Hemolytic uremic syndrome (HUS): history and implications

Part One

It appears that the E. coli O104 sproutbreak is starting to wind down, with more than 3,500 cases diagnosed to date and 39 deaths. Though sprouts remain the key source of the bacterium, a recent report also documents that human carriers helped to spread the organism (via H5N1 blog). In this case, it was a food service employee working at a catering company, who spread infection to at least 20 people before she even realized she was infected.

As with many infectious diseases, there are potential lingering sequelae of infection, which can occur weeks to years after the acute infection has cleared up. Like almost 800 others involved in this outbreak, the woman who unwittingly infected others via food developed hemolytic uremic syndrome, or HUS. We now know that the most common cause of HUS are bacteria such as STEC ("shiga toxin-producing E. coli"); the "shiga toxin" that they produce inhibits protein synthesis in the host and cause cell death. This can have systemic effects, and leads to clotting in affected organs--most commonly the kidneys, but other organs can also be affected. Dialysis may be necessary, and the infection can lead to kidney failure and the need for organ transplantation. There is already concern that, because of the huge numbers of HUS cases, many patients will have long-term kidney damage, including the potential need for additional organs (and possibly, re-vamping the way donations are made as well):

In previous E. coli outbreaks, up to half of patients who developed the kidney complication were still suffering from long-term side effects 10 to 20 years after first falling sick, including high blood pressure caused by dialysis.

In addition to possible kidney problems, people who have survived serious E. coli infections may also suffer from neurological damage, as the bacteria may have eaten away at blood vessels in the brain. That could mean suffering from seizures or epilepsy years after patients recover from their initial illness.

While it's common knowledge in the medical community now that STEC can lead to HUS, which can lead to chronic kidney issues, for many years, the link between E. coli and HUS was obscured. HUS first appears in the literature in 1955, but the link to STEC wasn't confirmed until the early 1980's. In the interim, myriad viruses and bacteria were examined, as well as genetic causes. (There are cases of HUS caused by host mutations and other etiologies, but they are much less common than HUS caused by STEC and related organisms). I'll delve into the history of HUS and look at a few studies which examined alternative hypotheses of causation, until finally STEC was confirmed as the causative agent. I'll also discuss what this means as far as discovering infectious causes of other "complex" and somewhat mysterious diseases whose causes are unknown, as HUS was a mere 30 years ago.

Part Two

The epidemiology of hemolytic uremic syndrome (HUS) was murky for several decades after it was first defined in the literature in 1955. In the ensuing decades, HUS was associated with a number of infectious agents, leading to the general belief that it was a "multifactorial disease"--one that had components of genetics and environment, much like we think of multiple sclerosis today, for example.

Several HUS outbreaks made people think twice about that assumption, and look deeper into a potential infectious cause. A 1966 paper documented the first identified outbreak of HUS, which occurred in Wales. The researchers examined a number of possible environmental factors the patients may have had in common--including food, water, and various toxins--but came up empty. They sum up:

Since it is almost invariably preceded by a gastrointestinal or respiratory illness, it seems probable that it represents a response to an infection. Although Gianantonio et al. (1964) have identified one possible causative virus, it may be that various infective agents can initiate the syndrome.

This idea held throughout the next 20-odd years, as numerous studies looked at both environmental and genetic effects that may be leading to HUS. A 1975 paper examined HUS in families, suggesting that there may be two types of HUS (which we now know to be true--the genetic form is less often associated with diarrhea, and tends to have a worse prognosis as I mentioned yesterday). But still, no definitive cause for either.

There were also a number of studies testing individuals for many different types of pathogens. A 1974 paper enrolled patients in the Netherlands between 1965 and 1970, but one of the inclusion criteria was a "history of a prodromal illness in which gastrointestinal or respiratory tract symptoms were present." The respiratory tract symptoms are mentioned in a number of papers, and were probably a red herring that sent people in search of the wrong pathogens for awhile. In this particular paper, they examined children for infection with a number of viral and bacterial pathogens, using either culture or serological methods (looking for antibodies which may suggest a recent infection). In that portion of the paper, they note a possible association with adenoviruses, but state that the data don't support a bacterial infection--a viral etiology was deemed more likely. Regarding basic epidemiology, they did note a few small clusters of cases in families or villages, as well as a peak in cases in spring/summer--as well as an increasing number of cases from the first year of their study to the last. The epidemiology of HUS was starting to become clearer, and the syndrome appeared to be on the rise.

Even as additional case reports occasionally targeted foods as a precursor to HUS outbreaks, it wasn't until the late 1970s and early 1980s that HUS really started to come into focus. In 1977, a paper was published identifying the "Vero toxin"--a product of E. coli that caused cytotoxicity in Vero cells (a line of African green monkey kidney cells, commonly used in research). Researchers were closing in.

Part Three

I left off yesterday with the initial discovery of "Vero toxin," a toxin produced by E. coli (also called "Shiga toxin" or "Shiga-like toxin"). Though this may initially seem unconnected to hemolytic uremic syndrome (HUS), the discovery of this cytotoxin paved the way for a clearer understanding of the etiology of this syndrome, as well as the mechanisms by which disease progressed. By the early 1980s, several lines of research pointed toward E. coli, and particularly O157:H7, as the main cause of HUS.

A 1982 Centers for Disease Control and Prevention MMWR report found a rare E. coli serotype, O157:H7, associated with hemorrhagic colitis following consumption of hamburgers. Similar results were reported in a 1983 Lancet paper, which found serotype O157 among their collection of verotoxin-producing strains. Another paper that same year from a Canadian group showed that O157:H7 was the second most common cytotoxic strain in their collection of over 2,000 E. coli isolates. The most common was serotype O26--more on that below. This paper also discussed an outbreak of hemorrhagic colitis that had occurred at a nursing home, with O157 identified as the cause. The evidence was mounting, but these were small studies and not always associated with HUS. Still, these papers collectively were suggestive of a connection between E. coli infection (especially with strains that produced the shiga/vero toxin), hemorrhagic colitis, and HUS.

In 1985, a new study came out which really helped to seal the deal. Rather than look only at cases in isolation, the authors designed a case-control study looking at patients with "idiopathic HUS" (in other words, HUS of unknown origin which started with diarrhea, rather than the other variant lacking this symptom). They ended up with 40 patients who qualified. They then picked a single control for each patient, matching them on age, sex, and season of the year. The controls were children either diagnosed with Campylobacter enterocolitis (and therefore, enterocolitis of a known cause) or were healthy children either from a local daycare center, or kids coming in for elective surgeries. Stools were collected from each group and tested for a variety of organisms, including vero toxin-producing E. coli (VTEC, also known as STEC for the shiga-like toxin nomenclature). They also tested for activity of the toxin itself in fecal samples. Finally, in the case patients, attempts were made to collect what are called "acute" and "convalescent" blood samples. These are samples taken when the patient is actually sick ("acute"), and then ones taken a few weeks later ("convalescent), to look at the presence of antibodies in the blood. If it was an infection by the suspected organism (in this case, STEC/VTEC), you should see a rise in antibodies the host produces that target the organism--for these kids, they were looking for antibodies to the shiga/vero toxin.

They found either vero toxin or VTEC in 60% of the case patients, but in none of the controls. Of the VTEC isolated, serotypes included O26, O111, O113, O121, and O157. For the latter, it was the most common type isolated (25% of the VTEC found). Of the patients who were negative for both VTEC and vero toxin, from those who had paired blood samples (12/16 of the remaining cases), 6 did show a rise in antibody titer against the vero toxin--suggesting they had been exposed and were producing antibodies to neutralize the toxin. So, for those keeping score, 75% of the cases had evidence of VTEC infection either by culture or serological techniques. It may not have been the nail in the coffin and there are certainly some flaws (the diversity of controls and lack of analysis of blood titers for the controls being two that pop out at me), but this paper went a long way toward establishing VTEC/STEC as the cause of HUS, which has been subsequently confirmed by many, many studies worldwide.

The most common vehicles of transmission of these organisms have also come into clearer focus since the 1950s, with almost all HUS/STEC outbreaks associated with food products; most common is still the O157:H7 serotype. O157 is a bit unique, in that this strain typically doesn't ferment sorbitol--as such, this is often used as a diagnostic feature that sets it apart from "normal" E. coli. However, as I mentioned above (and as the current outbreak has shown), a number of other serotypes besides O157:H7 can also cause HUS. Most of these don't appear to be as commonly associated with outbreaks--rather, they may more commonly cause sporadic disease where fewer people may become sick. Because these don't have the unique sorbitol-non-fermenting feature, these may be overlooked at a diagnostic lab. There are assays that can detect the Shiga-like toxin directly (actually, we now know there are multiple families of related toxins), but not all labs use these routinely, so it's likely that the incidence of infection due to non-O157 STEC is higher than we currently know.

HUS was once a mysterious, "complex" disease whose perceived etiology shifted almost overnight, as scientific advances go. What implications does this have for other diseases whose etiology is similarly described as HUS was 50 years ago? More on that tomorrow.

Part Four

As I've laid out in parts 1-3, the realization that a fairly simple, toxin-carrying bacterium could cause a "complex" and mysterious disease like hemolytic uremic syndrome came only with 30 years' of scientific investigation and many false starts and misleading results. Like many of these investigations, the true cause was found due to a combination of hard work, novel ways of thinking, and simple serendipity--being able to connect the dots in a framework where the dots didn't necessarily line up as expected, and removing extraneous dots as necessary. It's not an easy task, particularly when we've had mostly culture-based methods to rely on since the dawn of microbiology.

If you read start digging around in the evolutionary medicine literature, you'll see that one oft-repeated tenet is that many more "chronic" and "lifestyle" diseases are actually caused by microbes than we currently realize. (I'll note that there is active disagreement here in the field--one reason noted is that many of these diseases would decrease one's fitness and thus they are unlikely to be genetic, but many of them also have onset later in life than the prime reproductive years, so--still controversial). But whether you agree on the evolutionary reasoning or not, I think it's safe to say that those who make this claim (like the Neese & Williams book I linked) are probably right on the overall assertion that more and more of these "lifestyle/genetics" diseases are going to be actually microbial in cause than we currently realize.

Why do I agree with this claim? History is a great indicator. Many infectious diseases were thought to be due to complex interactions of genetics (or "breeding," "lineage," etc.) with "lifestyle." Think of syphilis and tuberculosis in the Victorian era. Syphilis (and many other diseases which we know now to be sexually-transmitted infections) was considered a disease which affected mainly the lower social classes ("bad breeding"), and was thought to be rooted in both family history as well as an over-indulgence in sex or masturbation. Tuberculosis, because it affected those throughout the income spectrum, was still blamed on "poor constitution" in the lower classes, but was a disease of the "sensitive" and "artistic" in the upper classes. It was also thought to be due to influences of climate in combination with genetics. Or, look to more recent examples of Helicobacter pylori and gastric ulcers, which were also ascribed to dietary habits and stress for a good 30 years before their infectious nature was eventually proven. And from that same era, HIV/AIDS--which even today, some are still all too ready to write off as merely a behavioral disease, rather than an infectious one.

So, we still view many of these diseases of unknown etiology as multi-factorial, "complex" diseases. And undoubtedly, genetic predisposition does play a role in almost every infectious disease, so I'm not writing off any kind of host/pathogen interplay in the development of some of these more rare sequelae, such as HUS as a consequence of a STEC infection. But looking back over history, it's amazing how many diseases which we view now as having a documented infectious cause were studied for years by researchers thinking that the disease was the result of exposure to a toxin, or diet, or behavior, or a combination of all three.

I've mentioned the example of multiple sclerosis in previous posts. Multiple sclerosis is an autoimmune disease; the body produces antibodies that attack and eventually destroy parts of the myelin sheath covering our nerves. The cause of MS, like HUS 40 years ago, is unknown, though it's thought to be a combination of genetics and environmental influences. Going through the literature, it seems like almost everything has been implicated as playing a causal role at one point or another: pesticides, environmental mercury, hormones, various other "toxins," and a whole host of microbes, including Chlamydia pneumoniae, measles, mumps, Epstein-Barr virus, varicella zoster (chickenpox), herpes simplex viruses, other herpes families viruses (HHV-6 and HHV-8), even canine distemper virus. They've done this looking at both microbe culture (from blood, brain tissue, CNS, etc.) as well as using serology and DNA/RNA amplification in various body sites. None have shown any strong, repeatable links to the development of MS--much like the spurious associations that were seen with adenovirus and HUS.

Although no microbial agent has been convincingly implicated to date, there are tantalizing hints that MS is caused by an infectious agent. There have been "outbreaks" of MS; the most famous occurred in the Faroe Islands in the 1940s. Studies of migrants show that the risks of developing MS seem to be tied to exposures in childhood, suggesting a possible exposure to an infectious agent as a kid. And one of the most common mouse models used to study MS has the disease induced by infection with a virus called Theiler's murine encephalitis virus (TMEV). If it can happen in mice, why not humans?

It might seem implausible that infection with some microbe could lead to the eventual neurological outcomes of MS, but again, examples abound of weird connections between microbes and health outcomes. For STEC, it might not be intuitively obvious at first glance how a fecal organism could be a cause of kidney failure. The respiratory bacterium Streptococcus pyogenes usually causes throat infections ("strep throat"), but if left untreated, it can also cause kidney damage (glomerulonephritis) or even heart failure due to rheumatic heart disease. A microbial cause of MS could lie in a virus, bacterium, parasite, or fungus--maybe one that we haven't even discovered yet, but that perhaps will pop up as we learn more and more about our metagenome. Perhaps 30 years down the road, the way we view many of these "complex" diseases will look as short-sighted as it does looking back at old HUS papers from today's vantage point.