Hidden hypotheses of epigenetic studies

This 2018 UK review discussed three pre-existing conditions of epigenetic genome-wide association studies:

“Genome-wide technology has facilitated epigenome-wide association studies (EWAS), permitting ‘hypothesis-free’ examinations in relation to adversity and/or mental health problems. Results of EWAS are in fact conditional on several a priori hypotheses:

  1. EWAS coverage is sufficient for complex psychiatric problems;
  2. Peripheral tissue is meaningful for mental health problems; and
  3. The assumption that biology can be informative to the phenotype.

1. CpG sites were chosen as potentially biologically informative based on consultation with a consortium of DNA methylation experts. Selection was, in part, based on data from a number of phenotypes (some medical in nature such as cancer), and thus is not specifically targeted to brain-based, stress-related complex mental health phenotypes.

2. The assumption is often that distinct peripheral tissues are interchangeable and equally suited for biomarker detection, when in fact it is highly probable that peripheral tissues themselves correspond differently to environmental adversity and/or disease state.

3. Analyses result in general statements such as ‘neurodevelopment’ or the ‘immune system’ being involved in the aetiology of a given phenotype. Whether these broad categories play indeed a substantial role in the aetiology of the mental health problem is often hard to determine given the post hoc nature of the interpretation.”


The reviewers mentioned in item #2 the statistical flaw of assuming that measured entities are interchangeable with one another. They didn’t mention that this problem also affects item #1 methodologies of averaging CpG methylation measurements in fixed genomic bins or over defined genomic regions. This was discussed in:

The reviewers offered suggestions for reducing the impacts of these three hypotheses. But will doing more of the same, only better, advance science?

Is it too much to ask of researchers whose paychecks and reputations depend on a framework’s paradigm – such as the “biomarker” mentioned a dozen and a half times – to admit the uselessness of gathering data when the framework in which the data operates isn’t viable?

The truth about complex traits and GWAS added another example of how this framework and many of its paradigms haven’t produced effective explanations of “the aetiology of the mental health problem”:

“The most investigated candidate gene hypotheses of schizophrenia are not well supported by genome-wide association studies, and it is likely that this will be the case for other complex traits as well.”

Researchers need to reevaluate their framework if they want to make a difference in their fields. Recasting GWAS as EWAS won’t make it more effective.

https://www.sciencedirect.com/science/article/pii/S2352250X18300940 “Hidden hypotheses in ‘hypothesis-free’ genome-wide epigenetic associations”

Advertisements

Preventing prostate cancer with a broccoli sprouts diet

This 2018 Oregon rodent study fed a 15% broccoli sprout diet beginning at four weeks of age to a mouse strain with a near-100% chance of developing prostate cancer:

“Broccoli sprouts reduced prostate cancer incidence and progression to invasive cancer. Broccoli sprout consumption also decreased histone H3 lysine 9 trimethylation in the ventral lobe (age 12 wk), and decreased histone H3 lysine 18 acetylation in all prostate lobes (age 28 wk).

The TRAMP model of prostate cancer was utilized because the tumors occur in the prostate epithelium and the tumor tissue histopathology closely mimics human disease. Additional advantages include that the tumors arise spontaneously and appear in ∼100% of mice.”


Like in utero prevention of breast cancer by a broccoli sprouts diet, this study had a problem measuring sulforaphane dosage. The relevant statements were:

“This 15% broccoli sprout diet had 400 mg SFN [sulforaphane]/kg diet, which was chosen because it is equivalent to 1 mg SFN/d which has been used in previous studies.

Food consumption was measured over the course of the study and no difference was found in the intake of food between the control and broccoli sprout–fed groups.”

To be “equivalent to 1 mg SFN/d” at a .4 mg sulforaphane/gram rate, the animals would need to eat 2.5 grams per day. “Food consumption was measured” but not disclosed.

Also, looking at the sulforaphane references, the study cited at http://cancerpreventionresearch.aacrjournals.org/content/early/2015/02/21/1940-6207.CAPR-14-0386.full-text.pdf for the “1 mg SFN/d” dosage was actually:

“4 week old male TRAMP mice were treated with PBS [phosphate-buffered saline] (control) or 1 mg SFN in PBS three times/week for 15-18 weeks.”

not “1 mg SFN/d.”

I downgraded the study’s rating because the researchers didn’t sufficiently quantify their findings to help humans, which is the basic purpose of any animal study. The study’s sulforaphane dosage was undefined, so no human equivalent dosage could be derived.

https://academic.oup.com/cdn/article/2/3/nzy002/4803105 “Broccoli Sprouts Delay Prostate Cancer Formation and Decrease Prostate Cancer Severity with a Concurrent Decrease in HDAC3 Protein Expression in Transgenic Adenocarcinoma of the Mouse Prostate (TRAMP) Mice”

The lack of oxygen’s epigenetic effects on a fetus

This 2018 Loma Linda review subject was gestational hypoxia:

“Of all the stresses to which the fetus and newborn infant are subjected, perhaps the most important and clinically relevant is that of hypoxia. This review explores the impact of gestational hypoxia on maternal health and fetal development, and epigenetic mechanisms of developmental plasticity with emphasis on the uteroplacental circulation, heart development, cerebral circulation, pulmonary development, and the hypothalamic-pituitary-adrenal axis and adipose tissue.

An understanding of the specific hypoxia-induced environmental and epigenetic adaptations linked to specific organ systems will enhance the development of target-specific inhibition of DNA methylation, histone modifications, and noncoding RNAs that underlie hypoxia-induced phenotypic programming of disease vulnerability later in life.

A potential stumbling block to these efforts, however, relates to timing of the intervention. The greatest potential effect would be accomplished at the critical period in development for which the genomic plasticity is at its peak, thus ameliorating the influence of hypoxia or other stressors.

With future developments, it may even become possible to intervene before conception, before the genetic determinants of the risk of developing programmed disease are established.”

Table 3 “Antenatal hypoxia and developmental plasticity” column titles were Species | Offspring Phenotypes of Disorders and Diseases | Reference Nos.

Hypoxia phenotypes


This review was really an ebook, with 94 pages and 1,172 citations in the pdf file. As I did with Faith-tainted epigenetics, I read it with caution toward recognizing the influence of the sponsor’s biases, and any directed narrative that ignored evidence contradicting the narrative, and any storytelling.

See if you can match the meaning of the review’s last sentence (“intervene before conception” quoted above) with the meaning of any sentence in its cited reference Developmental origins of noncommunicable disease: population and public health implications.

One review topic that was misconstrued was transgenerational epigenetic inheritance of hypoxic effects. The “transgenerational” term was used inappropriately by several of the citations, and no cited study provided evidence for gestational hypoxic effects through the  F2 grandchild and F3 great-grandchild generations.

One omitted topic was gestational hypoxic effects of caffeine. The first paper that came up for my PubMed search of “caffeine pregnancy hypoxia” was an outstanding 2017 Florida rodent review Long-term consequences of disrupting adenosine signaling during embryonic development that had this paragraph and figure:

“One substance that fetuses are frequently exposed to is caffeine, which is a non-selective adenosine receptor antagonist. We discovered that in utero alteration in adenosine action leads to adverse effects on embryonic and adult murine hearts. We find that cardiac A1ARs [a type of adenosine receptor] protect the embryo from in utero hypoxic stress, a condition that causes an increase in adenosine levels. 

After birth in mice, we observed that in utero caffeine exposure leads to abnormal cardiac function and morphology in adults, including an impaired response to β-adrenergic stimulation. Recently, we observed that in utero caffeine exposure induces transgenerational effects on cardiac morphology, function, and gene expression.”

The timing of in utero caffeine treatment leads to differences in adult cardiac function, gene expression, and phenotype. Exposure to caffeine from E6.5–9.5 leads the F1 generation to develop dilated cardiomyopathy with decrease % FS and increased Myh7 expression. In utero caffeine exposure from E10.5–13.5 leads to a hypertrophic cardiomyopathy in the F2 generation along with increased % FS and decreased Myh7 expression

Why was this review and its studies omitted? It was on target for both gestational hypoxia and transgenerational epigenetic inheritance of hypoxic effects!

It was alright to review smoking, cocaine, methamphetamine, etc., but the most prevalent drug addiction – caffeine – couldn’t be a review topic?


The Loma Linda review covered a lot, but I had a quick trigger due to the sponsor’s bias. I started to lose “faith” in the reviewers after reading the citation for the review’s last sentence that didn’t support the statement.

My “faith” disappeared after not understanding why a few topics were misconstrued and omitted. Why do researchers and sponsors ignore, misrepresent, and not continue experiments through the F3 generation to produce evidence for and against transgenerational epigenetic inheritance? Where was the will to follow evidence trails regardless of socially acceptable beverage norms?

The review acquired the taint of storytelling with the reviewers’ assertion:

“..timing of the intervention. The greatest potential effect would be accomplished at the critical period in development for which the genomic plasticity is at its peak, thus ameliorating the influence of hypoxia or other stressors.”

Contradictory evidence was in the omitted caffeine study’s graphic above which described two gestational critical periods where an “intervention” had opposite effects, all of which were harmful to the current fetus’ development and/or to following generations. Widening the PubMed link’s search parameters to “caffeine hypoxia” and “caffeine pregnancy” returned links to human early life studies that used caffeine in interventions, ignoring possible adverse effects on future generations.

This is my final curation of any paper sponsored by this institution.

https://www.physiology.org/doi/abs/10.1152/physrev.00043.2017 “Gestational Hypoxia and Developmental Plasticity” (not freely available) Thanks to coauthor Dr. Xiang-Qun Hu for providing a copy.

in utero prevention of breast cancer by a broccoli sprouts diet

This 2018 Alabama rodent study investigated the epigenetic effects on developing breast cancer of timing a sulforaphane-based broccoli sprouts diet. Timing of the diet was as follows:

  1. Conception through weaning (postnatal day 28), named the Prenatal/maternal BSp (broccoli sprouts) treatment (what the mothers ate starting when they were adults at 12 weeks until their pups were weaned; the pups were never on a broccoli sprouts diet);
  2. Postnatal day 28 through the termination of the experiment, named the Postnatal early-life BSp treatment (what the offspring ate starting at 4 weeks; the mothers were never on a broccoli sprouts diet); and
  3. Postnatal day 56 through the termination of the experiment, named the Postnatal adult BSp treatment (what the offspring ate starting when they were adults at 8 weeks; the mothers were never on a broccoli sprouts diet).

“The experiment was terminated when the mean tumor diameter in the control mice exceeded 1.0 cm.

Our study indicates a prenatal/maternal BSp dietary treatment exhibited maximal preventive effects in inhibiting breast cancer development compared to postnatal early-life and adult BSp treatments in two transgenic mouse models that can develop breast cancer.

Postnatal early-life BSp treatment starting prior to puberty onset showed protective effects in prevention of breast cancer but was not as effective as the prenatal/maternal BSp treatment. However, adulthood-administered BSp diet did not reduce mammary tumorigenesis.

The prenatal/maternal BSp diet may:

  • Primarily influence histone modification processes rather than DNA methylation processes that may contribute to its early breast cancer prevention effects;
  • Exert its transplacental breast cancer chemoprevention effects through enhanced histone acetylation activator markers due to reduced HDAC1 expression and enzymatic activity.

This may be also due to the importance of a dietary intervention window that occurs during a critical oncogenic transition period, which is in early life for these two tested transgenic mouse models. Determination of a critical oncogenic transition period could be complicated in humans, which may partially explain the controversial findings of the adult BSp treatment on breast cancer development in the tested mouse models as compared the previous studies. Thus long-term consumption of BSp diet is recommended to prevent cancers in humans.”


“The dietary concentration for BSp used in the mouse studies was 26% BSp in formulated diet, which is equivalent to 266 g (~4 cups) BSp/per day for human consumption. Therefore, the concentration of BSp in this diet is physiological available and represents a practical consumption level in the human diet.

Prior to the experiment, we tested the potential influences of this prenatal/maternal BSp regimen on maternal and offspring health as well as mammary gland development in the offspring. Our results showed there was no negative effect of this dietary regimen on the above mentioned factors (data not shown) suggesting this diet is safe to use during pregnancy.”


I downgraded the study’s rating because I didn’t see where the sulforaphane active content of the diet was defined. It’s one thing to state:

“SFN as the most abundant and bioactive compound in the BSp diet has been identified as a potent HDAC inhibitor that preferably influences histone acetylation processes.”

and describe how sulforaphane may do this and may do that, and include it in the study’s title.

It’s another thing to quantify an animal study into findings that can help humans. Normal people aren’t going to eat “4 cups BSp/per day” but we may take one capsule of a sulforaphane dietary supplement when the price is $.20 a day.

The study’s food manufacturer offers dietary products to the public without quantifying all of the active contents like sulforaphane. Good for them if they can stay in business by serving customers who can’t be bothered with scientific evidence.

These researchers shouldn’t have conducted a study using the same lack of details as the food manufacturer provided, though. They should have either tasked the manufacturer to specify the sulforaphane active content, or contracted the analysis.

Regarding timing of a sulforaphane-based broccoli sprouts diet for humans, the study also didn’t provide evidence for recommending:

“Thus long-term consumption of BSp diet is recommended to prevent cancers in humans.”

http://cancerpreventionresearch.aacrjournals.org/content/early/2018/05/15/1940-6207.CAPR-17-0423.full-text.pdf “Temporal efficacy of a sulforaphane-based broccoli sprout diet in prevention of breast cancer through modulation of epigenetic mechanisms”

Are there epigenetic causes for sexual orientation and gender identity?

This US 2018 review lead author was a gynecologic oncologist in private practice:

“Sexual orientation is biologically conferred in the first trimester of pregnancy. Gender identity is biologically conferred during the middle trimester of pregnancy.

Since the genitals differentiate in the first trimester, and the brain becomes imprinted in the latter half of gestation, it is possible for the fetal brain to be imprinted differently than the genitals. As children mature, this innate imprinting expresses as genital anatomy, gender identity, sexual orientation and other physiologic capabilities and natural preferences along a continuum, between masculine and feminine.

The evidence shows that both orientation and identity are biologic features that co-vary with a very large number of other biologic sexually dimorphic traits.”


A fetus’ development is influenced by survival reactions to their environment. Although fetal and placental responses to environmental stressors are relevant to sexual orientation and gender identity, the subject wasn’t explored.

Epigenetic adaptations to the prenatal environment involving microRNA were mentioned in a small subsection. But the review didn’t cite relevant studies involving DNA methylation, chromatin and histone modifications for epigenetic causes of and effects on sexual orientation and gender identity.

The authors included a half-dozen anecdotal quotations from personal correspondence that promoted their narrative. These came across as appeals to authority rather than evidence for scientific understanding of the subject.

It was insufficient for the review to note “a continuum between masculine and feminine” without also exploring evidence for an individual’s placement on the continuum, including possible epigenetic causes for sexual orientation and gender identity.

https://www.sciencedirect.com/science/article/pii/S009082581731510X “Biological origins of sexual orientation and gender identity: Impact on health” (not freely available)

The Not-Invented-Here syndrome

I have high expectations of natural science researchers. I assume that their studies will improve over time, and develop methods and experiments that produce reliable evidence to inform us of human conditions.

My confidence is often unrealistic. Scientists are people, after all, and have the same foibles as the rest of us.

I anticipate that researchers will keep abreast of others’ work around the world. If other groups in their research areas are developing better methods and exploring hypotheses that discover better applications for humans, why not adopt them in the interest of advancing science?

That’s not what happened with this 2018 UK rodent study. The rat model some of the coauthors have built their reputations on depends on disturbing rat pregnancies by administering glucocorticoids. But both the rat model and a guinea pig model in Do you have your family’s detailed medical histories? demonstrated that physicians who disturb their pregnant human patients in this way may be acting irresponsibly toward their patients’ fetuses and their future generations.

This study didn’t find mechanisms that explained transgenerational epigenetic birth weight effects through the F2 generation:

“Although the phenotype is transmitted to a second generation, we are unable to detect specific changes in DNA methylation, common histone modifications or small RNA [including microRNA] profiles in sperm.

The inheritance mechanism for the paternally derived glucocorticoid-reprogrammed phenotype may not be linked with the specific germline DNA, sRNA and chromatin modifications that we have profiled here.”


The linked guinea pig model was developed specifically to inform physicians of the consequences through the F3 generation of disturbing human pregnancies with glucocorticoids:

“Antenatal exposure to multiple courses of sGC [synthetic glucocorticoid] has been associated with hyperactivity, impaired attention, and neurodevelopmental impairment in young children and animals. It is imperative that the long-term effects of antenatal exposure to multiple courses of sGC continue to be investigated since the use of a ‘rescue’ (i.e. a second) course of sGC has recently re-introduced the practice of multiple course administration.”


If a study’s purpose is to investigate potential mechanisms of epigenetic inheritance, why not adopt a model that better characterizes common human conditions, regardless of which research group initially developed it?

The prenatal stress model used in The lifelong impact of maternal postpartum behavior is one model that’s more representative of human experiences. Those researchers pointed out in Prenatal stress produces offspring who as adults have cognitive, emotional, and memory deficiencies that:

“Corticosterone-treated mice and rats exposed to chronic stress are models that do not recapitulate the early programming of stress-related disorders, which likely originates in the perinatal period.”

Animal models that chemically redirect fetal development also “do not recapitulate the early programming of stress-related disorders.”

Other than research that’s done to warn against disrupted development, how can animal studies like the current study help humans when their models don’t replicate common human conditions? This failure to use more relevant models has follow-on effects such as human intergenerational and transgenerational epigenetic inheritance being denigrated due to insufficient evidence.

Of course there’s insufficient human evidence! Researchers developed and sponsors funded animal study designs that ensured there wouldn’t be wide applicability to humans. Few derivative human studies were developed and funded as a result.

https://genomebiology.biomedcentral.com/articles/10.1186/s13059-018-1422-4 “Investigation into the role of the germline epigenome in the transmission of glucocorticoid-programmed effects across generations”

What will it take for childhood trauma research to change paradigms?

This 2018 German human study found:

“DNA methylation in a biologically relevant region of NR3C1-1F [the glucocorticoid receptor gene] moderates the specific direction of HPA-axis dysregulation (hypo- vs. hyperreactivity) in adults exposed to moderate-severe CT [childhood trauma].

In contrast, unexposed and mildly-moderately exposed individuals displayed moderately sized cortisol stress responses irrespective of NR3C1-1F DNA methylation. Contrary to some prior work, however, our data provides no evidence for a direct association of CT and NR3C1-1F DNA methylation status.”


The study was an example of why researchers investigating the lasting impacts of human traumatic experiences won’t find causes, effects, and productive therapies until their paradigms change.

1. Limited subject histories

A. Why weren’t the subjects asked for historical information about their parents, grandparents, and great-grandparents?

The researchers had no problem using animal studies to guide the study design, EXCEPT for animal studies of the etiologic bases of intergenerational and transgenerational transmission of biological and behavioral phenotypes. Just the approximate places and dates of three generations of the German subjects’ ancestors’ births, childhoods, adolescences, and early adulthoods may have provided relevant trauma indicators.

B. Why are studies still using the extremely constrained Childhood Trauma Questionnaire? Only one CTQ aspect was acknowledged as a study design limitation:

“Our findings rely on retrospective self-report measures of CT, which could be subject to bias.”

But bias was among the lesser limiting factors of the CTQ.

The study correlated epigenetic changes with what the subjects selectively remembered, beginning when their brains developed sufficient cognitive functionalities to put together the types of memories that could provide CTQ answers – around age four. The basic problem prohibiting the CTQ from discovering likely most of the subjects’ historical traumatic experiences that caused epigenetic changes is that these experiences predated the CTQ’s developmental starting point:

  1. A human’s conception through prenatal period is when both the largest and the largest number of epigenetic changes occur, and is when our susceptibility and sensitivity to our environment is greatest;
  2. Birth through infancy is the second-largest; and
  3. Early childhood through the age of three is the third largest.

CTQ self-reports were – at best – evidence of experiences after age three, distinct from the  experience-dependent epigenetic changes since conception. If links existed between the subjects’ early-life DNA methylation and later-life conditions, they weren’t necessarily evidenced by CTQ answers about later life that can’t self-report relevant early-life experiences that may have caused DNA methylation.

2. Limited subject selection

The researchers narrowed down the initial 622 potential subjects to the eventual 200 subjects aged 18 to 30. An exclusion criteria that was justified as eliminating confounders led to this limitation statement:

“Our results might be based on a generally more resilient sample as we had explicitly excluded individuals with current or past psychopathology.”

Was it okay for the researchers to assert:

“Exposure to environmental adversity such as childhood trauma (CT) affects over 10% of the Western population and ranges among the best predictors for psychopathology later in life.”

but not develop evidence for the statement by letting people who may have been already affected by age 30 and received treatment participate in the study? Was the study design so fragile that it couldn’t adjust to the very people who may be helped by the research findings?

3. Limited consequential measurements

The current study design was very conformant to previous studies’ protocols. The researchers chose cortisol and specific DNA methylation measurements.

A. Here’s what Sex-specific impacts of childhood trauma had to say about cortisol:

“Findings are dependent upon variance in extenuating factors, including but not limited to, different measurements of:

  • early adversity,
  • age of onset,
  • basal cortisol levels, as well as
  • trauma forms and subtypes, and
  • presence and severity of psychopathology symptomology.”

The researchers knew or should have known all of the above since this quotation came from a review.

B. What other consequential evidence for prenatal, infancy, and early childhood experience-dependent epigenetic changes can be measured? One overlooked area is including human emotions as evidence.

There are many animal studies from which to draw inferences about human emotions. There are many animal models of creating measurable behavioral and biological phenotypes of human emotion correlates, with many methods, including manipulating environmental variables during prenatal, infancy, and early childhood periods.

Studies that take detailed histories may arrive at current emotional evidence for human subjects’ earliest experience-dependent changes. It’s not too big a leap to correlate specific historical environments and events, stress measurements, and lasting human emotions expressed as “I’m all alone” and “No one can help me” to better understand causes and effects.

CTQ answers aren’t sufficiently detailed histories.

4. Limited effective treatments and therapies

The current study only addressed this area in the final sentence:

“Given their potential reversibility, uncovering epigenetic contributions to differential trajectories following childhood adversity may serve the long-term goal of delivering personalized prevention strategies.”


Researchers: if your paradigms demonstrate these characteristics, why are you spending your working life in efforts that can’t make a difference? Isn’t your working life more valuable than that? What else could you investigate that could make a difference in your field?

I hope that researchers will value their professions enough to make a difference with their expertise. And that sponsors won’t thwart researchers’ desires for difference-making science by putting them into endless funding queues.

http://www.psyneuen-journal.com/article/S0306-4530(17)31355-0/pdf “Glucocorticoid receptor gene methylation moderates the association of childhood trauma and cortisol stress reactivity” (not freely available)