|Home | About | Journals | Submit | Contact Us | Français|
Although organochlorines have been reported more frequently in Parkinson’s disease (PD) brains than controls, the association with brain Lewy pathology is unknown. Honolulu-Asia Aging Study (HAAS) participants, exposed to organochlorines from a variety of sources during mid-life, represent a population well suited to determine the relationship of brain organochlorines with Lewy pathology in decedents from the longitudinal HAAS.
Study design included the measurement of 21 organochlorine levels in frozen occipital lobe samples from HAAS decedents. Alpha-synuclein immunostaining performed on 225 brains was used to identify Lewy bodies and Lewy neurites.
With the potential for spurious associations to appear between Lewy pathology and 17 organochlorine compounds found to be present in at least one brain, initial assessments identified heptachlor epoxide isomer b, methoxychlor, and benzene hexachloride b as being most important. Prevalence of Lewy pathology was 75% (6/8) among brains with any 2 of the 3 compounds, 48.8% (79/162) among those with 1, and 32.7% (18/55) for those with neither (P=0.007 test for trend). While findings persisted after removing cases with PD and dementia with Lewy bodies, and when adjustments were made for age at death, body mass index, pack-years of cigarette smoking, and coffee intake (p=0.013), results were insignificant when correcting for multiple testing.
While consistent with earlier accounts of an association between organochlorines and clinical PD, associations with Lewy pathology warrant further study.
Evidence supporting a role for organochlorine pesticides in the etiology of Parkinson’s disease (PD) comes from a variety of sources. Epidemiological studies demonstrate an association of organochlorine exposure with PD including a dose effect relationship1,2 and elevated serum levels of some organochlorines are associated with higher odds of developing PD.3,4 Pathological studies have also shown an association between brain organochlorines and PD.5,6 The organochlorine dieldrin induces proteasome dysfunction in rat mesencephalic dopaminergic cells over-expressing human alpha-synuclein promoting apoptotic cell death.7 It also leads to increased oxidative stress, increased expression of striatal alpha-synuclein, and impaired dopamine metabolism in exposed mice.8 Lastly, exposure to dieldrin and lindane combined, causes more dopaminergic neuron toxicity through induction of oxidative stress and mitochondrial dysfunction than exposure to only one agent.9
Lewy pathology in human brains is considerably more prevalent than clinical PD. Demonstrating an association of brain organochlorine levels with brain Lewy pathology would strongly support a role for organochlorine exposure in the etiology of PD. The Honolulu-Asia Aging Study (HAAS) cohort is an ideal population to examine this association for two reasons. First, HAAS participants were exposed to organochlorine insecticides in a number of ways during the period between the early 1940s, when these compounds were introduced commercially in Hawaii, until 1988, when the ban on sales of organochlorines was instituted in the United States. Second, the HAAS has an autopsy component with a large brain archive. The aim of this study is to determine if there is an association of brain organochlorine levels with presence of Lewy pathology in brains of deceased participants in the Honolulu-Asia Aging Study (HAAS) and to determine if presence of multiple organochlorines confers greater odds of having Lewy pathology.
The Honolulu Heart Program (HHP) is a longitudinal study of heart disease and stroke in a cohort of 8006 Japanese-American men born 1900–1919 and living on the island of Oahu, Hawaii in 1965 when the study began.10 Follow-up has continued through re-examinations and surveillance of hospital and death records. With establishment of the HAAS at the 1991 examination, research on diseases of aging including dementia and PD was initiated.11 Previously described standardized methods were used to identify all cases of PD in the cohort both at the 1991 examination and in follow-up examinations every 2 to 3 years.12,13 The study was approved by the Kuakini Medical Center Institutional Review Board and participants signed informed consents at all examinations.
The autopsy component was initiated in 1991.14 Autopsy was discussed with all HAAS participants and consent for autopsy was given by the closest living family member according to Hawaii state law.
Standardized gross and microscopic examinations of multiple brain regions were performed. Details have been published.15,16 This included immunohistochemical staining for alpha-synuclein 17 on sections of olfactory bulb, medulla, pons, midbrain, hippocampus, amygdala, and striatum at the level of the nucleus accumbens, basal forebrain, anterior cingulate gyrus, insula, mid frontal, mid temporal, inferior parietal, and occipital regions. Semi-quantitative pathology density analyses were carried out for each region examined, and staging was performed with some modifications to the methods of Braak and colleagues.18 For this analysis, presence of Lewy bodies or neurites was considered positive presence of Lewy pathology.
Frozen occipital lobe samples were analyzed at Research Triangle Institute for 21 organochlorines.19,20 Approximately 0.025–0.030 grams of tissue was weighed, transferred to a culture tube, and dried by grinding with one gram of anhydrous sodium sulfate, then extracted 3 times with 5 ml of hexane. Extracts were combined and extract volume was adjusted to 15 mL. A known aliquot of the extract was then removed for lipid analysis, while the remainder was concentrated to 1 mL. The extract was cleaned while being eluted with 3 solvent systems on a column of activated Florisil (partially deactivated before use with water). The first eluate (25 mL hexane) contained DDE and other non-polar pesticides. The second eluate (25 mL 10% ether in hexane) contained the more polar pesticides, including dieldrin and endrin. The third eluate (25 mL dichloromethane) contained pesticides such as endosulfan II. Fractions were concentrated individually to 1.0 mL each. A quantitation standard (PCB congener 119) was added, and the extracts analyzed by high resolution gas chromatography using an electron capture detector. Concentration was reported in parts per billion for 21 organochlorines. Organochlorine levels were considered to be not present for this analysis if they were reported to be “not detected” or “less than the level of calibration”.
Two-sample t-tests were used to compare the average characteristics between autopsied and non-autopsied decedents and between decedents with and without Lewy pathology (tables 1 and and2).2). Associations between an organochlorine compound and Lewy pathology were initially assessed using Fisher’s exact test with Lewy pathology categorized as absent (Braak stage = 0) versus present (Braak stage ≥ 1) (table 3). The actual Braak staging was not used for the latter assessments because the level of Braak staging was too limited for 10 of the 17 compounds that were detected in fewer than 10 of the 225 decedents. Subsequent stepwise analyses that considered associations with more than 1 compound, however, permitted the use of the actual Braak stage as a measure of Lewy progression (e.g., see figure). For these analyses, the effect of combinations of organochlorines that appeared to have a relationship with Lewy pathology was examined (e.g., exposure to none, 1, 2 or 3 compounds). The most important compound was first identified based on the initial results from the Fisher’s exact test. Additional compounds were considered through a 2-step process that began with the second compound with the strongest association with Lewy pathology. In such instances, ordinal logistic regression was used to model Braak staging as a function of the number of compounds detected (0, 1, or 2). 21 After adjustment for this compound, a third compound was selected based on its having the strongest association with Lewy pathology among the remaining compounds. The process was stopped after a third compound was identified because the number of decedents exposed to more than 3 compounds was small. To help confirm the association between the selected compounds and Lewy pathology, adjustments were made for confounding effects of age at death, body mass index, mid-life pack-years of cigarette smoking, and coffee intake. Further testing was based on removing cases of PD and dementia with Lewy bodies (DLB). Separate adjustment for each or the remaining compounds was made to help determine if the association between the 3 selected compounds and Lewy pathology was independent of the other compounds. Findings were further examined through the use of permutation tests that account for multiple testing, correlation, and non-normal distributions.22 For descriptive purposes (see figure), analysis of covariance models provided estimates of the prevalence of Lewy pathology (Braak stage ≥ 1) by the number of detectable compounds after age and risk factor adjustment.23 Body mass index was included among the possible confounders as fat is a potential depot for organochlorine compounds. All reported p-values were based on 2-sided tests of significance.
There were 225 brains with available organochlorine levels and Braak Lewy body staging. Of these, 122 were without Lewy pathology, 29 had PD or DLB, and 74 had incidental Lewy bodies (Braak stage >0 and no clinical history or PD or DLB). Mean age at death was 86.2 years (range = 75–99 years).
Table 1 compares characteristics of the autopsied sample for which organochlorine levels were available and deceased participants without an autopsy. There were no significant differences in mean age at death, pack-years of smoking, coffee consumed, or body mass index.
Characteristics of individuals with and without brain Lewy pathology were similar (Table 2). While those with Lewy pathology were slightly older, smoked less, and drank less coffee, differences were not statistically significant.
Among the 21 organochlorine compounds assessed, there were 3 that were not detected in any brains. These were benzene hexachloride a, endosulfan 1, and endosulfan 2. Data on g-chlordane, were inadequate for analysis due to technical problems resulting in large numbers of missing values. Among the remaining 17, clear positive associations with presence of Lewy pathology were noted for 9. There were negative associations for 4 and very little relationship with the other 4 (Table 3). The only statistically significant association was with benzene hexachloride b (p=0.043). Although only 4 brains had detectable levels, all had Lewy pathology while only 45% of those without detectable levels (99/221) had Lewy pathology.
For the remaining compounds positively associated with Lewy pathology, prevalence of heptachlor epoxide isomer b seemed important. Nearly half of those exposed had Lewy pathology versus 36.2% in those without detectable levels (p=0.095). While detection of methoxychlor was far less common (N=7), in its presence, 71.4% had Lewy pathology versus 45% in its absence (p=0.251).
Effects of DDT, hexachlorobenzene, and mirex may also be important, but detection of these compounds was noted in only a single decedent. Compounds with an inverse association were, benzene hexachloride g, endrin, isobenzan, and aldrin. Individually or in combination, these compounds did not have significant relationships with Lewy pathology.
In the stepwise process to select compounds with the strongest association with Lewy pathology, benzene hexachloride b was selected first because of its association in table 3 (p=0.043). The second was heptachlor epoxide isomer b. Here, prevalence of Lewy pathology was 33.9% (19/56) in the absence of both compounds, 49.1% (82/167) when one compound was present, and 100% (2/2) in the presence of both compounds (p=0.019). The third selected was methoxychlor, although improvement in predicting the prevalence of Lewy pathology is modest. None of the decedents had detectable levels of all 3. Figure A shows that the percent of Lewy pathology increased as the number of compounds detected increased (p=0.007). Findings remained significant after adjustment for multiple potential confounders including age at death, body mass index, and mid-life cigarette smoking and coffee intake (P=0.007), and, while also removing cases of PD and DLB (P=0.013) The independent association between the 3 selected compounds and Lewy pathology also persisted (p<0.05) after separate adjustment for the remaining unselected compounds. In each instance, the percent of decedents with Lewy pathology was more than doubled when there was exposure to any 2 of the organochlorines versus no exposure. Findings were similar when examining associations with the 29 cases of clinical PD or DLB, although results were not significant due to the limited sample size (p=0.08). Findings failed to persist after a permutation test leaving the possibility that positive results could be due to chance related to multiple comparisons.
Most of the findings seem to be due to presence of heptachlor epoxide isomer b (possibly due to its high frequency). For it to have a significant relationship with Lewy pathology, however, it needed to be in the presence of at least one of the other organochlorines (p=0.018 for benzene hexachloride b and p=0.018 for methoxychlor after risk factor adjustment). Overlap among these three compounds was not a large factor given that only 8 of 170 decedents with detectable levels of any one compound also had detectable levels of a second compound (Figure). Combinations involving the presence of other compounds were less informative, largely due to weaker associations, their rare detection, or because of associations that seemed largely determined by associations with the 3 compounds in the figure.
Analyses were also performed to examine the association of brain levels of the organochlorines with Lewy pathology. No clear dose effect was found, possibly related to the small number of brains with detectable levels for some compounds.
The original study design was to examine 21 organochlorine compounds and their association with Lewy pathology. In the absence of prior evidence favoring one compound over another, each compound was considered with equal weight, requiring multiple testing with a high risk for uncovering spurious associations. Findings highlight difficulties in designing and analyzing studies based on the selection of several risk factors perceived to be important in the assessment of a single hypothesis and the need to be as focused as possible.
Nevertheless, findings based on the multiple stepwise approach in a search for compounds related to Lewy pathology followed by adjustment for concomitant information and confirmation in a subset of decedents without PD or DLB are at least consistent with other reports of an association with clinical PD.
Most organochlorines belong to 1of 3 main classes, cyclodienes, including heptachlor and dieldrin; dichlorodiphenylethanes, including DDT and methoxychlor; and cylohexanes including benzene hexachloride b and g. These classes differ in toxic effects. No particular class dominated the association with Lewy pathology in the HAAS brains. Experimental evidence is strongest for the cyclodiene group that has been linked to a-synuclein aggregation, enhanced oxidative stress, and disruption of the ubiquitin-proteasomal system24 Among the cyclodienes in this analysis, presence of aldrin was inversely associated with Lewy pathology while others, such as heptachlor epoxide isomer B, had positive associations.
Organochlorines were widely and heavily used in Hawaii on pineapple plants through the 1960s and applied around homes and other buildings as termiticides. Hawaii is particularly sensitive to groundwater contamination from agricultural pesticides owing to the high permeability of the volcanic soil.25 In 1982 the commercial milk supply on the island of Oahu was found to be contaminated with heptachlor and the source for this contamination was chopped pineapple leaves fed to dairy cattle.26,27 Interestingly, a previous HAAS publication found that those consuming greater than 16 oz milk per day were significantly more likely to develop PD than nondrinkers (relative risk = 2.6, 96% confidence interval 1.1,6.4).13 Organochlorine use has been restricted in the United States since the mid-1970s and banned since 1988. The fact that organochlorine residues are detectable in the brains of deceased HAAS participants over 30 years following exposure may be related to bioaccumulation of these fat soluble chemicals with very long half lives.25 Alternatively, brain levels in HAAS decedents could reflect more recent exposure to persistent organochlorines in Oahu fish, water, and soil.
A recent US Geological Survey report on fish from Oahu streams found that in 1998 tissue concentrations of several organochlorine compounds exceeded guidelines.28 Although concentrations have decreased since the 1970s, levels remain high for some fish.
Earlier studies with small numbers of cases found that PD brains were more likely to have detectable levels of dieldrin6 and higher mean concentrations of lindane and dieldrin5 than Alzheimer’s disease (AD) or normal brains. In a recent small case control study of serum organochlorines and PD, beta hexachlorocyclohexane was more often detected in PD cases versus normal and AD controls.3 In a larger follow-up study, odds of PD were significantly higher among subjects with beta hexachlorocyclohexane levels above the detectable inter-quartile range.29 In another case control study, using prospectively collected serum, a higher mean concentration of only dieldrin out of five analyzed organochlorine pesticides was associated with higher odds of PD. 4
Epidemiological studies have reported associations between self reported pesticide exposures and clinical PD.1,2,30–33 A recent case control study reported that professional use of organochlorines was associated with over twice the odds of PD compared to those not exposed.1
Exposure to combinations of two pesticides has been associated with significantly more dopaminergic neuron injury than exposure to only one in animal models examining lindane and dieldrin9 and maneb and paraquat.34 In humans, exposure to both maneb and paraquat increased risk of PD by 75% in a case control study.31
While the means by which organochlorines may result in Lewy pathology are unknown, current lines of evidence suggest several potential mechanisms. Dieldrin is selectively neurotoxic to dopaminergic neurons in primary mesencephalic cultures35 and accelerates the rate of alpha-synuclein fibril formation in vitro.36 Exposure to dieldrin in mice is associated with a decrease in dopamine metabolites, an increase in markers of oxidative stress in the striatum and an increase in alpha-synuclein protein expression in the striatum.8 Alternatively, Lewy pathology may result from neuronal damage caused by organochlorines. For example, heptachlor alters dopamine transport in mice and it is speculated that this may increase susceptibility to other dopaminergic toxins.37
There are limitations to this study. The HAAS cohort is all male and of Japanese ancestry. Generalizations of findings to women and other ethnic groups cannot be made with certainty. However, studies including women and persons of European ethnicity have also reported associations between organochlorines and PD. Because organochlorine measurements were performed well after the initiation of processes leading to Lewy pathology, it is possible that these processes affect the organochlorine levels. For example, Lewy pathology may impede organochlorine clearance from the brain. However, to our knowledge, this has not been reported. Presence of other pesticides that may also be associated with Lewy pathology such as paraquat was not assessed. Detectable levels of most of the organochlorines in this report were often exceedingly rare, imposing an important limitation on our findings. Considering that exposures to organochlorines may have occurred over 30 years ago, it is noteworthy that compounds were detected frequently enough to demonstrate a relationship.
There are several strengths. Sampled brains were from a population based longitudinal study with well characterized information on potential confounders. Organochlorine measurement in frozen brain tissue provided an objective measure of presence of organochlorines. All the participants lived on the island of Oahu during the period that organochlorines were in heavy use. Finally, state of the art immunohistochemical staining for alpha-synuclein was used.
In summary, prevalence of Lewy pathology in brains from HAAS participants was higher in those brains with detectable levels of most organochlorines assessed although only one relationship was statistically significant. In particular, having detectable levels of any 2 of heptachlor epoxide b, methoxychlor, or b-BHC was associated with a high prevalence of Lewy pathology after adjustment for age at death, body mass index, cigarette smoking and coffee intake, and in a subset of decedents without PD or DLB. We cannot rule out the possibility that these results are due to chance based on multiple comparison analyses. While these findings support a role for organochlorines in the etiology of PD, further study is needed. Evidence might also provide direction in designing future studies with a more careful focus on selected compounds that seem most important in the current report.
This work was supported by United States Department of the Army, Grant No. DAMD17-98-1-8621, National Institutes of Health: National Institute of Neurological Disorders and Stroke Grant No. 5 R01 NS041265; National Institute on Aging Grant Nos. 1 U01 AG19349 and 5 R01 AG017155; and the Office of Research and Development, Medical Research Service Department of Veterans Affairs. This research was also supported in part by the Intramural Research Program of the NIH, National Institute on Aging The information contained in this paper does not necessarily reflect the position or the policy of the government, and no official endorsement should be inferred.
Financial Disclosure/conflict of Interest related to this manuscript:
Dr. Ross, received research support from the Department of Defense, NIA, NIH; NINDS, NIH; and the Department of Veterans Affairs.
Dr. Duda and Mr. Noorigian received research support from NINDS, NIH
Dr. Abbott received research support from the Department of Defense, NIA, NIH; and NINDS, NIH.
Dr. Pellizzari received research support from the Department of Defense.
Dr. Petrovitch received research support from the Department of Defense, NIA, NIH; and NINDS, NIH.
Drs. Miller and O’Callaghan received research support from NIH.
Dr. Tanner received research support from the Department of Defense, NIA, NIH; and NINDS, NIH.
Dr. Masaki received research support from the NIA, NIH.
Dr. Launer has nothing to disclose.
Dr. White received research support from the Department of Defense, NIA, NIH; and NINDS, NIH.
Author’s rolesGW Ross: conception and design, acquisition of data, interpretation of data, drafting of manuscript, obtaining funding
JE Duda: acquisition of data, interpretation of data, critical revision of manuscript
RD Abbott: analysis and interpretation of data, critical revision of manuscript, obtaining funding, statistical expertise
E Pellizzari: acquisition of data, critical revision of manuscript
H Petrovitch: conception and design, interpretation of data, critical revision of manuscript, obtaining funding
DB Miller: acquisition of data, interpretation of data, critical revision of manuscript, obtaining funding
JP O’Callaghan: acquisition of data, interpretation of data, critical revision of manuscript, obtaining funding
CM Tanner: acquisition of data, interpretation of data, critical revision of manuscript, obtaining funding
JV Noorigian: acquisition of data, critical revision of manuscript
K Masaki: acquisition of data, critical revision of manuscript, obtaining funding
L Launer: interpretation of data, critical revision of manuscript
LR White: conception and design, interpretation of data, critical revision of manuscript, obtaining funding