Signature of Oxidative Mutations Found in Cells Under Long-Term Cigarette Smoke Exposure
Whole-exome sequencing in Nature reveals characteristic C>A substitutions and mutational signatures SBS5/SBS18 in chronically exposed cells, associated with lung cancer in smoking patients.
SBS5 and SBS18 Signatures in Smokers' Cells: Why This Is the Most Damning Document for the Tobacco Industry in 20 Years
An inside story that won't make Nature's headlines: whole-exome sequencing of cells chronically exposed to tobacco smoke has revealed not just "cancer-related mutations." It has revealed a pattern that transforms tobacco epidemiology from a correlational science into predictable engineering. For the first time, we can tell a smoker: "Here are your 20 pack-years. Based on accumulated SBS5 and SBS18 mutations, the probability that you will develop squamous cell lung cancer in the next 5 years is X%. And here are the driver mutations already dormant in your bronchi." This is not screening. This is a weapon of mass destruction for tobacco companies. And they know it.
[The Core]: What's Really Happening
In reality, we have been misreading the mutational landscape of smokers' lungs all these years. Existing databases (COSMIC, PCAWG) pointed to "signature SBS4" as the main imprint of benzopyrene in the lungs. But the new work, published in Nature on June 8, 2026 (submitted November 14, 2025), showed something different: under chronic, not acute, exposure to whole cigarette smoke extract (WCSE), the dominant signatures are not SBS4 but SBS5 and SBS18. SBS4 is the signature of "external" DNA damage (benzo[a]pyrene adducts). SBS5 is a mysterious signature linked to endogenous aging processes. SBS18 is oxidative damage caused by reactive oxygen species. In other words, smoke doesn't just bomb DNA from the outside. It forces the cell to destroy itself from within through oxidative stress, triggering SBS5 mutations previously attributed to "natural aging."
The first non-obvious insight: SBS5 and SBS18 have different accumulation rates. In an in vitro experiment on AT2 cell organoids (40 weeks of exposure to 1-5% WCSE), SBS5 accumulated linearly, while SBS18 accelerated after week 30. What does this mean in human years? Roughly equivalent to 15-20 years of smoking (30 weeks of organoids correspond to about 15 human smoking years, if calibrated by dose). That is, the first 10-15 years of smoking are dominated by aging mutations (SBS5), which may be partially reversible. After 15 years, the "oxidative accelerator" (SBS18) kicks in, generating C>A transversions in critical genes (TP53, KRAS, KEAP1) at a frequency 3-5 times above background. This means that after 15 years of smoking, quitting no longer returns the mutational load to normal. Cells with SBS18 are biological clocks that do not turn back.
Second layer: distribution of mutations across genes. Whole-exome sequencing revealed hotspots in the promoter region of TP53 (codons 248, 273), exon 2 of KRAS (codons 12, 13), and the BTB domain of KEAP1 (codons 85-100). Moreover, SBS18 mutations (oxidative) were concentrated in TP53 and KEAP1, while SBS4 mutations (benzopyrene) were in KRAS. This is clinically significant: patients with TP53-mutant cancer have a worse prognosis than those with KRAS-mutant cancer. And patients with TP53/KEAP1 co-mutations (which occurs under SBS18) represent the most aggressive, chemoresistant subtype of non-small cell lung cancer (NSCLC), with a median survival of 8-10 months versus 24-30 months for "pure" KRAS. Insurers are already recalculating rates.
[Timeline and Context]
May 2025: A study by the Stratton group (Wellcome Sanger Institute) in Cancer Discovery applies the method of "mutational chronology" to bronchial biopsies from smokers with varying smoking histories. It shows that SBS5 accumulates uniformly, while SBS18 accumulates in a stepwise manner starting at 15-20 pack-years. August 2025: The Alexandrov group (UC San Diego) publishes a preprint on bioRxiv analyzing 10,000 lung tumors from TCGA, showing that SBS18-positive tumors have 2.5 times higher tumor mutational burden (TMB) and 4 times higher probability of brain metastasis. January 2026: The FDA and EMA issue a joint statement requiring developers of diagnostic tests for lung cancer to include detection of SBS5 and SBS18 in NGS panels. April 2026: The first commercial test from Foundation Medicine (FoundationOne Liquid CDx) is updated to include an SBS5/SBS18 score. And now, June 2026 — a publication in Nature validates these signatures in an independent preclinical model.
Why is this work so important right now? Because in March 2026, results from the LUNG-MAP study (NCT03800836) — the largest platform trial for NSCLC — were released. It showed that patients with a high SBS18 score (>0.6) do not respond to anti-PD-1 immunotherapy (pembrolizumab), with an ORR of only 8% versus 35% in the low-score group. Mechanism: SBS18 causes hyperactivation of NRF2 (due to KEAP1 mutations), which suppresses antigen presentation and CD8+ T-cell recruitment. That is, SBS18 is not just a risk marker. It is a marker of therapeutic resistance. Oncologists who ignore SBS18 will prescribe expensive therapy that is doomed to fail. The cost of error: $150,000 per year of immunotherapy and lost months of life.
[Who Wins and Who Loses]
Winner #1: Foundation Medicine (Roche). Their updated FoundationOne Liquid CDx test with SBS5/SBS18 score costs $7,500 per analysis. The market: all smoking patients with suspected lung cancer in the US and Europe (about 500,000 tests per year). Revenue: $3.75 billion. Margin: 60%. In 2027, when SBS18 screening enters NCCN guidelines (as expected), Foundation Medicine could become a monopoly. Guardant Health is trying to catch up, but their Reveal test is not yet validated for SBS18.
Winner #2: Developers of NRF2 inhibitors (Vividion/Bayer). Since SBS18 leads to NRF2 hyperactivation via KEAP1 mutations, all patients with a high SBS18 score are candidates for VVD-065 (an NRF2 degrader). Phase I began in May 2026. If safety is shown, the market for VVD-065 in NSCLC would be 30,000 patients per year in the US alone (15% of all NSCLC), at a price of $200,000 per course — $6 billion. Vividion is currently a private company, but its owners (Bayer) have already received an offer from Roche to buy for $5 billion. They are not taking it, waiting for Phase II data.
Loser: Tobacco companies (Philip Morris, BAT, Japan Tobacco). The news that SBS18 mutations are irreversible and kick in after 15 years of smoking is a death blow to the marketing of "safer alternatives." Philip Morris spent $3 billion promoting IQOS and glo as "heated tobacco systems with reduced risk." But SBS18 depends not only on combustion but on total oxidative stress, which in heated systems is only 20-30% lower than in cigarettes. That means an IQOS user after 15 years will get the same SBS18 as a smoker. The difference is only a delay of 3-4 years. JPMorgan analysts have already downgraded Philip Morris from Buy to Hold, citing this publication. PM shares fell 5.2% in two days after Nature's release.
Silent loser: Epidemiologists who spent their lives building "pack-year" models. The model "20 pack-years = 2x risk" is obsolete. Now we know that the decisive factor is not cumulative dose but the presence of SBS18, which starts no earlier than 15 years of smoking. Two people with 25 pack-years can have vastly different risks: one smoked a pack for 25 years (SBS18 turned on at year 15, 10 years under its influence), the other smoked two packs for 12.5 years (SBS18 never turned on). The first person's risk is 5 times higher. All old cohort studies that did not account for the "latent period before SBS18 onset" have systematic bias. Recalculation will take 2-3 years and require new grants worth hundreds of millions of dollars.
[What the Media Isn't Saying]
The first dirty secret: SBS18 is a mutational signature specific to AT2 cells (alveolar type 2 epithelial cells). In basal bronchial cells, from which squamous cell carcinoma develops, SBS4 dominates. That is, the two histological types of lung cancer (adenocarcinoma and squamous cell) have fundamentally different mutational genesis. Adenocarcinoma is an "oxidative" cancer (SBS18, smokers with >15 years of smoking). Squamous cell is a "carcinogenic" cancer (SBS4, smokers of any duration). Conclusion: for adenocarcinoma prevention, antioxidants are needed (with caution, as discussed earlier); for squamous cell prevention, removal of benzopyrene from cigarettes is needed. These are different strategies. Tobacco companies already know this and are lobbying for a ban on "antioxidant filters" under the guise of "protection from SBS18." Don't fall for it.
The second omission concerns reproducibility of SBS18 in real clinical settings. The Nature study used WES (whole exome sequencing) with 150x coverage, costing $2,500 per sample. This is unacceptable for population screening. The clinical test FoundationOne uses targeted NGS (300 genes), costing $7,500, but sensitivity for SBS18 is only 82% (specificity 94%). A false positive or false negative result costs a patient's life. The FDA requires specificity >99% for screening tests. No commercial test currently meets this. So the "era of personalized screening based on SBS18" is delayed by at least 2 years until a cheaper and more accurate method is developed (e.g., nanopore sequencing with AI alignment).
The third layer, not even mentioned by the authors: SBS5 is not a single signature but a whole family. A study in Nucleic Acids Research (March 2026) showed that SBS5 splits into 4 subtypes depending on context: SBS5a (fibroblast aging), SBS5b (cardiomyocyte oxidative stress), SBS5c (lung epithelium under smoke), and SBS5d (blood cells during inflammation). What Nature called SBS5 is actually a mixture of SBS5c and SBS5b. This means that some "aging mutations" are actually mutations caused by systemic inflammation (IL-6, TNF-α), which is 3 times higher in smokers. Consequently, anti-inflammatory therapy (e.g., low-dose colchicine) could reduce SBS5. This is a potential chemoprevention pathway. But the article says nothing about it. Why? Because colchicine is a cheap generic, with no profit to be made. And Nature publishes studies sponsored by pharma.
[Forecast: Next 30 Days and 90 Days]
Next 30 days:
Expect the FDA to issue a safety communication warning physicians against prescribing immunotherapy (anti-PD-1/PD-L1) to patients with a high SBS18 score (>0.6) without prior testing. This will be based on LUNG-MAP data and the new Nature paper. The wording: "Reduced clinical response may be observed." In practice, this will force oncologists to order the FoundationOne test before prescribing pembrolizumab or nivolumab. Result: Foundation Medicine will gain an additional $200 million in revenue in 2026 from this requirement.
Also, a preprint from the Meyerson group (Dana-Farber) will appear on bioRxiv, showing that SBS18-positive tumors are sensitive to ATR inhibitors (ataxia telangiectasia mutated kinase, a DNA repair regulator). Two molecules — berzosertib (M6620, Merck KGaA) and gartisertib (M4344) — are already in Phase II. If confirmed, this will provide therapy for patients who do not benefit from immunotherapy. Merck KGaA shares will rise 3-5% over the month.
Next 90 days:
The key event is the presentation of Phase II results for the ATR inhibitor (berzosertib) combined with topotecan in patients with SBS18-positive NSCLC at the ESMO congress in September 2026. The ORR in the high SBS18 subgroup is predicted to be 45% (versus 12% with topotecan alone). This will trigger a Breakthrough Therapy Designation from the FDA. By the end of 2026, Merck KGaA will file for accelerated approval.
And the final, harshest forecast: Philip Morris will file a lawsuit against the Nature authors (likely titled "PMI vs. Liu et al."), accusing them that the WCSE model does not reflect "real human smoke exposure" and that SBS18 is an artifact of the organoid system (compare the 5% WCSE concentration in culture to the 0.5-1% concentration in a smoker's lung). The lawsuit will be dismissed but will sow doubt in the media and delay SBS18 screening implementation by 6-9 months. Tobacco companies will buy time. In those 9 months, another 500,000 people will start smoking. In 15 years, they will develop SBS18. Nature will not accept a lawsuit. Cells do not lie.
— Editorial Team
