Saturday, February 14, 2015

Molecular Medicine

Down the rabbit hole: what medicine will be like in the future.

The primitive nature of current medicine is usually hidden behind white lab coats and alien machinery. But typically, the knowledge of what is wrong with you is sorely lacking. Cancer treatment consists of shots in the dark, killing all growing cells with horrible poisons that in most cases still do not resolve metastisis and only add a few months of sickened life. More broadly, all sorts of syndromes from diabetes to cronic fatigue, depression, and autoimmune processes, to name a few, have very murky causes and again only trial and error treatments that palliate more than cure. The dizzying round-about of nutrition and diet advice is a symptom of this pervasive ignorance.

One might see the progression as going from big to small: from the wilds of intuition, with its often cosmic / astrological concepts, to scientific anatomy, to the advent of microbiology and cellular biology. The next step is molecular, where the real foundations of biology lie. Whether we have truly mastered the microbiological level is open to question, of course, with the recent measles and ebola outbreaks, the continued scandal of hospital-acquired infections, and our thoughtless use of antibiotics in animals.

What would a better world look like? First, we would know what we are seeing and doing, not just sort of, in a we-gave-it-a-name kind of way, but in a mechanistic, engineering sense. Second, we would have technology to truly address the many defects that can arise out of our systems, at those causal points, rather than farther down the line. Wouldn't it be better to re-instruct cancer cells to become normal again, rather than killing them and all their more or less distant relations? The second goal is far more difficult than the first, but depends entirely on the first being fulfilled.

Molecular medicine (some call it "precision medicine") will be the way to address these issues, and has four elements. First, everyone's genome will be sequenced as a matter of course and be a core part of the personal medical record. Then, wayward growths, infections, fluids, and other samples will be not just subject to the ever more advanced lab tests for metabolic and histologic evaluation, (second element), but will also be sequenced and compared with that genome reference sequence (third element) to nail down molecular alterations that lie at the root of many diseases.

But what would these sequences tell us? However advanced and cheap the sequencing technology, it does no good without knowledge behind it- the biological models of which gene does what, what pathway causes which disease, which mutation causes what effect. That is the fourth and most important element. The war on cancer was only a down-payment on this knowledge, funding the profusion of molecular biology scholarship and technique that has blossomed since the discovery of the structure of DNA in 1953. Unlike the two sequencing elements, which have discernable end-points in terms of creating complete, inexpensive sequences from our DNA samples, this third is essentially open-ended and ever-developing. It may take another hundred years to work out in a full engineering sense this (alien) technology that is human biology.

All this came to mind when reading a recent paper on the properties of one human mutation. A gene called HOXB13 is about 4,000 nucleotides long and encodes a protein that is 277 amino acids long. If you knock out its function completely, the organism is dead ... the protein is essential for early development. (Mice were used to find this out, thankfully.) But less severe defects, like, say, the mutation of deoxycytidine (C) 407 to deoxyguanosine (G), which causes the encoded protein to change at position 84 from glycine (G) to glutamate (E), have far more subtle effects, which require population studies, statistics, and much else to figure out.

Such studies say that the mutation raises the holder's chances of developing cancer, about five-fold in the case of prostate cancer. The authors give a chart of some of the other cancers that have been studied:

Effect of the position 84 mutation of HOXB13 on various cancers. The frequency in column 4, which is the rate of finding this mutation in those who have cancer, can be compared to a general population frequency of this mutation of about 0.45% The higher frequencies generate Odds ratios > 1 which denote a positive association between the mutation and the cancer, and the p-value estimates from the sample sizes how significant this odds ratio, or correlation, is (lower is better). For prostate cancer, not listed, the odds ration is about 4.5, with very well-attested significance.
The odds ratio is perhaps the most easily interpretable statistic, being the odds of some effect, given the hypothesized cause. It is a simple ratio of the rate of the effect in the "with" population over the "without" population. A value of 1 means that there is no connection, and higher values mean that there is a positive correlation. Here, the odds ratio of 3.3 for kidney cancer claims that having the mutation raises the odds of this cancer. For prostate cancer, the odds ratio is 4.51. The P-value is a helpful associated statistic, which tells you how much confidence to place in the odds ratio, since a small population in the study, for instance, may create a very skewed odds ratio, with little significance. The lower the P-value, the greater the significance. So when we get down to oral cancer and leukemia, the association with this mutation is negligable in all respects.

The mutation is very rare, occurring about 0.3% in European populations, and most prevalent in Northwestern European / Russian populations. The rarity is doubtless because of its bad effects, killing its bearers at significantly higher rates than the norm. But it may have other, conceivably beneficial effects- so much is not known. This gene is part of the extremely interesting HOX group that are transcription activators that help the body interpret where its parts are, and activate organ growth during development. This one is strongly turned on in the embryonic tail and urogenital system, including the prostate.

Expression locations of HOXB13 in the embryonic mouse.  UGS = urogenital system. HG =  hindgut.

So, why isn't there a definitive effect? Why are only risks increased, and all these statistics deployed? What could we do to gain a more accurate prognosis? It is likely that if the other three billion nucleotides in our genomes are put through a similar analysis, covering each of their four possible alleles(!), we could gain much better predictive value for each person's genome as a whole. The present statistics were gathered over a population that is essentially random with respect to every position other than this single one, and it may well be that this cancer-promoting effect arises from the interaction between several more or less rare mutations, or at least from biological settings that are more specific than just a random sample, either in genetic terms or other environmental respects. Drug development and treatment is heading in this direction, focusing on specific genotypes that have the most to gain from a particular drug, even if in a totally random population with the disease at issue, that drug may have little discernible positive effect.

But additionally, knowledge doesn't obviate stochasticity / randomness, which is unavoidable in biology as it is in any other complex process. Cancers arise from environmental insults, chemical accidents, and behaviors as much as from innate genetics. In future medicine, the presence and effects of our living conditions would be visible by way of thorough biological testing, taking the typical blood or urine test to a new level of insight to assess what the immune system has seen, for instance, whether nascent cancers have released a few cells, or what subtle imbalances the metabolic system is dealing with; even what you had to eat for lunch.

Once, say, a cancer is detected, (in a very early stage, given a far more sensitive fluid and tissue testing system), the knowledge that a patient has mutations like the one above would inform treatment, which could be assembled from a shelf full of gene-specific medicines that shut off or turn on each encoded protein respectively, as the data indicate, or even create novel activities. We may someday even be able to re-program the DNA of our cells, directly correcting this and the dozens of other mutations that will have conspired to form that particular cancer.


  • What would you pay for a drug that actually cures?
  • Our brains are not our own.
  • Seminar on how the brain flushes & refreshes during sleep.
  • Federal debt is OK, and austerity is madness. Greek austerity is even worse.
  • The state of SDI and disability.
  • On the obsolescence of top-down capitalism.
  • The conscience of a conservative.
  • The unholy mess in Argentina.
  • "From 1960 to 1991, violent crime rose by 400 percent ..."
  • Terror is not entirely a foreign concept, especially to conservatives.
  • Are we seeing euthanasia of the rentier, at least on interest rates, if not on capital income?
  • On the role of history in economics.
  • "... those who have actually investigated the issue demonstrate that it is overwhelmingly the lenders and their agents who put the lies in “liar’s” loans."
  • Containment of Russia, version 4.9.2.0
  • Fossil fuels- is it love, or is it just high-maintenance?
  • Graph of the week: unemployment trends in selected countries.

No comments: