The world's first "brain chart": how the brain expands and atrophied from fetus to centenarian

The Frenchman De Montbeillard drew the first human growth curve at the end of the 18th century. From the birth of his son, he measured and recorded his son's height every six months until he was 18 years old, thus obtaining the most complete and measured growth curve. Since then, for more than 200 years, growth assessment has evolved into an important component of child health care.

Unfortunately, for our vital organ, the brain, there is no similar table of growth and development that can be used as a reference. Dr. Jakob Seidlitz, a neuroscientist from the University of Pennsylvania's Department of Psychiatry and Children's Hospital of Philadelphia, was immediately unhappy that when he took his 15-month-old son to a pediatrician for examination, he could only measure the rate of development from a height and weight chart, "It's shocking that doctors know so little about the biological information about this vital organ [of the brain]. ”

A recent study by Seidlitz et al. may change that. On April 6, local time, a study published online by the top academic journal Nature showed that an international team of more than 200 research institutions drew a chart of the brain covering the entire human life cycle, spanning from a 15-week-old fetus to a 100-year-old person. This chart shows how our brains expand rapidly early in life and then slowly shrink as we age.

Seidlitz and Dr Richard Bethlehem of the University of Cambridge's Autism Research Centre are co-corresponding authors of the paper. It's worth mentioning, however, that this is the result of a research project spanning six continents, bringing together what may be the largest ever dataset of nuclear magnetic resonance (MRI) from 123984 NMR scans of more than 100 different studies around the world, with 101457 NMR participants.

The scale of the study came as a shock to neuroscientists. Scientists have had to contend with reproducibility in similar studies, in part because of small sample sizes and the high cost of magnetic resonance, which means that the number of people involved in experiments is often limited. Angela Laird, a cognitive neuroscientist at Florida International University, commented, "The sheer amount of data they collected is very impressive and really sets a new standard for the field." ”

However, the research team also reminds that the brain charts obtained so far are only a "first draft" and cannot be used directly in the clinic. Bethlehem explains, "Our brain graphing study is still in a very early stage, but this shows that it is possible to create these tools by integrating huge data sets. He believes that these charts are already starting to provide interesting insights into brain development, "In the future, as we integrate more data sets and refine the charts, they may eventually become part of routine clinical practice." ”

Email labs around the world: Collect MRI datasets

Although we roughly know that the human brain undergoes a long and complex growth process from the fetal period to the age of 30, it gradually ages from about the age of 60. But there are currently no charts like growth curves to quantify age-related brain changes.

The research team believes that mental illness and Alzheimer's disease already constitute the largest health burden in the world, which highlights the importance of standardized brain charts and provides a basis for standardized quantification of brain structures throughout the life cycle.

Seidlitz mentioned that making brain charts involves multiple technologies and a large collaborative team. "With brain imaging data, things are a little more complicated than just taking out a tape measure to measure a person's height or head circumference." But researchers face enormous challenges.

Everyone's brain structure is very different, and studies of the human brain often require collecting large amounts of scan data to create a statistically significant set of authoritative growth charts. Bethlehem says it's not an easy task.

In the study, the researchers didn't personally perform thousands of scans, which would have taken decades and cost prohibitively high. They took a different approach, starting with previously completed neuroimaging studies. Bethlehem and Seidlitz emailed researchers around the world asking if they would like to share their neuroimaging data for the project.

They were surprised by the number of replies. They believe this may be because the COVID-19 pandemic has left researchers spending less time in the lab and more time checking email inboxes than usual.

In total, the team collected 123894 MRI scans of 101457 people, including fetuses between 15 weeks of pregnancy and 100-year-olds; including the brains of people with normal nervous systems, as well as the brains of people with various disorders (such as Alzheimer's) and neurocognitive differences (such as autism spectrum disorder).

Bethlehem said, "One of the things we've been able to do through a global effort is stitch together data for the entire lifespan. It allows us to measure the earliest and most rapid changes that occur in the brain, as well as the long,slow decline that occurs with age. ”

The team used standardized neuroimaging software to extract data from MRI scans, starting with simple volumes of gray matter and white matter, among others, and then expanding details such as the thickness of the cortex or the volume of specific areas of the brain. Brain graphs of the human life cycle were then created using a generalized additive model (GAMLSS) recommended by the World Health Organization (WHO) to model the location, proportion, and shape of nonlinear growth trajectories.

Overall, they estimate that it has taken about 2 million hours of computing time to analyze data close to 1 petabyte (petabyte). "It's really impossible without access to Cambridge's high-performance computing cluster." Seidlitz said.

At the same time, he stressed, "We believe that this work is still ongoing." This is the first step in establishing a standardized reference chart for neuroimaging. That's why we built this website and created a huge network of collaborations. We want to keep the charts updated and build these models on top of the new data available. "They've launched a website where they intend to update their growth charts in real time as they take more brain scans.

First Brain Chart: Very early research

Specifically, the GAMLSS model fitted MRI data for four major tissue volumes of the brain, including total cortical gray matter volume (GMV), total white matter volume (WMV), total subcortical gray matter volume (sGMV), and total cerebrospinal fluid volume (CSF).

Diagram of the human brain.

Key milestones observed by the research team include a rapid increase in the total volume of gray matter in the cerebral cortex (GMV, brain cells) starting in the second trimester, peaking at age 5.9, followed by a near-linear decline. The study notes that this peak is 2 to 3 years later than previous reports. They argue that previously reported relying on a smaller sample with greater age restrictions.

Total white matter volume (WMV, brain connections) also increased rapidly from the second trimester to early childhood, peaking at age 28.7. The decline began to accelerate after 50 years.

The total volume of subcortical gray matter (sGMV, which controls body functions and basic behaviors) shows an intermediate growth pattern compared to the total volume of gray matter in the cortex (GMV) and the total volume of white matter (WMV), reaching a peak at age 14.4 years. The study said that both the total volume of white matter (WMV) and the total volume of gray matter under the cortex (sGMV) peaked in line with previous neuroimaging and autopsy reports.

In contrast, cerebrospinal fluid (CSF) shows a growth trend before the age of 2 years, followed by a stable state before the age of 30, then grows slowly and linearly, and grows exponentially at age 60.

The study also pointed out that from the perspective of individual differences, individual differences in total gray matter volume (GMV) in the cortex gradually increased early and peaked at age 4 years, while total subcortical gray matter volume (sGMV) variability peaked in late puberty, and total white matter volume (WMV) varied the most between individuals around the age of 40 years. Interestingly, cerebrospinal fluid (CSF) varies the most between individuals as it approaches the end of human life.

In addition, the research team also used THE GAMLSS modeling method to assess the development of the average cortical thickness, total surface area and volume of 34 cortical regions of the whole brain. The results were as expected, with the total area of the human cerebral cortex closely related to the development of total volume (TCV) throughout the life cycle, with both measures peaking at age 11-12 years. In contrast, cortical thickness peaked significantly at age 1.7 years, consistent with previous observations that cortical thickness increased during perinatal period and decreased later in development.

The team also found significant regional differences in peaks in neurodevelopmental trajectories in different regions. The age at which the gray matter volume peaked in the 34 cortical regions varied greatly compared to the peak of the total volume of gray matter (GMV) in the 34 cortical regions, ranging from 2 to 10 years. The primary sensory region reaches the earliest peak volume and declines more quickly after the peak, while the frontotemporal combined cortex region reaches the volume peak later and declines more slowly after the peak.

Overall, the volumetric peak of the ventral-caudal region of earlier maturation is faster after peaking, while the volume peak of the later-maturing dorsal-beaked cortex is slower.

The team emphasizes that it is worth noting that this pattern of spatial development trajectory reproduces the gradient from the "basic perceptual sense-joint cortex" that has been closely related to multiple aspects of brain structure and function.

Neurodevelopmental milestones.

It is worth noting that in the long run, the research team hopes to use brain charts as a clinical tool. For example, Alzheimer's disease can lead to neurodegeneration and loss of brain tissue, so people affected by the disease may have less brain volume than their peers. However, it is clear from the brain chart that as we age, the size of the brain naturally decreases, while Alzheimer's patients do so much faster.

"As you can imagine, they're being used to help assess patients with conditions like Alzheimer's, allowing doctors to spot signs of neurodegeneration by comparing how quickly a patient's brain volume changes compared to their peers." Bethlehem said.

Of course, the paper also clearly reminds that the study does not mean that the ultimate goal of quantitatively and accurately diagnosing an INDIVIDUAL patient's MRI scan has been achieved in clinical practice. Their brain chart study is still in a very early stage.

The team also hopes to make brain charts more representative of the entire population, which requires more MRI data on the brain. The brain scan data they collected, mostly from North America and Europe, disproportionately reflected white, college age, urban and affluent people. The study included only 3 datasets from South America and 1 dataset from Africa, accounting for about 1 percent of all brain scan data used in the study.

Sarah-Jayne Blakemore, a cognitive neuroscientist at the University of Cambridge, said this limited the universality of the findings.

"However, current work demonstrates the principle that building standard charts to measure individual differences in brain structure can already be achieved on a global scale and throughout life." The research team believes that this study provides an open set of scientific resources for the neuroimaging research community and will accelerate further progress in the standardized quantitative evaluation of MRI data.