Dietary habits have been recognized as a modifiable risk factor for many types of cancer. For example, a meta-analysis including more than three million participants demonstrated that high adherence to the Mediterranean diet was related to a lower risk of all-cause cancer mortality (1). Furthermore, a separate study including about 800,000 participants revealed that those who best complied with a dietary approach to stop hypertension (characterized by, e.g., low intake of saturated fat) had a 19% lower relative risk of colorectal cancer (2). Finally, in a prospective study from Sweden following about 1,000 community-dwelling older men over a median period of 13 years, dietary habits characterized by low carbohydrate and high protein intake were associated with a ~50% reduced risk to be diagnosed with prostate cancer (3). However, the association between diet and prostate cancer in this Swedish study was only seen among men with accurate reports of energy intake (3). Food intake reports of poor validity can distort the relationships between nutrient intake and health (4).

In addition to the type of diet, the 24-h temporal eating pattern may affect cancer risk in humans. For example, in a study among 2,413 women with early-stage breast cancer, a habitual short nightly fast of <13 h was associated with a 36% higher risk for breast cancer recurrence (5). Furthermore, in a case-control study, habitual fasting for more than 11 h overnight was associated with 23% lower odds of prostate cancer (6). In addition, a prospective French study involving 41,389 day-working men and women found that regular late eating, defined as an eating episode after 21:30 h, was associated with an increased breast and prostate cancer risk (+48 and +120%, respectively) (7). Finally, a population-based case-control study in Spain observed that those who habitually fasted at least 2 h before their nighttime rest had a 20% lower risk for breast and prostate cancer (8).

The cancer incidence is greater in older adults (9). However, it has not been systematically studied among seniors whether meal timing characteristics affect the risk of developing and dying from cancer. Here, we examined in 70-year-old men if the daily eating interval, the timing of the daily eating interval, and the day-to-day variability in the timing of eating are associated with the risk of being diagnosed with and dying of cancer. Furthermore, since unreliable food logs represent a source of bias in nutritional epidemiology (10), we paid particular attention to those with reliable dietary reports.

Several studies have shown that short regular daily eating intervals (definitions range between ≤11 and <13 h) reduce the risk of cancer and prolong the cancer remission period among cancer survivors (5, 6). However, whether meal timing characteristics, such as the length and timing of the daily eating interval and regularity of daily eating patterns, are associated with the risk of developing and dying from cancer in men of advanced age is largely unknown. In the present prospective cohort of community-dwelling older men with a maximum follow-up of over 22 years, we show that those eating on average 13 h per day and exhibiting an average day-to-day variability in the timing of eating of about 50 min or more have an increased risk for fatal cancer. These associations were seen despite adjustment for well-known cancer risk factors, e.g., smoking, dietary habits, and daily energy intake. If confirmed in future studies with larger sample size that involve assessment at a younger age and both sexes, our results suggest that restricting the daily eating interval and reducing the day-to-day variability in the timing of eating may have some potential in the prevention of fatal cancer among seniors.

The observation that the timing of the daily eating interval, measured by its midpoint, was not associated with cancer contrasts with some but not all previous studies. For example, a shorter daytime eating window has been shown to reduce the risk of prostate cancer in those starting eating at 08:30 or before. However, no such association was seen for those having a later onset of the daily eating interval (6), suggesting that not only the length but also the timing of the eating interval may impact the future cancer risk. However, a study among breast cancer survivors showed that fasting at least 13 h per day reduced the risk of breast cancer recurrence, irrespective of the timing of the offset of the eating interval (5).

Although our study cannot establish causality, several possible mechanisms could explain the association of the length of the daily eating duration and the day-to-day variability in the timing of eating with the risk of fatal cancer. For example, even during a shorter fasting period that can occur within a single day, healthy cells can switch from an anabolic state toward maintenance and repair state. This may limit the growth potential and mutagenic process of pre-malignant or cancerous cells (21). In contrast, proliferating cancer cells fail to adapt to scarce nutrient conditions because of a relative increase in their metabolic activity (22). Furthermore, fasting decreases blood levels of insulin-growth factor I (23, 24), a hormone that activates cellular oncogenes involved in malignant cell proliferation (25, 26). These anticancer properties may also explain why intermittent fasting has attracted significant interest as a potential lifestyle strategy in the prevention and treatment of cancer (27, 28). However, it must be kept in mind that results from intermittent fasting studies in rodents are controversial and suggest potential detrimental effects in certain oncological conditions (29). Finally, a greater day-to-day variability in the timing of eating may concur with a greater risk that food intake coincides with metabolically inappropriate endogenous circadian times. Notably, a temporal mismatch between behavioral and endogenous metabolic rhythms, e.g., as occurs in jetlag, has been associated with cancer in animal models (30).

Several strengths apply to our study, including the prospective study design and the extended follow-up, and that the observed associations remained significant following adjustment for well-known cancer risk factors. However, limitations of our study include the observational design and that residual confounding cannot be excluded. Furthermore, no potential site-specific associations between cancer outcomes and meal timing characteristics were examined. Additionally, dietary habits were only monitored at the age-70 investigation. The present study involved only men. Thus, it remains unclear whether meal timing characteristics affect cancer outcomes in women, although preliminary evidence is in favor of this hypothesis. For example, a daily eating interval shorter than 13 h may prolong cancer remission, as suggested by an investigation of female breast cancer patients (5). Finally, 29% of the cohort developed cancer during the observational period, and 20% died with cancer as the primary cause. These numbers are higher than those usually reported by, e.g., the U.S. National Cancer Institute (9), and could be explained by selection bias. With all these limitations in mind, other population-based studies, including women and repeated assessments of dietary characteristics, are needed to evaluate the generalizability of our findings further.

According to the widely applied Goldberg cut-off criterion (4), ~47% of the participants in the present study provided inaccurate reports of energy intake. Compared to men with accurate reports of daily energy intake, those with unreliable food logs had a two-point greater BMI and an about 31% lower daily energy intake. The magnitude of misreporting of energy intake by older men in the present study was greater than typically seen for younger men. For example, the percentage of underreporters in studies using estimated food records ranged from 14.3 to 42% for men (31). The underlying reasons for the relatively high proportion of older men with inaccurate energy intake reports in the present study are unclear. Still, they may include respondent memory lapses and misrepresentation of portion size consumed. Thus, screening for unreliable dietary reports is necessary when studying the association of nutritional patterns, including meal timing, with health outcomes, especially when including seniors. If ignored, inaccurate dietary reports can otherwise contribute to the lack of replicability and inconsistency across studies in the field of nutritional research, including nutritional epidemiology (10, 32). Future studies should also compare the efficacy of various methods to measure food intake among seniors reliably.

Our study indicates that curtailing the length of the daily eating interval and reducing the day-to-day variability in the timing of eating may be potential behavioral strategies to reduce the fatal cancer risk among older men. However, to fully unlock its potential, studies are needed to test whether recommendations to time-restrict the 24-h eating interval and have regularly timed daily eating intervals can reduce the risk of death due to cancer.

The original contributions presented in the study are included in the article/Supplementary Materials, further inquiries can be directed to the corresponding author/s.

The studies involving human participants were reviewed and approved by the Regional Ethics Review Board at Uppsala University (https://etikprovningsmyndigheten.se/). The patients/participants provided their written informed consent to participate in this study.

EM and CB came up with the research idea. EM, LvE, TM, and CB analyzed data. EM wrote the paper. CB had primary responsibility for final content. All authors read and approved the final manuscript.

The authors' work is funded by the Novo Nordisk Foundation (NNF19OC0056777) and Swedish Brain Research Foundation (FO2020-0044). The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.