Our results show that, in this situation, there is no positive effect for FFM gain regardless of type of WP used, its protein percentage, or level of physical activity. On the other hand, the reduction in FM is significant, occurring only for WPC and for regular physical activity practitioners. In addition, it was found that this reduction presents a trend inversely proportional to protein content, being verified only in reduced values. Except for one analysis, the heterogeneity among studies was not significant.
4.1. Methodological Approach of the Problem
Some types of protein have the potential to influence the entire protein metabolism by modulating muscle development through physical activity. Its intrinsic characteristics, such as absorption rate, amino acid profile, hormonal response, and antioxidant potential, have a direct effect on muscle gain [42
]. Thus, the daily dose ingested becomes a determining factor for hypertrophy [16
In order to reach recommendation levels, most athletes regularly use WP as a food routine component [6
], and frequently this consumption, without a correct nutritional accompaniment, can occur in imprudent doses [9
]. In our analysis, the establishment of a single standard recommendation for WP users has two direct consequences: possible occurrence of health damage as a result of overdose (a situation not reported by any of RCTs included in this review) and lack of adequacy to the individual protein demand. An orientation established only in gross weight WP values does not consider the physical characteristics of each athlete [15
]. For this reason, aiming at better adequacy, some reviews stipulate daily protein intake values based on the individual’s body composition. To maximize skeletal mass gains and protein synthesis, values of ~0.4 g·kg−1
are indicated for young adults and ~0.6 g·kg−1
for older adults per meal, considering minimum of four throughout the day in order to reach daily consumption of approximately 1.6 g·kg−1
]. Another example is the indication for those who are in caloric restriction, which ranges from ~1.8 to 2.7 g·kg−1
or even from ~2.3 to 3.1 g·kg−1
of FFM [38
]. The studies included in the meta-analysis implemented WP doses between 0.24 and 1.28 g·kg−1
Although we chose to present the dose implemented in grams per kilogram of mass of the individual in each study (aiming to show more representative data of the individual daily consumption), not all selected RCT observed in the same way the food intake of participants, especially regarding the inclusion or not of protein supplementation values in the final calculation. In addition, although it is possible to infer that all adopted restriction on the consumption of other supplements, only two studies explained this result in the text [36
]. Consequently, it became unviable to reasonably assess the adequacy of the administered WP dose compared to values recommended for total proteins in literature for this population. However, we emphasize that the food consumption analysis of each study was the same used for all randomized groups and that, according to our evaluation, the intergroup results of none of the studies showed relevant differences in order to significantly influence participants’ body composition.
In addition, the dose (in grams) and the total time of supplementary nutritional intervention (in days of consumption) were also similar to studies found in other reviews [9
]. Although it does not allow making definitive conclusions regarding protein adequacy for physical activity practitioners, this result similar to literature allows inferring the linearity among intervention protocols that have been implemented by the scientific community. However, we consider that future RCTs should standardize the form of food consumption analysis of participants regarding presenting, preferably in grams per kilogram of mass, both the supplemented ingestion values and also the total amount disregarding the established by their intervention protocols.
Considering the importance of aspects related to protein supplementation, the nutrient timing is also one of the factors that have been gaining the focus of debate in the scientific community [45
]. In our review, this was one of the domains of analysis that varied most among studies, both in terms of portioning and distribution, being one of the main reasons that made us choose the random-effects model for the meta-analysis. The consensus regarding the real efficiency of each ingestion distribution model is still far from being reached, however, some qualitative reviews have indicated that the consumption of moderate doses of protein supplements in the pre-sleep moment, combined with physical activity throughout the day, favors the increase in muscle protein synthesis [45
]. Although it is necessary to consider the inherent difficulties of conducting randomized clinical trials adopting multiple protein portioning distributions, our investigation points to the need for greater standardization of future supplementation protocols in order to obtain further conclusions about it.
According to the values found, we chose to establish cutoff points for the percentage of WP supplemented in each study: 51%–80%, 81%–95%, and 96%–100%. Although limited due to the inherent heterogeneity among the studies, the option for this fractioning occurred after verifying a worrying data about the WP production chain: disregarding the study in which it was not possible to obtain these data [31
], when comparing values identified to cutoff points available in literature (already reported in the introduction of this article), it was found that the protein content evaluation, although all RCTs have performed it, it was not taken into account with one of the criteria for the distinction between WPC, WPH, and WPI in studies included in this review.
To exclude this bias, as previously described in Section 2.1
, the protein typology was confirmed using both classification criteria through descriptive information contained in the text of each study comparing these findings to the values on the product label on the website of each manufacturer. This allowed a more broad stratified analysis of data, also focused on bioavailability, but on the other hand, introduced a limiting factor in the research. This dichotomy shows that WP classification according to its typology is more usually related to the manufacturing process than to its final protein content. The divergence in the legislation that regulates the trade of food supplements in each country is notorious. Instead of having specific legal mechanisms to ensure their regulation, most countries include food supplements, such as WP, in subsets of existing legislation for biological, herbal or food compounds [49
]. In addition to hindering inspection, this delays progress in establishing uniform standards for the proper identification and classification of these products in the global food industry [6
A system of specification of WP variants according to both criteria brings the advantage of being more adequate to the reality of consumption of this supplement by physical activity practitioners, while the use of the manufacturing criterion only, can lead to the marketing of products with levels below adequate for hypertrophy achievement. The use of legislation aimed at standardizing the marketing of WP by adopting its classification by both protein percentage and fractionation method is therefore essential. Until then, a legal precedent has been set for WP manufacturers to include their products in categories that are not consistent with recommendations in scientific literature. Consequently, new systematic reviews on the subject should also investigate this aspect in order to categorize WP products within implemented interventions.
Regarding the nutritional aspects of the methodological approach, it became essential to evaluate the choice of supplementation protocol of control groups adopted by studies included in this review. As previously mentioned, the selected RCTs adopted commercial carbohydrate placebos of two types: dextrose and maltodextrin. Since all of them were in an isocaloric condition compared to the protein group and no study reported significant differences in participants’ adherence to food intervention, it could be conclude that the placebos used were virtually identical.
In addition to aspects related to food routine, physical activity, and parameters related to trainability are also factors that play an essential role in muscle gain [50
]. As part of the criteria adopted in the screening of this review, only studies involving physically active individuals were selected. This is because the metabolic physiology of the energy expenditure is different in sedentary individuals and, in addition to being regulated by genetic factors, its modulation is gradual according to the level of adaptation to the physical exercise intensity [50
Natural mechanisms of biological compensation such as increased energy expenditure at rest by thermoregulation increased hunger in fasting and intestinal growth modulation are usually expressed in individuals in transition from sedentary to physically active condition, and also from this to physical rigor required by more exhaustive and frequent training routines [51
]. Such mechanisms hinder sudden metabolic changes during transition and adaptation stages, so that the general metabolism always tends to be focused on ensuring the maintenance of the homeostatic energy state before routine change, whether of dietary origin or muscle work [51
]. Therefore, there is need to implement a period of adaptation to training in studies that have physical activity as protocol and, although it was a methodological similarity casually produced by screening, in our review, this step was present in all selected RCTs. Given its importance for the energy metabolism regulation, further reviews should evaluate the existence of this adaptive phase, defined as part of the inclusion criteria.
Nevertheless, the quantification of the losses of RCT participants included in the meta-analysis showed that, in general, 34.67% of them occurred in this adaptive period and that of these, 31.05% were directly related to physical activity. Since all individuals were physically active before participation the study and all participants had similar previous training routines, these results point to the need to introduce evaluation criteria for this adaptive period in future clinical trials in order to investigate their effectiveness.
Since the level of physical activity has a strong influence on the adaptive and/or physiological resistance to the increase in individual energy expenditure before new stimuli [57
], we chose to perform the analysis in three categories regarding this factor: resistance and/or recreational, regular, and athletic strength exercises. This classification took into consideration aspects related of training in each stage of this classification. Factors such as number of sets and replicates per set, execution speed, exercise ordination, and interval between executions, for example, differ according to the individual’s training stage so that the adequate choice is determining to increase the adaptive capacity of athletes [50
Seeking to apply this investigation to the daily life of most physically active individuals, the implementation of resistance and/or strength exercise was established as one of the inclusion criteria adopted in our study. An investigative assessment of intervention protocols, as previously pointed out (Table 4
), reveals a positive relationship between trainability characteristics and volume and intensity, which are, in turn, the main factors related to training efficiency [50
]. Moreover, the average total duration of the training routine implemented in studies was also similar to values found in other reviews [9
]. These results, combined with the fact that there were no statistically significant differences among groups regarding adherence to the training protocol, it could be concluded that there is homogeneity for this field of analysis.
However, an essential methodological failure was observed in the practice of physical exercise: while Naclerio et al. [37
] used the usual training routine of participants as a physical activity protocol (since they were professional athletes), and two other studies did not mention any information about it [31
], none other RCT established this restriction. In addition, of the studies showing physical activities outside the protocol, only that of Cribb et al. [32
] investigated data using a daily reminder filled by participants. Although the analysis of this record did not indicate significant differences (excluding data from individuals with incorrect and/or insufficient filling), it was considered that this gap introduced substantial bias of non-differential information, which is the main limitation of this meta-analysis.
In this systematic review, only clinical trials whose randomized individuals were adults at the time of acceptance were selected. However, some studies indicate possible metabolic differences even within this cutoff point (18–60), especially after 40 years of age [59
]. However, of studies included in this review, only one covered this age group [37
] and, therefore, a more in-depth analysis of the influence of age on the effects of WP would inevitably be limited. In addition, sensitivity analysis showed that the overall result does not change with the absence of data from this and any other clinical trial.
Finally, after assessing the risk of bias, it was concluded that RCTs were of relatively low methodological quality. However, some factors lead us to make an important observation regarding this result. Perhaps as a consequence of restrictive choices in narrative selectivity or the limit of characters imposed by several publishers, almost all the authors did not report in the text information which, although indirectly perceptible in some cases, would contribute to a more positive analysis of the risk of bias. A clear example is the omission of the restriction of parallel consumption of supplements other than those provided by the intervention protocol.
Considering the amplitude of the interpretative limits of the ROB 2.0 tool [28
], and even though it is possible to infer some assumptions through the analysis of other descriptive elements, the mere deduction of the occurrence of a methodological process does not provide sufficient support in this case to reach a definitive conclusion in this regard. Thus, it is emphasized that much of this result is due to the report quality, and not necessarily to the actual occurrence of methodological failures in conducting the studies included in this review.
4.2. Effect of Whey Protein on Body Composition
An analysis of the individual results of each RCT (summarily presented in Table 3
) reveals a stable increasing trend in FFM gain both for control and supplemented groups comparing their respective final and initial moments. However, the delta value was not significant (p
> 0.05) comparing groups with each other. This result was maintained in this global effect meta-analysis, as shown in Figure 3
a. As in each study, individuals were submitted to the same training conditions and isocaloric conditions compared the opposite group, this result demonstrates that the protein isolate derived from WP was not sufficient to provide resistance exercise practitioners a more considerable increase in FFM when compared to placebos. Therefore, it could be inferred that in this situation, the positive effect on body composition is more related to the efficient practice of physical activity than to protein supplementation.
In the individual FM evaluation in each RCT (Table 3
), a persistent gain tendency is verifiable for placebo groups while the opposite was observed for protein groups. In addition, studies included in the meta-analysis reported significant statistical difference (p
< 0.05) for FM loss in WP groups compared to placebo-controlled groups. Furthermore, a visual inspection of the graph of this global meta-analysis (Figure 3
b) reveals that this tendency in FM gain in the comparative groups is evident so that all diagrams point to WP supplementation benefits.
Comparing these findings with the effects on body composition described in similar meta-analyses that evaluated WP supplementation allied to resistance exercise, wide divergence of results was observed.
Schoenfeld et al. [61
], in a sample of 525 individuals in 23 studies, analyzed the influence of the daily protein intake distribution, with or without supplementation, comparing to controls with or without protein in their composition. There was no categorization for type of WP, and the caloric equivalence did not constitute the eligibility criteria. In this scenario, there were no significant differences in the global analysis either for hypertrophy (p
= 0.18) or for FFM in the model reduced to covariates (p
Naclerio and Larumbe [62
], in a meta-analysis that used calorie equivalent placebos compared to the protein group, but did not stratify results by type of WP, did not find significant statistical differences for body composition (FFM or FM) among studies that supplemented WP with no associations, while significant effect was verified for combined interventions (n
= 4, g = 0.468, 95% CI 0.003–0.934). In this sample composed of 192 participants in 9 studies, according to the adopted inclusion criteria, control groups could or could not be supplemented with compounds of protein origin, except for WP.
Nissen and Sharp [63
] adopted the presence of WP intervention as inclusion criteria, which may be associated, but only with non-anabolic substances. Placebos used could or could not be caloric and, moreover, there was also no analysis for type of WP. After screening, 48 designs in 40 clinical trials were selected, whose participants could or could not be previously trained with resistance exercise. Of these studies, four had physical activity as part of the intervention protocol and, in this sample, both LM and FM were not affected by protein supplementation (IC: 0.07–0.31%, p
= 0.31 and CI: 0.87–0.51%, p
= 0.66, respectively).
Miller et al. [64
] performed a stratified verification of results by type of WP (concentrated and isolated) and found no significant differences (p
> 0.05) in the body composition of participants. Nevertheless, placebos used were not paired for calories. In their global findings, a non-significant increase in LM in the order of 0.83 kg (95% CI, −0.36–2.03) was observed in a sample of seven studies.
Morton et al. [37
] in their metanalytical review (consisting of 1863 participants in 49 studies) in which there was no calorie equalization of comparative groups as well as no analysis for type of WP, protein supplementation promoted statistically significant changes in FFM (MD: 0.30 kg (0.09, 0.52), p
= 0.007) and FM (MD: −0.41 kg (−0.70, −0.13), p
On the other hand, the most recent meta-analysis performed on this subject [65
], still in the publication process, in an aggregate of 21 RCT in ~13.1 weeks of resistance training, detected significant effect for LM (p
= 0.01) with an increase of 0.46 kg, and significant reduction of FM 0.62 kg (p
= 0.004) in individuals supplemented with WP compared with placebos. An analysis of studies that composed this specific review reveals that only 33.33% (n
= 7) implemented food protocols in order to calorically equalize branches of each intervention. Of these seven clinical trials, five were also included in our meta-analysis [31
], with the others were Taylor et al. [39
] and Volek et al. [40
], who composed only our qualitative review for reasons already reported in this article. In addition to positively strengthen the screening performed here, it also reinforces the hypothesis verified in this meta-analysis that isolated and/or concentrated WP may not have a significant effect on FFM gain in comparison to non-protein isocaloric interventions.
4.3. Limitations, Strengths, and Quality of Evidence
This systematic review and meta-analysis have several factors that limit its direct application in clinical sport practice. RCTs that met all eligibility criteria had systematic failures in the reporting of information so that they were evaluated, in general, with low methodological quality. Moreover, intervention protocols were different from each other in terms of effecting, restricting, and evaluating or not food consumption and the practice of physical activity other than those proposed in this research. Although the protein composition of WP was evaluated, in one case [31
], it was not possible to obtain information from the corresponding author. Furthermore, it was considered that the confirmation analysis of WP typology, even in pairs and blindly, is inherently exposed to an evaluator’s bias.
Regarding the study sample, two points hinder the juxtaposition of our results. All essays included in the meta-analysis were performed only in men, which inevitably restricts our findings to only a portion of the physically active population. Another factor is that, considering the country of origin of each RCT (Table 1
), it was observed that all are among the 15 best evaluated in the global human development index [66
]. Considering that a substantial portion of recruited individuals were native, this result exposes our review to a significant geosocial limitation since, in addition to genetic components and environmental aspects, factors such as family income, cultural diversity and social habits play an essential role in the diet quality and diversification and have a potential influence on the metabolic profile of the population [67
On the other hand, our review and meta-analysis also has several strengths that reinforce the power of its findings: the amplitude and magnitude of the effects of WP supplementation on physical activity, both for FFM and FM was very similar among studies and, moreover, sensitivity analysis showed that the overall result is not altered by the absence of any RCTs, which therefore demonstrates strong consistency. The methods for body composition measurement were selected only by the gold standard, were in a condition of similarity among studies, and, moreover, were blindly performed in all clinical trials. The intervention protocols showed congruence and similarity for the primary analysis domains of proven direct influence on body composition. Placebo intervention of comparative groups was virtually identical among studies; and, finally, in almost all analyses, heterogeneity among RCTs can be considered low, very low or, in some cases, potentially null with I2 values equivalent to 0%.
Considering the balance between consistency, accuracy, and amplitude of results of each RCT, the implications of identified bias, and limitations of this review, we chose to apply the GRADE method in order to enable the better understanding of results found in this metanalytic study. Given that the publication bias was not statistically investigated due to the number of studies included in the quantitative analysis, scaling of domains related to similarities of estimates, overlap of confidence intervals, results of heterogeneity tests was performed, considering the fact that this systematic search also included the search for registration protocols for clinical trials in three scientific bases and that, of the 266 results assessed by eligibility criteria (total of 640), none was finalized and not published. This allowed us inferring a low risk of publication bias for all investigated outcomes.
After schematizing the evidence profile of studies included in this review, results were compiled in the form of a synthesis of conclusions that are available in Table 5
. Based on results, the following could be concluded:
Low power of evidence, in which WP does not significantly change FFM when compared to other non-protein isocaloric interventions and that concentrated and isolated WP, even with different protein levels, are virtually identical regarding the amplitude of this effect;
Moderate power of evidence, which after adapted to resistance exercise, the non-occurrence of FFM gain regarding WP supplementation, both in its concentrated and isolated form, does not depend on the level of physical activity;
High power of evidence, in which WP supplementation is favorable to the loss of FM and this reduction is observed only for WPC;
Moderate power of evidence, in which the depletive effect of FM occurs only in WPs of low protein content (51%–80%); and this result is verified only in regular practitioners of resisting and/or strength activity;
Very low power of evidence, in which there is an inversely proportional trend between protein WP content and FM reduction.
4.4. Future Recommendations
Evaluating the scope and diversity regarding multiple factors related to the use of WP, as well as the exchange of its importance in areas that go beyond sports such as public and preventive health, further new randomized clinical trials and meta-analyses are necessary for a better understanding of the effects of WP supplementation on relevant outcomes for human health, such as body composition. In order to contribute to the advancement of knowledge in this area, some recommendations discussed in our review are highlighted below that in our analysis, may support the methodological standardization of further studies.
After screening using eligibility criteria, there was only one randomized branch of one study that used WPH as intervention protocol. Although we recognize that, among all types, WPH has the highest financial cost. Our analysis is that studies with this supplementary variant are vital to advance knowledge in this area. Although all nutritional intervention protocols showed the analysis of participants’ food intake outside the study, these data were not assessed in the same way by RCTs. Thus, if a standardized laboratory diet developed by a qualified nutrition professional is not feasible, we believe that future studies present, preferably in grams per kilogram, both supplemented intake and total quantity values, disregarding values established in their intervention protocols. Furthermore, since some studies collected data from other feedings, in only three moments and even though food registration protocols already validated for various populations are abundant in scientific literature, we strongly recommend that this evaluation should be performed at least weekly during the research.
Even after decades of studies (the oldest WP supplementation clinical trial identified dates back to 1983), there are still some gaps. In our discussion, a possible influence of the moment of supplementation ingestion on muscle gain was identified and, in this sense, we consider that RCTs should adopt multi-time distribution so that new and more robust meta-analyses can be performed.
We also identified that placebos should be more standardized. Given that the effect of WP compared to controls is already well established, we recommend that new studies, including meta-analysis, also adopt caloric equivalence of placebo compared to groups supplemented with WP.
As previously discussed (Section 4.1
), it is essential to use combined legislation to standardize the marketing of WP by adopting its classification both by the fractionation criterion in the production chain and also by its final protein content. Meanwhile, we consider it valid that upcoming reviews adopt an analysis, in pairs and blind, of the WP type supplemented in the RCTs contrasting, whenever possible, the information described in the text with the centesimal composition available in the manufacturer’s website. Besides, we also consider it prudent for the next clinical trials to perform analyses of the WP composition by independent laboratories in order to corroborate the information of the product label to be supplemented.
Regarding physical activity, we believe that next RCTs in the area should adopt more selective analysis criteria in their protocols regarding the performance of parallel physical activity. Some studies included in this review, despite implementing this restriction, allowed the practice of recreational activities. In our evaluation, this concept can often be broadly interpreted, especially considering that, at the time of execution, participants are not supervised by the researcher’s team. In an attempt to reduce this bias, we recommend the choice of two most appropriate situations according to results obtained: (a) total restriction of non-labor physical activity (more indicated but less feasible methodologically); and (b) restriction of physical activity of the same typology implemented in the study, adopting as part of the protocol in this case, the daily completion of a detailed form with information on exercise intensity and volume in order to allow the statistical evaluation of data among groups.
In our review, substantial losses of participants during the adaptation training period were found, therefore, future studies should adopt criteria to assess the effectiveness of this adaptive step in order to ensure that all randomized individuals are comparable in terms of physical capacity to perform the exercise protocol. In addition, we believe that further reviews should adopt this phase as an inclusion screening criterion.
Almost all RCTs included here were assessed with low methodological quality due to the risk of bias and, as seen in Section 4.1
, it was considered that a large part of this result was an inevitable consequence of the quality of the study reports. As a result, we strongly recommend that the authors of clinical trials should adopt less rigid criteria in the selectivity of information to be published in order to enable not only a better understanding but also a better judgment of the possible risks of bias.
In our literature searches, it was identified that of the 114 RCTs included in our final screening and the 86 systematic reviews with meta-analysis, only 20.17% and 37.21% were included in publicly accessible digital platforms. The publication of a study not previously registered, in addition to increasing the chances of the occurrence of biases and hindering the proper understanding of its execution by the reader, exposes the scientific community to the risk of redundancy, so that, without this record, the likelihood of more than one research group performing similar studies is very high. The publication of a study protocol in advance in networks such as PROSPERO for reviews and Clinical Trials for randomized studies (both global and free) may significantly contribute to reducing the use of material and human resources in science.
Moreover, the need to extend the domains of analysis to other variables related to physical activity was also identified. We believe that investigating and reporting sociodemographic information on the population recruited in the next RCTs would significantly contribute to expanding the scientific knowledge in the sports area.
Finally, we would like to emphasize the latent need for RCTs to be carried out with the female public in the sports field. Since 2011, only in the African continent, for example, the physically active population between men and women was already comparable (83.8% and 75.7%, respectively) according to criteria defined by the World Health Organization [69
]. It is a consensus that, although growing, the number of studies focused on the sports field including women is unfortunately far from achieving a minimally realistic representation of the physical exercise population. We consider that this is one of the leading and most urgent challenges to be overcome in this area by the scientific community.