Throughout my blog posts, I’ve noted how challenging the science in this case is to digest, and wondered how well the jury could possibly be following it.
Today, Monsanto calls expert witness Dr. Lorelei Mucci, a leading cancer epidemiologist from Harvard, to do her own assessment of the epidemiology that she finds most compelling in addressing association between glyphosate and NHL. Monsanto has her spend an unusually long time explaining in detail how one determines the strength of an epidemiological study. She also describes, very clearly on that jumbo notepad, how to understand the statistical methods behind those determinations.
This classroom component visibly turns lights on in the heads of many members of the jury, and I believe some begin to comprehend the epidemiological studies in a much more sophisticated way. Mucci is a highly gifted teacher. What is concerning is that with the jury’s fresh ability to understand how to really read a study, Monsanto is able to spoon feed their relevant epidemiology in a visual format commensurate with the way in which Mucci taught.
Defense masterfully delivers free education, and the tools and analytical frameworks that they teach dovetail oh so conveniently with the arguments that they make for and against the studies that link glyphosate to cancer. It’s just like Microsoft’s strategy: We’ll give you Windows for free, but it will be hard to run anything by Apple.
With that, on to an A+ expert testimony. I’m going to do a deeper dive today, just because the game plan was so well executed by Monsanto. We all should know about it, like it or not. Rest assured that by the end of the day, I deem the Direct and Cross a wash. No worries.
THE TESTIMONY OF DR. LORELAI MUCCI
We have a lot of pink going again today! Bolanos rocks her pink collar (and on a Tuesday!), and Dr. Mucci dons a lovely, modest soft pink dress with sophisticated gray flowers printed throughout. Her blazer is a similar petal pink. Dr. Mucci’s tone of voice is clear, direct, and intentional, but with an underlying softness which makes her quite likable.
Dr. Lorelei Mucci is a cancer epidemiologist. She is a professor at the Harvard School of Public Health and head of the Cancer Epidemiology program at the Dana Farber/Harvard Cancer Center. She received her BS from Tufts, MPH from Boston University, and PhD at the Harvard School of Public Health. Mucci has published over 300 peer-reviewed research articles and written chapters for textbooks. Lombardi has a copy of the textbook on his desk, and asks her if she finds writing textbooks “fun.” She says that she indeed does.
Mucci was approached by defendant counsel to provide an independent evaluation of epidemiological studies. After looking through the studies on her own, she agreed to do an expert report. She has never testified before a jury and is being compensated $350/hour for her work.
Lombardi moves to designate Dr. Lorelai Mucci as an Expert in Cancer Epidemiology. Before Bolanos confirms the expert designation, Wisner clarifies with Mucci that she is only here to review epidemiological studies of NHL only, not animal nor mechanistic studies. Mucci agrees.
DIRECT EXAMINATION OF DR. LORELAI MUCCI
An overtly friendly version of Mr. Lombardi opens his questioning in an animated tone and something makes me want to call him a crossover of Pat Sajak and Alex Trebek.
Mucci begins by discussing the history of Epidemiology. Why is such an ingratiating historical lesson like this familiar? We learn about John Snow’s cholera study of the 1850s, in which Snow narrows down a cholera outbreak to a specific well. Wisner later makes a joke about John Snow from Game of Thrones, but it is entirely lost on Mucci. She must not be a fan.
IN CASE YOU ARE INTERESTED #1: A CRITICAL APPROACH TO ANALYZING EPIDEMIOLOGICAL STUDIES
Dr. Mucci discusses that in any epidemiology study, she considers if results could be influenced by confounding, bias, or chance:
- Confounding: In a glyphosate epidemiological study, a significant confounder is the simultaneous use of glyphosate and other pesticides. Race may be another confounder, in that we know that race itself doesn’t cause cancer, but race is a stand in for biological factors or social inequities. If a glyphosate study doesn’t adjust for other pesticides, it is very concerning.
- Bias: In collecting information from questionnaires for a study, an epidemiologist should question how reliable that information is. In case-control studies, the way in which cases are recruited may cause selection bias. Today, the plan is to talk about proxy bias. When an epidemiologist collects followup data on a person, that person could be sick or dead. In that case, a “proxy” – spouse of family member – will provide information on estimated pesticide exposure. Mucci points out that because a proxy who lost their family member may be “ruminating” on the cause of the cancer, he/she may not give a completely reliable exposure estimate.
- Chance: Chance refers to how likely the finding that you observe in your own data is due to chance. A significant factor that contributes to the role of chance in study results is when a study has a low number of “exposed cases,” meaning people who have NHL and have been exposed to glyphosate. In addition, an epidemiologist uses confidence intervals to look at the relative risk of NHL between exposed and unexposed people.
IN CASE YOU ARE INTERESTED #2: LESSON ON RELATIVE RISK AND CONFIDENCE INTERVALS
Confidence interval lesson:
CASE A: 1000 people, 100 exposed cases
CASE B: 1000 people, 500 exposed cases.
Assume that the Relative Risk (RR) of both studies is 1.5.
RR is calculated as: The risk of the disease in people who have been exposed to glyphosate/risk of the disease in those unexposed to glyphosate.
If there is no risk, the RR = 1.0
In CASE A, the RR is 1.5, but with a much wider range in confidence (confidence interval)
In CASE B, the RR is 1.5, but due to so many exposed cases, the confidence interval is much smaller because the study is more reliable with more information.
In other words, in CASE A, your estimate for the likelihood that one develops cancer from exposure is much less certain because you have fewer observations of exposed cases on which to measure the likelihood of cancer for that population. Likewise, in CASE B there are many more observations of exposed cases and thus your estimate of the risk of developing cancer in that population should have greater precision.
Dr. Mucci’s explanations and Mr. Lombardi’s facilitation are much appreciated. Most of the jury is urgently writing and looks relieved to have such high quality tutoring. Naturally, they also like the teacher and her ability to make the convoluted seem straightforward.
EXPLORATORY STUDIES ON NHL AND GLHPHOSATE
Mucci has prepared a chart that evaluates selected exploratory epidemiological pesticide studies.
For each study, Dr. Mucci describes:
- When the cases were diagnosed
- How many exposed cases were in the study
- Whether respondents answered questions by proxy or themselves (AKA BIAS)
- If the study adjusted for other pesticides (AKA CONFOUNDING)
- The Relative Risk Ratio/Confidence Interval (AKA CHANCE)
Look at that – EASILY understood because we just had excellent instruction so as to understand every portion of this chart. Grumble.
After lengthy discussion about each study, Mucci concludes that in looking at all five exploratory epidemiology studies, the information presented was fairly limited. She is concerned that there may be bias, confounding, and chance at play, even in those results that are positive and statistically significant.
GLYPHOSATE POOLED STUDIES
Yesterday, we heard about the NAPP study through Dr. Blair’s testimony. From that testimony, it appeared that the evidence that emerged from that study rejected any association between glyphosate and NHL. Mucci continues to develop that theme.
The NAPP study uses pooled data from the McDuffie and DeRoos studies. (1)
Wisner objects that the judge had ruled not to publish the NAPP study slide. Counsel is called into a sidebar. It is very loud whispering today, back behind the judge’s bench. Each time they take a sidebar, they huddle around a 1950s looking microphone linked to the recorder. Because Bolanos remains seated in the high seat, I can only see a dark brown flop of hair as she cranes down to the mic level. Her hair is enviously thick. The lawyers are on the ground level, and we can just barely see the tops of their heads.
Today, we may not see them but we hear them. Wisner says in a whispered-yell: “THIS IS A VIOLATION” and “THAT’S UNACCEPTABLE.” The jury has a little laugh.
The objection doesn’t work, and here we go. The Defense shows an isolated slide about the NAPP study to the jury. It was originally shown at a conference in Brazil in August of 2015. The study breaks the NHL cases into three specific lymphoma subtypes and a fourth category includes all other subtypes. None are the Mycosis Fungoides subtype from which Johnson suffers.
Mucci concludes from the data on the slide that when adjusted for three pesticides commonly used with glyphosate, the odds ratio is 1.45 and not statistically significant. She concludes that the paper shows the presence of confounding by other pesticides. When she takes that into account, there is no longer an association between glyphosate use and NHL.
Mucci also explains a slide from the same presentation that discusses proxy bias, and how it likely moves the risk ratio closer to the null hypothesis of no effect from glyphosate exposure.
GG Sidebar: The NAPP study still hasn’t been published in any journal, which is why the Defendant is referring to these isolated summary slides from a conference presentation. The NAPP study actually concluded that there is a positive association between glyphosate and NHL. As of now, the jury doesn’t know that to be the case. And many are still writing everything that Mucci says. (1)
GLYPHOSATE COHORT STUDIES
Because you have been missing it desperately, we are back to discuss the Agricultural Health Study (AHS), which the Plaintiff witnesses discussed at length over the last two weeks. Monsanto relies heavily on the 2018 AHS study by Andreotti to devalue any previous epidemiological studies and wage war on IARC. They also heavily depend on it in this case. (2)
Remember that this is the study that Dr. Neugut claims is like “measuring shit with a gold scale.”
I will spare you the belaboring of this study again. Mucci of course defends the imputation method used and concludes that there were “no associations between glyphosate use and NHL overall, or in any of its subtypes.”
The problem for the Plaintiff is that Mucci is so clear and concise that I fear that she has briefly become the dear epidemiological tutor to the jury. She established a relationship of sorts with the jury in the beginning of her testimony, and that bond may remain. She’s all relatable and dressed in pink to boot.
Mucci puts together her own meta-analysis, using the same method as IARC, but including the 2018 Andreotti study and the 2015 NAPP study. She excluded the Hardell study, which IARC used, because she claims it unreliable.
Results? The Mucci meta-analysis relative risk assessment concludes that there is no association between exposure to glyphosate and to NHL, based on whether individuals were ever exposed to glyphosate (as opposed to basing the analysis on the *duration* of glyphosate exposure). “Based on the epidemiological evidence, there is no causal association between exposure to glyphosate based herbicide and NHL risk.”
CROSS EXAMINATION OF DR. LORELAI MUCCI
Everyone in this courtroom must be aware that the direct was incredibly strong, and Wisner has catchup to do. No pressure. Just don’t drop that fly ball, 2 outs, final inning, 7th game of the World Series. Low stakes. I don’t envy him.
Wisner immediately points out that Mucci has been paid $100,000 to be a consultant on this case. The money is paid directly to Mucci, not to Harvard.
Mucci read pieces of the reports created by Dr. Portier and Dr. Neugut. She points out that both men agreed that there is limited evidence of carcinogenicity in the epidemiological studies. There is a several minute long back and forth between Wisner and Mucci, Wisner demanding her clarification of association vs causation, and what limited evidence means. She says that she doesn’t “understand what you are asking.” I have to say, I don’t really know either, I just see that he is pissed.
We move on to delve into Mucci’s juicy Textbook of Cancer Epidemiology. Wisner asks several questions about whether one can make a carcinogenicity determination even with limited epidemiology. Very nice, Wisner, weaving in that regardless of what she says about epidemiology, Plaintiff has some sweet animal and mechanistic analyses on his side. Mucci’s book uses IARC as a benchmark for criteria one would use to determine carcinogenicity, including that even with limited epidemiology, the classification of a Class 1 carcinogen can be made. In fact, there are 475 references in her textbook to IARC. Furthermore, Wisner claims that there are only around 2 references to EPA. Quite telling.
Next up, the claim that Mucci made that proxy bias tends to push Risk Ratios upwards. Wisner explains that Dr. Blair published that proxies will attenuate risk towards the null, so if included, are more likely to lower the Risk Ratio, not push it up. Mucci backs down and says that sometimes there are under or overestimates, but in this case, the bias was away from the null.
MUCCI’S META ANALYSIS AGAIN
Wisner points out that Mucci didn’t present any of the statistically significant results from the highlighted research papers in her exploratory studies chart, including Hardell. A long debate ensues surrounding the McDuffie study and whether or not proxies were used. Wisner and Mucci are evenly matched, and Mucci has no fear in responding with answers like: “I disagree with the authors completely” and “Not true.”
CONCLUSIONS OF THE CROSS-EXAMINATION
There are hours more of debate between Wisner and Mucci, so if any readers would like more detail, I am happy to provide more. I know that there are several epidemiologists reading this blog, so I don’t want to short change you!
At one point, Mucci slips and mentions a detail of Johnson’s mode of exposure, and Wisner pounces, asking her how she knows that detail about Johnson. With further pressing, Mucci admits that she is restating the words of Mr. Lombardi. This callout tries to paint Mucci as a parrot of the Monsanto legal team. It was a small moment, but an interesting one.
In summary, Wisner argued for interpretations of epidemiological studies that align with IARC interpretations. In my opinion, he did so very successfully, and softened the effect of Mucci’s strong testimony. Regarding post-IARC epidemiological research releases of NAPP and Andreotti, he successfully showed the jury that the NAPP study actually found some positive associations between glyphosate and NHL, and that the AHS study has flaws that even a scientist at Monsanto identified in 1997.
Mucci was incredibly well coached and prepared for the eventualities of this testimony. She did frequently look to the Defendant counsel table, for reassurance maybe?
When Wisner points out that she cites Dr. Neugut several times in her textbook, he alludes to him as The Grandfather of Cancer Epidemiology. Mucci says that she does not concur.
Today is Juror #8’s birthday! During the afternoon break, I hear loud singing out in the hallway. One of the alternates brought in pastel colored cupcakes from Susie Cakes to celebrate. I can tell that coming back into the courtroom to hear more epidemiology is probably not top of their list of desires. I love how close they have become.
Because it is on my mind… Earlier this year, the NIH gave Harvard scientist Kenneth Mukamel $100 million to fund a global study comparing people who drink with those who don’t. Its conclusions could have enshrined alcohol as part of a healthy diet. As it turned out, much of the money for the study came from the alcohol industry. The New York Times reported that officials at the National Institute on Alcohol Abuse and Alcoholism, part of the N.I.H., had solicited that funding from alcohol manufacturers, a violation of federal policy. (3) This scandal doesn’t involve Dr. Mucci, but I find it interesting that another Harvard researcher (who has published papers on cancer epidemiology), was involved. Also, an inside look at questionable sourcing of NIH grants.
Thank you for your emails and comments! I am reading them, but am so swamped with generating an article every day that I can’t respond quite yet. I will soon!
© 2018 Kelly Ryerson ALL RIGHTS RESERVED