by Jackson Weir
figures by Jasmin Joseph-Chazan

Why is cancer so difficult to cure? Why do available treatments only help a subset of patients? Why are some cancers more aggressive than others? These are questions that clinicians, scientists, and the public have pondered for generations. As it turns out, the answers are complicated because cancer biology is complicated. Luckily, new tools and technologies are helping us dissect the disease with unprecedented resolution, launching a new era in cancer research. Before we explore some of the latest tools changing how we study cancer, let’s first examine what makes cancer so complex.

Layers of cancer complexity

Human biology is a product of intricate layers of biological information. Atoms arrange into molecules, molecules interact to build cells, cells organize to form tissues, tissues coordinate to create organs, and organs harmonize into the beings we call humans. Unfortunately, cancer – a disease of uncontrolled cell growth – has pathology that permeates each of these layers of human biology. A harmful change, or mutation, in our DNA (the genetic instructional code of life) can alter how our molecules interact. These molecules then build dysfunctional cells, which can no longer organize properly into tissues. Instead, the abnormal cells proliferate into tumors, which can disrupt organ function, and ultimately, our health and wellbeing.

Consider an example of how cancer can manifest itself at different layers of organization (Figure 1). Smoking cigarettes may cause an accumulation of harmful mutations in the DNA of lung cells. These mutations may suppress cell growth control circuits within a mutated cell, initiating uncontrolled proliferation and subsequent tumor formation. The resulting mass of cells in the lung can result in inflammation, fluid build-up, and shortness of breath. After some time, cancer cells from the lung may spread to other organs in the body, leading to organ dysfunction beyond the lung and eventually system-level failure.

Figure 1: Different layers of organization affected by cancer. (1) Patient suffering stems from (2) organ dysfunction. Organ function is determined by (3) tissue organization. Tissues are formed by (4) cells, and cells are constructed from (5) molecules.

Heterogeneity at every level of organization

Heterogeneity in cancer presentation across each level of biology and between patients is part of what makes cancer so difficult to cure. At the molecular level, any number of genes could be mutated in a cancer-causing manner. In fact, the vast majority of the 20,000 genes in the human genome have been associated with cancer. Some of these mutations can be caused by patient exposure to different chemicals. Returning to our lung cancer example, smokers accumulate different signatures of mutations in their lungs compared to non-smokers as a result of exposure to tobacco. These distinct mutational signatures result in different genes becoming dysfunctional, and ultimately, separate trajectories of disease progression. At the cellular level, mutations can impact various pathways that determine normal cell function. For example, one study found that smokers are more likely to acquire a specific mutation in a gene known as KRAS. This mutation hijacks a normal cell growth and differentiation pathway in cells, leading to overactivation of growth signals and abnormal cell proliferation. Similarly, many other cellular pathways can become hijacked by different mutations, leading to unique cell abnormalities and growth patterns that cause cancer.

At the tissue level, the immune system’s involvement adds even more complexity. In addition to fighting viruses and bacteria, immune cells hunt and kill cancer cells. Depending on how well the immune system can detect the presence of cancer cells within tissues, a tumor might be considered “hot” or “cold”. Lung cancers are generally considered to be “hot” tumors because many immune cells infiltrate the cancerous region of the lung. Cancers that emerge from other tissues, such as the pancreas or the brain, are generally “cold” tumors, as they have less immune infiltration. We have known for a while now that the composition of immune cells surrounding tumors are predictive of prognosis and treatment response, with “hot” tumors generally responding better to immune-stimulating therapies. However, studying which immune cells surround which tumor cells, how they interact, and what role these interactions play in disease has remained challenging.

Spatial transcriptomics collects information across levels of organization

For a long time, cancer research was limited to studying single layers of biology in isolation. Molecular biologists studied the role of proteins and genes in cancer, cell biologists studied the difference in cellular mechanisms between cancer cells and healthy cells, and pathologists examined how tumors manifest themselves in tissues and organs. While these fields worked together to advance our understanding of cancer, it has historically been difficult to simultaneously study the molecular, cellular, and tissue levels of cancer. Newer methods can preserve cellular integrity, but they still lose positional information of cells within tissues. If we could scalably study what molecules are present in what cells, and where these cells are positioned in tissues, we could better unravel cancer pathology. 

Enter spatial transcriptomics – a set of new technologies that can collect information on the molecular, cellular, and tissue levels all at once! Transcriptomics tools work by measuring levels of RNA in a cell. RNA levels are proxies for gene activity, as RNA is a transcribed “copy” of the DNA genetic code used as instructions to build proteins, which then build the parts of cells. Spatial transcriptomics tools measure RNA levels in cells while preserving the spatial context of where these cells are in tissues.

There are generally two types of spatial transcriptomics methods: sequencing-based and imaging-based. Sequencing-based approaches (Figure 2) capture RNA molecules onto a glass slide, give each RNA molecule a spatial barcode (a short sequence of nucleic acids associated with a location), and sequence these RNA molecules to read both their identity (what gene the RNA came from) and spatial barcode (location of the RNA in its resident tissue) at once. Imaging-based approaches mount tissue sections on glass slides, add fluorescent probes that bind a predetermined list of RNA molecules, and use microscopy to determine the location and abundance of RNA molecules in tissue by the intensity of fluorescent probes. Altogether, these new techniques allow us to study how active genes are (molecular level), what cells they are in (cellular level), and where these cells are in space (tissue level), opening the door to unprecedented resolution in cancer research.

Figure 2: Sequencing-based spatial transcriptomics. Tissue sections are placed onto slides with spatially barcoded spots. RNA molecules are captured out of the tissue, given a spatial barcode, and sequenced. The sequencing data can then be reconstructed to learn molecular information (RNA levels), cellular information (cell types), and tissue information (tissue structure and organization) from samples.

Spatial determinants of cancer

While spatial transcriptomics tools are still relatively new technologies undergoing constant technological improvements, scientists have begun applying them to answer pressing questions in cancer biology. Take the tissue level of complexity, for example, where immune cells interact with cancer cells to shape disease trajectories. As it turns out, these interactions are spatially localized, occurring only in specific regions within cancerous tissue. Using spatial transcriptomics, we can study what types of immune cells localize with what types of cancer cells in a tissue, and what types of cells do not. We can also learn if there are differences between cells that are in close proximity to cancer cells and those that are farther away. By identifying potential interactions between cancer cells and surrounding cells, we can use this information to design drugs that promote the killing of cancer cells by the immune system.

A recent study did just that to study pancreatic cancer, a notoriously difficult disease to treat. The researchers identified three multicellular communities of cells across pancreatic cancer patients, each consisting of different mixtures of cancer cells, immune cells, and fibroblasts (cells involved in connective tissue formation). By integrating molecular, cellular, and tissue level information, the scientists were able to identify critical molecular interactions between nearby cells in these communities that may serve as promising targets for drug development.

Future perspectives

Spatial transcriptomics technologies are just the beginning of a changing tide in cancer research. New spatial methods collecting more information at the molecular and cellular level, with improved spatial resolution at the tissue level, are in development. Scientists are also working on clever ways to integrate temporal information into spatial tissue measurements. Hopefully, new spatial technologies will help us elucidate the complexities of cancer across space and time, opening the door to answering questions that have puzzled scientists for generations.


Jackson Weir is a second-year Ph.D. student at Harvard University in Fei Chen’s lab. He develops new spatial tools to study cancer.

Jasmin Joseph-Chazan is a fourth year PhD student in the Immunology program at Harvard Medical School.

For More Information:

  • Read about spatial transcriptomics being named Nature’s Method of the Year here.
  • Learn how scientists are using spatial transcriptomics to study the tumor microenvironment  here.
  • Read more about how spatial technologies are revolutionizing cancer research here.

This article is part of our special edition on diversity. To read more, check out our special edition homepage!

Leave a Reply

Your email address will not be published. Required fields are marked *