XMRV - Innocent on All Counts?

A bombshell has just gone off in the continuing debate over XMRV, the virus that may or may not cause chronic fatigue syndrome. Actually, 4 bombshells.

A set of papers out today in Retrovirology (1,2,3,4) claim that many previous studies claiming to have found the virus haven't actually been detecting XMRV at all.

Here's the rub. XMRV is a retrovirus, a class of bugs that includes HIV. Retroviruses are composed of RNA, but they can insert themselves into the genetic material of host cells as DNA. This is how they reproduce: once their DNA is part of the host cell's chromosomes, that cell is ends up making more copies of the virus.

But there are lots of retroviruses out there, and there used to be yet others that are now extinct. So bits of retroviral DNA are scattered throughout the genome of animals. These are called endogenonous retro-viruses (ERVs).

XMRV is extremely similar to certain ERVs found in the DNA of mice. And mice are the most popular laboratory mammals in the world. So you can see the potential problem: laboratories all over the world are full of mice, but mouse DNA might show up as "XMRV" DNA on PCR tests.

Wary virologists take precautions against this by checking specifically for mouse DNA. But most mouse-contamination tests are targeted at mouse mitochondrial DNA (mtDNA). In theory, a test for mouse mtDNA is all you need, because mtDNA is found in all mouse cells. In theory.

Now the four papers (or are they the Four Horsemen?) argue, in a nutshell, that mouse DNA shows up as "XMRV" on most of the popular tests that have been used in the past, that mouse contamination is very common - even some of the test kits are affected! - and that tests for mouse mtDNA are not good enough to detect the problem.

• Hue et al say that "Taqman PCR primers previously described as XMRV-specific can amplify common murine ERV sequences from mouse suggesting that mouse DNA can contaminate patient samples and confound specific XMRV detection." They go on to show that some human samples previously reported as infected with XMRV, are actually infected with a hybrid of XMRV and a mouse ERV which we know can't infect humans.
• Sato et al report that PCR testing kits from Invitrogen, a leading biotech company, are contaminated with mouse genes including an ERV almost identical to XMRV, and that this shows up as a false positive using commonly used PCR primers "specific to XMRV".
• Oakes et al say that in 112 CFS patients and 36 healthy control, they detected "XMRV" in some samples but all of these samples were likely contaminated with mouse DNA because "all samples that tested positive for XMRV and/or MLV DNA were also positive for the highly abundant IAP long terminal repeat [found only in mice] and most were positive for murine mitochondrial cytochrome oxidase sequences [found only in mice]"
• Robinson et al agree with Oakes et al: they found "XMRV" in some human samples, in this case prostate cancer cells, but they then found that all of the "infected" samples were contaminated with mouse DNA. They recommend that in future, samples should be tested for mouse genes such as the IAP long terminal repeat or cytochrome oxidase, and that researchers should not rely on tests for mouse mtDNA.

They're all open-access so everyone can take a peek. For another overview see this summary published alongside them in Retrovirology.

I lack the technical knowledge to evaluate these claims, no doubt plenty of people will be rushing to do that before long. (Update: The excellent virologyblog has a more technical discussion of these studies.) But there are a couple of things to bear in mind.

Firstly, these papers cast doubt on tests using PCR to detect XMRV DNA. However, they don't have anything to say about studies which have looked for antibodies against XMRV in human blood, at least not directly. There haven't been many of these, but the paper which started the whole story, Lombardi et al (2009), did look for, and found, anti-XMRV immunity, and also used various other methods to support the idea that XMRV is present in humans. So this isn't an "instant knock-out" of the XMRV theory, although it's certainly a serious blow.

Secondly, if the 'mouse theory' is true, it has serious implications for the idea that XMRV causes chronic fatigue syndrome and also for the older idea that it's linked to prostate cancer. But it still leaves a mystery: why were the samples from CFS or prostate cancer patients more likely to be contaminated with mouse DNA than the samples from healthy controls?

Robert A Smith (2010). Contamination of clinical specimens with MLV-encoding nucleic acids: implications for XMRV and other candidate human retroviruses Retrovirology : 10.1186/1742-4690-7-112