Infectobesity: is obesity transmitted through a common viral infection?

Image via Wikipedia

Today’s post comes from Audrey Richard.  You can learn more about Audrey at the bottom of this post.  Want to learn how to publish your own guest post on Obesity Panacea?  Click here


According to the World Health Organization, obesity is preventable. But is it really if we don’t know yet for sure everything that causes it? Is obesity “only” about diet, lifestyle and genetics? Could it be something that somehow you just catch, like a virus? Or, more precisely, could a virus make you “catch” obesity at some point?

A little more than a decade ago, Doctor Nikhil Dhurandhar demonstrated what he had suspected for a long time: a human virus can cause chickens to increase their number of fat cells (which, paradoxically, reduced their cholesterol levels, inducing what Peter has referred to as metabolically healthy obesity). The human virus Dhurandhar investigated was adenovirus 36 (Ad36), one of the many serotypes of human adenovirus, and known to cause respiratory illness mostly. Most interesting is the fact that, besides chicken, Ad36 also increases adiposity in mammals including marmosets and rats, with serum cholesterol and triglycerides being significantly lowered in each case (details on how increasing the number of fat cells – rather than the amount of actual fat in those cells – can actually improve metabolic health here). But what does this human virus do to humans as far as weight gain and obesity are concerned?

Several studies have tried to establish whether the link between adenovirus and obesity in animals existed in humans as well. As it is often the case with observational studies, some say it has an effect, others not so much (here and here for example). So, in the light of the extant literature, should we take for granted that there is a correlation link between Ad36 and obesity in humans and further investigate such a lead? This is one of the questions Doctors Yamada, Hara and Kadowaki are covering in their recent meta‑analysisAssociation of adenovirus 36 infection with obesity and metabolic markers in humans: a meta‑analysis of observational studies” published in PLOS One.

What did they do?

Searches, selection and assessment of validity

To perform this meta-analysis, Yamada and Kadowaki performed several rounds of selection using first Medline and the Cochrane Library to identify observational studies running from 1951 to April 22nd 2012 that tested the association between Ad36 and obesity, as well as body mass index (BMI) and various metabolic markers.

Here are their selection criteria:

Cohort studies, case-control studies, and cross-sectional studies evaluating the relation of obesity to Ad36 infection were considered eligible for inclusion if the following criteria were fulfilled: 1) the full text of report was published in English; 2) event numbers in each exposure category were reported; 3) the presence of Ad36 infection was reported; and 4) obesity events and/or BMI and other metabolic markers were reported.

The authors then examined the quality of the selected studies according to the STrengthening the Reporting of OBservational study in Epidemiology (STROBE) statement, endorsed by many journals. After this whole selection had been completed, of the 237 studies that had been identified, only 10 studies, all cross‑sectional studies published between 2005 and 2012, could finally be included in the meta‑analysis.

Figure 1 from Yamada et al., 2012, PLOS One. Flow diagram of study selection.

Data extraction and quantitative synthesis of data

Doctors Yamada and Hara independently extracted a set of characteristics from the 10 included studies as summarized in Table 1. While the types of parameters reported and the way of defining/assessing obesity could vary quite widely depending on the study and country, the presence of neutralizing antibodies was the marker of Ad36 infection in all cases. Besides, a number of particular points are highlighted:

Atkinson reported two independent studies performed in 502 adults and 28 sets of twins, respectively, in the same paper (Note: which explains why there are actually 11 lines in the Table). The mean BMI of the 502 adults (38.0 kg/m2) was higher than in the other studies. Trovato reported two studies that included patients with non-alcoholic fatty liver disease (NAFLD). The subjects of Broderick study included military personnel on active duty.

Table 1 from Yamada et al., 2012, PLOS One. Summary of the studies included in the meta‑analysis (click to see full size)

The investigators assessed the relation of Ad36 infection to obesity and/or BMI and other metabolic markers by calculating the weighted mean difference (WMD) and pooled odds ratio (OR) according to the DerSimonian‑Laird random effects meta‑analysis method. The authors also analyzed sensitivity (e.g. are the results still the same when the calculation is restricted to studies in just one age group or geographic region?) and sources of heterogeneity. They considered as significant a P value of less than 0.05, except for heterogeneity for which p=0.10 was enough to validate significance. Also, they complied with the guidelines for the meta‑analysis of observational studies in epidemiology and the PRISMA statement, “an evidence­‑based minimum set of items for reporting in systematic reviews and meta‑analyses” which is endorsed by many journals.

What did they find?

The 10 studies conforming to the criteria of selection gather 2, 870 subjects altogether and the meta‑analysis reveals the following:

Figure 2 from Yamada et al., 2012. Association of Ad36 with BMI. (click image to enlarge)

  •  The BMI of people that have already been infected by Ad36 is significantly higher than that of people that have not (N=9), with a WMD of 3.19, meaning that people showing antibodies neutralizing Ad36 have an average overweight of 3.19 kg compared with people with none. This is also true, with the WMD value varying though, when the analysis is restricted to studies focused on adults (N=7), studies in the USA (N=3) and those in Europe (N=4). However, studies performed with children only (N=2) or in Asia (N=2) do not show any significant association of Ad36 infection with BMI.
  • Ad36 infection correlates with a higher risk of obesity (N=6) (e.g. having a BMI > 30 kg/m2) but it is worth noting that P=0.047 “only”. Also, the results varied quite widely when the analysis was performed with respect to studies conducted in USA only, or Europe, or Asia.
  • None of the metabolic markers (triglycerides, total cholesterol, high‑density lipoprotein, blood glucose, waist circumstances and systolic blood pressure) is associated with Ad36 infection, except for low‑density lipoprotein for which the difference is significant but slight.

Given that most studies included in the analyses were published by just a few groups, the investigators determined whether or not this could somehow impact their findings since such studies could have reported observations made on overlapping cohorts. However, after additional analysis, they concluded that this aspect did not bias their general conclusions.

Furthermore, when the subjects were put into two groups with a mean BMI<27 or BMI≥27, they found that the heavier the people were, the more significant the relation of Ad36 infection with obesity and weight gain became.

According to the authors, the lack of association of Ad36 infection with any metabolic markers including waist circumstances suggests that the virus is related to subcutaneous more than visceral fat (more details on the differences between subcutaneous and visceral fat here). The outcomes of several recent studies actually provided insight as to how Ad36 could not only be associated with, but possibly cause weight gain and/or obesity. First, Ad36 infection accelerates the differentiation of human preadipocytes into adipocytes (here and here for open access papers). Second, Ad36 increases both lipid and glucose uptake, resulting in higher cellular lipid levels by stimulation of de novo lipogenesis (another open access paper here).

What is the take-home message?

Despite a certain number of limitations such as some heterogeneity among the studies, or lack of information about any kind of medication that could have contributed to obesity, the authors consider the results of their meta‑analysis as quite robust and tend to confirm the message the literature has been spreading for a while now: people who test positive for adenovirus 36 are more prone to weight gain and obesity than those that have never encountered the virus in their life. But as this work included observational cross‑sectional studies only, there is still the possibility that overweight or obese people may actually be more susceptible to this viral infection for some reason.

In addition, given that there does not appear to be an association between Ad36 and metabolic markers, infection may be linked with accumulation of relatively benign subcutaneous fat, which highlights the important distinction between adiposity and metabolic health. Anyway, the relationship between Ad36 and obesity in animals, including monkeys, is causal and Ad36 is increasingly shown to be able to impact lipogenesis in human cells. These findings suggest that further investigation may at some point lead to the demonstration of an infectious cause for weight gain and obesity in humans as well, at least to some extent.

Audrey Richard, PhD

About the author: Audrey graduated with a PhD in molecular and cell biology at the University of Lille 2, France. She owes the little oncolytic virus (H‑1 parvovirus) she was working on for that. She is now, among others, willing to learn as much as she can about other viruses and most importantly their interactions with hosts.


Related Posts Plugin for WordPress, Blogger...
This entry was posted in Guest Post, Obesity Research, Peer Reviewed Research and tagged . Bookmark the permalink.

5 Responses to Infectobesity: is obesity transmitted through a common viral infection?

  1. MarkH says:

    I tell you this looks like bullshit to me, for a couple of reasons.

    For one, adenoviral infections are typically cleared after formation of neutralizing antibodies. Presence of antibody should only indicate past exposure, and unlike retroviruses, they aren’t inserting DNA permanently into your genome. So, once virus is clear, infected cells are destroyed by the immune system etc., there isn’t a good reason for a persistent effect.

    Second, figure 2 is a pretty classic case of regression towards the mean. The first study shows the biggest result, with the subsequent studies showing smaller-and-smaller effects, with many CI overlapping with 1. This will pull you into the possibly statistically signficant, but likely clinically insignificant range. Obesity is about a lot more fat than being described here.

    So what you have to me is a biologically-implausible mechanism combined with results suggesting a massive correction from the original effect reported. Looks like garbage to me.

    • Travis Saunders, MSc, CEP says:

      Thanks for the knowledgeable comment, Mark!

      Naive question – any idea on how long it takes to clear an adenovirus out of a person’s system? If it’s long enough to cause adipocyte differentiation, then the above (e.g. small increase in adiposity, but no change in health risk) is exactly what you’d expect, and would be pretty similar to TZD treatment. Or is that implausible? This at least seems to fit with what’s going on in animals and humans, regardless of the actual magnitude of weight gain.

      Given that the virus doesn’t seem to result in increased metabolic risk, I’m more interested in it from a basic physiological perspective, rather than a health one. Fat cell differentiation is actually potentially useful from a clinical perspective, and I think people have been so focused on the idea that a virus could be linked with obesity that they haven’t actually realized the more interesting relationship with the number of adipocytes.

    • Audrey Richard says:

      Hi Mark,

      Thanks for your comment.
      It is certainly possible that your conclusions are the right ones in the end but you are wrong about one point at least: retroviruses aren’t the only viruses to insert their genome into human DNA.
      Not that I’m claiming that there IS viral infection, then insertion and then consequences on the host in a lifetime for sure but no, retroviruses aren’t the only viruses to be endogenized per se.

      Plus, there’s actually no need to look for something as fancy and new as the proved endogenization of virtually any kind of viral genetic material into the genome of hosts to possibly explain the long-term effects of viruses. As many other viruses (CMV, EBV, hepatitis B…), adenoviruses can undergo latent/occult infection, meaning they can be reactivated at some point because they are just hidden. So monitoring the presence of neutralizing antibodies certainly indicates past exposure to the virus but it doesn’t mean that the virus has been cleared for good. So there could be a reason for a persistent effect: persistent infection.

  2. MarkH says:

    Woah. Holdup there chief.

    It is certainly possible that your conclusions are the right ones in the end but you are wrong about one point at least: retroviruses aren’t the only viruses to insert their genome into human DNA.

    Lots of those papers are examples of ERVs, endogenous retrovirus insertion, which if it occurs with germline mutation (quite rare) can be a marker used to trace evolution through ERV insertions throughout genomes. The key is, these events are rare hence it’s more likely to have acquired an ERV from a common ancestor rather than a de novo mutation. And it’s true, retroviruses aren’t the only ones that do this, but viruses that mutate the genome are a small subset of viruses overall. Adenoviruses, as far as I can tell in the literature, do not cause permanent genetic modification.

    You say “many other viruses (CMV, EBV, hepatitis B…), adenoviruses can undergo latent/occult infection, meaning they can be reactivated at some point because they are just hidden. ‘

    I agree with CMV, EBV, hep B, etc., but not so much with adenoviruses. There are situations in which latency is found like asthma and COPD, but they generally are cleared by innate immunity which is what has hampered them as a gene therapy vector. Maybe ad36 is different, it’s possible. Maybe the innate immune response to ad36 is what generates the evidence so far of increased adiposity in infected animals. Possibly antibodies formed to ad36 activate an adipogenic pathway. I doubt it’s from persistent infection however. I would have to see evidence in the literature of failure of the innate immune system to clear ad36 as it does for other members of the family. The correlation between evidence of previous infectivity and obesity, and the animal studies are intriguing. It should be studied more, but “infectobesity” as a cause of human illness is far from proven for me, and before the term is spread (and people are given yet another excuse or worse declared potentially infectious for their weight), a lot more work should be done, some classical virology of the course of infection, the body’s immune response, viral levels, etc., should be done.

    I also doubt it’s a big effect on obesity in humans based on the data presented. The correlation is modest, and decreasing with each subsequent measure. Hence, regression towards the mean which appears to be about 3% more body mass. Obesity isn’t from 3% differences in body mass. So while statistically-significant correlations are found, I’m unconvinced of the clinical significance.

  3. Pingback: Links 10/7/12 | Mike the Mad Biologist