Emulating target trials of behavioral interventions among cancer survivors: an editorial commentary

Over 18 million people are living with cancer in the United States, and this number is projected to exceed 22 million by 2035 (1). These individuals, and their doctors and families, often want to know what they can do to improve their prognosis. In particular, cancer patients frequently ask how they should change their diet, exercise, or alcohol use. The article by McGee et al. published in Epidemiology seeks to help answer this question by modeling hypothetical pragmatic randomized controlled trials using data from three prospective observational cohort studies (2).

Several prospective cohort studies were designed to examine health behaviors (like diet and physical activity) and risk of cancer, including the Nurses’ Health Studies, Health Professionals Follow-up Study, European Prospective Investigation into Cancer and Nutrition, and Multiethnic Cohort Study. As these cohorts have aged and accrued cancer cases, researchers have taken the opportunity to examine health behaviors after diagnosis in relation to cancer-specific and all-cause mortality.

Such research is methodologically challenging. Achieving conditional exchangeability is a challenge when risk factors for mortality among cancer survivors are complex, often correlated, and occur at multiple levels (e.g., individual, hospital, neighborhood). Addressing time-varying covariates that could be both confounders for, and causal intermediates between, exposures and outcomes add further challenge, especially given the variable time span from cancer diagnosis to death (3). Achieving consistency for nutrition and physical activity exposures is also difficult. For example, a recommendation of 5+ servings/day of fruits and vegetables can be achieved in multiple ways (e.g., all from one fruit or from many different vegetables), each of which can result in different nutrient intakes that in turn produce different biochemical processes in the body. An added complication is the substantial measurement error in self-reported health behaviors (4). Positivity can be difficult to achieve in survivorship studies, given that these analyses often have limited sample sizes and many covariates. Further, selection and measurement biases are concerns when collecting burdensome self-reported information at repeated time points from people who are also managing cancer diagnoses (5).

Though these biases exist to varying degrees across studies, relatively consistent results have been reported regarding post-diagnosis health behaviors and cancer survival. Indeed, the American Cancer Society published its Nutrition and Physical Activity Guidelines for Cancer Survivors based on results from systematic reviews and meta-analyses (6). From this body of literature, they concluded that survivors should engage in regular physical activity and follow a healthy eating pattern rich in a variety of vegetables, fruits, and whole grains and low in processed foods, red meats, and sugar-sweetened beverages.

While not all observational findings have been confirmed in randomized controlled trials, the recent results from the Colon Health and Lifelong Exercise Change (CHALLENGE) trial add evidence in support of the potential impact of such health behaviors (specifically exercise) on cancer mortality. In this trial among 889 stage III colon cancer patients, those randomized to a 3-year structured exercise program had 7% longer overall survival at 8 years compared to those who received usual care [absolute risk difference: 90.3% vs. 83.2%; hazard ratio (HR) for death: 0.63, 95% confidence interval (CI): 0.43–0.94] (7). This result is markedly similar to the estimates reported by observational studies examining post-diagnosis physical activity in relation to colon cancer survival. In a pooled analysis of six prospective cohort studies using standard methodologic approaches for this field, engaging in 7.5 to <15 metabolic equivalent task-hours/week (MET-h/wk) of moderate-to-vigorous physical activity versus none was associated with a 39% lower rate of mortality among colon cancer survivors (HR: 0.61; 95% CI: 0.50–0.76) (8).

Additional questions remain, of course, that may be most efficiently answered using observational data. For example: the optimal timing (relatively to diagnosis and treatment), dose, and duration of behavior changes to impact clinical outcomes as well as elucidation of moderators and mediators. To answer these questions, researchers may consider applying the approach used by McGee et al. (2). McGee et al.’s rigorous approach specifies and emulates target trials (hypothetical pragmatic randomized controlled trials) to overcome some of the limitations of observational analyses in cancer survivorship: the interventions are well-defined and the timing of eligibility, intervention start, and onset of follow-up are aligned, limiting potential immortal time and prevalent user biases (5). Applying these methods, McGee et al. observed mortality risk differences indicating benefits of the modeled nutrition and physical activity intervention versus no intervention. The 20-year all-cause mortality risk differences comparing adhering to nutrition and physical activity recommendations vs. no intervention were −9.5% (95% CI: −11.7% to −7.2%) in breast cancer survivors and −9.2% (95% CI: −12.5% to −5.4%) in prostate cancer survivors.

As McGee et al. acknowledge, the target trial approach overcomes some, but not all, limitations of observational survival analyses. Critically, it still relies on the unverifiable assumptions of conditional exchangeability and the ability to appropriately adjust for loss to follow-up with measured variables. Additionally, the health behaviors of individuals who volunteer to participate in research studies may not represent the distribution of those behaviors in the general population. These challenges perhaps explain McGee et al.’s estimated effect of abstaining from alcohol. In their target trial emulation among breast and prostate cancer survivors, the estimated mortality risk difference for abstaining from alcohol vs. no intervention was 3.3% (95% CI: 2.2–4.4%) for breast cancer survivors and 4.8% (95% CI: 1.7–7.4%) for prostate cancer survivors, indicating higher risk of mortality among those hypothetically assigned to abstain from alcohol.

McGee et al. went to great lengths to identify and quantify remaining biases in their analyses. A strength of the target trial framework is that it allows investigators to test modifications to their protocols to understand potential sources of bias, such as unmeasured confounding and positivity violations. After numerous sensitivity analyses, their results were relatively stable. Yet, as the authors acknowledge, the finding that (hypothetically) assigning breast and prostate cancer survivors to abstain from alcohol substantially increases risk of death seems implausible. The implausibility becomes yet more striking when considering the low amounts of alcohol that the participants in these cohorts reported consuming—median intake was 0.1 servings/day among breast cancer survivors and 0.6 servings/day among prostate cancer survivors. This underscores that limitations of the observational datasets used for target trial emulation, such as unmeasured confounding, measurement error, selection bias, and limited generalizability, will still impact study results. The researchers conclude that it may not be possible to resolve the question of alcohol use after cancer diagnosis from observational studies. Given the interest in this question, and that about 50% of cancer survivors report consuming some alcohol (9), perhaps a randomized controlled trial in which cancer survivors who are moderate drinkers are randomized to abstain from alcohol is warranted. A study such as this is currently underway among the general population of adults over 50 years old in Spain (10).

In conclusion, McGee et al. demonstrate another tool in the toolbox for epidemiologists striving to understand the role of health behaviors in cancer survivorship. Evidence is accruing that nutrition and physical activity have an important role to play in helping cancer patients improve their prognosis, lower their risk of other cardiometabolic conditions, and lengthen their overall survival (6). Rigorous and reproducible data are needed to inform policies and pay structures to incorporate nutrition and exercise services into cancer care. Moreover, considering that only 0.2% of breast cancer survivors and 0.3% of prostate cancer survivors met the seven nutrition and physical activity recommendations specified by McGee et al. at baseline, continued investment in research on the potential health effects of nutrition and physical activity after cancer, and studies on how best to optimize, implement, and sustain behavioral interventions, are needed to improve cancer survivorship.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Annals of Cancer Epidemiology. The article has undergone external peer review.

Peer Review File: Available at https://ace.amegroups.com/article/view/10.21037/ace-2025-13/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://ace.amegroups.com/article/view/10.21037/ace-2025-13/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Wagle NS, Nogueira L, Devasia TP, et al. Cancer treatment and survivorship statistics, 2025. CA Cancer J Clin 2025;75:308-40. [Crossref] [PubMed]
2. McGee EE, Hernán MA, Giovannucci E, et al. Estimating the Effects of Lifestyle Interventions on Mortality Among Cancer Survivors: A Methodologic Framework. Epidemiology 2025;36:705-18. [Crossref] [PubMed]
3. Keil AP, Edwards JK, Richardson DB, et al. The parametric g-formula for time-to-event data: intuition and a worked example. Epidemiology 2014;25:889-97. [Crossref] [PubMed]
4. Willett W. Nutritional Epidemiology. New York: Oxford University Press; 1998.
5. Hernán MA, Dahabreh IJ, Dickerman BA, et al. The Target Trial Framework for Causal Inference From Observational Data: Why and When Is It Helpful? Ann Intern Med 2025;178:402-7. [Crossref] [PubMed]
6. Rock CL, Thomson CA, Sullivan KR, et al. American Cancer Society nutrition and physical activity guideline for cancer survivors. CA Cancer J Clin 2022;72:230-62. [Crossref] [PubMed]
7. Courneya KS, Vardy JL, O'Callaghan CJ, et al. Structured Exercise after Adjuvant Chemotherapy for Colon Cancer. N Engl J Med 2025;393:13-25. [Crossref] [PubMed]
8. Rees-Punia E, Teras LR, Newton CC, et al. Leisure-time physical activity after diagnosis and survival by cancer type: a pooled analysis. J Natl Cancer Inst 2025;117:1689-98. [Crossref] [PubMed]
9. Baughman C, Norman K, Mukamal K. Adherence to American Cancer Society Nutrition and Physical Activity Guidelines Among Cancer Survivors. JAMA Oncol 2024;10:789-92. [Crossref] [PubMed]
10. Martínez-González MA. Should we remove wine from the Mediterranean diet?: a narrative review. Am J Clin Nutr 2024;119:262-70. [Crossref] [PubMed]