There are some myths, bits of misinformation, or lies about medicine that I like to refer to zombie quackery. The reasons are obvious. Like at the end of a horror movie, just when you think the myth is finally dead, its rotting hand rises out of the dirt to grab your leg and drag you down to be consumed. Of course, the big difference between zombies and these bits of zombie quackery is that in most stories a single shot to the brain will kill the zombie. The same is not true of zombie quackery. You can empty clip after clip of reason, science, and logic into the “head” of the zombie quackery at point blank range, and the best you’ll do is to drive it away for a while, only tor rise up again when you least expect it.
Of course, antivaccine pseudoscience, in my experience, is one area of quackery that is rich, if not the richest, in zombie quackery and zombie memes. The same old lies keep popping up again and again and again, like Whac-A-Mole. Sure, they’ll sometimes go away for a while (or appear to go away for a while), but sooner or later the exact same misinformation, occasionally with minor alterations. Think the claim that the CDC conspired at Simpsonwood to “hide” that thimerosal in vaccines causes autism, a myth first popularized by antivaccine icon Robert F. Kennedy, Jr. back in 2005 whose rotting corpse recently been resurrected to shamble about like so many extras on The Walking Dead. One of the advantages of having been at this blogging thing for nearly a decade is that I’ve come to recognized many of these zombie memes immediately on sight. Where other people think they’re new, I recognize them as something old, often something I’ve written about before at least once, if not many times. The disadvantage, on the other hand, is a tendency to become jaded or bored with refuting the same nonsense over and over again. Sometimes I marvel that in December it will have been a decade since I started blogging and sixteen years since I started refuting online nonsense.
The latest zombie meme struggling to make a comeback is a particularly brain dead one, even by antivaccine zombie meme standards. It comes in the form of a press release on ChristianNewsWire for a new “study” (and I do use the term loosely, even though apparently it was published in a peer-reviewed journal) entitled New Study in Journal of Public Health and Epidemiology Correlates Autism Disorder Increase and Human Fetal DNA, Retroviral Agents in Vaccines. The first thing that you should notice about this press release is that it is on ChristianNewWire, whose news sources consist mainly of—you guessed it!—fundamentalist Christian and conservative Catholic organizations, with very few legitimate scientific organizations.
The next thing you should notice is who put out this press release: Katie Doan of the Sound Choice Pharmaceutical Institute. A quick click rapidly reveals that the SCPI believes vaccines cause autism and that it’s somehow related to “fetal DNA” in vaccines, complete with a list of “aborted fetal product”- This, you might recall, is part of another antivaccine zombie meme, namely the claim that vaccines are made using “aborted fetal tissue.” This comes from the simple fact that a human cell line originally derived from an aborted fetus decades ago is used to grow the viral stocks used to make some vaccines. This is such a non-issue that even the Catholic Church says it’s acceptable to use these vaccines because “the risk to public health, if one chooses not to vaccinate, outweighs the legitimate concern about the origins of the vaccine” and because parents “have a moral obligation to protect the life and health of their children and those around them.” That doesn’t stop radical antiabortionists from trying to represent cells hundreds of cell divisions removed from the original fetus from which they were derived as somehow being “fetal tissue” or “fetal parts,” rather than what they are: A cell line.
A second part of this antivaccine zombie meme is that it is in actuality DNA from these “fetal cells” that somehow gets into human neurons, recombines with the DNA there, producing foreign proteins that show up on the surface of the neurons and provoke an immune response, thus damaging the neurons. I’ve already explained in my usual painful detail how utterly ignorant of biology and homologous recombination one has to be to accept this hypothesis as anything other than incredibly implausible at best, with no evidence to support it, to boot.
So what are the hypothesis and conclusion of this “study” being touted? Let’s take a look at the press release and then go to the study itself:
A new study published in the September 2014 volume of the Journal of Public Health and Epidemiology reveals a significant correlation between autism disorder (AD) and MMR, Varicella (chickenpox) and Hepatitis-A vaccines.
Using statistical analysis and data from the US Government, UK, Denmark and Western Australia, scientists at Sound Choice Pharmaceutical Institute (SCPI) found that increases in autistic disorder correspond with the introduction of vaccines using human fetal cell lines and retroviral contaminants.
Even more alarming, Dr Theresa Deisher, lead scientist and SCPI founder noted that, “Not only are the human fetal contaminated vaccines associated with autistic disorder throughout the world, but also with epidemic childhood leukemia and lymphomas.”
Theresa Deisher? Where have I heard that name before? Oh, right. Here. She’s the founder of SCPI and has been laying down the serious stupid about “fetal DNA” in vaccines since at least 2009, when I first noticed her. It explains much about why the press release also mentioned the whole “CDC whistleblower” manufactroversy being flogged by the antivaccine movement right now. It also explains why this press release cherry picks information from an FDA presentation from 2005 by Keith Peden at the Division of Viral products on Issues Associated With Residual Cell-Substrate DNA.
Funny how they fail to note that 106 or 107 μg of cellular DNA would be needed to produce an oncogenic event, and that the oncogenic risk for 2 ng DNA would be around 5 x108 to 5 x 109. In actuality, it’s even higher than that, around 7.5 x1013 to 7.5 x 1014, given the size reduction of DNA to around 200 bp fragments that occurs. The authors in the study itself claim that there is anywhere from 142 ng to 2000 ng of “fetal DNA,” but, one notes, that they didn’t do PCR on this DNA to prove that it was fetal DNA, rather than DNA used to make the virus, nor did they show any gels demonstrating the claimed size of the DNA fragments. All they did was to use an ELISA for single- and double-stranded DNA and called it a day. From a molecular biology perspective, this is not nearly enough to prove that the DNA they are measuring, even assuming they are using the assay kits correctly, is in fact fetal DNA. Particularly amusing is this passage:
Notably, the viruses in the Meruvax, MMRII, and HAVRIX vaccines are mRNA viruses, not DNA viruses, and since the mRNA was degraded by heat treatment prior to oligonucleotide measurements, the DNA results are indeed specific for human DNA, the only DNA in the mRNA virus vaccines.
Of course, one wonders whether Deisher et al took the specificity of PicoGreen into account. Its specificity for dsDNA over RNA is not perfect; indeed, take a peak at the graph here. It’s clearly at least 100=fold more sensitive to dsDNA at 520 nm than it is for RNA, but remember, there’s a lot of RNA in a concentrated solution of RNA virus relative to the contaminating dsDNA. True heating the RNA will result in its degradation, but not as much as Deisher et al apparently think. Anyone who’s done plasmid preps the old fashioned-way before columns existed to remove RNA contamination knows that a lot of small RNA fragments remain in such preps, even after heating. If Deisher et al had really wanted to measure only DNA, they should have treated these vaccine vials with RNAse to guarantee that no small fragments of RNA were left. In any case, there’s no evidence presented that these small amounts of DNA are dangerous, much less that they have anything to do with autism. Remember, we’re talking about nanogram quantities, at most a microgram or two, injected intramuscularly. As I explained, the thought that such a tiny amount of DNA could cross the blood-brain barrier and get into neurons in sufficient quantities to actually recombine with host DNA sufficiently to cause neuroinflammation is incredibly implausible. To really do this rigorously would have required measuring the RNA, dsDNA, and ssDNA in several vials, and then to subject some of the vaccine solution to PCR using appropriate primers to prove the source of the RNA and DNA.
Of course, none of this really matters, at least for purposes of this study, because the investigators never demonstrate that there is a correlation between the introduction of “fetal human cell”-containing vaccines and increased rates of increase of autism prevalence. True, the authors do the mother of all studies confusing correlation with causation, complete with a lot of linear regressions between autism prevalence in multiple datasets and the uptake of vaccines such as Varivax and Hepatitis A for specific birth cohorts. It’s an excellent demonstration of the adage that if you look hard enough you can fit almost any data in a linear regression.
Then there’s the change point analysis. A change point is, as it sounds, a point where the slope of a curve changes suddenly, most commonly seen when one curve consistent with a straight line “changes slope.” Of course, the assumption that the relationships between change points between time and autism prevalence are linear is a rather dubious assumption right off the bat. Of course, as Mark Chu-Carroll put it:
One big catch here is that least-squares linear regression produces a good result if the data really has a linear relationship. If it doesn’t, then least squares will produce a lousy fit. There are lots of other curve fitting techniques, which work in different ways. If you want to treat your data as perfect, you can use different techniques to progressively fit the data better and better until you have a polynomial curve which precisely includes every datum in your data set. You can start with fitting a line to two points; for every two points, there’s a line connecting them. Then for three points, you can fit them precisely with a quadratic curve. For four points, you can fit them with a cubic curve. And so on.
Similarly, unless your data is perfectly linear, you can always improve a fit by partitioning the data. Just like we can fit a curve to two points from the set; then get closer by fitting it to three; then closer by fitting it to four, we can fit two lines to a 2 way partition of the data, and get a closer match; then we can get closer with three lines in a three way partition, and four lines in a four way partition, and so on, until you have a partition for every pair of adjacent points.
The key takeaway is that no matter what you data looks like, if it’s not perfectly linear, then you can always improve the fit by creating a partition.
Which appears to be exactly what Deisher et al did, as I described. Mark Chu-Carroll also described how using this sort of “iterative hockey stick” analysis is a completely inappropriate analysis for this sort of data, as well as how it is very difficult to identify real change points without massive data sets and a very dramatic change in slope, neither of which qualify here. Basically, this study appears to be more of the same thing as the last study; I look at it as the previous “study” put out by Deisher, only on steroids of stupid. My main thought was this: It took them over four years to produce this after their last study? Really?
Not surprisingly, Deisher et al reported change points very similar to what they found last time, too: 1980, 1988, 1996. These are the same change point years as last time, give or take at most a few months. I can’t resist recycling what I wrote last time, because it’s so freakin’ appropriate, although I’ll spare you quoting it exactly.
Apparently the tainted DNA from ground-up murdered babies is so powerful in causing autism that it can do so immediately. These “change points” correlate within a year to the introduction of vaccines made from ground up cells from murdered babies. For example, the rubella vaccine was approved in the U.S. in 1979, and the first changepoint detected was in 1980. The second dose of the MMR vaccine was added to U.S. recommendations in 1988, and in 1988 there was a changepoint. Then the chickenpox vaccine was recommended in 1995, and there was a changepoint in 1996. Autism is usually diagnosed between ages 2 and 4; so, unless the power of these evil tainted vaccines to contaminate the DNA of our precious children can also travel back in time, it’s hard to take correlations between these change points and vaccine introduction as anything more than spurious pseudo-correlations. It would be so hilarious if the consequences of such fear mongering weren’t so dire, although even then it’s still useful as a cautionary tale worthy of extreme mockery of how not to do linear regression and inflection point “hockey stick” analysis.