Effects of whey and soy protein supplementation on inflammatory cytokines in older adults: a systematic review and meta-analysis

Background and aims: Low-grade inflammation is a mediator of muscle proteostasis. This study aimed to investigate the effects of isolated whey and soy proteins on inflammatory markers. Methods: We conducted a systematic literature search of randomised controlled trials (RCT) through MEDLINE, Web of Science, Scopus and Cochrane Library databases from inception until September 2021. To determine the effectiveness of isolated proteins on circulating levels of C-reactive protein (CRP), IL-6 and TNF-α, a meta-analysis using a random-effects model was used to calculate the pooled effects (CRD42021252603). Results: Thirty-one RCT met the inclusion criteria and were included in the systematic review and meta-analysis. A significant reduction of circulating IL-6 levels following whey protein [Mean Difference (MD): −0·79, 95 % CI: −1·15, −0·42, I2 = 96 %] and TNF-α levels following soy protein supplementation (MD: −0·16, 95 % CI: −0·26, −0·05, I2 = 68 %) was observed. The addition of soy isoflavones exerted a further decline in circulating TNF-α levels (MD: −0·20, 95 % CI: −0·31, −0·08, I2 = 34 %). According to subgroup analysis, whey protein led to a statistically significant decrease in circulating IL-6 levels in individuals with sarcopenia and pre-frailty (MD: −0·98, 95 % CI: −1·56, −0·39, I2 = 0 %). These findings may be dependent on participant characteristics and treatment duration. Conclusions: These data support that whey and soy protein supplementation elicit anti-inflammatory effects by reducing circulating IL-6 and TNF-α levels, respectively. This effect may be enhanced by soy isoflavones and may be more prominent in individuals with sarcopenia.

Accruing adverse changes in muscle physiology across the lifespan may lead to reduced muscle mass and physical capacity, particularly after the fifth decade (3) , known as sarcopenia (4) . From the beginning of the fourth decade, muscle mass decreases by approximately 0·5 % every year. The multifactorial determinants of this phenomenon include reduced levels of anabolic hormones, chronic inflammation, degradation of the muscle contractile proteins, loss of regenerative capacity, altered neural activation, and mitochondrial dysfunction (5,6) . Sarcopenia is associated with an increased circulating pro-inflammatory signalling (i.e., higher levels of TNF-α and IL-6) (7,8) . In conjunction with sarcopenia, concomitant accumulation of adiposity has also been observed during ageing, representing sarcopenic obesity, which is also linked with elevated inflammatory markers (9,10) . Accelerating age-related muscle wasting is partially explained through systemically and locally elevated oxidative stress and reactive oxygen species (ROS) accumulation (11)(12)(13) . Excessive ROS levels may result in damaged muscle and DNA proteins, triggering the release of pro-inflammatory cytokines and leading to low-grade inflammation (14) . Interestingly, antioxidative properties derived from nutrients may prevent excess ROS inflation that could alter muscle proteostasis (15) . Hence, finding nutritional strategies to mitigate low-grade inflammation may be considered as a safe and effective strategy for the prevention and treatment of sarcopenia.
Albeit protein supplementation is associated with reduced circulating levels of pro-inflammatory cytokines (16,17) , different protein sources may exert distinct anti-inflammatory effects (18) . Specifically, soy food intake has been associated with lower circulating levels of IL-6 and TNF-α (19) ; however, the functional properties of whole foods may differ compared with nutrients in isolation (20) . In this regard, previous systematic reviews have observed a reduction of serum CRP levels following intact whey and soy protein supplementation (21,22) , while the addition of soy isoflavones has been linked with a decline in circulating IL-6 levels among postmenopausal women (23) . Thus, isolated sources of protein may elicit promising isolated anti-inflammatory responses, although the most effective source of intact protein in alleviating circulating pro-inflammatory cytokine levels remains to be fully elucidated. To date, no previous meta-analysis has investigated the effects of intact whey and soy protein supplementation on multiple inflammatory markers in older adults. The aim of this systematic review and meta-analysis is to investigate the effects of intact whey and soy protein supplementation on serum CRP, IL-6 and TNF-α levels in older adults.

Methods
This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (24) . The protocol of this systematic review and meta-analysis was registered in the PROSPERO International prospective register of systematic reviews (CRD42021252603).

Search strategy
Two independent reviewers (KP and MI) searched the MEDLINE, Web of Science, Scopus and Cochrane Library databases from inception until September 2021, using the following search terms: 'whey OR soy' in combination with 'older adults' and 'inflammation OR high sensitivity-C reactive protein OR C reactive protein OR IL-6 OR tumour necrosis factor-a'. The complete search strategy is presented in Supplementary Table 1. No restrictions in terms of geographical region were applied. Articles were written in English and discrepancies in the literature search process were resolved by a third investigator (MM).

Study selection
Studies in this systematic review and meta-analysis were included based on the following criteria: (1) they were RCT; (2) the intervention group received intact soy or whey protein supplements in oral form; (3) the comparator group received a placebo or a non-identical appropriate treatment; (4) circulating levels of CRP, IL-6 and/or TNF-α were assessed; (5) participants that took part in the intervention had a mean age ≥ 50 years old and (6) full text was written in English. Accordingly, studies were excluded if: (1) they were not randomised trials; (2) participants were institutionalised; (3) studies were missing the baseline and/or post-intervention outcome values; (4) whey and soy protein products were in peptide/whole-food form and (5) whey and soy protein supplements were consumed enterally (Supplementary Table 2). Finally, if studies were comprised of a comparator group of < 50 years of age, they were included in the analysis as long as the participant age was similar to the intervention group.

Data extraction and quality assessment
Two authors (KP and MI) extracted data independently on name of first author, date of publication, country of origin, study design, participant health status, gender, age, BMI, sample size, intervention type, dose and duration, daily energy and protein intake, serum high-sensitivity CRP (hs-CRP), CRP, IL-6 and TNF-α levels. CRP and hs-CRP units were converted to mg/l, while IL-6 and TNF-α values to pg/ml. Disagreements between authors on data eligibility were resolved by a third reviewer (MM). When studies contained multiple doses of protein supplementation, only the highest dose was considered as the intervention arm.
The quality of included studies was evaluated using the Riskof-bias 2 tool and the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) system approach. Riskof-bias 2 is a detailed and comprehensive tool to assess the risk of bias in randomised trials included in Cochrane Reviews, focussing on (1) the evaluation of randomisation process, (2) deviations from intended interventions, (3) missing outcome data, and (4) measurement of the outcome and selection of the reported result (25) . According to the Risk-of-bias 2 scoring system, study quality was defined as high, some concerns or low. Additionally, the GRADE approach involves the consideration of (1) within-study risk of bias, (2) directness of evidence, (3) study heterogeneity, and (4) precision of effect estimates and risk of publication bias, using four levels of quality (high, moderate, low and very low) (26) .

Data synthesis and statistical analysis
Our analysis reported on the differences among circulating inflammatory markers (hs-CRP, CRP, IL-6, and TNF-α) following whey and soy protein supplementation, when compared with individuals receiving placebo or a non-identical treatment. Quantitative data were treated as continuous measures and were combined by calculating the mean differences between outcomes from baseline and the follow-up period of each intervention. Statistical significance between the intervention and comparator groups was assessed using the random effects inverse-variance model. Missing standard deviations of outcomes were estimated depending on the availability of either CI, SE, t and P values or by calculating a correlation coefficient (Corr) from a known change from baseline standard deviation. A 95 % CI was used to calculate missing SD and considering the absence of studies with SD changes from baseline to follow-up, an extra analysis utilising a Corr value of 0·7 was performed (27) .
The statistical heterogeneity between studies was assessed using the overlap of their 95 % CI and expressed as measures of Cochran's Q (Chi-square test) and I 2 . Data classification as moderately heterogeneous was based on I 2 from 50 % to 74 %, and highly heterogeneous from 75 % and above (28) . Furthermore, sensitivity analysis was performed to evaluate the robustness of the reported statistical results by discounting the effect of confounding factors on outcome measures through a leave-one-out analysis. Studies with a high risk of bias and/or the study with the highest effect size were discounted through a leave-one-out sensitivity analysis. Publication bias was assessed using Begg's funnel plots and Egger's linear regression test (29) using R software. Data were meta-analysed and forest plots were drawn using Review Manager (RevMan 5.4.1). A P value of < 0·05 was considered statistically significant.

Subgroup and sensitivity analyses
Subgroup analyses were performed based on Corr equal to 0·7, age, BMI, treatment dose and duration, soy protein and isoflavone co-supplementation, soy protein supplementation during postmenopause, and whey protein supplementation in participants with sarcopenia and pre-frailty. Sensitivity analyses were performed using a leave-one-out analysis, excluding the study with the largest effect size and the study with the highest bias risk.

Search results and study characteristics
The initial search generated 5432 records, in which 5220 were excluded due to ineligibility issues and study duplicates. Following a full-text review of the remaining 212 studies, 153 articles were removed and 59 articles were sought for retrieval. In total, 45 full-text reports were assessed for eligibility. Acute studies and articles with missing or incomplete data were excluded from the analysis. Overall, 31 studies were included in the systematic review and meta-analysis ( Fig. 1).

Risk of bias and quality of evidence assessment
Out of 18 studies utilising whey protein supplements, 11 studies had an overall low risk of bias (20,30,32,35,(47)(48)(49)51,52,55,57,60) , five studies had some concerns (33,34,53,56,58) and two studies had a high risk of bias (31,59) . Specifically, one study was unblinded (53) and six studies did not provide any details on allocation treatment (33,34,52,56,58,59) , whereas although one study claimed there was allocation concealment, no further details were provided (48) . In addition, one study had a high risk of trial personnel being aware of participants' assigned intervention (31) . In two studies, there were some concerns regarding missing outcome data (35,59) . Finally, in two studies, the outcome assessment could have been influenced by knowledge of the intervention received (31,53) .
Traffic light plots were created using robvis visualisation tool. A detailed description of Risk-of-bias 2 traffic light plots for whey and soy protein supplementation studies are presented in Supplementary Tables 4 and 5, respectively. Finally, the GRADE system approach showed that the quality of evidence for the primary outcomes was moderate (Supplementary Tables 6a-d and 7a-d).
Effect of whey protein supplementation on circulating inflammatory markers analysis  . 2d); however, a high heterogeneity among studies was observed (I 2 = 96 %). Using Corr equal to 0·7 did not demonstrate any significant changes compared with the main analysis (online Supplementary Fig. 1a-d).

Subgroup analysis of whey protein supplementation trials
Subgroup analysis based on age revealed no significant changes in serum hs-CRP, CRP, TNF-α and IL-6 in adults < 60 and ≥ 60 years of age (online Supplementary Fig. 3a, d).

Subgroup analysis of soy protein supplementation trials
Using Corr equal to 0·7 did not demonstrate any significant changes compared with the main analysis (online Supplementary Figures 2a-d).

Sensitivity analysis based on effect size and bias risk
Sensitivity analyses using a leave-one-out strategy based on the effect size of whey protein (online Supplementary Fig. 14a-d) and soy protein supplementation studies (online Supplementary  Fig. 15a-d) did not alter outcome measures. Likewise, sensitivity analyses using a leave-one-out strategy for bias risk did not reveal any changes compared with the results from the main analysis (whey protein studies: Supplementary Fig. 16a-d; soy protein studies, Supplementary Fig. 17a, b).

Publication bias
Visual examination to test for asymmetry among studies for serum IL-6 and CRP levels using Begg's funnel plots are illustrated in Supplementary Fig. 18a, b and Supplement Fig. 18c, d, respectively. Egger's linear regression test revealed no evidence for publication bias in both the intervention (z = −0·6174, P = 0·5369) and the comparator group (z = −0·0367, P = 0·9708) for serum IL-6 levels following whey protein supplementation based on twelve RCT in this meta-analysis. Additionally, Egger's linear regression test also revealed no evidence for publication bias in the intervention group for serum CRP levels (z = −0·0043, P = 0·9966); however, an increased risk for publication bias was observed in the comparator group (z = 2·5193, P = 0·0118).

Discussion
This meta-analysis showed a significant decline in circulating IL-6 and TNF-α levels following whey and soy protein supplementation, respectively. Subgroup analysis based on age (< 60 years) revealed a significant reduction of serum TNF-α following whey protein consumption, while subgroup analysis accounting for sarcopenia and pre-frailty status also exhibited a significant reduction of serum IL-6. In addition, a decline in serum CRP levels was observed following a treatment duration of ≤ 8 weeks and in participants with BMI ≤ 25 kg/m 2 . Similarly, subgroup analyses based on age (≥ 60 years) and treatment duration of > 8 weeks showed a significant reduction of serum TNF-α following soy protein supplementation, while the addition of isoflavones exhibited further benefits by reducing serum CRP levels. Overall, these findings suggest that whey and soy protein supplementation may exert distinct anti-inflammatory properties, which are dependent on participant physiological characteristics, treatment duration, and addition of isoflavones.
A previous meta-analysis has demonstrated that whey protein may mitigate low-grade inflammation by decreasing serum CRP levels; however, the increased heterogeneity among studies may have influenced such findings (22) . Although a high degree of heterogeneity among studies was detected, our analysis revealed a significant effect of whey protein supplementation in reducing serum IL-6 levels. Noteworthy that insignificant results were found in the subgroup analyses on serum TNF-α according to age (< 60 vs. ≥ 60 years), BMI (< 25 vs. ≥ 25 kg/m 2 ) and treatment duration (≤ 8 vs. > 8 weeks) on serum CRP levels, our findings should be treated with caution due to the small number of studies. Interestingly, our subgroup analysis revealed significant benefits of whey protein supplementation on sarcopenia and pre-frailty, highlighting a significant decline in circulating IL-6 levels. The combination of these two populations was based on their identical characteristics in relation to muscle mass and strength, displaying a low degree of study heterogeneity. In this context, hospitalised patients with frailty have elicited a beneficial effect on reducing serum IL-6 following whey protein supplementation (63) , which may be explained by a concomitant increase in glutathione concentrations and a decrease in ROS accumulation (64) . Moreover, reduced serum IL-6 levels have also been demonstrated in individuals with sarcopenia by comparing a whey protein-based product to placebo; however, its nutrient content may have masked the effectiveness of whey protein in isolation (65) . Particularly, the combination of carotenoids, choline, vitamin A and E and Fe may exert anti-inflammatory effects (21,66,67) and act as confounders in assessing the efficacy of whey protein in alleviating lowgrade inflammation. In a subgroup analysis, one study combined whey protein with vitamin D, which may be partially responsible for serum IL-6 level reduction (68) . However, research is conflicting regarding the effects of vitamin D on reducing serum inflammatory markers in older adults (9,69,70) . Our findings suggest that a -0·98 pg/ml mean reduction in serum IL-6 concentrations of individuals with sarcopenia and pre-frailty may be of clinical relevance given a 0·7 pg/ml mean difference between younger and older populations based on cross-sectional data (71) . Therefore, whey protein supplementation may be a valuable dietary strategy to attenuate the progression of low-grade inflammation and exacerbation of sarcopenia and frailty risk. Considering the increased baseline pro-inflammatory profile in people with sarcopenia, the effects of intact protein supplementation may be more prevalent in these populations. However, due to the limited number of studies and their heterogeneous designs, our results regarding the effectiveness of whey protein in reducing circulating inflammatory markers in individuals with sarcopenia and frailty should be treated with caution.
Previous meta-analyses have revealed that soy-based protein foods and supplements may not alter serum inflammatory status (72,73) . However, these findings were based on flavonoidenriched foods (65) and postmenopausal women from which only serum CRP levels were measured (72) . Additionally, experimental studies have not observed a significant effect of soy food consumption on serum CRP levels (74) that may be attributed to the interaction of multiple nutrients contained in whole soy foods (75) compared with isolated sources (76) . Our analysis revealed a significant effect of soy protein supplementation in reducing serum TNF-α levels, which are in line with previous research (77,78) , although, insignificant reductions of serum IL-6 levels were displayed as reported previously (23) . Furthermore, subgroup analysis showed that the addition of isoflavones did not decrease serum CRP and IL-6 levels; however a significant reduction of serum TNF-α was observed. These results may be attributed to the bioactive substances in soy isoflavones (i.e. phenolic compounds, daidzein, and genistein) that exert antioxidant effects (79,80) through glutathione peroxidase regulation and reduction of ROS and malondialdehyde infiltration (81) . Although several soy isoflavone doses were administered in this systematic review, subgroup analysis based on dose was not feasible due to the low number of studies. Therefore, whether greater isoflavone quantities correspond to higher decreases of circulating inflammatory cytokines is currently unclear.

Limitations
Our study was prone to limitations. High variability regarding participant health status, isoflavone dose, and study sample size potentially accounted for the increased heterogeneity in multiple subgroup analyses. The sample size of studies did not allow for subgroup analyses based on healthy populations and individuals with comorbidities. Hence, definitive conclusions around specific conditions and healthy older populations cannot be extrapolated. In addition, the quality of evidence was moderate according to GRADE system approach, while several studies did not use a placebo group as a comparator. Finally, nutrient intake was not controlled in multiple studies, which may have influenced the participants' inflammatory profile. More importantly, the effects of vitamins, minerals, alcohol, and energy intake may be pivotal contributors in regulating pro-inflammatory cytokine status; hence, the scarcity of data on these parameters should be considered in future studies.

Conclusions
Systemic low-grade inflammation is a critical contributor to muscle proteolysis during ageing. Our study found a significant reduction of circulating IL-6 and TNF-α levels following whey and soy protein supplementation, respectively. These effects were particularly augmented with the addition of soy isoflavones and populations with sarcopenia and pre-frailty. Whey and soy protein supplementation may serve as a valuable dietary intervention in reducing serum inflammatory cytokine levels, however, more homogeneous studies are required to provide more reliable results on healthy populations and individuals with comorbidities.