International Consortium Completes First-Ever Proteomic Mapping of All 28 Human Cancer Types
Data covers over 5,000 tumors and reveals 13 new cancer subtypes requiring targeted therapy, published in Cell.
"Cancer Atlas 2.0: Why Proteomic Mapping of 28 Cancers Kills Old Oncology and Creates 13 New Billion-Dollar Markets"
Author: Venture Analyst in Personalized Medicine
Date: 2026-06-01
When the international consortium led by CPTAC (Clinical Proteomic Tumor Analysis Consortium) published the first complete proteomic atlas of 28 cancer types in Cell, covering over 5,000 tumors, medical media responded with the standard summary: "A new era of precision oncology." This is a catastrophically narrow view.
I have been analyzing omics technology markets since 2019, and what was published in this issue of Cell is not just a database. It is a fundamental redefinition of what a "cancer type" is. The discovery of 13 new subtypes that were indistinguishable by genome or transcriptome but radically different in protein profile and, most importantly, in sensitivity to targeted drugs means that all previous clinical trials stratified by genetic markers may have missed ideal targets.
Mainstream media will write about a "breakthrough in diagnostics." I will write about why Illumina (genomic sequencing) shares will drop 15% in the next 90 days, why Roche and Thermo Fisher will split the proteomic mass spectrometry market worth $8 billion, why 13 new subtypes mean 13 new opportunities for "patent reset" on old drugs, and why the biggest winners will not be diagnosticians but AI platform companies like DeepMind and PathAI.
1. [The Essence]: What Is Really Happening
This is not just about compiling a catalog. It is the first systematic evidence that the proteome (the entire set of proteins) carries diagnostic information fundamentally inaccessible to the genome. Until now, oncology lived in the paradigm "mutation = subtype = therapy." BRCA, EGFR, HER2, BRAF — these few hundred genes determined which targeted drug a patient would receive. But the genome does not tell you whether a protein is activated, does not reveal post-translational modifications, and does not indicate which signaling pathway is actually active in the tumor at a given moment.
CPTAC researchers used high-resolution mass spectrometry to analyze over 10,000 proteins per tumor across 5,000+ tumors. They integrated this data with genomics, transcriptomics, and clinical outcomes. They found that at least 13 previously known "genetic" cancer types actually split into two or three proteomic subtypes with completely different prognoses. For example, one subtype of pancreatic cancer, indistinguishable by DNA, had 4 times higher expression of the EGFR receptor and responded to cetuximab where standard chemotherapy was useless.
But the most important aspect is the "functional" information. The proteome shows which signaling pathways (RAS-MAPK, PI3K-AKT, p53) are truly hyperactivated in the tumor. This means inhibitors can be prescribed not based on a mutation (which may not exist) but on a "phosphoproteomic fingerprint" — an activity imprint. This expands the target audience for many targeted drugs by 2-3 times. A drug previously given to only 10% of patients with a specific mutation could be given to 30-40% of patients with a specific proteomic pattern, even without the mutation.
Numbers that did not make it into the press release: of the 28 cancer types studied, at least one "proteome-specific" vulnerability was found for 22 — a protein or pathway not obvious from the genome. And for 15 of these 22, FDA-approved drugs already exist that target these proteins. Thus, the CPTAC atlas de facto proposes "repurposing" 15 existing drugs for new indications, without the need to develop new molecules.
2. [Timeline and Context]: 10 Years and $100 Million
This project did not come out of nowhere. CPTAC was launched by the National Cancer Institute (NCI) in 2011. The first phase (2011-2016) covered 7 cancer types. The second phase (2016-2021) added 3 more. Only the third phase (2021-2026) completed mapping all 28 major types, including rare ones. The total budget over all years is about $380 million (NCI estimate). Compared to the Human Genome Project ($3 billion), this is peanuts. But the return is already comparable.
Key technological breakthroughs that made the atlas possible:
- 2018-2020: Development of standardized mass spectrometry protocols allowing data comparison across centers. Work published in Nature Methods became the foundation for all subsequent data harmonization.
- 2022: Creation of the public CPTAC data portal, where any researcher can download raw proteomic files. To date, the portal has over 15,000 registered users from 70 countries.
- 2024-2025: Integration of proteomic data with radiological and pathological images from The Cancer Imaging Archive (TCIA). This allowed correlation of "proteomic subtype" with what a radiologist sees on MRI or a pathologist under a microscope.
Why publish now, in May-June 2026? Because in January 2026, the final analyses for rare tumors (sarcomas, acute myeloid leukemia) were completed. The peer review process at Cell took 4 months — a record for such a massive work (main article 78 pages, supplementary 450 pages). Cell editors called this work "the most significant contribution to oncology in the decade since the completion of The Cancer Genome Atlas (TCGA)."
Additional context: in March-April 2026, several papers "warmed up" the audience for this atlas. The Path2Prot work from AACR showed that AI can predict proteomic markers from routine histological slides. This meant that using proteomic classification does not necessarily require mass spectrometry for every tumor — just train AI on CPTAC data, and it will "read" proteins directly from standard H&E slides. This dramatically lowers the barrier to adoption.
3. [Who Wins and Who Loses]: Billion-Dollar Shifts
Biggest winner — Thermo Fisher Scientific (TMO) and Bruker (BRKR). These are the two largest manufacturers of high-resolution mass spectrometers used in proteomics. Thermo Fisher owns the Orbitrap platform, used for 80% of CPTAC data. Bruker is catching up with the timsTOF platform. The clinical proteomics market is currently valued at $2.5 billion, but after the atlas publication, it is expected to grow to $8-10 billion by 2030. Thermo Fisher shares rose 4% on the day of publication, Bruker 6%. Analysts have raised target prices by 15-20%.
Second winner — companies developing AI for proteomic data analysis (DeepMind, PathAI, Tempus). DeepMind already has AlphaFold for protein structure prediction. Now they are launching AlphaProteome — an AI that predicts proteomic subtypes from genomic data. PathAI (heavily funded by Roche) received exclusive access to CPTAC data to train its pathology platform. PathAI's valuation after this deal rose from $2 billion to $3.5 billion (according to PitchBook).
Third winner — patients with rare and "molecularly silent" tumors. For 20-30% of solid tumors, standard genomic testing finds no driver mutations ("pan-negative"). These patients were previously told: "We have no target; we will treat with chemotherapy blindly." Proteomic analysis of the 13 new subtypes identified targets (e.g., MET amplification at the protein level, invisible in DNA) for 40% of these "ghost patients." Not a panacea, but a chance.
Biggest loser — Illumina (ILMN) and the genomic sequencing market for oncology. Illumina dominates tumor sequencing (80% market share, about $3 billion revenue). But the proteomic atlas shows that genomic information is insufficient for accurate stratification. If proteomics enters clinical routine, hospital budgets for cancer testing will be reallocated: less money for sequencing (done once) and more for mass spectrometry (more expensive and complex). Forecast: Illumina's growth slows from 12% to 5% over the next 3 years. Shares fell 7% in the week after publication.
Loser #2 — Roche in the IHC and FISH test segment (immunohistochemistry, fluorescence in situ hybridization). Currently, the standard for diagnosing many cancers is staining sections with antibodies (IHC) for HER2, ER, PR, PD-L1. But IHC can only look at 1-2 proteins at a time. Proteomics — thousands of proteins. If proteomic profiling becomes cheaper and faster, IHC tests costing $200-500 will become obsolete. Roche earns about $1.5 billion annually from selling antibodies and Ventana automated systems. This business is under direct threat.
Unobvious loser — companies developing drugs against "neoproteomic" targets. Sounds like an oxymoron, but let me explain. Many pharma companies (e.g., Merck KGaA) invested billions in targeted therapy against rare proteomic targets considered "unique" to one cancer type. The new atlas shows these targets often appear in 5-6 other cancer types. This is good for patients (the drug can be used more broadly) but bad for the originator, because competitors can file for a "new indication" for their generic and enter the market through accelerated approval. Patent protection for "targeted therapy" is built on "use in cancer X." If the atlas shows that cancer X is actually three different proteomic subtypes and only one responds, the patent can be challenged. Lawyers are already preparing lawsuits.
4. [What the Media Isn't Saying]: The Curse of Dimensionality, the Cost of Standardization, and "Who Will Pay"
Insight #1 — the most technical and most important: the "curse of dimensionality" problem. The CPTAC atlas contains terabytes of data. But in clinical practice, when you take a biopsy from a patient, you don't have the time or money to analyze 10,000 proteins. You need to select 50-100 "informative" proteins. Who will choose them? How will they be validated? The Cell paper proposes a "minimal panel" of 823 proteins that are most informative for all 28 cancer types. But 823 proteins are still too many for routine diagnostics. You need 20-30. Research groups are now racing to reduce this list to a clinically applicable one. This will take 2-3 years.
Insight #2: standardization is a nightmare. CPTAC data was generated over 10 years on different versions of mass spectrometers, by different technicians, with different protocols. That researchers managed to "stitch" them together is a scientific feat. But their harmonization methods (hybrid imputation + quantile normalization) only work for this data. If your clinical lab buys a new mass spectrometer, you cannot simply compare your data to CPTAC because protocols will diverge. You will need "reference standards" that have not yet been created. NCI knows this and in 2025 launched a $50 million program to develop "proteomic standards." Results are not yet available.
Insight #3 — cost and accessibility. Full proteomic analysis of one sample on a high-resolution mass spectrometer currently costs $2,000-3,000. This is roughly the same as whole exome sequencing ($1,000-2,000) plus RNA sequencing ($500-1,000). Insurers (Medicare, private) do not yet cover proteomics for routine cancer diagnostics. After the atlas publication, they will likely start pilot coverage programs for "indeterminate" cases (pan-negative tumors). But full coverage will take 3-5 years. Without coverage, $3,000 out of pocket is unaffordable for most.
Insight #4 — legal and ethical minefield: open data means anyone can use it for commercial purposes. CPTAC is a public resource. Any company can download the data and develop a diagnostic test based on it without paying NCI a cent. This is good for competition but bad for "monetization" by the discoverers. NCI cannot patent "proteomic subtypes" because they are discoveries, not inventions. As a result, billion-dollar profits from commercializing the atlas will go to Roche, Thermo Fisher, Illumina (yes, they both lose and win simultaneously), and a hundred startups. NCI and taxpayers who funded the research will see no royalties. Politicians in the US Congress are already raising this issue: "Why do we pay, and private companies profit?" The answer: that's how the public science funding system works. But this could change in the next 5 years.
5. [Forecast: Next 30 Days and 90 Days]
30-Day Forecast (June 2026):
First: June 14-18 — the American Society of Clinical Oncology (ASCO) conference in Chicago. A special plenary session "CPTAC Pan-Cancer Atlas: From Discovery to Clinic" will be held. Expect 5-7 groups to present "pilot" proteomic testing protocols for specific cancer types (lung, breast, colorectal). The most advanced protocols promise results in 7 days (vs. 3 weeks currently). This will attract the attention of practicing oncologists.
Second: June 20 — FDA will issue a draft guidance on "proteomic biomarkers for companion diagnostics." Until now, FDA has only approved genomic and IHC biomarkers. The new guidance will define what data is needed to validate a proteomic test. Expect stringent requirements (prospective clinical trials required). This will slow commercialization by 2-3 years but increase physician trust.
Third: June 25 — BioRxiv will publish a preprint from a group that analyzed CPTAC data using a new AI algorithm and found that at least 6 of the 13 new subtypes can be predicted from routine MRI without biopsy (proteomic radiomics). If confirmed, this changes everything: a patient could get a "proteomic subtype" without an invasive biopsy. Shares of companies developing radiomics (HealthLytix, Quantitative Insights) will rise 50% in a month.
90-Day Forecast (by September 2026):
By August, a paper in Nature Genetics will show that some of the new proteomic subtypes have an epigenetic (not genetic) nature. This means they can be "reversed" with drugs that alter DNA methylation (azacitidine, decitabine). The epigenetic drug market (estimated at $5 billion by 2030) will get a powerful new driver.
By September, Quantum-Si, a company developing single-molecule protein sequencing technology (an analog of Illumina but for the proteome), will announce its first clinical prototype, Proteus. If their platform can analyze 100 proteins for $500, it will kill mass spectrometry for routine diagnostics. Quantum-Si shares (QSI) are a bet on the "proteomic future." They are currently valued at $300 million, but with a successful Proteus launch, they could grow to $2 billion in 2 years.
The most important thing that will happen in the next 90 days is invisible to the public: Big Pharma (Roche, Novartis, Pfizer) will start internal projects for "proteomic reclassification" of their portfolios. Roche has trastuzumab (Herceptin) for HER2+ breast cancer. The CPTAC atlas shows that the HER2 protein is overexpressed (but no gene amplification) in 5 other cancer types. Roche will launch a "basket" trial (one molecule against many cancer types united by a proteomic feature). If successful, Roche will add $3-4 billion to Herceptin sales. If not, it will lose market share.
Analyst Verdict: The CPTAC proteomic atlas is the moment oncology stopped being "genomic" and became "functional." Invest in Thermo Fisher (TMO) and Bruker (BRKR) as the "picks and shovels" of the proteomic gold rush. Short Illumina (ILMN) — their genomic monopoly is crumbling. Watch Quantum-Si (QSI) as a risky but potentially high-reward bet on technology disruption. And if you are a practicing oncologist — start looking for a contract lab that performs mass spectrometry. Because in 2-3 years, you will not be able to answer a patient's question "what is my subtype?" without looking at the proteome. The genome will no longer help you. Proteins are the new language of cancer. And the CPTAC atlas is the first dictionary of that language. Now it's up to the translators.
— Editorial Team
