Molecular Fingerprints? On the Science and Ethics of Transcriptomic Research

A few weeks ago, I stumbled across a newspaper article on transcriptomic research. It stated that brains of persons with various psychiatric disorders share a common “molecular fingerprint.” The article refers to a publication by Gandal et al. in the journal Science.¹ In a UCLA press release on February 8th 2018, senior author Daniel Geschwind says that “[these] findings provide a molecular, pathological signature of these disorders, which is a large step forward.”²

First of all, how does transcriptomic, or gene expression, research work? Gene expression studies are seen as more informative than traditional genetic studies. Any given gene can be more or less active or even completely inactive. Gene expression studies therefore do not look at the genes themselves, but at their activity in the form of mRNA. mRNAs are basically blueprints of genes. They carry all the information necessary to manufacture proteins, which in turn determine cell function. The more mRNA is found in a cell, the more active the respective gene is considered to be.

Gene activity can be quantified, but not in absolute terms. It is computed by comparing two datasets. For example, in cancer research, cell activity of tumor cells can be compared to cell activity in normal cells. When a particular gene in the tumor cell is expressed in higher frequency (more mRNA) than in the normal cell, this is called an upregulation of the gene. When a particular gene in the tumor cell is expressed in lower frequency (less mRNA), this is called downregulation.

There are several problems with the study by Gandal, Geschwind, et al. First, differences in gene expression are neither specific for “mental illness” in general nor for a certain disorder. What the researchers find are mere statistical effects, not differences in gene activity found in each pair. The differences are therefore not specific for persons with certain psychiatric diagnoses and have no diagnostic value. However, the article in Science frames the findings in words such as “specificity,” “perturbations,” “dysregulation” and “transcriptomic severity.” This is completely uncalled for and very misleading.

Second, it is probably important to note that Gandal et al. included only published studies. This might have inflated the differences between patients and controls.

Third, Gandal et al. examined “networks” of coexpressed genes. However, samples that did not show these specific coexpression patterns were excluded from the analysis! This is a curious side note and you will only find out about this if you download and carefully read through the supplementary material.

Fourth, one third of the studies lack important information on the lives and deaths of the subjects. The study is based on 14 studies in total. Only four of these 14 studies report on controlling for use of psychotropic medication; in one of them this was only the case for one of two patient groups. According to the authors of one of the studies included, controlling for psychotropic medication is particularly important because many of these drugs have shown mitochondrial toxicity and downregulation of mitochondrial genes. In their conclusion, they write:

Mitochondrial dysfunction in BD (bipolar disorder) and SZ (schizophrenia) was observed even after the effect of pH was controlled. However, we conclude that this apparent dysfunction was due to the patients’ medication, especially antipsychotics.³

The authors of another study state that they

[…] cannot exclude that possibility that differences between cohorts of different disease duration are due to disparities in lifetime exposure to such drugs.⁴

Gandal et al. did not control for medication type and dose and treatment duration. They argue that a study on non-human primates has found that neuroleptic treatment, administered acutely or chronically, partially reverses “schizophrenic” gene-expression patterns as mimicked by phencyclidine, a drug given to induce psychosis-like symptoms. However, in that study,⁵“chronic” treatment lasted only four weeks. The persons whose brains were used in Gandal et al.’s research had taken psychotropic drugs for years or even decades. In one sample of “chronic patients” 33 years was the minimum (!) duration of neuroleptic treatment. Information on ECT, alcohol and other drug consumption is completely lacking.

These details are worth of scientific discussion. In this article, I want to primarily focus on something else: the ethics of transcriptomic research. When researching gene activity in a tumor, organ tissue can often be extracted via biopsy. You cannot do the same with brain tissue. Research on gene activity in human brains is dependent on donations. Thus, transcriptomic research in psychiatry is in need of dead people.

My first question when reading articles of that kind is: Who were these people? The only things we know about them from the study are, first, their diagnoses, and second, that they are dead. Most of us know someone, a fellow (ex-)patient or friend, who has died prematurely, way too often by suicide, but also from medication effects or from “unidentified causes.” Also, many psychiatric survivors have been victims of child abuse, and some have left their families of origin. Almost all have suffered greatly before their premature death. So, who were these people whose brains were used in Gandal et al.’s research? How did they live and die? How did the researchers gain permission to open their skulls and extract brain tissue for research purposes? The article itself contains no information on these questions. For any information on the sample, you have to download the supplementary material. Here, you learn that the study included samples from 14 studies in total. However, the supplementary material contains only information on the platform from which the datasets were retrieved and details on the methods of data extraction and computing. So, for information on the samples, you have to take a look at every single study. Which is what I did.

It is important to note here: The sample in the study by Gandal et al. does not consist of persons, but of gene expression data. In the original studies, samples are not comprised of persons either, but of brain tissue.

Here is what I found: In only one of the studies, ⁶ consent was explicitly received from the persons themselves while they were still alive. And even here this was only the case for part of the sample. The other part of the sample was derived from the Harvard Brain Bank. In the US, the legal framework for postmortem brain donations is called the Uniform Anatomical Gift Act. It includes a hierarchy of those authorized to make a donation of a deceased person’s brain. This can be (in this rank order) a healthcare agent appointed under a health care proxy, the legal spouse, a parent, the legal guardian, but also any other person “having the authority to dispose of the body,” whatever that means. Consent of the deceased person is not required for making a brain donation. For four of the studies, information on sample obtainment and consent is completely lacking. Some of the other studies simply state that brains were “obtained during autopsies” and consent was received by “families” or “next-of-kin.” One study says about sample acquisition:

Suicides, deaths occurring in group homes known to house individuals with psychiatric disorders, and deaths occurring in residences in which antipsychotic or mood stabilizing medications are found by the investigators are considered to be potential donors.⁷

So, from the point of view of the researchers, a potential donor is a dead person who might have never known they were a potential donor, but who has certain characteristics which are of interest to the researchers. Following that logic, a donation is not an active decision, but something that researchers can request from a relative or any person “having the authority to dispose of the body.”

Three of the studies were on autism and used brain tissue from the Autism Tissue Program (now Autism Brain Net). On its registration site It Takes Brains it tries to convince families that

(…) postmortem studies on human brain tissue represent the only way that researchers can gain a deeper understanding of autism on the genetic, cellular, and molecular levels.

“It Takes Brains” wants to “solve autism” and therefore “our mission is to urge families to make the heroic decision to register for brain tissue donation.” There is a hotline you can call “if a death occurs and you wish to begin the donation process.” On the website, a family of a teenager with autism is portrayed wearing green superhero capes and t-shirts with logos of the website. In a promotional video, the parents talk about their 14-year-old son with autism (who is still alive) and about their hopes of improving autism treatment by brain research. This quote from his mother is heartbreaking:

It’s gonna sound silly, but I hope that scientists absolutely love my son’s brain. I hope that they dive into it and they are wowed and amazed and excited and passionate to see what’s in there. I hope they can use my son’s brain to find the cure for autism, so that one other family doesn’t have to feel what I’ve had to feel for the past 14 years.

This mother who has obviously gone through a lot of pain has no idea that the scientists have absolutely no consideration for individual brains, nor for the individuals whose brains they are collecting. She does not know that the aim of the program is to collect large enough numbers of tissue samples to achieve adequate statistical power. This is tragic, because families trust these scientists who do not provide them with sufficient information to understand their work and make an informed decision. The obscure propaganda for brain donations as seen on It Takes Brains is highly unethical, even when targeted at families whose supposedly mentally ill relatives are still alive. Requests for brain donations are even more unethical when targeted at families who have just learned about the death of their loved ones. This certainly was the case for the majority of samples included in the study by Gandal et al.

I learned a lot from this enquiry. Transcriptomic research is one step forward in scientists’ quest to make the public believe in the myth of mental suffering as brain disease with a genetic basis. Scientists are being dishonest with the public, with families and with people with psychiatric diagnoses by telling us that there exist unique “molecular signatures” of mental disorders. The research from which these lies are derived is practically based on desecration of corpses. I urgently hope that people will realize that this research is not for us — it brings no direct or indirect advantages for persons in mental distress or their families — it is for the researchers.