With ambition abound, ICGC-ARGO marks the beginning of a new era of genomic oncology where pathology goes far beyond diagnosis and pathologists, like Mark Rubin, will be interrogating tissue samples to even greater depths.
Simply put, the promise of precision oncology will only be fully realized when all available data is used to discover subtle differences between tumors. That includes the wealth of digital images now routinely captured in pathology for clinical care, which, together with genomic data and information on clinical outcomes, can help clinicians get the full picture of their patient’s cancer.
As is the strategy of ICGC-ARGO, there will also be more samples for pathologists to analyze than ever before, but there’s obvious strength in numbers.
Where the ICGC 25K Initiative sequenced ‘static’ tumor samples collected at diagnosis, ICGC-ARGO will collect multiple tumor samples from every donor, from their diagnosis through and after therapy, to understand how cancers change and grow with time, and with treatment.
Digital images of every tissue sample submitted will also be collected, curated and linked to clinical and genomic data through the data platform.
Altogether, it will provide an unprecedented opportunity for pathologists to develop new insights into the classification of cancer.
“In ICGC-ARGO, we’re moving from genomic studies on mostly untreated patients to patients who are now undergoing standard of care cancer therapy,” says Professor Rubin, who leads ICGC-ARGO’s Pathology Working Group and Swiss Precision Oncology Program.
“This is a tremendous learning opportunity for pathologists,” he continues. “Because [in routine pathology] we mostly examine biopsy samples taken from localized cancer treated with surgery, and we don’t have a long experience evaluating metastatic samples taken after targeted therapy.”
Assembling digitized collections of sequential samples will, for example, enable pathologists to examine finer features of tumor morphology, such as the shape of cells’ nuclei, across larger cohorts than what was possible before, and especially in rare cancers and subtypes. This could refresh fundamental understandings about how tumors change as cancers progress.
“These are the types of things that could be captured in pathology reports [and images] but not yet utilized in genomic studies,” Rubin says. “Moving forward, we’d like to capture some of these nuances which play an important role in cancer pathology.”
It will require an integrated approach, Rubin says, with pathologists embedded in every ICGC-ARGO project correlating changes in tumor pathology with genomic patterns and clinical data. Synoptic reporting will also be indispensable to standardize the information described in pathology reports so it can be analyzed en masse. “Obviously, this will take a lot of coordination,” he says.
The long-term vision of the Pathology Working Group also involves applying artificial intelligence to pathology data. Working closely with computational scientists at ICGC-ARGO’s Data Coordination Centre, the group is planning to develop machine learning algorithms capable of extracting information from digital images, and identifying patterns among them for pathologists to interpret.
This approach could reveal new ways for pathologists to categorize tumors, and more effectively diagnose cancers. This is where genomics has previously excelled: in teasing out the genetic differences between subtypes of cancer thought to be one and the same. With AI, pathologists could stratify cancers further or detect similar distinctive features between them.
“We can’t possibly anticipate now what we’re going to discover,” Rubin says. “But I would imagine that we’re going to discover important subclasses that were missed using classical pathology approaches.”
It’s not just rare, understudied cancers which stand to benefit, Rubin says, but the approach could also redefine common cancers, where many questions still remain. For example, it is unclear why some tumors which lack a classic mutation profile still respond to targeted therapies, and yet other patients with known signature mutations, don’t.
“That’s really at the core of precision oncology, because we want to be able to identify patients who are most suitable to certain types of treatments and predict who might not respond,” says Rubin, echoing the ambition of ICGC-ARGO.
And he knows what it takes to deliver true precision cancer care and implement newfound discoveries into clinical practice.
Rubin, along with colleagues at Weill Cornell Medicine, developed the first whole exome sequencing test approved in New York State for detecting mutations which can be used to guide treatment decisions for patients with advanced cancers who have few treatment options left. The test, validated against nearly 60 different tumor types, involves screening more than 21,000 genes for actionable mutations, providing far greater coverage than targeted sequencing tests or small diagnostic panels which only capture a few genes at a time.
In much the same way, pathologists joining ICGC-ARGO will be a vital link between research and clinical care, Rubin says. He urges his fellow pathologists to get involved, because in time, they can help to implement new diagnostic standards and molecular tests born out of ICGC-ARGO in their own pathology departments. Returning results to the clinic is, after all, the only way to transform and ultimately personalize clinical care for people with cancer.