Toms River (69 page)

The chemical’s defenders got to do their own slide review, too. Dow Chemical and the Saudi Arabian company SABIC, a petrochemical behemoth that still made plastics from acrylonitrile and styrene, had formed an organization called the SAN Trimer Association (“SANTA” was its incongruously jolly acronym) and hired their own pathologists to second-guess the judgments of the government’s reviewers. For Dow and SABIC, the stakes were high: Regulatory agencies all over the world, starting with the U.S. Environmental Protection Agency, routinely rely on NTP studies in setting limits on chemical exposures.
¹⁰

The seemingly endless review process for SAN trimer finally ground to a conclusion in early 2011, nearly fifteen years after Floyd
Genicola’s triumphant unmasking of the mystery compound in the Parkway well field. Two groups—the NTP senior staff, and then its outside review board—would convene at the agency’s North Carolina headquarters to pass judgment on the trimer’s carcinogenicity. It was a tough call. The slides showed that, except for one type of tumor, cancer incidence among the three hundred trimer-fed rats was similar to the one hundred unexposed controls. The exception was brain and spinal tumors: Eight of the trimer-fed rats had them, compared to just one unexposed rat. That was startling because brain tumors normally were very rare in rats. They had been found in just four of the almost thirteen hundred Fischer rats used as unexposed controls in recent NTP studies. That meant that the tumor count in trimer-fed rats was almost nine times higher than expected. Even more interestingly, there seemed to be a weak yet discernible dose-response pattern in which rats fed higher doses were more likely to get tumors.
¹¹

The Gillicks and the other families had always thought of themselves as involuntary guinea pigs in an uncontrolled experiment, which is why they had pushed so hard for the multigenerational rat study. Now its results were showing that in at least one important sense, the trimer-fed rats really did resemble the thousands of Toms River children who had been exposed prenatally to much lower levels of SAN trimer in their drinking water. In both populations, certain tumors were much less rare than expected. Further, there were suggestions in the rat study, just as there had been in Jerry Fagliano’s case-control studies of the Toms River children, that those with the greatest exposure faced the highest risk.

Was SAN trimer truly a cancer-causer? As usual, the study population was too small to know for sure. If the pathologists had found a brain tumor in just one more high-dosed male rat—three instead of two—then the dose-response pattern would have been clear. But with just fifty male rats in that dose category, the statistical ambiguity could not be resolved.
¹²
Nor was there any way to know whether SAN trimer caused leukemia—always the key concern in Toms River—because the type of rat the National Toxicology Program had selected for testing was prone to leukemia and thus a poor test model.
¹³
Despite these shortcomings, there was no chance the agency was going
to further delay its decision by conducting additional tests to try to clear up the ambiguities—not after a decade of work and several million dollars already spent. After arguing about the data for hours at a meeting on January 24, 2011, the NTP’s senior staff reached its verdict: equivocal evidence of carcinogenicity, the middle of five possible ranks. It was a compromise, a nuanced response to provocative but highly uncertain evidence.

It held sway for exactly two days. When the NTP’s Board of Scientific Counselors convened on January 26 to consider the case, a raft of industry consultants were waiting to testify. Dow offered four speakers, including James Swenberg of the University of North Carolina, a former chairman of the advisory board he was now trying to influence. Batting cleanup was Joseph Haseman, a consultant who had spent thirty-three years at the NTP, most of them as its chief biostatistician. When it was his turn to speak, Haseman launched a stinging critique of the staff’s decision. Based on past precedent at NTP—precedent Haseman had established—the correct call for SAN trimer was “no evidence of carcinogenicity,” he insisted.

The families of Toms River had no hired guns to match them. When the panel’s chairman called on the next speaker, a familiar voice—both pleading and reproachful—issued from a speakerphone. It was Linda Gillick, calling from Toms River. “I really believe our children were the real rat study,” she said, her voice crackling across the silent room. “Trying to take what really happened in Toms River and duplicate it is really impossible.… I’m just concerned that at the end of all this time, effort and money, and all the lives that have been affected and lost, we won’t come out of here with true answers.” Bruce Anderson had put his comments in a letter to the NTP. “This cannot be allowed to happen to any children again,” he wrote. “We hope that the results of this testing will lead to more proactive protection.” By that standard, the staff’s proposed finding of “equivocal evidence” could only be seen as a disappointment, the same ambiguous mush the families had been hearing from government experts for almost twenty years.

The Board of Scientific Counselors was not interested in equivocation either, but its solution was not what Gillick or Anderson had in
mind. The coup de grâce came from board member Jerry M. Rice, who had been a senior scientist at the International Agency for Research on Cancer before becoming an industry consultant and academic. The Toms River cluster, he asserted, was probably a chance occurrence—no matter what Fagliano’s case-control studies had concluded back in 2001. “It’s well known that rare events cluster in space and time,” he said, adding that “an occasional brain tumor” in rats was trivial. He proposed changing the official classification to “no evidence of carcinogenicity.” The vote came a few minutes later: six to one for Rice’s proposal, with several board members abstaining because they had business ties to Dow.

In the back row of the room, a man named Bob Fensterheim shook hands with his expert witnesses and then left for his flight back to Washington, D.C. A former top lobbyist for the American Petroleum Institute, Fensterheim now was a consultant who specialized in helping chemical manufacturers in their battles with regulatory agencies. He ran no fewer than ten industry groups, with names like the Chlorinated Paraffins Industry Association and the Alkylphenols and Ethoxylates Research Council. His latest creation, SANTA, the SAN Trimer Association, could now fade into obscurity. Its work was done.

In the morning, styrene acrylonitrile trimer had been a possible carcinogen, at least worthy of future study to clear up the uncertainty. By the end of the day, it was not. For the Toms River families, another door had slammed shut. Science, and scientists, had let them down so many times before that it hardly even hurt anymore.

There is one way—one very important way—in which the passage of time has strengthened the Toms River families’ conviction that industrial chemicals played a role in their children’s illnesses: The leukemia cluster is gone.

As he promised the families he would back in 2001, Jerry Fagliano has carefully tracked new local cases of childhood cancer, just as Linda Gillick still does at Ocean of Love. As always with a rare disease in a small population, the case counts vary so much from year to year due to chance that the underlying trend is not always clear. To smooth out this instability, Fagliano uses five-year running averages.
What he has found for childhood leukemia is that incidence in Toms River, after peaking in the late 1980s, has fallen sharply since then and is now below the statewide average. When charted through 2009 (the last year that fully verified data is available), the graph looks like this:
¹⁴

It is an encouraging graph. Leukemia was the type of cancer that Fagliano’s case-control studies had most strongly associated with pollution, so if the study were valid, the number of leukemia cases presumably would fall as environmental conditions in Toms River improved. That is what seems to have happened: Rates declined at the same time the factory wound down its operations and the Parkway wells were filtered or shut down.

It is just another correlation, not a proven cause-and-effect relationship, but in the eyes of the families, it is no coincidence. The leukemia decline, many of them believe, demonstrates that pollution was responsible for the cluster and that their activism has thus saved lives. “There are so many fewer cases than we had before all the changes that took place in this town, and that’s the way we intend to keep it,” Gillick declared at a public meeting in 2011. Jerry Fagliano is less sure about the reasons but is encouraged by the trend. “It’s somewhat comforting,” Fagliano said, “in that leukemia was the one kind of cancer that really stuck out as being elevated when we did our case-control studies. The fact that it’s come back down to normal can
mean one of two things: Either the association we saw at that time was causal and we’ve removed the cause, or else we’re simply seeing a return to background rates because of chance. We hope the actions we took have made a difference for public health, but I can’t say we’ve proven it.”

One reason for Fagliano’s uncertainty is that the trend for all childhood cancers, not just leukemia, is murkier. After falling steadily for a decade, there was an abrupt spike: Eight cases were diagnosed in Toms River in 2004 and nine in 2005, up from the usual four or five. Was the spike just another random consequence of the natural variability of small-number statistics? The fact that the yearly counts have fallen back somewhat since then suggests that it was. So does the fact that no one type of cancer appears to be unusually elevated—unlike in the 1980s, when leukemia incidence jumped. But 2009 was another unusually bad year, with nine cases reported in Toms River, so Fagliano’s latest five-year running averages for all types of cancer counted together look especially alarming when charted:

The curve is likely to bend back down again soon because the last few years have been good ones: just two childhood cancer cases in 2010 and three in 2011. But those recent numbers are still unofficial and uncounted in Fagliano’s five-year running averages. Where rates will go next is anyone’s guess. “It is frustrating because the rates are so
unstable that there’s only so far you can take any interpretation,” Fagliano said. “It really points to the difficulty of making sense of cancer rates in relatively small populations over relatively short periods of time. The rates are going to bounce around, even over five-year intervals. With more time, it may become clearer. Maybe.”

So what was it, really? Was the Toms River childhood cancer cluster a mirage, an aberration, or a warning? Was it a consequence of nothing but a stunningly bad run of luck, like rolling snake eyes six times in a row? Or was it the product of pollution so horrendous and governmental neglect so extreme that the combination has never been replicated anywhere but Woburn? And what about the third possibility, the one raised by David Ozonoff’s dictum that a public health catastrophe is an effect so strong that even an epidemiological study can detect it? Could it be that the only unique thing about the Toms River and Woburn clusters is that anyone managed to recognize them?

Daniel Wartenberg thinks he knows. An epidemiologist at Rutgers, he has been studying clusters for more than thirty years. Wartenberg was not involved in Toms River, except for brief service as an outside reviewer of the case-control studies, but has participated in dozens of other cluster investigations. The reality of cluster studies, he argues, is that researchers are much more likely to underestimate the number of true, nonrandom clusters than overestimate them. Typical tests of statistical significance are so conservative, he explains, that clusters that fail those tests and are dismissed as mere products of bad luck are actually true clusters 20 percent of the time. The same tests lead to errors in the other direction—a random cluster wrongly identified as a true one—just 2.5 percent of the time.
¹⁵
He asks, “Should we be comfortable in missing eight true clusters for every false one we detect?”