Preview
ELECTRONICALLY FILED
Superior Court of California
County of Santa Barbara
Darrel E. Parker, Executive Officer
Jessica L. Grant (SBN: 178138) 7/1/2024 1:24 PM
John K. Sherk (SBN: 295838) By: Narzralli Baksh , Deputy
Kara M. Flageollet (SBN: 347646)
SHOOK, HARDY & BACON L.L.P.
555 Mission Street, Suite 2300
San Francisco, California 94105
Tel: (415) 544-1900 | Fax: (415) 391-0281
jgrant@shb.com
jsherk@shb.com
kflageollet@shb.com
Jason Zager (admitted pro hac vice)
SHOOK, HARDY &BACON L.L.P.
2555 Grand Boulevard
Kansas City, MO 64108
Tel: (816) 474-6550 | Fax: (816) 421-5547
jzager@shb.com
10 Attorneys for Defendant
MONSANTO COMPANY
11
12
13
SUPERIOR COURT OF THE STATE OF CALIFORNIA
14
FOR THE COUNTY OF SANTA BARBARA
15
16 MARK JON ROGERS, Case No. 23CV00945
17 Plaintiff, DECLARATION OF JESSICA GRANT IN
SUPPORT OF MONSANTO COMPANY’S
18 Vv. OPPOSITION TO PLAINTIFF’S MOTION
IN LIMINE NO. 27
19 MONSANTO COMPANY,
Date: July 10, 2024
20 Defendant. Time: 10:00 am
Dept.: 3
21 Judge: Honorable Thomas Anderle
22 Filed concurrently herewith: Monsanto
Company’s Opposition to Plaintiff’s Motion In
23 Limine No. 27
24
25
26
27
28
DECLARATION IN SUPPORT OF MONSANTO’S OPPOSITION TO PLAINTIFF’S MOTION IN LIMINE NO. 27
I, Jessica Grant, hereby declare as follows:
1 I am a partner at the law firm of Shook Hardy & Bacon, counsel for defendant
Monsanto Company (“Monsanto”) in the above-captioned action. I am over eighteen years of age
and am fully competent to make this declaration in support of Defendant’s Opposition to Plaintiff’s
Motion in Limine no. 27. I also make this declaration based on my personal knowledge and if called
as a witness, I could and would competently testify to the matters stated herein.
2 A true and correct copy of the June 20, 2024 order entitled Order Granting Motion to
Exclude Dr. Luoping Zhang and Granting Summary Judgment To Monsanto, in In re Roundup
Prods. Liab. Litig., No. 16-md-02741 is attached hereto as “Exhibit 1.”
10 3 A true and correct copy of Zhang, et. al., Exposure to glyphosate-based herbicides
11 and risk for non-Hodgkin lymphoma: A meta-analysis and supporting evidence (2019), is attached
12 hereto as “Exhibit 2.”
13 4. A true and correct copy of the January 6, 2020 EPA Memorandum entitled
14 Glyphosate: Epidemiology Review of Zhang et al. (2019) and Leon et al. (2019) publications for
15 Response to Comments on the Proposed Interim Decision is attached hereto as “Exhibit 3.”
16 5. A true and correct copy of excerpts from Dr. Kristan Aronson’s December 15, 2023
17 Expert Report entitled General cancer causation for exposure to glyphosate and glyphosate
18 formulations in the case styled Brazell v. Monsanto, No. 1922-CC11327 (Mo. Cir. Ct., St. Louis
19 Cnty.) is attached hereto as “Exhibit 4.”
20 I declare under penalty of perjury under the laws of the State of California that the foregoing
21 is true and correct.
22 Executed this 1" day of July, 2024 in San Francisco, California.
23 Respectfully submitted,
SHOOK, HARDY & BACON L.L.P.
24
By:
A
25
wo ca L. Grant
26
neys for Defendant
27 MONSANTO COMPANY
28
DECLARATION IN SUPPORT OF MONSANTO’S OPPOSITION TO PLAINTIFF’S MOTION IN LIMINE NO. 27
EXHIBIT 1
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 1 of 14
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
IN RE: ROUNDUP PRODUCTS MDL No. 2741
LIABILITY LITIGATION
Case No. 16-md-02741-VC
This document relates to: ORDER GRANTING MOTION TO
EXCLUDE DR. LUOPING ZHANG
Bulone v. Monsanto Co., Case No. 20-cv- AND GRANTING SUMMARY
03719-VC JUDGMENT TO MONSANTO
Re: Dkt. No. 18022
Monsanto’s motion to exclude the testimony of Dr. Luoping Zhang is granted. This order
assumes familiarity with the Court’s prior orders on general and specific causation and the Ninth
Circuit’s opinion in Hardeman. See generally, e.g., In re Roundup Products Liability Litigation,
390 F. Supp. 3d 1102 (N.D. Cal. 2018) (Pretrial Order No. 45, Dkt. No. 1596); In re Roundup
Products Liability Litigation, 358 F. Supp. 3d 956 (N.D. Cal. 2019) (Pretrial Order No. 85, Dkt.
No. 2799); Hardeman v. Monsanto Company, 997 F.3d 941 (9th Cir. 2021). It also assumes
familiarity with the record from the Daubert hearing on Zhang that took place on April 24, 2024.
See Hearing Tr. (Dkt. No. 18297).
Zhang’s presentation is based on a meta-analysis of the epidemiological literature that she
published in 2019. There are several issues with her opinion, each of which is an independent
ground for exclusion. First, Zhang’s meta-analysis is junk science. It has deep methodological
problems, not the least of which is that it doesn’t achieve its stated aim of examining only
“highly exposed” individuals. Instead, it examines an arbitrary selection of the available data.
Second, Zhang’s only engagement with the epidemiological literature is her meta-analysis, and
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 2 of 14
her meta-analysis can’t substitute for a reliable general causation opinion. In other words, even if
the meta-analysis were somehow useful in the field of epidemiology, it still couldn’t reliably
serve as a litigation expert’s sole engagement with the epidemiological literature. Both of these
problems are exacerbated by the fact that Zhang struggled to answer basic questions about her
own paper, and the studies discussed in that paper. Nor can either problem be cured by intoning
that the paper was peer-reviewed and developed before an expert became personally involved in
litigation—those facts are relevant, but a court can’t wave junk science through the Daubert gate
simply because it survived some prepublication peer-review process. For each of those reasons,
Zhang’s testimony must be excluded. And without Zhang, Bulone has no admissible general
causation opinion, so Monsanto is entitled to summary judgment.
I
Dr. Zhang is a biochemist and a professor emeritus of toxicology at the University of
California, Berkeley. She has been designated to testify as a general causation expert in the
Bulone case. Zhang is Bulone’s sole expert on general causation.
Zhang’s presentation is different from those of past general causation experts in that the
only opinions she disclosed are those contained in two journal publications that she co-authored.
See Zhang Expert Report (Dkt. No. 18113-1) at 2 (stating that her “opinions in this matter, and
the bases for them, are set out in the following,” and citing two articles, without further
comment); see also Order Denying Motion to Strike Expert Report of Luoping Zhang, Ph.D.
(Dkt. No. 17743). In effect, these two publications are Zhang’s expert report. Only one ofthese
studies, Zhang’s 2019 meta-analysis, grapples with the epidemiological evidence.! That study is
the focus of Monsanto’s attack on Zhang’s presentation. Because an admissible expert opinion
on general causation is necessary to survive summary judgment, and because a complete general
causation opinion must address the epidemiological evidence, Bulone’s case will stand or fall on
' See Zhang et al., Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin
Lymphoma: A Meta-Analysis and Supporting Evidence, 781 Mutation Research 186 (2019)
[“Zhang Meta-Analysis”] (Dkt. No. 18351-4).
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 3 of 14
whether Zhang’s 2019 meta-analysis is an admissible treatment of the epidemiological evidence.
See Pretrial Order No. 45 at 13 (“As the parties acknowledge, epidemiology is central to the
general causation inquiry, and where such evidence exists, it must be addressed by the experts.”)
(citing Norris v. Baxter Healthcare Corp., 397 F.3d 878, 882 (10th Cir. 2005)); see also id. at 6.
Zhang’s 2019 paper is a meta-analysis of six epidemiological studies of varying sizes,
methods, and quality: McDuffie (2001), Hardell (2002), De Roos (2003), Eriksson (2008), Orsi
(2009), and Andreotti (2018). The paper’s approach is based on the “a priori hypothesis” that, if
glyphosate were carcinogenic, people who had been exposed to more glyphosate would be more
likely to show an increased risk of NHL. See Zhang Meta-Analysis § 2.3. Accordingly, the
authors purported to conduct the meta-analysis by isolating the data from the highest-exposure
groups, where available, across each of the studies and analyzing them using a fixed-effect
statistical model. However, not all of the epidemiological studies that Zhang analyzed broke out
their data into groups based on exposure levels. Some of the underlying studies only reported
their results on an ever-exposed/never-exposed basis. For those studies, the meta-analysis simply
used all of the data. See id. § 2.6.
The results of the meta-analysis are presented in the form of a “meta-risk ratio” of 1.41,
with a 95% confidence interval of 1.13-1.75. Id. § 3.1. Based on this top-line finding, as well as
summary discussions of other kinds of animal studies and mechanistic studies, the meta-analysis
concludes that “[t]he overall evidence from human, animal, and mechanistic studies presented
here supports a compelling link between exposures to [glyphosate-based herbicides] and
increased risk for NHL.” Id. § 9.
Monsanto’s motion to exclude Zhang focused on an EPA critique of her meta-analysis
produced during the agency’s review of glyphosate’s registration under FIFRA. See Hsu Decl.
Ex. D, EPA Review (Dkt. No. 18022-5) at 8. The Daubert hearing brought some additional
issues to light, and the Court ordered further briefing. Bulone doesn’t dispute that the Court can
consider the issues raised at the hearing in ruling on Zhang, particularly given the supplemental
briefing. See Miller v. Baker Implement Co., 439 F.3d 407, 413 (8th Cir. 2006) (“A district
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 4 of 14
court’s Daubert inquiry need not take any specific form, and its sua sponte consideration of the
admissibility of expert testimony is permissible so long as the court has an adequate record on
which to base its ruling.”) (citing Kirstein v. Parks Corp., 159 F.3d 1065, 1067 (7th Cir. 1998))
Il
In this litigation, the general causation question is “whether a reasonable jury could
conclude by a preponderance of the evidence that glyphosate can cause NHL at exposure levels
people realistically could have experienced.” Pretrial Order No. 45 at 2. Presenting an admissible
general causation opinion requires “offering independent and relatively comprehensive opinions
that the epidemiological and other evidence demonstrates glyphosate causes NHL in some
people who are exposed to it.” Id. at 3. The admissibility of any opinion that Roundup is capable
of causing NHL has always been a “very close question.” Id. at 1. Reliably offering such an
opinion requires serious engagement with the relevant literature: general causation experts must
be able to “assess whether a study is credible, to explain why they relied on one study more than
another, and to articulate how they reached their conclusion in the face of conflicting evidence.”
Pretrial Order No. 288 (Dkt. No. 17504) at 5; see also Pretrial Order No. 45 at 35.
A
The most fundamental problem with Zhang’s testimony is that her meta-analysis was not
reliably performed. It has several methodological problems that, when taken together, make its
analysis of the epidemiological literature indefensible.
First, the meta-analysis was based on the hypothesis that, if glyphosate caused NHL, then
people with the highest levels of exposure would be most likely to get sick. Zhang and her
colleagues justified this hypothesis with reference to “the understanding that higher and longer
cumulative exposures are likely to yield higher risk estimates, given the nature of cancer
development.” Zhang Meta-Analysis § 2.2; see also Hearing Tr. at 122:3—15. They also argued
that data from high-exposure groups are less likely to be affected by confounders or to be diluted
by low-exposure data points. Zhang Meta-Analysis § 2.2. But the meta-analysis doesn’t grapple
with the fact that the Andreotti study, which it says is the highest-quality epidemiological study
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 5 of 14
available, doesn’t show any kind of dose-response relationship between glyphosate exposure and
NHL. See Pretrial Order No. 45 at 26 (discussing Andreotti, which reported risk ratios below 1
for every exposure quartile, with the highest risk ratio reported for the third-highest quartile);
Zhang Meta-Analysis, Table 2 (ranking the quality of the studies included in the meta-analysis
and listing Andreotti as the best study); see also Pretrial Order No. 45 at 40 (discussing the
ambiguous evidence regarding dose response issue).
Second, in applying its hypothesis, the meta-analysis ends up mixing different types of
studies and different types of data in a way that ultimately seems hard to justify. Only half of the
studies that Zhang considered broke out their results according to the subjects’ exposure levels:
Eriksson, McDuffie, and Andreotti. See Hearing Tr. at 50:6—9. For the remaining three studies—
which reported their results on an ever-exposed/never-exposed basis—the meta-analysis simply
incorporated all of the studies’ data. So the meta-analysis used only a portion of the data from
three of the studies, and the entirety of the data from the other studies. Id. at 59:9—-12 (“[The
Court]: So you took portions of the group of people studied in Eriksson and McDuffie and
Andreotti and put them together with all of the people studied in the other three studies? A:
Yeah.”). One result of this uneven mixing of studies is that the meta-analysis, in effect,
artificially limited its use of data from Andreotti, the study Zhang said is the strongest and the
one that showed no statistically significant association between glyphosate and NHL for any
level of exposure. This issue was noted in the EPA’s assessment of Zhang’s study. See EPA
Review at 8. Using a small portion of Andreotti’s data might be defensible if the meta-analysis
really took only a portion of the data from every study. But instead, the study methodologically
de-emphasizes high-quality data that would tend to reduce the “meta-risk ratio,” while using the
full datasets from most of the lower-quality studies.”
Third, the studies that did stratify their results by exposure level used different exposure
? The treatment of the Andreotti data is especially misleading given that the meta-analysis touts
itself as “the first meta-analysis to include the most recently updated Agricultural Health Study
(AHS) cohort,” ie., the Andreotti data. Zhang Meta-Analysis § 1.3; see also Hearing Tr. at
49:21-50:5.
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 6 of 14
cut-offs and measured exposure differently. As a result, it’s likely that some people excluded by
the meta-analysis had as much or more exposure than people who were included. This makes the
method even more arbitrary. For example, the high-exposure category in McDuffie included
people who reported at least two exposure days per year. Hearing Tr. at 55:3-7. If that cutoff
were applied to the Andreotti data, it would probably capture many of the subjects from that
study’s third quartile (38.74-108.4 cumulative lifetime days of exposure) and even some from
the second quartile (13.75-38.74 cumulative lifetime days of exposure).? But these people were
uniformly excluded from Zhang’s analysis. Here again, the meta-analysis’s application of its
high-exposure hypothesis didn’t, in practice, result in the inclusion of only highly exposed
people. Instead, it resulted in the arbitrary exclusion of data from Andreotti and the over-
representation of unadjusted data from the case-control studies.
That leads to the fourth point: the meta-analysis mixes results adjusted for the use of
other pesticides with unadjusted results. See Hearing Tr. at 80:8-81:17. Indeed, the meta-
analysis’s selection criteria sometimes led Zhang to choose unadjusted data even when adjusted
data was available, which seems inexcusable. That is what Zhang did with the Eriksson study:
Eriksson reported unadjusted results stratified by exposure level, and Zhang included that data.
However, the study also reported adjusted results on an ever/never basis, and she ignored that
data. Id. at 91:2-24. (And then Zhang’s paper misleadingly states that five of the seven studies
reported data adjusted for other pesticides, which implies that it counted Eriksson as a study that
reported adjusted data. But the data from Eriksson that Zhang actually used in the primary meta-
analysis was unadjusted. See Hearing Tr. at 94:21—95:7.) The inclusion of the unadjusted data
seems to have had a significant impact on the results, as evidenced by one of the paper’s
sensitivity analyses. That analysis substituted the Hohenadel study for the McDuffie study. The
Hohenadel study used the same subjects as the McDuffie study, but it adjusted for exposure to
3 Gabriella Andreotti et al., Glyphosate Use and Cancer Incidence in the Agricultural Health
Study, 110 Journal of the National Cancer Institute 1 (2018) (Dkt. No. 1136-1) at 3.
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 7 of 14
malathion, an insecticide. See Hearing Tr. at 103:2—19; see also Zhang Meta-Analysis, Table 6,
Row: “Other”. (The relevant McDuffie data is unadjusted for other occupational exposures.)
With that substitution, Zhang 2gSs & meta-risk ratio” became statistically insignificant. See id.
The issues with unadjusted data are especially important because they tend to undercut
the assumptions underlying Zhang’s entire method. Recall that part of the paper’s justification
for focusing on “high exposure” groups is that they are “less likely to be dominated by
confounding or other biases[.]” Zhang Meta-Analysis § 2.2. But, in this context, exposure to
other pesticides is the major confounder, and people with high exposure to glyphosate (who tend
to be agricultural workers or landscapers) are likely to have more exposure to other pesticides,
not less. See Pretrial Order No. 45 at 24. That means that when Zhang isolated unadjusted, “high-
exposure” data from McDuffie and Eriksson, she may have increased the effect of confounders
on the meta-risk ratio. At the very least, it isn’t likely she reduced their effect.
The upshot of these four points is that the meta-analysis’s methods are impossible to
justify—and that becomes clear when one starts to drill down into the details. The stated purpose
of the meta-analysis is to test the a priori hypothesis by examining what the totality of the
epidemiological evidence shows about NHL incidence “following high cumulative GBH
exposure.” Zhang Meta-Analysis § 3.1. But the study doesn’t really do that. Instead, it slices and
dices the available epidemiological evidence in a way that doesn’t actually isolate high-exposure
data. It takes all of the data from some studies that don’t report exposure levels; it excludes
subjects from Andreotti who were probably as highly exposed as subjects from McDuffie; and it
mixes whole datasets with partial ones, notably using only a small portion of the NHL cases
from the highest-quality study. Finally, the meta-analysis mixes adjusted and unadjusted data,
even choosing the latter in Eriksson even where adjusted data was available. See Hearing Tr. at
81:10-17./ In short, the meta-analysis restricts itself to a more or less arbitrary subset of the
4 This is not to suggest that it would never be appropriate for a meta-analysis to combine
adjusted numbers with unadjusted numbers. But to avoid using adjusted numbers for no apparent
reason is a different matter altogether. And when you combine that with the previous three
problems discussed above, it becomes clear that Zhang’s meta-analysis is simply not an
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 8 of 14
epidemiological data points. And it does so in a way that seems likely to exacerbate, rather than
ameliorate, problems with confounding. Even on its own terms, it is junk science.
B
There’s a second, independent problem. Even if Zhang’s paper could somehow be
considered a reliable meta-analysis, a reliable meta-analysis is not the same thing as a reliable
general causation opinion. In this respect, it matters that Zhang’s entire engagement with the
epidemiological studies is based on her meta-analysis. Accordingly, the right way to think about
Zhang’s presentation is not just to ask whether it amounts to a reliable meta-analysis (although it
must at least be that). Instead, one also has to ask whether the meta-analysis is a reliable
approach to answering the question: “Does Roundup cause NHL in humans?”
To see why these are different questions, consider that the authors of the McDuffie,
Eriksson, or Orsi studies probably have fair claims that their case-control studies were reliably
conducted. Those studies were peer-reviewed and published, and they’ve no doubt been widely
cited in the scientific debates about glyphosate. But that doesn’t mean they would make for
admissible general causation opinions. Imagine, for example, that a plaintiff offered Dr.
McDuffie herself as their sole general causation expert. Imagine further that at the expert
disclosure deadline, McDuffie served her 2001 study, describing it as a statement of her
“opinions in this matter, and the bases for them.” To survive a Daubert motion, it would not be
enough for McDuffie to show that she had reliably conducted a case-control study. Instead, she’d
have to explain why her methods in that study (now functioning as an opinion on general
causation) were a reliable way to decide, using epidemiological evidence, whether glyphosate
did or didn’t cause NHL.
The same is true here. Obviously, the difference between a meta-analysis and a case-
control study is that the former purports to synthesize the results of multiple studies. But that
doesn’t mean that a meta-analysis can necessarily stand in for a holistic general causation
objective, reliable scientific undertaking.
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 9 of 14
opinion, especially where the meta-analysis is designed to examine only a subset of the available
data. Just like the McDuffie study could be useful to epidemiologists without being able to
prove, by itself, that glyphosate causes cancer, a meta-analysis like Zhang’s might (if performed
reliably) offer a useful addition to the scientific literature without being able to prove anything
about general causation. Perhaps Zhang’s meta-analysis is best understood as a statistical
experiment designed to see what can be learned by massaging the data in a certain way. As such,
perhaps it could be useful to experts in some fashion. But here, the meta-analysis has been
offered as Bulone’s only opinion about the epidemiological evidence and general causation. That
means Zhang has opened the study up to the types of questions necessary to determine whether it
reliably analyzes that evidence—not just by the standards that govern epidemiological meta-
analyses, but by the standards that govern general causation testimony in this litigation.
Zhang’s meta-analysis can’t meet those standards. First, and maybe most fundamentally,
the paper doesn’t seem to offer the necessary conclusion about general causation. Zhang
concludes that there is a “compelling link between exposures to [glyphosate-based herbicides]
and increased risk for NHL.” Zhang Meta-Analysis § 9. Similarly, the study’s key figure, the
1.41 meta-risk ratio, is described as underscoring the fact that “exposure to GBHs is associated
with an increased risk of NHL.” Id. § 6. But saying there is a “compelling link” or an
“association” isn’t the same as saying that Roundup is capable of causing NHL in humans—
which is what a plaintiff's general causation evidence must enable the jury to conclude.® See
Pretrial Order No. 45 at 2; cf. also Pretrial Order. No. 263 (Dkt. No. 14432) at 2 (excluding a
5 The other paper that Zhang disclosed as a statement of her opinions, which deals with the
mechanistic evidence on glyphosate, also doesn’t offer an opinion about causation. It concludes:
The totality of evidence from mechanistic studies in human and
animal studies [sic] suggest that glyphosate and its formulations
possess several of the ten key characteristics of carcinogens. . . .
Overall . . . our findings our findings of strong evidence of
glyphosate’s ability to cause genotoxicity, epigenetic alterations,
oxidative stress, chronic inflammation, and endocrine disruption,
as well as its demonstrated perturbation of the gut microbiota
outline several avenues implicated in lymphomagenesis.
Zhang Expert Report (Dkt. No. 18113-1) at 65.
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 10 of 14
specific causation expert who concluded that ee it is possible’ glyphosate contributed to” the
plaintiff's NHL).
Second, because the meta-analysis was published in 2019, Zhang can’t offer an up-to-
date account of the epidemiological evidence on glyphosate and NHL. In the roughly five years
since the meta-analysis was published, further epidemiological studies have come out. Zhang
admitted that if she conducted the meta-analysis today, she would incorporate several of them.
See Hearing Tr. at 104:5—105:21 (“Q: So if you were starting today to do the same study with
your a priori hypothesis, you would use Pahwa instead of McDuffie and De Roos 2003; correct?
... A: That’s correct.”); id. at 110:18-111:12 (Zhang admitting that if she were “doing the same
analysis with the same a priori hypothesis,” she would incorporate updated data from three
studies that post-dated her 2019 meta-analysis). It also seems like, if Zhang applied the same
methods today, the more recent data points would tend to reduce her “meta-risk ratio.” See id. at
109:9—23. At any rate, an expert testifying in 2024 couldn’t reliably opine that glyphosate is
carcinogenic if they only considered epidemiological evidence published before 2019.
Finally, the methodological problems described in the preceding section show not only
that the paper is an unreliable meta-analysis but also that it’s unreliable as a standalone treatment
of the epidemiological evidence. If an expert offering a more conventional Bradford Hill—
oriented general causation opinion arbitrarily limited the scope of the data they examined, elided
the distinction between adjusted and unadjusted data, and simply assumed the existence of a
dose-response relationship, their opinion would be inadmissible. Cf. Pretrial Order No. 288 at 2—
5 (excluding general causation opinion where, among other things, the expert “cherry-picked the
findings of the epidemiology studies, reporting only certain odds ratios—those most favorable to
his ultimate opinion,” and misstated whether the McDuffie and Eriksson studies were adjusted
for other pesticides). The same has to be true here.
To be clear, none of this is to suggest that a well-performed meta-analysis could never
independently support an expert opinion on causation. One could imagine a well-conducted
epidemiological meta-analysis that yielded such a high risk-ratio that no further work would be
10
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 11 of 14
necessary to conclude that a particular substance can cause an illness. It’s just that Zhang’s meta-
analysis can’t support a finding that Roundup causes NHL, even assuming it makes some sort of
contribution in the world of epidemiology.°
Cc
Zhang’s testimony, rather than addressing the problems with her opinion, made it even
less reliable. At the hearing, Zhang struggled to answer basic questions about her meta-analysis
or the handful of studies it incorporated. For example, she couldn’t remember the names of all
six studies that she analyzed. See id. at 48:8-51:17. She couldn’t remember important facts about
how the high-exposure groups were defined in each study. See id. at 51:20-52:20, 55:3-57:20.
She couldn’t remember whether key data points were adjusted or unadjusted. See id. at 93:25—
94:20. She couldn’t remember which studies were used in some of the sensitivity analyses, and
she was therefore unable to explain problems with them that were raised on cross-examination.
See id. at 100:11—102:21. This is clear enough from the transcript of the hearing; it’s even more
obvious from viewing the recording.
Bulone says that Zhang didn’t expect the hearing to get into issues that weren’t raised in
Monsanto’s motion, and that she may have been flustered by a change in the hearing format from
in-person to Zoom. Some allowances can be made for that. But, frankly, nothing can excuse
Zhang’s inability to answer basic and straightforward questions about her own paper and the
epidemiological literature about which she is being offered as an expert. Zhang’s lack of
familiarity with her own opinion is itself a basis to exclude her testimony. See Pretrial Order No.
288 at 4 (excluding general causation opinion in part because the expert “was not able to discuss
his report, the literature, or his opinion without reading directly from his notes or being supplied
the answers by counsel through leading questions or displayed exhibits”).
To the extent that Zhang did address the substantive issues described above, her
° It’s possible to characterize the problems discussed in this section as issues of “fit” rather than
reliability. See Daubert v. Merrell Dow Pharmaceuticals, Inc. (Daubert II), 43 F.3d 1311, 1315
(9th Cir. 1995); Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 597 (1993).
However they’re understood, they require exclusion.
11
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 12 of 14
responses were inadequate. Generally speaking, Zhang defended the paper by saying that its
methods had been used in other meta-analyses, that its methodological choices were all made
clear in the paper, that certain limitations were a function of the small body of epidemiological
evidence, and that the paper’s sensitivity analyses supported its headline “meta-risk ratio”
finding.’ But none of Zhang’s answers responded directly to the question of why, as a
substantive matter, these issues shouldn’t shake confidence in the paper’s methods. See, e.g.,
Hearing Tr. at 65:11-66:23; 71:7—72:8; 82:22-84:22. For example, the fact that the paper clearly
states its selection criteria doesn’t mean that applying them to slice up the epidemiological data
makes for a reliable analysis. In short, Zhang wasn’t able to defend her own study’s methods, let
alone explain why it could serve as an independent analysis of the epidemiological evidence.
D
In his supplemental response, Bulone argues that Monsanto is asking the Court to
impermissibly substitute its judgment about the quality of the meta-analysis for the judgment of
the scientific community. He argues that the methods underlying the meta-analysis (including
most of practices discussed above, like mixing adjusted and unadjusted data) are common in
epidemiological meta-analyses. See Plaintiff's Response to Monsanto’s Supplemental Brief (Dkt.
No. 18351) at 12-20; see also Zhang Declaration (Dkt. No. 18351-2). He says that studies
7 4 word on the sensitivity analyses. Pointing to them was the closest that Zhang came to
substantively defending the study’s methods. But in crucial places, the sensitivity analyses end
up reproducing the problems with Zhang’s method rather than resolving them. Consider the
sensitivity analysis that examined only the three studies that stratified data by exposure level. See
Zhang Meta-Analysis, Table 6, Row: “High level’. Necessarily, that analysis still mixes
unadjusted data from Eriksson and McDuffie with adjusted data from Andreotti, and it still
leaves out subjects from Andreotti that likely had as much exposure as subjects from McDuffie.
Or consider the analysis that purportedly examined only the studies with adjusted data. The
paper says that it considered four such studies. See id., Table 6, Row: “Adjusted”. But only three
of the studies—De Roos (2003), De Roos (2005), and Hardell—reported adjusted, stratified data.
The fourth study included must have been Eriksson, as Zhang acknowledged. See Hearing Tr. at
100:25—101:1. But that means that either the study authors used the unadjusted high-exposure
data from Eriksson in this meta-analysis (which would make it completely bogus) or they used
Eriksson’s ever/never data (which seems unlikely, because that would be inconsistent with the
paper’s method of selecting the high-exposure groups where available). Either way, that
sensitivity analysis doesn’t do what it claims to do. Zhang was unable to explain this discrepancy
and did not know which data was used in that sensitivity analysis. See id. at 100:11—102:19. The
paper itself doesn’t resolve the confusion, either. See Zhang Meta-Analysis, Table 6, n.7.
12
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 13 of 14
employing such methods are frequently published in peer-reviewed journals, including
prestigious ones. See id. And he argues that Zhang’s meta-analysis has all the standard indicia of
reliability under Daubert: it was published in a peer-reviewed journal, it was produced
independently of litigation, Zhang has been invited to present the paper at various scientific
societies, and the paper has been cited many times. See id. at 8-12.
These points can’t make the study’s glaring methodological flaws disappear. It’s true that
the meta-analysis was published in a peer-reviewed journal and prepared before Zhang got
personally involved in any Roundup cases (although the paper seems so results-driven that one
could be forgiven for doubting that its authors were really thinking “independently” of the
nationwide litigation). But while those facts are entitled to consideration, they don’t mean that
blind deference to Zhang is appropriate. There are a vast number of peer-reviewed journals out
there. Pre-publication editorial peer review, just by itself, is far from a guarantee of scientific
reliability. See Valentine v. Pioneer Chlor Alkali Co., 921 F. Supp. 666, 674—76 (D. Nev. 1996)
(distinguishing between “editorial peer review” and “true peer review,” and arguing that the
latter—which includes the scientific reception of an article after publication, efforts to replicate
its findings, etc.—is what’s really important under Daubert).® Plus, part of the point of
centralizing cases like these in an MDL is that the Court, over many years, can develop a fluency
in the relevant scientific literature. Where that fluency helps identify serious problems in an
expert opinion, it doesn’t make sense for the Court to ignore them.
Second, the other meta-analyses that Zhang cites, which also use an “a priori hypothesis”
to isolate high-exposure groups in epidemiological studies, don’t help her as much as she would
like. These studies may share some of the basic features of Zhang’s paper, but that isn’t enough
to save a study so riddled with problems. The other papers also aren’t methodologically
8 See also Effie Chan, Note, The Brave New World of Daubert: True Peer Review, Editorial Peer
Review, and Scientific Validity, 70 N.Y.U. Law Review 100 (1995); Smith, Peer Review: A
Flawed Process at the Heart of Science and Journals, 99 Journal of the Royal Society of
Medicine 178 (2006) (arguing that there are many shortcomings of pre-publication peer review,
including that it is not a good system for catching errors, is prone to publication bias, and often
operates inconsistently).
13
Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 14 of 14
comparable to Zhang’s in important ways. For one thing, all of the cited meta-analyses examined
at least twice as many epidemiological studies as Zhang’s glyphosate meta-analysis, and often
many more. Perhaps if Zhang had a lot more studies to feed into her meta-analysis, she might
have a defensible claim to be roughly capturing “high-exposure” data points. As it is, her
analysis does no such thing.
Finally, Bulone’s arguments do nothing to address the problems with using Zhang’s
meta-analysis as a standalone analysis of the epidemiological evidence. One can’t assume that
peer reviewers were vetting a study’s ability to play a certain role in an expert opinion, and
there’s no reason to think that was the case here. In short, simply intoning that a paper was peer-
reviewed and that it was (apparently) produced before the author was thinking about getting
involved in litigation isn’t enough to compensate for methodological problems as glaring as
those described here. And it certainly isn’t enough to show that Zhang’s paper can support a
reliable general causation opinion.
iil
Monsanto filed a motion for summary judgment in connection with its motion to exclude
Dr. Braunstein, Bulone’s specific causation expert. See Dkt. No. 17598. That motion mentioned
the necessity of summary judgment in the absence of a specific causation expert. Id. at 9-10. The
absence of an admissible general causation opinion requires the same result, as Bulone’s counsel
conceded. Hearing Tr. at 87:14—25. Accordingly, summary judgment is granted for Monsanto on
the grounds that Bulone lacks any admissible evidence of general causation.
IT IS SO ORDERED.
Loe
Dated: June 20, 2024
VINCE CHHABRIA
United States District Judge
14
EXHIBIT 2
Accepted Manuscript
PUSAN
Title: Exposure to Glyphosate-Based Herbicides and Risk for
Non-Hodgkin Lymphoma: A Meta-Analysis and Supporting
Evidence
Authors: Luoping Zhang, Iemaan Rana, Rachel M. Shaffer, =
Emanuela Taioli, Lianne Sheppard
PII: $1383-5742(18)30088-7
DOI: https://doi.org/10.1016/j.mrrev.2019.02.001
Reference: MUTREV 8261
To appear in: Mutation Research
Received date: 22 September 2018
Revised date: 2 February 2019
Accepted date: 5 February 2019
Please cite this article as: Zhang L, Rana I, Shaffer RM, Taioli E, Sheppard L, Exposure
to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma: A Meta-
Analysis and Supporting Evidence, Mutation Research-Reviews in Mutation Research
(2019), https://doi.org/10.1016/j.mrrev.2019.02.001
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma:
A Meta-Analysis and Supporting Evidence
Luoping Zhang, lemaan Rana*, Rachel M. Shaffer®, Emanuela Taioli*, Lianne
Sheppard>4
4 Division of Environmental Health Sciences, School of Public Health, University of
California Berkeley, Berkeley, USA
> Department of Environmental and Occupational Health Sciences, University of
Washington, Seattle, USA
© Institute for Translational Epidemiology and Department of Population Health Science
and Policy, Icahn School of Medicine at Mount Sinai, New York, USA
4 Department of Biostatistics, University of Washington, Seattle, USA
*Corresponding Author:
Luoping Zhang, PhD
Division of Environmental Health Sciences
School of Public Health
2121 Berkeley Way West, #5117
Berkeley, CA 94720-7360
Tel: (510) 643-5189
Fax: (510) 642-0427
Email: luoping@berkeley.edu
Abstract
Glyphosate is the most widely used broad-spectrum systemic herbicide in the world.
Recent evaluations of the carcinogenic potential of glyphosate-based herbicides (GBHs)
by various regional, national, and international agencies have engendered controversy.
We investigated whether there was an association between high cumulative exposures
to GBHs and increased risk of non-Hodgkin lymphoma (NHL) in humans. We conducted
anew meta-analysis that included the most recent update of the Agricultural Health Study
(AHS) cohort published in 2018 along with five case-control studies. Using the highest
1
exposure groups when available in each study, we report the overall meta-relative risk
(meta-RR) of NHL in GBH-exposed individuals was increased by 41% (meta-RR = 1.41,
95% Cl, confidence interval: 1.13—-1.75). For comparison, we also performed a secondary
meta-analysis using high-exposure groups with the earlier AHS (2005), and we
determined a meta-RR for NHL of 1.45 (95% Cl: 1.11-1.91), which was higher than the
meta-RRs reported previously. Multiple sensitivity tests conducted to assess the validity
of our findings did not reveal meaningful differences from our primary estimated meta-
RR. To contextualize our findings of an increased NHL risk in individuals with high GBH
exposure, we reviewed available animal and mechanistic studies, which provided
supporting evidence for the carcinogenic potential of GBH. We documented further
support from studies of malignant lymphoma incidence in mice treated with pure
glyphosate, as well as potential links between GBH exposure and immunosuppression,
endocrine disruption, and genetic alterations that are commonly associated with NHL.
Overall, in accordance with evidence from experimental animal and mechanistic studies,
our current meta-analysis of human epidemiological studies suggests a compelling link
between exposures to GBHs and increased risk for NHL.
(273/300 words)
Abbreviations: AHS, Agricultural Health Study; c-NHEJ, canonical non-homologous end
joining pathway; Cl, confidence interval; EDC, endocrine disrupting chemical; EFSA,
European Food Safety Authority; EPA, Environmental Protection Agency; ETS,
environmental tobacco smoke; GBHs, glyphosate-based herbicides; IARC, International
Agency for Research on Cancer; IFN-y, interferon gamma; IL-2, Interleukin-2; JMPR,
2
Joint Meeting on Pesticide Residues by the Food and Agriculture Organization of the
United Nations and World Health Organization; meta-RR, meta-analysis relative risk;
mg/kg/day, milligrams per kilogram per day; MM, multiple myeloma; NHL, non-Hodgkin
lymphoma; OR, odds ratio; ppm, parts per million; PRISMA, Preferred Reporting Items
for Systematic Reviews and Meta-Analysis; RR, relative risk.
Keywords: Glyphosate, pesticide, Roundup, Ranger Pro, carcinogenesis, and meta-
analysis.
1. Background
1.1 Global Usage of Glyphosate-Based Herbicides
Glyphosate is a highly effective broad spectrum herbicide that is typically applied
in mixtures known as glyphosate-based herbicides (GBHs) and commonly sold under the
trade names of Roundup® and Ranger Pro®. Use of GBHs has increased dramatically
worldwide in recent decades. In the United States alone, usage increased nearly sixteen-
fold between 1992 and 2009 [1]. Most of this increase occurred after the introduction of
genetically modified glyphosate-resistant “Roundup-ready” crops in 1996 [1]. In addition,
there have been significant changes in usage. In particular, the practice of applying GBHs
to crops shortly before harvest, so-called “green burndown,” began in the early 2000s to
speed up their desiccation; as a consequence, crops are likely to have higher GBH
residues [2]. By the mid-2000s, green burndown became widespread, and regulatory
agencies responded by increasing the permissible residue levels for GBHs [3, 4].
1.2 Ubiquitous Exposure in Humans
Glyphosate and its metabolites persist in food [5-7], water [8], and dust [9],
potentially indicating that everyone may be exposed ubiquitously. Non-occupational
exposures occur primarily through consumption of contaminated food, but may also occur
through contact with contaminated soil [9], dust [9] and by drinking or bathing in
contaminated water [8]. In plants, glyphosate may be absorbed and transported to parts
used for food; thus, it has been detected in fish [5], berries [6], vegetables, baby formula
[7], and grains [10], and its use as a crop desiccant significantly increases residues. GBH
residues in food persist long after initial treatment and are not lost during baking.
Limited data exist on internal glyphosate levels among GBH-exposed individuals
[11]. Average urinary glyphosate levels among
Preview
ELECTRONICALLY FILED Superior Court of California County of Santa Barbara Darrel E. Parker, Executive Officer Jessica L. Grant (SBN: 178138) 7/1/2024 1:24 PM John K. Sherk (SBN: 295838) By: Narzralli Baksh , Deputy Kara M. Flageollet (SBN: 347646) SHOOK, HARDY & BACON L.L.P. 555 Mission Street, Suite 2300 San Francisco, California 94105 Tel: (415) 544-1900 | Fax: (415) 391-0281 jgrant@shb.com jsherk@shb.com kflageollet@shb.com Jason Zager (admitted pro hac vice) SHOOK, HARDY &BACON L.L.P. 2555 Grand Boulevard Kansas City, MO 64108 Tel: (816) 474-6550 | Fax: (816) 421-5547 jzager@shb.com 10 Attorneys for Defendant MONSANTO COMPANY 11 12 13 SUPERIOR COURT OF THE STATE OF CALIFORNIA 14 FOR THE COUNTY OF SANTA BARBARA 15 16 MARK JON ROGERS, Case No. 23CV00945 17 Plaintiff, DECLARATION OF JESSICA GRANT IN SUPPORT OF MONSANTO COMPANY’S 18 Vv. OPPOSITION TO PLAINTIFF’S MOTION IN LIMINE NO. 27 19 MONSANTO COMPANY, Date: July 10, 2024 20 Defendant. Time: 10:00 am Dept.: 3 21 Judge: Honorable Thomas Anderle 22 Filed concurrently herewith: Monsanto Company’s Opposition to Plaintiff’s Motion In 23 Limine No. 27 24 25 26 27 28 DECLARATION IN SUPPORT OF MONSANTO’S OPPOSITION TO PLAINTIFF’S MOTION IN LIMINE NO. 27 I, Jessica Grant, hereby declare as follows: 1 I am a partner at the law firm of Shook Hardy & Bacon, counsel for defendant Monsanto Company (“Monsanto”) in the above-captioned action. I am over eighteen years of age and am fully competent to make this declaration in support of Defendant’s Opposition to Plaintiff’s Motion in Limine no. 27. I also make this declaration based on my personal knowledge and if called as a witness, I could and would competently testify to the matters stated herein. 2 A true and correct copy of the June 20, 2024 order entitled Order Granting Motion to Exclude Dr. Luoping Zhang and Granting Summary Judgment To Monsanto, in In re Roundup Prods. Liab. Litig., No. 16-md-02741 is attached hereto as “Exhibit 1.” 10 3 A true and correct copy of Zhang, et. al., Exposure to glyphosate-based herbicides 11 and risk for non-Hodgkin lymphoma: A meta-analysis and supporting evidence (2019), is attached 12 hereto as “Exhibit 2.” 13 4. A true and correct copy of the January 6, 2020 EPA Memorandum entitled 14 Glyphosate: Epidemiology Review of Zhang et al. (2019) and Leon et al. (2019) publications for 15 Response to Comments on the Proposed Interim Decision is attached hereto as “Exhibit 3.” 16 5. A true and correct copy of excerpts from Dr. Kristan Aronson’s December 15, 2023 17 Expert Report entitled General cancer causation for exposure to glyphosate and glyphosate 18 formulations in the case styled Brazell v. Monsanto, No. 1922-CC11327 (Mo. Cir. Ct., St. Louis 19 Cnty.) is attached hereto as “Exhibit 4.” 20 I declare under penalty of perjury under the laws of the State of California that the foregoing 21 is true and correct. 22 Executed this 1" day of July, 2024 in San Francisco, California. 23 Respectfully submitted, SHOOK, HARDY & BACON L.L.P. 24 By: A 25 wo ca L. Grant 26 neys for Defendant 27 MONSANTO COMPANY 28 DECLARATION IN SUPPORT OF MONSANTO’S OPPOSITION TO PLAINTIFF’S MOTION IN LIMINE NO. 27 EXHIBIT 1 Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 1 of 14 UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA IN RE: ROUNDUP PRODUCTS MDL No. 2741 LIABILITY LITIGATION Case No. 16-md-02741-VC This document relates to: ORDER GRANTING MOTION TO EXCLUDE DR. LUOPING ZHANG Bulone v. Monsanto Co., Case No. 20-cv- AND GRANTING SUMMARY 03719-VC JUDGMENT TO MONSANTO Re: Dkt. No. 18022 Monsanto’s motion to exclude the testimony of Dr. Luoping Zhang is granted. This order assumes familiarity with the Court’s prior orders on general and specific causation and the Ninth Circuit’s opinion in Hardeman. See generally, e.g., In re Roundup Products Liability Litigation, 390 F. Supp. 3d 1102 (N.D. Cal. 2018) (Pretrial Order No. 45, Dkt. No. 1596); In re Roundup Products Liability Litigation, 358 F. Supp. 3d 956 (N.D. Cal. 2019) (Pretrial Order No. 85, Dkt. No. 2799); Hardeman v. Monsanto Company, 997 F.3d 941 (9th Cir. 2021). It also assumes familiarity with the record from the Daubert hearing on Zhang that took place on April 24, 2024. See Hearing Tr. (Dkt. No. 18297). Zhang’s presentation is based on a meta-analysis of the epidemiological literature that she published in 2019. There are several issues with her opinion, each of which is an independent ground for exclusion. First, Zhang’s meta-analysis is junk science. It has deep methodological problems, not the least of which is that it doesn’t achieve its stated aim of examining only “highly exposed” individuals. Instead, it examines an arbitrary selection of the available data. Second, Zhang’s only engagement with the epidemiological literature is her meta-analysis, and Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 2 of 14 her meta-analysis can’t substitute for a reliable general causation opinion. In other words, even if the meta-analysis were somehow useful in the field of epidemiology, it still couldn’t reliably serve as a litigation expert’s sole engagement with the epidemiological literature. Both of these problems are exacerbated by the fact that Zhang struggled to answer basic questions about her own paper, and the studies discussed in that paper. Nor can either problem be cured by intoning that the paper was peer-reviewed and developed before an expert became personally involved in litigation—those facts are relevant, but a court can’t wave junk science through the Daubert gate simply because it survived some prepublication peer-review process. For each of those reasons, Zhang’s testimony must be excluded. And without Zhang, Bulone has no admissible general causation opinion, so Monsanto is entitled to summary judgment. I Dr. Zhang is a biochemist and a professor emeritus of toxicology at the University of California, Berkeley. She has been designated to testify as a general causation expert in the Bulone case. Zhang is Bulone’s sole expert on general causation. Zhang’s presentation is different from those of past general causation experts in that the only opinions she disclosed are those contained in two journal publications that she co-authored. See Zhang Expert Report (Dkt. No. 18113-1) at 2 (stating that her “opinions in this matter, and the bases for them, are set out in the following,” and citing two articles, without further comment); see also Order Denying Motion to Strike Expert Report of Luoping Zhang, Ph.D. (Dkt. No. 17743). In effect, these two publications are Zhang’s expert report. Only one ofthese studies, Zhang’s 2019 meta-analysis, grapples with the epidemiological evidence.! That study is the focus of Monsanto’s attack on Zhang’s presentation. Because an admissible expert opinion on general causation is necessary to survive summary judgment, and because a complete general causation opinion must address the epidemiological evidence, Bulone’s case will stand or fall on ' See Zhang et al., Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma: A Meta-Analysis and Supporting Evidence, 781 Mutation Research 186 (2019) [“Zhang Meta-Analysis”] (Dkt. No. 18351-4). Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 3 of 14 whether Zhang’s 2019 meta-analysis is an admissible treatment of the epidemiological evidence. See Pretrial Order No. 45 at 13 (“As the parties acknowledge, epidemiology is central to the general causation inquiry, and where such evidence exists, it must be addressed by the experts.”) (citing Norris v. Baxter Healthcare Corp., 397 F.3d 878, 882 (10th Cir. 2005)); see also id. at 6. Zhang’s 2019 paper is a meta-analysis of six epidemiological studies of varying sizes, methods, and quality: McDuffie (2001), Hardell (2002), De Roos (2003), Eriksson (2008), Orsi (2009), and Andreotti (2018). The paper’s approach is based on the “a priori hypothesis” that, if glyphosate were carcinogenic, people who had been exposed to more glyphosate would be more likely to show an increased risk of NHL. See Zhang Meta-Analysis § 2.3. Accordingly, the authors purported to conduct the meta-analysis by isolating the data from the highest-exposure groups, where available, across each of the studies and analyzing them using a fixed-effect statistical model. However, not all of the epidemiological studies that Zhang analyzed broke out their data into groups based on exposure levels. Some of the underlying studies only reported their results on an ever-exposed/never-exposed basis. For those studies, the meta-analysis simply used all of the data. See id. § 2.6. The results of the meta-analysis are presented in the form of a “meta-risk ratio” of 1.41, with a 95% confidence interval of 1.13-1.75. Id. § 3.1. Based on this top-line finding, as well as summary discussions of other kinds of animal studies and mechanistic studies, the meta-analysis concludes that “[t]he overall evidence from human, animal, and mechanistic studies presented here supports a compelling link between exposures to [glyphosate-based herbicides] and increased risk for NHL.” Id. § 9. Monsanto’s motion to exclude Zhang focused on an EPA critique of her meta-analysis produced during the agency’s review of glyphosate’s registration under FIFRA. See Hsu Decl. Ex. D, EPA Review (Dkt. No. 18022-5) at 8. The Daubert hearing brought some additional issues to light, and the Court ordered further briefing. Bulone doesn’t dispute that the Court can consider the issues raised at the hearing in ruling on Zhang, particularly given the supplemental briefing. See Miller v. Baker Implement Co., 439 F.3d 407, 413 (8th Cir. 2006) (“A district Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 4 of 14 court’s Daubert inquiry need not take any specific form, and its sua sponte consideration of the admissibility of expert testimony is permissible so long as the court has an adequate record on which to base its ruling.”) (citing Kirstein v. Parks Corp., 159 F.3d 1065, 1067 (7th Cir. 1998)) Il In this litigation, the general causation question is “whether a reasonable jury could conclude by a preponderance of the evidence that glyphosate can cause NHL at exposure levels people realistically could have experienced.” Pretrial Order No. 45 at 2. Presenting an admissible general causation opinion requires “offering independent and relatively comprehensive opinions that the epidemiological and other evidence demonstrates glyphosate causes NHL in some people who are exposed to it.” Id. at 3. The admissibility of any opinion that Roundup is capable of causing NHL has always been a “very close question.” Id. at 1. Reliably offering such an opinion requires serious engagement with the relevant literature: general causation experts must be able to “assess whether a study is credible, to explain why they relied on one study more than another, and to articulate how they reached their conclusion in the face of conflicting evidence.” Pretrial Order No. 288 (Dkt. No. 17504) at 5; see also Pretrial Order No. 45 at 35. A The most fundamental problem with Zhang’s testimony is that her meta-analysis was not reliably performed. It has several methodological problems that, when taken together, make its analysis of the epidemiological literature indefensible. First, the meta-analysis was based on the hypothesis that, if glyphosate caused NHL, then people with the highest levels of exposure would be most likely to get sick. Zhang and her colleagues justified this hypothesis with reference to “the understanding that higher and longer cumulative exposures are likely to yield higher risk estimates, given the nature of cancer development.” Zhang Meta-Analysis § 2.2; see also Hearing Tr. at 122:3—15. They also argued that data from high-exposure groups are less likely to be affected by confounders or to be diluted by low-exposure data points. Zhang Meta-Analysis § 2.2. But the meta-analysis doesn’t grapple with the fact that the Andreotti study, which it says is the highest-quality epidemiological study Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 5 of 14 available, doesn’t show any kind of dose-response relationship between glyphosate exposure and NHL. See Pretrial Order No. 45 at 26 (discussing Andreotti, which reported risk ratios below 1 for every exposure quartile, with the highest risk ratio reported for the third-highest quartile); Zhang Meta-Analysis, Table 2 (ranking the quality of the studies included in the meta-analysis and listing Andreotti as the best study); see also Pretrial Order No. 45 at 40 (discussing the ambiguous evidence regarding dose response issue). Second, in applying its hypothesis, the meta-analysis ends up mixing different types of studies and different types of data in a way that ultimately seems hard to justify. Only half of the studies that Zhang considered broke out their results according to the subjects’ exposure levels: Eriksson, McDuffie, and Andreotti. See Hearing Tr. at 50:6—9. For the remaining three studies— which reported their results on an ever-exposed/never-exposed basis—the meta-analysis simply incorporated all of the studies’ data. So the meta-analysis used only a portion of the data from three of the studies, and the entirety of the data from the other studies. Id. at 59:9—-12 (“[The Court]: So you took portions of the group of people studied in Eriksson and McDuffie and Andreotti and put them together with all of the people studied in the other three studies? A: Yeah.”). One result of this uneven mixing of studies is that the meta-analysis, in effect, artificially limited its use of data from Andreotti, the study Zhang said is the strongest and the one that showed no statistically significant association between glyphosate and NHL for any level of exposure. This issue was noted in the EPA’s assessment of Zhang’s study. See EPA Review at 8. Using a small portion of Andreotti’s data might be defensible if the meta-analysis really took only a portion of the data from every study. But instead, the study methodologically de-emphasizes high-quality data that would tend to reduce the “meta-risk ratio,” while using the full datasets from most of the lower-quality studies.” Third, the studies that did stratify their results by exposure level used different exposure ? The treatment of the Andreotti data is especially misleading given that the meta-analysis touts itself as “the first meta-analysis to include the most recently updated Agricultural Health Study (AHS) cohort,” ie., the Andreotti data. Zhang Meta-Analysis § 1.3; see also Hearing Tr. at 49:21-50:5. Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 6 of 14 cut-offs and measured exposure differently. As a result, it’s likely that some people excluded by the meta-analysis had as much or more exposure than people who were included. This makes the method even more arbitrary. For example, the high-exposure category in McDuffie included people who reported at least two exposure days per year. Hearing Tr. at 55:3-7. If that cutoff were applied to the Andreotti data, it would probably capture many of the subjects from that study’s third quartile (38.74-108.4 cumulative lifetime days of exposure) and even some from the second quartile (13.75-38.74 cumulative lifetime days of exposure).? But these people were uniformly excluded from Zhang’s analysis. Here again, the meta-analysis’s application of its high-exposure hypothesis didn’t, in practice, result in the inclusion of only highly exposed people. Instead, it resulted in the arbitrary exclusion of data from Andreotti and the over- representation of unadjusted data from the case-control studies. That leads to the fourth point: the meta-analysis mixes results adjusted for the use of other pesticides with unadjusted results. See Hearing Tr. at 80:8-81:17. Indeed, the meta- analysis’s selection criteria sometimes led Zhang to choose unadjusted data even when adjusted data was available, which seems inexcusable. That is what Zhang did with the Eriksson study: Eriksson reported unadjusted results stratified by exposure level, and Zhang included that data. However, the study also reported adjusted results on an ever/never basis, and she ignored that data. Id. at 91:2-24. (And then Zhang’s paper misleadingly states that five of the seven studies reported data adjusted for other pesticides, which implies that it counted Eriksson as a study that reported adjusted data. But the data from Eriksson that Zhang actually used in the primary meta- analysis was unadjusted. See Hearing Tr. at 94:21—95:7.) The inclusion of the unadjusted data seems to have had a significant impact on the results, as evidenced by one of the paper’s sensitivity analyses. That analysis substituted the Hohenadel study for the McDuffie study. The Hohenadel study used the same subjects as the McDuffie study, but it adjusted for exposure to 3 Gabriella Andreotti et al., Glyphosate Use and Cancer Incidence in the Agricultural Health Study, 110 Journal of the National Cancer Institute 1 (2018) (Dkt. No. 1136-1) at 3. Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 7 of 14 malathion, an insecticide. See Hearing Tr. at 103:2—19; see also Zhang Meta-Analysis, Table 6, Row: “Other”. (The relevant McDuffie data is unadjusted for other occupational exposures.) With that substitution, Zhang 2gSs & meta-risk ratio” became statistically insignificant. See id. The issues with unadjusted data are especially important because they tend to undercut the assumptions underlying Zhang’s entire method. Recall that part of the paper’s justification for focusing on “high exposure” groups is that they are “less likely to be dominated by confounding or other biases[.]” Zhang Meta-Analysis § 2.2. But, in this context, exposure to other pesticides is the major confounder, and people with high exposure to glyphosate (who tend to be agricultural workers or landscapers) are likely to have more exposure to other pesticides, not less. See Pretrial Order No. 45 at 24. That means that when Zhang isolated unadjusted, “high- exposure” data from McDuffie and Eriksson, she may have increased the effect of confounders on the meta-risk ratio. At the very least, it isn’t likely she reduced their effect. The upshot of these four points is that the meta-analysis’s methods are impossible to justify—and that becomes clear when one starts to drill down into the details. The stated purpose of the meta-analysis is to test the a priori hypothesis by examining what the totality of the epidemiological evidence shows about NHL incidence “following high cumulative GBH exposure.” Zhang Meta-Analysis § 3.1. But the study doesn’t really do that. Instead, it slices and dices the available epidemiological evidence in a way that doesn’t actually isolate high-exposure data. It takes all of the data from some studies that don’t report exposure levels; it excludes subjects from Andreotti who were probably as highly exposed as subjects from McDuffie; and it mixes whole datasets with partial ones, notably using only a small portion of the NHL cases from the highest-quality study. Finally, the meta-analysis mixes adjusted and unadjusted data, even choosing the latter in Eriksson even where adjusted data was available. See Hearing Tr. at 81:10-17./ In short, the meta-analysis restricts itself to a more or less arbitrary subset of the 4 This is not to suggest that it would never be appropriate for a meta-analysis to combine adjusted numbers with unadjusted numbers. But to avoid using adjusted numbers for no apparent reason is a different matter altogether. And when you combine that with the previous three problems discussed above, it becomes clear that Zhang’s meta-analysis is simply not an Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 8 of 14 epidemiological data points. And it does so in a way that seems likely to exacerbate, rather than ameliorate, problems with confounding. Even on its own terms, it is junk science. B There’s a second, independent problem. Even if Zhang’s paper could somehow be considered a reliable meta-analysis, a reliable meta-analysis is not the same thing as a reliable general causation opinion. In this respect, it matters that Zhang’s entire engagement with the epidemiological studies is based on her meta-analysis. Accordingly, the right way to think about Zhang’s presentation is not just to ask whether it amounts to a reliable meta-analysis (although it must at least be that). Instead, one also has to ask whether the meta-analysis is a reliable approach to answering the question: “Does Roundup cause NHL in humans?” To see why these are different questions, consider that the authors of the McDuffie, Eriksson, or Orsi studies probably have fair claims that their case-control studies were reliably conducted. Those studies were peer-reviewed and published, and they’ve no doubt been widely cited in the scientific debates about glyphosate. But that doesn’t mean they would make for admissible general causation opinions. Imagine, for example, that a plaintiff offered Dr. McDuffie herself as their sole general causation expert. Imagine further that at the expert disclosure deadline, McDuffie served her 2001 study, describing it as a statement of her “opinions in this matter, and the bases for them.” To survive a Daubert motion, it would not be enough for McDuffie to show that she had reliably conducted a case-control study. Instead, she’d have to explain why her methods in that study (now functioning as an opinion on general causation) were a reliable way to decide, using epidemiological evidence, whether glyphosate did or didn’t cause NHL. The same is true here. Obviously, the difference between a meta-analysis and a case- control study is that the former purports to synthesize the results of multiple studies. But that doesn’t mean that a meta-analysis can necessarily stand in for a holistic general causation objective, reliable scientific undertaking. Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 9 of 14 opinion, especially where the meta-analysis is designed to examine only a subset of the available data. Just like the McDuffie study could be useful to epidemiologists without being able to prove, by itself, that glyphosate causes cancer, a meta-analysis like Zhang’s might (if performed reliably) offer a useful addition to the scientific literature without being able to prove anything about general causation. Perhaps Zhang’s meta-analysis is best understood as a statistical experiment designed to see what can be learned by massaging the data in a certain way. As such, perhaps it could be useful to experts in some fashion. But here, the meta-analysis has been offered as Bulone’s only opinion about the epidemiological evidence and general causation. That means Zhang has opened the study up to the types of questions necessary to determine whether it reliably analyzes that evidence—not just by the standards that govern epidemiological meta- analyses, but by the standards that govern general causation testimony in this litigation. Zhang’s meta-analysis can’t meet those standards. First, and maybe most fundamentally, the paper doesn’t seem to offer the necessary conclusion about general causation. Zhang concludes that there is a “compelling link between exposures to [glyphosate-based herbicides] and increased risk for NHL.” Zhang Meta-Analysis § 9. Similarly, the study’s key figure, the 1.41 meta-risk ratio, is described as underscoring the fact that “exposure to GBHs is associated with an increased risk of NHL.” Id. § 6. But saying there is a “compelling link” or an “association” isn’t the same as saying that Roundup is capable of causing NHL in humans— which is what a plaintiff's general causation evidence must enable the jury to conclude.® See Pretrial Order No. 45 at 2; cf. also Pretrial Order. No. 263 (Dkt. No. 14432) at 2 (excluding a 5 The other paper that Zhang disclosed as a statement of her opinions, which deals with the mechanistic evidence on glyphosate, also doesn’t offer an opinion about causation. It concludes: The totality of evidence from mechanistic studies in human and animal studies [sic] suggest that glyphosate and its formulations possess several of the ten key characteristics of carcinogens. . . . Overall . . . our findings our findings of strong evidence of glyphosate’s ability to cause genotoxicity, epigenetic alterations, oxidative stress, chronic inflammation, and endocrine disruption, as well as its demonstrated perturbation of the gut microbiota outline several avenues implicated in lymphomagenesis. Zhang Expert Report (Dkt. No. 18113-1) at 65. Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 10 of 14 specific causation expert who concluded that ee it is possible’ glyphosate contributed to” the plaintiff's NHL). Second, because the meta-analysis was published in 2019, Zhang can’t offer an up-to- date account of the epidemiological evidence on glyphosate and NHL. In the roughly five years since the meta-analysis was published, further epidemiological studies have come out. Zhang admitted that if she conducted the meta-analysis today, she would incorporate several of them. See Hearing Tr. at 104:5—105:21 (“Q: So if you were starting today to do the same study with your a priori hypothesis, you would use Pahwa instead of McDuffie and De Roos 2003; correct? ... A: That’s correct.”); id. at 110:18-111:12 (Zhang admitting that if she were “doing the same analysis with the same a priori hypothesis,” she would incorporate updated data from three studies that post-dated her 2019 meta-analysis). It also seems like, if Zhang applied the same methods today, the more recent data points would tend to reduce her “meta-risk ratio.” See id. at 109:9—23. At any rate, an expert testifying in 2024 couldn’t reliably opine that glyphosate is carcinogenic if they only considered epidemiological evidence published before 2019. Finally, the methodological problems described in the preceding section show not only that the paper is an unreliable meta-analysis but also that it’s unreliable as a standalone treatment of the epidemiological evidence. If an expert offering a more conventional Bradford Hill— oriented general causation opinion arbitrarily limited the scope of the data they examined, elided the distinction between adjusted and unadjusted data, and simply assumed the existence of a dose-response relationship, their opinion would be inadmissible. Cf. Pretrial Order No. 288 at 2— 5 (excluding general causation opinion where, among other things, the expert “cherry-picked the findings of the epidemiology studies, reporting only certain odds ratios—those most favorable to his ultimate opinion,” and misstated whether the McDuffie and Eriksson studies were adjusted for other pesticides). The same has to be true here. To be clear, none of this is to suggest that a well-performed meta-analysis could never independently support an expert opinion on causation. One could imagine a well-conducted epidemiological meta-analysis that yielded such a high risk-ratio that no further work would be 10 Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 11 of 14 necessary to conclude that a particular substance can cause an illness. It’s just that Zhang’s meta- analysis can’t support a finding that Roundup causes NHL, even assuming it makes some sort of contribution in the world of epidemiology.° Cc Zhang’s testimony, rather than addressing the problems with her opinion, made it even less reliable. At the hearing, Zhang struggled to answer basic questions about her meta-analysis or the handful of studies it incorporated. For example, she couldn’t remember the names of all six studies that she analyzed. See id. at 48:8-51:17. She couldn’t remember important facts about how the high-exposure groups were defined in each study. See id. at 51:20-52:20, 55:3-57:20. She couldn’t remember whether key data points were adjusted or unadjusted. See id. at 93:25— 94:20. She couldn’t remember which studies were used in some of the sensitivity analyses, and she was therefore unable to explain problems with them that were raised on cross-examination. See id. at 100:11—102:21. This is clear enough from the transcript of the hearing; it’s even more obvious from viewing the recording. Bulone says that Zhang didn’t expect the hearing to get into issues that weren’t raised in Monsanto’s motion, and that she may have been flustered by a change in the hearing format from in-person to Zoom. Some allowances can be made for that. But, frankly, nothing can excuse Zhang’s inability to answer basic and straightforward questions about her own paper and the epidemiological literature about which she is being offered as an expert. Zhang’s lack of familiarity with her own opinion is itself a basis to exclude her testimony. See Pretrial Order No. 288 at 4 (excluding general causation opinion in part because the expert “was not able to discuss his report, the literature, or his opinion without reading directly from his notes or being supplied the answers by counsel through leading questions or displayed exhibits”). To the extent that Zhang did address the substantive issues described above, her ° It’s possible to characterize the problems discussed in this section as issues of “fit” rather than reliability. See Daubert v. Merrell Dow Pharmaceuticals, Inc. (Daubert II), 43 F.3d 1311, 1315 (9th Cir. 1995); Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 597 (1993). However they’re understood, they require exclusion. 11 Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 12 of 14 responses were inadequate. Generally speaking, Zhang defended the paper by saying that its methods had been used in other meta-analyses, that its methodological choices were all made clear in the paper, that certain limitations were a function of the small body of epidemiological evidence, and that the paper’s sensitivity analyses supported its headline “meta-risk ratio” finding.’ But none of Zhang’s answers responded directly to the question of why, as a substantive matter, these issues shouldn’t shake confidence in the paper’s methods. See, e.g., Hearing Tr. at 65:11-66:23; 71:7—72:8; 82:22-84:22. For example, the fact that the paper clearly states its selection criteria doesn’t mean that applying them to slice up the epidemiological data makes for a reliable analysis. In short, Zhang wasn’t able to defend her own study’s methods, let alone explain why it could serve as an independent analysis of the epidemiological evidence. D In his supplemental response, Bulone argues that Monsanto is asking the Court to impermissibly substitute its judgment about the quality of the meta-analysis for the judgment of the scientific community. He argues that the methods underlying the meta-analysis (including most of practices discussed above, like mixing adjusted and unadjusted data) are common in epidemiological meta-analyses. See Plaintiff's Response to Monsanto’s Supplemental Brief (Dkt. No. 18351) at 12-20; see also Zhang Declaration (Dkt. No. 18351-2). He says that studies 7 4 word on the sensitivity analyses. Pointing to them was the closest that Zhang came to substantively defending the study’s methods. But in crucial places, the sensitivity analyses end up reproducing the problems with Zhang’s method rather than resolving them. Consider the sensitivity analysis that examined only the three studies that stratified data by exposure level. See Zhang Meta-Analysis, Table 6, Row: “High level’. Necessarily, that analysis still mixes unadjusted data from Eriksson and McDuffie with adjusted data from Andreotti, and it still leaves out subjects from Andreotti that likely had as much exposure as subjects from McDuffie. Or consider the analysis that purportedly examined only the studies with adjusted data. The paper says that it considered four such studies. See id., Table 6, Row: “Adjusted”. But only three of the studies—De Roos (2003), De Roos (2005), and Hardell—reported adjusted, stratified data. The fourth study included must have been Eriksson, as Zhang acknowledged. See Hearing Tr. at 100:25—101:1. But that means that either the study authors used the unadjusted high-exposure data from Eriksson in this meta-analysis (which would make it completely bogus) or they used Eriksson’s ever/never data (which seems unlikely, because that would be inconsistent with the paper’s method of selecting the high-exposure groups where available). Either way, that sensitivity analysis doesn’t do what it claims to do. Zhang was unable to explain this discrepancy and did not know which data was used in that sensitivity analysis. See id. at 100:11—102:19. The paper itself doesn’t resolve the confusion, either. See Zhang Meta-Analysis, Table 6, n.7. 12 Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 13 of 14 employing such methods are frequently published in peer-reviewed journals, including prestigious ones. See id. And he argues that Zhang’s meta-analysis has all the standard indicia of reliability under Daubert: it was published in a peer-reviewed journal, it was produced independently of litigation, Zhang has been invited to present the paper at various scientific societies, and the paper has been cited many times. See id. at 8-12. These points can’t make the study’s glaring methodological flaws disappear. It’s true that the meta-analysis was published in a peer-reviewed journal and prepared before Zhang got personally involved in any Roundup cases (although the paper seems so results-driven that one could be forgiven for doubting that its authors were really thinking “independently” of the nationwide litigation). But while those facts are entitled to consideration, they don’t mean that blind deference to Zhang is appropriate. There are a vast number of peer-reviewed journals out there. Pre-publication editorial peer review, just by itself, is far from a guarantee of scientific reliability. See Valentine v. Pioneer Chlor Alkali Co., 921 F. Supp. 666, 674—76 (D. Nev. 1996) (distinguishing between “editorial peer review” and “true peer review,” and arguing that the latter—which includes the scientific reception of an article after publication, efforts to replicate its findings, etc.—is what’s really important under Daubert).® Plus, part of the point of centralizing cases like these in an MDL is that the Court, over many years, can develop a fluency in the relevant scientific literature. Where that fluency helps identify serious problems in an expert opinion, it doesn’t make sense for the Court to ignore them. Second, the other meta-analyses that Zhang cites, which also use an “a priori hypothesis” to isolate high-exposure groups in epidemiological studies, don’t help her as much as she would like. These studies may share some of the basic features of Zhang’s paper, but that isn’t enough to save a study so riddled with problems. The other papers also aren’t methodologically 8 See also Effie Chan, Note, The Brave New World of Daubert: True Peer Review, Editorial Peer Review, and Scientific Validity, 70 N.Y.U. Law Review 100 (1995); Smith, Peer Review: A Flawed Process at the Heart of Science and Journals, 99 Journal of the Royal Society of Medicine 178 (2006) (arguing that there are many shortcomings of pre-publication peer review, including that it is not a good system for catching errors, is prone to publication bias, and often operates inconsistently). 13 Case 3:16-md-02741-VC Document 18659 Filed 06/20/24 Page 14 of 14 comparable to Zhang’s in important ways. For one thing, all of the cited meta-analyses examined at least twice as many epidemiological studies as Zhang’s glyphosate meta-analysis, and often many more. Perhaps if Zhang had a lot more studies to feed into her meta-analysis, she might have a defensible claim to be roughly capturing “high-exposure” data points. As it is, her analysis does no such thing. Finally, Bulone’s arguments do nothing to address the problems with using Zhang’s meta-analysis as a standalone analysis of the epidemiological evidence. One can’t assume that peer reviewers were vetting a study’s ability to play a certain role in an expert opinion, and there’s no reason to think that was the case here. In short, simply intoning that a paper was peer- reviewed and that it was (apparently) produced before the author was thinking about getting involved in litigation isn’t enough to compensate for methodological problems as glaring as those described here. And it certainly isn’t enough to show that Zhang’s paper can support a reliable general causation opinion. iil Monsanto filed a motion for summary judgment in connection with its motion to exclude Dr. Braunstein, Bulone’s specific causation expert. See Dkt. No. 17598. That motion mentioned the necessity of summary judgment in the absence of a specific causation expert. Id. at 9-10. The absence of an admissible general causation opinion requires the same result, as Bulone’s counsel conceded. Hearing Tr. at 87:14—25. Accordingly, summary judgment is granted for Monsanto on the grounds that Bulone lacks any admissible evidence of general causation. IT IS SO ORDERED. Loe Dated: June 20, 2024 VINCE CHHABRIA United States District Judge 14 EXHIBIT 2 Accepted Manuscript PUSAN Title: Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma: A Meta-Analysis and Supporting Evidence Authors: Luoping Zhang, Iemaan Rana, Rachel M. Shaffer, = Emanuela Taioli, Lianne Sheppard PII: $1383-5742(18)30088-7 DOI: https://doi.org/10.1016/j.mrrev.2019.02.001 Reference: MUTREV 8261 To appear in: Mutation Research Received date: 22 September 2018 Revised date: 2 February 2019 Accepted date: 5 February 2019 Please cite this article as: Zhang L, Rana I, Shaffer RM, Taioli E, Sheppard L, Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma: A Meta- Analysis and Supporting Evidence, Mutation Research-Reviews in Mutation Research (2019), https://doi.org/10.1016/j.mrrev.2019.02.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Exposure to Glyphosate-Based Herbicides and Risk for Non-Hodgkin Lymphoma: A Meta-Analysis and Supporting Evidence Luoping Zhang, lemaan Rana*, Rachel M. Shaffer®, Emanuela Taioli*, Lianne Sheppard>4 4 Division of Environmental Health Sciences, School of Public Health, University of California Berkeley, Berkeley, USA > Department of Environmental and Occupational Health Sciences, University of Washington, Seattle, USA © Institute for Translational Epidemiology and Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, USA 4 Department of Biostatistics, University of Washington, Seattle, USA *Corresponding Author: Luoping Zhang, PhD Division of Environmental Health Sciences School of Public Health 2121 Berkeley Way West, #5117 Berkeley, CA 94720-7360 Tel: (510) 643-5189 Fax: (510) 642-0427 Email: luoping@berkeley.edu Abstract Glyphosate is the most widely used broad-spectrum systemic herbicide in the world. Recent evaluations of the carcinogenic potential of glyphosate-based herbicides (GBHs) by various regional, national, and international agencies have engendered controversy. We investigated whether there was an association between high cumulative exposures to GBHs and increased risk of non-Hodgkin lymphoma (NHL) in humans. We conducted anew meta-analysis that included the most recent update of the Agricultural Health Study (AHS) cohort published in 2018 along with five case-control studies. Using the highest 1 exposure groups when available in each study, we report the overall meta-relative risk (meta-RR) of NHL in GBH-exposed individuals was increased by 41% (meta-RR = 1.41, 95% Cl, confidence interval: 1.13—-1.75). For comparison, we also performed a secondary meta-analysis using high-exposure groups with the earlier AHS (2005), and we determined a meta-RR for NHL of 1.45 (95% Cl: 1.11-1.91), which was higher than the meta-RRs reported previously. Multiple sensitivity tests conducted to assess the validity of our findings did not reveal meaningful differences from our primary estimated meta- RR. To contextualize our findings of an increased NHL risk in individuals with high GBH exposure, we reviewed available animal and mechanistic studies, which provided supporting evidence for the carcinogenic potential of GBH. We documented further support from studies of malignant lymphoma incidence in mice treated with pure glyphosate, as well as potential links between GBH exposure and immunosuppression, endocrine disruption, and genetic alterations that are commonly associated with NHL. Overall, in accordance with evidence from experimental animal and mechanistic studies, our current meta-analysis of human epidemiological studies suggests a compelling link between exposures to GBHs and increased risk for NHL. (273/300 words) Abbreviations: AHS, Agricultural Health Study; c-NHEJ, canonical non-homologous end joining pathway; Cl, confidence interval; EDC, endocrine disrupting chemical; EFSA, European Food Safety Authority; EPA, Environmental Protection Agency; ETS, environmental tobacco smoke; GBHs, glyphosate-based herbicides; IARC, International Agency for Research on Cancer; IFN-y, interferon gamma; IL-2, Interleukin-2; JMPR, 2 Joint Meeting on Pesticide Residues by the Food and Agriculture Organization of the United Nations and World Health Organization; meta-RR, meta-analysis relative risk; mg/kg/day, milligrams per kilogram per day; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; OR, odds ratio; ppm, parts per million; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis; RR, relative risk. Keywords: Glyphosate, pesticide, Roundup, Ranger Pro, carcinogenesis, and meta- analysis. 1. Background 1.1 Global Usage of Glyphosate-Based Herbicides Glyphosate is a highly effective broad spectrum herbicide that is typically applied in mixtures known as glyphosate-based herbicides (GBHs) and commonly sold under the trade names of Roundup® and Ranger Pro®. Use of GBHs has increased dramatically worldwide in recent decades. In the United States alone, usage increased nearly sixteen- fold between 1992 and 2009 [1]. Most of this increase occurred after the introduction of genetically modified glyphosate-resistant “Roundup-ready” crops in 1996 [1]. In addition, there have been significant changes in usage. In particular, the practice of applying GBHs to crops shortly before harvest, so-called “green burndown,” began in the early 2000s to speed up their desiccation; as a consequence, crops are likely to have higher GBH residues [2]. By the mid-2000s, green burndown became widespread, and regulatory agencies responded by increasing the permissible residue levels for GBHs [3, 4]. 1.2 Ubiquitous Exposure in Humans Glyphosate and its metabolites persist in food [5-7], water [8], and dust [9], potentially indicating that everyone may be exposed ubiquitously. Non-occupational exposures occur primarily through consumption of contaminated food, but may also occur through contact with contaminated soil [9], dust [9] and by drinking or bathing in contaminated water [8]. In plants, glyphosate may be absorbed and transported to parts used for food; thus, it has been detected in fish [5], berries [6], vegetables, baby formula [7], and grains [10], and its use as a crop desiccant significantly increases residues. GBH residues in food persist long after initial treatment and are not lost during baking. Limited data exist on internal glyphosate levels among GBH-exposed individuals [11]. Average urinary glyphosate levels among