Lifestyle behaviours are not associated with haemolysis: results from Donor InSight

Abstract

Background

Lifestyle behaviours such as physical activity, sedentary behaviour and dietary habits have been shown to influence blood lipid levels, and both lifestyle and blood lipids may be associated with haemolysis during storage of blood products. We aimed to investigate whether lifestyle behaviours are associated with degree of haemolysis in red cell concentrates (RCC), and if such associations are mediated by low-density lipoprotein (LDL) cholesterol and triglyceride levels.

Materials and methods

Cross-sectional analyses were performed in data from 760 Dutch blood donors participating in Donor InSight, an observational cohort study. Linear regression analyses were conducted to assess associations of lifestyle behaviours with haemolysis levels in RCC 28 days after blood sampling. Lifestyle behaviours included moderate-to-vigorous physical activity and sedentary behaviour measured by accelerometry, and self-reported intake of a selection of foods potentially related to blood lipids, i.e. consumption of eggs, meat, nuts and fish. Potential mediating roles of both LDL cholesterol and triglyceride levels were investigated separately. All analyses were adjusted for relevant confounders.

Results

No statistically significant nor substantial associations of any of the lifestyle behaviours with haemolysis in RCC were found, nor were there any associations between lifestyle behaviours and blood lipids. We did find consistent positive associations of LDL cholesterol and triglyceride levels with haemolysis in RCC during storage.

Discussion

In this large cohort, blood lipid levels were consistently associated with haemolysis in RCC. Nonetheless, there was no evidence for an association between lifestyle behaviours and haemolysis in RCC, or for mediating effects by blood lipid levels.

INTRODUCTION

Blood supply organisations depend on donors who can give blood or plasma that meet quality requirements. An important criterion for blood quality is that plasma must have a clear to slightly turbid appearance before freezing and high blood lipid levels are the primary cause of non-conformity^1,2. Another quality feature of blood products relates to haemolysis levels. Different manufacturing methods of blood components lead to differences in free haemoglobin levels in plasma units and red cell concentrates (RCC) immediately after processing. For RCCs, the degree of haemolysis gradually increases during cold storage. For red blood cells, European guidelines allow a maximum of 0.8% haemolysis at the end of the storage period². Haemolysis leads to free haemoglobin, which might be toxic for recipients^3,4. Furthermore, the haemolysed cells must also be cleared from circulation, which is not desirable in critically ill patients. The debate continues about whether longer-stored blood negatively affects clinical outcomes^5–8. Whole blood and plasma donations that are classified as lipaemic or haemolytic are discarded. Visits of donors whose blood products are rejected or of low quality represent a significant burden for blood supply organisations. Knowledge on factors underlying these undesirable outcomes is the first step towards diminishing them.

Donor characteristics including sex, age, the presence of subclinical diseases and lifestyle behaviours such as smoking have been associated with increased levels of haemolysis during storage in RCC^9–12. Exercise-induced intra-vascular haemolysis has been reported in literature but this was mainly in athletes, and in particular, in endurance athletes such as long distance runners^13,14. In addition, lipaemic plasma is associated with increased haemolysis in RCC¹, even when RCC are resuspended in an additive solution with a low remaining plasma content. Plasma lipid levels are partly influenced by lifestyle behaviours. Moderate-to-vigorous physical activity (MVPA) is associated with higher high-density lipoprotein (HDL) cholesterol and a decline in levels of low-density lipoprotein (LDL) cholesterol and triglycerides^15,16. Sedentary behaviour (i.e. any waking activity characterised by an energy expenditure ≤1.5 metabolic equivalents [MET] and a sitting or reclining posture¹⁷) is associated with higher LDL cholesterol and triglyceride levels and lower HDL cholesterol levels^18,19. High dietary intakes of saturated fat and cholesterol are associated with more unfavourable blood lipid levels, whereas foods high in unsaturated fat, and especially omega-3 fatty acids, are associated with more favourable blood lipid levels^20–23. Hence, lifestyle behaviours of donors may be relevant determinants of rejection of blood products due to high blood lipid contents or high levels of haemolysis during storage.

As lifestyle behaviours are associated with blood lipid levels and blood lipid levels with haemolysis, we aimed to test the hypothesis that healthy lifestyle behaviours (i.e. higher levels of physical activity, lower levels of sedentary behaviour, more consumption of foods high in unsaturated and omega-3 fatty acids, and less consumption of foods high in saturated fat) are inversely associated with haemolysis in RCC 28 days after blood sampling. We further tested the hypothesis that the associations between lifestyle behaviours and haemolysis in RCC are partly mediated by blood lipids.

MATERIALS AND METHODS

Study-design and study population

The current study was a cross-sectional analysis of data from Donor InSight (DIS)-III, the second follow up of an observational cohort study of whole blood and plasma donors in the Netherlands; DIS-I started in 2007–2009²⁴. Donor eligibility criteria at Sanquin (the only organisation authorised to collect and supply blood products in the Netherlands) include a minimum age of 18 years and eligibility to donate according to several criteria, as assessed using a donor health questionnaire (DHQ) before each donation. A total of 6,140 donors who participated in DIS-I and/or DIS-II were invited to participate in DIS-III between April 2015 and December 2016. Of these, 2,551 participants completed the general questionnaire and provided a blood sample. To objectively measure physical activity and sedentary behaviour, 1,944 DIS-III participants were invited to also wear an accelerometer for seven consecutive days; 760 provided complete accelerometer data. This study on lifestyle behaviours and haemolysis included participants who had accelerometer data. The Medical Ethical Committee in the Amsterdam Academic Medical Center approved DIS-III and all participants gave their written informed consent.

Measures

Haemolysis and blood lipid levels

Non-fasting whole blood samples were collected in a 2 mL ethylene-diamine-tetra-acetic acid (EDTA) and a 3 mL lithium heparin tube from the diversion pouch. This pouch collects the first 20–30 mL of a donation and is routinely used for screening and blood typing purposes. For DIS-III, a venepuncture was performed if no donation was provided.

The primary outcome variable was haemolysis level 28 days after collection of the blood sample, expressed as the percentage of free haemoglobin of the total haemoglobin present in the red blood cells after correction for haematocrit. Potentially mediating variables were LDL cholesterol and triglyceride levels.

Preparation of miniature red cell concentrates and plasma samples

We used a model system to study haemolysis in miniature RCC, reflecting the degree of haemolysis in the RCC prepared by standard Dutch blood bank procedures from the corresponding whole blood unit.

As our standard RCCs are resuspended in saline adenine glucose mannitol (SAGM) and have only a low amount of residual plasma, the conditions to produce miniature RCC were selected to reflect the composition of our standard RCCs. The miniature RCC differ with respect to being leucoreduced and not leucodepleted as only buffy coat was removed, without any additional filtration step, and the storage container, was not a polyvinylchloride di(2-ethylhexyl)phthalate (PVC-DEHP) container, but an Eppendorf cup (VWR International, Amsterdam, the Netherlands). Increased haemolysis has previously been found in blood bags without phthalate-based plasticisers, and the Eppendorf cups used in this study have no plasticisers at all^25–27. Leucocyte reduction in stored RCCs has a favourable effect on haemolysis due to the lower accumulation of cytokines and reduced release of enzymes by leucocyte^28,29. Due to these negative storage effects, haemolysis was measured after 28 days of storage instead of after the conventional 35 days. Post-hoc analyses were conducted to gain better insight into the actual agreement of haemolysis between these two storage methods. The miniature RCC haemolysis levels were 3.42 time higher than those of standard blood units. Furthermore, exponentially plotting the standard blood unit measurements indicated haemolysis, as measured in 28-day stored miniature RCCs, resembled that of a 63-day stored standard whole blood unit (see Online Supplementary Content and Figure S1)³⁰. Total haemoglobin was measured from the 2 mL EDTA tube with a haematology analyser (XT 2000T, Sysmex, Kobe, Japan) after homogenisation. To produce the miniaturised RCC, the sample was centrifuged for five minutes at 2,000 g; 600 μL of the lower erythrocyte pellet was diluted with 400 μL SAGM to a haematocrit of approximately 60% and stored in 1.5 mL Eppendorf cups at 2–6°C for 28 days (under similar conditions as standard RCC). At day 28 of storage, the full blood count was repeated, followed by centrifuging the Eppendorf cups for 5 minutes at 18,000 g. The supernatant was transferred to a clean Eppendorf cup and centrifuged again (18,000 g, 5 minutes). For measurement of free haemoglobin, 50 μL were pipetted into a 96-well plate and supplemented with 200 μL of distilled water and homogenised. Free haemoglobin of the RCC was determined at 415 nm by a spectrophotometer (EON^™ plate reader, BioTek^™, Winooski, VT, USA).

Total cholesterol (TC), HDL cholesterol and triglycerides (TG) were determined by enzymatic colourimetric methods using the plasma from the lithium heparin tubes (Cobas C, Roche/Hitachi, Basel, Switzerland). LDL cholesterol levels were calculated using the Friedewald formula:

$$\text{TC\hspace{0.17em}cholesterol}-\text{HDL\hspace{0.17em}cholesterol}-(\text{TG}/2.2)$$³¹.

Lifestyle behaviours

Physical activity was operationalised as mean minutes per day of MVPA (≥3 MET). Sedentary behaviour (≤1.5 MET) was also expressed in mean minutes per day. Both MVPA and sedentary behaviour were measured by means of accelerometers (wGT3X-BT and GT3X Actigraph, Pensacola, FL, USA). Troiano Adult (2008) cut-off points for MVPA and sedentary behaviour were used³². To calculate mean minutes per day per category, the total number of minutes were divided by the number of valid days. A day was considered valid if the accelerometer was worn for at least ten hours, for a minimum of four valid days³³. A date as close to the blood donation date as possible was sought. Dietary behaviour was estimated using items of a short food frequency questionnaire (FFQ ). The FFQ used for this study was originally designed to assess dietary iron intake, but covered a number of foods high in saturated fat and/or cholesterol or high in unsaturated and omega-3 fatty acids for which earlier research had established a relation with blood lipids. These foods were fish and nuts (rich sources of unsaturated and omega-3 fatty acids), for which an inverse association with LDL cholesterol and triglycerides had been found^34,35, and meat and eggs (rich sources of saturated fat [meat] and dietary cholesterol [both meat and eggs]) for which earlier research had found a positive association with LDL cholesterol and triglycerides³⁶. All food items, questions and answer categories are presented in the Online Supplementary Content and Table SI.

Co-variates

The following self-reported variables were tested as potential confounders in the analyses: smoking status (yes/no), sex, age, and the use of lipid-modifying medication. Medication was classified according to the World Health Organization recommended Anatomical Therapeutic Chemical classification (ATC) system. All drugs with ATC code “C10 lipid modifying agents” were considered lipid-modifying medication.

Statistical analysis

Descriptive statistics are presented as mean ± standard deviation (SD) or, in the case of a skewed distribution, as median and interquartile range (IQR).

Missing data

Missing data were assumed to be missing at random. Item non-response ranged from 0.1% (lipid-modifying medication) to 6.2% (smoking) with 91% of the participants with complete data; therefore, multiple imputation was performed on an item-score level using Predictive Mean Matching³⁷. All missing data in variables that were used for analyses were imputed. A total of ten imputed datasets were generated as recommended by White et al.³⁷. Alcohol use was not included in the analysis and therefore not imputed, but did serve as a predictor of the missing data. Time spent in sedentary, light, moderate and vigorous activity had no missing data, but were also used in the imputation model to predict the missing values.

Mediation analyses

To assess whether lifestyle behaviours were associated with haemolysis four weeks after blood sampling, and whether blood lipid levels mediated these associations, mediation analyses (with linear regression) were made using the framework of Preacher and Hayes and the work of Baron and Kenny^38,39. A graphical representation of the mediation design is provided in Figure 1. Panel A shows the association of X (lifestyle behaviours) with Y (haemolysis) and is marked with the letter c. Panel B shows three pathways: the c′-pathway indicates the association of X on Y after adjusting for M (blood lipid levels). Consistent with the definition of Baron and Kenny³⁹, M is considered to be a mediator if X significantly predicts Y (c-pathway), and X significantly predicts M (a-pathway), while M significantly predicts Y adjusting for X (b-pathway). It was hypothesised that regression coefficients of the lifestyle behaviours and haemolysis (X–Y) would diminish after M was added to the model. In the present analyses, we performed mediation analyses with three lifestyle behaviours (physical activity, sedentary behaviour and food items; X) and haemolysis (Y) separately from each other mediated by LDL cholesterol and triglycerides (M). Because of collinearity between LDL cholesterol and triglyceride levels, LDL cholesterol and triglycerides have been analysed separately⁴⁰. Finally, a mediation analysis with all lifestyle behaviours in one model was performed to test whether the lifestyle behaviours were independently associated with haemolysis.

Figure 1.

Mediation analysis framework

(A) Association of X on Y. (B) X is associated by Y through M.

To increase readability of the analyses, regression coefficients for associations with haemolysis were expressed in 100ths of a percent. A covariate was considered a confounder if the regression coefficient of the determinant changed with >10%. p<0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 23.0 (SPSS Statistics, IBM, Armonk, NY, USA).

RESULTS

A total of 1,269 (65%) of the donors invited to wear an accelerometer were interested in taking part in this part of the study. Of these, 800 randomly chosen donors were contacted and asked to participate and they received an accelerometer by post. Complete accelerometer data was provided by 760 participants; an overview of reasons for not providing data is shown in Figure 2. Table I provides characteristics of the study sample. The majority of the participants were female (n=411, 54%) and active donors, meaning that they were registered as available to receive invitations to give a donation (n=571, 75%). Median haemolysis level in miniature RCC 28 days after blood sampling was 1.26% (IQR: 1.00–1.62%).

Figure 2.

Flow chart of accelerometer study

DIS: Donor InSight Study.

Table I.

Participants' characteristics

Participants	760
Male	349 (46)
Age (years)	50.6±13.1
Donor status (activea)	571 (75)
PA - light (minutes per day)	309 (259–369)
PA - moderate (minutes per day)	27 (18–41)
PA - vigorous (minutes per day)	0 (0–2)
MVPA (minutes per day)	29 (19–45)
Sedentary time (minutes per day)	550 (491–600)
Current smoker	62 (9)
Alcohol consumption
None	89 (12.6)
< once a per week	166 (23.5)
1–2 days a week	188 (26.6)
3–5 days a week	153 (21.6)
Almost every day	111 (15.7)
Haemolysisb (%)	1.26 (1.00–1.62)
LDL cholesterol (mmol/L)	2.92±0.84
Triglycerides (mmol/L)	1.27 (0.93–1.74)

Values are expressed as number, N (%), mean ± standard deviation, or median (interquartile range).

^aDonors who could be invited to donate according to the blood bank information system, eProgesa;

^bhaemolysis in miniature red cell concentrates 28 days after blood sampling.

PA: physical activity; MVPA: moderate to vigorous physical activity; LDL: low-density lipoprotein.

Lifestyle behaviours and haemolysis in RCC

The results obtained with linear regression analyses are presented in Table II for the model with LDL cholesterol and triglycerides as potential mediators. Analyses were adjusted for the following confounders: age, sex, smoking, and lipid-modifying medication. The hypothesised inverse associations of lifestyle behaviours and haemolysis in RCC were found for MVPA, fish and nut consumption; however, these associations were not statistically significant. Meat consumption was positively but not statistically significantly associated with haemolysis in RCC. The expected positive associations of the other lifestyle behaviours considered unhealthy and haemolysis in RCC were not found, and these were also not statistically significant. The results of the models with all lifestyle behaviours showed similar results.

Table II.

Mediation analysis^a: associations of lifestyle behaviours with haemolysis in red cell concentrates

	LDL mediation model
	c-pathway		c'-pathway		a-pathway		b-pathway
	Lifestyle behaviour-haemolysisa		Lifestyle behaviour-haemolysisa		Lifestyle behaviour-LDL		LDL-haemolysisa
	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)
MVPA	−1.78	(−3.66–0.10)	−1.77	(−3.65–0.11)	0.02	(0.00–0.04)	7.95	(2.90–13.00)
SB	−0.10	(−0.61–0.41)	−0.13	(−0.64–0.37)	0.01	(0.00–0.01)	7.78	(2.58–2.90)
Meat and egg consumption
Eggs	−16.50	(−43.74–10.74)	−14.20	(−41.44–13.04)	−0.17	(−0.56–0.22)	7.53	(2.46–12.60)
Meat	2.60	(−5.63–10.83)	1.80	(−6.43–10.05)	0.11	(0.00–0.22)	7.53	(2.46–12.60)
Fish and nuts consumption
Fish	−24.79	(−54.41–4.83)	−24.19	(−53.43–5.05)	0.07	(−0.33–0.47)	7.79	(2.74–12.84)
Nuts	−10.33	(−38.24–17.58)	−8.62	(−36.43–19.19)	−0.03	(−0.42–0.36)	7.79	(2.74–12.84)
TG mediation model
	c-pathway-adjusted		c'-pathway		a-pathway		b-pathway
	Lifestyle behaviour-haemolysisa		Lifestyle behaviour-haemolysisa		Lifestyle behaviour-TG		TG – haemolysisa
	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)
MVPA	−1.78	(−3.66–0.10)	−1.45	(−3.29–0.39)	−0.01	(−0.03–0.02)	17.69	(12.39–23.00)
SB	0.10	(−0.61–0.41)	−0.10	(−0.59–0.39)	0.00	(0.00–0.01)	17.82	(12.50–23.14)
Meat and egg consumption
Eggs	−16.50	(−43.74–10.74)	−14.36	(−40.82–12.12)	−0.08	(−0.43–0.27)	17.75	(2.74–12.37)
Meat	2.60	(−5.63–10.83)	0.12	(−8.01–8.24)	0.21	(0.10–0.31)	17.75	(2.74–12.37)
Fish and nuts consumption
Fish	−24.79	(−54.41–4.83)	−23.37	(−52.22–5.48)	0.07	(−0.33–0.47)	17.71	(12.39–23.03)
Nuts	−10.33	(−38.24–17.58)	−3.21	(−30.56–24.13)	−0.03	(−0.42–0.36)	17.71	(12.39–23.03)

^aAll models are adjusted for age, sex, smoking and lipid modifying medication. To increase readability of the analyses regression, coefficients for associations with haemolysis were expressed in 100ths of a percent. Food items are presented per 100 grams per day, MVPA and SB are presented per 10 minutes.

MVPA: moderate-to vigorous-physical activity; SB: sedentary behaviour; LDL: low density lipoprotein cholesterol; TG: triglycerides.

Lifestyle behaviours and haemolysis with potential mediators

The expected deminishment in the regression coefficients for associations of lifestyle behaviours and haemolysis in RCC after incorporating LDL cholesterol as potential mediator was only found for the unhealthy food items and fish consumption (Table II). The regression coefficient of egg consumption decreased marginally from β= −16.50 (95% CI: −43.74–10.74) to β= −14.20 (95% CI: −41.44–13.04). All associations were statistically insignificant. A similar pattern was found with triglycerides as potential mediator (Table II). The model with all lifestyle behaviours showed the same results, regardless of the potential mediator (Table III).

Table III.

Mediation analysis^a: associations of all lifestyle behaviours with haemolysis in red cell concentrates

LDL mediation model
All lifestyle behaviours	c-pathway-adjusted		c'-pathway		a-pathway		b-pathway
	Lifestyle behaviour-haemolysisa		Lifestyle behaviour-haemolysisa		Lifestyle behaviour-LDL		LDL-haemolysisa
	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)
MVPA	−1.93	(−3.91–0.05)	−1.97	(−3.95–0.00)	0.02	(−0.01–0.05)	7.93	(2.86–13.01)
SB	−0.25	(−0.78–0.29)	−0.29	(−0.82–0.25)	0.01	(0.00–0.01)	7.93	(2.86–13.01)
Meat and egg consumption
Eggs	−12.27	(−40.65–16.11)	−1.02	(−38.52–18.08)	−0.18	(−0.57–0.20)	7.93	(2.86–13.01)
Meat	1.29	(−7.05–9.62)	0.04	(−7.93–8.67)	0.12	(0.01–0.23)	7.93	(2.86–13.01)
Fish and nuts consumption
Fish	−20.02	(−50.68–10.63)	−2.03	(−50.68–10.00)	0.16	(−0.25–0.58)	7.93	(2.86–13.01)
Nuts	−8.47	(−36.57–19.62)	−0.65	(−34.54–21.44)	−0.07	(−0.46–0.32)	7.93	(2.86–13.01)
Triglycerides mediation model
All lifestyle behaviours	c-pathway-adjusted		c'-pathway		a-pathway		b-pathway
	Lifestyle behaviour-haemolysisa		Lifestyle behaviour-haemolysisa		Lifestyle behaviour-LDL		LDL-haemolysisa
	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)	Beta	(95% CI)
MVPA	−1.93	(−3.91–0.05)	−1.63	(−3.56–0.31)	0.00	(−0.03–0.02)	18.59	(13.41–23.76)
SB	−0.25	(−0.78–0.29)	−0.25	(−0.77–0.28)	0.00	(0.00–0.01)	18.59	(13.41–23.76)
Meat and egg consumption
Eggs	−12.27	(−40.65–16.11)	−10.63	(−38.25–17.00)	−0.07	(−0.43–0.30)	18.59	(13.41–23.76)
Meat	1.29	(−7.05–9.62)	−1.10	(−9.21–7.09)	0.21	(0.10–0.31)	18.59	(13.41–23.76)
Fish and nuts consumption
Fish	−20.02	(−50.68–10.63)	−20.34	(−50.26–9.59)	0.13	(−0.26–0.52)	18.59	(13.41–23.76)
Nuts	−8.47	(−36.57–19.62)	−1.45	(−28.99–26.09)	−0.34	(−0.70–0.03)	18.59	(13.41–23.76)

^aAll models are adjusted for age, sex, smoking and lipid modifying medication.

To increase readability of the analysis regression coefficients for associations with haemolysis were expressed in 100ths of a percent. Food items are presented per 100 grams per day, MVPA and SB are presented per 10 minutes. MVPA: moderate-to vigorous-physical activity; SB: sedentary behaviour; LDL: low density lipoprotein cholesterol; TG: triglycerides.

Lifestyle behaviours and blood lipids

We found meat consumption was statistically significantly associated with LDL cholesterol and triglyceride levels, β=0.11 (95% CI: 0.00–0.22) and β=0.21 (95% CI: 0.10–0.31) respectively, indicating that an increase of 100 g in meat intake is associated with an increase of 0.11 mmol/L in LDL cholesterol. The expected inverse association of healthy lifestyle behaviours and blood lipid levels was only found for fish consumption; however, this was not statistically significant. Non-significant positive associations were found for haemolysis in RCC with time spent in sedentary behaviour and with meat consumption.

Blood lipids and haemolysis

Higher LDL cholesterol levels were significantly associated with higher levels of haemolysis in RCC after adjusting for sedentary behaviour (β=7.78, 95% CI: 2.58–2.90). The regression co-efficients for the association of LDL cholesterol and haemolysis in RCC changed slightly when adjustments for lifestyle behaviours were made, but remained statistically significant. Effect sizes found for the association of triglyceride levels and haemolysis in RCC (β=17.75, 95% CI: 12.50–23.14) were higher than the effect sizes for the association of LDL cholesterol and haemolysis in RCC. This indicates that a 1 mmol/L higher triglyceride level is associated with 1.78% higher haemolysis (as previously stated, analyses of associations with haemolysis were expressed as 100ths of a percent to increase the readability). Table III shows that, in the model with all lifestyle behaviours and haemolysis, LDL cholesterol was also significantly associated with haemolysis in RCC (β=7.93, 95% CI: 2.86–13.01).

DISCUSSION

In this study, no evidence was found for associations between measured lifestyle behaviours and haemolysis levels in RCC during storage. Lifestyle behaviours were not associated with blood lipid levels and no statistically significant associations of lifestyle behaviours with haemolysis in RCC were found with blood lipids as mediators. We did find that both LDL cholesterol and triglycerides were significantly associated with haemolysis levels in RCC during storage. As according to the criteria of Baron and Kenny³⁹ significance of all previously mentioned associations is required for a variable to be a mediator, we did not find evidence that LDL cholesterol or triglycerides were mediators in the association of lifestyle behaviours on haemolysis in RCC during storage.

We hypothesised that physically active donors would provide red blood cells with lower levels of haemolysis during storage, mainly focusing on the indirect effect of physical activity on haemolysis in RCC through blood lipid levels. In contrast to earlier findings on physical activity and blood lipids in men, in a systematic review of randomised controlled trials on exercise and blood lipids by Kelley and Kelley¹⁵, donors who spent more minutes in MVPA did not have lower levels of LDL cholesterol and triglycerides; associations were inconsistent and non-significant (−0.01 and 0.02 mmol/L respectively for triglycerides and LDL cholesterol)¹⁵. This discrepancy between the present study and the systematic review by Kelley and Kelley¹⁵ might be due to differences between study populations.

Mean triglycerides and LDL cholesterol at baseline were 1.5 and 3.7 mmol/L, respectively, in studies reported in the systematic review. In the present study, blood lipid levels were lower (median triglycerides 1.3 mmol/L, mean LDL cholesterol 2.9 mmol/L); even the donors who were less physically active had relatively low blood lipid levels. The target populations of the randomised controlled trials included in the review by Kelley and Kelley¹⁵ more often consisted of participants with an already increased (cardiovascular) disease risk or even diagnosed diseases. Due to eligibility screening and self-selection, donors are generally a “healthier” subset of the general population^41,42. This might have caused a lack of variation in blood lipid levels, as these are well within the normal range. The same accounts for physical activity; an absolute increase of 10 minutes MVPA in a person who is already engaged in the recommended 150 minutes of MVPA per week probably does not lower blood lipid levels as much as the 10 minutes more MVPA in a person who is not physically active⁴³. The median time spent in MVPA was 203 minutes per week in our study, which is significantly higher than the (updated) recommendation of 150 minutes per week⁴⁴. Another Dutch population-based cohort study reported similar levels of objectively measured physical activity, namely an average of 202 minutes MVPA per week⁴⁵.

Both triglycerides and LDL cholesterol were significantly associated with haemolysis levels in RCC. As residual plasma in RCC can have a high impact on haemolysis levels, the design of the haemolysis model was set up to have comparable, low residual plasma levels as in our standard RCC. Because the amount of residual plasma differs between studies with RCC produced by different methods, different associations between donor blood lipids and haemolysis can be expected in other studies. Our findings are in line with previous research by Bashir et al., where haemolysis levels for red cells in lipaemic and non-lipaemic plasma were compared after 24 and 48 hours in different storage conditions⁴⁶. In this study, the associations were stronger, which could be explained by storage of RCCs in 100% plasma, whereas in our study the percentage plasma was approximately 15%. De Korte et al. also reported a major difference in haemolysis level after 35 days between lipaemic and non-lipaemic donations, with more haemolysis in the lipaemic donations^47,48. Triglyceride levels were measured and were more than three times higher in the lipaemic donations as compared to the non-lipaemic donations. Despite the limited range in blood lipids and haemolysis levels, and the restricted number of donors studied, the associations of lipids with haemolysis were strong and consistently significant.

The degree of haemolysis in the miniaturised RCCs was high with a median of 1.26%, as compared with the European guidelines that state a maximum haemolysis of 0.8% in blood bags. As explained previously, our miniaturised RCC samples were non-leucodepleted and were stored in Eppendorf cups, which both have negative effects on storage quality conditions^25–27. Although the storage quality in Eppendorf cups is obviously lower, an agreement is seen between measurements obtained from both storage conditions (the miniature RCC haemolysis is 3.42 times higher than the standard whole blood unit haemolysis) (see Online Supplementary Content and Figure S2).

To our knowledge, this is the first study on lifestyle behaviours and haemolysis in RCC in a large cohort of donors. The key strength of this study lies in the thorough assessment of physical activity and sedentary behaviour of donors, using accelerometry rather than less reliable questionnaire data⁴⁹. In addition, the sample size with regard to the haemolysis measurements is a strong point of the study; previous laboratory studies that investigated associations of lipid levels with haemolysis in RCC used smaller numbers^46,48. Another asset was the high willingness (65%) of DIS-III participants to participate in the accelerometer study.

There are also a number of limitations. One drawback of our study is that we analysed dietary behaviour using four items of a questionnaire that was not designed originally to estimate blood lipid intake and absorption. We chose these four items based on strong evidence of epidemiological studies³⁴. The questionnaire aimed to assess average intake; however, it would have been informative to also know how much foods high in saturated lipids and cholesterol were consumed prior to donation⁵⁰. The lack of significant associations might be a consequence of the way dietary behaviour was estimated in this study. Furthermore, although haemolysis levels were quite high in this study, the variation was relatively small. This is likely due to the selection of participants of the observational DIS cohort. Selection for this study was not based on previous haemolysis levels as in another study¹², which could be a reason for less variation but may better reflect haemolysis levels in a general donor population.

The findings of this study suggest that the lifestyle behaviours of donors investigated are not associated with haemolysis levels during storage of RCC, and these lifestyle behaviours thus do not appear to represent a cause for any major concern in the present population of Dutch blood donors. However, because blood lipid levels were consistently associated with haemolysis in RCC, evaluating blood lipid levels may be helpful in selecting donors in the future in case mean haemolysis levels rise in the blood product pool. This would prevent outliers if phthalate-plasticised blood collection bags are prohibited.

CONCLUSION

In this population of Dutch blood donors, lifestyle behaviours are not associated with haemolysis levels in miniature RCC 28 days after blood collection. We also did not find evidence of mediating effects of LDL cholesterol or triglyceride levels. Nonetheless, both LDL cholesterol and triglyceride were strongly associated with haemolysis levels in RCC which warrants replication and further exploration into potential implications.