Vertebral variation, Hox genes, development, and cancer


First, a tiny bit of quantitative morphological data you can find in just about any comparative anatomy text:

mammal number of vertebrae
cervical thoracic lumbar sacral caudal
horse 7 18 6 5 15-21
cow 7 13 6 5 18-20
sheep 7 13 6-7 4 16-18
pig 7 14-15 6-7 4 20-23
dog 7 13 7 3 20-23
human 7 12 5 5 3-4

The number of thoracic vertebrae varies quite a bit, from 9 in a species
of whale to 25 in sloths. The numbers of lumbar, sacral, and more caudal vertebrae also show considerable variation. At the same time, there is a surprising amount of invariance in the number of cervical vertebrae in mammals — as every schoolkid knows, even giraffes have exactly the same number of vertebrae in their necks as we do. What makes this particularly striking is that other vertebrates have much more freedom in their number of cervical vertebrae; swans can have 22-25. I was idly wondering why mammals were so limited, and stumbled onto a couple of papers that addressed exactly that question (Galis & Metz, 2003; Galis, 1999). Galis’s explanation is that it is a developmental constraint that may have something to do with the incidence of cancer.

Development is an intricately choreographed process that treads a dangerous line. On one side is stability; but development is in many ways a destabilizing process, in which cells have to change their path and form new tissues, and stability is not compatible with it. On the other side is chaos, unregulated proliferation — cancer. During development, the organism has to foster proliferation and change to a greater degree than it can tolerate later, and that loosening of constraints represents a danger. Galis suggests that one reason we mammals may always have 7 cervical vertebrae is that the regulatory genes that specify the number of vertebrae are coupled to processes that otherwise regulate cell fates, and that modifications to those genes that would cause variation in vertebra number would also lead to unacceptable increases in the frequency of embryonal cancers.

This isn’t at all an improbable idea. Genes exhibit bewilderingly complex patterns of expression, and pleiotropy (the regulation of multiple phenotypic characters by a single gene) is the rule, not the exception. The Hox genes, the particular genes that control the identity of regions along the length of the animal, are known to switch on and off in proliferating mammalian cell lines in culture. Perhaps the Hox genes involved in defining cervical vertebrae are somehow also involved in controlling cell proliferation, making them dangerous targets for evolution to tinker with?

Galis provides several lines of evidence that this is the case. To see whether variation in cervical vertebra number leads to increased incidence of cancer, we need to look for instances of variation in mammalian vertebrae.

There isn’t much variation in cervical vertebra number, though. There is an exception: sometimes, the 7th cervical vertebra is found to undergo a partial homeotic transformation and forms a pair of ribs, which are normally found only on thoracic vertebrae. Humans develop cervical ribs with a frequency of about 0.2%; do they also develop cancers? The answer is yes, with a frequency 125 times greater than the general population.

Another place to look would be in phylogenetic variation — between groups rather than within a population. It turns out that there are two groups of mammals that do have a non-canonical number of cervical vertebrae: one manatee genus and two genera of sloths. No data is available on frequencies of embryonal cancers in either, and Galis reports that manatees at least seem to have a low incidence of cancer. One explanation is that both sloths and manatees have exceptionally slow metabolic rates, which in itself will reduce the frequency of cancer, since it will reduce the rate of oxidation damage; the idea is that this low cancer rate may have made these organisms more tolerant of variation in these genes.

An open question is how birds can have greater variability in the number of cervical vertebrae — they certainly don’t have low metabolic rates. One suggestion is that the coupling between these particular Hox genes and a predilection for cancer is unique to mammals. Another possibility is that birds possess other, unidentified mechanisms that reduce free radical production, reduces oxidative damage, and makes them relatively cancer-free. Galis cites several studies that show that birds do seem to be less severely afflicted with cancers than us mammals.

It’s an interesting idea, but the evidence so far is a collection of correlations. I’d be interested in seeing some direct analyses of the role of patterning genes on carcinogenesis. Still, it’s the first answer I’ve seen to explain why such a peculiar restriction in morphology should be nearly universal within a whole class of animals, when other classes allow so much more diversity.


Galis, F and JAJ Metz (2003) Anti-cancer selection as a source of developmental and evolutionary constraints. BioEssays 25:1035-1039.

Galis, F (1999) Why do almost all mammals have seven cervical vertebrae? Developmental constraints, Hox genes, and cancer. J Exp Zool (Mol Dev Evol) 285:19-26.


  1. says

    Wow! I never thought of it this way.

    Birds also have tremendously powerful immune systems compared to mammals and that may help in stemming some of the cancers, perhaps?

  2. JDLee says

    When you mention sloths and metabolic correlations with oxidative damage, you seem to be focusing on possible Hox-controlled DNA repair mechanisms. But oxidative DNA damage is not likely to be a significant cause of embyonic cancers; these are mostly the result of misregulation of cell growth and proliferation mechanisms, and Hox genes are definitely involved. While I don’t personally work on Hox genes, and I haven’t read the papers that you cite, I’m pretty sure that there’s a large body of work published on Hox gene knockouts, which might provide experimental evidence for the link that you’re suggesting. It should be straightforward to survey that literature for reports of increased cancer rates. There may even be data linking Hox genes to cell cycle pathways or other proliferation controls. Cheers,


  3. Nick says

    Following up to Jeff’s comment. Ribs on the seventh cervical vertebra would be a posterior transformation. If I remember correctly, knockout of Hox genes tends to cause anterior transformations. Posterior homeotic tranformations are often due to loss of polycomb-group genes which normally inhibit HOX activity. There seems to be a fairly rich literature on polycomb-group genes as oncogenes or tumor suppressors and their role in controlling cell proliferation.

    See, for example Raaphorst, F.M. (2005) Deregulated expression of polycomb-group genes in human malignant lymphomas and epithelial tumors. Hum Mol Genet 14(suppl 1): R93-R100.

  4. jesst says

    There is some evidence to suggest there might be something odd going on in the giraffe neck – see Solounias, N. 1999. The remarkable anatomy of the giraffe’s neck. J.
    Zoology 247:257-268.
    “Mammalian cervical vertebrae 6 and 7 and thoracic vertebra 1 possess many distinguishing characteristics. In the giraffe, bone morphology, muscle origins and insertions, as well as the location of the brachial plexus (described as many osteological and some soft tissue characters) are identical to those in other mammals but are all displaced posteriorly by one vertebra. There are two exceptions to these observations: the pre-sacral vertebral count is unchanged when compared with that in the okapi and C7 supports the first rib. Thus, one vertebra has been added in the neck of the giraffe between cervical 2 and 6, and some type of structural blending has occurred in the region of the first rib. The junction of the giraffe neck with the thorax is unusual and results in a protruding forelimb. It is possible that the unusual position of the neck relates to balancing of a cantilevered neck and head upon a relatively slight body. Characteristic drinking postures may have also influenced the observed anatomical modifications.”

    Has anyone looked at cancer rates in giraffes?

  5. lannejhang says

    I have just had a quick look at the Bioessays article and don’t find this correlation convincing at all. They only correlate high cancer with a transformation of the 7th cervical vertebra into a rib. Yet, the variations you see in the number of other vertebrae, like the thoracic vertebrae of horses and cows, are most probably not the result of a transformation, but of an addition of vertebrae. So the common variation seems to be addition, and NOT transformation of vertebrae. And this point is not addressed in the article.

  6. Nick says


    Is there a meaningful distinction between transformation and addition of vertebrae? Somites appear and differentiate starting at the anterior end and proceeding posterior. Their differentiation into different types of vertebrae is presumably controlled by overlapping sets of Hox and other genes. One way to add a thoracic vertebra would be to increase the size of the field that expresses the “thoracic” genes and push the “lumbar,” “sacral,” and “caudal” fields more posterior. This would transform the identity of the somites that normally form the first lumbar, sacral, and caudal vertebrae. There would be no change in the number of lumbar and sacral vertebrae, an increase in the number of thoracic vertebrae, and a decrease in the number of caudal vertebrae Since the number of caudal vertebrae is so variable anyway, the net effect of those transformations would be addition of a thoracic vertebra.

  7. says

    I should say that this is a very phenomenological study — they don’t know anything about the actual causal connections between patterns of Hox expression and cancer rates.

  8. says

    My first guess about conservation of cervical number would have been that it had something to do with nerve branching. As I recall, the nerves that branch from the cervical vertebra are strictly ordered and have well-understood, and biologically vital, destinations, as evidenced by what happens when you break the spinal cord in a particular place.

    Maybe our cervical number is conserved because if it isn’t, we get the wrong nerves in the wrong places and the resulting embryo isn’t viable.

    But that’s Aristotelian biology, and hence useless. I’m perfectly happy to read a theory backed up by actual data.

  9. Chilean says

    So, maybe an alteration can produce both cancer and vertebral alterations. Another alteration can cause both mental deficiency along with polidactyly. Doesn’t mean they are inseparable. Manatees and sloths show they aren’t.
    Are we going to come up with other hypotheses of cloaked adaptationism like this for everything? the five legs of the starfishes, the fix multiple of petals, the 8 rhombomeres…I guess a clever hypothetical and irrefutable trade-off can always be imagined, huh?
    Let’s hope some room will remain for the possibility that purely structural explanations exist

  10. says

    I notice a small bug:

    old_pharyngula.gif has a title that says it was posted on October 28, 2003. The link goes to the old Pharyngula site that says it was posted on October 28, 2003

  11. miko says

    It seems that alteration in vertebrae number would be more likely to involve shifts in Hox gene expression domains, not gains or losses in Hox biochemical functions (a Hox knockout or change in activity). This scenoario is harder to reconcile with a possible Hox/cancer connection.

    As a counter example, you have all these developmental patterning molecules and their receptors (Wnts, hedgehogs, etc), that definitely contribute to cancer cell behavior, but many of these control highly labile mammalian features (limb morphology, e.g.).

    On the other hand, I remember a recent hypothesis that Hox gene expression might be regulated through a special form of chromatin remodelling that explains the correlationship between the gene locations in Hox clusters and their expression domains along the A-P axis. Maybe defects/alterations in chromatin regulation underly some Hox expression changes; this would seem consistent with a cancer link.

  12. says

    Having first proposed the radical idea that “cancer selection” — lethal cancer in developing animals — played a central role in the evolution of Bilaterians in two letters published in the Journal of Theoretical Biology in 1983 and 1984, and having expanded on that idea in my book Cancer Selection: the new theory of evolution which was published in 1992, I’ve been gratified to note that it is now receiving some recognition, including a perfunctory citation in the 2003 Gallis and Metz paper.

    More information is available at my website. (The site does not yet reflect comments and references to my idea which have either been published recently or are scheduled for publication in coming months.)

  13. says

    Funny, just the other day I found out that I have the extra cervical ribs. I don’t know if it’s related to the herniated disc that brought me to the radiology department in the first place, but it’s at least plausible.

  14. Marilu says

    I have a full pair of cervical ribs and so does my 3-year-old daughter. My sister and my father each have one. Your article would imply that we might each have an 125% greater chance of contracting some type of cancer, although I gather that these are only hypotheses currently and there aren’t any clinical studies. How worried should I be?

  15. says

    Your article would imply that we might each have an 125% greater chance of contracting some type of cancer, although I gather that these are only hypotheses currently and there aren’t any clinical studies. How worried should I be?

    Hi, Marilu–it’s very difficult to talk precisely about something that is so newly proposed, and–as you note–hasn’t been tested yet. Additionally, cancer is many diseases, and multiplying a low-risk cancer by a factor of 125 is going to still be a lot less risk than it would be for a cancer with a correspondingly high incidence. So it’s hard to talk specifically about something so broad as “cancer”.

    The following figures were taken from and are about breast cancer, but it’s how you would assess it for any kind of cancer.

    During 2000-2004, women aged 20-24 years had the lowest breast cancer incidence rate, 1.4 cases per 100,000 women, and women aged 75-79 years had the highest incidence rate, 464.8 cases per 100,000.3

    If those figures hold at the time your daughter is that age, and if this hypothesis is borne out, and if all other risk factors are equal–a few large assumptions, to be sure–then at ages 20-24, you and she would be expected to have a 175 out of 100,000 chance of developing breast cancer. A lot worse chances than 1.4 out of 100,000 to be sure, but there are still 99,825 chances out of 100,000 that you would not get it.

    At 75-79, again holding all the previous assumptions, you and she might have a 58,100 out of 100,000 chance of developing breast cancer–again, much worse than the 464.8 of people without that same risk, but also a 41,900 chance out of 100,000 of not developing it.

    So how much should you worry? It’s probably worth following in the research literature to see how it develops–does it pan out biologically, or is it just a fluke correlation–and worth making your health-care provider aware of, just to keep in the back of his or her mind. Maybe you start screening earlier than the usual recommended age as a result. But I’d venture to say that if you’re losing any sleep over it, that’s too much worry at this very early stage.

    Note that I am not a physician, and this is not medical advice–just an interested layperson. Hell, I probably got the arithmetic wrong, too–but you get the idea, anyway.

  16. Wolfgang says

    There is a major flaw in Galis’ work, which, in addition was wrongly cited by PZ Myers:

    Galis calculated an excess risk of 120 (not 125!) for neonatal and childhood cancers, not for cancers in adults, if cervical ribs are present. But this calculation has a major flaw and is completely inacceptable: She based her assumption on a work of Schumacher et al. from 1991, but compares apples with pears: Instead taking the cervical rib counts from both, cases and controls, out of the Schumacher study, the took only the cases numbers from Schumacher and compares them with the cervical rib prevalence in the general population from other studies. If she have had taken the Schumacher numbers for both, cases and controls, only an exess risk of 2-6 for certain neonatal and childhood cancers can be calculated. This is congruent with two later studies (Dutch and American) on the same topic as Schumachers.

    So you should relax: The probability for early childhood cancer is rather low, even if the child has cervical ribs!

  17. Katy says

    You took the words out of my mouth, Marilu. I have a full pair of cervical ribs as well. As far as I know, I’m the only one in my family who has any – but I’ll certainly keep an eye on any future children and keep this information in the back of my mind. I’ve got enough health problems to worry about for myself as it is. I don’t need to add any more to my list…lol

  18. grama duria says

    Ihave just find out I have cervical ribs I m a grama How woried should i be for my g kids I habe not told my kids yet Ifigured i I did not Die up to 55y old I shoulddo more reserc on it d

  19. Rosalind Watterson says

    We just found out today that the lump on my 10 month old granddaughter’s left side of her neck is a cervical rib. You can’t really see it but you can feel that it is there.

    My daughter read about the childhood cancer statistics and she is extremely upset. Can someone explain this to us in simple terms how they came up with this 125% chance and how does that relate to our granddaughter. What precautions should we take?

    Next week she is being operated on for a perforated anus. Could this be related?

    She looks like a beautiful healthy baby girl.

    Rosalind Watterson

  20. Kara says

    It’s just been confirmed that my 6 yr. old daughter has two cervical ribs. She’s had a suspicious lump on the left side of her neck/shoulder for 2 years now that we have long thought felt a lot like a bone. She’s had a CT scan and ultrasound to try and identify it. Finally, an x-ray has revealed not one, but two cervical ribs – a matching pair.

    The information on this page is very disturbing, but I wonder if the supposed rarity of this “condition” and its correlation with cancer is accurate. After all, if you don’t feel or see the rib(s), how would you know that you have it (and therefore be part of the statistic)? There may be many more of us walking around with cervical ribs than is known about and with cancer rates being so high, it skews the data — in other words, they only find the ribs after you’ve been diagnosed with cancer, but there may be many others who have them who will never know it.

    Does this make sense to anyone else besides me?

  21. wrdwyz says

    For Kara, from 2008, if it still matters: I am now 68 years old with bilateral cervical ribs, about to have surgery for carotid artery plaque removal, but w/ no sign of cancer.

    For blogger: post no. 25 seems to be spam, care to delete?

    For anyone: I was told from childhood that the ribs impinged on carotids and I might have “problems” in pregnancy. I never became pregnant but wonder if the C7 ribs might have anything to do with blockage in both internal carotids, now 85% on the right, 8mm long above bifurcation of common carotid, at level of C2-3 up to the mandible.

  22. Janine, Mistress Of Foul Mouth Abuse, OM says

    Wrdwyz, people really do not go back and check on years old posts. There are thousands of post to go through if that were the case and Pharyngula is enough of a time sink as it is. As for the spam, that is part of the reason why PZ instituted registration, so he can more easily find and delete the spammers.

    But if you want to bring up this discussion to the general audience, you could bring it up in the undead thread. It is the designated spot for open discussion, any and all topics are welcome.