Sweet Beverages and Cancer: A Scoping Review of Quantitative Studies

Sweet beverages (SB) are the largest single source of calories (1) in the United States and globally one of the leading behavioral risk factors contributing to the increase in attributable deaths and disability adjusted life years (2). Often defined as drinks with no nutritional benefit (e.g., soda/pop, fruit-flavored drinks), SB have been linked to obesity (3, 4), type II diabetes (5, 6), and coronary heart disease (7) in a substantial body of literature. However, despite plausible mechanisms through which excessive SB consumption can increase cancer risk, fewer studies have examined this relationship.

The amount of sugar provided by any type of SB can contribute to excessive energy intake and increased adiposity, which can contribute to obesity. Potential mechanisms through which obesity can increase cancer risk include greater insulin resistance, hormonal imbalances, and angiogenesis (8, 9). There are also mechanisms through which SB can contribute to cancer directly, such as by increasing dietary glycemic index and insulin glucose (10, 11).

The fructose provided by SB also has distinct mechanistic properties. Unlike glucose, fructose is predominantly metabolized in the liver. Human studies have reported fructose intake's contribution to impaired glucose tolerance, insulin resistance, and hyperinsulinemia (12–16). As such, higher fructose can alter hepatic insulin and lipid metabolism, which can have a direct impact on cancer risk. Yet, specific types of SB have varying ratios of glucose, fructose, and sucrose; and these ingredients can vary in different regions where manufactured (17). For example, over the past 30 years in the United States, carbonated drinks (CD) and sugar-added fruit, sports, and energy drinks are most commonly sweetened with high fructose corn syrup, which is typically 45% glucose and 55% fructose (18, 19). Yet, common SB sweeteners differ in other regions, for example, Australia (i.e., sugarcane-derived sucrose that is 50% glucose and 50% fructose) and Europe (i.e., sucrose-rich sugar beet; ref. 17). Finally, in contrast to manufacturer-added sugars, 100% fruit juices have natural sugars in the form of fructose.

In addition, artificially sweetened beverages typically provide no energy and no sucrose, glucose or fructose. Also, different types of artificial sweeteners have distinct chemical characteristics (e.g., saccharin vs. aspartame) and properties (20, 21). Because of these differences, consumption of artificially sweetened beverages can further confound and complicate interpretation of cancer risk (22–24). Because of the lack of a consistent beverage classification system, disentangling the contributions of beverages and potential cancer-related mechanistic actions is a notable challenge within existing cohort and case–control studies, as well as other reviews.

Preliminary research has linked SB to the increased risk of several adiposity-related cancers (25–33) and two recent reviews of SB and cancer have reported mixed results (34, 35). However, these reviews do not entirely account for the potential mechanistic actions between SB and cancer through including artificial sweeteners with SB, or not fully analyzing SB association with adiposity-related cancers versus non–adiposity-related cancers (34, 35). Thus, we conducted a scoping review (36) of quantitative studies that included distinct conceptualizations of SB based on the mechanistic literature (e.g., added sugar vs. natural sugars/100% fruit juice), where artificial sweeteners were not included in analyses. We also examined associations between SB and adiposity-related cancers, versus non–adiposity-related cancers.

We used the PRISMA Scoping Review Guidelines to conduct a scoping review of quantitative studies of SB and cancer.

The sample of published studies was drawn from a systematic search conducted in March 2020 of six bibliographical databases: PubMed, Medline, Web of Science, SCOPUS, Psychology and Behavioral Sciences Collection, and Embase. The search included: sugar-sweetened beverage, fruit juice, soda, soft drink, pop, carbonated beverage, energy drink, and cancer (e.g., sugar-sweetened beverage AND cancer, fruit juice AND cancer). After removing duplicates, the search yielded 2,799 records (Fig. 1).

Title and abstracts were used to conduct the screening process. Eligible studies were peer-reviewed articles reporting a quantitative association between SB intake and a cancer-related health outcome. SB was defined as beverages with added sugars, 100% fruit juices (not vegetable juices or a mix of fruit and vegetable juices or extracts), or fruit drinks that were not 100% fruit juice. Studies were excluded that were not quantitative, were published in languages other than English, conducted studies with animals, examined only specific juice extracts, did not include analyses specific to SB intake, or did not specifically examine relationships between SB and a cancer outcome. The references of studies selected for inclusion were searched for other relevant studies, and a citation search for the studies included was conducted to search for subsequent follow-up studies. This search process yielded 52 studies. These studies were then reviewed in detail to exclude those that did not report the independent effect of nonartificial sweeteners, yielding 38 studies.

Two authors independently coded each study, with discrepancies resolved through discussion. Each study was coded for the year it was published, year the data were collected, sample size, gender(s), age, race/ethnicity, location, sampling method, research design, cancer outcome, and SB conceptualization. Level of statistical adjustment for the SB-cancer relationship was also coded as either using bivariate analysis only or multivariate analysis controlling for confounding variables. Results were coded as worse cancer outcome, better cancer outcomes, and mixed results.

Descriptive statistics were conducted using SAS version 9.3 to assess: (i) study characteristics; (ii) study findings about SB by cancer type (adiposity-related cancer vs. not); and (iii) study findings about cancer by SB type (CD; 100% fruit juices; sugar-added fruit, sports, and energy drinks). We examined associations between SB and specific cancers if there were at least five studies per cancer type.

The data generated in this study are available within the article and its Supplementary Data.

Thirty-eight separate studies examined the SB and cancer relationship (Table 1). Sample sizes ranged from 302 to 487,922, with a median sample size of 35,376. Both genders were included in most studies (76%, k = 29/38) with 3% (k = 1/38) sampling men only (37) and 21% (k = 8/38) sampling women only (32, 38–44). All but one study included adults (97%, k = 37/38), with one including only children/youth (less than 18 years old; ref. 45). For studies conducted within the United States (k = 24), race/ethnicity of the samples included 75% (k = 18/24) White or predominantly White (at least 80% of the sample; refs. 27, 32, 37–41, 44, 46–54), 17% (k = 4/24) mixed race/ethnicity (45, 55–57), 4% (k = 1/24) other race/ethnicities (58), and 4% (k = 1/24) not stating race/ethnicity (59). Over one-third used samples outside of the United States (37%, k = 14/38), with the majority being European (k = 9/14; refs. 23, 28, 60, 61, 62, 63, 64, 65–67).

The majority of studies were longitudinal (36/38), with 79% prospective longitudinal (30/38) and 5% retrospective longitudinal (2/38; refs. 58, 59). A total of 16% were cross sectional (6/38; refs. 48, 57, 60, 61, 63, 68). A total of 74% used a cohort design (28/38), 24% used a case–control design (9/38; refs. 45, 48, 52, 55, 58–61, 63), and 3% used some other design (1/38; ref. 57). Over half (63%; k = 24/38) used a procedure to seek a representative sample of their population, whereas 37% (k = 14/38; refs. 30, 41, 43, 46–48, 50, 53, 61, 62, 65–67, 69) did not. The data source was reported to be nationally representative for 5% (k = 2/38; refs. 57, 68) of studies. All but one study (ref. 61; 97%; k = 37/38) used multivariate analysis.

SB was conceptualized as: (i) CD; (ii) 100% fruit juice; and (iii) sugar-added fruit, sports, and energy drinks. Studies also conceptualized SB as a combination of the three categories mentioned above, including: (iv) CD, sugar-added fruit, sports, and energy drinks; and (v) CD, 100% fruit juice, sugar-added fruit, sports, and energy drinks.

Some studies reported multiple analyses of several SB conceptualizations in a single paper, thus are represented in more than one of the following SB analyses. Nearly half (45%; k = 17/38; refs. 27, 28, 45, 46, 48, 49, 52–56, 59, 60, 62–64, 69) reported the independent effect of CD. Half reported the independent impact of 100% fruit juice (50%; k = 19/38; refs. 23, 28, 37–40, 42, 44, 45, 49, 50, 55, 56, 58, 60, 64–67). About half reported analyses whereby SB was conceptualized as CD, and sugar-added fruit, sports, and energy drinks (45%; k = 17/38; refs. 23, 30, 32, 37–39, 41, 43, 46–48, 50, 51, 57, 65, 66, 68). SB was conceptualized in analyses as sugar-added fruit, sports, and energy drinks in 5% (k = 2/38; refs. 32, 48). Similarly, SB was conceptualized in analyses as 100% fruit juice and sugar-added fruit, sports, and energy drinks in 8% (k = 3/38; refs. 23, 32, 50) of studies.

Food frequency questionnaires (FFQ) were used most often to assess SB intake (66%; k = 25/38). A total of 5% of studies (k = 2/38; refs. 23, 57) used dietary recalls only, or the Dietary Health Questionnaire (5%; k = 2/38; refs. 37, 60). The remainder (24%; k = 9/38; refs. 45, 52, 58, 59, 61, 65–67, 69) reported using a mixture of assessments.

Outcomes of 11 different cancers were assessed. Cancer-related outcomes included mortality and/or incidence for all studies except for one, which examined risk of colorectal polyps (38). A total of 74% (28/38; refs. 23, 27, 28, 30, 32, 38, 39, 41–43, 46, 48–52, 54–56, 59, 61–63, 65–69) examined SB's association with adiposity-related cancers. A total of 10/38 (26%; refs. 23, 47, 50, 51, 55, 57, 62, 64, 68, 69) studies examined the outcomes across a variety of cancers. A total of 7/38 (19%; refs. 28, 46, 48, 54, 56, 59, 65) studies focused on pancreatic cancer. Five (13%; 5/38; refs. 30, 38, 39, 41, 61) examined colorectal cancer. Four (11%; 4/38; refs. 27, 44, 45, 53) focused on hematopoietic cancers. There were 2/38 (5%) studies for each of the following cancers: melanoma or squamous cell (40, 58), biliary tract/liver/gallbladder (60, 66), renal (42, 49), and esophageal and gastric (52, 63); and one (3%) study for each of the following cancers: endometrial (32), prostrate (37), breast (43), and thyroid (67).

Nearly half of the studies (47%; k = 18/38) measured cancer outcomes through data record, including cancer registry data, the National Death Index, and other records. About one-third (29%; k = 11/38; refs. 30, 37, 40, 43, 48, 50, 52, 58, 60, 61, 63) of the studies assessed cancer outcomes through medical records. One study used self-reported data (3%, k = 1/38; ref. 57), whereas 21% (k = 8/38; refs. 23, 27, 38, 39, 51, 53, 54, 59) used other measurements.

Of 38 studies, 37 reported multivariate analyses. Associations between SB and cancer by different conceptualizations of SB are reported in Table 2. Across the 37 studies reporting multivariate analyses, regardless of how SB or cancer was conceptualized (all cancers), SB's relationship with worse cancer outcomes was reported in 43% (k = 16/37; refs. 23, 27, 28, 30, 37, 40, 43, 47, 51, 54, 55, 60, 62, 66, 68, 69) of studies. In contrast, 30% (k = 11/37; refs. 41, 42, 46, 48, 49, 52, 53, 56, 58, 64, 67) reported no significant relationships between SB and cancer outcomes. Mixed relationships between SB and cancer were reported in 19% (k = 7/37; refs. 32, 38, 39, 45, 50, 61, 65) of studies, and SB was related to better cancer outcomes in 8% (3/37; refs. 44, 57, 63).

A total of 28 studies examined SB's relationship with adiposity-related cancers. Twenty-seven studies used multivariate analyses, whereas one used bivariate analyses. Across all 27 studies that used multivariate analyses, 44% (k = 12/27; refs. 23, 27, 28, 30, 43, 51, 54, 55, 62, 66, 68, 69) reported SB's relationship with poorer cancer outcomes, whereas one study (4%; ref. 63) reported SB's relationship with beneficial cancer outcomes. A total of 22% (k = 6/27; refs. 32, 38, 39, 50, 61, 65) reported mixed relationships, and 30% (k = 8/27; refs. 41, 42, 46, 48, 49, 52, 56, 67) reported no significant relationships.

Seven studies analyzed SB's relationship with pancreatic cancer. Six studies reported multivariate analyses, and one reported bivariate analyses (59). Among studies reporting multivariate analyses (28, 46, 48, 54, 56, 65), three (k = 3/6; 50%; refs. 46, 48, 56) reported no significant relationships between SB and pancreatic cancer, whereas two (k = 2/6; 33%; refs. 28, 54) reported SB's relationships with poorer outcomes. One study (k = 1/6; 20%; ref. 65) reported mixed relationships. Five studies examined SB's relationship with colorectal cancer, with three (k = 3/5; 60%; refs. 38, 39, 61) reporting mixed relationships, one (k = 1/5; 20%; ref. 30) reporting SB's relationship with poorer cancer outcomes, and one (k = 1/5; 20%; ref. 41) reporting no significant relationships.

A total of 17 studies analyzed SB as CD, and multivariate analyses were reported in 16 (27, 28, 45, 46, 48, 49, 52–56, 60, 62–64, 69). Across these 16 studies, regardless of how cancer was conceptualized (all cancers), 50% (k = 8/16; refs. 27, 28, 54, 55, 60, 62, 69) reported SB's associations with poorer cancer outcomes. A total of 44% (k = 7/16; refs. 45, 46, 48, 49, 52, 53, 56, 64) reported no significant relationships, whereas 6% (k = 1/16; ref. 63) reported SB's association with better cancer outcomes.

Twelve studies analyzed CD's association with adiposity-related cancers. Among these 12, a total of 58% (k = 7/12; refs. 27, 28, 54, 55, 62, 69) reported CD's relationship with poorer adiposity-related cancer outcomes. One study (8%; ref. 63) reporting CD's relationship with better adiposity-related cancer outcomes. A total of 33% (k = 4/12; refs. 46, 48, 49, 52, 56) reported no significant relationships.

Five studies examined CD's association with pancreatic cancer. Two (k = 2/5; 40%; refs. 28, 54) reported significant relationships between CD and poorer cancer outcomes and three (k = 3/5; 60%; refs. 46, 48, 56) reported no significant relationships. There were no studies that analyzed colorectal cancer outcomes where SB was conceptualized as CD's.

Nineteen studies analyzed SB as 100% fruit juice. In multivariate analyses, 53% (k = 10/19; refs. 28, 37, 39, 42, 49, 56, 58, 60, 64, 67) reported no significant relationships with cancer (all cancers). A total of 21% (k = 4/19; refs. 23, 40, 50, 66) reported SB's relationship with poorer cancer outcomes, and 21% (k = 4/19; refs. 38, 44, 45, 65) reported SB's relationship with better cancer outcomes. One study (5%; ref. 55) reported mixed relationships.

Twelve studies reported relationships between 100% fruit juice and adiposity-related cancers. Among these studies, 50% (6/12; refs. 28, 39, 42, 49, 56, 67) reported no significant relationships, where 25% (3/12; refs. 23, 50, 66) reported poorer cancer outcomes. A total of 17% (k = 2/12; refs. 38, 65) reported 100% fruit juice's relationship with beneficial cancer outcomes, whereas 8% (1/12; ref. 55) reported mixed relationships.

Three studies examined associations between 100% fruit juice and pancreatic cancer, reporting no significant relationships in two studies (k = 2/3; 67%; refs. 28, 56) and significant relationships with better cancer outcomes in one study (k = 1/3; 33%; ref. 65). Two studies examined the relationship between 100% fruit juice and colorectal cancer. One study (k = 1/2; 50%; ref. 38) reported beneficial cancer outcomes, where one (k = 1/2; 50%; ref. 39) reported no significant relationships.

Two studies (32, 48) analyzed cancer's (all cancers; cancers were all adiposity-related cancers) relationship with SB, where SB was conceptualized as sugar-added fruit, sports, and energy drinks. Both studies reported no significant relationships.

One study examined the relationship between sugar-added fruit, sports, and energy drinks and pancreatic cancer and reported no significant relationships (k = 1/1; 100%; ref. 48). There were no studies that examined colorectal cancer outcomes where SB was conceptualized as sugar-added fruit, sports, and energy drinks.

Seventeen studies analyzed associations between cancer (all cancers) and SB as CD plus sugar-added fruit, sports, and energy drinks. A total of 56% (k = 9/17; refs. 23, 30, 32, 37, 39, 43, 47, 51, 68) reported SB's relationship with poorer cancer outcomes, and 35% (k = 6/17; refs. 41, 46, 48, 50, 65, 66) reporting no significant relationships. One study (6%; k = 1/17; ref. 57) reported better cancer outcomes, and one (6%; k = 1/17; ref. 38) reported mixed relationships.

Fourteen studies analyzed relationships between SB and adiposity-related cancers. A total of 50% (k = 7/14; refs. 23, 30, 32, 39, 43, 51, 68) reported SB's relationship with poorer cancer outcomes; fewer studies (43%; k = 6/14; refs. 41, 46, 48, 50, 65, 66) reported nonsignificant relationships. One (7%; ref. 38) reported mixed relationships.

SB was not significantly associated with pancreatic cancer in the three studies (k = 3/3; 100%; refs. 46, 48, 65) that examined these relationships. Four studies examined SB and colorectal cancer where SB was conceptualized as CD combined with sugar-added fruit, sports, and energy drinks. Two studies (k = 2/4; 50%; refs. 30, 39) reported SB's association with negative cancer outcomes, whereas one study (k = 1/4; 25%; ref. 38) reported mixed relationships and one study (k = 1/4; 25%; ref. 41) reported no significant relationships.

Three studies analyzed the association between cancer (all cancers; cancers were all adiposity-related cancers) and SB conceptualized as CD plus sugar-added fruit, sports, and energy drinks, and 100% fruit juice. Two studies reported SB's relationship with poorer cancer outcomes (23, 32), whereas one reported no significant relationships (50).

There were no studies that examined pancreatic or colorectal cancer outcomes where SB was conceptualized as a combination of CD, sugar-added fruit, sports, and energy drinks, and 100% fruit juice.

In comparison with current reviews of SB and cancer, this systematic review adds to the literature by distinguishing different SB based on plausible mechanisms impacting cancer risk. We distinguished SB by the type of sugar, including manufacturer added fructose and naturally occurring fructose, and excluded beverages without fructose or glucose (artificial sweeteners); and delineated how manufacturer added fructose, in particular, could directly and indirectly increase cancer risk.

Thus, unlike a prior review (35), we removed artificial sweeteners from our analyses of SB given the markedly different metabolic pathways artificial sweeteners have compared with nonartificial or sugar-based sweeteners. By examining 100% fruit juice separately, we also explored associations between SB with manufacturer-added fructose and cancer overall, and examined adiposity-related cancers by combining all studies with adiposity-related cancers in our analyses, which has not been done in prior reviews (34, 35).

Across all conceptualizations of SB, a greater proportion of studies examining CD reported SB's relationship with poorer cancer outcomes. Compared with other SB, CD's higher amounts of manufacturer added fructose may increase cancer risk both directly and indirectly; directly through contributing to impaired glucose tolerance, insulin resistance, and hyperinsulinemia, and indirectly through contributing to obesity through providing excess calories; CD are the most widely consumed SB (70) and thus contribute the most calories compared with other SB. When examining studies that reported SB's association with adiposity-related cancers, a greater percentage of studies reported CD's relationship with poorer cancer outcomes compared with studies including non–adiposity-related cancers.

In contrast to CD, the vast majority of studies analyzing 100% fruit juice reported no significant relationship with cancer, yet about 20% reported beneficial cancer outcomes. This suggests that the natural sugars and nutrients provided by 100% fruit juice, and lack of added fructose or HCFS, may contribute differently to cancer risk, relative to all SB sources. Given the potential different mechanistic actions of added sugars and HCFS in the development of cancer risk, teasing out the role of 100% fruit juice is an important area of future study.

Limitations of this analysis occurred in sampling, including potential coverage biases from studies not published in English and the lack of diverse sampling in U.S.-based studies. Within the United States, samples were predominately White; and outside of the United States, samples were predominately European, which limits the generalizability of this review. Potential confounders that may impact SB's relationship with cancer (i.e., overall diet) were also not controlled for across all studies.

The main themes in the current literature about this topic are evidence for significant associations between SB and cancer outcomes, with a greater proportion of studies reporting SB's relationship with poorer cancer outcomes. A greater proportion of studies that conceptualized SB as CD reported more deleterious outcomes compared with other conceptualizations. A higher proportion of studies reported significant associations between CD and adiposity-related cancer outcomes. There was a lack of significant relationships between 100% juice and cancer outcomes, and between SB and better cancer outcomes.

SB are not the same and contain different types of sugars or sweetening agents that are linked to diverse metabolic pathways. Regional differences in SB ingredients and manufacturing further complicate investigations and interpretations. Future studies that acknowledge these nutritional differences within SB are needed to understand how SB contribute to cancer risk, which would have important implications for practitioners and policy makers. Cancer's complexity also merits future studies examining SB intake by cancer type, as there are currently not an adequate number of studies to fully understand relationships between SB and certain types of cancer. Conducting studies in more diverse populations is critically important to increase external validity and shed much needed insight into those groups that bear a disproportionate burden of both SB intake and cancer.

No disclosures were reported.

K.H.-c. K. Yeary: Conceptualization, formal analysis, supervision, validation, methodology, writing–original draft, writing–review and editing. A.J. Quisenberry: Data curation, validation, writing–review and editing. M.G. Hall: Data curation, validation, writing–review and editing. H. Yu: Formal analysis, validation, writing–review and editing. W.A.E. Henry: Data curation, validation, writing–review and editing. E.M. Rodriguez: Data curation, validation, writing–review and editing. J.M. Zoellner: Conceptualization, data curation, validation, writing–original draft, writing–review and editing.

The research received no specific grant from any funding agency, commercial or not-for-profit sectors.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.