Table 2.

Design and interpretation concerns for studies examining energy balance and cancer recurrence and survival

•Determination of the optimal cancer outcome to evaluate, for example, recurrence, progression to metastases, or cancer death. The decision rests on the following factors:
‐An ability to capture biology and the greatest clinical impact
‐The type of treatment and management a patient receives, which is partly dependent on the pathologic characteristics of the tumor at diagnosis
‐The time required to observe outcomes
•Inclusion of a comparison group to account for social interaction/placebo effects, which may be particularly important for trials targeting patient-reported outcomes.
•Control for confounding by pathologic characteristics (e.g., cancer stage, and grade); if obesity is associated with worse pathologic characteristics of the cancer, and these characteristics are strong prognostic factors, then adjustment is essential to determine whether obesity independently influences outcome.
•Control for confounding or effect modification for factors that may co-occur with obesity, for example, physical activity, energy intake, diabetes and other comorbid conditions, and smoking
•Knowledge that treatment and management of cancer might affect the interpretation of the results if:
‐The presentation or selection of treatment options varies by weight status
‐The treatment success varies by weight status, for example, whether the chance of positive surgical margins and thus recurrence varies by weight status
‐Potential for reverse causation as some treatments, for example, hormonal therapy, can lead to an increase in body fat accumulation, especially centrally, and metabolic perturbations
•Recognition that the effects of energy balance on cancer may differ depending on:
‐Cancer type and stage
‐Treatment
‐Race/ethnicity of the host
‐Body fat distribution
‐Other cofactors (e.g., smoking, comorbidity, medications)
‐Awareness that findings of studies conducted in non-Hispanic white survivors of nonmetastatic breast, prostate, or colorectal cancers (i.e., most studies conducted to date) may not generalize to survivors of other race/ethnicity, cancer type, or advanced cancer.
•Awareness that the effects of adiposity on cancer may differ depending on:
‐Body fat distribution and extent of body fatness
‐Volitional vs. nonvolitional weight loss
‐Rapid vs. slower weight change
‐Intermittent vs. continual exposure
•Realization that the effects of negative energy balance on cancer may differ depending on:
‐Diet composition
‐Type of physical activity
‐The magnitude of energy deficit
•Awareness that the measurement of diet and physical activity is difficult, and discrepancies in methods pose a challenge for pooling of data or carrying out meta-analyses.
•Consideration of the relationships between obesity, comorbidity, and treatment (which may or may not be independent) in the analysis and interpretation of results.
•Consideration within the study design and analysis to reduce and account for potential measurement error (e.g., energy intake, physical activity, and obesity).
•Lack of control for all components of energy balance (i.e., both energy intake and physical activity, as well as BMI).
•Lack of characterization of the study population about accrual (enrollees vs. larger pool of cancer survivors) and attrition (completers vs. dropouts).
•Adherence and long-term change in behavior.
•Adequate power to detect significant associations.