A clinically annotated proteomic and histological resource of 1,000 colorectal cancers
This study presents the largest multimodal resource of 1,000 treatment-unexposed colorectal cancer cases, integrating deep proteomics, genomics, and histopathology to define six clinically relevant molecular subgroups that improve patient stratification and establish benchmarks for computational pathology.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Colorectal cancer is a disease of great variety. Even when two patients have tumors that look identical under a microscope and are at the same stage of spread, their bodies can react to the disease in completely different ways. One might respond well to treatment and live for many years, while another might face a rapid decline. For decades, doctors have relied on a standard system of staging that measures how deep a tumor has grown and whether it has spread to lymph nodes or distant organs. They also check for a few specific genetic changes, such as whether the tumor has a defect in its DNA repair system. While these tools are useful, they often fail to explain why patients with seemingly similar cancers have such different outcomes. The tumor is not just a single block of cells; it is a complex ecosystem where thousands of proteins work together to drive growth, resist drugs, or interact with the immune system. To truly understand why one cancer behaves differently from another, scientists need to look at this entire protein landscape rather than just a few isolated clues.
A team of researchers has now created the most detailed map of this landscape to date. They gathered tissue samples from 1,000 patients with colorectal cancer who had not yet received any treatment. For every single patient, the team collected three types of information: a deep analysis of the proteins inside the tumor, a scan of specific genes known to drive the disease, and a high-resolution digital image of the tumor's physical structure. By combining these three views, they were able to sort the 1,000 tumors into six distinct groups based on how their proteins were working together. This approach revealed that the tumors are not just a few broad categories, but a spectrum of functional states that predict how the disease will progress and how the patient will respond to chemotherapy.
The most significant discovery was that the researchers could split the largest group of tumors—those with chromosomal instability, a common feature where cells have many errors in their DNA structure—into three separate subgroups. Previously, these were often treated as a single category. The new analysis showed that one of these subgroups behaves in a very slow, indolent manner and is associated with a favorable outcome for the patient. Another subgroup, however, was found to be rare but extremely aggressive, with patients in this group facing a steep decline in survival after the second year of diagnosis. The study also confirmed two other groups: one that is rich in immune activity and another that is dominated by structural proteins and is often resistant to standard treatments.
These six groups are defined by coordinated programs of proteins, meaning that the behavior of the tumor is driven by a complex network of molecular signals rather than a single genetic mutation. The researchers found that looking at just one or two genes was not enough to tell these groups apart. Instead, the tumors had to be classified by analyzing hundreds of proteins at once. This distinction is crucial because it explains why some patients with the same genetic mutations have different fates; their tumors are running on different protein programs. For instance, the aggressive subgroup was found to have a unique signature involving energy regulation and a loss of specific signaling pathways, setting it apart from all other types of colorectal cancer.
The study also explored whether these complex protein patterns could be predicted simply by looking at the tumor's appearance under a microscope. The researchers trained computer models to analyze the digital images of the tissue slides. They found that the models could reliably identify the broad categories of the disease, such as whether the tumor had a defect in DNA repair or whether it belonged to the chromosomally unstable group. While the models were less precise at identifying the six specific protein subgroups, they performed significantly better than when trying to guess individual genetic mutations just from the image. This suggests that the physical shape and texture of a tumor hold clues about its internal molecular machinery that are invisible to the naked eye but detectable by artificial intelligence.
Perhaps most importantly, the researchers linked these protein groups to how patients responded to chemotherapy. They found that patients with the large, chromosomally unstable tumors generally benefited from chemotherapy, while those with the immune-active group showed a borderline benefit. In contrast, patients with the subgroup rich in structural proteins saw no survival advantage from the treatment. This finding helps explain why some patients do not improve with standard care and suggests that future treatments might need to be tailored to these specific protein groups rather than just the general stage of the cancer.
The team made all their data, including the protein measurements, genetic profiles, and thousands of digital microscope images, available to the scientific community. This resource serves as a new reference point for understanding colorectal cancer. It moves the field beyond simple staging and single-gene testing toward a more nuanced view where the functional state of the tumor is the primary guide for treatment. By revealing that the disease is composed of these six distinct functional states, the study provides a clearer path for developing targeted therapies and for understanding why some cancers are stubbornly resistant to treatment while others are surprisingly manageable. The work does not offer an immediate cure, but it provides a much sharper lens through which to view the disease, turning a confusing mix of outcomes into a structured map of biological possibilities.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.