What this study adds
Perfluorooctanoate and perfluorooctanesulfonate are synthetically produced chemicals which have been found carcinogenic in animal studies, whereas there is limited evidence from human studies to draw any conclusions. Prostate and bladder cancers are among the most common cancers, and perfluorooctanoate and perfluorooctanesulfonate have been suggested to be associated with both disease incidence and mortality, but no studies have investigated their association with cancer survival. This first study found no association among prostate cancer patients and an inverse association among bladder cancer patients. However, as the first study of its kind, the findings require replication before any firm conclusions can be drawn.
Perfluorooctanoate (PFOA) and perfluorooctanesulfonate (PFOS) are synthetically produced chemicals, which are used in a range of industries and consumer products as surfactants, due to their water- and oil-resistant properties.1 From a public health perspective, however, concern was raised when the compounds were found to be ubiquitously present in the environment due to their persistent nature. Furthermore, they bioaccumulate in the food chain, have long half-lives in humans,1,2 and are found in human blood samples from the general population all over the world.3,4
Animal studies have shown an increased incidence of liver, pancreas, and testicular tumors in exposed rodents,1,4 whereas studies on PFOA and PFOS and cancer in human populations have generally produced conflicting results, and they are inadequate to draw strong conclusions regarding carcinogenity in humans.4–6 Hence, the International Agency for Research on Cancer has classified them as a “possible carcinogen” (Group 2B), primarily based on evidence for testicular, renal, and kidney cancer.6 However, as the number of human studies are relatively limited, conducted in diverse populations, and most with few cancer cases, further studies are required.1,4
Prostate and bladder cancers are among the most common cancers: prostate cancer is the second most common nonskin cancer in men worldwide7 and bladder cancer the sixth.8 Some studies have suggested an association with PFOA and/or PFOS in relation to disease incidence or mortality for both cancers,9–17 whereas others have not.11,13,15,18 However, most studies, especially those on mortality, have been hampered by a limited number of cases. No study has previously investigated the association with survival after a cancer diagnosis. Given the high occurrence of both prostate and bladder cancer, it is very relevant to investigate factors that may affect survival from these diseases. In the Diet, Cancer and Health (DCH) cohort, in which the present study is conducted, we have previously investigated the association between PFOA/PFOS and prostate and bladder cancer incidence, finding no association with bladder cancer and a suggestion of a direct association with prostate cancer in categorical analyses, only.13
The mechanisms rendering PFOA and PFOS involved in the initial development of cancer may differ from its role in cancer survival. However, extremely few studies have investigated the role of PFOA/PFOS in relation to cancer promotion and progression. Experiments with rats fed with PFOS or PFOA have indicated increased cell growth and proliferation, but it is difficult to extrapolate from rodents to humans.5 In vitro experiments using a human breast cell line (MCF-10A) showed higher growth after PFOS treatment, which also stimulated MCF-10A cell migration and invasion.19 Another in vitro experiment showed that treatment with PFOS stimulated proliferation of human glioblastoma cells (T98G).20 Thus, the evidence is too limited to conclude anything in relation to cancer survival. At present, no strong biological hypotheses support an association between PFOA/PFOS and prostate and bladder cancer survival; however, the present study is meant as hypothesis-generating, as the long follow-up time accumulated in our cohort gives us a unique opportunity to investigate such a potential association. Hence, the aim of the present study was to investigate the association between PFOA and PFOS concentrations in plasma samples and subsequent mortality in two survivor cohorts of prostate and bladder cancer patients, respectively, among participants in a large population-based Danish cohort.
Material and methods
A detailed description of the DCH cohort is published previously.21 Briefly, 160,725 Danes were invited to participate from 1993 to 1997. Inclusion criteria were residence in the greater Copenhagen or Aarhus area, 50–64 years of age, and no previous cancer diagnosis in the Danish Cancer Registry. In total, 57,053 participants (52% women) accepted and were included into the study, representing 7% of the Danish population in this age group. At baseline, all participants filled in lifestyle and food frequency questionnaires and were subjected to anthropometrical measurements. They also provided urine and blood samples. Blood samples were collected as 30 mL of blood. The samples were immediately protected from sunlight and left for 30 minutes on ice. Afterwards, tubes were centrifuged for 10 minutes at 3,000 rpm and a temperature of 5°C. The samples were then aliquoted into cryotubes, including 6 × 1 mL plasma. Within 2 hours of collection, samples were stored at −20°C, and at the end of the day they were placed in liquid nitrogen vapor at −150°C.
The study was approved by the local ethical committees of Copenhagen and Frederiksberg Municipalities. All participants provided written informed consent, and the study was conducted according to the Helsinki Declaration.
Participants have been followed up in Danish registries on cancer and mortality since baseline, and between baseline and July 1, 2006, 688 men were diagnosed with prostate cancer (ICD-10 code C61), and 308 men and women were diagnosed with bladder cancer (ICD-10 code C67) as their first cancer in the Danish Cancer Registry.22
PFOA and PFOS analyses
Plasma concentrations of PFOA and PFOS were measured by use of high-pressure liquid chromatography coupled to tandem mass spectrometry in the 3M Toxicology Laboratory, as described in Ehresman, 2007.23 The laboratory was blinded to disease status. One PFOA value was below the lower limit of quantification (L) of 1 ng/mL and were assigned the value 1/√2, according to the formula L/√2 for replacement of nondetectable values.24 Double determinations of PFOA and PFOS concentrations of 50 random plasma samples were done to test the replication uncertainty of the laboratory measurements. The laboratory was blinded to this series of parallel measurements. Mean coefficients of variation, indicating the replication uncertainty, were very low (5.9% for PFOA and 1.8% for PFOS).
We investigated overall mortality after a prostate or bladder cancer diagnosis as two separate outcomes. Information regarding vital status was collected by linkage to the Danish Civil Registration System.25 As a supplementary analysis, we investigated also prostate- and bladder cancer–specific mortality. Information on cause of death was collected from the Cause of Death Registry.26
At study baseline, all participants filled in a food frequency and a lifestyle questionnaire, and anthropometric measures were collected by trained personnel. Information on prognostic factors and treatment was retrieved from medical records for the prostate cancer patients. Tumor stage was determined by digital rectal examination, whereas lymphadenectomy and/or bone scan were performed, when clinically indicated. Primary treatment was defined as the first initiated treatment, including active surveillance and watchful waiting. All data retrieved from the medical records pertained to the period from diagnosis to first initiated treatment. Missing information on clinical stage in the medical records was supplemented with information from the Cancer Registry, which holds information on TNM staging since 2004. Gleason score was gradually introduced throughout the period, and we made no attempt to combine with the old grading system.
The selection of covariates was done based on a review of existing literature, biological plausibility, and availability of data. We included lifestyle factors and socioeconomic characteristics as covariates in the models, whereas prognostic factors and treatment information were included only in analyses of effect modification, as these were considered intermediate factors on the pathway from cancer diagnosis to mortality.
Cox proportional hazards models, estimating mortality rate ratios (MRR), and 95% confidence intervals (CI) were used to investigate the association between plasma concentrations of PFOA and PFOS and survival. Time since diagnosis was used as the underlying time scale. All analyses of bladder cancer were stratified by sex to allow for separate underlying hazards by sex. Follow-up started at date of cancer diagnosis and continued until date of death, emigration, or February 1, 2016, whichever came first.
Linearity of all continuous variables was investigated by graphical evaluation using linear spline models with knots at the 25th, 50th, and 75th percentiles among deceased.27 No deviation from linearity was observed. The proportional hazards assumption of the Cox models was tested by graphical inspection, which revealed no deviance from this.
MRRs were calculated as crude, that is, adjusted only for calendar year at diagnosis (linear) and age (linear; and for bladder cancer also stratified by sex; Model 1), and additionally for smoking status (never/former/current/missing), alcohol intake (linear, g/day), abstainers (yes/no), waist circumference (linear, cm), leisure-time sports (yes/no/missing), education (≤7/8 to 10/>10 years/missing), and area-level socioeconomic status based on municipality/district information on education, work market affiliation, and income (low/medium/high/missing; Model 2).
Stratified analyses were conducted for sex (bladder cancer only), smoking status (never, former, current), education (low, medium, high), Primary treatment (none, conservative, intended curative, palliative, missing; prostate cancer only), Gleason score (<7, ≥7, missing; prostate cancer only), T-stage (T1, T2, T3+, missing; prostate cancer only), N-stage (N0, N1, missing; prostate cancer only), and M-stage (M0, M1, missing; prostate cancer only), in order to explore effect modification. P values for heterogeneity across strata were calculated using χ2 tests.
All other tests were based on the likelihood ratio test statistic. Two-sided 95% CI were calculated based on Wald’s test of the Cox regression parameter, that is, on the log ratio scale. P values <0.05 were considered statistically significant. The procedure PHREG in SAS, version 9.3, was used for all statistical analyses (SAS Institute Inc., Cary, NC).
Between baseline and July 1, 2006, 996 persons in the DCH cohort were diagnosed with a primary prostate (n = 688) or bladder cancer (n = 308). Of these, we excluded one person who had missing information on all the included lifestyle covariates. This left 995 persons for the present study, 688 prostate cancer cases, and 307 bladder cancer cases. Of these, 356 prostate cases and 149 bladder cancer cases died during follow-up, with prostate or bladder cancer stated as the underlying cause of death for 232 and 66 of these, respectively.
Those who died during follow-up, regardless of whether they were prostate or bladder cancer cases, had a shorter education, were more likely to be current smokers, and drank more alcohol. However, deceased bladder cancer cases were also more likely be teetotalers. Furthermore, deceased participants were less likely to participate in leisure-time sports and to live in areas of high socioeconomic status. Finally, they were slightly older at diagnosis, and deceased bladder cancer cases were more likely to be male than the entire population of bladder cancer cases. For both cancers, deceased patients had a lower median PFOS concentration at baseline, whereas the PFOA concentrations was very similar among all and deceased prostate cancer cases, but deceased bladder cancer cases had a lower median PFOA concentration at baseline than the entire population of bladder cancer cases (5.71 ng/mL vs. 6.45 ng/mL). Deceased prostate cancer patients presented with a more severe diagnosis and more often received palliative rather than intended curative treatment (Table 1).
The median follow-up time for all prostate cancers was 9.9 years (5–95 percentiles: 0.9–16.1), and the corresponding numbers for bladder cancer was 11.0 (0.6–17.8) years.
The Spearman correlation coefficient between PFOA and PFOS concentration was 0.65 for prostate cancer cases and 0.75 for bladder cancer cases.
The association between PFOA and PFOS concentrations and overall mortality for prostate and bladder cancer patients are presented in Table 2. There was no association between PFOA and PFOS concentrations and survival among prostate cancer patients—neither in continuous, nor in categorical analyses. For bladder cancer cases, there was an inverse association between both PFOA and PFOS and subsequent mortality in both categorical and continuous analyses.
We examined the interaction between smoking, education, and sex (bladder cancer only) in Table 3. For bladder cancer, the only significant interaction was between PFOA concentration and smoking (P = 0.04), confining the inverse association to former smokers. For prostate cancer, there was a significant interaction was between PFOA and education; showing an association between PFOA concentration and mortality among those with a high education, but not those with a middle or short education. There were no interactions for PFOS. We further examined interaction between prognostic factors and treatment for prostate cancer cases, where this information was available, but we found no association, neither with PFOA or PFOS (Table s1; http://links.lww.com/EE/A11).
When examining the association between PFOA and PFOS concentrations and prostate- and bladder cancer-specific mortality, we found associations similar to overall mortality, with an inverse association for both PFOA and PFOS in relation to bladder cancer but no association for prostate cancer (Table 4).
The present study investigated the association between PFOA and PFOS concentrations in plasma samples at time of cohort enrolment and mortality after a prostate or bladder cancer diagnosis in middle-aged Danes from the general Danish population. We found no association with neither overall nor prostate cancer–specific mortality among prostate cancer cases, whereas for bladder cancer cases, we found an inverse association in relation to both overall and bladder cancer-specific mortality. There were some suggestions of interaction with education and smoking for PFOA only. There was no interaction with prognostic factors among prostate cancer patients.
A primary study strength is the long and complete follow-up of all included patients, by means of information on date and cause of death from the well-validated Danish Civil Registration System25 and Cause of Death Registry.26 For the included prostate cancer cases, we also had detailed information on prognostic factors and treatment from their medical records. And for both bladder and prostate cancer cases, we were able to include information on a range of lifestyle and socioeconomic factors that are known to be associated with mortality, based on extensive baseline questionnaire data filled out by all participants.
In contrast, the study is limited by a relatively low number of events in the analyses on bladder cancer and in subgroup analyses, rendering limited statistical power. The PFOA and PFOS concentrations were measured in one sample from each participant, only, which was taken at cohort entry (1993–97). Since the study included cases diagnosed until July 1, 2006, and follow-up for mortality continued until February 1, 2016, some participants had their sample taken decades before the outcome of interest. Given the estimated half-lives of 2–4 years of PFOA and PFOS in general, nonoccupationally exposed populations,28–30 this may entail exposure misclassification, as the concentration at this point in time may not reliably reflect the concentration at the clinically relevant time period. In a general population of middle-aged adults, we would expect few change in diet and lifestyle—two known factors affecting the concentration.31,32 However, being diagnosed with cancer may result in changes in lifestyle and diet, which could affect the concentration of PFOS and PFOA during the survival period. Also, having had more than one sample per person could have provided a more precise measurement of PFOA and PFOS concentrations. We expect any measurement error to be random, and therefore it should influence the estimates toward the null. A limitation on the substudy in bladder cancer patients is that we had no information on prognostic factors or treatment, which we expect to influence survival stronger than PFOA/PFOS concentration. For prostate cancer patients, however, where this information was available, we did not find an interaction with prognostic and treatment variables in the effect of PFOA/PFOS on survival. However, a number of the included prostate cancer patients lacked information on one or more of the prognostic or treatment variables (Table 1). The proportion was particularly high among those who died during follow-up. This may be because there was no time to collect data from men who presented with advanced disease and died shortly afterwards or there might have been no need to do so, as they were only offered palliative treatment. Men with missing data on a specific tumor characteristic were excluded from these interaction analyses. Sensitivity analyses excluding men with missing tumor characteristics produced a very similar result (results not shown).
The association between PFOA and PFOS exposure and prostate and bladder cancer has previously been examined in a number of relatively diverse studies, most of which have focused on incidence, and a few on mortality, whereas we have not identified any previous survival studies. Our findings in a survival study may not be directly comparable to findings from mortality studies, as mortality studies represent both the general disease incidence in the population, as well as the survival from the disease, whereas our study only reflects the survival from the disease, but not the disease incidence. And in relation to the incidence studies, there is no guarantee that the findings are comparable to our findings in a survival study, as the association with PFOA/PFOS concentrations may not necessarily be the same for incidence and survival. With this in mind, we have examined the existing literature on PFOA/PFOS and prostate and bladder cancer incidence and mortality
We identified three studies on prostate cancer mortality in occupationally exposed workers, with two small studies in the same population (6 and 16 deaths) finding an association,9,10 whereas a larger study, including 72 deaths, did not find an association.15 With regards to prostate cancer incidence, we identified two studies in occupationally exposed workers,11,12 a geographical analysis of residents living near a PFOA producing plant,14 and two studies in general populations,13,18 but the results of these were inconsistent. A study on PFOA and PFOS and prostate-specific antigen levels did not find an association.33 For bladder cancer, two studies in occupationally exposed workers suggested an increased bladder cancer mortality, but based on 3 and 16 deaths, only.15,16 Regarding bladder cancer incidence, we identified three studies: two in occupationally exposed populations,11,17 and only one in the general population, which was conducted in the DCH cohort, from which the cases of the present study hails,13 but again the results were inconsistent.
Hence, the finding of an inverse association between PFOA and PFOS and subsequent mortality among bladder cancer cases and the lack of any association for prostate cancer cases in the present study are difficult to compare to the above studies. This is further complicated by the fact that most of these are studies on occupational exposure, which in several cases is estimated from a job-exposure matrix or is based on job title or working place, which may not necessarily be a very good proxy for blood concentrations, as used in the present study. Furthermore, the exposure levels in occupational studies will most likely be much higher than in the present study, and this may entail different associations found. Thus, this is the first study investigating prostate and bladder cancer mortality in relation to measured PFOA and PFOS concentrations in a population-based sample, and reproduction of the findings are encouraged before firm conclusions can be drawn.
Our finding of an inverse association with bladder cancer mortality may seem counterintuitive and could be ascribed to limited statistical power. However, it is a relatively consistent finding with a suggestion of a dose–response association also in the quartile analyses, and it may thus be a true association. One possible explanation could be that preclinical bladder cancer was already available in some cases at baseline when the blood sample was taken and that this distorted the PFOA and PFOS concentrations. Hence, we conducted a sensitivity analysis, in which we excluded cases diagnosed within the first 2 years after baseline. However, this produced a similar inverse association (Table s2; http://links.lww.com/EE/A11). A previous study investigating determinants for PFOA and PFOS concentrations in men from the same cohort as the cases from the present study hail found that never smokers had higher PFOA and PFOS concentrations compared to current smokers, with former smokers having intermediate levels. They also found an inverse association between alcohol and body mass index (BMI) in relation to PFOA and PFOS concentrations.34 Hence, one could speculate that the inverse association found with bladder cancer in the present study could be explained by residual confounding from smoking, alcohol consumption, and obesity, which are all well-known risk factors for survival. One would expect, though, that the same residual confounding should have also been present in the analyses of prostate cancer. However, smoking is a very prominent risk factor for bladder cancer, accounting for approximately half of all cases,35 and is also known to affect survival after a bladder cancer diagnosis36 and could therefore be a stronger confounder also in relation to mortality than for prostate cancer, where smoking is a less strong risk-factor.37 Additional adjustment for smoking duration (years) and smoking intensity (g/day) for bladder cancer cases who had this information available did bring the MRRs marginally closer to 1.00, albeit there was still a significantly reduced risk of death: PFOA, 0.79, 95% CI (0.63–0.99); PFOS, 0.72, 95% CI (0.55–0.95), per IQR. Similar adjustment for smoking duration and intensity only marginally affected the estimates for prostate cancer (results not shown). Also, the interaction analyses suggested an interaction between PFOA and smoking in relation to bladder cancer, with a significant inverse association between PFOA concentrations and mortality in bladder cancer patients only among former smokers, which could potentially be explained by residual confounding of smoking, as we see no association among never smokers. Finally, despite the above-mentioned sensitivity analysis, the inverse association with bladder cancer could be explained by reverse causation: Preclinical bladder cancer may entail microscopic hematuria over a prolonged period, and blood loss have been found to result in lower PFOA/PFOS concentrations in relation to other situations entailing blood loss, such as, for example, menstruation, and venesection.38,39
We can, however, not dismiss the inverse association found with bladder cancer mortality, and we thus encourage reproduction of this in future studies, preferably with prognostic and clinical variables available for bladder cancer patients, as we had only for prostate cancer patients in the present study, as we are aware that the extent of disease at time of diagnosis is a much stronger risk factor for survival than the PFOA/PFOS concentration. Unfortunately, however, we did not have these data available for bladder cancer patients in the present study.
We also find an interaction between PFOA and education in relation to prostate cancer, with a direct association with mortality among highly educated prostate cancer patients only. This may be a chance finding due to multiple testing as it is not consistent between PFOA and PFOS and lack a sound underlying explanation.
Information on covariates in the present study were included from a baseline questionnaire filled in before participants were diagnosed with their cancer. Participants may have changed their lifestyle after a cancer diagnosis. A previous study in the DCH cohort investigated changes in tobacco smoke, alcohol intake, and BMI over the first five years of follow-up. The study found no significant difference in positive health behavior changes between men with prostate cancer, compared to cancer-free men. In contrast, in men with a nonprostate cancer, the study found that the cancer patients decreased their BMI and smoked less compared to cancer-free men.40 This suggests that the covariates may be a good proxy for lifestyle between cancer diagnosis and end of follow-up for prostate cancer patients in present study but may be less so for bladder cancer patients, suggesting that additional residual confounding by lifestyle could draw the inverse association seen in the bladder cancer study. This does, however, require that the change in lifestyle is nonrandom and associated with the PFOA and PFOS concentrations, which were measured at baseline, before the cancer diagnosis, rendering this speculative.
Finally, some studies have shown that dietary intake may be the major human exposure pathway of PFOA and PFOS in nonoccupationally exposed populations.31,32 Consumption of both fish and eggs have been suggested determinants of body concentrations.31,32,34,41–43 As these are furthermore considered healthy food items, which could have beneficial effects in relation to cancer survival, they might confound the association between PFOA/PFOS and survival after a breast and prostate cancer diagnosis in the present study. However, inclusion of information on fish and egg consumption in the models of the present study did not affect the association (results not shown).
In conclusion, the results of the present study suggest that a higher plasma concentration of PFOA and PFOS may be inversely associated with overall as well as bladder cancer–specific mortality after a bladder cancer diagnosis, among a population-based sample of middle-aged persons with moderate concentrations. In contrast, PFOA and PFOS concentrations were not associated with overall or prostate-specific mortality among prostate cancer patients. As the first study of its kind, these findings require replication before firm conclusions can be drawn.
Conflict of interest statement
The authors declare that they have no conflicts of interest with regard to the content of this report.
Sources of funding
The data material used for the present study was established as part of a previous study (PMID: 19351918; DOI: 10.1093/jnci/djp041), which was supported by a grant from the International Epidemiology Institute (no grant number), which received funding from the 3M Company. The present study was undertaken as part of our normal work at the Danish Cancer Society Research Center, thus supporting the study with our salary. Neither the 3M Company nor the International Epidemiology Institute had any role in design of the present study, interpretation of the results, or writing of the paper.
1. Lindstrom AB, Strynar MJ, Libelo EL. Polyfluorinated compounds: past, present, and future.Environ Sci Technol201145197954–7961
2. Lindstrom AB, Strynar MJ, Libelo EL, Field JA. Guest comment: perfluoroalkyl acid focus issue.Environ Sci Technol201145197951–7953
3. Kannan K, Corsolini S, Falandysz J, et al. Perfluorooctanesulfonate
and related fluorochemicals in human blood from several countries.Environ Sci Technol200438174489–4495
4. Arrieta-Cortes R, Farias P, Hoyo-Vadillo C, Kleiche-Dray M. Carcinogenic risk of emerging persistent organic pollutant perfluorooctane sulfonate (PFOS): a proposal of classification.Regul Toxicol Pharmacol20178366–80
5. Chang ET, Adami HO, Boffetta P, Cole P, Starr TB, Mandel JS. A critical review of perfluorooctanoate
exposure and cancer risk in humans.Crit Rev Toxicol201444suppl 11–81
6. Benbrahim-Tallaa L, Lauby-Secretan B, Loomis D, et al. Carcinogenicity of perfluorooctanoic acid, tetrafluoroethylene, dichloromethane, 1,2-dichloropropane, and 1,3-propane sultone.Lancet Oncol2014159924–925
7. Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012.Int J Cancer20151365E359–E386
8. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012.CA Cancer J Clin201565287–108
9. Gilliland FD, Mandel JS. Mortality among employees of a perfluorooctanoic acid production plant.J Occup Med1993359950–954
10. Lundin JI, Alexander BH, Olsen GW, Church TR. Ammonium perfluorooctanoate
production and occupational mortality.Epidemiology2009206921–928
11. Steenland K, Zhao L, Winquist A. A cohort incidence study of workers exposed to perfluorooctanoic acid (PFOA).Occup Environ Med2015725373–380
12. Olsen GW, Burlew MM, Marshall JC, Burris JM, Mandel JH. Analysis of episodes of care in a perfluorooctanesulfonyl fluoride production facility.J Occup Environ Med2004468837–846
13. Eriksen KT, Sorensen M, McLaughlin JK, et al. Perfluorooctanoate
plasma levels and risk of cancer in the general Danish population.J Natl Cancer Inst20091018605–609
14. Vieira VM, Hoffman K, Shin HM, Weinberg JM, Webster TF, Fletcher T. Perfluorooctanoic acid exposure and cancer outcomes in a contaminated community: a geographic analysis.Environ Health Persp20131213318–323
15. Raleigh KK, Alexander BH, Olsen GW, et al. Mortality and cancer incidence in ammonium perfluorooctanoate
production workers.Occup Environ Med2014717500–506
16. Alexander BH, Olsen GW, Burris JM, Mandel JH, Mandel JS. Mortality of employees of a perfluorooctanesulphonyl fluoride manufacturing facility.Occup Environ Med20036010722–729
17. Alexander BH, Olsen GW. Bladder cancer
in perfluorooctanesulfonyl fluoride manufacturing workers.Ann Epidemiol2007176471–478
18. Hardell E, Karrman A, van Bavel B, Bao J, Carlberg M, Hardell L. Case-control study on perfluorinated alkyl acids (PFAAs) and the risk of prostate cancer
19. Pierozan P, Karlsson O. PFOS induces proliferation, cell-cycle progression, and malignant phenotype in human breast epithelial cells.Arch Toxicol2018922705–716
20. Merritt RL, Foran CM. Influence of persistent contaminants and steroid hormones on glioblastoma cell growth.J Toxicol Environ Health A200770119–27
21. Tjonneland A, Olsen A, Boll K, et al. Study design, exposure variables, and socioeconomic determinants of participation in Diet, Cancer and Health: a population-based prospective cohort study of 57,053 men and women in Denmark.Scand J Public Health2007354432–441
22. Storm HH, Michelsen EV, Clemmensen IH, Pihl J. The Danish Cancer Registry—history, content, quality and use.Dan Med Bull1997445535–539
23. Ehresman DJ, Froehlich JW, Olsen GW, Chang SC, Butenhoff JL. Comparison of human whole blood, plasma, and serum matrices for the determination of perfluorooctanesulfonate
(PFOA), and other fluorochemicals.Environ Res20071032176–184
24. Hornung RW, Reed LD. Estimation of average concentration in the presence of nondetectable values.Appl Occup Environ Hyg19905146–51
25. Pedersen CB. The Danish Civil Registration System.Scand J Public Health2011397 suppl22–25
26. Helweg-Larsen K. The Danish Register of Causes of Death.Scand J Public Health2011397 suppl26–29
27. Greenland S. Dose-response and trend analysis in epidemiology: alternatives to categorical analysis.Epidemiology199564356–365
28. Bartell SM, Calafat AM, Lyu C, Kato K, Ryan PB, Steenland K. Rate of decline in serum PFOA concentrations after granular activated carbon filtration at two public water systems in Ohio and West Virginia.Environ Health Persp20101182222–228
29. Russell MH, Waterland RL, Wong F. Calculation of chemical elimination half-life from blood with an ongoing exposure source: the example of perfluorooctanoic acid (PFOA).Chemosphere2015129210–216
30. Li Y, Fletcher T, Mucs D, et al. Half-lives of PFOS, PFHxS and PFOA after end of exposure to contaminated drinking water.Occup Environ Med201875146–51
31. Johansson JH, Berger U, Vestergren R, et al. Temporal trends (1999–2010) of perfluoroalkyl acids in commonly consumed food items.Environ Pollut2014188102–108
32. Jian JM, Guo Y, Zeng L, et al. Global distribution of perfluorochemicals (PFCs) in potential human exposure source-A review.Environ Int201710851–62
33. Ducatman A, Zhang J, Fan H. Prostate-specific antigen and perfluoroalkyl acids in the C8 health study population.J Occup Environ Med2015571111–114
34. Eriksen KT, Sorensen M, McLaughlin JK, Tjonneland A, Overvad K, Raaschou-Nielsen O. Determinants of plasma PFOA and PFOS levels among 652 Danish men.Environ Sci Technol201145198137–8143
35. Leppert JT, Shvarts O, Kawaoka K, Lieberman R, Belldegrun AS, Pantuck AJ. Prevention of bladder cancer
: a review.Eur Urol2006492226–234
36. Rink M, Crivelli JJ, Shariat SF, Chun FK, Messing EM, Soloway MS. Smoking and Bladder Cancer
: A Systematic Review of Risk and Outcomes.Eur Urol FocusAug 20151117–27
37. De Nunzio C, Andriole GL, Thompson IM Jr., Freedland SJ. Smoking and prostate cancer
: a systematic review.Eur Urol Focus20151128–38
38. Harada K, Inoue K, Morikawa A, Yoshinaga T, Saito N, Koizumi A. Renal clearance of perfluorooctane sulfonate and perfluorooctanoate
in humans and their species-specific excretion.Environ Res2005992253–261
39. Lorber M, Eaglesham GE, Hobson P, Toms LM, Mueller JF, Thompson JS. The effect of ongoing blood loss on human serum concentrations of perfluorinated acids.Chemosphere2015118170–177
40. Karlsen RV, Bidstrup PE, Christensen J, et al. Men with cancer change their health behaviour: a prospective study from the Danish diet, cancer and health study.Br J Cancer20121071201–206
41. Hlouskova V, Hradkova P, Poustka J, et al. Occurrence of perfluoroalkyl substances (PFASs) in various food items of animal origin collected in four European countries.Food Addit Contam Part A Chem Anal Control Expo Risk Assess201330111918–1932
42. Guerranti C, Perra G, Corsolini S, Focardi SE. Pilot study on levels of perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) in selected foodstuffs and human milk from Italy.Food Chem20131401–2197–203
43. Papadopoulou E, Poothong S, Koekkoek J, et al. Estimating human exposure to perfluoroalkyl acids via solid food and drinks: Implementation and comparison of different dietary assessment methods.Environ Res2017158269–276