The Institute of Medicine (IOM) just released its report on the possible relationship between SV40, known to have once contaminated the polio vaccine, and cancer.  The committee got a lot right in this mostly excellent review in which they concluded that the "evidence is inadequate to accept or reject a causal relationship".  Nevertheless, some aspects of the report warrant discussion.

Although the committee left open, and rightly so, the possibility that a relationship might yet be proven to exist, and admitted to supportive biological evidence, in my opinion they did not go far enough in communicating what is known about the strength and weight of the evidence in favor of there being a connection.

In the abstract to their executive summary, they reported that "Although SV40 has biological properties consistent with a cancer-causing virus, it has not been conclusively established whether it might have caused cancer in humans."  While they are probably right to say that it has not been proved that SV40 causes cancer in humans, it is a huge understatement to say that "SV40 has biological properties consistent with a cancer-causing virus". There is a large and considerable body of scientific evidence supporting the notion that SV40 is associated with cancer, in both animals and humans.  As leading SV40 scientist Carbone titled one of his studies, referring to the preponderance of the evidence, "Simian virus 40 and human tumors: It is time to study mechanisms."

They then go on to say, "Studies of groups of people who received polio vaccine during 1955-1963 provide evidence of no increased cancer risk".

Talk about sins of omission!  While it is true that there are studies which provide evidence of no increased cancer risk, it is equally true that there are studies which provide evidence that such a relationship exists, some of them even having been discussed in the report (e.g.., Fisher et al, Farwell et al), although apparently ignored in order to be able to make that statement.

One seemingly relevant article was not even included in their review,  a 1984 article by Farwell et al, in which it was concluded that  "...an excessive number of children born in the period 1954 to 1958 have developed medulloblastomas."  Why was this article omitted?

Saying the epidemiologic studies in this review "provide evidence of no increased cancer risk" seems particularly arguable, however,  given that the committee found every single epidemiologic study they reviewed to be weak, on both sides of the argument.

 I have no problem with their conclusion that the "evidence was inadequate", given the apparent paucity of good epidemiologic data, but it would have been better to have made just that point, to have reported that the few results that exist are mixed and why, than to have made it incorrectly seem that all available evidence was against there being a relationship.

Equally important are the reasons why some studies, especially more recent ones, show no evidence of increased risk.  Among the most important of them is the fact that SV40 has now spread beyond those who received the polio vaccine. As noted, "...the assumption that persons who received polio vaccine after 1963 were unexposed to SV40 may not be accurate if sources of exposure other than contaminated IPV exist... A limited number of people are known to have been exposed to SV40 through other vaccines...Evidence of SV40 exposure has also been detected in serologic samples obtained before 1955 and from studies of persons too young to have received contaminated polio vaccine. Detection of SV40 in persons too young to have received contaminated polio vaccine suggests the possibility of continuing transmission of SV40 through means other than the polio vaccine. Possible sources of exposure to SV40 are person-to-person transmission, animal-to-person transmission, and laboratory exposure to SV40." 

Consequently, unless study participants are accurately tested for evidence of exposure to SV40, they might well be in the wrong study group, thus confounding the results and making them meaningless.  (Which is one of the reasons the IOM correctly recommended no epidemiologic studies until the methods of detection and measurement improved.)

(By the way, had the question been assiduously studied when it was first discovered that SV40 contaminated the vaccine, before it had spread beyond the polio-vaccinated, epidemiologic studies would not have faced this particular obstacle to the degree they do now.)

And while the committee did provide the caveat that "The presence of antibody might also indicate protective immunity against SV-40 induced oncogenesis", nowhere in their recommendations is there a call for better understanding of the meaning and significance of SV40 antibody production, including whether or not the assumption that antibody production necessarily follows exposure is correct, the possible misunderstanding of which could well result in additional confounding.

Perhaps most troubling, however, is the large number of references not included in the review by the IOM, including many cited in my March column on SV40 and cancer. 

Why were so many relevant articles not included in the IOM review?  Shouldn't any review by the IOM be exhaustive?  Given the import attributed to any of their findings, what justification could the IOM possibly have for leaving out even one relevant study?

In the end, the committee concluded that because of the "flawed" epidemiologic evidence, and in spite of the "moderate" to "strong" biological evidence, "evidence was inadequate to conclude whether or not the contaminated polio vaccine caused cancer". 

This conclusion may be reasonable, based on the state of the evidence.  Regardless, a point needs to be made about the weight they give to the two types of evidence.  Their failure to give more weight to biological evidence was at least partly based on their stated position that "Epidemiologic studies carry the most weight in a causality assessment", and their apparent view that the problems with the studies examined in this review generally revolved around their poor design (i.e., being "flawed"), rather than in anything inherently weak in that kind of study.  But all epidemiologic studies are not created equal.  According to Epidemiology in Medicine, "Intervention studies (a.k.a. "experimental", "clinical", or "prospective cohort"  trials) are often considered as providing the most reliable evidence from epidemiologic research."  Case-control and non-prospective cohort studies do not carry the same weight, nor are they necessarily superior to evidence gleaned from studies of biological mechanisms. 

In an earlier column, a similar point was made about problems associated with rejecting or minimizing biological evidence due to an alleged lack of epidemiologic supportive evidence.  And while in this case biological evidence was not rejected outright in favor of epidemiologic evidence, the committee underplayed the strength of the association, even though they readily acknowledged the "wealth of (biological mechanism) literature" available to them.   The sheer size of the biological evidence in favor of there being a relationship between SV40 and both human and animal cancers should probably have counted for more than was evidenced in their conclusion.

Also worth noting is the fact that while no member allegedly had a financial interest in the issues discussed in this review, it was disclosed that "all the committee members share the view that immunization is generally beneficial".  Perhaps different conclusions would have been drawn if they had had a better balance of members, including a few who were neutral about the benefits of vaccination, as well as others who leaned against them, in order to avoid the obvious potential for bias.  Maybe then we might have received a report completely worthy of the attention this one will, in all likelihood, receive.

The truth is, the only way to conclusively prove causality in humans is through human experimentation, preferably through prospective, randomized, double-blind, placebo-controlled longitudinal studies.  That would mean administering SV40 to some and not to others, and comparing the results over time.  Clearly, this is not going to happen, nor should it.  The next best thing would be to study over time otherwise matched groups with knowledge of and/or evidence of exposure to SV40 virus, comparing the outcomes, and determining what, if any, biological mechanisms explain any differences, to whatever extent they occur. 

The strength of the biological mechanism evidence definitely raises a red flag, regardless of the weakness of the epidemiological evidence.  The recommendations the committee made will go a long way toward correcting what is missing.  But until the problems are resolved, and resolved they must be, which detract from our ability to conduct meaningful epidemiologic studies, all we are left with is the "moderate" to "strong" biological mechanism evidence, and our conclusion, until then, should be one of guarded but heightened concern.

Sandy Mintz