The literature selection process for animal and human epidemiological studies for this systematic review is shown in Figure 1. We first assessed the strength of evidence for an association between declines in semen quality and exposure to PCBs and individual PCB congeners. Next, we used data from eligible studies to derive a reference dose to be used in mixture risk assessment for declines in semen quality.
Strength of evidence: Experimental studies in laboratory animals
Study selection and evaluation
Overall, we identified 33 publications that assessed links between semen quality in vivo and exposure to PCBs. Of these, 15 publications reported on declines in semen quality in vivo upon treatment with individual PCB congeners. Because some studies examined two PCB congeners, we extracted data for a total of 18 separate experimental observations for individual congeners. The studies were conducted in rats, mice or goats. We identified four studies examining the effects of PCB-77 (Faqi et al. 1998a, 1998b, Hsu et al. 2004, Huang et al. 1998), two studies on those of PCB-118 (He et al. 2020, Kuriyama and Chahoud 2004), four studies which looked at PCB-126 (Faqi et al. 1998b, Oskam et al. 2005, Wakui et al. 2007, 2010), two studies for PCB-132 (Hsu et al. 2007, 2003), one study for PCB-149 (Hsu et al. 2003), two studies reporting on PCB-153 (Oskam et al. 2005, Xiao et al. 2010), two on PCB-169 (Wolf et al. 1999, Xiao et al. 2011) and one report on PCB-180 (Alarcón et al. 2021). All these studies were selected for the data extraction process.
An additional 18 studies which described the effects of PCB mixtures, commercial (Aroclor 1242, 1254, and 1260) or other PCB mixtures were identified. We did not fully evaluate the 18 studies which tested the effects of PCB mixtures (Aroclor 1242, 1254, and 1260 or 1:1 PCB-101/-118) because they were unsuitable for derivation of congener-specific reference doses. However, we summarise their findings in support of the overall evidence. Two studies in mice tested 1:1 mixtures of PCB-101 and -118 and both found decreases in sperm viability (Fiandanese et al. 2016, Pocar et al. 2012). Three studies reported increases in daily sperm production upon treatment with Aroclor mixtures, two of those tested Aroclor 1242 (Fielden et al. 2001, Kim 2001) and one Aroclor 1242 and 1254. Two studies of Aroclor 1254 in rats observed no effects on the examined semen parameters (Gray et al. 1993, Sager et al. 1991). The remaining 11 studies all observed adverse effects upon treatment with Aroclor. Only one tested Aroclor 1260 in rats, and reported decreases in sperm count, motility and daily sperm production (Aly et al. 2016). All others tested Aroclor 1254 and adverse effects on various sperm parameters, including number, concentration, motility and morphology were reported (Aly et al. 2009, 2016, Anbalagan et al. 2003, Cai et al. 2015, 2011, Sanders et al. 1977). Furthermore, Aroclor 1254 was used to induce declines in semen quality to test beneficial co-exposures in five studies (Atessahin et al. 2010, Güleş and Eren 2016, Krishnamoorthy et al. 2007, 2013, Mazen and Zidan 2017).
We evaluated the internal validity of the 18 experimental observations from the 15 studies which investigated individual congeners by carrying out a risk of bias analysis. All the studies met the key appraisal elements with a rating of “probably low” or “definitely low risk” (Table 1). None of the studies were disqualified due to failure of other elements. The only element which received rankings of “definitely high risk” was inadequate reporting on funding sources and conflicts of interest (14 studies). “Probably high risk” was assigned to the 8 studies that used soy containing diets, and due to a lack of information on attrition and detection in one study (Table 1).
Congener-specific studies
PCB-77: Of the four studies examining PCB-77, three were conducted in rats (Faqi et al. 1998a, 1998b, Hsu et al. 2004) and one in mice (Huang et al. 1998). All four studies were rated as “probably low” or “definitely low risk” in all key elements, and one had only one “definitely high risk” for another element and was assigned to TIER 1, or High confidence (Faqi et al. 1998a). The other three had “probably high” or “definitely high risk” ratings in two of the remaining elements and thus were assigned to an overall Medium confidence (TIER 2) (Faqi et al. 1998b, Hsu et al. 2004, Huang et al. 1998). One rat study which found an increase in daily sperm production upon treatment with PCB-77 only tested one PCB dose (0.1 mg/kg/d at GD15 via maternal gavage) and was therefore excluded from consideration as a basis for deriving a reference dose (Faqi et al. 1998b). The other rat study from this group established a decrease in daily sperm production with a LOAEL of 18 mg/kg/d (Faqi et al. 1998a). However, this study used s.c. injection of PCB-77 and was therefore only included as evidence for a link between PCB-77 and reduced semen quality but not considered for derivation of a reference dose. The only mouse study did not report on the purity of the compound, but was still considered TIER 2 because PCB-77 was analytically confirmed in the treatments (Huang et al. 1998). This study found no effect on semen quality. Finally, Hsu et al. (2004) reported a decline in semen quality upon i.p. injection of PCB-77 (NOAEL = 2 mg/kg/d) and was considered for derivation of a reference dose (Hsu et al. 2004).
PCB-118: Of the two studies of PCB-118, one was conducted in the rat (Kuriyama and Chahoud 2004), and the other in mice (He et al. 2020). We rated the key elements of the study by Kuriyama and Chahood (2004) as “probably low risk” or “definitely low risk”, but failed some of the additional elements and therefore assigned an overall Medium confidence (TIER 2) (Kuriyama and Chahoud 2004). The mouse study was “definitely low” or “probably low risk” in all elements and was therefore considered TIER 1 or of High confidence (He et al. 2020). Both studies reported a decline in semen quality. However, the rat study only tested one dose of PCB-118 (0.375 mg/kg/d) and was thus not taken forward for reference dose derivation (Kuriyama and Chahoud 2004). The mouse study which reported a LOAEL of 0.02 mg/kg/d was used to derive a reference dose for PCB-118 (He et al. 2020).
PCB-126: There were four studies of PCB-126, of which one was conducted in goats (Oskam et al. 2005) and the other three in rats (Faqi et al. 1998b, Wakui et al. 2007, 2010). The goat study and two of the rat studies were rated at an overall confidence level of “High” (TIER 1) due to all elements being evaluated as “definitely low” or “probably low risk” (Oskam et al. 2005, Wakui et al. 2007, 2010). The third rat study was evaluated as “definitely low” or “probably low risk” in the key elements but had some other elements rated lower and was thus assigned to TIER 2, Medium confidence (Faqi et al. 1998b). The goat study (Oskam et al. 2005) and one rat study (Faqi et al. 1998b) both reported no effect of PCB-126 on semen quality, however, both studies also only tested one dose and would not have qualified for derivation of a reference dose. Of the other two studies one did not show significant effects, but a trend towards declining semen quality (Wakui et al. 2007). These trends were confirmed in a later study by the same group after including higher doses, and we used their NOAEL of 2.50E-05 mg/kg/d to derive a reference dose (Wakui et al. 2010).
PCB-132: Both studies we identified for PCB-132 were conducted in rats (Hsu et al. 2007, 2003) and were evaluated as “definitely low” or “probably low risk” in the key elements but had some other elements rated lower and were therefore considered to be of Medium confidence (TIER 2). Both studies used i.p. injection of PCB-132 and reported declines in semen quality. One study was conducted in juvenile rats and determined a LOAEL of 9.6 mg/kg/d (Hsu et al. 2003) whereas the second studied prenatal exposure to PCB-132 (LOAEL = 1 mg/kg/d) (Hsu et al. 2007). Both studies were considered for derivation of a reference dose.
PCB-149: The only available study on PCB-149 was evaluated as “definitely low” or “probably low risk” in the key elements but had other elements rated lower and was assigned to TIER 2 (Medium confidence) (Hsu et al. 2003). This study was conducted in juvenile rats, used i.p. injection of PCB-149 and estimated a NOAEL of 9.6 mg/kg/d which was used to derive a reference dose.
PCB-153:Of the two studies reporting on PCB-153, one was conducted in goats (Oskam et al. 2005) and the other in rats (Xiao et al. 2010). The goat study was evaluated as “definitely low” or “probably low risk” in all elements and rated at an overall confidence level of “High” (TIER 1) (Oskam et al. 2005). The rat study was also assigned to High confidence (TIER 1) as only one additional element was rated lower (Xiao et al. 2010). The goat study (Oskam et al. 2005) did not find any effects of PCB-153 on semen quality and it also tested only one dose and would not have qualified for derivation of a reference dose. The rat study was conducted in pups and reported a NOAEL of 0.025 mg/kg/d which was used to derive a reference dose (Xiao et al. 2010).
PCB-169: The two studies that examined associations between declines in semen quality and PCB-169 exposure were conducted in rats (Wolf et al. 1999, Xiao et al. 2011). Both were rated as “definitely low” or “probably low risk” in the key elements. One had only one additional element rated at definitely high risk and was therefore of overall High confidence (TIER 1) (Xiao et al. 2011). The second study was rated lower at two other elements and was thus assigned to TIER 2 (Medium confidence) (Wolf et al. 1999). This study examined prenatal exposure to PCB-169 exposure and found declines in semen counts (Wolf et al. 1999). However, it tested only one dose (1.8 mg/kg/d) and was thus not used to derive a reference value. The second rat study used neonatal exposures and reported declines of semen quality with a LOAEL of 0.025 mg/kg/d which was used to derive a reference value (Xiao et al. 2011).
PCB-180: PCB-180 was orally administered to rats during gestation (Alarcón et al. 2021). We assessed this study as “definitely low” or “probably low risk” in the key elements, but lower in other elements and thus assigned to Medium confidence (TIER 2). However, the focus of the study was on other endpoints and declines in sperm counts were only observed in three out of seven animals, and only in those with damage to the seminiferous tubule sperm counts. Furthermore, sperm counts were only assessed at one, the highest, exposure dose (250 mg/kg/d). Therefore, no reference value could be derived for PCB-180.
Overall study confidence ratings
Overall, eight of the 18 studies on individual PCB congeners were assigned to TIER 1 (High confidence). These included one study investigating PCB-77, one study on PCB-118, three on PCB-126, two on PCB-153 and one testing PCB-169. The remaining ten studies were rated as Medium confidence (TIER 2), mainly because they had been rated as “definitely high risk” due to deficient reporting on funding sources or conflict of interest and an assessment of “probably high risk” due to the use of soy-based diet an in one case lack of information on the methods and timepoint for endpoint measurements. None of the studies were considered to be of Low confidence (TIER 3) since they all were rated at a sufficiently low risk in all key and other elements of the assessment. A detailed summary of all risk of bias assessments and confidence ratings is shown in Table 1.
Evidence synthesis
A summary of the study evaluations for all individual PCB congeners is shown in Table 2. Of the 18 observations, the majority described some adverse effect on selected semen quality parameters, while four studies reported no effects (Faqi et al. 1998b, Huang et al. 1998, Oskam et al. 2005) and one study even observed an increase in daily sperm production (Faqi et al. 1998a).
We rated the overall evidence of an effect of PCB-77 on semen quality as Slight: One TIER 1 (Faqi et al. 1998b) and one TIER 2 study (Hsu et al. 2004) showed declines in semen quality, but these effects were not seen in other studies (Faqi et al. 1998a, Huang et al. 1998).
The evidence for declines in semen quality after PCB-118 exposure was assessed as Robust. The two available studies, one a TIER 1 study (He et al. 2020), the other a TIER 2 study (Kuriyama and Chahoud 2004), both reported disrupted sperm parameters.
The overall evidence for links between PCB-126 and deteriorations of semen quality is Moderate: Of the four available studies, two high confidence (TIER 1) studies observed a decrease in sperm counts. Due to low administered doses the effects in one study did not reach statistical significance (Wakui et al. 2007), but significant effects were seen in a follow-up study with higher doses (Wakui et al. 2010). One TIER 1 study (Faqi et al. 1998b) and one TIER 2 study (Oskam et al. 2005) did not demonstrate effects, but tested only one dose which may well have precluded detection of changed semen parameters.
The two TIER 2 studies examining PCB-132 (Hsu et al. 2007, 2003) reported declines on semen quality. We did not identify additional TIER 1 studies, but the evidence for declines in semen quality associated with PCB-132 exposures was consistent and we therefore ranked it as Moderate.
We identified only one study which tested PCB-149 (Hsu et al. 2003). This TIER 2 study in rats described decreases in sperm quality, and accordingly, we considered the overall strength of evidence to be Moderate.
The two studies that examined PCB-153 exposures were rated as high confidence (TIER 1). One of them (Oskam et al. 2005) was carried out in goats (see also PCB-126) and did not find any effects on semen quality. In this study only one dose of PCB-153 was tested, which was described as low dose. Thus, the absence of effects in this study is not conclusive. The second study was conducted in rats and found a decrease in daily sperm production (Xiao et al. 2010). Due to the clear effects in the high confidence study in rats, the explanation for the lack of effects in the goat study and in absence of further supporting or conflicting evidence, we considered the evidence for PCB-153 to be Moderate.
The effects of PCB-169 were investigated in two studies, one of overall high (Xiao et al. 2011) and the second of medium (Wolf et al. 1999) confidence. Both studies described declines in sperm counts. In the absence of conflicting evidence, the overall evidence for PCB-169 was regarded as Robust.
One study examined PCB-180 and found decreases in sperm counts in a subgroup of animals in the treatment group (Alarcón et al. 2021). Although the study was of overall medium confidence, sperm counts were only assessed at the highest dose tested and the findings were equivocal. Therefore, in absence of additional studies, we consider the evidence for PCB-180 to be Indeterminate.
Strength of evidence: Human epidemiological studies
Study selection and evaluation
We identified 23 human epidemiological studies from the full text screening which were selected for data extraction and RoB assessment (Table 3). Most of these studies measured multiple PCB congeners, often in combination with other organochlorines or additional POPs. A few focused on single congeners, such as PCB-153 (Giwercman et al. 2007, Haugen et al. 2011, Lenters et al. 2015, Richthoff et al. 2003, Rignell-Hydbom et al. 2004, Toft et al. 2006). Combinations of PCBs-118, -138, -153 and -180 with other POPs were measured in six studies (Abdelouahab et al. 2011, Dallinga et al. 2002, Den Hond et al. 2015a, Hauser et al. 2002, 2003, Vitku et al. 2016). The remaining ten publications looked at a larger set of PCBs (Emmett et al. 1988, Kobayashi et al. 2017, Magnusdottir et al. 2005, Mínguez-Alarcón et al. 2017, Mumford et al. 2015, Paul et al. 2017, Petersen et al. 2015, 2018, Pines et al. 1987, Vested et al. 2014, Weiss et al. 2006).
The ideal assessment of exposure to PCBs would be in maternal serum during pregnancy, as foetal development is a critical time period for semen quality in adulthood (Skakkebaek et al. 2015). Only one of the eligible studies met these criteria, which measured PCBs in maternal serum, collected in pregnancy week 30 and semen quality in the sons (19-21 year old) (Vested et al. 2014).
In adult men, the duration of spermatogenesis is around 75 days plus an additional 12 days of maturation. Because PCBs bioaccumulate in fatty tissues, it is likely that existing exposures last over the entire period of spermatogenesis. The exposure assessment element in studies with a general description of sampling, extraction and analytical techniques was rated as “adequate” (Dallinga et al. 2002, Den Hond et al. 2015a, Emmett et al. 1988, Giwercman et al. 2007, Mumford et al. 2015, Petersen et al. 2018, Pines et al. 1987, Weiss et al. 2006). The studies which provided detailed descriptions of quality assurance and analytical performance were evaluated as “good” with respect to the exposure aspect (Abdelouahab et al. 2011, Haugen et al. 2011, Hauser et al. 2002, 2003, Kobayashi et al. 2017, Lenters et al. 2015, Magnusdottir et al. 2005, Minguez-Alarcon et al. 2017, Paul et al. 2017, Petersen et al. 2015, Richthoff et al. 2003, Rignell-Hydbom et al. 2004, Toft et al. 2006, Vested et al. 2014, Vitku et al. 2016).
We assessed outcome measurement elements in relation to adherence to established quality standards described in the WHO guidelines (WHO 2010). These guidelines recommend the analysis of core semen parameters (number, concentration, motility and morphology). If all these parameters were analysed according to WHO standards, we evaluated the outcome measurement as “good” (Abdelouahab et al. 2011, Den Hond et al. 2015a, Giwercman et al. 2007, Haugen et al. 2011, Hauser et al. 2002, 2003, Lenters et al. 2015, Mumford et al. 2015, Paul et al. 2017, Petersen et al. 2018, 2015, Rignell-Hydbom et al. 2004, Toft et al. 2006, Vested et al. 2014). Studies which lacked details about the methods (Pines et al. 1987) or only reported sperm numbers (Emmett et al. 1988) were rated as “poor”. All other studies, which conducted the outcome measurement according to WHO guidance, but did not provide all details on sampling and analysis or did not include sperm morphology measurements were evaluated as “adequate” (Dallinga et al. 2002, Kobayashi et al. 2017, Magnusdottir et al. 2005, Minguez-Alarcon et al. 2017, Richthoff et al. 2003, Vitku et al. 2016, Weiss et al. 2006).
Studies which selected participants from the general population with no apparent selection bias were rated as “good” (Giwercman et al. 2007, Haugen et al. 2011, Lenters et al. 2015, Mínguez-Alarcón et al. 2017, Petersen et al. 2015, Rignell-Hydbom et al. 2004, Toft et al. 2006, Vested et al. 2014). One study included infertile patients without control groups and was therefore evaluated as "critically deficient” (Weiss et al. 2006). One study provided limited information on participant selection and was rated as poor in relation to participant selection (Kobayashi et al. 2017). Another study which was part of a series of publications only referred to the description of the recruitment process in another publication was rated as “adequate/poor” (Emmett et al. 1988). The remaining studies were from fertility clinic or occupational settings and were classed as “adequate” (Abdelouahab et al. 2011, Dallinga et al. 2002, Den Hond et al. 2015b, Hauser et al. 2002, 2003, Magnusdottir et al. 2005, Paul et al. 2017, Pines et al. 1987, Rignell-Hydbom et al. 2004, Vitku et al. 2016).
We evaluated the quality of control for confounding by checking whether the following factors were accounted for: age, abstinence time, smoking history, body mass index and chronic disease status (Sánchez-Pozo et al. 2013). Alcohol use and stress could also be considered, but are less well established. The majority of eligible studies took account of the key confounders and accordingly were rated as “good”. Where the key confounders were considered but some details were missing, we rated the study as “adequate” (Giwercman et al. 2007, Hauser et al. 2002, Vitku et al. 2016). Studies which did not provide information on abstinence time were evaluated as “poor” (Abdelouahab et al. 2011, Dallinga et al. 2002, Kobayashi et al. 2017, Pines et al. 1987, Weiss et al. 2006).
When examining associations between declines in semen quality and exposure to PCBs, semen parameters should be analysed as continuous parameters to avoid misclassifications. Furthermore, sufficient detail should be provided, such as confidence intervals and standard errors, in addition to significance. Most of the studies fulfilled these criteria and were evaluated as “good” for data analysis. Weiss et al. (2006) did not provide sufficient detail on the analysis and did not show their data and was therefore rated as “critically deficient”. If the data were dichotomised or some minor details on the analysis and results were not provided, the studies were rated as “adequate” (Abdelouahab et al. 2011, Emmett et al. 1988, Giwercman et al. 2007, Haugen et al. 2011, Hauser et al. 2002, 2003, Kobayashi et al. 2017, Mumford et al. 2015). Studies with missing details to warrant an “adequate” rating were rated as “poor”(Dallinga et al. 2002, Den Hond et al. 2015a, Magnusdottir et al. 2005, Pines et al. 1987).
Overall study confidence ratings
We assigned overall study confidence ratings based on the ratings in the individual study evaluation elements. In cases where all or at least four of the evaluation aspects were rated as “good” and one as “adequate”, we assigned an overall “High” confidence rating. If two or three elements were rated as “good” and the remaining ones as “adequate” or maximally one as “poor”, we allocated an overall confidence of “Medium”. If two elements were considered to be “poor” in addition to “adequate” or “good” ratings, the overall confidence was pegged at a rating of “Low”. Where three or more ratings were “poor” or found to be “critically deficient”, the overall confidence was classed as “Uninformative”. We applied these decision rules to the eligible studies and arrived at the overall confidence ratings summarised in Table 3.
Evidence synthesis
The outcomes of the 23 eligible epidemiological studies are summarised in Table 3. Nine studies reported null findings. One of these was judged to be “Uninformative” (Weiss et al. 2006). Two studies with null results were of “low” overall confidence (Emmett et al. 1988) (Abdelouahab et al. 2011), two of “medium” confidence (Magnusdottir et al. 2005, Den Hond et al. 2015) and four studies were of “high” confidence (Minguez-Alarcon et al. 2017, Petersen et al. 2015, 2018, Vested et al. 2014).
Among the studies which reported effects, a diverse picture emerged. Four studies report mixed findings, with declines in semen quality for some PCB congeners or PCB metabolites, and improved semen parameters for other congeners in exposed populations compared to controls. One study which reported no effects for the congeners, declines in quality for metabolites and improvements for the sum of PCBs was rated as “low” confidence (Dallinga et al. 2002). The study by Mumford et al. (2015) was of “medium” confidence and reported mix of declines or improvement for semen parameters, dependent on congener (Table 3). Two studies that found mixed results depending on congener and outcome measure were of “high” confidence (Paul et al. 2017, Toft et al. 2006).
We identified three studies which only report improved semen parameters in exposed populations compared to controls for some parameters. Two of those were of “medium” confidence (Magnusdottir et al. 2005, Vitku et al. 2016) and one study was of “high” confidence (Haugen et al. 2011).
The remaining eight studies all reported declines in semen quality for one or more parameters. One of these studies was considered to be “Uninformative” (Pines et al. 1987) and a second was judged to be “low” confidence (Kobayashi et al. 2018). We identified three “medium” confidence studies that reported declines in semen quality (Giwercman et al. 2007, Hauser et al. 2002, 2003) and an additional three “high” confidence studies (Lenters et al. 2015, Richthoff et al. 2003, Rignell-Hydbom et al. 2004).
Overall weight of evidence from human and experimental studies
There is Robust evidence from animal studies that PCB congeners -118 and -169 exposures lead to declines in semen quality. For congeners -126, -132 and -153 the evidence is Moderate. The evidence for PCB-77 from animal studies is only Slight and for PCB-180 the evidence was Indeterminate. In humans, only one study was available which measured PCB exposure during foetal life and assessed the semen quality in adults, and this study did not find any changes. Overall, the evidence from human epidemiological studies in adults is mixed and not all individual congeners have been examined. We did not identify human evidence for PCBs-77, -132, and 149. PCB-153 was investigated in several studies and the majority found declines in semen quality parameters, in line with the animal evidence, although studies reporting improved parameters do exist. One epidemiological study that included PCB-126 and another including PCB-169 supported the evidence from animal studies. For PCB-118 the human evidence was weak but generally in support of the animal studies. The evidence for PCB-180 from epidemiological studies was equivocal. Overall, the evidence from human studies is sufficiently robust to support hazard identification for some congeners and the commercial mixtures. We therefore used the evidence from animal studies to derive a reference dose for declines in semen quality for selected PCB congeners with sufficient evidence.
Derivation of reference doses for declines in semen quality for PCB-118, -126, -132, -149, -153 and -169
We derived reference doses for PCB congeners with a Moderate or Robust evidence rating from animal studies and where there was no conflicting human evidence. Consequently, we estimated reference doses for PCB-118, -126, -132, -149, -153 and -169. PCB-77 and 180 were excluded as their confidence rating did not reach Moderate. Where studies reported data from three or more different dose groups, we attempted BMD modelling to estimate a BMDL5. However, none of the selected studies provided adequate data and therefore we decided to use the NOAEL values as PoDs for all PCB congeners. Where only a LOAEL value was reported, we extrapolated a NOAEL using an AF of 3. Studies which only tested one dose were considered as supporting evidence, but were excluded from the derivation of reference doses. Table 4 shows the PoDs derived from the studies which were included in the calculation of reference dose values.
PCB-118: One TIER 1 study qualified for derivation of a reference dose for PCB-118 (He et al. 2020). In this study PCB-118 was orally administered to mice during gestation (daily from Gestational Day (GD) 7.5 to GD 12.5). Two dose groups were exposed, and the authors reported a LOAEL of 20 µg/kg/d for declines in sperm with normal morphology. Using an AF of 3, we extrapolated a NOAEL of 6.67 µg/kg/d. By using the toxicokinetic parameters for PCB-118 ( t1/2,a = 117 days, Fabs,a = 0.9 for the mouse and t1/2,h = 3395 days, Fabs,h = 1 for the human) we first calculated the cumulative critical body burden at the NOAEL in the mouse before estimating the EHDI. The critical body burden was 35.5 µg/kg/d and the estimated EHDI was 0.00725 µg/kg/d. By applying the AF of 2.5, we derived reference dose value of 0.0029 µg/kg/d (Table 4).
PCB-126: The reference value for PCB-126 was derived from one TIER 1 rat study which used 3 dose groups, and repeat administration from GD13 to GD19 (Wakui et al. 2010). The study determined a NOAEL of 0.25 µg/kg/d for declines in sperm numbers. With the kinetic parameters for PCB-126 ( t1/2,a = 100 days, Fabs,a = 0.9 for the rat and t1/2,h = 584 days, Fabs,h = 1 for the human) we estimated the critical body burden as 0.154 µg/kg/d and the corresponding EHDI as 0.00018 µg/kg/d. Applying the AF of 2.5 resulted in a reference dose value of 0.000073 µg/kg/d for PCB-126 (Table 4).
PCB-132: We identified two TIER 2 rat studies which were eligible for inclusion in the derivation of a reference dose for PCB-132 (Hsu et al. 2007, 2003). Both studies used a single i.p. administration in two dose groups, one during foetal development (GD15) (Hsu et al. 2007) and in juvenile animals at Postnatal Day (PND) 15 (Hsu et al. 2003). One of these studies reported a LOAEL of 1000 µg/kg/d for reductions in sperm numbers, which was extrapolated to a NOAEL of 333.33 µg/kg/d by using an AF of 3 (Hsu et al. 2007). The other study observed a higher LOAEL of 9600 µg/kg/d for declines in motility, which we extrapolated to a NOAEL of 3200 µg/kg/d (Hsu et al. 2003). Both studies used a single administration, thus, using an absorption of 90% in rodents, we calculated the critical body burden of PCB-132 at PoD by multiplying the NOAEL with the absorbed fraction, resulting in a body burden of 300 µg/kg/d (Hsu et al. 2007) or 2880 µg/kg/d (Hsu et al. 2003). Applying the toxicokinetic parameters for PCB-132 (t1/2,a = 100 days, Fabs,a = 0.9 for the rat and t1/2,h = 3650 days, Fabs,h = 1 for humans) we calculated EHDI values of 0.057 µg/kg/d (Hsu et al. 2007) and 0.547 µg/kg/d (Hsu et al. 2003). The reference doses were derived using an AF of 2.5, resulting in values of 0.0228 µg/kg/d (Hsu et al. 2007) and 0.219 µg/kg/d (Hsu et al. 2003). The lower value derived from the gestational exposure study (0.0228 µg/kg/d) was chosen as reference dose for PCB-132 (Table 4).
PCB-149: The reference dose for PCB-149 was derived from the TIER 2 study in juvenile rats which also tested PCB-132 (Hsu et al. 2003). The authors used a single i.p. administration at PND 15 and reported a NOAEL of 9600 µg/kg/d for reductions in sperm motility and velocity. Assuming 90% absorption, we calculated a critical body burden of 8640 µg/kg/d. With the toxicokinetic parameters for PCB-149 (t1/2,a = 100 days, Fabs,a = 0.9 for the rat and t1/2,h = 3650 days, Fabs,h = 1 for humans), we estimated an EHDI of 1.641 µg/kg/d. Using the AF of 2.5 we calculated the reference dose value of 0.656 µg/kg/d (Table 4).
PCB-153: We used one TIER 1 rat study with two dose groups and repeat administration in pups (PND3) to derive a reference dose value for PCB-153 (Xiao et al. 2010). The study determined a NOAEL of 25 µg/kg/d for reductions in daily sperm productions as PoD. The PCB-153 toxicokinetic parameters ( t1/2,a = 113 days, Fabs,a = 0.9 for the rat and t1/2,h = 5256 days, Fabs,h = 1 for the human) were used to calculate the critical body burden in the animal (111 µg/kg/d) and the corresponding EHDI (0.0147 µg/kg/d). We applied the AF of 2.5 to derive the reference dose value of 0.00586 µg/kg/d (Table 4).
PCB-169: One TIER 1 study in the rat was available to derive a reference dose for PCB-169 (Xiao et al. 2011). Using repeat oral dosing from PND1 to 7 in 3 dose groups, the authors reported a LOAEL of 25 µg/kg/d for decreases in sperm numbers and daily sperm production. We extrapolated the NOAEL (8.33 µg/kg/d) by applying an AF of 3. We estimated the critical body burden in the rat and the EHDI using the kinetic parameters for PCB-169 ( t1/2,a = 117 days, Fabs,a = 0.9 for the mouse and t1/2,h = 3395 days, Fabs,h = 1 for the human). The cumulative critical body burden had a value of 51.2 µg/kg/d, resulting in an EHDI of 0.0133 µg/kg/d. Finally, we applied the AF of 2.5 to account for differences between humans, resulting in a reference dose value of 0.00533 µg/kg/d for PCB-169 (Table 4).
Comparison of reference doses with PCB exposures
To evaluate whether current exposures to specific PCB congeners exceed any of the above reference doses for deteriorations in semen quality, we used exposure data from the European Union.
The average exposures of European adults to PCB-169 via food are around 0.00079 ng/kg/d, but these can increase to 0.0024 ng/kg/d (mean and 95th percentile LB intake for adults, calculated from the percentage contribution of individual congeners to sums of dl-PCBs (EFSA 2018)). Both these values are far below the reference dose of 5.33 ng/kg/d.
For PCB-126, the average exposures via food are around 0.0035 ng/kg/d, with high levels rising to 0.01 ng/kg/d (EFSA 2018). Whereas the average value is well below the reference dose of 0.073 ng/kg/d, the high exposure is less than an order of magnitude below the reference dose, resulting in a Risk Quotient of 0.14.
Average exposures to PCB-118 via food are around 0.576 ng/kg/d, with high exposures up to 1.7 ng/kg/d (EFSA 2018). Both these values are relatively close to the reference dose of 2.9 ng/kg/d, resulting in Risk Quotients of 0.2 and 0.59 respectively. We did not identify exposure levels for PCB-132, -149, or -153. PCB-153 is frequently assessed as part of the sum of 6 indicator PCBs, which also includes PCB-118. PCB levels in food are highly correlated and PCB-153 is often present at levels up to three times higher than PCB-118 (EFSA 2005). Thus, as a worst-case assumption average and high exposures to PCB-153 via the diet could be estimated to be around 1.7 ng/kg/d and 5.1 ng/kg/d respectively. This would also put the exposures close to the reference value of 5.86 ng/kg/d with Risk Quotients of 0.29 and 0.87 for average and high exposures, respectively. No exposures for PCB-132 and -149 could be retrieved, however, these congeners are not part of common indicator PCB groups and are with their higher reference doses of 22.8 ng/kg/d (PCB-132) and 656 ng/kg/d (PCB-149) likely of lower concern.
The overall Hazard Index for PCB-118, -126, -153 and -169 for average exposures observed in European adults would be 0.54, relatively close to the value of 1. For the higher exposure scenario, the Hazard Index is 1.58 and therefore exceeding the index value of 1.
