’s also nowhere near as clear as Duesberg claims whether aneuploidy is a cause or a consequence of carcinogenesis. For one thing, there are at least two examples that I’m aware of (which means there are probably more than that) of groups generating tumor cells that do not have significant or widespread aneuploidy, demonstrating that aneuploidy may not be a prerequisite for carcinogenesis, in direct conflict with Duesberg’s hypothesis. In addition, there’s a very interesting article from Don Cleveland’s lab in the January Cancer Cell that suggests that aneuploidy can promote carcinogenesis under some circumstances (an observation that seems supportive of Duesberg’s hypothesis) and act as a tumor suppressor under others (an observation that is arguably not).

What really irks me about Duesberg with respect to his ideas about cancer is that he may be on to something, but he can’t seem to stop himself from the same black-and-white, either-or thinking that apparently led him down the road of HIV crankery, nor can he seem to resist massively overselling his hypothesis as the be-all and end-all hypothesis to explain cancer initiation and progression. As I said at the beginning of my post, whenever someone postulates theirs as The One True Cause of Cancer, my skeptical antennae start twitching, and Duesberg’s aneuploidy hypothesis is no exception. Cancer is a complex and resourceful foe, not to mention that it’s hundreds of different diseases, not a single disease. Duesberg neglects a variety of other new hypotheses for causes of carcinogenesis that hold equal or greater promise than the chromosomal chaos hypothesis. Among these are cancer stem cells, tumor angiogenesis, and the aforementioned metabolic hypothesis of cancer (a.k.a. the Warburg effect). He even neglects what I consider to be a far more fascinating and sophisticated version of the chromosomal hypothesis, specifically Tom Misteli’s concept that derangements in the higher order three dimensional structure of chromosome territories can lead to cancer by alterations in gene expression.

Duesberg’s supporters may look at the relative neglect of chromosomal structure as a controller of gene expression and a potential cause of cancer and wonder why it was neglected for so long. The analogy I like to make is to politics. It is said that politics is the art of the possible. To me, science is the study of what it is possible to study. Two to three decades ago, we figured out how to study individual genes; so that’s what we studied, even though we soon realized that such reductionist techniques did not give the complete picture of cancer. Less than 10 years ago, gene chips, coupled with improvements in statistical analysis and increases in computing power that made it possible to analyze the data produced from such huge experiments, allowed us to look at the expression of the entire genome at once, leading to a richer understanding of the changes in gene expression that occur during cancer. The more sophisticated techniques and understanding did not invalidate what we had learned before; it complemented and extended it. Similarly, we now have the tools to probe chromatin structure at a level of detail never before possible; consequently we are now looking at chromatin structure in cancer. Our increasing ability to probe the detailed structure of chromosomes will likely now complement and extend what we have learned about cancer through the study of mutations in individual genes. Walter Giaretti asks:

…I would like to pose the question if the “aneuploidy theory” of cancer in relationship with the “mutation theory” still remains as controversial as in the near past. Don’t we have now enough experimental evidence that cancer originates and progresses with the contribution of both gene mutations and aneuploidy?

Duesberg’s failure is that he doesn’t seem willing to accept that the answer to this question is almost certainly “yes.” As Giaretti puts it:

It is likely that new studies directly comparing DNA copy number and gene expression will be performed in the near future on the role of aneuploidy in cancer, on what genetic events may induce chromosomal instability and on the validation of novel criteria for early diagnosis. It is predictable that these studies will vanish the conflicting views that either aneuploidy or gene mutations are a unique cause of the origin and progression of cancer negating the role of the alternative mechanism. Today, these conflicting interpretations are increasingly being abandoned to let a more complex mixed paradigm take over from previous concepts. In brief, ideas stemming from the old Boveri theory and from the modern theories may soon be seen as cooperative and equally important to cancer.

There are indeed deficiencies in our current understanding of cancer initiation and progression, but there’s no reason that gene mutations and aneuploidy couldn’t both contribute to these processes. Indeed, I’d be surprised if it were otherwise. Duesberg seems too dogmatic and wedded to his hypothesis to see the big picture.