The study, published in the journal Clinical Gastroenterology and Hepatology, documents a 20 to 30 percent increase in all forms of liver disease among people who reported late-night eating compared with those who did not. The association held for two distinct conditions: metabolic dysfunction-associated steatotic liver disease, commonly known as fatty liver disease, and cirrhosis, the advanced scarring of liver tissue that represents one of the most serious outcomes of chronic liver injury. The finding is notable because previous work had already tied late-night snacking to obesity and heart disease, but a clear signal connecting meal timing to liver pathology had remained elusive.

“A few studies have hinted at this idea but this is among the first long-term analyses to explicitly draw the connection to fatty liver disease,” said lead author Robert Wilechansky, M.D., a hepatologist at OHSU who specializes in liver transplantation. His motivation, he explained, came from a striking epidemiological fact: as much as 30 percent of the U.S. population is affected by fatty liver disease, a prevalence that has climbed alongside rising rates of obesity and metabolic syndrome. If a behavior as simple and modifiable as the timing of the evening meal contributed to that burden, the public health implications could be considerable.

To investigate the question, the researchers turned to three of the largest and longest-running nutritional epidemiology projects in existence: the Nurses’ Health Study, the Nurses’ Health Study II, and the Health Professionals Follow-up Study. From these nationwide datasets, the team extracted diet and lifestyle information self-reported by almost 200,000 participants. Rather than analyzing individual nutrients or total caloric intake, the investigators grouped participants according to their eating behaviors, with particular attention to whether food was consumed after dinner or after bedtime, and then searched for correlations between those behavioral patterns and reports of fatty liver disease or cirrhosis.

The statistical signal that emerged was consistent across the pooled cohorts. People who reported eating late at night showed a 20 to 30 percent higher rate of liver disease overall than those who confined their eating to earlier hours. The researchers emphasize that this is a correlation drawn from observational, survey-based data, which means it cannot by itself prove that late-night eating causes liver damage. Self-reported dietary habits are also broadly defined and subject to recall limitations. Even so, the size of the cohorts and the consistency of the association give the finding a weight that smaller, shorter studies could not achieve, and the authors describe it as a starting point for more definitive research.

“It’s important to remember, of the people who developed fatty liver disease, only a small fraction develop serious forms of the condition like liver cirrhosis,” Wilechansky said. “Even so, the big take-away here is that confining your calorie intake to several hours before bedtime might be beneficial.” That framing matters for how the results should be interpreted. Fatty liver disease exists on a spectrum, ranging from benign fat accumulation to inflammation, fibrosis, and ultimately cirrhosis or liver cancer. Most people with hepatic steatosis never progress to end-stage disease, but because the affected population is so large, even a modest shift in the fraction who progress would translate into a substantial number of additional cases of serious liver illness.

The biological rationale for why meal timing might matter to the liver rests on the science of circadian rhythms, the roughly 24-hour cycles that govern physiology in virtually every organ. The liver is not a passive filter that processes nutrients identically around the clock. Its enzymatic machinery, its glucose and lipid metabolism, and its detoxification pathways all fluctuate in anticipation of feeding and fasting cycles that evolved over millions of years. When food arrives during the biological night, a period when the internal clock is promoting sleep rather than digestion, the organ must process an energy load for which it is not physiologically prepared.

“We are seeing over and over again that the timing of behaviors, whether it’s sleep, eating, or exercise, really matters,” said co-author Andrew McHill, Ph.D., associate professor and sleep and circadian physiologist at OHSU. “For example, we have shown that eating during the night, when the internal biological clock is promoting sleep, reduces the ability of your body to metabolize foods as efficiently and can eventually lead to obesity, diabetes and now poor liver health.” McHill also pointed to previous research in animal models showing that the liver is among the organs most susceptible to misalignment between eating and sleep timing, a vulnerability that the new human data now appear to mirror.

The findings also resonate with what is known about shift workers, a population that Wilechansky highlighted as an important context for the research. Night shift workers experience chronic misalignment of the body’s natural day-night cycle, and epidemiological studies have consistently documented elevated rates of obesity, cardiovascular disease, and metabolic disorders among them. Their eating patterns are frequently displaced into the biological night, making them a natural experiment in circadian disruption. The new study suggests that the liver may be a previously underappreciated casualty of that disruption, and that the risks extend beyond shift workers to anyone who regularly eats in the hours when the body expects to be fasting and asleep.

Because the study is observational, the OHSU team is already moving toward a more rigorous test of the hypothesis. Wilechansky is part of a group at OHSU recruiting participants for a clinical study that will monitor eating behaviors directly, rather than relying on retrospective questionnaires, and will assess whether shifting meals earlier in the day improves clinical measures of underlying fatty liver disease. Such an intervention trial could establish whether meal timing is not merely a marker of other unhealthy habits but a causal lever that patients can pull to protect their livers. “We know the liver is very much in tune with the body’s circadian rhythms,” Wilechansky said. “So we’re interested in changing the timing of eating in a way that could make a difference in health without changing the composition of what people eat.” If that proposition holds up under controlled conditions, the prescription for liver health could become remarkably simple: not necessarily a new diet, but the same diet, eaten earlier. “Our findings add to the growing body of literature linking late-evening eating patterns to adverse metabolic and liver outcomes,” the authors write, a conclusion that, if confirmed, would place the humble midnight snack squarely on the list of modifiable risks to one of the body’s most indispensable organs.

Subject of Research: The association between late-night eating and the risk of fatty liver disease and cirrhosis

Article Title: Late-night eating may be bad for your liver, study suggests

Article References: Late-night eating may be bad for your liver, study suggests. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: liver disease, fatty liver disease, MASLD, cirrhosis, circadian rhythm, late-night eating, meal timing, metabolism, OHSU, Nurses' Health Study, sleep, observational study

The research team, led by investigators at Brigham and Women’s Hospital and Harvard Medical School, drew on data from the Nurses’ Health Study, established in 1976 with 121,701 participants, and the Health Professionals Follow-up Study, established in 1986 with 51,529 participants. Across these cohorts, the researchers identified 4,420 people who developed colorectal cancer during follow-up. For 1,208 of these cases, archived tumor tissue was available, allowing the team to measure LINE-1 methylation levels directly in the cancer cells. After a median follow-up of 16 years among surviving patients, the study documented 776 deaths from any cause, including 343 deaths specifically attributable to colorectal cancer.

LINE-1, short for long interspersed nucleotide element-1, is a transposable element, a genetic relic that appears in hundreds of thousands of copies scattered throughout the human genome. In healthy cells, these elements are normally silenced by chemical tags called methyl groups attached to the DNA. When cancer cells lose this methylation, a phenomenon known as hypomethylation, the genome becomes destabilized. LINE-1 methylation serves as a convenient surrogate marker for this genome-wide loss of methylation, and previous work has linked LINE-1 hypomethylation in colorectal tumors to genomic instability, worse prognosis, reduced response to certain chemotherapies, and even the emerging epidemic of early-onset colorectal cancer diagnosed before age 50.

To measure methylation, the researchers extracted DNA from formalin-fixed, paraffin-embedded tissue blocks, carefully focusing on tumor areas identified by a pathologist blinded to clinical information. They then performed bisulfite treatment, polymerase chain reaction, and pyrosequencing, quantifying methylation across four CpG sites within the LINE-1 element. The resulting scores, ranging from 0 to 100 percent, were normally distributed and categorized as low (below 60 percent), intermediate (60 to below 68 percent), or high (68 percent and above). The team also characterized other molecular features of each tumor, including microsatellite instability, the CpG island methylator phenotype, and mutations in the KRAS, BRAF, and PIK3CA genes, all of which were accounted for in the statistical models.

Smoking status at diagnosis was classified as never, past, or current, based on questionnaires completed before the cancer diagnosis. The researchers then applied a sophisticated statistical framework: multivariable Cox proportional hazards regression with inverse probability weighting, a technique that used data from all 4,420 cases to correct for potential selection bias arising from the fact that tumor tissue was available for only a subset of patients. The models adjusted for a long list of potential confounders, including age, body mass index, alcohol consumption, diet quality, physical activity, aspirin use, family history, and tumor characteristics. The team also adopted an unusually stringent significance threshold of P less than 0.005, following recommendations from a panel of methodologists seeking to reduce false-positive findings.

The central result was a statistical interaction between smoking and tumor LINE-1 methylation. The association between smoking status and mortality differed depending on the tumor’s methylation level, with a P value for interaction of 0.050 for overall mortality and 0.017 for colorectal cancer-specific mortality. Among patients whose tumors showed low LINE-1 methylation, below 60 percent, current smokers at diagnosis had an 80 percent higher risk of death from any cause compared with never smokers, corresponding to a hazard ratio of 1.80 with a 95 percent confidence interval of 1.19 to 2.73. For colorectal cancer-specific mortality in the same low-methylation group, the hazard ratio was 1.55, though its confidence interval crossed one. In sharp contrast, among tumors with high LINE-1 methylation of 68 percent or more, current smoking showed no clear association with either endpoint, with hazard ratios of 1.33 for overall mortality and 0.93 for cancer-specific mortality.

The authors propose a biological explanation rooted in the mechanics of genome maintenance. Experimental studies have shown that disruption of DNA methyltransferase 1, the enzyme responsible for copying methylation patterns during cell division, can trigger global DNA hypomethylation, which in turn promotes chromosomal instability, including loss of heterozygosity at the TP53 tumor suppressor locus. Tumors that have already accumulated this kind of instability may be especially vulnerable to additional mutagenic insults from the carcinogens in cigarette smoke. In other words, patients whose tumors ended up LINE-1 hypomethylated may have been more susceptible all along to the damaging effects of tobacco, and continued smoking after cancer develops may accelerate an already unstable genome toward lethal progression.

An immune dimension may compound the effect. Prior research has linked LINE-1 hypomethylation to reduced infiltration of T cells into colorectal tumors, hinting at an immunosuppressive tumor microenvironment. Separately, smoking has been associated with colorectal cancer subtypes that harbor fewer T cells and macrophages, implying that tobacco suppresses the effector immune cells that would otherwise attack the tumor. Taken together, the study’s results suggest that smoking and LINE-1 hypomethylation may interact to jointly weaken the tumor-immune relationship, producing a particularly aggressive disease course in patients who carry both exposures. This convergence of genomic instability and immune evasion could explain why the smoking penalty appears only in the low-methylation subgroup.

The study has notable strengths, particularly its foundation in molecular pathological epidemiology, a discipline that integrates lifestyle data collected prospectively over decades with detailed tumor molecular profiling. Repeated biennial questionnaires captured smoking, diet, and physical activity long before diagnosis, and the inverse probability weighting approach guarded against bias from missing tissue data. Sensitivity analyses without weighting, or restricted to early-stage cancers, produced consistent results. Still, the authors acknowledge limitations: post-diagnosis smoking changes could not be analyzed because many current smokers quit after diagnosis; treatment and recurrence data were limited; and the interaction findings did not meet the strict P less than 0.005 threshold, prompting the team to emphasize biological plausibility alongside statistical evidence. The sample size also precluded stratification by additional molecular markers such as microsatellite instability status.

Looking forward, the researchers argue that emerging technologies could sharpen the picture. Locus-specific methylation assays combined with machine learning can predict genome-wide methylation more accurately than the four-CpG pyrosequencing average used here, and nanopore sequencing now allows direct, real-time reading of long DNA fragments without amplification bias, opening the door to comprehensive profiling of repetitive element methylation. If larger studies confirm the interaction, LINE-1 methylation could become a practical biomarker for identifying which patients with colorectal cancer have the most to gain from smoking cessation, adding a molecularly targeted dimension to a modifiable behavior. For now, the message is clear: in a subset of colorectal tumors marked by epigenetic destabilization, cigarette smoke appears to do its worst damage.

Subject of Research: The interaction between cigarette smoking and tumor LINE-1 DNA methylation levels in colorectal cancer survival

Article Title: Smoking and colorectal cancer survival in relation to tumor LINE-1 methylation levels: a prospective cohort study

Article References: Kishikawa, J., Ugai, T., Fujiyoshi, K., Chen, Y., Haruki, K., Liu, L., Arima, K., Akimoto, N., Hamada, T., Inamura, K., Kosumi, K., Twombly, T. S., Shi, S., Lau, M. C., Du, C., Li, P., Guo, C., Väyrynen, J. P., Väyrynen, S. A., … Wu, K. (2022). Smoking and colorectal cancer survival in relation to tumor LINE-1 methylation levels: a prospective cohort study. Epigenetics Communications, 2(1), Article 4. https://doi.org/10.1186/s43682-022-00012-y

Image Credits: AI Generated

DOI: 10.1186/s43682-022-00012-y

Keywords: colorectal cancer, smoking, LINE-1 methylation, DNA methylation, epigenetics, cancer survival, molecular pathological epidemiology, genomic instability, tumor immunology, prospective cohort study, biomarker, Nurses' Health Study