Wednesday, January 29, 2014

Age of disease onset is telling us something

Age of onset and what it tells us about disease
We tend to think of diseases as either genetic or not genetic, in the latter case due to infection or self-inflicted exposure to things we know we shouldn't do like smoke or overeat, or just plain bad luck.  Well, except that hundreds of millions of dollars have been spent trying to find genetic causes for these 'not genetic' diseases, to almost no avail.  But maybe there's a way to expand our understanding of causation a bit, by taking into account information we've had for a long time but don't always consider, either because we don't think to or because we don't know what it means.

Many or even most traits, including diseases and disorders but normal traits as well, have characteristic age of onset patterns, albeit with more or less variation around the mean.  Cystic fibrosis is usually manifest in infancy, while Huntington's disease usually begins to show its effects in middle age, although with a lot of variation.  Most cancers have typical ages of onset patterns, whether or not they are familial forms of the disease.  Epilepsy can first occur at any age.  Puberty generally starts at, well, puberty, and menopause in middle age.

And age at onset may well tell us something about causation -- for example, early onset can (though need not) indicate a strong genetic contribution to the disease, as with dementia or colon or breast cancer, for example.  But even diseases associated with clear, strong genetic risk, such as Huntington's disease, or familial breast cancer, can have delayed and variable age of onset, for both known and unknown reasons.  These diseases are not inherent, not present at birth, and this means that something has to happen in addition to the expression of a faulty gene to cause disease.

Age at onset for most disorders shows 'smooth', regular patterns in populations -- the disorders don't just wait until some age and then pop up.  This regularity suggests a pathogenic process at work, rather than just a 'cause'; it can indicate the extent to which exposure to environmental factors contributes to risk, even when there's a strong genetic component.  The genetic variant doesn't just 'give' you the disease, but instead it affects the pathogenic process.  The figure shows the regular increase, but with details specific to each organ site, for cancers.  Inherited risk typically increases the slope of the risk function, that is, accelerates risk to higher levels more rapidly than in persons who haven't inherited elevated risk.


Cancers strike at different ages, though with characteristic increase with age (log-log scale) Weiss, KM, Chakraborty, R, 1989.  

The general idea is the process is in part due to the accumulation of mutations, during life, in the tissue involved.  If the disease requires some number of such 'hits', but you've inherited one or more, you need to be exposed to fewer hits than someone who inherited none, and it thus takes less time for the disease to develop in you.  However, the details show that this is not a totally simple story by any means.

Retinoblastoma
Retinoblastoma is a childhood tumor of the retina of the eye.  It is found in children with a genetic predisposition in the RB1 gene named because it was discovered in this context, though of course its natural function is not 'for' eye cancer.  Children can be born with one or more tumors in one or both eyes, but it also occurs in children without familial risk.  Age at onset is generally before three, younger in children with the familial form of the disease.  The onset patterns of this disease triggered Al Knudsen in about 1970 to propose the two-hit model of cancer -- his idea was that tumors develop when two different genes suffer harmful mutations.  In this case, as it turned out, the key factor was two 'hits' in RB1, one in each copy that the fetus carries (one copy inherited from each parent).  After childhood, retinal cells don't grow and aren't really vulnerable to being 'transformed' into cancer precursors.  If you inherit two good copies it is most unlikely that any single retinal cell will have the bad luck to experience mutational damage to both copies, so spontaneous RB is rare.

But if the fetus has inherited one damaged and one good copy, there are enough millions of retinal cells ('retinoblasts') that it is likely that more than one, and/or at least one in each eye, will suffer this single new hit required to transform the cells.  Since the disease appears more or less at birth, the idea was that this is a two-hit, single-gene type of cancer.  There are other cancers with early childhood onset that seem mainly to involve one gene.  (How just one gene gone awry can lead to something as complex as cancer is generally not known, though there are some ideas.)

Familial breast and ovarian cancers
By contrast, some variants in the BRCA1 and 2 genes strongly raise the risk of breast and ovarian cancers, and even though age of onset is generally younger in women who inherit one of the risk variants than in women without a clear family history, tumors still don't develop for decades after puberty, when breast tissue begins to grow. This is because it requires numerous mutational changes, either inherited or acquired during a woman's lifetime, before any single breast cell is transformed and then divides and divides, forming a cancer.  Presumably exposure to hormones after menarche, during pregnancy and lactation, which also stimulates division of breast cells, perhaps components of the diet, and other possible risk factors all increase risk, and exposure to these factors varies among at-risk women.  And not all familial cases are early onset -- some only strike their at-risk victims at the more typical ages for non-familial breast cancer.

In addition, though they can confer very high lifetime risk, even in women who inherit a seriously damaging BRCA1 or 2 gene variant, this by itself is not enough to cause disease, as 10 to  40% or so of women with a variant associated with disease won't in fact have developed breast cancer by age 70. Indeed, BRCA1 and 2 are not cancer genes per se; they don't cause cells to divide uncontrollably.  They are genes that code for proteins that repair damaged DNA, or destroy cells that can't be repaired, preventing them from becoming cancerous.  So, in a very real sense, even if it's genes gone awry that directly lead to cancer, it's environmental exposures -- age of menarche, number of births, breast-feeding, dietary factors -- that one might say enable cancer to arise in these at-risk women.  So life-history as well as random exposure factors can affect the age at which such a partially genetic trait can arise.  And cancers are not the only example.

Psychiatric diseases
Many psychiatric diseases characteristically strike in late adolescence; schizophrenia, bipolar disease, severe depression and so forth, though with a lot of variation in onset age.  Why this is isn't at all clear.  The genetic contribution to these sorts of traits is not likely to be a two-hit kind of model -- genetic risk triggered by an interacting environmental factor -- because if genes are an important contributor to these traits, they seem to involve multiple genes, each with small effect, as well as, presumably, environmental factors.  The puberty effect is, one would think, due to changes in gene expression and cell behavior that happen then, but how or why this works is not known.  However, to the extent that genes are involved, the process must be more complex than just mutations in genes, and indeed finding major relevant genes for many psychiatric diseases has been a largely fruitless search.

Some, however, such as many severe intellectual impairments that are essentially present from birth are, not surprisingly, largely due to variants of single genes that affect normal brain development and function.  Again, this is consistent with a process involving a mix of effects in the delayed-onset cases, and a more direct developmental-genetic effect in the cases present at birth.  Supporting this is that in the latter instances we can often find the responsible gene.

Huntington's disease
Huntington's is very clearly a genetic disease, with a known cause, at least a predominant and highly predictable cause, in the number of 3-nucleotide (CAG) repeat sequences in part of the 'Huntingtin' gene.  The number of copies of this triplet repeat had for some years been thought to be highly predictive both of the occurrence and the onset age of HD.  The reason was not really understood, but in any case onset usually is after age 40 or 50 -- though it can be as early as infancy or as late as old age -- and with such strongly predictive gene variants it was curious that their effect could take so long to be manifest in a serious way. What process is involved is not yet clear, even though mouse models do exist that can be studied experimentally.

Even more perplexing is that recently careful study of more ample data have shown that the onset-age/copy-number relationship is far less clear than had been originally thought, and various modifier genes have been identified that may have an effect.  Still, why the genetic anomaly is present at birth and yet the disease doesn't manifest for decades isn't entirely clear, as far as we know.  Neurological damage of some kind clearly first needs to accumulate.  According to a 5 year old paper, it may have to do with oxidative stress.
Although no one specific interaction of mutant huntingtin has been suggested to be the pathologic trigger, a large body of evidence suggests that, in both the human condition and in HD mice, oxidative stress may play a role in the pathogenesis of HD. Increased levels of oxidative damage products, including protein nitration, lipid peroxidation, DNA oxidation, and exacerbated lipofuscin accumulation, occur in HD. Strong evidence exists for early oxidative stress in HD, coupled with mitochondrial dysfunction, each exacerbating the other and leading to an energy deficit.    
This is, however, as much guessing as knowledge.  One obvious consideration is whether HD is actually multigenic.  The genomic background -- variation in many other genes -- may affect the speed at which the neural problems arise and become more widespread during the victim's life.  The lifetime risk in those inheriting a damaged Huntingtin gene is manifest in ways that depend on other genes with which ht interacts in the brain.

Not too long ago, type 2 diabetes, the non-insulin dependent form of the disease, was typically a disease of middle age.  Indeed, it was called 'adult' diabetes.  But once again, the water becomes murkier, because even if there can be a strong if generally unidentified genetic component, environmental risk factors clearly have a major effect because children are now getting this disease in large numbers, presumably because of the sharp rise in childhood obesity.  Some, at least, of the physiological mechanisms are known.  One can hypothesize that if obesity were controlled not only would there be a large drop in the life time risk, but also a reversal in the age of onset to much later once again.  Those who, while not obese and eating healthfully, still got early type 2 diabetes would be expected to have strong genetic risk factors.  But that's hard to find when so much of our population overeats and under-exercises.

But what about when we can't explain it?
The above are diseases for which age at onset can be fairly convincingly explained, at least in general terms.  A disease can develop in utero or not long after birth because genetic susceptibility is pretty much all it takes, or the disease doesn't develop for decades because environmental exposures must accumulate, or must interact with genetic risk (anomalous protein, anomalous error correcting, and the like).  In such cases, a disease or risk of occurrence may be accelerated by genetic risk, but we think it's not accurate to say it's 'caused' by it.  The genetic variant 'contributes' to the risk process.  Indeed, there are people with 'causal' variants of many even devastating diseases who are disease-free.  How frequent this is is unknown, but accumulating information from the ever-increasing number of whole genomes of healthy people makes it clear that it's not insignificant.  Most of us seem to be 'at risk' of at least one genetic disease we haven't got.  

But, again, there are what seem to be single-gene diseases that are latent until, say, puberty even though the abnormal protein is present at birth.  The periodic paralyses are an example.  Rare apparently single-gene disorders, this family of diseases causes attacks of weakness, or partial to total paralysis, under specific, often predictable conditions -- heat, dietary triggers, illness, exercise, rest after exercise and so forth.  We've written before about our theory that English poet Elizabeth Barrett Browning may have had this disorder.  

There are three forms of periodic paralysis; hyperkalemic and hypokalemic periodic paralysis, and Andersen Tawil.  All involve disruptions in how cells respond to voltage differentials across the cell membrane, which are determined by potassium and sodium concentrations on both sides of the membrane.  Attacks are often sudden and severe, but can be interrupted with either an infusion of potassium or of sugar, usually oral, depending on which form of the disorder the person has.  

Variants in three different ion channel genes have been linked to these disorders.  Each variant causes the ion channel -- or rather, the millions of occurrences of the ion channel in skeletal muscle cell membranes throughout the affect person's body -- to respond aberrantly.  Usually a person has only one copy of the defective gene, so that half of the ion channels coded for by this gene would be normal, and half aberrant.

But still, half the ion channels have been defective since birth.  Why do so many people with these disorders not experience weakness or paralysis until puberty?  What is it about being awash in hormones that sets things off?  Age at onset must be telling us something about how this, and other diseases with a similar onset pattern, are triggered.  It's just not clear what.  And, equally perplexing, why do some family members have the presumptive causal variant and never have attacks?  A very similar story can be said of many forms of epilepsy, which involve ion channel genes expressed in neural cells.

And then, what causes puberty?  That is, what triggers the cascade of hormones that results in such dramatic physical growth and change 10 or 12 years after birth?  Or in recent decades, as early as age 8, a sharp decrease that would suggest that there are at least some environmental triggers involved.  Genes that trigger hormonal secretions have been identified (e.g., the hypothalamic Kiss1 system described here and here), but what triggers the triggers a decade or so after birth?

The age at which a trait or disease is manifest has a lot to say about the relative importance of life experience in the disease.  And this can be true whether or not the trait or disease is thought to be due to a single gene, many genes or no genes.  

No comments: