Discussion:
Mean BMI was significantly decreased (p<0.05) in the hypocaloric, high-protein diet groups as compared to the control diets groups, in contrast, no significant changes (p>0.05) were found in mean LDL-C, HDL-C, TAG, and TC levels among diet groups. This finding suggests that hypocaloric, high-protein diets have an unclear effect on blood lipid levels as compared to other non-high-protein hypocaloric diets. In agreement with our findings, previous studies have noted that blood lipid levels were only improved within high-protein and control diet groups individually however no significant changes were found in lipid levels between diet groups (Azadbakht et al. , 2013; Amini et al. , 2016). This is in contrast to the stated hypothesis as well as other studies’ findings that suggest that higher-protein diets improve blood lipids, including decreasing LDL-C, TAG, TC and increasing HDL-C levels (Hu, 2005; Layman et al. , 2003; Layman and Baum, 2004). An indication for a meta-analysis of these studies is therefore justified to uncover possible combined significant changes of blood-lipid levels between diets. The results within these studies are also speculated to be confounded by dietary composition which was limited in information, for example if a diets was richer in saturated fatty acids it would likely raise lipid levels, an opposing effect to diets richer in unsaturated fatty acids (Müller et al. , 2003). Mean BMI was significantly decreased (p<0.05) in three of four studies (Abete et al. , 2009; D.A. et al. , 2015; Johnstone et al. , 2011) between the high-protein and control diet groups (-2.28 kg/m2 vs -1.85 kg/m2respectively) which is in agreement with other studies involving hypocaloric, high-protein diets (Halton and Hu, 2004; Wycherley et al. , 2012; Leidy et al. , 2015). A plausible explanation to the increased weight loss is that hypocaloric, high-protein diets lessen decreases in REE as compared to other energy restricted diets (Babaet al. , 1999; Mikkelsen, Toubro and Astrup, 2000; Wycherleyet al. , 2012). This lessened decrease in REE in high-protein diets may be explained by an increase in mitochondrial oxidation pathways specific to dietary protein intake as theorized by Abete, Parra and Martinez, 2009. Additionally, dietary protein intake increases dietary thermogenesis (Halton and Hu, 2004) and contributes to lessened REE decreases in high-protein diets (Westerterp, 2004; Tentolouriset al. , 2008). The increased weight loss in the high-protein diets may also be due to these diets being more satiating than non-high-protein diets as suggested in previous studies (Stubbs et al. , 1996). This finding agrees with the decrease in caloric intake between actual and stated caloric intake values in high-protein diet groups compared to control diet groups (Table 3). Indicating voluntarily limited intake of food in the high-protein diet groups as compared to control. Qualitative questionnaires in the context of satiety are therefore warranted to explore appetite control in these diets. On the other hand, increased gluconeogenesis as a result of the lower-carbohydrate composition (Westerterp-Plantenga et al. , 2009) in high-protein as compared to control diet groups of all studies (Table 3) may have increased weight loss through additional energy expenditure as theorized in the study by Johnstone et al . This study produced a significantly greater weight loss in the low-carbohydrate high-protein (LC-HP) diet group compared to the moderate-carbohydrate, high-protein diet group (MC-HP) (Table 3, Table 2). In disagreement to this explanation however, the high-protein low-carbohydrate diet group in the study by Petrisko et al did not produce a statistically significant mean weight loss as compared to the control diet. This may be because actual dietary carbohydrate values (19.1%) (Table 3) were higher than stated dietary carbohydrate values (10%), therefore possibly decreasing gluconeogenesis, and hence less energy expenditure. Another possibility for the increased weight loss between diet groups is due to increased mean total body water (TBW) losses in lower-carbohydrate diet groups compared to the higher-carbohydrate diet groups as theorized in the study by Johnstone et al. Even though in this study TBW loss was not significant between diets (Table 4), there was a significantly greater loss in mean free-fat mass (FFM) in the LC-HP diet group as compared to the MC-HP diet group. The increased mean FFM loss may reflect greater mobilisation of hepatic glycogen stores as a result of the low-carbohydrate content in the diet, causing an associated water loss as suggested in previous studies (Yang and Van Itallie, 1976; Bilsborough and Crowe, 2003). Therefore, weight loss may have been centred around the carbohydrate rather than the high-protein composition of the diets, hence reducing the clarity of effects of hypocaloric, high-protein diets on weight loss. Reflecting on bias in the studies, the only study that had complete control over the prescribed diets was the study by Johnstone et al. The other studies’ protocols involved either providing dietary recipes to follow or food on an outpatient basis, therefore not having as much dietary input control as in the study by Johnstone et al. Additionally three-day self-reported food logs to measure compliance were done towards the end of all studies. In the study by de Luis et al,participants had a 100% retention rate whilst being on a supposed 1193.8 daily caloric restriction for 9 months. The food log was self-reported as being more or less within dietary guidelines of the study, however the actual mean weight loss was around -8.4 to -5.0 kg after 9 months. This weight loss is relatively low considering that participants had a calculated average BMR of 1747 kcal/day (Harris-Benedict Equation), which proposes a weight loss of at least -21.3 kg if the 9 month diet plan was adhered to (-553.2 kcal deficit/day). The lesser-than-expected weight loss may indicate participants falsly self-reporting dietary intakes and not completely adhering to the prescribed diets. In most studies, participants received a fixed caloric intake based on averages in weight, height, age, and sex of all participants, hence caloric deficits were not individualized, leading to possible confounding bias in primary outcome measures. The study by Abete et al was the only study to measure BMR adjusted for physical exercise in individual participants via indirect calorimetry and provided diets with adjusted caloric requirements to individuals based on these BMR values. Other studies estimated average physical exercise via activity logs, which predisposes the same biases as with food logs. Therefore in order to accurately estimate individual participants BMR as adjusted for physical exercise, dietary studies should calculate BMR via indirect calorimetry as recommend by other studies (Abete, Parra and Martinez, 2009; Lam and Ravussin, 2017). Providing individualized diets through indirect calorimetry may have also solved the confounding bias present in all studies with a wide age range of participants (Table 3), the wide age range is deemed a confounding bias as BMR decreases linearly with age (Shimokata and Kuzuya, 1993). Furthermore, all participants had voluntarily signed up to be in all these studies, which meant that these were obese adults who were motivated enough to start dieting, which may prove less external validity of studies as obese adults in the general population may not be as motivated to diet and therefore possibily have different retention rates to the obese participants in these studies. Finally, the internal validity of the studies presented in this systematic review is low as a result of non-individualized caloric intakes, high risk of biases within studies (Figure 2), different dietary compositions and other aforementioned confounders. Studies with larger sample sizes and different study designs such as longitudinal studies are warranted to increase external validity, as well as studies with a particular focus on other cardiovascular risk factors such as hypertension and asking for complete dietary nutrient composition in these studies would provide a more complete outlook on the effects that hypocaloric, high-protein diets have on cardiometabolic health.