Data were analysed in R version 3.6.3. Primary used for modelling were ‘lme4’ and ‘lmerTest’ ( 21 – Reference Bates, Mächler and Bolker23 ) . Linear mixed-effects models were conducted to compare changes in the biomarkers of inflammation from baseline and the end of each dietary phase (weeks 8 and 16) to determine if the changes were significantly different for the Plant-based phase compared to the Animal phase, adjusting for the fixed effect of diet order (i.e. study arm).
Data for inflammatory biomarkers were collected from participants’ blood samples at baseline, week 8, and week 16. The data were analysed from Olink Proteomics using the 92 inflammation-related protein biomarker panel ( https://www.olink.com/products/inflammation/ ). Specific inflammatory biomarkers were selected for primary analysis based on a review of previous literature regarding biomarkers that were both available within the Olink panel and that had been reported to change in response to increased consumption of plant-based foods: interleukin-6 (IL-6), interleukin-18 (IL-18) and tumour necrosis factor (TNF) ( Reference Eichelmann, Schwingshackl and Fedirko15 – Reference Menzel, Jabakhanji and Biemann17 ) . We also selected three additional biomarkers that were reported to improve in previous nutrition studies in response to dietary change ( Reference Wastyk, Fragiadakis and Perelman19 ) . The other three biomarkers were: interleukin-12 subunit beta (IL-12B), interleukin-10 (IL-10) and transforming growth factor beta (TGF-β) for the main analysis based on these being some of the most commonly analysed biomarkers of inflammation from the literature on dietary patterns and inflammation ( Table 2 ) ( Reference Wastyk, Fragiadakis and Perelman19 ) . The remaining eighty-six markers were analysed as a secondary analysis, and significance was assessed after controlling for a false discovery rate of <0·05 ( Reference Benjamini and Hochberg20 ) . Note that all biomarkers are expressed in units of normalised protein expression (NPX).
Randomisation to one of the two diet sequences (Plant→Animal v. Animal→Plant) was performed in block sizes of 4 by an independent statistician. Pairs (spousal or parent–child) were randomised in block sizes of 2. Participants did not learn of their diet sequence until they completed all baseline measures and surveys. Laboratory technicians and study staff conducting blood and stool analyses were blinded to the diet sequence.
The SWAP-MEAT (ClinicalTrial.gov registration number: NCT 03718988) was a randomised crossover design that involved participants eating at least 2 or more daily servings of plant-based meats for 8 weeks (Plant-based phase) followed by 2 or more daily servings of animal meats for 8 weeks (Animal phase) (or vice versa). A single serving size was 3–4 ounces per product. Complete details of the nutritional profile of the plant-based meat and animal meat products can be found in the published manuscript of the original study ( Reference Crimarco, Springfield and Petlura7 ) . The animal meats used in the present study were supplied from a San Francisco-based organic foods delivery service; the red meat sources were grass-fed. The plant-based meats used in the present study were supplied from Beyond Meat. Both types of meats were provided to the participants throughout the study. Other than the study products, participants were instructed to keep all dietary habits as consistent as possible between the two phases, such that the only dietary changes were the consumption of the plant-based meat products or the animal-based meat products. Participants consented to allow any frozen and archived blood samples that were available after completing the main study analyses to be used to examine additional factors that were been specified in the original protocol. Complete details of the original study have been reported elsewhere ( Reference Crimarco, Springfield and Petlura7 ) . Participant enrollment began on 5 December 2018 and continued through 9 July 2019. The date of final follow-up data collection was on 5 December 2019.
The participants ate similar amounts of calories, macronutrients and servings of plant-based meat or animal meat products (~2·5 and 2·6 servings per day, respectively) throughout the intervention. Complete details of adherence are available elsewhere ( Reference Crimarco, Springfield and Petlura7 ) . Table 2 presents the mean values of the six biomarkers selected for the main analysis and Supplementary Table S1 presents the mean values for all of the biomarkers. Overall, none of the biomarkers in main analysis comparing the between-phase change scores reached statistical significance ( Table 3 and Supplementary Table S2).
Discussion and conclusion
This secondary analysis of the SWAP-MEAT randomised crossover trial assessed changes in biomarkers of inflammation when participants ate two or more daily servings of plant-based meats v. animal meats for 8 weeks each. Overall, none of the change scores between the two diet phases were significantly different, and for the within-phase paired scores, only four out of ninety-two inflammatory biomarkers reached statistical significance. Additionally, the four biomarkers all slightly increased regardless of diet phase and order. These results do not support our hypothesis, that biomarkers of inflammation would improve during the plant-based meat diet. It could be that the duration of the diet protocols was not long enough to observe any notable changes. In Menzel et al.'s review of plant-based diets and their association with biomarkers of inflammation, the authors noted that there were no significant changes in C-reactive protein (CRP) concentrations for any of the studies that assessed vegetarian diets for a duration of 6 months or less(Reference Menzel, Jabakhanji and Biemann17). Their sensitivity analysis indicated that the impact of vegetarian diets on CRP was more obvious for participants following the diet for at least 10 years(Reference Menzel, Jabakhanji and Biemann17). While CRP is not included in the Olink panel, some of the other biomarkers in our analysis (IL-6 and TNF) were also reported to improve in previous meta-analyses(Reference Eichelmann, Schwingshackl and Fedirko15–Reference Menzel, Jabakhanji and Biemann17); therefore, it is possible that the length of time for substituting plant-based meats in place of red meats within this study was too short to achieve detectable improvements in systemic inflammation(Reference Haghighatdoost, Bellissimo and de Zepetnek24).
A second explanation for the modest changes in biomarkers of inflammation is that the mere substitution of plant-based meats over animal meats is not sufficient enough by itself to improve systematic inflammation. Only replacing animal meats (which comprised ~25 % of our participants’ energy intake in this study), but keeping all other dietary components the same, may not be enough to improve chronic inflammation, since the majority of participants’ foods were self-selected and likely did not change during the intervention. An overall healthy, plant-based diet may have yielded different results, since the focus of plant-based diets is to maximise the consumption of nutrient-dense foods like fruits, vegetables, nuts, wholegrains and seeds, while minimising the consumption of processed foods and animal foods(Reference Tuso, Ismail and Ha25). In general, highly processed foods are associated with increased weight gain and an increased risk for a number of leading chronic health conditions. The plant-based meat products used in our study would fit the definition of ‘ultra-processed’ foods, due to a number of added ingredients used to enhanced the flavor and texture of the meats. Therefore, it is possible that the plant-based meats may not have any benefit to improving inflammatory biomarkers due to the products being highly processed. However, others have noted that plant-based meats ‘are often perceived as low-quality, ultra-processed foods. However, we argue that the mere industrial processing of ingredients of plant origin does not make a PBMA product ultra-processed by default(Reference Toribio-Mateas, Bester and Klimenko26).’ Additionally, recent work conducted by van Valiet et al. on the metabolomic profiling that compared a plant-based beef substitute to grass-fed beef demonstrates the complexity in trying to compare and understand nutritional adequacy of plant-based meat products to animal meats, since both foods assessed both contained and lacked differing nutrients with potential benefits to health(Reference van Vliet, Bain and Muehlbauer27).
Other potential benefits of substituting plant-based for animal meats have been reported. A recent randomised clinical trial noted positive changes in the gut microbiome among participants that substituted ~4–6 servings of plant-based meats per week in place of animal meats(Reference Toribio-Mateas, Bester and Klimenko26). Also, reductions in animal meat consumption, particularly red and processed meats, have been widely reported to have beneficial impacts on environmental sustainability(Reference Van Vliet, Kronberg and Provenza28).
The strengths of the SWAP-MEAT study were minimal missing data from all participants, the crossover design of this trial allowed each participant to serve as their own control, and the strong dietary adherence to both dietary conditions. There were also some limitations to the study. Only the plant-based meat and animal meat products were controlled, while the remainder of the diet was self-selected, which limits the researchers’ control over the portion sizes or specific food selection. However, this does allow for increased external validity and generalizability. Our sample was ~66 % female and existing research on gender differences in inflammation and immune functioning suggest that women have higher levels of inflammation and poorer immune functioning(Reference Cartier, Côté and Lemieux29,Reference Klein and Flanagan30) . This may impact the generalizability and interpretation of our results. Finally, the present study utilised only Beyond Meat products. There are numerous plant-based meat products with different ingredients or formulations, therefore the findings may not be generalisable to the consumption of other plant-based meat products.
In conclusion, while the results of the main trial indicated several improvements in CVD risk factors, including TMAO, for the plant-based meats, no differences in the selected biomarkers of inflammation were observed(Reference Crimarco, Springfield and Petlura7). Future research may benefit from longer study duration periods.