Toms River (83 page)

11.
As an example of how these adjustments worked, a boy who lived in a home that got an average of 30 percent of its water from the Parkway wells during his mother’s pregnancy and during his life before a cancer diagnosis would normally be classified as having a “medium” exposure to Parkway water, but if the boy’s mother reported that during her pregnancy she had drunk an unusually high number of glasses of water each day, the boy would be bumped up into the “high” category.

12.
The total number of children—702—was fewer than 727 because of overlap between the 528 children in the birth record study and the 199 children in the interview study.

13.
Frederica P. Perera and I. Bernard Weinstein, “Molecular Epidemiology and Carcinogen-DNA Adduct Detection: New Approaches to Studies of Human Cancer Causation,”
Journal of Chronic Diseases
35 (1982): 581–600.

14.
Frederica Perera and her collaborators found that mice and rabbits injected with benzo(a)pyrene had more DNA adducts than control animals did. But dogs exposed to cigarette smoke for twenty months did not. In humans, they found high adduct concentrations in the lung tissue of five lung cancer patients, all heavy smokers. But twenty-two other smokers did not have more adducts, and there was no dose-response pattern: Heavier smokers did not have more adducts. Frederica P. Perera et al., “A Pilot Project in Molecular Cancer Epidemiology: Determination of Benzo(a)pyrene-DNA Adducts in Animal and Human Tissues by Immunoassays,”
Carcinogenesis
3:12 (1982): 1405–10.

15.
Mutated TP53 genes are in about half of all cancers, including neuroblastomas, melanomas, and many other tumors. Timothy E. Baroni et al., “A Global Suppressor
Motif for P53 Cancer Mutants,”
Proceedings of the National Academy of Sciences
101:14 (April 6, 2004): 4930–35.

16.
Richard Albertini and R. B. Hayes, “Somatic Cell Mutations in Cancer Epidemiology,”
IARC Scientific Publications 142
(International Agency for Research on Cancer, 1997), 159–84.

17.
Barry Finette’s mentor at the University of Vermont, Richard Albertini, had pioneered the use of the HPRT gene in cancer research in the late 1980s. One of the gene’s many advantages as a biomarker is that it is located on the X chromosome. Since there is normally only one active X chromosome in a human cell, just one mutation event is enough to guarantee a heritable mutation, which makes HPRT a more sensitive biomarker for genetic damage than genes on paired chromosomes. Albertini also identified mutational mechanisms in HPRT that seemed to apply to many other genes, supporting the idea that HPRT really was a good surrogate for genome-wide damage.

18.
Barry Finette’s mutant frequency study was published as Pamela M. Vacek et al., “Somatic Mutant Frequency at the HPRT Locus in Children Associated with a Pediatric Cancer Cluster Linked to Exposure to Two Superfund Sites,”
Environmental and Molecular Mutagenesis
34 (2005): 339–45.

1.
To understand how Jerry Fagliano calculated adjusted odds ratios in the Toms River case-control studies, consider this example: Nine case children and twenty-three healthy controls in the interview study were in the “high” exposed category because their mothers had gotten more than 50 percent of their water during pregnancy from the Parkway well field. Twenty-six cases and 110 controls were in the “low” exposure category because their mothers had gotten less than 10 percent of their water from the Parkway wells during pregnancy. Calculating a “crude” odds ratio is simply a matter of comparing the odds that high-exposed women would have a child with cancer, versus low-exposed women. The simple equation looks like this: (exposed cases/unexposed cases)/(exposed controls/unexposed controls) = odds ratio. Applied to the Parkway data for high-exposure mothers, the math is (9/26)/(23/110), yielding an odds ratio of 1.66. Jerry Fagliano then adjusted that result in a way that helped to account for the influence of possible confounding factors such as the age and sex of each study child. After this final calculation, known as conditional logistical regression, the adjusted odds ratio is slightly higher: 1.68. In other words, the odds of a case child having been highly exposed to Parkway water were 68 percent greater than the odds that a healthy child had been exposed.

2.
All of the odds ratios described in this chapter come from data tables in
Case-Control Study of Childhood Cancers in Dover Township (Ocean County), New Jersey, Volume III: Technical Report Tables and Figures
.

3.
New Jersey Department of Health and Senior Services,
Case-Control Study of Childhood Cancers in Dover Township (Ocean County), New Jersey, Volume II: Final Technical Report
(January 2003), 23. Like many of the others in the study, this odds ratio fell short of statistical significance because the lower bound of its very wide confidence interval—0.79 to 274—was below 1.0.

4.
The same small-numbers problem stymied the health department’s attempts to assess the consequences of the groundwater pollution that tainted hundreds of backyard wells in the neighborhoods like Oak Ridge Estates and Pleasant Plains in the 1960s and early 1970s, before water mains were fully extended to those areas. Eleven such neighborhoods had been identified by investigators, but so much time had passed that they could track down just two exposed cases from that era—too few to generate a credible odds ratio.

5.
In order to create high-, medium-, and low-exposure categories for the air pollution model, its developers at Rutgers first analyzed air exposures for the residences of the control families. The family at the midpoint was assigned a 1.0 rating on a scale of “relative air impact units.” Under the same scale, the top 25 percent of controls in the interview study were at or above a rating of 2.18. They then applied the same scale to the case families, placing cases and controls into three categories: low (below 1.0), medium (1–2.17), and high (2.18 and above). For the birth record study, the “cut point” separating low from medium was 1.5; for separating medium from high exposure, the cut point was 3.04.

6.
The size of the Toms River legal settlement has never been disclosed but, based on court filings, it is very likely that the total was somewhere between $35 and $40 million, before deductions for the lawyers’ 25 percent fee and other reimbursable expenses. Records from judicial hearings, held only for the twenty-nine families with children still under age eighteen, show that those families were awarded approximately $11.4 million in gross payments. The payments went to children and siblings and also to parents for reimbursable costs (such as lost wages, uninsured medical bills, and trips to distant hospitals). After deductions for legal fees and expenses, the net payout was about $8.4 million, or an average of about $290,000 per family. Assuming that those twenty-nine families were roughly representative of the forty families not subject to the judicial hearings, then the total gross payout was approximately $27.1 million. In addition, there was a “recurrence fund” of $5 million ($3.75 million after legal fees) to compensate any surviving children who relapsed before 2016, at which time the proceeds would be distributed to all the surviving children. There were also additional payments to parents beyond expense reimbursement. Thus, the gross amount of the settlement in 2001 (not including subsequent interest on funds held in trust for minors) was probably somewhere between $35 million and $40 million. The court records also show that some families with children under eighteen received as much as $501,570, after deductions for fees and expenses, while others got as little as $91,954. The lawyers donated $150,000 from their fees to TEACH, the environmental advocacy group formed by the families.

7.
Dow, Ciba, and United Water have never revealed what percentage of the overall settlement each company paid, but Dow (as Union Carbide’s corporate successor) likely paid the largest share because the Parkway well contamination was so well documented.

8.
Like the Toms River settlement, the Woburn settlement was supposed to be secret. Unlike in Toms River, however, its $8 million size was immediately leaked to the press. Each of the eight Woburn families ultimately received about $435,000, while the payouts to the sixty-nine Toms River families varied widely but probably averaged about $290,000 per family. In Toms River, families of children with leukemia
received some of the largest settlements—in some cases, larger than what the Woburn families received—because the scientific evidence of environmental causation was stronger for leukemia than for other cancers. The Woburn lawsuit included only leukemia cases.

9.
New Jersey Department of Health and Senior Services,
Case-Control Study of Childhood Cancers in Dover Township, Volume I
(January 2003), 18.

10.
For the four nervous system cancer cases diagnosed before age five, the 9.00 odds ratio for high exposure to Ciba air emissions came close to achieving statistical significance because the lower bound of the 95 percent confidence interval was 0.86, just under 1.0. The upper bound was an extremely high 94.2.

11.
For birth record study children diagnosed with cancer before age five, the odds ratio for high prenatal exposure to Ciba air emissions was 2.95, with a 95 percent confidence interval of 0.92 to 9.45.

12.
Tom Feeney and Mark Mueller, “Crusading Mom Shrugs Off Vindication,”
Star-Ledger
, December 19, 2001.

1.
Carol S. Rubin et al., “Investigating Childhood Leukemia in Churchill County, Nevada,”
Environmental Health Perspectives
115:1 (January 2007): 151–57.

2.
Craig Steinmaus et al., “Probability Estimates for the Unique Childhood Leukemia Cluster in Fallon, Nevada, and Risks near Other U.S. Military Aviation Facilities,”
Environmental Health Perspectives
112:6 (May 2004): 766–71. The estimate that a childhood cancer cluster as large as Fallon’s would occur randomly only once every 22,000 years in a country the size of the United States (which had about 77 million children in 2003) is based on a case count of eleven childhood leukemias diagnosed in Fallon and the surrounding area from 1999 to 2001. The authors’ estimate that there was only a 1-in-232 million chance that the Fallon cluster was random is a sharp contrast to the Woburn childhood leukemia cluster, which the authors calculate had a 1-in-120 probability of occurring by chance. The authors did not attempt to estimate the odds that the Toms River cluster was random.

3.
U.S. Centers for Disease Control and Prevention,
Biosampling Case Children with Leukemia (Acute Lymphocytic and Myelocytic Leukemia) and a Reference Population in Sierra Vista, Arizona: Final Report
(November 30, 2006).

4.
Florida Department of Health,
Acreage Cancer Review, Palm Beach County
(August 28, 2009), subsequently revised as
Acreage SIR Recalculation Population Estimate Methods and Results
. Both are available on the Palm Beach County Health Department website. Because The Acreage grew so fast during the years at issue, the state calculated incidence rates using four different methods of estimating the local population. For the four cases of pediatric brain cancer diagnosed from 2005 to 2007, the 95 percent confidence intervals for the four methods were (1.7–11), (1.5–9.8), (1.3–8.2), and (1.3–8.2). Thus, by any of those methods, the cluster was statistically significant. Since three of the four cases were in girls, the incidence ratios and confidence intervals for girls were higher still.

5.
Stephen P. Hunger et al., “Improved Survival for Children and Adolescents with Acute Lymphoblastic Leukemia Between 1990 and 2005: A Report from the
Children’s Oncology Group,”
Journal of Clinical Oncology
, published online March 12, 2012. Gleevec, the Novartis trade name for the drug imatinib, is especially effective in treating children with ALL who carry the Philadelphia chromosomal translocation, the article notes.

6.
The biggest dosimeter success story so far is for aflatoxin, a naturally occurring fungal secretion that is carcinogenic and often contaminates peanuts and other grains, especially in Africa and China. In the early 1990s, John Groopman of The Johns Hopkins University confirmed that aflatoxin-DNA adducts can be readily detected in urine and are a superb measure for identifying populations at high risk for liver cancer. See John D. Groopman et al., “Molecular Epidemiology of Aflatoxin Exposures: Validation of Aflatoxin-N7-Guanine Levels in Urine as a Biomarker in Experimental Rat Models and Humans,”
Environmental Health Perspectives
99:3 (March 1993): 107–13.

7.
One clue about the SUOX gene is that it produces an enzyme that is inhibited by tungsten exposure. That may be significant because unusually high levels of the naturally occurring metal have been found in the blood of Fallon children, both those with cancer and matched healthy controls. While case-control studies have not associated tungsten exposure with cancer in Fallon or anywhere else, it is possible that the gene variation may make certain children more vulnerable to tungsten-related toxicity. See Karen K. Steinberg et al., “Genetic Studies of a Cluster of Acute Lymphoblastic Leukemia Cases in Churchill County, Nevada,”
Environmental Health Perspectives
115:1 (January 2007): 158–64.

8.
Barry Finette’s Toms River research did uncover some suggestive evidence, but he was not sure what it meant, if anything. He found that small changes in three genes—a detoxifier on Chromosome One known as EPHX1, another detoxifier on Chromosome Eleven called GSTP2, and a DNA repairer on Chromosome Fourteen called APEX1—more than doubled cancer risk among children who did not live in Toms River. For Toms River children, however, the presence of the three genetic alterations did not significantly change their cancer risk. A more intriguing result concerned the gene ABCB1, on the long arm of Chromosome Seven. A drug transporter gene, it affects a cell’s ability to expel invading molecules that are capable of crossing complex bodily barriers, including the placenta that separates mother from child. Finette found that Toms River children were more likely to have one version of the ABCB1 gene than out-of-town children. But those differences in genotype did not match up with differences in cancer risk—at least, not that could be elucidated from such a small sample.