Type 2 diabetes (T2D) is a chronic condition characterized by elevated glucose levels, or hyperglycaemia (1). If left untreated, chronic hyperglycaemia can lead to an increased risk of cardiometabolic complications and early death (1). In the past 30 years, the prevalence of T2D has increased from approximately 108 million to 422 million adults worldwide (2). The projections estimate that by 2045, 700 million individuals will develop diabetes (3). Therefore, understanding how T2D progresses is critical for developing preventative methods to lower its incidence worldwide.

An unhealthy diet is considered one of the key risk factors for the onset of T2D (4). Whilst the benefits of managing some dietary risk factors, such as carbohydrate intake, have been well researched (5) the role of dietary fats is less clear and this is likely due to the varying sources of fat in our diet (6). Higher polyunsaturated fatty acids (PUFA) intake has been linked to lower fasting blood glucose levels (7) and reduced T2D incidence (8, 9), whilst higher saturated fatty acid (SFA) intake has been linked to the development of insulin resistance (10). However, SFA intake may also differentially affect the risk of developing T2D, depending on its food source; SFA from meat may increase risk (11, 12) whereas SFA from dairy may decrease risk of T2D (13). Moreover, some studies also report no association between dietary fat type and risk of T2D (6, 14). The majority of studies to date have focused on a single nutrient approach (10, 14–16). However, as foods and nutrients are not eaten in isolation, understanding of the role of dietary fat within the context a dietary pattern is warranted.

There are three main delineations of dietary pattern methodologies: a priori, empirical and a posteriori (16, 17). The a priori, or hypothesis-oriented approach, derives dietary patterns using a pre-defined criterion (16) whilst empirical methods use a data-driven approach to create dietary patterns according to the combinations of foods consumed in a specific population (16). Extensive research has been conducted to investigate the association between a priori and empirical dietary patterns and T2D (18, 19). However, the use of a posteriori dietary pattern methods to understand T2D is only beginning to be explored.

A posteriori methods, such as reduced rank regression, combine both hypothesis-oriented and data driven approaches to create patterns of food intake according to pre-defined response variables (16, 20, 21). The response variables selected for use in reduced rank regression are known to be on the causal pathway between diet and the health outcome of interest, which can include biomarkers or nutrient intake. While increasing research has used reduced rank regression to investigate the association between diet and T2D using biomarkers as response variables (22–25) few studies have used nutrient intakes as response variables (16, 26–28). An understanding of how dietary fats, within the context of overall dietary patterns, impact on T2D risk is important for advancing T2D research beyond a single nutrient-focus. Therefore, the aim of this scoping review was to systematically search and synthesize the literature regarding the association between dietary fats and risk of T2D while using reduced rank regression.

The initial search identified a total of 270 records. After removal of duplicates, titles and abstracts of 207 records were screened and 188 were excluded for not meeting our pre-defined inclusion criteria. Full texts for 19 reports were screened. Of these, 11 articles were excluded: n = 5 exposure didn't match inclusion criteria (reduced rank regression response variable didn't include dietary fat), n = 4 did not have a full-text available, n = 1 was a duplicate of another study and n = 1 had an ineligible study design (meta-analysis). In total, eight studies were deemed eligible and were included in the present review (Figure 1).

This is the first scoping review to synthesize information on the associations between dietary fats and risk of T2D in an a posteriori dietary patterns context. The main findings are that of the eight articles included, five high-fat dietary patterns were identified that were positively associated with either T2D incidence or elevated fasting glucose, insulin and/or HOMA levels. All five dietary patterns were characterized by other nutrients in addition to dietary fats, such as low dietary fiber and high dietary energy density, characterized by low intake of fruit and vegetables, reduced fat dairy products and higher intake of processed meats and butter. The only dietary pattern that used exclusively dietary fat as the response variables (namely PUFA, MUFA and SFA) did not find any associations with T2D risk. This suggests that intake of unhealthy fats is often accompanied by lower intake of fruits, vegetables and other fiber-rich foods, which may need to be considered when deriving a posteriori dietary patterns for assessing T2D risk. Therefore, consumption of healthy dietary fats for the prevention of T2D should be encouraged as part of a healthful overall dietary pattern.

Although all studies in this review used dietary fat as one of the response variables, the way these were used varied between studies. Regardless of whether total fat, or individual dietary fats were used, most dietary patterns generated included high factor loadings for meat and dairy (16, 26, 28, 30, 33). This is consistent with literature on the main food sources of saturated fat, where high processed meats intake has been consistently associated with higher risk of T2D and cardiovascular diseases (34, 35) whereas findings for dairy are mixed, and often depend on whether full fat or low fat dairy products are considered separately (36). For the two studies that included saturated and unsaturated fats as response variables, high factor loadings for nuts and seeds were also observed. However, findings from these studies were conflicting; Shin et al. (33) reported that a dietary pattern high in nuts, seeds, fat and soybean was associated with higher odds of developing GDM, whilst Brayner et al. (28) reported that a dietary pattern high in nuts, seeds and butter was not associated with T2D incidence. This difference may be partially attributable to the sample populations, since Shin et al. (33) examined GDM in pregnant women and Brayner et al. (28) examined T2D incidence in healthy adults. Taken together, these findings suggest that the underlying foods that contribute to a high-fat dietary pattern should be taken into consideration to ensure these foods are lower in SFA.

Outcomes from this review align with literature on the effect of dietary fats on fasting serum glucose and insulin levels. The study by Appannah et al. (30) reported an association between a dietary pattern characterized by high intakes of high-SFA foods, such as processed meat, chocolate, savory snacks and fried foods, and higher fasting serum glucose and insulin, which could indicate insulin resistance. This supports evidence from randomized controlled trials (RCT), which have shown that high-SFA diets can decrease insulin sensitivity (37, 38). For example, a 12-week RCT of 486 European adults investigated two high-fat (either high in SFA or high in MUFA) and two low-fat (with or without 1.2 g/day n-3 PUFA supplementation) isocaloric diets. In healthy individuals, the high MUFA diet increased participants' insulin response to glucose, whilst the high SFA diet reduced insulin response (37). Similarly, a meta-analysis of RCTs reported that replacing 5% of energy from SFA with PUFA lead to significant decreases in glucose, HbA1c and HOMA levels. Although the mechanisms behind high-SFA diets and insulin resistance are still unclear, there is evidence to suggest that it might be linked to ceramide production (39). High-SFA diets increase ceramide production, which in turn can impair insulin signaling (39). High-SFA diets have also been suggested to be more inflammatory and chronic low-grade inflammation can inhibit insulin action (40, 41). Although RCTs can provide insight into the causal relationship between dietary fats and glucose or insulin, prospective cohort studies can better understand these associations in a dietary pattern context. Moreover, as fasting glucose and insulin are important biomarkers in T2D development, more prospective studies investigating their long-term associations are needed to better elucidate the associations between dietary fat and T2D incidence in an a posteriori dietary pattern context.

Consistent with this review, evidence suggests that intake of specific nutrients, in addition to dietary fats, may play an important role in the development of T2D (42). Higher dietary fiber, found in fruits and vegetables, for instance, has been shown to modulate blood glucose concentration following a meal (43) and has thus been linked to a lower risk of T2D (5). Conversely, higher intake of carbohydrate, especially refined carbohydrate has a strong positive association with T2D risk (5) High dietary sodium is also linked to insulin resistance (44). Dietary patterns such as Dietary Approaches to Stop Hypertension (DASH) and the Mediterranean diet have dietary fat as a key focus, but are also a rich source of other key nutrients, such as fiber and antioxidants (19). A systematic review and meta-analysis of 48 studies has shown that a higher adherence to either DASH (RR: 0.81; 95% CI: 0.72, 0.92) or Mediterranean diet (RR: 0.87; 95% CI: 0.82, 0.93) was associated with substantial reduction in T2D incidence. Both dietary patterns encourage a high consumption of fiber-rich foods such as fruits and vegetables, along with lower intakes of SFA rich foods such as red meat and processed foods (45, 46). Therefore, the combination of dietary fats with other dietary components may be a stronger predictor for T2D, than any nutrient in isolation. However, as only one study was identified that derived dietary patterns based on dietary fat alone, further prospective evidence is needed to confirm this.

The present review provides supporting evidence for the role of energy-dense dietary patterns as a risk factor for T2D development (47). Half of the studies included in this review included energy density as a response variable (27, 30–32) which has an established positive association with obesity risk (48) whereas associations with T2D are less clear (49). Energy dense foods are also often high in SFA and added sugars, low in fiber and have a high glycaemic load, all of which have been linked to a higher risk of T2D (50). Interestingly, the study by Brayner et al. (28) in which only dietary fats were used to derive dietary patterns, there was no association with T2D. However, there was an association between higher SFA-rich foods and lower PUFA-rich foods with increased odds of developing overall and abdominal obesity. Therefore, although there are no clear associations between dietary fats and T2D, it may be that obesity is on the causal pathway as a key risk factor for T2D development (51). As many as 85% of individuals with T2D have either overweight or obesity (52). Further, obesity can contribute to low-grade chronic inflammation (53) whereas PUFA intake can counteract pro-inflammatory pathways (54).

This review has identified some discrepancies in the design and reporting of reduced rank regression methods that limit the interpretability of results. As the number of dietary patterns generated depended on the number of response variables, studies included in this review ranged from three to four response variables (16). This created challenges when the number of patterns reported differed, as the rationale for which dietary patterns to report also varied. Often, the dietary patterns that explained the most variation in the response variables were reported, regardless of how much was explained by the other dietary patterns. However, this cut off ranged from >10% to 20% (26, 28, 30, 31) of explained variation, with some studies not providing a numerical cut off (16, 27, 32, 33), The number of food groups also varied, ranging from 13 to 50. Evidence from principal component analysis, another data-driven dietary pattern method, has shown that changes in the number of food groups can influence the associations with health outcomes (16). Thus, more consistent definition of this reporting of minimum cut off points and the rationale for classifying food groups would improve the reporting of this dietary pattern method in the literature.

This review acknowledges some strengths and limitations. Firstly, as a scoping review, this study followed systematic searching methods based on PRIMSA-ScR guidelines. Secondly, it identified reporting discrepancies in the use of reduced rank regression that should be addressed. In line with scoping review methodology, no critical appraisal of the evidence has been conducted, and therefore the quality of studies has not been considered. However, most studies were large prospective studies that adjusted for demographic characteristics, dietary intake and lifestyle risk factors.

In conclusion, this scoping review has identified eight studies using reduced rank regression to derive dietary patterns based on dietary fat. Overall, findings suggest that high-fat dietary patterns, especially high-SFA containing foods, were positively associated with T2D incidence and glucose, insulin and HOMA levels. While dietary fat may be an important predictor of T2D risk, evidence from a dietary pattern perspective suggests that foods such as processed meats, low-fiber cereals and low intakes of fruit and vegetables are likely to contribute to an increased T2D risk. Therefore, consumption of healthy dietary fats for the prevention of T2D should be encouraged as part of a healthful overall dietary pattern.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

BB and LM performed the screening of articles. BB drafted the manuscript. KL reviewed first draft of the manuscript. All authors contributed to the design of the study, the interpretation and critical evaluation of the review, provided edits, and approved final submission of the manuscript.