The research questions were formulated in cooperation with other relevant expert groups. The effects or associations marked with * should be reviewed in cooperation with or in the relevant expert groups (e.g. infants and children, elderly, pregnant and lactating women).

What is the dietary requirement of protein and protein of different dietary sources for adequate growth, development, and maintenance of body functions, mainly based on N-balance studies?

We included studies with protein intake from foods, but excluded studies with isolated protein as supplements, and studies based on the intake of amino acids. The protein intake could be expressed as animal protein, vegetable protein and/or total protein (animal + vegetable).

The search terms were established in collaboration with a librarian and are shown in Appendix A. The databases used were PubMed and SweMed (the latter was used to identify Nordic papers not published in PubMed). The main search included the period January 2000–January 2011. An additional search was run in Medline through the PubMed platform in January 2012 in order to update the search with the most recent papers published from January to December 2011.

The 5,718 abstracts from the initial search were screened by one of the authors (ANP) in order to exclude the clearly ineligible abstracts and to select abstracts that should be directed to the other expert groups. This left 1,483 abstracts to be screened in pairs by the three members of the protein expert group. In July 2011, a member of the initial expert group resigned. The two remaining experts (ANP and EB) made the first screening. All articles suggested by at least one of the two were ordered as full-text papers. In August 2011, the expert group was supplemented with a third expert (JK) who participated in the second screening of the full-text papers. The experts made the second screening in pairs, and papers suggested by at least one expert were included in the quality assessment. The quality assessment was done according to the principles in the guidelines (7). Briefly, a quality assessment tool specific for the study type was used to grade the papers as A (high-quality study with very low level of potential bias); B (some bias, but not enough to invalidate the results); C (significant bias and weaknesses that may invalidate the results). After the quality grading, evidence tables were constructed with a description and the quality assessment of each study. Finally, for each evaluated outcome the grading of evidence was based on summary tables and a four-class grading: convincing (high), probable (moderate), suggestive (low), and no conclusion (insufficient). The minimum requirement for ‘suggestive’ was two studies showing an association and no conflicting results. If some studies showed a non-significant (neither positive nor negative) association, it was decided that for ‘suggestive evidence’ the number of results showing an association was required to be at least two times higher than those showing no association.

The studies used for the grading of evidence for protein requirements based on N-balance studies are a meta-analysis including 19 balance studies (3), a controlled metabolic study of three 18-day periods (8), and a controlled single blinded short-term study with high versus usual protein (UP) intake (9), quality graded as B, A, and B, respectively (see Appendix C, Table C1).

Rand et al.'s meta-analysis from 2003 (3) included 19 N-balance studies of eucaloric diets with at least three test protein intakes. They found no significant differences in requirements between adult age, sex, or source of dietary protein, but they also stated that the data did not provide sufficient power to detect possible differences. The median-estimated protein requirement of good-quality protein was 0.66 g per kg body weight (BW) per day, and the estimated recommended dietary allowance (RDA) was set to 0.83 g good-quality protein/kg BW per day (97.5th percentile). Campbell et al. (8) tested young versus old, and men versus women in a controlled metabolic study with a low-protein (0.5 g/kg BW), medium-protein (0.75 g/kg BW), and high-protein (HP) (1 g/kg BW) diet. The N-balance was not different between the four groups, and the estimated requirement expressed per kg BW was not significantly different for the young versus old or men versus women. Mean protein requirement was lower for older women versus older men, but when expressed per kg fat-free mass (FFM), there was no significant difference. For all subjects combined, the adequate protein allowance was estimated to be 0.85±0.21 g/kg BW per day and not statistically different from the estimate of 0.83 g/kg BW per day, as suggested by Rand et al. (3).

A short-term study was also included because of an HP intake in the test meal versus UP intake (9). Young men and women (UP: 1.04 g/kg BW and HP: 2.08 g/kg BW) versus old men and women (UP: 0.89 g/kg BW and HP: 1.79 g/kg BW) were tested for 10 days on each diet in a cross-over design. There was no age-related difference in N-balance. Nevertheless, there was concern about an HP diet corresponding to ca. 24 E% in the elderly because of a potentially negative effect on the kidney function expressed as a lack of increase in glomerular filtration rate (GFR) from a reduced value among the elderly.

In summary, the evidence is assessed as probable regarding the estimated average requirement and the subsequent RDA of 0.83 g good-quality protein/kg BW per day (Table 1). The evidence of potential adverse effects of an HP diet (ca. 24 E%) is regarded as inconclusive (Table 1).

The evaluation of the association between protein intake and mortality among healthy individuals is based on seven prospective cohort studies with nine populations included (10–16), four papers quality graded as B (10, 14–16), and three papers quality graded as C (11–13) (see Appendix C, Table C2). Such an association might be expected as a result of a possible association between protein intake and cancer or cardiovascular diseases, as described in the following sections.

Fung et al. (10) used pooled data from the Nurses’ Health Study (NHS) and Health Professionals’ Follow-up Study (HPFS) and, based on a food frequency questionnaire (FFQ), they created a low-carbohydrate (LC) score from deciles of the energy percentage (E%) of fat, protein, and carbohydrate (CH). They also made an animal LC score (animal fat and animal protein) and a vegetable LC score (vegetable fat and vegetable protein). The range of intake of total protein was ca. 15–23 E%. They found that a high overall low-carbohydrate–high protein (LCHP) score was associated with an increase in all-cause mortality, and that a high LCHP score based on animal protein and animal fat, was even more positively associated with all-cause mortality, cardiovascular disease mortality, and cancer mortality, while the LCHP score based on vegetable protein and vegetable fat was associated with lower all-cause mortality and cardiovascular disease mortality. Thus, the health effects of LC diets may depend on the sources of protein and fat. The authors also emphasized that the presented LC scores were not designed to mimic any particular versions of the LC diets in the popular literature, and therefore the risk estimates did not correspond with any versions of LC diets in the population.

In the Prevention of Renal and Vascular ENd-stage Disease (PREVEND) study (11), the focus was on mortality, cardiovascular events, and renal outcomes. The protein intake was calculated from two 24-h urinary urea excretions and expressed as protein intake in g/kg ‘ideal’ BW, i.e. after correcting BW to a body mass index (BMI) corresponding to 22. Thus, the level of protein intake could not be assessed, because the correction probably overestimated intakes, and because of no correction for possible loss of urine in the collections. They found quintiles of protein intake inversely associated with all-cause mortality and non-cardiovascular mortality.

In the Iowa Women's Health Study (12), total, animal and vegetable protein E% in quintiles from an FFQ were isoenergetically substituted for CH. The range of intake was from 14 E% in the lowest quintile to 22 E% in the highest. No association between the intake of total and animal protein, respectively, and mortality (all-cause, coronary disease mortality and cancer mortality) was observed in the multivariate models. Vegetable protein was inversely associated with coronary disease mortality, and substituting vegetable protein for animal protein was also inversely associated with coronary disease mortality.

The Swedish Women's Lifestyle and Health cohort study (13) used energy-adjusted increasing protein and decreasing CH intake in deciles, and a combination of them in an LCHP score. The range of protein intake in deciles was from 10 to 23 E%. Increasing protein intake was associated with increased risk of cardiovascular disease mortality, and the combination score was even more predictive. The combination score was also positively related to increased risk of all-cause mortality. The associations were more pronounced for cardiovascular mortality in women aged 40–49 years at baseline compared to women aged 30–39 years at baseline. The dietary assessment was based on an FFQ and the mean energy intake of ca. 6.4 MJ indicated under-reporting. The authors emphasized that the presented LCHP score did not address the potential short-term effects of LCHP diets in the control of BW or insulin resistance, but they did draw attention to the potential long-term adverse health effects of a diet generally low in CHs and high in protein, especially with respect to cardiovascular health.

Based on FFQ, The Health Professionals’ Follow-Up Study examined quintiles of protein E% (total, animal, and vegetable) substituted for an isocaloric amount of CHs and the association with fatal and non-fatal ischemic heart disease (IHD). An inverse relationship of vegetable protein to fatal IHD was found (14).

Prentice et al. (15) pooled two cohorts from the Women's Health Initiative, and from FFQ they used ‘biomarker calibrated’ protein intake in gram per day and protein E%. They found protein E% inversely related to coronary heart disease mortality, while the relation to protein intake in gram per day was non-significant.

In the Greek cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC) study, Trichopoulou et al. (16) evaluated the association between mortality and a habitual LCHP diet expressed as a score based on deciles of energy-adjusted intake. The range of intake of total protein was ca. 10–20 E%. The LCHP score was positively related to all-cause mortality, and expressed as a 2-unit increase, also positively related to cardiovascular deaths. The dietary assessment was based on an FFQ and under-reporting was particularly present among women. The authors emphasized that the data evaluated the health consequences of long-term habitual dietary intakes and should not be interpreted as indicating that short-term use of LCHP diets is detrimental to health.

For mortality, the relationship between protein intake per se and all-cause mortality is regarded as inconclusive, while the evidence is assessed as suggestive regarding an increased risk of all-cause mortality in relation to an LCHP diet with total protein intake of at least 20–23 E% in three studies, including four prospective cohorts (Table 2).

For cardiovascular mortality, the evidence is assessed as suggestive for an inverse relation to vegetable protein intake based on three studies with four prospective cohorts (Table 2).

Regarding protein intake and the sources of protein (animal versus vegetable) and the relation to cancer deaths, the evidence is assessed as inconclusive (Table 2).

The evaluation of the association between protein intake and breast cancer is based on one prospective cohort study (17) and two nested case-control studies (18, 19), quality graded as C, B, and C, respectively (see Appendix C, Table C3). A possible association with cancer could be explained by an increased production of growth factors, such as insulin-like growth factor I (19) and/or the formation and absorption of carcinogens produced during cooking or processing of meat (17).

In the Nurses’ Health Study (17), there were no statistically significant associations between breast cancer and total, animal or vegetable protein intake, based on FFQ. In a nested case-control study by Sala et al. (18), the odds ratio of having a high-risk mammographic parenchymal pattern in the highest tertile of total protein intake, based on 7-day diaries, was twice that of women in the lowest tertile, while an Italian nested case-control study (19) found no relationship between breast cancer and total, animal or vegetable protein, based on FFQ, but with no information about the total energy intake.

Thus, the evidence is assessed as inconclusive regarding the relation of protein intake to risk of breast cancer (Table 3).

The evaluation of the association between protein intake and colorectal cancer is based on one meta-analysis (20) and three case-control studies (21–23), quality graded as C, B, B, and C, respectively (see Appendix C, Table C4). A possible association between the protein intake and colorectal cancer is most commonly explained by a production of carcinogens arising either from cooking or from preservation of meat products (22).

In the meta-analysis (20), the main focus was on fat, and not protein, in relation to cancer. Some of the included studies regarding protein intake were not relevant in a Nordic setting, and only a few studies included animal protein, while most studies included foods (meat). They found no significant association between animal protein and colorectal cancer. Two of the case-control studies used colorectal adenomas (as precursors for colorectal cancer) as the outcome and used hospital controls as well as healthy controls. None of them found a statistically significant relation to total protein intake (21, 22) or animal protein intake (21). The third case-control study found no statistically significant association between colorectal cancer and total protein intake (23).

The evidence is assessed as inconclusive regarding the relation of protein intake to risk of colorectal cancer (Table 3).

The following associations between protein and cancer diseases are based on just one study (see Appendix C, Table C5) and thus regarded as inconclusive (Table 3). All of the studies used FFQ.

The evaluation of the association between protein intake and laryngeal cancer is based on an Italian case-control study with hospital controls, quality graded as C (24). They found a statistically significant increased risk for total and animal protein intake and a slightly reduced risk with increased vegetable protein intake.

The evaluation of the association between protein intake and non-Hodgkin's lymphoma is based on a US case-control study with population-based controls, quality graded as C (25). They found a statistically significant increased risk with increasing animal protein intake and a reduced risk with increased vegetable protein intake.

The evaluation of the association between protein intake and esophageal and gastric cancer is based on a US case-control study with population-based controls, quality graded as B (26). They found a statistically significant increased risk with increased intake of total and animal protein and a reduced risk with increased vegetable protein intake.

The evaluation of the association between protein intake and ovarian cancer is based on a Canadian case-control study with population-based controls, quality graded as B (27). They found no statistically significant association with total protein intake.

The evaluation of the association between protein intake and pancreatic cancer is based on an Italian case-control study with hospital controls, quality graded as C (28). They found a statistically significant increased risk related to animal protein intake, but no statistically significant associations with total and vegetable protein intake.

The evaluation of the association between protein intake and prostate cancer is based on a Canadian case-control study with population-based controls, quality graded as B (29). They found no statistically significant association with total protein intake.

The evaluation of the association between protein intake and renal cell cancer risk is based on a pooled analysis of 13 prospective cohort studies (30), quality assessed as B. They found no statistically significant associations with the E% of total, animal or vegetable protein in quintiles. The intake level of the quintiles was not reported, thus the intake cannot be assessed.

The evaluation of the association between protein intake and coronary heart disease is based on four prospective cohort studies (11, 14) (15, 31), quality graded as C, B, B, and B, respectively (see Appendix 3, Table C6). It has been suggested that the possible association between cardiovascular disease and protein intake is caused by the effect of protein intake on BP, plasma LDL, and weight maintenance (14).

In the PREVEND study (11), the focus was on mortality, cardiovascular events, and renal outcomes. As mentioned earlier, the protein intake was calculated from two 24-h urinary urea excretions and expressed as protein intake in g/kg ‘ideal’ BW, i.e. after correcting BW to a BMI corresponding to 22. Thus, the level of protein intake could not be assessed, because the correction probably overestimated intakes, and because of no correction for possible loss of urine in the collections. They reported a statistically significant (non-linear) relationship between protein intake, as a continuous variable, and cardiovascular events, based on a Cox regression analysis corrected for confounding variables. However, we could not reproduce the reported statistical significance of the association between protein intake in quintiles and cardiovascular events according to their Table 2, and we have therefore chosen not to include this association in our table and conclusion.

Based on FFQ, The Health Professionals Follow-Up Study (14) examined quintiles of protein E% (total, animal, and vegetable) substituted for an isocaloric amount of CHs and the association with fatal and non-fatal IHD. The lowest quintile of total protein was 14.6 E% and the highest quintile 22.5 E% substituted for CH. They found no association between protein E% and risk of total IHD (non-fatal and fatal), but a higher intake of total and animal protein E% was associated with increased risk of IHD in a subgroup of ‘healthy’ men (without baseline hypertension, diabetes, and hypercholesterolemia).

Based on FFQ, the Nurses’ Health Study (31) used a low-CH score (low CH and high fat and HP diet) from energy percentages, and they also separated protein and fat into animal- and vegetable-based sources. They found an inverse relation to coronary heart disease when the score was based on vegetable sources. In separate analyses of each macronutrient, no statistically significant associations were found between total, animal, or vegetable protein and coronary heart disease.

Two cohorts from the Women's Health Initiative study were pooled in the analysis by Prentice et al. (15), where they used ‘Biomarker Calibrated’ protein intake and protein E%. They found that protein E% was associated with an increased risk of coronary heart disease.

The evidence is assessed as inconclusive regarding the relationship between protein intake and risk of coronary heart disease (Table 4).

The evaluation of the association between protein intake and fatal/non-fatal strokes is based on two prospective cohort studies (32, 33), quality graded as C and B, respectively (see Appendix C, Table C7).

In the Nurses’ Health Study (32), they used quintiles of energy-adjusted total, animal, and vegetable protein in gram per day based on FFQ, and found no relation to strokes. However, in the subgroup with intraparenchymal hemorrhages, the risk was inversely associated with animal protein.

Based on FFQ, the Health Professionals’ Follow-Up Study (33) examined quintiles of protein E% (total, animal, and vegetable) substituted for an isocaloric amount of CHs and the association with fatal and non-fatal strokes. The lowest quintile of total protein was 14.6 E% and the highest quintile 22.5 E% substituted for CHs. They found no association between protein E% from any source of protein substituted for CH and risk of total strokes (non-fatal and fatal).

The evidence is assessed as inconclusive regarding the relation of protein intake to risk of stroke (Table 4).

The association between protein intake and BP is based on one feeding study (34), quality graded as B, two prospective cohort studies (35, 36), both quality graded as B, and two meta-analyses (37, 38), quality graded as C and B, respectively (see Appendix C, Table C8).

The OmniHeart randomized trial compared the effect of healthy diets with partial substitution of CHs with either protein (about half from vegetable sources) or monounsaturated fat in adults with pre-hypertension or hypertension stage 1 (34). In a subgroup analysis of the 40% Caucasians, there was a statistically non-significant inverse association between the protein diet partially substituted for CHs. In a subgroup analysis of pre-hypertensive participants, the protein diet lowered BP significantly, but the analysis was not controlled for race (only 40% Caucasians) and BW (only 21% were not overweight or obese), and thus not comparable to a healthy Nordic population.

The Spanish SUN cohort of university graduates (35) found an inverse relationship between risk of hypertension and vegetable protein intake expressed in quintiles of energy-adjusted gram per day based on a Spanish version of FFQ. The Chicago Western Electric Study (36) used the dietary history method, and also found an inverse relationship between BP change and vegetable protein intake expressed in E%, but they did not control for potassium and fiber, and thus it is difficult to separate the influence of vegetable protein per se.

Liu et al.'s meta-analysis from 2002 (37) included nine cross-sectional studies and two prospective cohort studies (one study with adults and one with children). They found an inverse association between dietary protein intake and BP in men and women, and the association was dependent on the dietary assessment method. Evidence from the longitudinal studies was limited. A recent meta-analysis from 2011 (38) of 25 randomized controlled trials (RCTs) analyzed the effect of soya protein versus control on BP. Soya protein intake ranged between 18 and 66 g/day with a median of 30 g/day. The analysis found that soya protein reduced BP, and that the difference to the control groups was more pronounced in hypertensive groups, in trials using CH as the control diet versus casein/milk in the control diet, in parallel design, and with intervention duration of at least 12 weeks.

The evidence is assessed as inconclusive regarding the relation of total and animal protein intake to BP, but it is assessed as suggestive regarding the inverse relation of vegetable protein to BP (Table 4).

The evaluation of the effect of protein on serum lipoprotein risk factors for coronary heart disease is based on two meta-analyses of RCTs with soya protein (39, 40), quality graded as B and A, respectively (see Appendix C, Table C9).

The meta-analysis from 2008 (39) of 30 RCTs covered the period 1995–2007. They concluded that ‘the inclusion of modest amounts of soya protein (ca. 25 g) into the diet of adults with normal or mild hypercholesterolemia resulted in a small, highly significant reduction in total and LDL cholesterol equivalent to ca. 6% LDL reduction’. The most recent meta-analysis of 43 RCTs also included the impact of study design (40). They found that parallel studies scored higher in study quality than cross-over studies, and that the parallel RCTs were associated with significantly greater improvements in LDL values. A sub-analysis also showed that studies with highest baseline LDL had greater reductions than studies with the lowest values. Thus, the effect may be smaller in normocholesterolemic subjects. Overall, they found that 15–30 g soya protein (1–2 servings per day) had a positive impact on LDL cholesterol.

The evidence is assessed as probable to convincing regarding the effect of soya protein on LDL cholesterol (Table 4).

The evaluation of the association between protein and bone health is based on five prospective cohort studies, two randomized controlled studies, and two systematic reviews and meta-analysis (see Appendix C, Table C10). Such an association could be explained by the effect of protein-associated acid load or effect of protein intake on calcium retention and/or increases in insulin-like growth factor I (47, 49).

Three cohort studies (41–43), all quality graded as C, and a systematic review and meta-analysis (44), quality graded as B, were identified based on the association between protein and BMD or bone loss. The cohort studies included young women, postmenopausal women, and older men and women. In the cohort study of young women (41), the main focus was on contraceptive use in relation to BMD, while the relation to protein intake was a secondary analysis. Based on FFQ, they found no longitudinal effect of total, animal, and vegetable protein E% on changes in BMD. In the cohort of postmenopausal women (42), they found higher bone loss related to a higher animal/vegetable (A/V) protein ratio and also no relation to total protein intake after 7 years follow-up. The women had a median protein intake of 17 E%, but the intake data were weakened by a very low reported energy intake (mean ca. 5 MJ), estimated from FFQ. Among older men and women in the Framingham Osteoporosis Study (43), a lower E% of total and animal protein was associated with a higher bone loss after 4 years, and the highest quartile of total protein intake (1.2–2.8 g/kg BW) was associated with lower bone loss. They used an FFQ, and there was no information about the total energy intake. The systematic review and meta-analysis (44) concluded that the overall impression was a small benefit of protein on bone health based on cross-sectional and supplemental studies. The analysis was weakened by limited information about the quality of the dietary assessment methods.

The evidence is assessed as inconclusive regarding the relation of protein intake to bone loss (Table 5).

Three prospective cohort studies (42, 45) (46), quality graded as B, B, and C, respectively, and a systematic review and meta-analysis (44) quality graded as B, were identified based on the association between protein and risk of fractures (see Appendix C, Table C10).

Based on a validated FFQ, a French study of postmenopausal women with a habitual HP intake (45), there was no overall association between fracture risk and total protein intake. In the presence of low calcium intake (<400 mg/1,000 kcal), there was an increased risk of fractures related to energy-adjusted total and animal protein as well as gram per kg BW, while energy-adjusted vegetable protein was associated with a decreased fracture risk. In the Framingham Offspring Study of men and women (46), there was no overall association between fracture risk and total protein intake based on FFQ. Animal protein intake was associated with an increased fracture risk provided a low (<800 mg) calcium intake and a decreased risk of fractures provided a high (>800 mg) calcium intake. The Study of Osteoporotic Fractures in postmenopausal women (42) found increased risk of hip fractures related to high animal protein intake and high A/V ratio estimated from FFQ. When the model was adjusted for BMD, the relation of A/V ratio to fracture risk became non-significant. The systematic review and meta-analysis (44) found no relationship between protein intake and risk of fractures, neither in the cohort studies nor in the supplemental studies.

The evidence is assessed as inconclusive regarding the relation of protein intake to risk of fractures (Table 5).

A systematic review and meta-analysis assessed the relation of dietary acid load to bone health (47), quality graded as C because of the lack of information about dietary intake methods or intervention (see Appendix C, Table C10). The analysis did not support the hypothesis that ‘acid’ from the diet causes osteoporosis or that an ‘alkaline’ diet prevents osteoporosis. The systematic review also indicated that higher protein intake and animal protein were not detrimental to calcium retention. The ideal protein intake for bone health could not be determined.

The evidence is assessed as inconclusive regarding the relation of protein intake (acid load) to the risk of osteoporosis or calcium retention (Table 5).

Two intervention trials including postmenopausal women (48, 49), quality graded as B and A, respectively, were identified for the association between protein and calcium and bone metabolism (see Appendix C, Table C10). Harrington et al. (48) used a high-sodium–high-protein diet versus a low-sodium–UP diet in a randomized cross-over trial. Thus, it was difficult to separate the effect of protein per se. Nevertheless, they found that a high-sodium HP diet led to increased urinary calcium loss and increased bone resorption. In a high-quality feeding trial by Hunt et al. (49), high- (20 E%) or low- (10 E%) protein intake was combined with high- (1,510 mg) and low-calcium (675 mg) intake in a randomized four interventions’ cross-over design. They found that the combination of HP and low-calcium diet increased calcium retention, and it also resulted in an increase in IGF-1, an anabolic peptide hormone stimulating bone formation.

The evidence is assessed as inconclusive regarding the relation of protein intake to an overall effect on calcium and bone metabolism at normal intakes of sodium and calcium.

The evaluation of the association between protein and energy intake is based on one prospective cohort study (50) and two intervention studies (51, 52), all quality graded as B (see Appendix C, Table C11). Such an association could be explained by a satiating effect of protein (51).

In the Amsterdam Growth and Health Longitudinal Study (50), 350 men and women were studied at the ages of 13, 32, and 36 years by the use of a dietary history method. During the 23-year follow-up, there was a decrease in energy intake of 125 kJ for men and 152 kJ for women for every increase in protein E% intake. The association between protein and energy intake was about three times stronger than the association between fat and energy intake. In this paper, the energy intake was only reported at the age of 36. Rumpler et al. (51) measured energy intake in 12 men during ad libitum food intake of two out of three treatments in two 8-week periods: drinks, based on foods providing 2.1 MJ, were included in a high-CH, high-fat, or HP diet. The HP diet included an additional 27 g protein/day. After the 8-week periods, there was no change in the energy intake for any of the macronutrients. In the discussion, the authors mentioned the possibility that the addition of 27 gram protein might have been insufficient to induce a change in the energy intake. In a strictly controlled intervention study of 19 weight-stable men and women (52), a weight-maintaining diet with 15 E% protein for 2 weeks was compared to a weight-stable diet with 30 E% protein in 2 weeks followed by an ad libitum diet for 12 weeks with 30 E% protein. The CH content of the diets was kept constant and thus, the fat content varied considerably, from 20 to 35 E%. The energy intake was unchanged during the weight-maintaining and weight-stable periods, but decreased during the 12-week ad libitum HP diet resulting in a significant weight loss of 4.9±0.5 kg. Since they tested an HP-low-fat diet, it was difficult to separate the effect of protein per se.

The evidence is assessed as inconclusive regarding the relation of protein to change in energy intake (Table 6).

Overall, the evaluation of the association between protein and BW and body composition is based on seven prospective cohort studies (50, 53–58), quality graded as C, C, B, B, B, B, and C, and two intervention studies (52, 59), quality graded as C and B, respectively (see Appendix C, Table C12). Such an association could be explained by a protein-induced increased thermogenesis and satiety (56).

Regarding the evidence of a relationship between protein intake and changes in BMI, two of the cohort studies and one controlled trial were used. In college students (53), a very simple frequency question about protein consumption per day was not associated with a 1-year change in BMI. Based on the dietary history method, the Chicago Western Electric Study (54) found that quartiles of animal E% protein intake was positively related to risk of overweight (BMI ≥ 25) and obesity (BMI ≥ 30), while vegetable protein was inversely related to obesity among men aged 40–55 years at baseline. In an RCT, 15 physically active men were prescribed an HP diet: 1.9 g/kg BW (22 E%) versus a normal diet (NP): <1.3 g/kg BW (15 E%) in 6 months (59). The main focus and power calculation were on vascular reactivity, but they also measured BMI and body composition and found no statistically significant effect on association with BMI despite a significant decrease in BW of 2 kg in HP (3.5% of baseline BW) versus 0.7 kg in the NP group (1% of baseline BW).

The evidence is assessed as inconclusive regarding the relation of protein intake to change in BMI (Table 7).

Regarding the evidence of a relationship between protein intake and changes in BW, three prospective cohort studies and two controlled trials are used. Halkjaer et al. (56) used country-specific FFQs from 89,432 men and women from six EPIC cohorts that were also included in the Diogenes project. After 6.5 years, they found weight gain to be significantly positively associated with total and animal protein intake. In the Danish Glostrup Population Studies and MONICA1 (57), the focus was on the energy density and fiber in relation to 5-year BW changes in adult men and women, but they also found a statistically non-significant positive association with the protein E%, assessed via weighed 7-day food records. Among US women consisting of 51% Caucasians, Sammel et al. (58) used a very simple FFQ with protein intake as ‘servings per day’. They found no statistically significant association with 4-year BW gain of ≥5 kg. In Ferrara et al.'s small RCT (59), 15 physically active men were prescribed an HP diet: 1.9 g/kg BW (22 E%) versus a normal diet (NP): <1.3 g /kg BW (15 E%). After 6 months, they found a significant decrease in BW of 2 kg in HP (3.5% of baseline BW) versus 0.7 kg in the NP group (1% of baseline BW). In another small but strictly controlled intervention study in 19 weight-stable men and women (52), the participants were on a weight-maintaining diet (15 E% protein, 35 E% fat, 50 E% CH) for 2 weeks, an isocaloric diet (30 E% protein, 20 E% fat, 50 E% CH) for 2 weeks and then an ad libitum diet (30 E% protein, 20 E% fat, 50 E% CH) for 12 weeks. During the ad libitum diet, the BW loss was 4.9±0.5 kg.

The evidence is assessed as inconclusive regarding the relation of protein intake to change in BW (Table 7).

Regarding the evaluation of the evidence of an association between protein intake and changes in waist circumference (WC), two prospective cohort studies are included. In the Danish Diet, Cancer and Health Study (55), where they used a validated FFQ designed for the study, the 5-year change in WC was inversely associated with E% of total and animal protein intake, while the Diogenes project (56) found no association between protein E% and a 6.5-year change in WC, based on country-specific FFQs.

The evidence is assessed as inconclusive regarding the relation of protein intake to change in WC (Table 7).

One RCT included body composition as the outcome. In Ferrara et al.'s small RCT (59) where 15 physically active men were prescribed an HP diet: 1.9 g/kg BW per day (22 E%) versus a normal diet (NP): 1.3 g/kg BW per day (15 E%) for 6 months, no association between protein intake and change in fat mass or FFM was found.

The evidence is assessed as inconclusive regarding the relation of protein intake to change in body composition (Table 7).

The evaluation of the association between protein and renal function based on GFR is based on two prospective cohort studies (11, 60), both quality graded as C, and two short-term intervention studies (9, 61), both quality graded as B. The short-term studies were included because of the HP content in the intervention diet (see Appendix C, Table C13). An association between protein intake and renal function may be explained by the increase in GFR observed after an increase in protein intake, which in the long-term may lead to increased glomerular pressure (62).

In a 7-year prospective cohort study among healthy individuals, Halbesma et al. (11) found no association between protein intake and decline in GFR, as estimated from plasma creatinine (eGFR). The cohort was separated into two groups according to 24-h urinary albumin ≥ or<10 mg/day. In the Nurses’ Health Study with 11-year follow-up among healthy women, Knight et al. (60) also found no association between protein intake and decline in eGFR in participants with a normal eGFR at baseline. Among women with mild kidney insufficiency at baseline, the decline in GFR was related to protein intake, significant also for non-dairy protein intake. The protein intake was based on an FFQ, and there was no information about the total energy intake. The cross-over study by Frank et al. (61) showed an increase in GFR with HP intake among young healthy male participants. This was confirmed by Walrand et al. (9) who studied both sexes of young participants. In contrast, Walrand found no increase in GFR among healthy elderly of both sexes.

The evidence is assessed as inconclusive regarding the relation of protein to renal function based on GFR (Table 8).

The evaluation of the association between protein and renal function based on microalbuminuria is investigated in the same studies as GFR, namely two prospective cohort studies (11, 60), both quality graded as C, and one intervention study (61), quality graded as B, but also in a short-term study by Jakobsen et al. (63), quality graded as A (see Appendix C, Table C13).

The two cohort studies (11, 60) found no association between protein intake and urinary albumin excretion. The experimental cross-over study (61) among young healthy male volunteers showed that 7 days of HP intake (2.4 g/kg BW per day) considerably increased urinary albumin excretion (from 9 to 18 mg/day), as compared to a control protein intake of 1.2 g/kg BW per day. There were no changes in renal blood flow, renal vascular resistance, BP, or plasma levels of renin, aldosterone, or angiotensin II. The 3-week study of a similar increase in protein intake in young males (63) found no increase in urinary albumin excretion.

The evidence is assessed as inconclusive regarding the relation of protein to renal function based on microalbuminuria (Table 8).

The evaluation of the association between protein and risk of kidney stones is based on two 8- or 10-year prospective cohort studies (64, 65), both quality graded as C (see Appendix C, Table C14). Overall, there was no association between protein intake and kidney stone formation. One of the studies (65) found a higher risk with increased animal protein intake among men with a BMI<25, but no explanation could be offered for this observation. Thus, it cannot be entirely ruled out that an HP intake may promote kidney stone formation in normal weight men, but this suggestion is weakened by the low quality of the study.

The evidence is assessed as inconclusive regarding the relation of protein to risk of kidney stones (Table 8).

The evaluation of the association between protein intake and the onset of T2D is based on four prospective cohort studies (66–69), all quality graded as B (see Appendix C, Table C15). An association may be explained by the effect of amino acids on insulin sensitivity (68).

In the Health Professionals’ Follow-up Study of middle-aged men followed for 20 years (66), a LC-high total protein and fat score based on E% was associated with an increased risk of T2D, and the risk was even higher when the score was based on animal sources, mainly red and processed meat. The lowest quintile of total protein intake was 15.7 E% and the highest quintile 21.5 E%. The Nurses’ Health Study of middle-aged women followed for 20 years (67) found no association between an LC-high total protein and fat score and risk of TD2, except for a decreased risk when the score was based on vegetable sources. The lowest quintile of total protein intake was 14.7 E% and the highest quintile was 18.4 E%. The EPIC-Potsdam Study (68) in men and women found a decreased risk of T2D for each isoenergetic 5 E% higher contribution by CHs at the expense of protein, i.e. an increased risk related to an LCHP diet. Also, the EPIC-NL study among men and women (69) found an increased risk of T2D with increasing protein intake per 10 gram and with an isoenergetic substitution of 5 E% protein with CHs, resulting in an LCHP diet, and also when the protein intake was based on animal sources.

All of the diet results were obtained by FFQs.

Only one small study (70), quality graded as C, addressed the association between protein intake and blood glucose (see Appendix C, Table C15) and thus no conclusion can be drawn.

The evidence is assessed as suggestive regarding the relation of total and animal protein intake to increased risk of T2D, based on long-term LCHP diets, including one study with an LCHP-high-fat diet, while the evidence is assessed as inconclusive regarding the relation of total protein to fasting blood glucose (Table 9).

The evaluation of the impact of physical training on protein requirement is based on three clinical trials (71–73), quality graded as C, B, and B, respectively (see Appendix C, Table C16). Increased protein use for building and repair of muscle tissue in periods of strength training, together with an increase in protein oxidation during endurance training, have been suggested as potential mechanisms underlying an association between training status and protein requirement.

The effect of aerobic exercise training on whole-body protein turnover during a set level of protein intake was tested in a study of seven young men and women (pooled) using stable isotope methodology (71). Protein intake was adjusted to 0.88 g/kg BW per day during a 2-week adaptation to the study diet. Thereafter, the subjects participated in 4 weeks of endurance training (walking and running 4–5 times per week at 85% of maximal heart rate), while following the study diet. The data indicated improved protein utilization in response to the exercise training; improved N-balance, decreased protein oxidation, and a tendency toward an improvement in non-oxidative leucine deposition (measurement of whole-body protein synthesis). The study may be underpowered for the rate of appearance of leucine (measurement of whole-body protein breakdown). No non-exercise control group was included, and only one level of protein intake was studied. For N-balance, no measurement was carried out on the completeness of urine collection.

A longer training study was performed by Hartman et al. (72) who studied the response to 12 weeks of resistance exercise training (whole-body split routine five times/week) in eight young men. Whole-body nitrogen flux (Q), protein synthesis (PS), protein breakdown (PB), and net protein balance (NPB=PS-PB) was measured by a stable isotope tracer of glycine before and after the exercise program during a 5-day period with controlled macronutrient intake (1.2 g protein/kg BW/day). Reductions were found in both PS and PB after the training program, whereas the net balance between synthesis and breakdown improved, suggesting that dietary requirements for protein in resistance trained formerly novice athletes, are not higher, but rather lower after resistance training.

Thalacker-Mercer et al. (73) analyzed 4-day dietary records on 60 participants previously clustered (K-means cluster analysis) as non-, modest-, and extreme-responders to 16 weeks of high-intensity resistance training (3-day/week), based on the magnitudes of change in m. vastus lateralis myofiber cross-sectional area. Despite marked variations in responses in the different groups, no differences were found among clusters in daily intake of protein or other macronutrients. The authors concluded that the observed protein intakes (ca. 1.1 g/kg BW/day in modest and extreme) were sufficient to facilitate modest and extreme muscle growth during resistance training. There may have been under-reporting of energy intake in all clusters, based on the reported intake relative to BW.

The evidence is assessed as suggestive for the effect of training on whole-body protein retention, but inconclusive regarding the effect of physical training on protein requirements (Table 10).