Saturday, May 18, 2013

War on cancer: update from the front

Some promising, and frustrating, data from the genomics revolution.

We might each very soon get our genomes sequenced, and this will provide a wealth of information about our ancestry as well as our susceptibility to many diseases and other conditions. This is quite static data ... get sequenced once, and your medical file is set for life- those basic facts are not going to change, even if our ability to interpret those genetic sequence facts is growing by the day and will continue to grow for decades, if not centuries.

But cancer is different- it is a genetic disease, a matter of mutations that waylay the normal course of cellular management from its what's-best-for-the-organism discipline to a descent into a mad Darwinian greed. To really tell what is going on, each cancer would have to be sequenced. Like HIV, whose mutations continue as the disease progresses, evading each drug hurled at it in turn, cancer mutations accumulate over time in cancer cells as well, making a dynamic genomic landscape.

Science magazine recently ran a magisterial, long, and unusually clear, review of cancer genomics. While sequencing individual cancers is not yet routine clinical practice, (other than for a few select markers), for research purposes it has been going on for some time, and we now have mountains of data. The authors made quite a few interesting points.

Sequence any cancer, and you get a mess. The tissues are heterogeneous, full of normal and mutated cells. The cancerous cells are a dog's breakfast of early and late cells, with some people theorizing that relatively few "cancer stem cells" are the real replicating drivers, and most of the other cells in the tumor in various stages of stasis or death. Even when you isolate the real, core, fastest-growing cells, they are again a mess, full of mutations that have nothing to do with the problem of cancer.

Indeed, the authors mention that genome sequences from highly mutagenized sites like lung cancers of smokers have ten times the number of mutations as those from lung cancers from non-smokers. Which gives you some idea of the incredibly mutagenic drive that smoking constitutes, and how much mutagenesis it takes to dramatically increase cancer incidence. It takes a lot of hits, and even then some smokers live to a ripe old age.

Tumors vary tremendously in their scale of gross mutation, from only a handful in an entire genome (common in pediatric cancers) to ten to a hundred in most types of tumors, up to a thousand or more in the most mutation-rich tumor of all, colorectal cancer.

So after a great deal of work, researchers have screened out all the noise and the garbage and come up with the genes that really drive cancer, out of our genomes of 23,000-odd genes. And this is the good news- there are only, roughly, 138 "cancer genes" responsible, in some mutated or altered state, for every known case of cancer that has been analyzed. Each tumor typically has a handful of these, which it has accumulated extremely slowly, over many years.

These genes tend to encode master controllers of the cell cycle, cell survival, cell differentiation, and DNA damage repair. For instance, ATM encodes a protein that senses DNA damage and halts the cell cycle in response. Obviously the kind of gene you want on your side, but one that gets in the way of cancer progression. It is frequently mutated in leukemias and lymphomas.

The 12 general classes of the 138 genes whose mutation or overproduction drives cancer growth. Some positively drive growth, while most are inactivated from their normal function of inhibiting cell growth.

The bad news is that few of these genes are easily targeted by drugs. The majority of these 138 drive cancer by virtue of being mutated into inaction, which is to say that they are tumor suppressors in their normal state. The typical gene mutation truncates these proteins- the remnant folds badly when it is made and is promptly tossed into the cellular recycling bin. There is little a drug can do for (or against) a protein that is not doing anything or is absent. Only when we have true gene therapy reliably injectable into these (highly inaccessible) cells would such a defect be truly fixable.

The ones that can be effectively targeted by drugs are oncogenic enzymes which are overproduced or specifically mutated into overactivity. The Ras kinase is a classic example, where a specific mutation of codon 12 or 13 from glycine to another amino acid renders this signalling protein deaf to upstream pathways that turn it off, by inactivating an enzymatic function that constitutes its "reset" switch. It becomes an always-on signaller, telling its cell (falsely) that external growth factors are always there, so go ahead and grow, grow, grow.

This is the kind of thing that can be targeted with drugs, not to turn the protein's reset switch back on, but to block its other actions so that it no longer does harm. This KRAS gene is mutated in about 30% of human cancers, so one can appreciate the usefulness to a cancer cell of having a good deal of mutagenesis going on, perhaps via another mutation in the DNA repair machinery, since this specific defect would otherwise be extraordinarily rare- much harder to come by than a truncating mutation.

The authors hold out hope that, since each of the un-druggable tumor suppressor gene products function in larger cellular pathways of control, other proteins can be found downstream from these inactivated tumor suppressors that might be usefully targeted by drugs:
"All of the known driver genes can be classified into one or more of 12 pathways (Fig. 7). The discovery of the molecular components of these pathways is one of the greatest achievements of biomedical research, a tribute to investigators working in fields that encompass biochemistry, cell biology, and development, as well as cancer. 
We believe that greater knowledge of these pathways and the ways in which they function is the most pressing need in basic cancer research. Successful research on this topic should allow the development of agents that target, albeit indirectly, defective tumor suppressor genes. Indeed, there are already examples of such indirect targeting."

Unfortunately, the fact that there are so few core driver genes for cancer, itself militates somewhat against this view. If there were so many pressure points in the pathways of cellular control, we would see more of them reflected in oncogenesis. By all means, we need to gather all the knowledge we can, but magic bullets are going to be hard to come by.

The bottom line is that cancer, while far more complicated than the singular word naively indicates, still has an underlying "muta-genetic" pattern that can be used for definitive diagnosis in the coming molecular age, where genomes and individual cancers will be sequenced as a matter of routine. Once we devise maybe a couple hundred magic bullets to various oncogenes and related pathways, we may be able to treat cancer on an individualized basis much like HIV- with a customized cocktail of several drugs that, in combination, will forestall recurrence indefinitely. Currently, there are maybe twenty such drugs, many of which have poor efficacy or other issues, not to mention astronomical expense, so we have a long way to go.

A related point from this paper is that metastasis does not seem (at current knowledge) to involve novel or special mutations. The authors observe that cancer takes decades to develop, slowly accumulating its growth-promoting mutations, and that cancers slough off circulating cells in prodigious numbers, more so the larger they are. Thus a careful diagnosis of the original tumor, or any decendent, should suffice to characterize a cancer completely, and to stop it no matter how disseminated, given the specifically tailored and combined drugs that are envisioned above.


  1. Silver bullets, eh? Forty three years after Nixon declared War on Cancer?

    "The authors observe that cancer takes decades to develop, slowly accumulating its growth-promoting mutations....."

    Here's a v interesting alternate view from Paul Davies, wherein these growth promoting mutations are seen to come from somewhere else, back in Deep Time. I commend it your attention:

  2. Hi, Croc-Chuck-

    Thanks for reading, and for sending a very interesting link. Their proposal is not complete nonsense, but it also does not differ much from the "rogue cell hypothesis" either. They state:

    "Our central hypothesis is that cancer is an atavistic state of multicellular life. Atavisms occur because genes for previously existing traits are often preserved in a genome but are switched off, or relegated to non- coding (‘junk’) segments of DNA."

    Well, expecting such functions to lie dormant for a billion years or more is a lot to expect. I think it is more reasonable to think that our cells still have lots of cell-autonomous functions, with a layer of cooperative control on top. Being rogue means reverting to a more primitive state, whether that is single-cell or some messy version of simple multicellular colonialism doesn't make very much difference. Both are very far from the current design of multicellularity. Tumors get very poor angiogenesis, cooperate very poorly with other cells, throw off lots of dying cells- it is not an optimized state, other than possessing the ability to grow, which is, willy-nilly, the only reason we care about them.

    "Cancer can be triggered in a wide variety of ways, but once it becomes established it is extremely hard to reverse."

    Well, this is a pretty empty statement. My post above is about how the triggers are not a wide variety, but a surprisingly limited set of ways, which might be seen to consitute the core of metazoan 2.0 overlay controls to keep cells down on the farm, as it were. The difficulty of reversing cancer is the whole point of the "rogue cell" view, and of our limited molecular technology.

    "The first is the failure to account for the rather high degree of cooperative organization among cancer cells. The most striking example of this is angiogenesis, in which an entire tumor builds its own blood supply for the common good of all the tumor cells. A more contentious example concerns evidence that a small population of highly malignant cancer cells can be held in check by less malignant cells."

    This does not seem like a helpful view, to me. Angiogenesis is a clearly high-level metazoan function, and tumors do it very poorly, in confused and inefficient fashion. That is why they rely on glycolysis rather than respiration, for the most part, and typically have lots of necrosis and other problems. The colonial 1.0 metazoans like volvox or planaria have no blood supply.. it is irrelevant. As for mutual inhibition in large tumors ... that is far more likely due to apoptosis and cell death and lack of nutrients, than it is any orchestrated atavistic regulation. Why do people die from cancer??

    "Why don’t the vast majority of mutations in tumor cells lead to mal-adaptation and death, as is the case for healthy cells?"

    Well, they do- these authors haven't evidently been paying attention. They are a bit like theo-evolutionists who think of a ladder of life leading smoothly to the pinacle of it all- us, ignoring all the death and contingency along the way.

    Anyhow, there are many reasonable connections between these author's views and the more mainstream view of cancer, particularly regarding the toolkit of metazoan-specific controllers that go awry in cancer. Learning more about metastisis will probably be the key area of research that will differentiate between these views.

  3. Burk

    I found the paper v interesting, and note parallels of 'metazoa 1.0' and prokaryotes, eg, 'quorum sensing' of bacteria in colonies, and the switching off of their 'regular' genes and switching on of others, when they are present in sufficient numbers to form biofilms.

    So this inter-cellular 'cooperativity', albeit primitive, seems to be a characteristic of all life. Think of fungi and slime molds - another example.

    I look forward to Biophilia every week, and your 'Links' set. Keep up the great work.

  4. Thanks very much! I did enjoy the metazoa 1.0 paper. It is a fascinating question whether cancer cells are typically cooperative in any significant & distinct way.

  5. I think its more fundamental than your reply indicates. I invoke Lynn Margulis' endosymbiosis theory-her original 'case' was the mitochondrion; today there is emergent thinking that all eukaryote cellular organelles owe their existence to this phenomenon.