n-3 Long chain PUFA (n-3 LCPUFA) have been widely studied due to their associations with benefits to human health including lowering inflammation(Reference Calder1,Reference Calder2) , decreasing the risk of CVD and events(Reference Mori and Woodman3–Reference Innes and Calder8) and improving cognition(Reference Frensham, Bryan and Parletta9,Reference Ciappolino, Mazzocchi and Botturi10) . The bioactive n-3 LCPUFA are EPA and DHA; for these fatty acids (FA) to exert their actions, they require incorporation into the bloodstream and delivery to cells and tissues(Reference Calder11).
Synthesis of EPA and DHA from the essential FA α-linoleic acid is low in humans, and therefore consumption of preformed EPA and DHA from the diet is preferred(Reference Arteburn, Hall and Oken12). In the UK, it is recommended that adults consume EPA and DHA in the form of fish to provide the equivalent of at least 450 mg of EPA + DHA/d (two × approximately 140 g portions per week, one of which is to be oily/fatty fish)(13). Other global or national recommendations for EPA and DHA intake by adults are in the range 250–600 mg/d (see Table 5 of Calder(Reference Calder11)). However, the UK population, in addition to those of the USA and many other countries, consumes well below the recommended amounts of oily fish and of EPA and DHA(Reference Micha, Khatibzadeh and Shi14). It is estimated that only 18·9 % of countries achieve the minimum daily recommendation of 250 mg EPA + DHA(Reference Micha, Khatibzadeh and Shi14). Here, supplementation would be an alternative option to diet to increase intake of bioactive EPA and DHA.
EPA and DHA are commercially available in supplements in multiple chemical forms as TAG, ethyl esters (EE) or NEFA. The chemical form in which EPA and DHA are administered has been shown to have some impact on their incorporation into the bloodstream, cells and tissues, most noticeably when consumed in the absence of a fatty meal. This is particularly observed when EPA and DHA are administered as EE, which are used in many clinical trials in which unsupervised daily dosing of capsules occurs over several weeks/months(Reference El-Boustani, Colette and Monnier15–Reference West, Kindberg and Hustvedt17). This suggests that the optimal absorption of these FA is dependent on digestive processes, which act in response to intake of dietary fat. Therefore, variation in the fat content of the meal consumed alongside the capsules or if the capsules were taken on an empty stomach may be one of the reasons why some studies fail to find beneficial effects of n-3 LCPUFA(Reference Rice, Bernasconi and Maki18).
Considering this, the development of alternative delivery systems to improve the bioavailability of EPA and DHA in the absence of dietary fat has been of great interest. Pre-emulsification of n-3 LCPUFA in TAG form and self-micro-emulsifying delivery systems (SMEDS) which promote spontaneous emulsification of n-3 FA in EE form in the gastric environment have been shown to result in superior incorporation into blood and cell lipid pools in comparison with other TAG and EE formulations, respectively(Reference West, Kindberg and Hustvedt17,Reference Raatz, Johnson and Bukowski19–Reference Bremmell, Briskey and Meola22) . Other approaches to enhancing uptake of EPA and DHA are less well researched but include the use of a mixture of monoglyceride, diglyceride and TAG(Reference Bremmell, Briskey and Meola22,Reference Banno, Doisaki and Shimizu23) . Investigation in rats showed improved appearance of DHA in the lymph using monoglyceride and diglyceride formulations in comparison with TAG and EE formulations(Reference Banno, Doisaki and Shimizu23), but no such improvement was seen when using a diglyceride and TAG DHA formulation in comparison with diglyceride and TAG(Reference Tamai, Murota and Maruyama24). A single study in humans reported significantly enhanced bioavailability of EPA using a mixture of monoglyceride, diglyceride and TAG in comparison with TAG and monoglyceride formulations(Reference Wakil, Mir and Mellor25). These studies suggest the use of monoglycerides in combination with diglyceride and TAG is key in n-3 LCPUFA bioavailability, but this requires further evaluation as evidence for both EPA and DHA, and comparison with bioavailability of EE n-3 LCPUFA in humans is lacking.
Therefore, the present study aimed to compare the use of a novel alternative glyceride formulation of EPA + DHA, comprising of a mixture of monoglyceride, diglyceride and TAG in a ratio of 1:2·5:1 thought to enhance solubility, digestion and therefore availability of EPA and DHA, with an EE-EPA + DHA formulation, in the absence of dietary fat, on EPA and DHA concentrations in blood plasma over a period of 12 h following single dosing. It was hypothesised that the glyceride formulation would result in higher concentrations of EPA and DHA in blood plasma and that the maximal concentration would be achieved faster than seen with the EE-EPA + DHA formulation.
Methods
Subjects
All procedures involving human subjects were approved by the London – Brighton & Sussex Research Ethics Committee (REC 19/LO/0939) and the study is registered at www.isrctn.com (study ID: ISRCTN18364209). The trial was conducted according to the principles of the Declaration of Helsinki, and all participants gave written informed consent prior to enrolment.
Ten healthy male and ten healthy female participants aged 50–70 years were enrolled into the study. Inclusion criteria for participation were as follows: age between 50 and 75 years, BMI (kg/m2) 20–35, self-reported oily fish intake of <1 portion/week and an omega-3 index (erythrocyte EPA + DHA(Reference Harris26)) determined from a screening blood sample of ≤6·5. Exclusion criteria were any chronic medical condition, cancer within the last 2 years, gastrointestinal problems, allergy to fish, smoking, pregnancy or lactation, or consumption of any n-3 FA or other lipid supplements. The study took place between October 2019 and February 2020.
Study design and supplements
The present study was a double-blind randomised crossover trial that investigated an n-3 glyceride formulation with 500 mg EPA + 200 mg DHA per capsule (BASF AS), alongside an EE formulation (also providing 500 mg EPA + 200 mg DHA per capsule) (BASF AS) both in soft gelatin capsules. The FA compositions of the EE and glyceride mixtures were identical. They contained (g/100 g FA): 18 : 1n-9, 0·8; 18 : 2n-6, 1·5; 18 : 3n-3, 0·9; 18 : 4n-3, 3·7; 20 : 3n-3, 0·2; 20 : 4n-6, 2·3; 21 : 5n-3, 1·8; 22 : 5n-6, 0·8; 20 : 5n-3 (EPA), 53·2; 22 : 5n-3 (docosapentaenoic acid (DPA)), 3·0; 22 : 6n-3 (DHA), 23·4; and other minor FA (all < 0·4), 3·1. Both formulations provided about 28 mg of DPA. The distribution of EPA and DHA within the monoglyceride, diglyceride and TAG is not known.
As this was a crossover design, all participants received both treatments but in random order. Blinding, randomisation and supplement packaging were completed by the Research Pharmacy at Southampton General Hospital, Southampton, UK, by individuals independent of the researchers involved in the study. Treatment group blinding was maintained until completion of statistical analysis of all data.
Participants attended the National Institute for Health Research Wellcome Trust Clinical Research Facility, Southampton General Hospital, Southampton, UK, on three occasions. The first was a screening visit during which participants provided written informed consent prior to having their weight and height measured and providing a non-fasting blood sample. This sample was used to determine erythrocyte omega-3 index (EPA + DHA as a % of total erythrocyte FA(Reference Harris26)), with a value of ≤6·5 required for study enrolment. Participants who met all inclusion and exclusion criteria were enrolled into the study and randomly assigned. These participants made a further two clinic visits, both in the fasted state (≥10 h without food or drink other than water). At both of these visits, participants attended the clinic at approximately 07.30 hours at which a cannula was inserted into a forearm vein. Participants provided a 0 h blood sample after which they ingested a single dose (four capsules totalling 2·8 g EPA + DHA) of one of the study formulations with water. A member of the research nursing team observed capsule ingestion to ensure compliance and record time of consumption. Further blood samples were collected at 1, 2, 3, 4, 5, 6, 8 and 12 h post-supplement ingestion. Low-fat meals with decaffeinated tea or coffee were given directly after the 3 and 6 h samples were collected, and again at approximately 10 h. These meals consisted of two slices of toast without butter or alternative spread but with jam, and either a small apple or orange accompanied by tea or coffee made with skimmed (0·1 % fat) milk, totalling 2·3 g fat for the entire meal. A more substantial meal was provided after the 12 h blood sample was collected; participants could choose either sandwiches, fruit or cake and a juice drink, or a pre-made meal such as curry or pasta to heat and consume at home. The third clinic visit took place at least 2 weeks after the second clinic visit (the average time between the second and third clinical visits was 26·1 (sd 2·4) d); all procedures were the same at this visit except that the alternative supplement was taken.
Sample preparation
Blood was collected into tubes containing EDTA and directly stored on ice before being processed within 1 h of collection. Plasma was prepared by centrifugation of blood samples collected at all time points at 3000 g for 15 min at 4°C and analysed for EPA and DHA contents. Plasma and erythrocyte samples were stored at −80°C before analysis.
Fatty acid composition analysis
Total lipid was extracted from 0·5 ml plasma with 5 ml chloroform–methanol (2:1, v/v) containing 0·2 m BHT(Reference Bligh and Dyer27). A quantity of 1 ml of 1 m sodium chloride was added and the sample vortex mixed prior to centrifugation at 1500 g for 5 min at room temperature. The lower solvent phase was aspirated and evaporated under N2 at 40°C. The lipid extract was re-dissolved in 0·5 ml toluene and FA were released from esterified lipids and simultaneously derivatised to methyl esters by incubation with 2 % H2SO4 in methanol for 2 h at 50°C to form fatty acid methyl esters(Reference Fisk, West and Childs28). The samples were then neutralised and fatty acid methyl esters transferred into hexane for analysis by GC; fatty acid methyl esters were separated on a BPX-70 fused silica capillary column (30 m × 0·2 mm × 0·25 µm; manufactured by SGE) in a HP6890 gas chromatograph fitted with a flame ionisation detector. GC run conditions were as described elsewhere(Reference Fisk, West and Childs28). Dipentadecanoyl phosphatidylcholine added into the initial plasma sample was used as an internal standard for quantification purposes and a Supleco 37 Component FAME Mix was used as a calibration reference standard (Sigma-Aldrich).
Other laboratory analyses
The plasma concentrations of TAG, total cholesterol, HDL-cholesterol and glucose at study entry were measured using enzyme-linked colorimetric assays (Alpha laboratories and Microgenics GmbH) on a Konelab 20 autoanalyzer in accordance with manufacturer’s instructions. Non-HDL-cholesterol at study entry was calculated by subtracting HDL-cholesterol from total cholesterol measurements. Plasma insulin concentration at study entry was measured by ELISA (Access Ultrasensitive Insulin kit; Beckman Coulter). Plasma C-reactive protein concentration at study entry was measured by using a high sensitivity ELISA kit (CRP Latex kit; Beckman Coulter).
Statistical analysis
The study sample size was estimated according to the anticipated change in EPA + DHA content of plasma over the 12 h period following single dosing. Assuming a similar difference between the two formulations as was seen in the previous study comparing SMEDS and a standard EE formulation(Reference West, Kindberg and Hustvedt17), a sample size of twenty was estimated to give 90 % power to detect this difference as statistically significant by pairwise comparison with significance defined as P ≤ 0·05.
EPA, DPA and DHA in plasma are expressed as relative (percentage of total FA) and absolute (μg/ml of plasma) concentrations. All FA data were normalised against baseline concentrations and the distribution of all data checked. Any skewed data were normalised by logarithmic transformation. Incremental AUC (iAUC), maximum concentration change (Cmax) and time point at which Cmax was achieved (Tmax) were calculated using GraphPad Prism version 7 (GraphPad). Repeated-measures ANOVA was performed on all time-course data controlling for possible confounding effects of sex. There were no significant confounding effects of age or BMI on either relative or absolute EPA, DPA, DHA or EPA + DHA time-course data (P ≤ 0·172 and ≤ 0·659, respectively). Therefore, these variables were not controlled for in the analyses. The Kruskal–Wallis test was performed to compare baseline characteristics between sexes and to compare plasma iAUC, Cmax and Tmax between treatments as log transformation was unable to correct the skewed distribution of these data. Correlation of metabolic measurements at study entry with iAUC, Cmax and Tmax data was investigated using Spearman’s correlation. All statistical analyses were conducted using SPSS version 26 (IBM Corporation), and significance was defined as P ≤ 0·05.
Results
Participant characteristics
Fig. 1 shows the flow of participants through the study and numbers of participants in each treatment group. Twenty participants (ten male and ten female) consumed both single dose treatments at least 2 weeks apart with the order of these randomly assigned. Capsule ingestion was observed by research nurses resulting in 100 % compliance. Table 1 details participant characteristics at study entry. Participants included in the study had a mean age of 59·1 (sd 0·7) years (range 50–70 years), mean BMI of 26·95 (sd 0·73) kg/m2, consumed on average 2·55 (sd 0·31) portions of oily fish per month and had a mean screening omega-3 index of 5·35 (sd 0·19). Female participants had significantly higher BMI and total cholesterol, and lower blood glucose than male participants (Table 1). There were no significant differences in erythrocyte omega-3 index or in baseline relative or absolute concentrations of EPA, DPA, DHA or EPA + DHA between male and female participants (Table 1).
Fig. 1. Consolidated Standards of Reporting Trials (CONSORT) diagram of participant inclusion and flow through the study. EE, ethyl ester; glyceride, mixture of monoglyceride, diglyceride and TAG EPA + DHA.
Table 1. Characteristics of the participants at study entry*
(Median values and 25th, 75th percentiles; percentages)
HOMA2-IR, homeostatic model assessment of insulin resistance; hsCRP, C-reactive protein measured with a high-sensitivity assay.
* All participants consumed both treatments in a crossover design separated by at least a 2-week period. Capsule consumption was supervised resulting in 100 % compliance. P for comparison of male and female participants; significance defined as ≤0·050.
EPA, docosapentaenoic acid and DHA incorporation in plasma with single dosing
The glyceride-EPA + DHA supplement resulted in a significantly greater increase in the absolute (µg/ml) and relative (%) concentrations of EPA, DPA, DHA and EPA + DHA in the plasma total lipid pool within 12 h following single dosing than seen with the EE-EPA + DHA supplement (all P < 0·001, Figs. 2 and 3). Significantly higher Cmax, lower Tmax and greater iAUC for concentrations of EPA and DHA in the plasma following single dosing with the glyceride supplement were observed when compared to single dosing with the EE supplement (P ≤ 0·013, Figs. 2 and 3, Table 2). Significantly higher Cmax and greater iAUC for concentrations of DPA in the plasma following single dosing with the glyceride supplement were also observed when compared to single dosing with the EE supplement (P ≤ 0·023, Figs. 2 and 3, Table 2). An average 9·7-fold greater Δ concentration of EPA, 3·0-fold greater Δ concentration of DPA, 2·8-fold greater Δ concentration of DHA and 5·3-fold greater Δ concentration of EPA + DHA were achieved with single dosing of the glyceride supplement in comparison with the EE supplement (P < 0·001, Fig. 2, Table 2). The time at which the maximal concentration of EPA, DHA and EPA + DHA was achieved occurred on average 4, 1 and 2 h earlier, respectively, following single dosing with the glyceride supplement in comparison with the EE supplement (P < 0·001, Fig. 2, Table 2). The time at which the maximal concentration of DPA was achieved did not differ following single dosing with the glyceride supplement or the EE supplement (P ≥ 0·520, Fig. 2, Table 2).
Fig. 2. Changes from baseline in absolute concentration of plasma total EPA, docosapentaenoic acid (DPA), DHA and EPA + DHA after a single dose of glyceride-EPA + DHA or EE-EPA + DHA in healthy older adults. Values are medians and interquartile ranges. Glyceride-EPA + DHA n 20, EE-EPA + DHA n 20. EE, ethyl-ester; glyceride, mixture of monoglyceride, diglyceride, and TAG. 
, Treatment 1; 
, treatment 2.
Fig. 3. Changes from baseline in relative concentration of plasma total EPA, docosapentaenoic acid (DPA), DHA and EPA + DHA after a single dose of glyceride-EPA + DHA or EE-EPA + DHA in healthy older adults. Values are medians and interquartile ranges. Glyceride-EPA + DHA n 20, EE-EPA + DHA n 20. EE, ethyl ester; glyceride, mixture of monoglyceride, diglyceride, and TAG. , Treatment 1; , treatment 2.
Table 2. Summary of changes (Δ) in plasma EPA, docosapentaenoic acid (DPA) and DHA concentrations over 12 h after a single dose of EE-EPA + DHA or glyceride-EPA + DHA in healthy adults*
(Median values and 25th, 75th percentiles (P25, P75))
EE, ethyl ester EPA + DHA; Cmax, maximum concentration change; glyceride, mixture of monoglyceride, diglyceride and TAG EPA + DHA; iAUC, incremental AUC; Tmax, time at which Cmax occurs.
* P values were derived by using the Kruskal–Wallis test.
† The ratio of glyceride-EPA + DHA compared with EE-EPA + DHA for iAUC and Cmax.
Increase in the absolute concentrations of EPA, DPA and DHA over the 12 h period was reflected in an increase in the relative concentration of these n-3 LCPUFA (P < 0·001, Fig. 3, Table 2). The Cmax and iAUC for the relative concentration of EPA, DPA and DHA in the total plasma lipid pool were greater in participants when consuming the glyceride supplement in comparison with the EE supplement. The time at which the maximal concentrations of EPA and DHA were achieved occurred on average 1 h earlier with the glyceride supplement in comparison with the EE supplement (P < 0·001, Fig. 3, Table 2). There was no difference in the time at which the maximal concentration of DPA was achieved between the glyceride and the EE supplement.
Metabolic measures were not significantly correlated with iAUC, Cmax or Tmax data for either relative or absolute EPA, DHA or EPA + DHA (P ≥ 0·172, all).
Discussion
n-3 LCPUFA are found to be esterified into TAG and phospholipids in many commercially available supplements and in food. In this form, the FA require solubilisation and hydrolysis in the upper gastrointestinal tract, known as digestion, before they are available for absorption. The digestive process involves the secretion of bile and pancreatic enzymes which is stimulated by the presence of fat within a meal. Therefore, consuming supplements on an empty stomach or alongside a meal which is low in fat impairs the digestion and absorption of EPA and DHA into the bloodstream and so makes less EPA and DHA available to cells and tissues(Reference Davidson, Johnson and Rooney29).
The incorporation of EPA and DHA into cells and tissues, in addition to the bloodstream, is pivotal for their action and associated benefits to human health(Reference Calder11). Using esterified n-3 LCPUFA in trials in which participants undergo chronic intervention may result in varying or null findings, which could be due to consumption of the supplements on an empty stomach or in the absence of a fatty meal, such as alongside a low-fat breakfast. Therefore, an alternative formulation that requires less input from digestive processes to deliver bioactive EPA and DHA could be of great benefit. Free EPA and DHA, which would require less emulsification and no hydrolysis for effective absorption, and in situ emulsification or ‘self-emulsifying’ EE n-3 LCPUFA formulations, have been investigated and found to increase the appearance of EPA and DHA in the blood stream after single dosing alongside a low-fat meal in comparison with standard EE n-3 LCPUFA formulations(Reference Qin, Nyheim and Haram20–Reference Bremmell, Briskey and Meola22,Reference Davidson, Johnson and Rooney29,Reference Lopez-Toledano, Thorsteinsson and Daak30) . In accordance with this, we have previously shown that SMEDS EE-EPA and -DHA enhanced appearance of EPA and DHA into total plasma and FFA pools in comparison with standard EE-EPA and -DHA over 24 h following single dosing with a low-fat meal(Reference West, Kindberg and Hustvedt17). Furthermore, repeated daily dosing of SMEDS EE-EPA and -DHA over a 12-week period resulted in a greater increase in EPA and DHA in plasma, mononuclear cells and erythrocytes, with a greater increase in omega-3 index, in comparison with standard EE formulation(Reference West, Kindberg and Hustvedt17).
The present study supports the use of an alternative formulation to SMEDS-EE that enhances the availability of EPA and DHA in the absence of a fatty meal. A novel glyceride formulation of EPA + DHA was used comprising of a mixture of monoglyceride, diglyceride and TAG in a specific ratio (1:2·5:1) that improves solubilisation of the lipids and enhances emulsification of the oil in the stomach. This enhanced emulsification will allow for better access of digestive enzymes, pivotal for the release of free FA and monoglycerides. These FA may then be more readily absorbed during passage through the small intestine resulting in a more effective absorption, thereby improving the bioavailability of EPA and DHA in the absence of a fatty meal. Once absorbed, the n-3 LCPUFA may circulate in the bloodstream esterified into TAG as components of chylomicrons, VLDL released from the liver, cholesteryl-rich lipoproteins and as phospholipids as a component of the phospholipid monolayer that stabilises such lipoproteins in the aqueous environment(Reference Heath, Karpe and Milne31,Reference Heath, Karpe and Milne32) . Therefore, EPA and DHA in total plasma were measured, as this contains a combination of these forms.
The present study shows that compared with the EE-EPA + DHA formulation, the use of a glyceride-EPA + DHA formulation significantly increases EPA, DPA and DHA to a greater extent and EPA and DHA reach a maximal concentration more rapidly over a 12 h period following single dosing. It is thought that EPA and DHA derived from the gut appear in the bloodstream approximately 4 h following consumption(Reference Heath, Karpe and Milne31,Reference Heath, Karpe and Milne32) . We previously showed that the largest difference in total plasma EPA and DHA between self-emulsifying EE-EPA and -DHA (SMEDS) in comparison with standard EE-EPA and -DHA occurred at 4 h suggesting improved gastrointestinal handling of the SMEDS formulation(Reference West, Kindberg and Hustvedt17). This has been reproduced in the present study in which we show the largest difference in total plasma EPA and DHA between the glyceride formulation and EE formulation to occur at 4 h also, suggesting there is improved gastrointestinal handling of the glyceride formulation.
These results suggest that the glyceride mixture may be suitable for use in individuals with chronic gastrointestinal problems in which ‘normal’ absorption is impaired and therefore, EE formulations which are absorbed best with fat may not be appropriate or tolerated. However, this remains to be tested and further investigation into use of this formulation in such individuals would be of great interest. In addition, these results further suggest suitable use for supplementation in older adults in which changes in pancreatic enzyme secretion occur with increasing age and may result in less effective absorption(Reference Rémond, Shahar and Gille33).
The use of an n-3 glyceride formulation in trials in which participants are responsible for their own diet alongside supplementation may also be of benefit as absorption would not be dependent on the fat content of the meal consumed alongside the supplements and these could be taken on an empty stomach. Davidson et al. (Reference Davidson, Johnson and Rooney29) found that the superior absorption of a FFA formulation of EPA and DHA in comparison with EE-EPA and -DHA was negated when consumed with a fatty meal suggesting it would not matter if participants consumed the capsules alongside meals of varying fat content. Further tests would be required to conclude whether similar findings would be observed with the glyceride mixture, but this could potentially provide greater efficacy of trials using n-3 supplements. A final implication of these results is that much lower doses of EPA and DHA would be required if administered via glyceride formulation to achieve similar concentrations achieved by EE supplementation.
The present study has several strengths. First, participants were recruited with the use of the omega-3 index as a criterion; a value of ≤6·5 was required for inclusion. Second, consumption of the capsules was observed by a member of the research nursing team to ensure compliance. Third, participant retention was high; all participants completed both treatments providing all samples. These strengths provide confidence in our findings. One limitation is that we studied only single dosing; however, a previous study(Reference West, Kindberg and Hustvedt17) demonstrated that effects of a SMEDS formulation of EPA and DHA EE shown with single dosing were also seen with repeated daily dosing out to 12 week. Nevertheless, it will be important to study the new glyceride formulation with repeated daily dosing. Another limitation is that we studied only a single dose of n-3 LCPUFA. A third limitation is that we studied adults aged 50–70 years and so we cannot generalise the findings to those of other ages. The glyceride mixture was a ratio of monoglyceride, diglyceride and TAG (1:2·5:1) with an identical FA composition to that of the EE formulation; however, the distribution of EPA, DPA and DHA amongst monoglyceride, diglyceride and TAG within the glyceride mixture is not known and is therefore a final limitation of the study.
In conclusion, a glyceride formulation of EPA + DHA resulted in higher total plasma EPA and DHA over a 12 h period following single dosing and achieved a maximal concentration faster than observed with an EE formulation of EPA + DHA. Therefore, a mixture of monoglyceride, diglyceride and TAG can provide superior absorption and higher amounts of bioactive EPA and DHA than a commonly available EE formulation.
Acknowledgements
We thank Christiaan Gelauf, who assisted with the FA analysis.
This research was funded by BASF AS.
The authors’ responsibilities were as follows: P. C. C., G. M. K., S. O. H.: designed the study; H. L. F.: recruited participants, carried out the intervention, processed samples, conducted laboratory and statistical analyses, and drafted the manuscript; P. C. C.: supervised all research and finalised the manuscript; all authors: read and approved the final manuscript. The funders had no role in the conduct of the study, the analysis of the results or the decision on where to publish the findings.
G. M. K. and S. O. H. are employees of BASF AS. P. C. C. has acted as a consultant to BASF AS. H. L. F. has no conflicts to declare.
