Radiogenomics is Changing the Landscape of Cancer Research
Leveraging radiation therapy and genetic data to find more accurate results in cancer diagnoses, prognosis and personalized medicine.
How this blog is broken down:
- Introduction to cancer with me rambling philosophically (please feel free to skip this part haha).
- Discussing the current problem in cancer research.
- A solution of said problem.
Part I: The beauty and the beast in our bodies.
(skip this straight to Part II if you want to jump right into the problem)
Our biology is beautiful and dangerous.
Let’s take a moment to think about how beautiful being alive is.
Being able to breathe fresh air, smell sweet flowers, solve math problems, make friendships, feel the wet rain and so many more experiences transcend and wash over our senses everyday.
Wow being alive is crazy.
You start from a single celled organism to now a complex machine that has the capacity to think and feel?!
Humans are beautiful, if you don’t think so then take a closer look at the intricate designs on your fingerprints or the fluid movements you take to walk.
The biology that makes us who we are is truly an extraordinary gift.
But because we don’t understand it well enough or know how to properly take care of it, it can become dangerous.
Our body is a complicated system, with an average of 30–40 trillion cells all working together to make you a functional being. (And that’s excluding the bacteria that lives in/on us!)
That’s more cells in our body than stars in the observable universe.
Going to pull out some philosophy…
In taoism and many similar eastern philosophy or religions the body, mind and soul are interwoven into the tapestry of our universe. There is no separation between our physical bodies and the “energy” of the universe.
“Without leaving my house, I know the whole universe.” ~ Lao-Tzu
Within us lives a whole microcosm of a universe with life and death permeating our whole existence until our physical bodies die.
Millions of cells die and get reborn every second just like stars explode and get reborn from gas and dust.
The complexity of our biological system is as vast as our universe, we barely understand ourselves on a granular level.
Just like there are so many physics discoveries to be made about our universe, our body is a huge mystery.
The largest mystery that we haven’t been able to solve is cancer. When sometimes cells become corrupt and cause our own death.
Why is cancer so hard and complicated? A dive into chaos theory.
Chaos theory is extremely interesting to think about when viewing how cancer as a disease works.
It allows us to understand that things of chaotic nature will remain unpredictable but there are underlying laws that govern chaos and allow it to emerge in complicated systems.
If everything is one interconnected system, then randomness and unpredictability increases due to chaos but you can still find patterns that repeat themselves allowing us to find some pathways to prediction.
Okay if you don’t understand what I mean, let me break it down.
CANCER X CHAOS THEORY
Cancer as a disease can be seen as chaotic and unpredictable to a degree. We have definitely built many tools to help treat cancer but the fact it still exists in humans is evidence of the randomness that causes the problem.
TL;DR on what is cancer:
It is a genetic mutation or defect in a cell’s ability to die or divide in a stable manner.
Malignant: Cells divide uncontrollably, may enter the bloodstream and harm the functionality of our vital organs and very often lead to death if not prevented in time.
Benign: Cells divide uncontrollably however have very low chance of invading the rest of the body so mortality rate is much higher.
It’s a hard problem for two reasons:
- No one solution exists for cancer.
- There is no way to pinpoint exactly why someone develops cancer.
Chaos emerges because its a system with many initial deterministic parameters (genetic mutations, family history, environment, lifestyle, geography, etc.) that dramatically affect the randomness of the outcome of the disease so it becomes unpredictable.
Cancer becomes a complicated, chaotic disease that is hard to predict and decipher. But this doesn’t mean we can’t build smarter tools to augment our understanding of the complexity of cancer.
Part II: The current problem in research and data modeling.
The economic cost of cancer globally exceeded $1 trillion in 2010 and it has been growing ever since.
Every year billions of dollars have been donated to cancer research in finding a cure but we haven’t made drastic progress in finding a solution based on how much money is spent.
This is because our approach to cancer research has become inefficient and redundant. We need new approaches to tackle this complicated problem and look at it from a different angle.
How research is currently done:
a) Cancer is broken down into smaller sub-areas or specific types of cancer to study and test in a lab.
b) Millions of dollars is spent doing slow, methodical, step-by-step processes of trying to find drugs/treatments.
c) 95% of all laboratory discoveries fail in clinical trials of humans.
(You can view stats behind failure rate of clinical trials here.)
d) Of the 5% of treatments that pass clinical trials, only 51% of these are affordable enough for the common man and this number has been decreasing since last recorded in 2016.
The problem: This is an inefficient process that isn’t returning enough results in accordance to the money spent in research.
It’s starting to look like people are just putting money in a black hole without looking into the actual output of their investment.
I’m not disqualifying the importance of lab testing and research, that will always be a fundamental part of finding a cure for cancer. However, there must be a method understand cancer in a different way, become more preventive over a reactive disease.
A study by Stanford university provides evidence that cancers can grow in your body undetected for over 10 years before they are discovered in our most sophisticated tests.
This is absolutely CRAZY!
Cancer research is not accelerating at the rate needed to find a cure, we have to look at different approaches to tackle this problem.
Whenever there are inefficiencies in a process, it always means there are opportunities to make it better.
The opportunity: Lab tests generate a lot of valuable data that has a lot potential to be used in a manner to map the complexity of cancer in models.
Bioinformatics and computational biology are becoming more important tools to understand cancer.
Data modeling is becoming a vital tool in how we understand the complexity in cancer and an integral part of research.
Detecting and treating cancer early significantly impacts the survival rate of the victim.
Cancer is a fast moving disease that evolves over time because genetic mutations add up and cells continue becoming more and more damaged with each replication cycle.
This means that cancer in its early stages will be very different weeks or months laters as it evolves into a more deadly situation.
Using computational tools we can track the progression of a certain type of cancer at a better rate
These tools have potential to become more precise and help cancer as a disease becomes more preventative and less reactive.
Part III: Radiogenomics as a tool to tackle cancer.
Radiogenomics is the intersection of radiology (cancer imaging) and gene expression data to find meaningful insights in cancer diagnoses and treatment.
We are lucky to be alive in a world where biomedical research enables us to collect vast amounts of data from cancer research labs.
Data can enable us to parse through the complexity of cancer at a deeper level if we use the right tools to examine it.
But not all of this data is complete and characterized to help uncover knowledge related to improved patient outcomes who suffer from cancer.
New tools like radiogenomics could help tackle this problem.
TL;DR of Diagnostic Radiology
Definition: Using radiation imaging to understand the inside of your body better and define its structures.
Basically help shed light on what the heck is happening inside someone’s body because that information is many times hidden on the outside.
Examples:
- Screen for tumours or other abnormalities caused by various diseases.
- Monitor the response your body is having from treatments
- Diagnose the cause of your sickness which can’t decipher just by looking at you
Some common methods of diagnostic imaging is CT scans, MRI, x-rays, etc.
TL;DR of Genomics Data
Definition: It is the genetic data extracted from a genome about the DNA structure, single or multiple gene expression, patterns in gene expression, mutations and other various characteristics.
Since the launch of projects like The Human Genome Project in the 1990’s we have developed many tools and inventions that can breakdown the way our genetic code plays a part in our everyday functions.
This is extremely valuable to cancer research because it fundamentally is a disease caused by irregularities in our genetic code.
Breakthrough technology has allowed us to gain insights into biomarkers (molecules) that can point to abnormal processes happening within our cells.
Examples of biomarkers:
- Proteins
- Gene mutations (changes)
- Gene rearrangements
- Extra copies of genes
- Missing genes
- Other molecules
Now imagine marrying these 2 technologies together!
Genomics + Imaging = Radiogenomics
So radiogenomics is trying to give us wider scope and knowledge about cancer by matching its imaging features with the underlying genetic processes.
Both techniques of diagnosing and understanding cancer provide different insights.
Genetic data is captured from very small samples of tumour tissue and then looked at as an average by dividing the tumour into smaller regions. However this process is does not reflect the heterogeneity of cancerous tumors.
This means that some parts of the tumour could have vastly varying genetic information because tumours don’t have all the same cells in them making it heterogeneous.
Imaging technology can capture parts of the heterogeneity of tumour cells and give insights into the phenotypes, characteristics more related to the physical features of the tumour and how it’s growing.
Leveraging both sources of data can help build better prediction models of patient outcomes and reveal hidden insights about tumour progression and response to treatment plans.
Radiogenomics provides a pathway to take limited datasets and acquire useful knowledge about new areas of cancer research.
What to consider when developing a radiogenomics approach:
Radiogenomics focuses on creating association maps to find relationships between diverse datasets in which it can find correlations between imaging phenotypes or genetic biomarkers.
3 questions to consider when developing radiogenomic maps:
a) What problem or question do want answered, will it benefit largely through the examination of various different datasets?
b) Are the right resources available, do you have enough data to build strong models of cancerous tumours?
c) Is there already a well defined methodology in creating an association map that can be leveraged to answer the question?
Machine learning is applied throughout the model for better predictive classification.
However since this field is still developing there are limitations in it’s accuracy due to result of “good” data and efficient machine learning models which are broader and transferable to different types of datasets.
Applications:
- Understanding the biological correlations behind image phenotypes.
- Understanding how a biological process is reflected in imaging.
- Defining biomarkers or biological substitutes that attempt at solving unmet clinical outcome related questions.
3 Main key takeaways from this article:
- Cancer is a complicated, heterogenous problem and requires more sophisticated tools or algorithms to understand more granulary
- Cancer research is not developing solutions at the rate we need it to so there is a large need for computer aided softwares to help breakdown the complexity of cancer and arrive at novel research breakthroughs
- Radiogenomics is a new tool that combines imaging and genetic data to approach cancer tumours in a holistic way and uncover new insights about patient outcomes and help increase their survival rate.
I love learning about complicated problems in the world with large doses of complexity and this is my first take at understanding cancer and how we can overcome it.
Stay tuned for more articles and a guide on how to create your own radiogenomic models at home!
Thanks for paying attention, follow my medium for more articles and connect with me on linkedin and twitter. ❤️
