*** The USEPA Weight-of-Evidence Classification shown in Submodule 5D is the old system from 1986. The newest USEPA scheme is quite a bit closer to the IARC classification scheme:

• “carcinogenic to humans”,
• “likely to be carcinogenic to humans”,
• “suggestive evidence of carcinogenic potential”,
• “inadequate information to assess carcinogenic potential”, and
• “not likely to be carcinogenic to humans”.

Answer, from EPA Q&A book.
How does EPA characterize a chemical's potential for human carcinogenicity under the draft final Guidelines ? The draft final Guidelines - recommend that an agent's human carcinogenic potential is described in a weight-of-evidence narrative . The narrative summarizes the full range of available evidence and describes any conditions associated with conclusions about an agent's hazard potential. For example, the narrative may explain that a chemical appears to be carcinogenic by some routes of exposure but not others (e.g., by inhalation but not ingestion). Similarly, a hazard may be attributed to exposures during sensitive lifestages of development but not at other times. The narrative also summarizes uncertainties and key default options that have been invoked.

To provide additional clarity and consistency in weight-of-evidence narratives, the draft final Guidelines suggest a set of standard weight-of-evidence descriptors to accompany the narratives. The draft final Guidelines emphasize that risk managers should consider the full range of information in the narratives and not focus exclusively on the descriptors. As in the case of the narratives, descriptors may apply only to certain routes of exposure, dose ranges, and durations of exposure. The fivedescriptors are:

Carcinogenic to Humans

Likely t o Be Carcinogenic to Humans

Suggestive Evidence of Carcinogenic Potential

Inadequate Information to Assess Carcinogenic Potential

Not Likely t o Be Carcinogenic to Humans

This proposed approach differs from that used in EPA's 1986 cancer guidelines in its recommendation that carcinogenic potential be discussed in a narrative and in the particular descriptors employed.

**Q. Why are cohort studies the most popular ones ? Apart from the reason that affected people are easier to identify and that they involve occupational exposures. Any statistical reason for cohort studies being the most popular ?
A. They are the soundest, philosophically, because prospective cohorts are the least biased. On the other hand, they are the most expensive. For rare diseases, enormous cohorts would have to be followed, and this would mean the greater expense for rarer diseases.

**Q. Is Maximum Tolerated Dose only related to loss of body weight? There are no other factors to consider?
A. Any obvious toxicity would be used. But MTD is a 3-month test. Chemicals that might be acutely toxic over a few days are given at much smaller doses for three months, and weight loss is the most common effect observed.

**Q. "In regulatory policy, the 'virtually safe dose' (VSD) corresponding to a maximum, hypothetical risk of one cancer in a million, is estimated from bioassay results using a linear model, which assumes that cancer causation is directly proportional to the dose and that there are no unique effects of high doses. To the extent that carcinogenicity in rodent bioassays is due to the effects of high doses for the nonmutagens, and a synergistic effect of cell division at high doses with DNA damage for the mutagens, then this model is inappropriate. Regulatory agencies are moving slowly to take mechanism and nonlinearity into account, e.g., US EPA." This is an intriguing statement. From what I have been able to observe, it is difficult to see how the methods have changed, and if they have, to what? It appears as though the same kind of animal test procedural standards are in effect today, or at least that's what the MSDS sheets seem to reflect (where toxicological information exists). Question: what changes are being made by the EPA to take the nonlinear nature of the dose-response relationship into account? Is a new method under development for determining "safe" exposure levels?
A. There have been several major policy shifts, and the EPA has umbrella documents that express the current philosophy. You can go here and read the latest: http://www.epa.gov/ORD/WebPubs/carcinogen/ . The practical effect of Ames and Gold was that EPA used to say, "5 mg or MEK in your drinking water will increase the changes of cancer to 1 in one million." Now they add that this is a projection and the real risk "may be lower or zero."

**Q. Quantal versus graded data?
A. Quantal means yes or no data. The effect is there or it isn't. Graded means the effect can be any number. Death is quantal, weight loss is graded.

**Q. The concept of probit totally lost me. Why is 5 the number that is used? Is it always 5 and what is the purpose of it?
A. To avoid the negative numbers. Makes it easier to see.

** Q. The genotoxicity tests all seem so uncertain.  Are these really the best tests they have?
If “positive results in bacterial assays are useful in suggesting the necessity of further testing,” but “negative genotoxicity in bacterial tests doesn’t necessarily mean the chemical will not cause genotoxicity in mammalian systems,” what is the point of the bacterial assays?  Does this mean they don’t conduct further testing on chemicals that produce negative results using bacterial assays? 
A. As a practical matter, they are very helpful, but you asked for it.  If the genotoxicty tests show a chemical for human consumption or exposure is mutagenic, it is unlikely to be used.  The chemical manufacture will run from it.  If however, the chemical is already in use in a useful or profitable product, they will test it thoroughly to be able to “prove” it does not cause cancer. 

**Q. I found Dr. Ames last point in his paper on “The Causes and Prevention of Cancer” very interesting.  He states that the fears surrounding the use of pesticides on fruits and vegetables are unsubstantiated and basically insinuates the organic movement is going to cause more harm than good.  While it doesn’t change the fact that I will continue to support organic and local agriculture, it does open my mind to another view point, even if he is over compensating for the flaws in his test.
A. He was not against your garden, but pointing out that is the government required “pesticide free,” fresh fruits and veggies would cost more, so that people won’t eat enough fruits and vegetables because of the cost.

**Q The tests that seem like they would give the more realistic results are the whole animal tests, and even with those, the administered doses of chemicals given are so high, it seems like nearly any substance could be found to be a potential carcinogen.  Reading about the different types of tests for potential genotoxicity was very sobering.
A. It is possible to test an important chemical with lab animal tests, say a two year exposure of several hundred animals, it is just expensive. 

*Q. Based on the two articles by Ames, it seems that the basic premise behind the testing that we do to determine the carcinogenicty of chemicals is pretty much a wash. Since the compounds, including radiation, basically kill cells which promotes increased cell division, which can lead to mutations, which can then lead to cancer, we basically can't separate the dose effect from the actual poison effect using the standard test for carcinogenicty. Although, Ames mentioned that there are some chemicals that apparently do cause mutations through other means, he didn't go into any details. Do you know what any of them are, is there a list of them somewhere, and is there some other test or does most of the data come from epidemiological studies, such as the smoking studies?
A. There are many tests of mutagenicity, I gave you an overview. They all have the same problem; they may not be related to human results. In general, several different mutagenic tests must be positive, before a substance is pronounced a mutagen. Epidemiology is proof, if it can be done. For rare cancers or chemicals to which only a few people are exposed, it can't be done

*Q. about statistical methods: I think that "direct form data" method is more accurate. why linear regression and using probit are used so wide?
A. Because testing is long and expensive. Often there are not data points that bracket the LD50 or LD-whatever.

*Q. And knowing the dose we can predict the rate of mitogenesis or damage to DNA…Everything depends on the dose, that's the main principle of toxicology.
A. Remember, mitogenesis is cellular replication, which does not damage DNA, as such. Just that agents that are mitogens, cause extra replication, perhaps before DNA can repair itself.

*Q. And knowing the dose we can predict the rate of mitogenesis or damage to DNA…Everything depends on the dose, that's the main principle of toxicology.
A. Remember, mitogenesis is cellular replication, which does not damage DNA, as such. Just that agents that are mitogens, cause extra replication, perhaps before DNA can repair itself.

Q. The more we get into this class, the more I am realizing that toxicology, epidemiology, and the whole health industry is not a very exact science. Aren't we making an awful lot of assumptions in module four? For example, linear regression makes many assumptions, such as the mean of the probability of e is zero, you must have constant variance in the probability distribution, and all errors must be independent. Seems that there are a lot of assumptions here that would be hard to guarantee in a biological system. Of course, there's no guarantees in STATs, we just try to make a best guess. And what about the RfD calculation? That seems to fall somewhere between voodoo and taro cards. How in the world did they develop those values and again, there seems to be an awful lot of uncertainty here.
A. You picked that up real quick. I had to go back to school and study for 5 years to figure that out. But don't shoot the messenger. I'm just telling you how it's done.

* Q. Ship yard workers have a very high incidence of long term tobacco use. They also have a long term history of exposure to asbestos. How would you design a retrospective study to determine if exposure to asbestos imparts additional risk of lung cancer? This risk would be above that already posed by tobacco use. All of the study subjects are ship yard workers.
A. That’s been done, although I don’t know about shipyard workers.  Retrospective studies are always suspect.  People are frequently inaccurate about their habits, such as smoking, and very bad at estimating exposures.  In scientific studies of odors, office workers who are lightly exposed will report heavy exposures, since they “notice” the smell, while production workers who are exposed all the time will not report any exposure, since they don’t notice.   When the work was done with asbestos, they studied similar current work situations and used that to estimate what the historical exposure were – that’s about the best you can do.

Q. It seems like cohort studies would be more valuable than any of the lab tests, but there are so many variables involved.
A. Human cohorts?  Impossible for most substances.

Q. Are epidemiologists involved in clinical trials of medications that are not yet approved by the FDA?  Since some experimental drugs have the risk of causing more harm than good, it seems that the experimental population would be a perfect cohort for the epidemiologist.
A.  Not usually.  The drug companies use statisticians and many others, but the sample size is too small.  After the drug has been in use for several years, they do another set of studies, which are closer to what you have in mind.  Almost all drugs have unwanted effects and are toxic – however for a drug there is a benefit that weighs against the harm.  That is not so with environmental chemicals.

Q. How does one become an epidemiologist?  Do most people get a Master’s in Public Health, become a Medical Doctor with a specialization in epidemiology, or study statistics? 
A. Epidemiology was once a fad and has faded.  Few MD’s get a degree in public health, although some do.  There is much more money in treating patients than in worrying about populations. You can get an MS in public health from many backgrounds.  All MPH people have studied statistics, at least at a basic level.  Here is an excellent school:
http://www.sph.unc.edu/