Our results show that serum PCB concentration is more strongly associated with elevation of blood pressure than with any other factor except age, and that this relationship applies both to those individuals who are not taking antihypertensive medication and to the subset of persons whose blood pressure is within the normotensive range. We have previously reported that serum PCB levels are significant risk factors for hypertension using logistic regression analysis (Goncharov et al. 2010
), and the present results show that the association is found over any range of blood pressure. The association appears to be specific to PCBs, because we found little or no relationship between blood pressure and levels of organochlorine pesticides.
The relationship with blood pressure was strongest with ortho
-substituted PCB congeners with two or more chlorines. Although we found no significant relationship for the sum of all mono-ortho
congeners, we found weak but significant associations for three mono-ortho
congeners, all of which have TEF values. Unfortunately, we did not have measurements of the most potent non-ortho
dioxin-like PCB congeners. We found no evidence for a relationship between estrogenic congeners as identified by DeCastro et al. (2006)
and blood pressure.
Anniston residents have relatively high levels of serum PCBs. It is difficult to compare levels of total PCBs across different studies because investigators analyze for different congeners, and some report wet-weight and others lipid-adjusted values. Needham et al. (2005)
and the 2003–2004 National Health and Nutrition Examination Survey (NHANES) report (Centers for Disease Control and Prevention 2009
) provide levels of some single congeners, and, of the congeners common to our study, levels of PCB congeners 99, 118, 153, 156, 170, 180, and 183 in our subjects are greater than those of the 90th percentile of levels in the general U.S. population. Only the level of PCB-74 was lower in this sample of Anniston residents than that reported for the 90th percentile of the general U.S. population. This is an interesting congener that has often been found to reflect fish consumption (Fitzgerald et al. 2004
There are limitations to our study. By its cross-sectional nature, this study cannot prove causality no matter how strong the associations. There also are some limitations in information that should be considered, such as diet and better measures of levels of exercise and smoking.
There has been some evidence for an association of elevated serum PCBs with blood pressure since the study by Kreiss et al. (1981)
of residents of Triana, Alabama, who were exposed by consuming fish downstream from an industrial facility. In a study of PCB exposure from living in the vicinity of waste sites, Stehr-Green et al. (1986)
found a dose-dependent but not statistically significant relationship between serum PCB levels and high blood pressure. Huang et al. (2006)
found elevated rates of hospitalization for hypertension among residents of New York who lived near hazardous waste sites containing persistent organic pollutants, primarily PCBs.
Two recent investigations have used the 1999–2002 NHANES to determine the relationship between exposure to various persistent organic pollutants and diagnosed hypertension. Ha et al. (2009)
reported that the sum of three dioxins and three furans showed significant elevations in rates of hypertension in women, whereas serum PCBs tended toward positive associations in men. Everett et al. (2008a)
investigated associations between hypertension and individual PCB congeners and found the strongest associations with dioxin-like PCBs (PCBs 126 and 118) but also found significant relationships with PCB congeners 74, 99, 138+158, 170, and 187. Everett et al. (2008b)
expanded their study to include NHANES data through 2004 and found significant associations with PCB congeners 74, 118, and 126. In the expanded data set of Everett et al. (2008b)
, the associations for PCB congeners 99, 138+158, 153, 156, 169, 170 180, and 187 were elevated but not statistically significant. In a study focused on the metabolic syndrome in the Japanese population, Uemura et al. (2009)
reported significant associations between hypertension and PCB congeners 105, 114, 118, 123, 126, and 167 but not with PCB congeners 156, 157, 169, or 189. They also found significant associations with levels of four dioxins and two furans.
Our results are in general agreement with those of these previous studies, even though in this study we are investigating blood pressure rather than hypertension and not all studies measured the same PCB congeners. Despite not monitoring the most potent dioxin-like congeners, we did find significant associations with levels of PCBs 156, 157, and 189, all of which have some dioxin-like activity. Although we found significant associations between blood pressure and some 20 congeners, we observed a clear pattern of stronger associations with those congeners with multiple ortho chlorines. Of these, PCB congeners 170, 172, 178, 180, 183, 187, 196+203, and 199 showed statistically significant associations in both the second and third tertile with both systolic and diastolic blood pressure. Our results differ somewhat from the conclusions above in that we found a dominance of associations with multi-ortho-substituted congeners, not dioxin-like mono-ortho congeners. It is also striking that the association between serum PCB levels and blood pressure appears to be a low-dose effect, in that we found little difference between the parameter estimates when comparing the first and second tertiles versus comparing the first and third tertiles in all those not on antihypertensive medication (, ). This is less apparent in normotensive individuals. If these associations are indeed present at relatively common serum PCB concentrations, the public health significance of the relationships may be considerable. Clearly, it is very important that these possibilities be tested systematically in a prospective investigation. There also is a need for investigations that can identify the mechanisms that might underlie these associations.
We found little if any association between blood pressure and levels of chlorinated pesticides. Siddiqui et al. (2002)
reported an association between maternal blood pressure and levels of γ-HCCH and p
′-DDT in breast milk. However, overall our results do not support a major role for chlorinated pesticides in regulation of blood pressure.
The finding that the association between serum PCB levels and blood pressure is present in individuals with normal blood pressure is important and raises the important question of what is “normal.” This observation implies that even background levels of PCBs are altering physiological processes and suggests that there is no threshold for action. Our previous studies in a Native American population showed relationships between serum PCB levels and “normal” levels of thyroid hormone (Schell et al. 2008
) and testosterone (Goncharov et al. 2009
), and also performance on cognitive tests in a “normal” range in adolescents (Newman et al. 2009
) and adults (Haase et al. 2009
). For these physiological effects, as with blood pressure, a number of external factors can influence levels or performance, but the observation that serum PCB levels can influence them indicates that PCB levels may cause physiological changes and contribute to the development of diseases even when they are not the sole cause.