Association of Diet Quality With Survival Among People With Metastatic Colorectal Cancer in the Cancer and Leukemia B and Southwest Oncology Group 80405 Trial

IMPORTANCE Diet has been associated with survival in patients with stage I to III colorectal cancer, but data on patients with metastatic colorectal cancer are limited. OBJECTIVE To examine the association between diet quality and overall survival among individuals with metastatic colorectal cancer. DESIGN, SETTING, AND PARTICIPANTS This was a prospective cohort study of patients with metastatic colorectal cancer who were enrolled in the Cancer and Leukemia Group B (Alliance) and Southwest Oncology Group 80405 trial between October 27, 2005, and February 29, 2012, and followed up through January 2018. EXPOSURES Participants completed a validated food frequency questionnaire within 4 weeks after initiation of first-line treatment for metastatic colorectal cancer. Diets were categorized according to the Alternative Healthy Eating Index (AHEI), Alternate Mediterranean Diet (AMED) score, Dietary Approaches to Stop Hypertension (DASH) score, and Western and prudent dietary patterns derived using principal component analysis. Participants were categorized into sex-specific quintiles. MAIN OUTCOMES AND MEASURES Multivariable hazard ratios (HRs) and 95% CIs for overall survival. RESULTS In this cohort study of 1284 individuals with metastatic colorectal cancer, the median age was 59 (interquartile range [IQR]: 51-68) years, median body mass index was 27.2 (IQR, 24.1-31.4), 521 (41%) were female, and 1102 (86%) were White. There were 1100 deaths during a median follow-up of 73 months (IQR, 64-87 months). We observed an inverse association between the AMED score and risk of death (HR quintile 5 vs quintile 1, 0.83; 95% CI, 0.67-1.04; P = .04 for trend), but the point estimates were not statistically significant. None of the other diet scores or patterns were associated with overall survival. CONCLUSIONS AND RELEVANCE In this prospective analysis of patients with metastatic colorectal cancer, diet quality assessed at initiation of first-line treatment for metastatic disease was not associated with overall survival. JAMA Network Open. 2020;3(10):e2023500. doi:10.1001/jamanetworkopen.2020.23500 Key Points Question Is diet quality at initiation of first-line treatment for metastatic colorectal cancer associated with overall survival? Findings In this cohort study including 1284 people with metastatic colorectal cancer enrolled between 2005 and 2012, diet quality (assessed by the Alternative Healthy Eating Index, Alternate Mediterranean Diet score, Dietary Approaches to Stop Hypertension diet score, a Western dietary pattern, and a prudent dietary pattern) was not associated with overall survival after a median follow-up of 73 months. Meaning The results of this study suggest that overall diet quality assessed at initiation of first-line treatment for metastatic colorectal cancer was not associated with overall survival. + Invited Commentary Author affiliations and article information are listed at the end of this article. Open Access. This is an open access article distributed under the terms of the CC-BY License. JAMA Network Open. 2020;3(10):e2023500. doi:10.1001/jamanetworkopen.2020.23500 (Reprinted) October 30, 2020 1/9


Introduction
Colorectal cancer is the second leading cause of cancer death in the US. Approximately 22% of patients have metastatic disease at diagnosis, and 5-year survival for these individuals is only 14%. 1 Thus, there is a need to identify factors associated with survival of patients with metastatic colorectal cancer.
Among people with stage I to III colorectal cancer, a Western dietary pattern, 2 high glycemic load, 3 and sugar-sweetened beverages 4 are associated with higher risk of recurrence and death, whereas long-chain n-3 fatty acids, 5-7 dark fish, 5 fiber, 8 whole grains, 9 nuts, 10  Therefore, we examined overall diet quality and patterns assessed at the start of initial treatment for metastatic colorectal cancer in relation to overall survival. Our primary hypothesis was that a Western dietary pattern would be associated with shorter overall survival in people with metastatic colorectal cancer. 2

Study Design and Patients
This cohort study used data from individuals who were enrolled in Cancer and Leukemia Group B of the protocol included diet as a secondary objective that received institutional review board approval. Eligibility criteria, recruitment procedures, and outcomes of the study have been previously reported 12 ; briefly, participants had to be 18 years or older, have untreated locally advanced or metastatic colorectal cancer, have an Eastern Cooperative Oncology Group performance status of 0 or 1, and have normal laboratory values (eg, hepatic, renal, or hematologic). Patients were excluded if they had undergone major surgery in the previous 4 weeks or minor surgery in the previous 2 weeks. Prior adjuvant chemotherapy (up to 6 months) must have been completed at least 12 months before recurrence. Patients in the data set were deidentified. There was no difference in survival between treatment arms. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 13 An optional survey including a validated food frequency questionnaire (FFQ) [14][15][16][17][18] was completed by 1354 participants within 4 weeks after treatment initiation; this subset had similar characteristics to the total study population. To reduce measurement error in diet, we excluded patients missing more than 70 items on the FFQ, as well as women with an estimated caloric intake less than 500 or greater than 3500 kcal per day and men with estimated caloric intake less than 600 or greater than 4200 kcal per day, which resulted in 1284 individuals who were eligible for analysis.

Dietary Assessment
The FFQ assessed average intake of specified portions of approximately 130 items during the past 3 months, using as many as 9 response options, ranging from 0 to 6 or more servings per day. [14][15][16][17][18] We computed nutrient intake by multiplying the frequency of consumption of each food by the nutrient content of the specified portion using values from the US Department of Agriculture. 19 We adjusted for energy using the residual method. 20

Diet Quality Scores
We assessed 3 diet quality scores and 2 dietary patterns, as previously described. 21 The Alternative Health Eating Index (AHEI) is scored from 0 to 110 and is based on vegetables (excluding potatoes), fruits, whole grains, nuts and legumes, long-chain n-3 fatty acids, polyunsaturated fatty acids, sweetened beverages and juice, red and processed meat, trans fat, sodium, and alcoholic drinks. The Alternative Mediterranean Diet (AMED) is scored from 0 to 9 and is based on vegetables, fruits, nuts, whole grains, legumes, fish, ratio of monounsaturated to saturated fat, red and processed meat, and alcohol. The Dietary Approaches to Stop Hypertension (DASH) is scored from 0 to 45 and is based on fruits, vegetables, nuts and legumes, low-fat dairy, whole grains, sodium, sweetened beverages, red and processed meats, and sweets and desserts. Higher scores indicate healthier diets for these 3 scores.
Principal component analysis was used to identify common dietary patterns based on the FFQ data collected in our study population, as previously described. 2 The Western dietary pattern was characterized by higher intake of dairy, refined grains, condiments, red meat, and sweets and desserts. The prudent pattern was characterized by high intake of vegetables, legumes, and fruit.
Dietary pattern scores were generated for each participant based on intake of each item and the item's factor loading and standardized to mean 0 and standard deviation 1. Higher scores indicate diets more consistent with that specific pattern.

Covariates
Comprehensive data on potential confounding factors were obtained through medical record review and the participant questionnaire. Clinical data (eg, performance status, tumor location, prior resection, prior treatment for nonmetastatic disease, and a history of diabetes) and sociodemographic factors (age, sex, and race/ethnicity) were obtained from the medical record.
Methods to determine KRAS status of participants' tumors have been previously described. 12,22 Lifestyle factors, including body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), weight change in the past 6 months, and physical activity, 23,24 were obtained through the questionnaire.

Main Outcome and Follow-up
Our primary outcome was death from any cause. 12 Data collection occurred continuously from 2005 until the data set was locked in January 2018.

Statistical Analysis
Participants in CALGB/SWOG 80405 were pooled and analyzed as a prospective cohort. We used multivariable Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% CIs. [25][26][27] Overall survival was defined as time from completion of the FFQ until death or the end of follow-up.
We categorized the diet quality scores and patterns into sex-specific quintiles. Our basic model included age (years), sex, and daily energy intake (kcal). Our second model was additionally adjusted for race/ethnicity, performance status, protocol chemotherapy, primary tumor unresected, diabetes, treatment arm, KRAS, 12,22 tumor sidedness, weight change in the previous 6 months, BMI, and weekly total physical activity (metabolic equivalent task hours per week). A missing indicator was used to account for missing data in KRAS and tumor sidedness. One individual was missing data on BMI, 3 were missing data on physical activity, and 20 were missing data on weight change in the previous 6 months; these were assigned to the most common categories for multivariate adjustment. We evaluated the proportional hazards assumption using the Schoenfeld residual method. 28 To test for evidence of a linear trend (P trend), we modeled the median values of the diet score and pattern quintiles as ordinal variables in the multivariate Cox proportional hazards regression models and used a Wald test. 29, 30 We also explored whether associations differed by age vs no). For each, we added a cross-product term between the potential modifier and exposure of interest to our multivariate models and used Wald tests to evaluate whether there was evidence that associations varied by level of the modifier.
Reverse causation bias may occur if individuals change their diet before death because of their underlying illness. To assess potential bias resulting from reverse causation, we explored whether results changed when events that occurred 90 days or less after FFQ administration (n = 135; 12%) were excluded. This approach applies a minimum time between the exposure assessment and outcome ascertainment and thus reduces the potential for the outcome to influence the exposure.
Two-sided P values <.05 were considered statistically significant. Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc).

Discussion
In this prospective cohort study of 1284 individuals with metastatic colorectal cancer, there was no statistically significant association between diet quality or pattern assessed at initiation of first-line treatment for metastatic disease and overall survival. Patients and clinicians often seek advice on whether diet changes or other modifiable factors can impact outcomes. Although there are increasing data on modifiable factors, such as exercise and diet, in people with colorectal cancer, there remains a paucity of data to guide patients with advanced and metastatic disease. In a recent study with this same cohort of patients, 24 physical activity was observed to be associated with longer survival. In contrast, the current study did not find an association between overall diet quality and survival. Thus, although data are limited to 1 cohort so far, efforts to help patients adopt and maintain a physical activity routine may be more important than suggesting changes to their overall dietary pattern at the time of initiation of treatment for metastatic colorectal cancer.

Strengths and Limitations
This study has strengths, including its large sample size and number of events, comprehensive lifestyle and treatment data, and complete follow-up.
The study also has limitations. There was only a one-time, baseline measure of diet.
Consequently, these results may have been attenuated by nondifferential measurement error in diet, and it remains possible that diet before or after initiation of first-line treatment for metastatic disease may affect survival in these patients. In addition, reverse causation is a concern; people who feel ill before death may change their diet. However, all participants in this study had adequate bone marrow, liver, and renal function at enrollment and all but 1 had an Eastern Cooperative Oncology Group performance status of 0 or 1 when they completed the FFQ. Furthermore, the results did not change when patients who died within 90 days after the FFQ were excluded. A final limitation is that the majority of participants in this study were White, and the results may not be generalizable to people of other races/ethnicities.

Conclusions
Diet quality and patterns at initiation of first-line therapy for metastatic colorectal cancer were not associated with overall survival in this large prospective cohort study. Studies with repeated measures of diet, as well as studies examining diet in relation to quality of life, among patients with metastatic colorectal cancer are needed. Research examining diet in relation to colorectal cancer survival in more diverse populations is also needed.