Thinking about Harris v. CSX Transportation, Inc. and trying to understand how a court could come to believe that an educated guess that has never been tested, or one that has been repeatedly tested and serially refuted, could nevertheless constitute scientific knowledge I thought I'd reread Milward v. Acuity Specialty Products: Advances in General Causation Testimony in Toxic Tort Litigation by Carl Cranor. It was published earlier this year in a Wake Forest Law Review issue devoted to advancing the thinking that produced Milward and now Harris. In it he sets out his argument that (a) "[t]he science for identifying substances as known or probably human carcinogens has advanced substantially" over the last quarter century and (b) where science leads, courts should follow. (Cranor you'll recall is a philosopher and "the scientific methodology expert for plaintiffs in Milward".)
Cranor begins by asking you to imagine having been diagnosed with "early stage bladder cancer that had been caused by MOCA (4,4'-methylenebis(2-chloroaniline)" following exposure to the chemical in an occupational setting. He then reveals that though "IARC classifies MOCA as a known human carcinogen" many judges would nevertheless deny you your day in court because they don't understand the "new science" for identifying the etiology of cancer. You see, while IARC concluded that "[t]here is inadequate evidence in humans for the carcinogenicity of 4,4'-methylenebis(2-chloroaniline)" its overall evaluation was that "4,4'-methylenebis(2-chloroaniline) is carcinogenic to humans (Group 1)" notwithstanding! And its rationale? MOCA (sometimes bearing the acronym MBOCA just to confuse things) is structurally similar to aromatic amines that are urinary bladder carcinogens (especially benzidine), several assays for genotoxicity in some bacteria and fruit flies have been positive, in rats and dogs MOCA can form DNA adducts, mice and rats and dogs exposed to MOCA develop cancer (dogs actually develop urinary bladder cancer), one of those DNA adducts found in dogs exposed to MOCA was found in urothelial cells from a worker known to have been exposed to MOCA, and finally an increased rate of chromosomal abnormalities in urothelial cells have been observed in some people exposed to MOCA.
At that point I stopped re-reading Cranor's paper and started looking into MOCA.
If MOCA really is a human urinary bladder carcinogen and if thousands of people have been exposed to MOCA in their work for many decades why is there no evidence of an increased risk of malignant urinary bladder cancer among them? Cranor claims the reason IARC concluded that there's "inadequate evidence" for MOCA being a human carcinogen is because "there are no epidemiological studies". Are there no such studies? If workers exposed to MOCA develop the same DNA adducts demonstrated in dogs and if four out of five dogs exposed to MOCA develop bladder cancer then where are all the human cases? And what's the story with the dogs?
It turns out there is an epi study of MOCA-exposed workers. The study was initiated in 2001 and its results were published four years ago. Only one death from bladder cancer was identified and it was not known whether the man was a smoker (a leading cause of bladder cancer). There was one bladder cancer registration for a man who had survived his cancer but he was a former smoker. Finally, there was one case of noninvasive, or in situ, bladder carcinoma; that case was excluded from analysis as there is no reference population that has been screened for benign tumors from which a background rate can be generated (take note of this case of a benign tumor, the significance of which no one can say as it will shortly become important). None of the findings allowed the researchers to reject the null hypothesis, i.e. that MOCA doesn't cause bladder cancer in humans.
Then there's "Bladder tumors in two young males occupationally exposed to MBOCA". This study was conceived because "[i]n addition to their chemical similarity MBOCA and benzidine have similar potency to induce bladder tumors in beagle dogs, the species considered to be the best animal model for humans", because 9,000 to18,000 workers were being exposed to it, because it was not regulated as a carcinogen and because they had a group of 540 workers - workers whose smoking history was known, who hadn't been exposed to other bladder carcinogens and who had been screened for signs of bladder cancer and monitored for exposure since 1968. Why were they screened and monitored?. Benzidine and 2-naphthylamine are aromatic amines that long before 1968 were known to consistently produce very high numbers of malignant bladder cancers among exposed workers (with the incidence of malignant bladder cancer reaching 100% in one study of very highly exposed workers) so it was reasonably conjectured that all aromatic amines might cause malignant bladder cancer.
Of the 540 workers none had died of or had symptoms of bladder cancer. However, two workers had been identified for follow up after screening and biopsies on each revealed non-invasive papillary tumors of the bladder. Because, again, there is no reference background population it was impossible to say whether the finding meant the risk of non-malignant asymptomatic tumors among MOCA workers was higher, lower or the same as expected among the unexposed. Nevertheless, after returning to muse about those MOCA-exposed beagles that had developed papillary carcinomas of the bladder the authors concluded that "[t]he detection of the two tumors in young, nonsmoking males is consistent with the hypothesis that MBOCA induces bladder neoplasms in humans."
And that's it for evidence of bladder cancer in humans from studies of humans exposed to MOCA - which is to say that nobody has ever found anyone exposed to MOCA to be at an increased risk of dying from bladder or even at an increased risk of developing clinically apparent bladder cancer. But at least it kills beagles, right?
Before I looked at the animal studies on MOCA I assumed there'd be lots, that they used modern techniques and that they'd been replicated; likely several times. Not so. IARC cited no rodent study done since the 1970s. It's summary of testing listed one study on mice, five for rats and just one for dogs. The mice were fed lots of MOCA for 2/3 of their lives and many developed hemangiosarcomas (mice and dogs get it, you don't and the finding probably isn't relevant to humans in any event) and liver tumors. The rats were fed lots of MOCA in different regimens that varied by dose and protein sufficiency. In one study they got common liver and lung tumors. In the next only liver tumors. In the third lung, liver and even mesotheliomas. Lung, liver and mammary gland tumors were found in the fourth. Plus lung, mammary gland, Zymbal's gland, liver and hemangiosarcoma in the fifth. The rates were often strongly affected by protein insufficiency status. Neither mouse nor rat developed bladder cancer. But those beagles sure did. Well, four beagles did anyway. But that's not what killed them.
In the late 1960s six one year old beagles were started on 100 mg of MOCA three days a week. Six weeks later the dosing was increased to five days a week. Another six beagles which were fed no MOCA served as controls. Time passed. Man landed on the Moon, Watergate brought down a President, the Vietnam War ended, the PC was launched and the first fiber optic cable laid and yet five of the six MOCA beagles carried on (one having died along the way of an unrelated infection). Eventually, as the dogs approached their second decade of life, urinalysis suggested possible neoplasia so one was killed and dissected. It was healthy (other than the usual ailments associated with being 65 dog-yrs old) but it did have a papillary bladder cancer - one that had neither metastasized nor invaded surrounding tissues. Eight months later, having enjoyed it is hoped seventy happy dog years of life, the remaining four beagles, undaunted and asymptomatic to the last, were also killed and dissected. Three of the four had non-invasive, non-malignant bladder cancer. None of the controls, which met a similar fate, had non-invasive, non-malignant bladder cancer.
And that's it. Five dogs in a single study (never replicated) that began more than 45 years ago and ended before PCR, etc. were fed MOCA for a lifetime and never noticed, and perhaps never would have noticed, any effect but for having been "sacrificed" once their biggest mortality risk became old age.
OK. Four of five dogs fed MOCA in a study done long ago developed a non-malignant, non-invasive bladder cancer whereas none of the unexposed dogs developed the condition. Two humans out of 500+ exposed developed the same non-malignant, non-invasive disease and in a different study one human exposed to MOCA had a DNA-adduct like that of an exposed dog. Setting aside the growing skepticism about the usefulness of animal studies let's assume this decades-old study proves that exposure to MOCA causes non-invasive and non-malignant bladder cancer. So what to make of it?
First there's the issue of screening bias. You need look no further than the Japanese and Canadian childhood neuroblastoma screening studies to understand that lots of people, including children, get otherwise deadly cancers that simply go away on their own and that screening in such cases merely increases morbidity without decreasing mortality.
Second there's the whole rationale for labeling MOCA a human carcinogen because its metabolites look like those of benzidine (which does cause malignant bladder cancer). So it walks like a duck and quacks like a duck. It still doesn't cause malignant bladder cancer. Shouldn't that give IARC pause? If you decide MOCA's a carcinogen because metabolism turns it into the same things that benzidine gets turned into shouldn't you be scratching your head when it doesn't cause malignant bladder cancer in mice, rats, dogs or humans? Isn't a little humility in order?
Finally, what do you do with such a causal claim in a toxic tort case? If you don't know how many people get non-malignant, non-invasive bladder cancer how do you know whether MOCA increases, decreases or has no impact on the risk of contracting it? In other words, if you don't know what the background rate for non-malignant, non-invasive bladder cancer is, and you don't know by how much MOCA increases the risk, how can you ever say it's more likely than not a cause of any bladder cancer, much less a particular plaintiffs?
You can't, and that's why the Milward plaintiffs lost when they got back to the trial court. They could make no argument other than to conflate risk with causation. Shorn of the fallacious assertion that any risk is post hoc necessarily causative they couldn't say why benzene was more likely than not the cause of decedent's acute promyelocytic leukemia. They simply couldn't make a coherent argument as to why one putative cause among many was the most likely in the absence of any evidence about which cause was the most common.
In the end Milward serves only as a bad example. An example of what happens, of the time and money wasted, when the law tries to outrun science.