Introduction

Osteoporosis is a systemic skeletal system disorder, characterised by diminishment of both bone architecture and bone mass, precipitating in an increase in bone fragility, risk of fracture [1, 2] and increased hospital admission and co-morbidities [3]. There is a significant social cost associated with hip fractures in Ireland, with the total in-patient cost alone being €58 million per annum for those over 65 years, which is set to increase [4]. Bone mineral density (BMD) is an indirect measurement tool for diagnosing osteoporosis, which is itself routinely measured using dual-energy x-ray absorptiometry (DXA), with osteoporosis defined as a T-score ≤ −2.5 (2 · 5 standard deviations below the young adult’s average) [5]. The fracture-risk assessment tool (FRAX) is used to judge an individual’s 10-year risk of developing a hip fracture and/or major osteoporotic fracture [6].

Weight bearing exercise, genetic factors, tobacco smoking, alcohol intake, age, hormone levels, gender, race, and nutrition all contribute to bone health [7,8,9,10,11,12,13]. Nutrients, including micronutrients—calcium, vitamin D, phosphorous and magnesium—are key components in the process of bone formation. However, macronutrients such as protein are an often neglected bone element and contributor to bone health through muscle mass and falls prevention, and the quantifying of its consumption in the population is important to physicians when considering treatment bone-related diseases [13,14,15,16,17,18,19,20]. Furthermore, dietary protein may also influence bone mineral content and density through its contribution to greater muscle contractions [21,22,23,24].

Protein accounts for approximately 30% of the mass of bone tissue and 50% of its volume [25]. The RDA of dietary protein is 0.6–0.8 g/kg per day for adult males and females, although studies have suggested that this figure could be increased to 1.0 g/kg or higher, especially for elderly populations [26, 27]. Certain mechanisms have been postulated while others have been established in relation to the association of protein with calcium absorption and urinary secretion. Protein has been observed to increase insulin-like growth factor (IGF-1) hormone production and secretion [13]. IGF-1 contributes variously to bone growth, including triggering chondrocyte proliferation and differentiation in the epiphyseal growth plate, involvement in osteoblast proliferation, differentiation and mineralization, trabecular and cortical bone formation and collagen synthesis [28,29,30]. It has also been observed that a decreased concentration of IGF-1, which can result from reduced dietary protein intakes [31], can indirectly lead to a decrease in calcium and phosphorous absorption in the intestine [32].

Dietary protein intake and its influence on the body’s calcium balance, and therefore bone growth, are complex. An increase in protein consumption has been associated with an increase in urinary calcium concentrations [33, 34]. It has been suggested that during the oxidation of sulphur-containing amino acids, protons are released, which increase renal acid excretion and acid production [35]. This translates to the oxidation process contributing to the body’s net acid load, causing increased bone dissolution, with the overall acid-base balance of the body being an important contributor to bone health [36]. The increased bone resorption of calcium was thought to act as a buffer to this increased acid load, with increased renal acid excretion of calcium as a result [37,38,39]. But more recent studies using a much larger dataset contradicted this, showing no significant difference between the biochemical indicators of bone turnover or whole body calcium retention in high protein or low protein diets [40]. This contradiction could be explained, among others, by the fact that foods which contain protein also contain hypocalciuric elements such as phosphorous [41,42,43]. Kerstetter et al. reported a strong association between dietary protein intake and intestinal calcium absorption [44] while conversely, both Surdykowski et al. and Fenton et al. concluded that increased protein intakes in the diet are not associated with significant blood pH changes and dispute the acid-base hypothesis [45, 46].

This confusion may have been due to the assumption of some studies that an increase in urinary calcium related to changes in bone turnover or bone health generally. Gaffney-Stomberg et al. reported a net gain of calcium with a high protein diet in rats, even when the resultant increase of urinary calcium was taken into account, with no changes in indicators of bone turnover [47]. In human isotopic studies, this trend has been repeated [40, 44, 48].

Advances have been made in understanding the underlying mechanism to which protein, or more accurately, the amino acids can increase calcium absorption. It is possible that calcium and protein act synergistically at the level of the cell, with one relying on the other to exist for the maintenance and growth of bone [30]. The calcium-sensing receptor (CaR), which is present in multiple organs, has been shown to be significant in maintaining calcium levels as well as contributing to skeletal growth and maintenance. Amino acids can increase the sensitivity of these receptors to calcium, therefore resulting in an increased uptake of calcium through these cells [13, 30].

There is considerable debate on whether protein quality, either vegetable or animal protein, is an important contributor to protecting and enhancing bone health. In some studies, vegetable protein was found to confer a greater benefit when compared with animal protein [49]. There is a degree of inconsistency in studies exploring the superiority of protein from vegetable rather than animal sources [13]. Furthermore, a potential disadvantage of consuming protein from vegetable sources is that vegetables may not contain all of the essential amino acids, unlike animal sources.

There is much controversy on the ideal quantity of protein intake and type among different population groups (especially older adults), as well as the most accurate and efficient means of evaluating dietary protein intake. The “one size fits all” model of adequate dietary protein intake among all age and population groups is not appropriate and may have unintended consequences for the elderly population, who require more dietary protein [27].

Therefore, our aims are to (1) review the evidence of the relationship between BMD and protein quantity in animal subjects as well as assessing this relationship among those ≥50 years of age and (2) assess the association between the concentration of protein consumed in the diet and bone fracture (hip fracture).

Materials and methods

We searched studies on various databases, including ScienceDirect, PubMed and EBSCO. The search strategy involved using specific terms: “Protein, bone, elderly, BMD” and “Protein, bone, elderly, fracture risk”. The reference list of each study and previous review articles was also examined. The search was not restricted regarding the date of the studies, with any articles meeting the inclusion criteria included in the review. The initial population was further narrowed, with any literature which included populations under 50 years of age being discounted. The majority of studies which assessed BMD utilised DXA scanning, with a small number of studies using single-photon absorptiometry or ultrasound technology [50,51,52]. One other study used dual photon absorptiometry for the baseline readings, with the 4-year follow-up using DXA [53]. Studies fit for qualification in the review process were also assessed by two reviewers with clinical and laboratory expertise. There was no journal article which posed a language restriction problem, and no authors had to be contacted for the literature review. All studies that were included in the main review had to be published, and reference was made to literature related to opinions in the discussion. A small proportion of unpublished literature was included only in the analysis of the conclusions of the studies, rather than the studies/trials themselves.

For assessing the relationship between protein quantity in the diet and BMD, we included studies which measured BMD using a DXA scan at sites such as the radius, femoral neck, total hip, lumbar spine, total hip and total body. For assessing the protein quantity and fracture risk/history association, publications which analysed hospital records were reviewed, depending on each individual study. Fractures caused by events such as road traffic accidents were discounted in many of the studies. The review of the assessment of dietary intake was carried out in the context of the studies assessing an individual’s protein consumption and BMD and fracture risk/history, respectively. We reviewed all publications on patients older than 50 years which related to protein quantity and fracture risk as well as BMD. There was a requirement for the subjects to be greater than 50 years, with studies discounted even if the average age was above 50 years but the range included those under 50 years, e.g. Sahni et al. [54]. Studies which included a maximum of two variables were included in the study, with the review of the studies taking this into account in the conclusions that were reached. This exclusion criteria and the process of selection is displayed in the flowchart below (Fig. 1 ).

Fig. 1
figure 1

Exclusion criteria and the process of selection

Full size image

To identify the volume of protein consumed in an individual’s diet, dietary assessment tools such as the Food Frequency Questionnaire (FFQ), dietary records and food diaries are used in the screening of large populations and to accurately identify those who are at risk due to their consumption patterns.

Results

BMD (Tables 1 and 2)

Of the 287 titles that we identified, 18 studies aimed to evaluate the relationship between protein quantity intake in the diet and BMD. Animal studies (Table 2) composed 4 of these, while 14 were on human subjects (Table 1). Of these studies, 13 found a positive correlation between protein and BMD, while 3 found either no relationship or possibly negative relationship. Two studies exhibited inconclusive results.

Table 1 Comparison of studies identified that evaluate the effect of protein quantity in the diet on bone mass (BMD)-human subjects
Full size table
Table 2 Comparison of studies identified that evaluate the effect of protein quantity in the diet on bone mass (BMD)-animal subjects
Full size table

Regarding the animal studies, one of these displayed no association while the remaining three showed a positive correlation. All of the population in the animal studies were growing or adult rats, either male or female. None were ovariectomized. Of the positive results, the studies focussed on the low BMD effects of a low-protein diet, along with negative effects on cancellous bone mass and trabecular thickness in some instances also found [58]. A control was used in all the animal studies, which was 15% casein for 2 of the studies and 18% and 24% protein diet for the remaining studies, representing adequate levels of protein in the diet.

Of the 14 studies on human subjects >50 years, 10 found a positive association between protein quantity in the diet and BMD. Of these, eight had protein as an independent variable in their assessment. The two remaining studies had a variation in calcium or vitamin D intake. Aside from these, the remaining 8, some of which were wide-ranging epidemiological studies, do show a strong positive correlation, independent of other factors. There was a significant variation in the number of subjects in each of the studies. Chan et al. had the highest number of participants, with 1225 men and 992 women. Geinoz et al.’s study of 74 hospitalised patients had the lowest number of participants. In the studies which portrayed a positive correlation, a significantly higher BMD was found in the femoral neck and lumbar spine of the human subjects [50, 53, 57], as well as total hip [51, 55] and total body [27, 59] of those who were not protein malnourished.

Bone fracture

A total of seven studies were identified which aimed to evaluate the relationship between protein quantity and bone fracture risk in those over 50 years (Table 3). Of these, four studies found a positive correlation between dietary protein quantity and bone fracture risk. One study found no association, with two studies with mixed results, depending on the ratio of animal to vegetable protein consumed.

Table 3 Comparison of studies identified that evaluate the association between protein quantity in the diet and bone fracture risk/history
Full size table

Wengreen et al. observed that those who consumed the highest quartile of protein (17.4–30.8% of total energy) had a 65% reduction in the risk of hip fracture. The study also reported that the association between protein consumption and its protective effects from fracture risk dissipated with increased age. The authors also observed that a higher protein intake conferred a protective effect from hip fracture in 50–69-year olds, with this protective effect disappearing in the 70–89-year age bracket [61].

Discussion

Protein quantity in the diet and BMD

The results from these studies convey the positive change in BMD with an increase in protein consumption, with 72% of the studies consistent with this finding. The majority of the studies in this review observed a positive relationship between protein consumption and BMD in older adults and the importance of the interaction between protein and calcium in bone formation and maintenance. This emphasises the need for adequate intake of all key nutrients in the prevention of bone-related diseases such as osteoporosis. However, not all studies were in agreement—2 of the 14 studies observed no association between consumption and BMD [21, 34].

For example, Kerstetter et al. in a combination of 13 younger and postmenopausal women observed no association between protein intake and femoral neck and lumbar spine BMD [44]. However, in this study, the relationship between a ratio of protein and calcium with bone was assessed, rather than protein on its own.

Importantly and as previous studies have suggested that although BMD is a good marker of bone strength, it is only one factor in the process, with bone mineral mass and architecture among other important contributors to bone strength [64]. Even with these, variance on the development of osteoporosis can occur. Therefore, assessing the role of intrinsic tissue quality may be a more accurate measure of the risk of osteoporosis developing [65]. Furthermore, the measurement and assessment of important structural proteins such as collagen and bone turnover (and the technical challenges that accompany this) in determining overall bone health should be considered for future studies.

Protein quantity in the diet and risk of fracture

The majority of studies reviewed displayed a positive relationship between protein consumption and fracture risk [59,60,61,62]. Many of the studies which displayed a positive correlation between protein quantity and fracture risk also emphasised the importance of correct calcium intake as well as protein intake in the diet [54]. However, this interdependent phenomenon between protein and calcium was not as evident among more elderly participants (aged >50 years [54]) which reinforces the hypothesis that the interaction of protein and calcium is complex in bone formation and stability [30].

The results of Wengreen et al. who investigated the different associations between protein intake and bone fracture risk between different age groups in the elderly population also requires further research. The renal tubular re-absorption and intestinal absorption of calcium are decreased among the elderly population, and as protein has been shown to potentially influence these mechanisms, the change in dietary protein would have to reflect this alteration in the elderly population [66]. This also emphasises the need for early intervention to maximise bone health before this life stage is reached, as previously stated by both Zhu et al. and Delmi et al. [56, 67].

Dargent-Molina et al. who observed a negative association between protein intake and bone health had calcium intakes which were designed to be low among the participants, while protein intakes were either moderate or high. The low calcium intake, rather than the moderate or high protein intake, could have contributed to the risk of the increased fractures. Interestingly, this study which looked at fracture risk of postmenopausal women (40–65 years) at baseline and 15 years later found that high animal protein intake actually conferred a greater fracture risk in the presence of reduced calcium intake whilst displaying a benefit in the presence of adequate calcium intake [36]. The complex interaction these two nutrients have is supported by Sahni et al. (which was not included in this review as the age range included subjects who were pre-menopausal), where there was a negative correlation between animal protein and risk of hip fracture in the presence of adequate calcium intake, as opposed to an increased risk of fracture associated with moderate to high intake of animal protein in the presence of low calcium intake in the diet [54].

As observed with BMD, the ratio of animal to vegetable protein may also play a role in the prevention of hip fractures. Sellmeyer et al. observed that elderly women with a high animal-plant ratio had a higher risk of hip fracture (relative risk = 3.7, P = 0.04) [63]. The acid-load hypothesis or a different, unknown mechanism may help explain this observed difference. However, this requires further exploration to elicit whether the issue is one of the animal protein vs plant protein or rather a question of macronutrients and micronutrients [13].

It has been observed that elderly fracture patients undergo deterioration in nutritional status during their hospital stay, with a negative association between low dietary protein intake and the risk of bone fracture and longer rehabilitation [61, 68, 69]. Furthermore, their functional outcome after bone fracture is well documented in literature [59, 60, 67, 70, 71]. Therefore, protein intake could also be a determinant of a patient’s morbidity and mortality and the length of their stay in hospital [70, 72, 73]. For example, protein-energy malnutrition has been found to be an outcome determinant for the patient, affecting immune function, among others [74, 75]. In three studies, protein energy malnutrition was identified in a significant percentage of admissions to long-term care settings, an orthopaedic surgery department and a community resident home, respectively [76,77,78].

Strengths and limitations of this review

From an initial literature review of 7941 papers, 25 were found to meet inclusion criteria. A total of 25 papers were assessed to examine the relationship between protein, BMD, and hip fracture, with reference to quantity rather than quality of protein. This review is one of the few studies that focussed predominantly on bone health in the over 50 age group. As there is a degree of variation within the FFQ format itself (with some FFQs going into more detail and using slightly different tools compared to others), the conclusions of this review had to reflect the degree of non-standardisation when deciding its effectiveness [79]. Furthermore, it was unknown if some of the studies included the impact of polypharmacy on absorption and utilisation of micronutrients which is a limitation of this paper.

Conclusions

The results of this review suggest that the quantity of protein consumed in the diet is associated with a greater BMD in the elderly. Furthermore, it was observed that protein bestowed a protective effect from the risk of fracture. These observations suggest that there is a need for a greater effort in ensuring that those >50 years meet their protein requirements as a means of bone protection [27]. Although there are weaknesses associated with the FFQ, the most common assessment tool in these studies, there are currently no equally alternative methods which have displayed greater utility in data attainment in this regard, in large studies. Ideally, the combination of biomarkers with BMD would yield the most accurate results by far, but there is limited cost and time, especially in the clinical environment. Improvements in the current FFQ model, such as creating a more detailed questionnaire, may prove to be more accurate in eliciting information [80]. As a result, the FFQ will continue to be the mainstay method of choice until further evidence suggests the need for a complete diet record of micro and macro nutrition.