Oral perfluorooctane sulfonate (PFOS) lessens tumor development in the APC min mouse model of spontaneous familial adenomatous polyposis

In these two preliminary studies of male and female APC^min mice, total tumor number decreased significantly with increasing PFOS dose, with the highest dose groups showing the largest effects. The observed dose–response relationship was particularly evident in larger tumors, suggesting a possible inhibitory effect of PFOS on tumor formation. Since PFOS administration in our mouse model was initiated prior to tumor development, the observed reduction in tumor burden may reflect effects on both tumor initiation and tumor progression. Notably, the number of 1–3 mm tumors decreased significantly with increasing PFOS dose (Fig. 1c), suggesting that PFOS may not only inhibit development, but may halt progression and even possibly induce tumor regression. If PFOS induces tumor regression, reduction of larger tumors may lead to a corresponding increase in the number of tumors??1 mm diameter, thus potentially attenuating the observed effect of PFOS on total tumor number and helping to explain the stronger effects observed for larger tumors. Collectively, these findings suggest that PFOS has a significant, dose-dependent inhibitory effect on gastrointestinal tumor formation in this established genetic mouse model of familial adenomatous polyposis.

Results of these preliminary experimental studies are broadly consistent with findings from our recent epidemiological investigation in a large population of Appalachian adults exposed to PFOA-contaminated drinking water. In this cross-sectional study, serum PFOS levels showed a strong, inverse dose–response association with prevalent colorectal cancer that remained robust after adjustment for multiple potential confounders [12]. However, while findings of this epidemiological investigation likewise suggest a possible protective effect of PFOS on colorectal cancer, the cross-sectional nature of the data limit causal inference. Although implications for non-familial colorectal cancer remain unclear, the current study in a mouse model of familial adenomatous polyposis offers the first experimental evidence that chronic exposure to PFOS in drinking water can reduce formation of gastrointestinal tumors, and that these reductions are both significant and dose-dependent.

In our present animal study, we provided PFOS in the drinking water to simulate chronic human PFOS exposure [20, 21]. Liver enzymes can be induced by PFOS exposure in mice [22], and toxicity indicated by weight loss [23] was observed here. In other studies, higher PFOS doses have been administered over shorter periods by oral bolus [5, 24] without evident toxic effects; however, our data suggest toxicity likely develops at a lower overall dose when PFOS is delivered slowly over time [5, 24]. Here progressive weight loss was observed at doses of 200 mg/kg or higher, indicating this dose is near the maximum tolerated dose for this mouse strain and delivery method over this time frame. Fortunately, measurement of plasma PFOS levels in male mice at 15 weeks indicated that drinking water administration at all doses resulted in plasma levels substantially higher than those associated colorectal cancer reduction in humans [12]. In common, PFOS appeared beneficial in human colorectal cancer and in APC^min mice. However, the former effect was observed in humans which metabolize PFOS differently from rodents, and where the effect was substantially based on the acquired non-familial form of colorectal cancer. In the mouse model, PFOS undergoes a greater degree of metabolism, has a shorter half-life, and counters a genetic predisposition to colorectal cancer. Further studies should investigate the adverse consequences of this agent under therapeutic conditions; even so, this study provides a possible direction to pursue in regard to familial colorectal cancer.

Although PFOS is widely distributed in the environment [25, 26] and has been detected in human populations worldwide [9, 27–30], non-occupational blood levels in humans are well below those reported toxic in lab animals [31–33]. The half-life of PFOS is reported to be??40 days in mice [34], and contrasts dramatically with the estimated 4–5 year half-life documented in humans [35]. It appears that PFOS in rodents is handled in a manner similar to fatty acids, and consequently induces hormonal, peroxisomal and P450 enzyme gene activation [36]. The increased PFOS half–life in humans compared to rodents may be in part due to PFOS inhibition of human cytochrome activity [37]; cytochrome activity inhibition could also reduce the influence of toxic metabolites which may explain higher degrees of toxicity as commonly reported in rodents. In addition, PFOS renal reabsorption and recycling have also been shown to contribute to a long half-life in humans and monkeys [38]. Cancer risk with prolonged chronic exposure was suggested at high doses in rats [11], although consistent evidence for elevated tumor risk with PFOS exposure in humans is lacking [11].

Each PFAS has a unique biological and toxicological profile that limits extrapolation across compounds or model species [11]. Potential mechanisms of PFOS action relevant to its effect on tumor development and progression are still ill-defined, but are not surprising given the large number of genes (e.g. ~400 in rats) PFOS appears to influence [36]. Possibilities include anti-inflammatory effects via prostanoid pathways, PPAR receptor mediated actions, immune effects, or other as yet unrecognized mechanisms. PFOS may serve an anti-inflammatory role via its influence on downstream transcriptional regulators such as NF-?B [6]. Phospholipase A2 is inhibited by PFOS in rats; this could, in turn, block the production of arachidonic acid as a substrate for prostaglandindin H synthase elaboration of prostanoids [36]. Similarly, PGE₂ has been shown to be a potent inducer of adenoma formation in APC^min mice [39], and tumor growth [40] was similarly increased by an agonist, where both were mediated through the PPAR? receptor. Adenoma formation by PGE₂ was removed in mice missing this receptor [41]. PFOS also significantly stimulates both PPAR? and PPAR? [42], which could also modulate tumor growth [43, 44]. PFOS serves as a partial agonist and induces PPAR? mediated effects at high doses [45], other effects via other PPAR receptor isotypes [46], and produces significant immunomodulatory effects in mice [47]. In PPAR? KO mice, PPAR? independent nuclear receptor mediated pathways and down-stream effects were noted [6], including suppression of T-cell dependent antibody production, and modulation of immune cell and cytokine synthesis (e.g. TNF? and IL-6) [6]. While PFOS appears to stimulate mouse and human PPAR receptors [42, 46] when screened in cell lines, robust in vivo evidence for direct PPAR receptor mediation is lacking, and human PPAR? expression is considerably reduced compared to in rodents; if so, this may thus lessen the importance of this pathway in human familial adenomatous polyposis or acquired colorectal cancer [6]. Even so, PPAR? stimulation in human colorectal cancer lines is moderately pro-inflammatory and stimulates prostaglandin H synthase-2 expression [48, 49]. The potential impact of PFOS directly on the Wnt-?-catenin signal transduction pathway also warrants closer examination [50]. The “min” defect causes catenin retention and ultimately the Wnt gene group to become canonically activated; gastrointestinal polyp formation is one direct consequence [51]. The putative role of PFOS in blocking this process would be a plausible mechanism to explain its efficacy in this model system. Alternative pathways could also be affected [52]. Other influences of PFOS, by interacting with dietary constituents [53, 54], or steroidogenic enzyme and hormone disruptive effects cannot be ruled in or out [55]. Here, both mouse genders benefited from PFOS exposure, arguing against a differential role as pertains to specific sex hormones.

In recent human cross-sectional studies, chronic PFOS exposure has been associated with modest, adverse changes in serum lipid profiles [56–58]. Similarly, elevated blood levels of PFOS have been associated with increased likelihood of early onset menopause [59] and altered thyroid function [59]. However, in contrast to findings from animal studies, including our study in APC^min mice, significant adverse effects of PFOS exposure have been rarely been documented in humans, even in pregnant women and highly exposed fluorochemical plant workers [16, 58, 60, 61]. Even with its complex disposition in humans, PFOS appears well tolerated at environmental levels (e.g. Danish National Birth Cohort Study [62]: PFOS mean 35.3 ng/ml; range 6.4–106.7 ng/ml; elsewhere 0–30 ng/ml [15, 63–65],) levels also associated with colorectal cancer protection [66]. Occupational exposures in chemical factory workers are up to 40 times higher, yet adverse outcomes at these levels, even among at risk pregnant women, are rare [16, 60]. At high occupational exposure levels, an association between PFOS and bladder cancer was reported [65]; however, this association was questioned more recently [16, 58, 61]. Therefore, although PFOS has been decried as an environmental contaminant, it might still have therapeutic value at low levels. If shown to be effective for colorectal cancer prevention and/or treatment in humans, PFOS may offer an option that is significantly safer, lower cost, and less toxic than alternative therapies [67].

One potential limitation relates to reverse causality, i.e., the possibility that tumor formation may reduce PFOS absorption, and thus, blood PFOS levels. However, this is unlikely given the massive surface area of the gastrointestinal tract and the inherent lipid solubility of PFOS. Moreover, serum PFOS generally correlates well with liver concentrations [61], suggesting that serum is a reasonably good systemic indicator of PFOS exposure in humans [68]. Similarly in a mouse model of familial adenomatous polyposis, blood levels corresponded to target dosing levels and predicted tumor reduction, again indicating that PFOS absorption reflects oral exposure. Another theoretical argument is that a parent compound is broken down to make PFOS, and it may be this compound, rather than PFOS itself that led to the reduction in colorectal cancer observed in our previous epidemiological study [12]. However, PFOS exposure had a clear benefit here.

Strengths of this study include identifying a beneficial role for PFOS as a chemo-preventive or therapy in a familial model of colorectal cancer, irrespective of gender. In addition, the development of a slow delivery method and estimates of an effective working dose range by this delivery method were determined. Finally, toxic effects were easily identified using simple body weight monitoring. Limitations include the absence of more dose groups to better define the dose–response relationship, with the eventual goal of developing a reliable PFOS therapeutic profile.

While effective against this animal model of familial adenomatous polyposis, PFOS efficacy in the more common acquired form of colorectal cancer is another important focus for future investigations. Planned studies will seek to determine a mechanism, such as might be derived from isolated cell cultures, an optimal dose in vivo using xenograft models. A mechanistic understanding of PFOS action, from the impetus provided here, may lead to successful new colorectal cancer treatment approaches.