Early Depression Risk: How biology and experience shape teen mental health
- Zuzanna Zajkowska
- 3 hours ago
- 5 min read
Back in 2023, we published a scientific article, along with a blog version, showing that the body’s immune response is linked to adolescent depression, and that these biological signals look different in boys and girls. We ended by asking whether bringing together what we know about adolescents’ backgrounds and their biology could help us identify who is at risk of developing depression.
And here we are today, trying to answer that very question.
If you find yourself wondering why it took us two years to get here, well, that’s because things in research take time (something I’m sure my fellow researchers can relate to). But patience does pay off, and I hope I can invite you, my dear reader, to be curious as to why.
Let’s dive into the story.
The Question We Couldn’t Yet Answer
The article I mentioned was part of the IDEA project, which stands for Identifying Depression Early in Adolescence, and is supported by MQ Mental Health. The main aim of the project was to understand which adolescents are at increased risk of developing depression, with a particular focus on low- and middle-income countries (LMICs).
I am a postdoctoral researcher at King’s College London, and I was very fortunate to be part of the IDEA project led by Professor Mondelli and Professor Kieling.
As part of IDEA, we developed a composite risk score (IDEA-RS) based on 11 sociodemographic factors, such as early life stress, family relationships, and substance use, to help us see if we could predict who would develop depression at an early stage. Over the course of three years, we used this score to recruit and study 100 adolescents (known as the IDEA RiSCo cohort), of whom 50 were scoring very high and 50 scoring very low at the IDEA-RS. Our recent article showed that the IDEA-RS was successful in predicting which adolescents would develop depression within three years of assessment.
Looking for the Answers
With that in mind, we set out to answer this question: Can we improve this prediction even more by combining sociodemographic risk with neurobiological markers? And if yes, could we create a biological “risk score” that complements the existing one?
We wanted to add neurobiological markers such as increased inflammation or changes in the activity of certain brain areas like the amygdala - a key threat-processing region - to our prediction score because we know from existing research that some biological changes are linked to depression. We also know that the kynurenine pathway is linked to depression. This is the process by which the body breaks down tryptophan (important in mood and sleep regulation) into chemicals such as kynurenic acid and quinolinic acid, which can either protect (kynurenic acid) or harm (quinolinic acid) the brain. Last year, we published an article (and a blog) using the same IDEA RiSCo cohort to show that an imbalance in this pathway was already present in adolescents at risk for, or experiencing, depression, particularly in girls.
Taking all that into account, in this study we set out to look at inflammation in the body – a part of the immune system’s natural response to stress, illness, or injury. We measured this using blood proteins called cytokines, which act as messengers in the immune system. Some cytokines, including interleukin (IL)-2, IL-6, IL-12p70, and tumour necrosis factor (TNF)-α, have been linked to depression in previous studies, particularly when their levels are persistently elevated.
Next, we looked at brain function using a type of brain scan called functional MRI (fMRI). This allows us to see how active different parts of the brain are while someone is performing a task. We focused on a small brain region called the amygdala, which plays an important role in detecting and responding to emotional signals, especially fear and threat. We measured how strongly the amygdala reacted when adolescents viewed faces showing fear, sadness, or anger.
Finally, we examined a biological process known as the kynurenine pathway. As before, we looked again at the kyneurine pathway - this time, at the balance between the protective kynurenic acid and the harmful quinolinic acid. The ratio between them (known as the KA/QA ratio) gives an indication of whether the system is more tilted toward protection or vulnerability. A lower ratio may suggest reduced neuroprotection. After all of these baseline assessments, we followed adolescents for three years to see if any of them developed depression.
What Did We Find?
On its own, the IDEA-RS could discriminate who would later develop depression with moderate accuracy. That was helpful, but was it better when we added neurobiological markers? The short answer is yes.
When we started adding biological information, step by step, prediction kept improving.
The most important result came when we combined everything together: the sociodemographic risk score (IDEA-RS) plus all eight biological measures (the four cytokines, the KA/QA ratio, and the three amygdala reactivity measures). That combined model was able to discriminate adolescents who would develop depression with substantial accuracy - we shifted from moderate to excellent prediction. In other words, biology meaningfully sharpened the picture.
A Biological Risk Score: Making it Rractical
Statistics are useful, but if the long-term goal is a tool that can inform prevention, we also need something that is simple to apply and easy to replicate. So, we developed a biological risk score, called the IDEA-BIO-RS.
We combined all the biological markers into one overall biological risk score, allowing us to group adolescents as biologically lower or higher risk for developing depression.
Even with biology alone, the difference was striking: in the biological high-risk group, 36% developed depression over three years, while in the biological low-risk group only 3% did.
But the most clinically meaningful results emerged when we combined biological and sociodemographic risk.
When Two Kinds of Risk Agree, Risk Becomes Clearer
We grouped adolescents into four categories: low risk on both scores, high risk on both, or high risk on one but not the other.
Among adolescents who were low risk on both the sociodemographic score and the biological score, none developed depression during follow-up. At the other extreme, among adolescents who were high risk on both, 44% developed depression within three years. Those who were high on one score but not the other fell in between.
This pattern matters because it suggests two things at once. First, the combination can help identify a group where risk is high enough that targeted prevention could be justified. Second, it may also help identify a group where risk is very low, which is equally important if we want screening tools to be more precise and avoid unnecessary anxiety.
What This Does and Doesn’t Mean
What this study shows is that integrating biology with sociodemographic context improves prediction in a meaningful way, and that a relatively simple biological risk score can complement an existing sociodemographic model.
It’s equally important to say what this study is not. It’s not evidence for a single biomarker of depression. Depression is too complex and varied for that. Different people may reach similar symptoms through different pathways. That is precisely why composite scores are so appealing - they take into account the complexity of real life and its multidimensional nature.
Why I Find This Hopeful
In mental health, prevention often feels like something we talk about more than something we do. Studies like this are one way to make it more concrete. If we can identify adolescents who are likely to develop depression within a few years – not perfectly, but better than chance and better than sociodemographic risk alone – then we can begin building prevention pathways that are timely, targeted, and fair.
And I think that is why it was worthwhile to wait a couple of years. My hope is that, with a little patience, we can see this making real difference in people's lives.
