Artificially and Sugar-Sweetened Carbonated Beverage Consumption Is Not Associated with Risk of Lymphoid Neoplasms in Older Men and Women

Abstract

Background: Concern about the carcinogenic potential of aspartame was raised after an increase in lymphomas and leukemia was reported in an animal study at doses similar to human exposure. Two prospective cohort studies published after the report found inconsistent results for estimated aspartame intake, artificially sweetened beverage consumption, and risk of lymphoid neoplasms.

Objective: The objective of this study was to examine associations of artificially and sugar-sweetened carbonated beverage consumption (for comparison) and aspartame intake with risk of non-Hodgkin lymphoma (NHL) overall and by major histologic subtype in the Cancer Prevention Study-II Nutrition Cohort.

Methods: Among 100,442 adult men and women who provided information on diet and lifestyle factors in 1999, 1196 NHL cases were verified during a 10-y follow-up period. Cox proportional hazards regression was used to estimate multivariable-adjusted RRs and 95% CIs.

Results: In women and men combined, there were no associations of consumption of ≥1 (355 mL) servings/d of artificially (RR: 0.92; 95% CI: 0.73, 1.17; P-trend: 0.14) or sugar- (RR: 1.10; 95% CI: 0.77, 1.58; P-trend: 0.62) sweetened carbonated beverages with NHL risk, compared to no consumption (P-heterogeneity by gender: 0.11–1.00). Similarly, aspartame intake was not associated with NHL risk (RR: 1.02; 95% CI: 0.84, 1.24; P-trend: 0.69, top vs. bottom quintile). Associations with NHL subtype (multiple myeloma, diffuse large B-cell lymphoma, chronic lymphocytic leukemia/small lymphocytic lymphoma, and follicular and other B-cell lymphoma) were generally null.

Conclusion: These findings do not support associations of daily consumption of artificially or sugar-sweetened carbonated beverages, or aspartame, with NHL risk.

Introduction

Incidence rates of non-Hodgkin lymphoid neoplasms increased from the mid 1970s until 2005, most steeply from 1975 to 1990 (1,2), and since 2005 incidence rates have stabilized (1,3). Lymphoid neoplasms are a related but heterogeneous group of cancers with varied clinical and morphologic features (4), and different incidence patterns by subtype suggest unique risk factors (2). Apart from infection and severe immunosuppression (5,6), however, their etiology is poorly understood. Accumulating evidence suggests a possible role of tobacco (7), alcohol (8), BMI (9–12), type 2 diabetes (13), and physical inactivity (14,15) in the etiology of certain non-Hodgkin lymphoma (NHL)⁴ subtypes. Few dietary risk factors have been identified, but recent evidence raises concerns of an association with higher consumption of the artificial sweetener aspartame and its major dietary source, artificially sweetened carbonated beverages, and risk of hematopoietic cancers (16,17).

Non-nutritive sweetener consumption is increasing in the United States among both adults and children (18), and aspartame (L-α-aspartyl-L-phenylalanine methyl ester) is an artificial sweetener used in many low-calorie, low-carbohydrate, sugar-free products. Aspartame was first approved for restricted use in dry foods in 1981, in carbonated beverages in 1983, and for general purposes in 1996 (19). Although aspartame is one of the most extensively studied food additives, questions remain regarding its safety. The major gut hydrolysis products of aspartame are L-phenylalanine, aspartic acid, and methanol (20). Methanol can be metabolized to formaldehyde, a documented human carcinogen (21). In 2006, an Italian study (16) of 1800 rats found that rats fed aspartame over their lifetimes developed more lymphomas and leukemia than controls, in a dose-dependent manner, starting at a dose that was lower than the FDA's acceptable daily intake level for humans of 50 mg/kg body weight (22). However, a European Food Safety Authority review of this animal study dismissed the findings because of concerns that the animals had higher than normal rates of infection and inflammatory conditions, as well as uncertainty about correctness of diagnoses (20).

In the United States, the majority of aspartame found in the food supply is in low-calorie beverages (19). Two prospective cohort studies examined the relation between artificially sweetened beverages and incidence of NHL, with one study finding no association (23) and the other suggesting a positive association in men only, especially for multiple myeloma (17). The latter study also showed a higher risk of NHL with higher consumption of sugar-sweetened carbonated beverages (17), included as a control, suggesting that components of carbonated beverages other than aspartame may be associated with NHL risk. In addition to its value as a control for the vehicle (e.g., packaging, other ingredients in carbonated beverages), sugar-sweetened carbonated beverage consumption is of interest because it may indirectly influence NHL risk through an association with obesity and weight gain (10,24,25).

To clarify these associations, we examined the relation between artificially and sugar-sweetened carbonated beverages, aspartame intake, and risk of NHL and its major subtypes in the Cancer Prevention Study (CPS)-II Nutrition Cohort, a large prospective study of U.S. men and women aged 47–95 (median age: 69 y).

Methods

Study population.

The CPS-II Nutrition Cohort is a prospective study of cancer incidence and mortality in the United States, established in 1992 (26). At enrollment, participants completed a mailed self-administered questionnaire including information on demographic, medical, diet, and lifestyle factors. Follow-up questionnaires to update exposure information and to ascertain newly diagnosed cancers were sent biennially starting in 1997. The Emory University Institutional Review Board approves all aspects of the CPS-II Nutrition Cohort.

Follow-up for this analysis began on the date of completion of the 1999 follow-up questionnaire, when a 152-item modified Willett FFQ, which included specific questions regarding carbonated beverage consumption, was first administered (26). From the 151,344 Nutrition Cohort participants who returned the 1999 questionnaire, we excluded 19,150 who did not complete the 1999 FFQ, 3286 who were lost to follow-up (did not complete any follow-up questionnaires after 1999 and were not deceased), 25,823 who reported a history of cancer (other than nonmelanoma skin cancer) on any survey before analytic baseline, and 52 with an unknown diagnosis date. Also excluded were those participants who provided incomplete or improbable FFQ data (n = 2039), and who left the entire beverage section (n = 348) or carbonated beverage intake questions (n = 204) blank. After exclusions, a total of 100,442 (43,350 men and 57,092 women) remained.

Follow-up for each subject began on the date of the returned 1999 survey and continued until the date of cancer diagnosis, censoring (for loss to follow-up after the first follow-up interval, unverified self-report of lymphoid neoplasm, or new report of other cancer diagnosis), death, or 30 June 2009, whichever came first.

Case ascertainment.

Of the 1196 incident cases of lymphoid neoplasms in this analysis, 941 cases were initially identified by self-report (26) and subsequently verified by obtaining medical records or by linkage with state cancer registries (27). An additional 216 cases were identified through computerized linkage with the National Death Index (28); additional information on 133 of these cases was obtained through subsequent linkage with state cancer registries. The remaining 39 cases were identified during verification of another reported cancer.

International Classification of Diseases (ICD) for Oncology, Second and Third Edition (ICD-O-2 and ICD-O-3) codes were used to classify lymphoid neoplasms into the histologic subtypes defined by the 2001 WHO classification of tumors of hematopoietic and lymphoid tissues (2). Based on the recommendations of the InterLymph Pathology Working Group (29), ICD-O-2 and ICD-O-3 codes were grouped into the following categories: diffuse large B-cell lymphoma (DLBCL) (n = 258); chronic lymphocytic leukemia/small lymphocytic lymphoma (n = 267); follicular lymphoma (n = 142); multiple myeloma (n = 198); and other B-cell lymphomas (n = 192). Other subtypes were not examined separately: T-cell (n = 45) and NHL not otherwise specified (n = 94).

Assessment of carbonated beverage and aspartame intake.

Diet was assessed in 1999 and updated in 2003 with use of a modified Willett FFQ (26). In both 1999 and 2003, mean consumption of artificially and sugar-sweetened carbonated beverages [“1 glass, bottle, or can (355 mL)”] during the past year was queried with use of frequency categories ranging from “never” to “≥4 per day.” Reported intake of specific artificially and sugar-sweetened carbonated beverage types listed (i.e., cola with caffeine, other carbonated beverages with or without caffeine) were summed. The carbonated beverage questions were validated previously in U.S. women (30) and men (31) with use of multiple food records; attenuated correlation coefficients were 0.73–0.74 for artificially sweetened carbonated beverages (31) and 0.36–0.55 (noncola type) and 0.84 (cola type) for sugar-sweetened carbonated beverages (30,31). Participants were asked about “use of NutraSweet or Equal (1 packet) (not Sweet 'N Low)” (manufactured by the NutraSweet Corporation, formerly Searle and Co.). Frequency responses ranged from “never” to “≥6 per day.” Total aspartame intake was calculated with use of the following values: 180 mg aspartame/355 mL (1 serving) of low-calorie cola with caffeine, 90 mg/355 mL of other low-calorie soda with caffeine, and 70 mg/355 mL of other low-calorie soda without caffeine; and 20 mg aspartame per packet of NutraSweet or Equal reported, as used previously (17).

Long-term soda consumption patterns were examined with use of information provided in the parent CPS-II mortality cohort (32), 10 y before study baseline. In 1982, participants were asked to report their usual consumption of artificially and sugar-sweetened carbonated beverages/d as “cups, glasses, or drinks.” We included responses ranging from none to 10 servings/d, assuming 355 mL per serving. We did not calculate aspartame for this exposure. Participants were instructed to write “1/2” if they consumed the beverage less than daily.

Statistical analysis.

Five categories of artificially and sugar-sweetened carbonated beverage consumption ranged from nonuse of that beverage (referent group) to ≥1 355-mL cans (1 serving)/d for overall NHL. For NHL subtypes, the top category of consumption was ≥5 355-mL cans (or ≥1.78 L)/wk because of smaller numbers. All beverage exposures were cumulatively updated, where estimates from the baseline assessment in 1999 predicted NHL risk from 1999–2003, and the mean of 1999 and 2003 consumption predicted risk from 2003 to 2009. In addition, we examined associations of both beverage types with use of mutually exclusive categories, with nondrinkers of either type as the referent group. Aspartame was examined with use of nonconsumption as the referent category and quartiles among all others. We examined long-term soda consumption patterns separately for artificially and sugar-sweetened carbonated beverages as nonconsumption in 1982 and 1999 (referent); high consumption in 1982 (355 mL to 3.55 L/d or 1–10 drinks/d) and 1999 (≥1.78 L or ≥5 cans/wk); and other combinations. Cox proportional hazards regression was used to estimate HRs and 95% CIs for NHL risk of categories of consumption compared to nonconsumption, adjusting for age at baseline with use of the stratified Cox procedure with 1-y age strata. P values for linear trend were estimated by creating a continuous variable with use of an ordinal value for each category.

We included known and probable risk factors as covariates in the multivariable model, including smoking status (never, current, former, missing), BMI (<18.5, 18.5 to <25.0, 25.0 to <30.0, and ≥30.0 kg/m², or missing) and history of diabetes (yes/no). Energy intake (quintiles) was included in all models, and models of diet and sugar-sweetened carbonated beverages consumption controlled for the other beverage. Other potential confounders considered, but not included because they did not change the risk estimates by 5% (top vs. bottom category) compared to the age-adjusted RR, for any main exposure, were education, race, weight, height, waist circumference, recreational physical activity, sitting time, family history of hematopoietic cancer, hormone replacement therapy use (women), use of nonsteroidal anti-inflammatory drugs, cholesterol-lowering medication, alcohol consumption, and intake of beef, processed meat, animal protein, total milk, saturated fat, fruits, and vegetables, and tea or coffee consumption.

Sensitivity analyses excluded the first 2 y of follow-up. Potential effect modifiers examined included age, BMI, and alcohol use. Statistical interaction was assessed by creating an interaction term between continuous exposure variables and the potential effect modifier and evaluated with use of the likelihood-ratio test (33). To test for violation of the Cox proportional hazards assumption, data were examined both graphically and by creating interaction terms between exposures and time, and evaluated with use of the likelihood-ratio test (33). Results from 2-sided chi-square tests were considered statistically significant at P < 0.05. All analyses were conducted with use of SAS version 9.3 (SAS Institute).

Results

The age range of CPS-II participants was 47–95 y (median age: 69.0 y). At baseline, men and women in the cohort consumed a mean (10–90th percentile) of 795 mL (0–2.49 L) or 2.24 (0–7.0) cans of artificially sweetened carbonated beverages/wk, and a mean of 358 mL (0–1.12 L) or 1.01 (0–3.2) cans of sugar-sweetened carbonated beverages/wk. Median intakes were 174 mL (0.5 cans) and 99.4 mL (0.3 cans)/wk, respectively. Daily mean aspartame intake was 46.6 (0–144) mg, and median intake was 10 mg.

Age-standardized baseline characteristics of the participants in this analysis according to artificially sweetened beverage consumption are provided in Table 1. The proportion of men in the beverage categories generally increased with higher amounts of artificially sweetened carbonated beverage consumption. Artificially sweetened carbonated beverage drinkers were slightly younger, heavier, more sedentary, and more likely to have a higher waist circumference and history of diabetes. They were also more likely to be former smokers and to be taking nonsteroidal anti-inflammatory drugs. The diet of artificially sweetened carbonated beverage drinkers was slightly higher in energy and animal protein; it also contained less sugar-sweetened carbonated beverages and milk.

TABLE 1.

Baseline characteristics by artificially sweetened carbonated beverage consumption among 100,442 men and women in the CPS-II Nutrition Cohort, 1999–2009¹

		Artificially sweetened carbonated beverage intake2
	Overall	Never	>0–3 cans/mo	1–4 cans/wk	5–6 cans/wk	≥1 can/d
Sociodemographic characteristic
n	—	30,215	30,230	24,267	5,346	10,384
Gender, %
Men	43.2	43.0	39.5	44.8	48.3	48.0
Women	56.8	57.0	60.5	55.2	51.7	52.0
Age, y	69.2 ± 6.1	69.6 ± 6.2	69.9 ± 6.0	69.1 ± 6.0	68.1 ± 5.8	67.0 ± 6.0
Race, %
White/White-Hispanic	97.8	97.6	97.7	97.8	98.0	98.1
Black/Black-Hispanic	1.1	1.1	1.1	1.1	1.0	0.9
Other/missing	1.2	1.3	1.2	1.2	1.0	1.0
Education, %
Less than high school	5.2	5.6	4.9	5.0	5.2	4.9
High school degree	25.2	25.5	24.8	25.9	24.4	24.5
Some college/trade school	28.5	28.8	28.8	27.8	28.9	28.6
College graduate	40.5	39.5	41.0	40.6	41.0	41.3
Family history of hematopoietic cancer, %
No	96.6	96.6	96.5	96.4	96.9	96.6
Yes	3.4	3.4	3.5	3.6	3.1	3.4
History of diabetes, %
No	90.1	95.1	92.1	87.7	82.4	79.7
Yes	9.9	4.9	7.9	12.3	17.6	20.3
BMI (kg/m2), %
<18.5	1.5	2.5	1.5	0.9	0.7	0.5
18.5 to <25.0	39.9	47.6	43.5	34.3	28.4	25.5
25.0 to <30.0	37.5	33.6	36.5	41.0	42.7	41.1
≥30.0	15.8	11.0	13.3	18.2	22.9	27.5
BMI at age 18, kg/m2	21.2 ± 2.9	20.8 ± 2.8	21.1 ± 2.8	21.4 ± 3.0	21.7 ± 3.2	21.9 ± 3.3
Weight change from 1982 to 1999 baseline, kg	3.3 ± 8.0	2.8 ± 7.4	3.1 ± 7.5	3.6 ± 8.2	4.0 ± 9.0	4.7 ± 9.6
Waist circumference, cm	91.7 ± 13.3	89.7 ± 13.1	90.5 ± 13.0	93.1 ± 13.1	95.1 ± 13.2	96.2 ± 13.9
Physical activity (MET-h/wk),3 %
0 (no activity)	5.0	6.2	4.5	4.2	4.9	5.3
>0 to <17.5	60.5	60.6	61.1	60.1	60.0	59.7
≥17.5	32.7	31.5	32.7	33.9	33.5	33.5
Sitting time (h/d), %
<3	54.1	54.4	56.1	54.7	51.0	47.3
3–5	28.5	28.5	28.0	28.2	29.9	29.3
≥6	14.3	14.0	12.7	13.8	15.6	20.5
Cigarette smoking status, %
Never	42.0	44.4	44.2	40.9	36.1	34.1
Current	4.4	6.0	3.6	3.0	4.2	5.1
Former	45.8	42.1	44.3	47.9	51.9	53.5
NSAID use, %
Never	7.4	9.6	7.4	5.7	5.8	5.3
Current	62.9	59.1	62.2	65.6	66.6	67.2
Former	19.6	21.5	20.1	18.0	17.5	17.4
HRT use,4 %
Never	30.1	32.7	28.9	29.0	28.5	28.9
Estrogen replacement therapy only	32.7	31.6	33.8	32.7	33.4	32.0
Combined HRT	15.6	14.4	15.1	16.6	16.6	18.3
Dietary characteristics
Median artificially sweetened carbonated beverage intake,2 cans/d	0.1	0	0.1	0.42	0.84	1.2
Alcohol intake (drinks/d),5 %
Nondrinker	25.0	27.3	23.5	22.4	24.1	28.9
≤1	54.2	51.1	56.0	57.2	54.4	50.4
>1	20.9	21.6	20.5	20.4	21.5	20.7
Energy intake, kcal/d	1735 ± 552	1769 ± 560	1685 ± 536	1723 ± 543	1740 ± 539	1808 ± 587
1999 Sugar-sweetened carbonated beverage intake,2 cans/wk	1.0 ± 2.4	1.7 ± 3.3	0.8 ± 1.9	0.7 ± 1.5	0.6 ± 1.6	0.6 ± 2.1
1982 Sugar-sweetened carbonated beverage intake,2 cans/wk	4.0 ± 6.6	4.1 ± 6.8	3.2 ± 5.7	3.8 ± 6.1	4.8 ± 7.7	5.4 ± 8.2
1982 Artificially sweetened carbonated beverage intake,2 cans/wk	5.3 ± 8.1	2.3 ± 5.9	4.2 ± 7.0	6.0 ± 7.6	7.9 ± 9.2	10.3 ± 10.9
Aspartame intake, mg/d	46.6 ± 88.7	2.1 ± 10.2	10.7 ± 14.0	49.0 ± 31.5	116.5 ± 41.3	238.7 ± 151.7
Animal protein intake, g/d	44.4 ± 17.5	43.1 ± 17.3	43.6 ± 17.1	45.3 ± 17.2	46.6 ± 17.7	47.7 ± 19.0
Fruit and vegetable intake, servings/d	33.2 ± 16.8	32.9 ± 17.4	33.5 ± 16.6	33.2 ± 16.2	32.6 ± 15.9	33.2 ± 17.3
Milk intake, servings/wk	6.2 ± 6.2	6.3 ± 6.5	6.6 ± 6.4	6.2 ± 6.0	5.6 ± 5.7	5.4 ± 5.8
Total sucrose, g/d	42.1 ± 21.5	45.1 ± 23.1	40.8 ± 20.2	40.9 ± 20.2	39.6 ± 21.1	40.8 ± 22.5
Glycemic load	12.2 ± 4.4	12.6 ± 4.5	11.9 ± 4.3	12.1 ± 4.3	11.9 ± 4.3	12.4 ± 4.7
Glycemic index	52.7 ± 3.3	52.9 ± 3.3	52.5 ± 3.3	52.7 ± 3.2	52.8 ± 3.6	52.9 ± 3.8
Carbohydrates, % of energy	53.3 ± 8.4	53.9 ± 8.5	53.9 ± 8.3	53.1 ± 8.0	51.7 ± 8.4	51.4 ± 9.0

¹Values are means ± SDs unless otherwise indicated; some percentages do not add up to 100% because of missing data or rounding. CPS, Cancer Prevention Study; HRT, hormone replacement therapy; MET, metabolic equivalent; NSAID, nonsteroidal anti-inflammatory drug.

²One can is equivalent to 355 mL.

³METs are defined for each type of exercise-related physical activity as a multiple of MET of sitting quietly for 1 h.

⁴Percentages reflect HRT use among women only.

⁵Drinks are defined as containing 355-mL beer, 114-mL wine, or 43-mL liquor.

The association between consumption of artificially sweetened carbonated beverages with NHL in men and women combined is shown in Table 2. Daily consumption was not associated with risk of overall NHL nor with any major subtypes except for a borderline inverse trend with risk of multiple myeloma, although the CIs were wide and not statistically significant (RR: 0.70; 95% CI: 0.42, 1.17 for ≥1.78 L or ≥5 cans/wk vs. none; P-trend: 0.05). Daily consumption of sugar-sweetened carbonated beverages was not associated with total NHL risk (Table 3). The RRs for the association of sugar-sweetened carbonated beverages with risk of DLBCL, follicular, and other B-cell lymphomas were elevated in the top vs. reference category, but results were not statistically significant (P-trend: 0.13–0.41). For chronic lymphocytic leukemia/small lymphocytic lymphoma, risk was elevated for >0–1.07 L (>0–3 cans)/mo compared to nondrinkers, but there was no evidence of a dose-response association. Aspartame consumption, estimated from artificially sweetened carbonated beverages and NutraSweet/Equal packets, was not associated with risk of NHL overall (Table 4). Although the RR for DLBCL was statistically significantly higher in quintiles 2 and 3, risk was not significantly higher at greater levels of intake, and there was no trend (P = 0.51). Results excluding the first 2 y of follow-up, and removing BMI from the models, were similar (not shown).

TABLE 2.

Multivariable-adjusted RRs and 95% CIs for the association between artificially sweetened carbonated beverage consumption and the incidence of NHL and NHL subtypes among 100,442 men and women in the CPS-II Nutrition Cohort, 1999–2009¹

	Artificially sweetened carbonated beverage intake category2
	Nondrinkers3	>0–3 cans/mo	1–4 cans/wk	5–6 cans/wk	≥1 can/d	P-trend4	Continuous intake, per 1 can/d
All NHL (including multiple myeloma)
Cases/person-years	331/226,131	424/256,262	303/224,796	40/30,974	98/72,834	—	—
RR (95% CI)5	1.00 (—)	1.14 (0.98, 1.32)	0.97 (0.82, 1.14)	0.77 (0.56, 1.08)	0.92 (0.73, 1.17)	0.14	1.00 (0.98, 1.01)
All NHL (excluding multiple myeloma)
Cases/person-years	272/226,131	349/256,262	261/224,796	33/30,974	83/72,834	—	—
RR (95% CI)5	1.00 (—)	1.13 (0.96, 1.33)	1.02 (0.86, 1.22)	0.79 (0.55, 1.14)	0.97 (0.75, 1.26)	0.45	1.00 (0.98, 1.02)
NHL subtype
Multiple myeloma
Cases/person-years	59/226,131	75/256,262	42/224,796	22/103,808	—	—	—
RR (95% CI)5	1.00 (—)	1.15 (0.81, 1.63)	0.71 (0.47, 1.07)	0.70 (0.42, 1.17)	—	0.05	0.97 (0.92, 1.01)
DLBCL
Cases/person-years	67/226,131	90/256,262	71/224,796	30/103,808	—	—	—
RR (95% CI)5	1.00 (—)	1.23 (0.89, 1.70)	1.12 (0.80, 1.59)	0.92 (0.59, 1.45)	—	0.89	1.01 (0.98, 1.04)
CLL/SLL
Cases/person-years	80/226,131	92/256,262	68/224,796	27/103,808	—	—	—
RR (95% CI)5	1.00 (—)	0.97 (0.72, 1.32)	0.89 (0.64, 1.23)	0.71 (0.45, 1.12)	—	0.15	0.99 (0.95, 1.02)
Follicular lymphoma
Cases/person-years	41/226,131	46/256,262	36/224,796	19/103,808	—	—	—
RR (95% CI)5	1.00 (—)	1.08 (0.70, 1.66)	0.97 (0.61, 1.54)	0.98 (0.55, 1.74)	—	0.85	1.00 (0.96, 1.04)
Other B-cell lymphoma
Cases/person-years	46/226,131	72/256,262	51/224,796	23/103,808	—	—	—
RR (95% CI)5	1.00 (—)	1.41 (0.97, 2.06)	1.24 (0.82, 1.88)	1.13 (0.67, 1.91)	—	0.59	1.00 (0.96, 1.04)

¹CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; CPS, Cancer Prevention Study; DLBCL, diffuse large B-cell lymphoma; NHL, non-Hodgkin lymphoma.

²One can is equivalent to 355 mL; highest intake category for all subtypes is ≥5 cans (≥1.78 L)/wk.

³Reference category.

⁴ P-trend calculated by assigning an ordinal value to each category and modeling as a continuous variable.

⁵Multivariable model adjusted for age at baseline, gender, history of diabetes, BMI, smoking status, energy intake, and sugar-sweetened beverage intake.

TABLE 3.

Multivariable-adjusted RRs and 95% CIs for the association between sugar-sweetened carbonated beverage consumption and the incidence of NHL and NHL subtypes among 100,442 men and women in the CPS-II Nutrition Cohort, 1999–2009¹

	Sugar-sweetened carbonated beverage intake category2
	Nondrinkers3	>0–3 cans/mo	1–4 cans/wk	5–6 cans/wk	≥1 can/d	P-trend4	Continuous intake, per 1 can/d
All NHL (including multiple myeloma)
Cases/person-years	404/297,676	525/334,090	206/147,954	26/10,928	35/20,349	—	—
RR (95% CI)5	1.00 (—)	1.10 (0.96, 1.26)	0.96 (0.80, 1.14)	1.38 (0.92, 2.08)	1.10 (0.77, 1.58)	0.62	1.00 (0.98, 1.03)
All NHL (excluding multiple myeloma)
Cases/person-years	328/297,676	448/334,090	172/147,954	24/10,928	26/20,349	—	—
RR (95% CI)5	1.00 (—)	1.18 (1.02, 1.37)	1.02 (0.83, 1.24)	1.64 (1.08, 2.51)	1.05 (0.70, 1.59)	0.34	1.01 (0.98, 1.04)
NHL subtype
Multiple myeloma
Cases/person-years	76/297,676	77/334,090	34/147,954	11/31,276	—	—	—
RR (95% CI)5	1.00 (—)	0.77 (0.56, 1.08)	0.71 (0.46, 1.09)	0.92 (0.47, 1.80)	—	0.21	1.00 (0.93, 1.06)
DLBCL
Cases/person-years	90/297,676	103/334,090	50/147,954	15/31,276	—	—	—
RR (95% CI)5	1.00 (—)	0.98 (0.73, 1.31)	1.07 (0.74, 1.54)	1.41 (0.79, 2.52)	—	0.41	1.03 (0.97, 1.08)
CLL/SLL
Cases/person-years	79/297,676	138/334,090	41/147,954	9/31,276	—	—	—
RR (95% CI)5	1.00 (—)	1.43 (1.07, 1.90)	0.88 (0.59, 1.31)	0.77 (0.38, 1.58)	—	0.53	0.92 (0.85, 1.00)
Follicular lymphoma
Cases/person-years	51/297,676	52/334,090	30/147,954	9/31,276	—	—	—
RR (95% CI)5	1.00 (—)	0.98 (0.66, 1.47)	1.36 (0.83, 2.21)	1.70 (0.79, 3.62)	—	0.13	1.04 (0.98, 1.10)
Other B-cell lymphoma
Cases/person-years	58/297,676	92/334,090	30/147,954	12/31,276	—	—	—
RR (95% CI)5	1.00 (—)	1.36 (0.97, 1.92)	1.02 (0.64, 1.63)	1.82 (0.94, 3.53)	—	0.26	1.02 (0.95, 1.08)

¹CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; CPS, Cancer Prevention Study; DLBCL, diffuse large B-cell lymphoma; NHL, non-Hodgkin lymphoma.

²One can is equivalent to 355 mL; highest intake category for all subtypes is ≥5 cans (≥1.78 L)/wk.

³Reference category.

⁴ P-trend calculated by assigning an ordinal value to each category and modeling as a continuous variable.

⁵Multivariable model adjusted for age at baseline, gender, history of diabetes, BMI, smoking status, energy intake, and artificially sweetened carbonated beverage intake.

TABLE 4.

Multivariable-adjusted RRs and 95% CIs for the association between aspartame intake and the incidence of NHL and NHL subtypes among 100,442 men and women in the CPS-II Nutrition Cohort, 1999–2009¹

	Aspartame intake quintile (median intake at baseline, mg/d)2
	Q13 (0)	Q2 (3.6)	Q3 (12.6)	Q4 (35.8)	Q5 (145)	P-trend4	Continuous intake, per 50 mg/d
All NHL (including multiple myeloma)
Cases/person-years	230/160,783	266/163,104	260/161,065	234/162,918	206/163,126	—	—
RR (95% CI)5	1.00 (—)	1.29 (1.08, 1.54)	1.27 (1.06, 1.52)	1.10 (0.91, 1.32)	1.02 (0.84, 1.24)	0.69	0.99 (0.95, 1.03)
All NHL (excluding multiple myeloma)
Cases/person-years	187/160,783	221/163,104	215/161,065	203/162,918	172/163,126	—	—
RR (95% CI)5	1.00 (—)	1.31 (1.08, 1.60)	1.29 (1.06, 1.58)	1.18 (0.96, 1.45)	1.07 (0.86, 1.33)	0.83	1.00 (0.96, 1.04)
NHL subtype
Multiple myeloma
Cases/person-years	43/160,783	45/163,104	45/161,065	31/162,918	34/163,126	—	—
RR (95% CI)5	1.00 (—)	1.18 (0.77, 1.81)	1.18 (0.77, 1.81)	0.74 (0.46, 1.20)	0.83 (0.51, 1.33)	0.14	0.93 (0.84, 1.04)
DLBCL
Cases/person-years	40/160,783	63/163,104	56/161,065	50/162,918	49/163,126	—	—
RR (95% CI)5	1.00 (—)	1.82 (1.22, 2.72)	1.62 (1.07, 2.45)	1.38 (0.90, 2.11)	1.39 (0.90, 2.16)	0.51	1.02 (0.94, 1.10)
CLL/SLL
Cases/person-years	59/160,783	58/163,104	55/161,065	52/162,918	43/163,126	—	—
RR (95% CI)5	1.00 (—)	1.05 (0.73, 1.52)	1.02 (0.70, 1.48)	0.95 (0.65, 1.39)	0.85 (0.56, 1.29)	0.40	0.96 (0.88, 1.05)
Follicular lymphoma
Cases/person-years	27/160,783	31/163,104	27/161,065	29/162,918	28/163,126	—	—
RR (95% CI)5	1.00 (—)	1.41 (0.84, 2.39)	1.22 (0.71, 2.10)	1.21 (0.71, 2.09)	1.20 (0.69, 2.11)	0.72	1.03 (0.94, 1.13)
Other B-cell lymphoma
Cases/person-years	35/160,783	37/163,104	48/161,065	43/162,918	29/163,126	—	—
RR (95% CI)5	1.00 (—)	1.18 (0.74, 1.89)	1.58 (1.01, 2.46)	1.39 (0.88, 2.20)	1.01 (0.60, 1.68)	0.63	0.94 (0.85, 1.05)

¹CLL/SLL, chronic lymphocytic leukemia/small lymphocytic lymphoma; CPS, Cancer Prevention Study; DLBCL, diffuse large B-cell lymphoma; NHL, non-Hodgkin lymphoma; Q, quintile.

²Gender-specific aspartame quintiles (mg/d) are as follows: 1999 men = 0, >0 to 6.8, >6.8 to 24.2, >24.2 to 82.3, >82.3; updated mean [(1999+2003)/2] men = ≤0.2, >0.2 to 6.8, >6.8 to 27.0, >27.0 to 82.3, >82.3; 1999 women = 0, >0 to 5.0, >5.0 to 17.5, >17.5 to 70.0, >70.0; updated mean women = 0, >0 to 5.4, >5.4 to 19.6, >19.6 to 63.4, >63.4.

³Reference quintile.

⁴ P-trend calculated by assigning an ordinal value to each quintile and modeling as a continuous variable.

⁵Multivariable model adjusted for age at baseline, gender, history of diabetes, BMI, smoking status, energy intake, and sugar-sweetened carbonated beverage intake.

Results for men and women separately are shown in Supplemental Tables 1–3. Associations were not statistically significantly heterogeneous by gender (P-heterogeneity: 0.11–1.0) for NHL overall and for NHL subtypes). However, in women, the trend for an association between artificially sweetened carbonated beverages and overall NHL was inverse (P = 0.03), but the RR was not statistically significant for the highest intake category (RR: 0.75; 95% CI: 0.52, 1.08). This was likely driven by an inverse association with multiple myeloma (RR: 0.50; 95% CI: 0.20, 1.26; P-trend: 0.04), because the association with all NHL excluding multiple myeloma was RR: 0.83; 95% CI: 0.56, 1.22; P-trend: 0.14. However, nonsignificant inverse associations for artificially sweetened carbonated beverages were observed for all subtypes in women (P-trend: 0.10–0.43), except other B-cell lymphomas. For women, but not men, the association between aspartame intake and multiple myeloma was slightly inverse, but not so for all NHL.

When artificially and sugar-sweetened carbonated beverages were characterized with use of mutually exclusive categories, results remained similar: compared with participants who did not report drinking either carbonated beverage in 1999, those who drank only artificially sweetened carbonated beverages (29.1%) or only sugar-sweetened carbonated beverages (21.9%) were not at risk of NHL. For artificially sweetened carbonated beverages, compared with nonconsumers, the RR for occasional consumers (>0 but <1.78 L or >0 to <5 cans/wk) was 1.15 (95% CI: 0.90, 1.46), and for ≥1.78 L/wk the RR was 0.84 (95% CI: 0.61, 1.16). For sugar-sweetened carbonated beverages, the RR for occasional (>0 but <1.78 L/wk) vs. nonconsumers was 1.11 (95% CI: 0.86, 1.42), and for ≥1.78 L/wk the RR was 1.30 (95% CI: 0.88, 1.93).

The associations according to long-term carbonated beverage consumption are shown in Table 5. In women, long-term regular consumption of artificially sweetened carbonated beverages was associated with a lower NHL risk (RR: 0.53; 95% CI: 0.32, 0.87), compared with long-term nondrinkers. Women who consistently consumed sugar-sweetened carbonated beverages had more than double the risk of NHL compared with nondrinkers (RR: 2.29; 95% CI: 1.21, 4.32). These models additionally included weight change from 1982–1999, although risk estimates were similar without this variable. Results were similar when education or alcohol consumption at either time point was included in the models.

TABLE 5.

Multivariable-adjusted RRs and 95% CIs for the association between long-term artificially sweetened and sugar-sweetened carbonated beverage intake (reported in 1982 and 1999) and the incidence of NHL among 100,442 men and women in the CPS-II Nutrition Cohort, 1999–2009¹

	Long-term carbonated beverage intake category2
	No past and no current intake3	Low past and low current intake4	High past intake/ low current intake5	Low past intake/ high current intake6	High past and high current intake
Artificially sweetened carbonated beverage
Men
Cases/person-years	61/36,917	135/70,788	56/26,870	25/18,448	37/20,864
RR (95% CI)2	1.00 (—)	1.08 (0.79, 1.48)	1.25 (0.84, 1.87)	0.87 (0.54, 1.41)	1.17 (0.74, 1.83)
Women
Cases/person-years	68/47,815	141/123,774	90/71,300	21/20,289	24/34,902
RR (95% CI)2	1.00 (—)	0.82 (0.61, 1.11)	0.94 (0.67, 1.33)	0.77 (0.46, 1.27)	0.53 (0.32, 0.87)
Sugar-sweetened carbonated beverage
Men
Cases/person-years	35/25,163	206/105,475	73/40,696	24/9,768	16/10,684
RR (95% CI)2	1.00 (—)	1.42 (0.98, 2.06)	1.33 (0.87, 2.03)	1.89 (1.10, 3.26)	1.19 (0.64, 2.22)
Women
Cases/person-years	71/64,484	176/139,904	36/34,664	5/6,223	12/6,210
RR (95% CI)2	1.00 (—)	1.30 (0.98, 1.74)	1.13 (0.74, 1.72)	0.89 (0.36, 2.24)	2.29 (1.21, 4.32)

¹CPS, Cancer Prevention Study; NHL, non-Hodgkin lymphoma.

²Mutually exclusive categories; multivariable models adjusted for age at baseline, gender, smoking status, BMI, weight change from 1982 to 1999, history of diabetes, energy, and long-term intake of the opposite carbonated beverage.

³Non-use in 1982 and 1999 (reference category).

⁴Low or no intake at either time point. Low past intake defined as occasional or <355 mL (<1 can)/d in 1982; low current intake defined as >0 to <1.78 L (>0 to <5 cans)/wk in 1999.

⁵High past intake = 355 mL to 3.55 L (1–10 cans)/d in 1982; low current intake may include those with no intake.

⁶High current intake = ≥1.78 L (≥5 cans)/wk in 1999; low past intake may include those with no intake.

Age modified the association between artificially sweetened carbonated beverage consumption and NHL (including multiple myeloma, P-interaction: 0.02): among participants aged ≥70 y at baseline, associations were inverse (RR: 0.80; 95% CI: 0.66, 0.98 for each additional 355-mL can/d), whereas no association was observed among participants aged <70 y (RR: 1.05; 95% CI: 0.93, 1.20).

Discussion

In this large prospective cohort of U.S. adults, consumers of ≥1 artificially or sugar-sweetened carbonated beverages/d were not at higher risk of NHL or its subtypes during a 10-y follow-up period with intake measured at baseline and updated 4 y later. Although an inverse trend with artificially sweetened carbonated beverage consumption and risk of multiple myeloma was observed in women, the number of cases was small and the CIs were wide and nonsignificant. Aspartame intake, estimated from artificially sweetened carbonated beverage consumption and use of aspartame packets, also was not associated with NHL risk.

Controversy surrounding the safety of aspartame after its approval largely focused on the interpretation or quality of experimental data (34). These issues were again ignited when, in 2006, a very large experimental study including 1800 Sprague-Dawley rats found that rats fed aspartame at 0–4000 mg/kg body weight starting at 8 wk of life had a higher occurrence of lymphomas and leukemia, as low as 20 mg/kg body weight in females (16). In a subsequent study by the same research team, aspartame doses of 0, 20, or 100 mg/kg body weight were fed to 70–95 male and female rats from fetal day 12 through the lifespan; at necroscopy, a higher number of lymphomas/leukemia and mammary carcinomas were reported in rats given the highest dose (35). A re-examination of the studies by the European Food Safety Authority in 2013 concurred with earlier reviews (36–38) that higher chronic tissue inflammation and uncertainty of the diagnosis of some tumor types clouded interpretation of the studies' findings (20). The panel concluded that there was no reason to revise the current acceptable daily intake of 40 mg/kg per body weight per day, which is slightly lower than the current U.S. acceptable daily intake of 50 mg/kg per body weight per day. However, aspartame's safety remains a topic of debate (39) and of public health interest.

Our findings of null associations of diet soda and aspartame with NHL risk in men and women are consistent with those from the NIH-AARP Diet and Health Study, a prospective cohort of older U.S. men and women (23). However, a more recent analysis from the Nurses' Health Study and Health Professionals' Follow-up Study (HPFS) found that artificially and sugar-sweetened carbonated beverages and estimated aspartame intake were associated with a higher risk of lymphoma and leukemia in men but not in women (17). The authors speculate that men may be more susceptible to the effects of aspartame than women because of higher alcohol dehydrogenase activity (17). Alcohol dehydrogenase is the only human enzyme capable of metabolizing methanol to formaldehyde (40). Overall, reasons for the different findings among studies are unclear but may include the multiple updating of dietary information during follow-up in the HPFS, potentially leading to more accurate assessment of aspartame and carbonated beverages. However, the current study used a dietary assessment instrument comparable to that used in the HPFS and Nurses' Health Study and updated exposure once, 4 y after baseline, and did not observe a higher risk of lymphoma in men. It is also possible that positive associations in men in the HPFS were due to chance.

The inverse association with artificially sweetened carbonated beverage consumption and NHL risk in women over the long-term, observed herein, may be due to confounding by other unknown exposures because the authors are not aware of animal or human evidence to suggest an independent protective role. In models stratified by gender (Supplemental Table 1), the borderline inverse association of artificially sweetened carbonated beverage consumption with NHL risk in women is influenced in part by the association with multiple myeloma. However, these estimates are based on 85 women who developed multiple myeloma during follow-up and therefore may be due to chance. A large pooled analysis recently reported a positive association between BMI and risk of multiple myeloma in women only (12). Because consumers of artificially sweetened beverages in our cohort gained rather than lost weight over time, residual confounding from better weight control would not explain the inverse associations observed. Alcohol consumption is associated with lower NHL risk (8); however, alcohol consumption did not vary by artificially sweetened carbonated beverage intake (Table 1) and was not a confounder in our analyses.

In our primary analysis, consumption of sugar-sweetened carbonated beverages was not associated with NHL risk. Although RRs were elevated for follicular lymphoma and other B-cell lymphoma, CIs were wide and the trends nonsignificant. This is in contrast to the HPFS analysis, where associations with sugar-sweetened carbonated beverages paralleled the artificially sweetened carbonated beverages association, with 66% higher risks of NHL among men who were daily vs. nondrinkers of sugar-sweetened carbonated beverages (17). These results may be confounded by other behaviors related to carbonated beverage consumption in male health professionals (e.g., shift work or sleep patterns). However, in our secondary analysis of long-term beverage consumption, we observed an ∼1.9-fold increased risk among men who consumed low intakes of sugar-sweetened carbonated beverages in 1982 but higher intakes in 1999 (but not in men with consistently high intakes) and a 2-fold higher risk of NHL in women who consistently reported higher intake of sugar-sweetened beverages in 1982 and at study baseline in 1999. Approximately 15% of total energy intake in the United States is from added sugars, with soda and energy drinks contributing 36% of added sugars (25). Consumption of sugar-sweetened beverages is linked to increasing energy intakes over time (41). Emerging evidence suggests that a high BMI (9–12), diabetes (13), and low physical activity (15) may be risk factors for some NHL subtypes. Therefore, long-term associations observed between consumption of sugar-sweetened carbonated beverages, if real, may be mediated in part by their contribution to energy balance, and/or metabolic sequelae. However, risk estimates were not markedly different in models controlling for BMI or weight change.

The strengths and weaknesses of this study deserve mention. Our cohort comprises older, mostly white, middle-class adults with a moderate intake of artificially sweetened beverages and a fairly low intake of sugar-sweetened carbonated beverages, and exposure was assessed only twice during follow-up. This would tend to misclassify the exposures which would limit our ability to detect an association. Nevertheless, we were able to compare categories of at least daily consumption to nonconsumption, and secondary analyses examined long-term consumption patterns with use of historical data from 10 y before study baseline. We had adequate power (≥80%) to test previously observed positive associations with most exposures and outcomes. However, power to observe a RR of 1.3 for NHL in men only who consumed ≥355 mL (1 can) of artificially sweetened beverages per day was 53%; this may have hampered our ability to replicate this association. Nevertheless, we observed no suggestion of a positive association of artificially sweetened carbonated beverages or aspartame intake in men as reported previously (17). The median age of our participants at baseline was 69 y (range: 47–95 y); we cannot rule out whether aspartame or soda consumption is associated with risk from earlier life exposure. The number of cases, especially in gender-specific models of NHL subtypes, was small, making risk estimates less stable and chance findings possible. An advantage of this analysis was the coding of NHL subtypes in a manner consistent with the proposed InterLymph (29) coding system to enable comparison and pooling with other epidemiologic studies.

We do not know the exact artificial sweetener used in the carbonated beverages consumed in our cohort, and contributions from artificially sweetened noncarbonated beverages, yogurts, or ice cream were not ascertained in our study. However, the majority of aspartame in the U.S. food supply is found in artificially sweetened carbonated beverages (19), especially diet cola beverages, and aspartame is the most widely used low-calorie sweetener in diet carbonated beverages in the United States (42). Moreover, follow-up for the 3 studies that evaluated associations of estimated aspartame intake with risk of hematopoietic cancers (17,23), including the current study, occurred before recent decreases in aspartame demand and use in the food supply (43,44). Current trends show a growth in high-intensity sweetener use, including increases in other noncaloric sweetener use (43,44), in response to public concern over excessive sugar consumption (44). Future studies may be limited in their ability to examine higher levels of aspartame consumption if current trends continue, and they should assess the quantity and type of sweeteners used to evaluate the health impact of various noncaloric sweeteners in the food supply.

Overall, results of this study suggest that moderate consumption levels of artificially and sugar-sweetened carbonated beverages are unlikely to increase the risk of NHL in older men and women. Because of continuing concern over the safety of artificial sweeteners and their increasing use in the food supply, as well as concern over the consumption of added sugars, additional studies of younger individuals and individuals with higher consumption of both artificially and sugar-sweetened carbonated beverages and sweeteners are needed.

Supplementary Material

Supplemental Table 1.

Multivariable-Adjusted Rate Ratios (RR) and 95% Confidence Intervals (CI) for the Association Between Artificially-Sweetened Carbonated Beverage Consumption and the Incidence of Non-Hodgkin Lymphoma (NHL) and NHL Subtypes Among 43,350 Men and 57,092 Women, by Gender, in the Cancer Prevention Study II Nutrition Cohort, 1999-2009.¹

Abbreviations

CPS

Cancer Prevention Study

DLBCL

diffuse large B-cell lymphoma

HPFS

Health Professionals' Follow-up Study

ICD

International Classification of Diseases

NHL

non-Hodgkin lymphoma