How Does Sleep’s Molecular Footprint in the Blood Affect Gastrointestinal Cancers?
In this article
Plasma Proteins Link Sleep Quality to Gastrointestinal Cancers
UK Biobank: A gold mine for modern epidemiology
UK Biobank is a prospective cohort biobank of half a million participants, comprising more than 502.000 people (five hundred and two thousand), aged 37-73, recruited across the United Kingdom between 2006-2010. Questionnaires, physical examinations, blood samples, urine samples and genetic data were collected at the baseline (entry) visit; participants were subsequently followed over the long term through national health records. Over the past 15 years, this infrastructure has become one of the world’s most productive sources of epidemiological data; thousands of publications have been based on UK Biobank data.
Liu and colleagues used two separate cohorts from this pool:
STUDY ARCHITECTURE — A TWO-LAYER ANALYSIS
- First layer — sleep score analysis in 472.105 participants: A sleep health score based on five sleep characteristics (sleep duration, chronotype, insomnia, snoring and daytime sleepiness) was classified into three categories: healthy (≥4 points), intermediate (2-3 points) and poor (≤1 point).
- Second layer — proteomic analysis in 52.920 participants: Using Olink Proximity Extension Assay technology, 2.911 plasma proteins were measured as part of the UK Biobank Pharma Proteomics Project. This is the world’s largest proteomic cohort dataset.
- Machine learning approach: The model was trained on 70% of the sample and validated on the remaining 30%. This is a gold-standard approach that protects the results against overfitting.
- Statistical adjustment: Multivariable adjustment included demographic, lifestyle and clinical factors, as well as family history. The results are therefore much more robust than the simple correlation that "good sleep lowers cancer risk".
- Outcome: Incident cancer, particularly gastrointestinal cancer, was linked to the sleep score and proteomic signature.
The anatomy of the UK Biobank sleep health score
The sleep score used in the study is far more nuanced than a simple question about "how many hours you sleep". Five components are assessed together, each reflecting a different dimension of sleep health:
1. Sleep duration (sleep duration)
Healthy range for adults: 7-8 hours per night. Both short sleep (<7 hours) and long sleep (>9 hours) increase risk, with a U-shaped relationship. It is not only "too little sleep" that is pathological, but "too much sleep" as well.
2. Chronotype (chronotype)
Morning–evening preference. A "morning person" (morning chronotype) is considered healthy; an "evening person" (evening chronotype) is regarded as deviating from the natural course of the circadian rhythm, negatively affecting metabolic and immune balance.
3. Insomnia (sleeplessness)
Difficulty falling asleep, waking during the night and waking early. Chronic insomnia disrupts cortisol rhythms, increases systemic inflammation (CRP, IL-6) and weakens immune surveillance.
4. Snoring (snoring)
A marker of sleep apnoea; obstructive sleep apnoea (OSA) produces cycles of hypoxia and reoxygenation, increasing oxidative stress, sympathetic activation and systemic inflammation. Multiple meta-analyses have shown OSA to be an independent cancer risk factor.
5. Daytime sleepiness (daytime sleepiness)
An objective clinical indicator of sleep deprivation and poor sleep quality. High Epworth scores → markers of untreated OSA, chronic sleep restriction and circadian disturbance.
Taken together, these five components provide a multidimensional picture of a person’s sleep health. Rather than a single question asking "how is your sleep?", assessment proceeds through five separate gateways. Including these five components in the CheckTheLife application’s questionnaire means alignment with the most up-to-date scientific standard for sleep health.
The relationship between sleep score and cancer risk
The first-layer analysis, examining the relationship between a conventional sleep score and cancer in 472.105 people, produced two major findings:
In other words, someone who sleeps healthily reduces their risk of developing liver cancer by one third compared with someone who sleeps poorly; their lung cancer risk also falls to a statistically significant degree. The real surprise, however, came in the second layer: the proteomic signature analysis.
The proteomic score: Sleep’s molecular footprint in the blood
Using machine learning models, the investigators identified the proteins most strongly correlated with the sleep health score. For each of the five sleep components, correlations were found with between 578 and 2.042 proteins (two thousand and forty-two). These protein sets were then combined to create an overall sleep proteomic signature comprising 303 plasma proteins . This signature showed highly significant validation in the test set (Pearson r = 0,24, p < 0,00000001).
An important conceptual point: a correlation of 0,24 is not numerically very high, but the p-value is extraordinarily small. This means the relationship is not weak; rather, no single protein reflects sleep health. Instead, sleep leaves a global signature on the biochemical composition of the blood. Assessing 303 proteins together makes it possible to read the biological shadow of a person’s sleep quality.
The truly striking finding was that this proteomic score was much more strongly associated with gastrointestinal cancers than the conventional sleep score. The following table shows cancer risk reductions among individuals in the highest tertile of the proteomic score compared with those in the lowest tertile:
SLEEP PROTEOMIC SCORE — HAZARD RATIOS FOR GASTROINTESTINAL CANCER RISK
Look closely at the table: all six gastrointestinal cancers show statistically significant risk reductions with biologically large effect sizes. For liver cancer, the risk reduction is 71%: individuals whose proteomic signature indicates "good sleep" have a liver cancer risk of not even approximately one third that of individuals whose signature indicates "poor sleep". This is one of the largest effect sizes observed for lifestyle factors in modern epidemiology.
Circadian rhythm, melatonin, inflammation and immune surveillance
Liu and colleagues’ presentation focused primarily on statistical associations, but the existing literature on cancer neuroscience and circadian oncology describes the possible molecular foundations of the sleep–cancer relationship in detail. It is important to recognise that these pathways operate in an interconnected manner rather than separately:
POSSIBLE MECHANISMS BEHIND THE SLEEP–CANCER RELATIONSHIP
- 1. Circadian clock genes and tumour suppressors. Core clock genes such as BMAL1, PER2, CRY1 and CLOCK activate tumour-suppressor pathways, particularly p53, in synchrony with time. Sleep disturbance disrupts this synchronisation; in some tumour types, loss of clock genes accelerates the expansion of cancer stem cells.
- 2. Melatonin deficiency. Melatonin, released from the pineal gland only in darkness, is a potent antioxidant and immunomodulatory hormone. Exposure to light at night and poor sleep quality reduce melatonin AUC, weakening the DNA damage response and cellular repair. A correlation between low melatonin levels and increased risk of breast, colon and liver cancers has repeatedly been demonstrated in the established literature.
- 3. Systemic inflammation — increased IL-6, TNF-α and CRP. Chronic sleep insufficiency continuously activates the hypothalamic–pituitary–adrenal (HPA) axis and sympathetic nervous system, causing low-grade systemic inflammation. Liver (HCC), pancreatic and colorectal cancers are among the leading inflammation-driven cancers.
- 4. Impaired immune surveillance. NK cell activity peaks during sleep. Sleep deprivation acutely reduces NK activity by 30-70% (classic studies from the Irwin laboratory). NK cells are the primary immune sentinels that detect tumour cells at an early stage.
- 5. Insulin resistance and metabolic syndrome. Insufficient sleep increases insulin resistance, glucose dysregulation, leptin–ghrelin imbalance and abdominal fat accumulation. This metabolic pattern is a recognised risk factor, particularly for liver (NAFLD → HCC), pancreatic and colorectal cancers.
- 6. Disruption of the microbiome–gut axis. Circadian rhythms govern daily changes in the gut microbiome. Shift workers and individuals with chronic sleep disturbance have reduced microbiome diversity, increased permeability and a deteriorating inflammatory cytokine profile.
- 7. The circulating tumour cell (CTC) peak. The 2022 study by Diamantopoulou and colleagues in Nature showed a marked increase in circulating tumour cells during sleep. In other words, metastasis is a biologically nocturnally active process. Sleep disturbance disrupts the functioning of this circadian window.
CONCEPTUAL SUMMARY — WHY GASTROINTESTINAL CANCERS IN PARTICULAR?
The strongest signals in the study were seen in gastrointestinal cancers: oesophageal, gastric, colon, liver, gallbladder and pancreatic cancers. This is no coincidence. The gastrointestinal system is the body’s most circadian-sensitive system: eating and drinking cycles, gastric motility, biliary and pancreatic secretion, the gut microbiome and hepatic metabolism all follow daily rhythms. When these are disrupted, cumulative "metabolic–inflammatory fatigue" naturally builds up, affecting cancer risk in gastrointestinal tissues.
The strongest signal, for liver cancer (a 71% risk reduction), is particularly meaningful: the liver is the body’s most strongly circadian organ, with approximately 10% of genome-wide gene expression oscillating over time. When sleep is disrupted, the liver’s metabolic buffering capacity gradually declines, creating conditions conducive to the NAFLD → MASH → HCC sequence.
An honest assessment: What we know and what we do not
Although the study is very strong, it requires an honest reading. Important limitations include:
LIMITATIONS AND DISCUSSION
- Correlational, not causal. This is an observational cohort study. It does not provethat improving sleep directly reduces cancer risk; it only shows a strong association. Randomised trials of sleep interventions with cancer outcomes are technically and ethically very challenging.
- UK Biobank selection bias. UK Biobank participants are healthier, better educated and socioeconomically more advantaged than the general British population. The results may not be directly extrapolatable to populations with lower socioeconomic status.
- Self-reported sleep data. The sleep score relies on participants’ own reports rather than actigraphy, an objective device-based measurement. This carries risks of recall bias and social desirability bias.
- Measurement at a single time point. Both sleep and proteomic data largely come from baseline; the effects of long-term changes in sleep were not fully captured.
- Confounding factors. People who sleep well generally also show other healthy behaviours, such as a better diet, more exercise and lower tobacco and alcohol consumption. Statistical adjustments partially separated these effects, but eliminating all confounding is impossible.
- The 303-protein signature is difficult to interpret. The specific proteins comprising this signature, particularly inflammatory, metabolic and immune markers, have not yet been fully published: the DDW abstract provided abbreviated information. Publication of the full article is awaited.
Nevertheless, the study’s strength is undeniable: 472.105 patients + 2.911 proteins + multiple cancer types + robust statistical adjustment + machine learning validation. Evidence on this scale is rare in modern epidemiology.
The sleep questionnaire module in the lifestyle assessment application we developed is a tool that brings the full counterpart of the DDW 2026 findings into clinical practice
CheckTheLife, an interactive application within the drozdogan.com ecosystem, assesses cancer risk and health determinants. Its design includes a questionnaire module measuring sleep quality under a heading we call "the psychological determinants of health" . When we designed this module, the existing scientific literature had provided epidemiological evidence for the sleep–cancer link, but there was not yet mechanistic evidence at the level of "the biological signature that sleep leaves in the blood".
This study presented at DDW 2026 strengthens the scientific foundation of CheckTheLife’s sleep module. We now know that asking about sleep quality is not a soft precaution but rather a hard biological intervention that measurably affects the biochemical composition of the blood, systemic inflammation, immune surveillance and gastrointestinal cancer risk.
The five components covered by the application’s sleep questionnaire—sleep duration, chronotype, insomnia, snoring and daytime sleepiness—are the same five components of the UK Biobank sleep health score. This means that our design is aligned with the methodological standards of the world’s largest epidemiological sleep study .
THE CLINICAL SIGNIFICANCE OF CHECKTHELIFE’S SLEEP MODULE — REFRAMING IN LIGHT OF DDW 2026
When a CheckTheLife user completes the sleep module and receives a result of "poor sleep health", that result now means the following:
"According to the model derived from data on 472.105 people in UK Biobank, your sleep pattern may be causing your plasma proteomic signature to fall into the ‘poor sleep’ category. Over the years, individuals in this category have a statistically significantly higher risk of oesophageal, gastric, colorectal, liver, gallbladder and pancreatic cancers than those in the healthy sleep category—by 50-70% for some cancer types."
This is a much more concrete scientific message than telling a user that "sleep matters". Numerical, measurable, personalised and biologically grounded.
Practical interventions that could be recommended based on CheckTheLife’s sleep module—sleep hygiene education, routines aligned with chronotype, reduced screen exposure, referral for OSA screening and normalisation of melatonin rhythms—can now be considered not merely in terms of "quality of life", but within cancer prevention .
⚡ WHY DOES IT MATTER?
The study presented by Yangyang Liu and the MSKCC team at DDW 2026 marks an important junction in the modern literature on cancer prevention. It combines 472.105 patients + a 303-protein signature + consistent, large, statistically significant risk reductions across six gastrointestinal cancers. The sleep–cancer link is no longer merely an epidemiological observation, but a measurable biological reality reflected in plasma proteomics. Considering the approximately ~46.000 new gastrointestinal cancer cases (forty-six thousand) annually in Türkiye, assessing sleep quality and intervening offer potential for a real public health benefit. CheckTheLife’s sleep module directly matches this scientific framework and has now gained stronger clinical legitimacy through the DDW 2026 findings.
✅ THE STRENGTH OF THE EVIDENCE AND METHODOLOGY
UK Biobank, a prospective cohort of half a million participants, is a gold mine of modern epidemiology. Olink Proximity Extension Assay measurements of 2.911 proteins represent the world’s largest proteomic cohort dataset. The machine learning model follows the gold standard of 70% training/30% validation. Multivariable adjustment covers demographic, lifestyle and clinical factors and family history. Proteomic signature validation: Pearson r = 0,24 + p < 10⁻⁸. A consistent direction across six gastrointestinal cancers: oesophagus HR 0,65, stomach HR 0,46, colorectal HR 0,72, liver HR 0,29, gallbladder HR 0,46 and pancreas HR 0,47. None of the confidence intervals cross 1,0, indicating strong statistical significance.
⚠️ LIMITATIONS AND POINTS REQUIRING CAUTION
A correlational study: it does not provecausality, but shows a strong association. UK Biobank participants are healthier than the general British population (volunteer bias). Sleep data are self-reported, not measured by actigraphy. Measurement was taken at a single time point, so long-term changes were not captured. Confounding factors—people who sleep well generally lead healthier lives—could not be fully separated. The specific components of the 303-protein signature have not yet been fully published. The DDW abstract did not clearly state its limitations. Extrapolation to populations with lower socioeconomic status carries risks. The sequence "improving sleep → cancer prevention" has still not been proven in a randomised intervention study.
🩺 IMPLICATIONS FOR CLINICAL PRACTICE
At Memorial Göztepe and other oncology centres, questions about sleep quality should become a standard part of routine medical history taking, particularly for patients at high risk of gastrointestinal cancers: a history of Helicobacter pylori, chronic viral hepatitis, NAFLD, familial colorectal cancer or chronic pancreatitis. The sleep–cancer relationship can be communicated to patients on an evidence-based basis: rather than merely saying "sleep matters", we can now say that "DDW 2026 data showed numerically that healthy sleep reduces liver cancer risk by up to 71%". The sleep module among CheckTheLife’s lifestyle modules can be used within a cancer prevention programme, and the application’s results can form the basis for discussing practical interventions with patients: a sleep plan aligned with chronotype, disciplined screen use, referral for OSA screening and melatonin assessment. For international patient coordinators, sleep assessment should be added to the routine prevention profile in Central Asian, Caucasian and Middle Eastern cohorts, particularly populations in which risk factors for pancreatic, gastric and liver cancers have a different distribution than in the West.
❓ QUESTIONS FOR THE FUTURE
When will Liu and colleagues’ full publication appear (NEJM, JAMA or Lancet)? Which specific proteins comprise the 303-protein signature, and how are the inflammatory, metabolic, immune and circadian markers distributed? Will the same paradigm be validated for lung, breast and prostate cancers? Are randomised trials of sleep interventions with cancer outcomes methodologically feasible and ethical? Is the relationship between sleep health score and gastrointestinal cancer in the Turkish population comparable with that in the European cohort, particularly in the context of a Turkish lifestyle in which evening chronotype is more common? Could CheckTheLife application data be analysed academically to develop a sleep–cancer risk model for the Turkish population? How can major oncology centres such as Memorial Göztepe integrate questions about sleep quality into routine medical history categories? How will insurance systems assess sleep intervention programmes within cancer prevention budgets?
Sources
- Liu Y, et al. Sleep proteomic score and gastrointestinal cancer risk in the UK Biobank. Memorial Sloan Kettering Cancer Center, New York. Digestive Disease Week (DDW) 2026, presentation in Chicago, May 2026.
- Medscape Medical News. Blood Proteins Link Poor Sleep to Increased Cancer Risk. 16 May 2026.
- UK Biobank. Pharma Proteomics Project (UKB-PPP) — Olink Proximity Extension Assay, plasma protein data from 54.306 participants (fifty-four thousand, three hundred and six). Uppsala, Sweden: Olink Analyses Service.
- Sudlow C, Gallacher J, Allen N, et al. UK Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Medicine 2015;12(3):e1001779.
- Diamantopoulou Z, Castro-Giner F, Schwab FD, et al. The metastatic spread of breast cancer accelerates during sleep. Nature 2022;607(7917):156-162. (The peak in circulating tumour cells during sleep.)
- Irwin MR. Why sleep is important for health: a psychoneuroimmunology perspective. Annual Review of Psychology 2015;66:143-172. (Foundational literature on the relationship between NK cell activity and sleep.)
- Sulli G, Lam MTY, Panda S. Interplay between circadian clock and cancer: new frontiers for cancer treatment. Trends in Cancer 2019;5(8):475-494. (A review of circadian biology and oncology.)
- Hadadi E, Taylor W, Li XM, et al. Chronic circadian disruption modulates breast cancer stemness and immune microenvironment. Nature Communications 2020;11(1):3193.
- Walker WH 2nd, Walton JC, DeVries AC, Nelson RJ. Circadian rhythm disruption and mental health. Translational Psychiatry 2020;10(1):28.
- Le K, Liu Y, Bai R, et al. Large-scale plasma proteomics reveals bidirectional associations between sleep patterns and inflammatory bowel disease. BMC Medicine 2025;23:539. (A preceding study in the UK Biobank literature on sleep, proteomics and gastrointestinal disease.)
- Turkish Sleep Medicine Society. National report on the epidemiology of sleep disorders in Türkiye. Ankara: TUTD, 2024.
- Turkish Thoracic Society. Clinical guidelines for obstructive sleep apnoea syndrome. Ankara: TTD, 2024 update.
- Ministry of Health of the Republic of Türkiye. Türkiye National Cancer Registry Data 2023 — Gastrointestinal cancer incidence. Ankara: General Directorate of Public Health, 2025 edition.
- Kırca Ö. CheckTheLife — cancer risk assessment and lifestyle application, sleep module. drozdogan.com ecosystem, 2024-2026.
This article is intended as medical news and information and does not provide individual treatment advice. The DDW 2026 presentation provided information at the conference abstract level; a full peer-reviewed publication is awaited. The study has an observational cohort design and does not prove causality, showing only a strong statistical association. The direct cancer-preventive effect of sleep interventions has not been proven in randomised clinical trials. CheckTheLife is not a medical diagnostic or treatment tool, but a tool for individual risk awareness and lifestyle assessment; its results should be discussed with a physician. Individuals experiencing sleep disorders are advised to consult a sleep medicine specialist or pulmonologist, particularly in cases of snoring, suspected OSA or chronic insomnia. In Türkiye, sleep medicine services are available through the Turkish Sleep Medicine Society network and Ministry of Health sleep centres.
Was this article helpful?
Your feedback helps us improve our content.
Reader comments
Share your comment
Your name and comment are published only after administrator approval. Your email address is never shown to visitors.
Be the first to comment.