By Deborah Borfitz

July 17, 2025 | In a bid to demonstrate the promise of multiomics in enhancing oncology clinical trials and real-world cancer care, researchers in Switzerland have had a technology-intensive project underway for eight years now that has started to bear fruit. Their latest feat was to show the value and clinical feasibility of a new kind of trial using diagnostics to predict the most effective therapy for treating melanoma rather than testing individual drugs. 

Nine different molecular biological technologies were used in parallel to measure the properties of tumors biopsied from 116 patients, enabling a precise treatment decision in four weeks (Nature Medicine, DOI: 10.1038/s41591-025-03715-6). The technical turnaround time with the platforms was a brief two weeks and in 75% of cases the resulting recommendations informed the therapy choice of treating specialists, according to Andreas Wicki, M.D., Ph.D., professor of oncology at the University of Zurich and senior physician and deputy director of the clinic for medical oncology and hematology at University Hospital Zurich (USZ). 

The project was conducted at the department of dermatology at USZ, the department of oncology at University Hospital Basel, and the department of oncology and hematology at Kantonsspital Baselland. The three hospitals belong to two of five university healthcare centers serving Switzerland’s populace of 10 million.   

The first-of-its-kind study is part of a larger prospective, nonrandomized observational project of the Tumor Profiler research consortium designed to assess the relevance of functional, single-cell and bulk omics readouts to molecular tumor board decision-making, he says. It follows a similar exercise in a small cohort of acute myeloid leukemia patients, which succeeded in demonstrating how the approach can reveal drug resistance mechanisms and treatment vulnerabilities (Nature Communications, DOI: 10.1038/s41467-024-53535-4). A paper specific to ovarian cancer is also expected to be published soon, Wicki reports. 

In the latest study, better response rates and progression-free survival were seen with the treatments driven by tumor profiling than in a non-randomized, combined exact- and propensity-matched comparator group of patients who did not participate in the program, he says. A deeper dive into the data suggested that “the benefit to the overall population is very limited, but for a couple of patients there was a huge benefit.” These were individuals who had exhausted standard-of-care therapies and went on to a specific, biomarker-based therapy and were still alive after four years.  

The data science work has yet to be done to learn how the many biomarkers interact, for example, to reveal the mechanisms of “synthetic lethality” whereby a combination of two genetic events results in cell death, says Wicki. This would create a poly-marker-based model that could eventually be deployed in clinical trials. 

Currently, the second phase of the project is underway at all the country’s university hospitals for five of the most prevalent solid cancers. These include melanoma as well as cancers with less tumor tissue available for testing—breast, colorectal, lung, and ovarian cancers. 

Tumor Analysis 

Since all Swiss university hospitals introduced next-generation sequencing (NGS) around 2015, the following year they began thinking about what else might be done to further improve cancer diagnostics, monitoring, and therapy, Wicki says. That is when their representatives first met to come up with the plan to build Tumor Profiler, which involves a collaboration between academic institutions—USZ, University of Zurich (UZH), ETH Zurich, and University Hospital Basel—and Roche.  

“We’re basically technology-agnostic,” he continues. Tumor Profiler is open to using any platform that can meet a series of criteria, including importantly the tight timeline for producing results. 

One of the big takeaways from the subsequent study is that “at least at an academic center, physicians are ready to take up the information and act on [it],” says Wicki. As a practical matter, this happens meaningfully only when there’s a choice “between standards or beyond standards” versus when there is only one option to consider. 

The information obtained from the nine technologies—single-cell genomics and transcriptomics, targeted spatial proteomics, cytometry by time of flight, mass spectrometry proteotyping, drug phenotyping, iterative indirect immunofluorescence imaging, targeted next-generation DNA sequencing, and digital pathology—was intended to create a comprehensive picture of the biological processes in tumors. Of these, only NGS and digital pathology are part of current diagnostic standards, Wicki says. 

There are also cost considerations, although the price of multiomics is coming down. When the project began in 2018, the cost for the first patient was about 140,000 Swiss francs (roughly the same in U.S. dollars), but this has since dropped to approximately one-tenth that level, he adds. 

Moreover, all those technologies may not be needed by every patient. In a simulation exercise, researchers showed that “for most patients it could be sufficient to have two, three, or four technologies,” says Wicki. The question that will ultimately need addressing is which technologies may be redundant and therefore make the overall approach less cost intensive. 

Markers of Response

Time-wise, the study was able to show that the recommendations provided by tumor profiling were available after four weeks—fast enough for most patients with cancer—inclusive of one week on the front end for doing the biopsy work and another week on the backend when the case was discussed at the tumor board by multidisciplinary cancer experts, says Wicki. In between was the technical turnaround time for the multiomics testing.  

The focus was on how long it would take for the tumor analysis to be available and how treating physicians assess the individual treatment recommendations, which were derived from 43,000 data points per sample, he adds. Tissue samples from patients were sent out for analysis by the various technology platforms located at different recognized specialty centers around the country. 

It was the first time treatment predictions made by the nine technologies were used in a real-world healthcare setting. From the large pool of resulting multiomics data, recommendations were based on 54 individual markers known to be associated with tumor response to certain therapies, says Wicki. It remains to be sorted out how those biomarkers may impact one another.  

Using a kind of reverse engineering, the research team tried to draw conclusions from the biology back to the drugs allowed to be used clinically, he explains. A molecular summary report, based on the different technology-specific recommendations, was created from raw data. 

Most of the 43,000 biomarkers were not used simply because “we don’t understand what the data points mean, and we don’t know whether they matter,” says Wicki. Machine learning models will be needed to leverage all the data in predicting therapy outcomes, and this will be a future focus of the Profiler consortium.  

The long-term hope here is to personalize the choice of technologies for patients and their cancer. Another project being run with the Swiss Institute of Technology (EPFL) in Lausanne is looking at the question of how to optimize the diagnostic workflow, he notes. 

The process begins with tissue slides stained with hematoxylin and eosin, a “very simple and very cheap” fundamental laboratory technique, which machine learning can already use to predict the presence of microsatellite instability—signifying a defect in the DNA mismatch repair system—in more than 90% of cases, says Wicki. Bayesian statistics could significantly aid in the efficient ordering of tests independent of the Tumor Profiler project. 

Multiomics Trials

In Switzerland, 180 oncology drugs have been approved and available for clinical use. A trial that tests predictions rather than drugs is important because of the many molecular subtypes of cancer that need to be represented in phase 3 clinical trials, making them complex and lengthy to complete, says Wicki. 

The idea is not to replace the drug trials, but to deploy the diagnostic-based treatment predictions in that context, he continues. As it is, between 5,000 and 6,000 compounds are in the oncology pipeline with 90% of them expected to fail, which still leaves 500 to 600 new drugs needing to find their role in clinical oncology the next decade or so. 

“There is no way we can test all those drugs in sequence or in parallel in many different molecular groups of patients,” says Wicki. “That’s the clinical need and it is also sort of the scientific need because... you will never have two samples of the [exact] same cancer genotype in your lab.” 

Among the advantages of the treatment predictions is that clinical trials could stop seeking to have a largely homogenous population, meaning more patients would be potentially eligible to participate, Wicki continues. The practical hope for clinical care is the avoidance of “running therapies through a process of trial and error. Just increasing the probability of response from 60% to 80% to 70% to 90% will have a considerable impact on patients, and on costs,” he adds, referencing an analysis currently underway. 

Now that the bottleneck of producing multiomics data is “almost gone” in terms of the affordability and turnaround time, he says, attention is turning to standardization of the clinical testing and bringing the insights into randomized controlled trials for formal assessment. “Probably the benefit at the beginning is going to be smaller,” says Wicki, much like the early days with antibiotics when they were effective against a limited range of bacteria.  

Like broad-spectrum antibiotics in subsequent years, multiomics trials will ultimately emerge that could scale the pace of discoveries, he says. “I don’t think it’s improbable to see at least the first of those trials with a smaller level of benefit running is the next five years.” 

In the meantime, Wicki says he encourages physicians to bring back to patients the benefits of the scientific output that is now available. The missing connection between the lab and the clinic has for many years slowed therapeutic innovation to a level that may no longer be ethically sustainable.