Interview with Dr. J. William Hirzy, Senior Vice President, EPA Headquarters Union, July 2000.

The following is an excerpt of Michael Connett's interview with Dr. J. William Hirzy, Senior Vice President of EPA's Headquarters Union in Washington DC. The interview took place on July 3, 2000, a couple days after Hirzy testified before the US Senate calling for an independent review of the tumor slides from the National Toxicology Program's bioassay for fluoride. It is Hirzy's and the Union's belief that the downgradings of the tumors was politically motivated and not scientifically defensible.

Q: How was the study conducted? What did they do?

Hirzy: Groups of 50 animals, rats and mice, were dosed with fluoride in their drinking water either at zero, eleven, forty-five, or seventy-nine parts per million fluoride over their entire lives. Normally that is essentially a two year period. And then the animals were autopsied... and their health status was tabulated including tumors and so forth.

Q: And the study was done by an independent contractor?

Hirzy: Done by Batelle laboratories.

What were Batelle’s findings?

Hirzy: There were tumors in the oral cavities of the test animals. There was osteosarcoma (bone cancer) in male rats. There was also an incidence, a relatively high incidence of rare liver tumors in mice, not the normal kind of liver tumor that is seen. This was hepatocholangiocarcinoma which was a relatively rare and newly discovered tumor actually. The pathologist Melvin Reuber who discovered that was actually part of the review group that worked on this bioassay.

The tumor incidence that was originally reported would have supported a finding of clear evidence of carcinogenicity for sodium fluoride, based on all these tumors that I mentioned. There was also some osteosarcomas not of the bone but found in soft tissue which is a rarity. Taking all of those tumors into account, it would have led to a finding of clear evidence of carcinogenicity. However, a commission was appointed by the Department of Health and Human Services that came down to rescue fluoridation from such adverse findings.

Q: The American Dental Association claimed that the rats were fed levels of sodium fluoride which humans don’t receive in their diet, 79 parts per million, 45 parts per million, and so the results were irrelevant to humans because the doses were so high. How would you respond to that point?

Well, it’s common practice in cancer bioassays, in order to limit the number of animals that are used, to give doses that are generally always in excess of what humans get. It’s a matter of statistics basically.

The level of regulatory concern at EPA generally is one person in a million, one person in a hundred thousand, one person in ten thousand, one person in a thousand developing cancer from a particular exposure given the kinds of exposures to a chemical that humans might get. Now in order to mimic that exactly in an animal population you’d have to study a million animals, or a hundred thousand animals, or ten thousand animals, or a thousand animals and this begins to get completely unreasonable in terms of use of animals, cost, and the like.

So what we do is rather to limit the number of animals but increase the dose on the assumption that there’s, what we call, a linear dose response curve...

We do preliminary studies to find out how much of the test material the animals can tolerate without showing overt signs of toxicity. We say that that’s the 'maximum tolerated dose' that the animals can have and we say that that’s going to be the upper limit of dosing in this study, with the idea of loading up the animal with the chemical so that with 50 animals in a group the likelihood of them developing cancer, if this material is carcinogenic, is going to be measurable within 50 animals.

And then from the 'maximum tolerated dose' the doses are scaled down so that we have a range of doses below the tolerated dose down to a zero control, so that one could plot from zero exposure in the control group to the maximum tolerated dose and then plot the response based on these different doses. One normally gets some sort of a dose response curve if the material is carcinogenic.

And that’s what was seen in the case of sodium fluoride. The highest dose level by the way, 79 parts per million, is relatively low compared to the doses that are given to the animals in other corresponding bioassays. It is not at all unusual in a cancer bioassay for animals to get a 1,000 or 10,000 times the dose that humans get. In this bioassay, the level of 79 parts per million of fluoride was only literally 79 times greater than the so-called 'optimum' level. It’s only about 20 times higher than the level which the EPA says is ok to have in your drinking water. That’s a remarkably close level to have the highest dose level in an animal study and to have a positive response for cancer in the anticipated target organ, namely the bone, it’s a remarkable finding.

Q: So have other chemicals been defined as carcinogenic based on studies like this?

Hirzy: Oh yes.

Q: Where they actually dosed the rats at higher levels than they dosed with fluoride?

Hirzy: Yes...And the dose levels are much higher compared to what the human exposures are, as opposed to the situation here. Here the animals got overt, frank bone tumors, bone cancer, malignant cancer of the bone, at a level only twenty times higher than the EPA says is safe to drink. That is a remarkable narrow margin of safety.

Table: Comparison of fluoride dose causing bone cancer with doses used to test other suspected carcinogens and their Maximum Contaminant Levels. From "Fluoride: risks and benefits? Disinformation in the service of big industry." David R. Hill. http://www.fluoridation.com/calgaryh.htm


Daily Dose

Maximum Contaminant Level (MCL)




Vinyl chloride



Carbon tetrachloride












Red dye #3



Q: Now, is it true that it’s very difficult to produce osteosarcomas in rats?

Hirzy: Yes, it’s a relatively rare tumor in rats and again Dr. Marcus has written of this in his memorandum pointing out the very few incidences in a control population of other bioassays in which osteosarcoma was reported. It’s a very, very low incidence tumor.

And I believe that the only other chemical that Dr. Marcus knew of that could produce osteosarcoma that quickly was radium.

Hirzy: Radium. That’s right.

Q: And is that a known...

Hirzy: That’s a known human carcinogen as well.

Q: And in the Batelle lab they found a certain kind of liver cancer that you had mentioned.


Q: This was a very rare form of cancer I understand.

Hirzy: That’s right. Hepatocholangiocarcinoma is what it was. It’s basically a rare cancer of the bile duct in the liver.

And in normal cancer studies if a rare cancer is found in the dosed group of animals, does that usually generate increased interest?

Hirzy: Absolutely. Yes indeed, it does. It increases the evidence of carcinogenicity if a rare cancer shows up in the dosed group and not in the controls.

And has the EPA or any government health organization followed up to find out, to better understand this relationship between fluoride and this rare liver cancer?

Not that I’m aware of.

...There’s other evidence for carcinogenicity besides this bioassay. There’s a substantial body of mutagenesis studies, that is studies where a chemical’s ability to cause changes in DNA have been proven. Also, fluoride is known to be an enzyme poison in any number of enzyme systems. And repair of damaged DNA by enzymes is an important anti-cancer mechanism that the body has.

If some of these repair enzymes are inhibited by fluoride, it could indicate fluoride’s ability to work as what we call a 'promoter' as opposed to a 'primary' carcinogen.

Carcinogenesis is a very interesting and complex subject. There are so called 'complete carcinogens' - chemicals that in fact damage the DNA and then allow an aberrant cell to reproduce wildly and develop a malignant tumor. There are other substances which are called 'promoters' which do things like inhibit repair enzymes and once an incipient cancer forms - if these repair enzymes are not able to go in, sense the aberrant cell or the aberrant strands of DNA, and either repair the DNA or kill the cell - then the tumor is allowed to grow. Those are classified as promoters.

It may well be that fluoride works in both ways as a primary carcinogen and as a promoter.

And it would be a promoter of cancer by damaging or inhibiting the enzymes that can be protective?

That’s right.

Q: If these tumors in the rats weren’t downgraded by the National Toxicology Program, what would have been the ramifications for fluoridation?

It would have been over. If in fact the classification of Probable Human Carcinogen, which would have flowed from the finding of clear evidence here, it would have required that the EPA set a Maximum Contaminant Level Goal of zero. All carcinogens have a Maximum Contaminant Level Goal of zero. That is, you recall, the health based standard, the one EPA sets saying that at this level we anticipate no adverse health effects in the entire population with an adequate margin of safety. For carcinogens, the policy is no level of exposure is safe, so the MCLG is zero. If it’s zero, it means you can’t add any of this stuff to any water supply and that’s the end of fluoridation.