Advantages of beginning exposure during intrauterine development
Over the years, the NTP has tested chemicals for possible carcinogenicity in about 600 2-year bioassays, most often using both sexes of two strains of rodents, rats and mice. Yet only 7 (~ 1%) of those involved in utero/perinatal exposures:
The developing fetus is particularly sensitive to effects of hormones and estrogenic (and other) chemicals. For instance, prenatal exposure of humans to diethylstilbestrol (DES) provides a classic example of transplacental carcinogenesis predicted in animals (Newbold and McLachlan 1982
) and confirmed in humans (Herbst et al. 1971
; Palmlund 1996
; Palmer et al. 2006
). Because exposures to phytoestrogens and synthetic hormonally active compounds, such as bisphenol A (BPA), are increasing in diets of infants and children, these materials need to be investigated in prenatal and perinatal bioassays. Newbold et al. (2001)
evaluated the endocrine disruptor genistein in a 5-day neonatal exposure in CD-1 mice and after 18 months found increases in adenocarcinomas of the uterus. They then designed expanded bioassay with three exposure regimens: a
) continuous exposure from conception through 2 years of age; b
) exposure from conception through 20 weeks (140 days) followed by a control diet to 2 years of age; and c
) dosing from conception through weaning (21 days) followed by the control diet to 2 years. Increases in tumors (mammary gland and pituitary gland tumors) were observed only in the first experiment.
An in utero
exposure protocol used to study effects of TBHQ (Abdo and Kari 1996
; NTP 1997a
) 2.5 years after exposure began should be considered by the NTP and regulatory agencies as a model for cancer bioassays. Female F0
rats were fed diets containing vary-ing amounts of TBHQ, beginning 2 weeks before cohabitation with males and continuing through gestation and lactation until F1
pups were weaned. F1
rats continued to receive postweaning diets containing TBHQ for 123 weeks for males and 129 weeks for females. Interestingly, despite this inclusive exposure protocol from conception up to nearly 30 months of age, there was no evidence of carcinogenicity. In fact, decreases in mammary gland tumors occurred in both sexes. These findings certainly contradict the notion that studies using in utero
exposure with follow-up for about 2.5 years invariably result in carcinogenic activity.
In contrast, recent studies of the widely used plasticizer additive BPA indicate that pre-natal exposures can induce obesity in offspring (Dolinoy et al. 2007
), a phenomenon that has not been observed with postnatal exposures alone. Prenatal exposure of pregnant Agouti Avy mice to BPA, which is a major component of polycarbonate plastics and epoxy resins, significantly reduced DNA methylation (Dolinoy et al. 2007
). Others have reported that prenatal BPA also increases mammary tumor development and other developmental abnormalities (Soto et al. 2008), and may increase recurrent miscarriages (Sugiura-Ogasawara et al. 2005
). Thus, experiments of prenatal exposure to BPA find that such exposures imprint mammary cells, leaving them especially sensitized to later cell growth and hormonal stimulation—characteristics also found in tumors. It may also be relevant that low levels of BPA have been found to activate genes in noncancerous breast cells in a way that mimics that seen in highly aggressive breast cancer (Dairkee et al. 2008
), indicating that BPA plays a critical role both prenatally and postnatally.
Advantages of studies with longer observation periods versus 2-year studies
Experimental studies should be designed with optimum sensitivity to identify likely adverse health problems throughout humans’ increasing life span. Humans, of course, consume or are exposed to countless natural and synthetic substances during gestation, nursing, and the rest of their lives. In modern societies, proportionally more people are living until their 70s, 80s, and 90s, long after prenatal and childhood exposures and retirement from workplace exposures. Because most long-term rodent carcinogenesis studies do not involve in utero exposure and are intentionally terminated after 2 years (104 weeks) of exposure, they cannot shed light on the effects of chemicals on embryos/fetuses/neonates or “elderly” animals. Likewise, studies truncated after 2 years of exposure do not allow sufficient latency periods for late-developing tumors, such as the 80% of all human cancers that occur after 60 years of age. Because a 2-year-old rat is roughly equivalent to a 60- to 65-year-old person, conventional 2-year-long bioassays cannot detect tumors that will develop later in life.
Although some researchers have suggested reducing the duration of rodent studies to 18 months (e.g., Davies et al. 2000
), scientists at the National Institute of Environmental Health Sciences reject that recommendation, noting that animals in such studies would be the equivalent of humans only 30–50 years of age and would reduce statistical power (Bucher 2002
; Haseman et al. 2001
; Kodell et al. 2000
). Instead, a number of researchers have recommended extending the duration of rodent studies to ≥ 30 months (Davis 2007
; Haseman et al. 2001
; Huff 2002
; Huff et al. 2007
; Maltoni 1995
; Soffritti et al. 2002
). In some cases, as the Ramazzini Foundation sometimes does, exposure to industrial chemicals is stopped after 2 years (“retirement age”), and the animals are then observed until ≥ 30 months of age to determine delayed or persistent carcinogenic activity (Maltoni 1995
; Soffritti et al. 2002
). Ceasing exposure at 2 years without monitoring tumor development for additional time cannot estimate the impact of food additives, drugs, and other chemicals on humans who die in their 70s or later.
Advantages of 2-year studies over studies with longer observation periods
Two-year studies do have certain advantages over longer studies. First, because these studies are standard protocols and designs, a great deal of experience has been gained from them, including information on tumor formation in historical controls.
Second, these briefer studies yield somewhat quicker results, which is useful for protecting public health.
Third, shorter tests are somewhat, but not proportionately, cheaper to conduct. However, because of the numerous fixed costs of animal studies (animals, postmortem pathology, etc.), longer studies are not commensurately marginally costlier. In any case, economy should not trump sensitivity and the reliability of results.
Ancillary costs of longer studies include educating researchers on how to conduct them and carrying out more frequent inspections at the later stages for moribund or sickly animals to minimize autolysis in animals that die naturally at night and may not be discovered immediately. However, certain tissues are affected more than others, and tumors would remain relatively inviolate; moreover, even in 2-year studies, only 60–70% of animals remain alive at the end.
Fourth, some tumor rates increase, and old-age lesions may complicate discerning nonneoplastic effects.
Finally, whereas tumor incidences in various strains of control animals are known on the basis of numerous studies, it would take time to develop a body of data on tumor incidence and other abnormalities for these longer time periods.
Some researchers have argued that higher rates of spontaneous tumors and other factors would undermine the value of longer-term studies (Solleveld and McConnell 1985
), but those assertions are not based on actual bio-assays—comparing, for instance, the relative abilities of 24-month, 30-month, and lifetime bioassays to detect the effects of known carcinogens. Most important, any “savings” accruing from current testing protocols are only illusory if false-negative results lead to the exposure of millions of people to substances that caused cancer or other maladies.
To maximize the knowledge gained from costly full-lifetime studies, protocols should be expanded to provide for periodic sacrificing to determine time-to-tumor and biological sampling to determine internal doses, metabolite levels, genetic alterations, and other data relevant to characterizing the pharmacokinetic and pharmacodynamic activity of toxicity and noncancer disease.