Viscometer validation studies for routine and experimental hemorheological measurements

Abstract

BACKGROUND:

Viscosity measurement is challenging due to the internal properties of blood and the artifacts deriving from the various viscometer systems.

OBJECTIVE:

We aimed to determine the pitfalls of a cone-plate viscometer (Brookfield DV-III Ultra LV) before starting measurements and compare it to our capillary type model (Hemorex Hevimet 40). Effects of sample storage and thermal calibration were assessed as well.

METHODS AND RESULTS:

Intra-observer variability was studied by 10 replicate measurements of 7 blood samples, mean coefficients of variation were less than 5%. Instruments were compared by measuring 26 blood samples, an average difference of 7% in WBV and 10% in PV was observed. 9 blood samples were stored at 4°C, 22°C and 37°C up to 48 hours to study the effect of storage on viscosity values. WBV at 50 and 100 s^-1 became significantly lower after 3 hours at 37°C (p < 0.05). WBV at higher shear rates and PV remained constant at all temperatures. To evaluate the possibility of measuring one sample at different temperatures, 8 blood samples were measured at 40°C with the device calibrated both at 20°C and 40°C; no significant difference was observed.

CONCLUSIONS:

Thorough validation studies are required before starting experimental and routine viscosity measurements.

Keywords

Capillary rotational cone-plate storage reproducibility temperature

1 Introduction

Numerous hematological, vascular and pulmonary diseases are associated with increased whole blood (WBV) and plasma viscosity (PV) [1], both identified as primary cardiovascular risk factors [2 –9] as well as variables with prognostic significance in certain clinical conditions [10]. WBV is an important determinant of blood flow resistance, its elevated level may cause a disturbance in tissue perfusion [11, 12], which is more likely in clinical conditions associated with decreased vascular compensation mechanism, such as in severe stenosis on the basis of atherosclerosis [1]. PV is an important factor of flow resistance in the microcirculation, mediating shear stress toward the endothelium as a consequence of the axial migration of RBCs [12, 13].

However, the usefulness of viscosity measurement in various clinical cases, blood viscosity is not a routinely measured macrorheological parameter because of its troublesome implementation. Difficulties of viscosity assessment originate both from the nature of blood and the properties of the various available viscometer systems. Whole blood is a non-Newtonian fluid, its viscosity is shear rate dependent, thus one viscosity value is insufficient to characterize a sample, therefore a shear rate – viscosity profile should be attained. This profile is affected by several factors, including RBC aggregation and deformability [14]. Blood plasma is a Newtonian fluid, therefore it can be more easily measured, although in certain clinical conditions (e.g. hematological disorders) or measurement settings surface film artifacts may alter the results.

Viscosity measurements are performed ex vivo. Preparation and handling of samples from blood drawing till the measurement may change the intrinsic properties of the sample and hence the observed results may not reflect the ones in the vessels. Moreover, environmental conditions (e.g. temperature, humidity) can also affect the measuring instrument as well, leading to error.

Beyond the natural complexity of blood, several artifacts may complicate the measurements. These artifacts are generated by surface tension, plasma proteins, phase separation and RBC aggregation [14]. Bias of artifacts is more visible at low shear rates where one tries to observe low intrinsic forces. Various types of instruments have been developed to measure viscosity, though capillary and rotational viscometers are the most frequently used ones [15]. These instruments, depending on their construction, are sensitive to artifacts to different extents, moreover, device-specific artifacts may also be present. Hence different instruments aiming to measure the same parameter may produce different results. This should be kept in mind when switching to another method.

2 Objectives

Our laboratory has recently acquired a Brookfield DV-III Ultra LV programmable rotational viscometer (Brookfield Engineering Labs; Middleboro, USA). Starting working with a different type of instrument is always challenging due to the above-mentioned reasons. Before any experimental or clinical measurements can be done, several calibration and validation measurements are required. In our paper we focus on the most important steps of our process, helping others to cope with the same situation.

Our Brookfield rotational viscometer is equipped with a cone-plate system, using a CP40 spindle. 500μl sample size is required for a single measurement. The gap between the cone and plate must be precisely adjusted to achieve ideal chamber geometry. The operating temperature is maintained by an external circulating bath (TC650-MX, Brookfield Engineering Labs; Middleboro, USA). Samples can be pre-incubated in the external bath before injecting the samples into the instrument (this is important in order to achieve proper volume in the chamber).

The device can be controlled via computer (Rheocalc v3.3. Build 49-1, Brookfield Engineering Labs; Middleboro, USA) allowing the creation of automatic measuring algorithms. In this rotational viscometer the same element generates shear rate and measures shear stress simultaneously, thus after changing the rotational speed, some time is required until the torque of the coiled spring reaches a steady-state. Viscosity for a given shear rate should be measured when fluctuation of torque terminates. Identification of this time is necessary to create an automatic measuring algorithm.

In particular cases the demonstration of different temperature conditions (e.g. fever, acral hypothermia, Raynaud-syndrome) is necessary, possibly on the same sample. The thermal expansion may change the cone-plate distance, making the measurement invalid due to the alteration of chamber geometry. The thermal expansion of blood must be also taken into consideration, therefore, the effect of temperature must be studied.

Reproducibility is one of the most important factors in clinical and experimental measurements. Variances of results may originate from the instrument itself, from the ambient (e.g. dust, vibration, room temperature) and the measurer (e.g. inaccurate setting of the cone-plate distance, incorrect sample volume and sample preparation, etc.). If the device is used by more than one person, the inter-observer differences also have to be taken into consideration.

Samples are usually not measured instantly after sampling, thus the stability of the measured viscosity values also has to be checked at different time points and relevant temperatures. Samples are typically stored at room temperature if intended to be measured within working hours. If samples can not be measured, they are generally put in the refrigerator and measured on the next working day. In experimental situations incubation for longer hours at body temperature may also be necessary.

3 Subjects and methods

3.1 Subjects, blood samples

Blood was obtained by sterile antecubital venipuncture into EDTA (10.8 mg K2-EDTA / 6.0 ml) containing Vacutainer^® tubes from healthy, non-smoker volunteers between the ages of 18 and 40 in the early morning. The venipuncture was performed in accordance with the latest hemorheological guideline [14]. None of the volunteers had regularly taken any medications or had surgery within 1 month. The study was approved by the Regional Ethics Committee of the University of Pecs (Approval No.: 5336) and undertaken with the understanding and consent of each subject. The measurements were performed at native hematocrit values. Prior to each measurement, the samples were incubated for 5 minutes in the external bath (T = 37°C) and after injecting 500μl of it into the chamber, for further 3 minutes (T = 37°C) at 50 s^-1 constant shear rate.

3.2 Torque stability time

The viscosity of a Newtonian calibration fluid was measured at 25, 50, 100 and 200 s^-1 shear rates. The shear rate was rapidly increased to the desired value from zero while the measurement program obtained viscosity value in every second (the shortest interval available by the program). Stability was considered when the fluctuation of the value stopped.

3.3 Temperature effect

6 ml of blood from 8 donors (4 females, 4 males) was used to measure WBV at 50, 75, 100, 200 and 400 s^-1 shear rates. The system was cooled down to 20°C, the cone-plate distance was adjusted, sample was injected. Then the system was heated up to 40°C without the alteration of calibration or change of the sample (the sample was sheared constantly at 50 s^-1 to avoid sedimentation of RBCs), then WBV was measured again. After that the sample was removed, the chamber was cleaned, the geometry was re-set at 40°C, a new sample was injected and viscosity values were acquired.

3.4 Reproducibility

From 7 volunteers (1 female, 6 males) 30 ml of blood was drawn. 10 replicate WBV (at 50, 75, 90, 100, 150, 200, 300, 500 s^-1 shear rates) and PV (at 500 s^-1 shear rate) measurements were carried out on each sample. After each measurement, the sample was immediately replaced with a new one from the same blood pool, gently mixed before being injected into the system.

3.5 Storage

30 ml of blood from 9 donors (2 females, 7 males) was collected to perform WBV (at 50, 100, 200, 500 s^-1) and PV (500 s^-1) measurement at the following time points and temperatures: baseline (22°C), 2 hours (22°C), 3 hours (37°C), 4 hours (22°C), 6 hours (37°C), 8 hours (4 and 22°C), 24 hours (4°C) and 48 hours (4°C). Prior to each measurement, the samples – pending their testing in vertical position – were gently mixed.

3.6 Comparison

Comparison studies were carried out to compare the Brookfield DV-III Ultra LV and Hevimet 40 viscometers.

The Hevimet 40 capillary viscometer (Hemorex Ltd.; Budapest, Hungary) consists of a capillary, connected to a vertical glass tube, surrounded by high specific heat capacity oil which maintains stable 37°C temperature. 620μl of blood is injected into the system and released to flow out. Forty diodes along the vertical tube register the height of the fluid column against time. Momentary shear rate, shear stress and apparent viscosity values are calculated, values for 10–240 s^-1 shear rates are inter/extrapolated by the measurement program.

Viscometers were calibrated with the same Newtonian calibration fluid (Hemorex Ltd., viscosity: 3.56 mPa·s ± 3%, No. 130703) and were securely mounted and levelled on a vibration-free table.

12 ml of blood from 26 donors (9 females, 17 males) was drawn, WBV at 50, 100, 150, 200 s^-1 and PV at 500 s^-1 shear rate were measured within 2 hours from sampling.

3.7 Statistical procedures

Shapiro-Wilk test was used to check normality, f-test for equivalence of variation between analyzed groups. Data were compared by two independent samples t-test and paired samples t-test. A two-tailed p value below 0.05 was considered statistically significant. To assess the agreement between the viscosity data obtained by the instruments, Pearson correlation and Bland-Altman method (confidence interval: 95%) was used. Analyses were carried out on IBM\copyright SPSS\copyright Statistics v23.

4 Results

4.1 Torque stability

The device required 9 seconds at 10/s and 10 seconds at 200/s to achieve stable viscosity values (Fig. 1).

Fig.1

Torque % values to time at rapid shear rate increase at 10 and 200/s (37°C).

4.2 Temperature effect

The observed viscosity values with unchanged and re-set cone-plate distance are shown in Table 1. There were no statistically significant differences between the two setups, although at lower shear rates the device measured – not significantly – higher viscosity values if it was not set at the appropriate temperature.

Table 1
Temperature effect on viscosity values

WBV (mPas) at 40°C

Shear rate (s^-1) 50 75 100 200 400

Calibrated at 20°C 4.81 ± 1.99 4.29 ± 1.48 3.97 ± 1.21 3.45 ± 0.78 3.12 ± 0.57

Calibrated at 40°C 4.29 ± 0.63 3.92 ± 0.54 3.71 ± 0.49 3.33 ± 0.41 3.09 ± 0.37

Difference 0.52 (12.2%) 0.37 (9.3%) 0.26 (6.9%) 0.12 (3.7%) 0.03 (1.1%)

p-value 0.36 0.36 0.40 0.46 0.74

WBV (mPas) at 40°C
Shear rate (s^-1)	50	75	100	200	400
Calibrated at 20°C	4.81 ± 1.99	4.29 ± 1.48	3.97 ± 1.21	3.45 ± 0.78	3.12 ± 0.57
Calibrated at 40°C	4.29 ± 0.63	3.92 ± 0.54	3.71 ± 0.49	3.33 ± 0.41	3.09 ± 0.37
Difference	0.52 (12.2%)	0.37 (9.3%)	0.26 (6.9%)	0.12 (3.7%)	0.03 (1.1%)
p-value	0.36	0.36	0.40	0.46	0.74

Mean ± SD.

4.3 Reproducibility

Results of reproducibility studies are presented in Fig. 2. Mean CV levels were less than 5% at all shear rates. In Donor 1, 2 and 4 there was significant negative correlation between shear rate and CV values (Donor 1: –0.891; Donor 2: –0.753, Donor 4: –0.765). Mean CV level of plasma viscosity at 500 s^-1 shear rate was 2.74 ± 0.73.

Fig.2

Coefficient of variation on 10 replicate measurements of whole blood viscosity. Data are shown as a mean ± standard deviation (37°C).

4.4 Storage

The effects of storage are shown in Fig. 3 and Fig. 4. The Hct of the samples was not adjusted, thus the shown standard deviations reflect differences in hematocrit (44.3% ± 2.9%) rather than errors of measurements. WBV after 3 hours at 37°C was significantly lower at 50 and 100 s^-1 shear rates (p < 0.05). In all other cases no significant difference was observed. PV remained constant at all temperatures.

Fig.3

Effect of storage on whole blood viscosity at different temperatures (absolute changes compared to baseline). Data are shown as a mean ± standard deviation. *significant difference (p < 0.05).

Fig.4

Effect of storage on plasma viscosity at different temperatures (absolute changes compared to baseline). Data are shown as a mean ± standard deviation. *significant difference (p < 0.05).

4.5 Comparison

The results are presented in Table 2 and Fig. 5. The capillary viscometer measured around 7% higher WBV and 10% higher PV values compared to the rotational one. At lower shear rates the difference in WBV was higher. At 50/s shear rate the correlation value was 0.67, while at the higher shear rates it was above 0.8 (100/s: 0.82, 150/s: 0.84, 200/s: 0,81). Bland-Altman analysis shows the above described systematic difference, but no visible trends can be seen in connection with viscosity values (Fig. 6).

Table 2
Whole blood and plasma viscosity values measured by Hevimet 40 and Brookfield DV-III Ultra viscometers (37°C)

Shear rate (s^-1) Hevimet 40 Brookfield

Whole blood (mPas)

50 4.81 ± 0.67 4.52 ± 0.52

100 4.30 ± 0.50 3.95 ± 0.39

150 4.01 ± 0.45 3.58 ± 0.30

200 3.94 ± 0.42 3.54 ± 0.33

Plasma (mPas)

500 1.28 ± 0.12 1.14 ± 0.08

Shear rate (s^-1)	Hevimet 40	Brookfield
Whole blood (mPas)
50	4.81 ± 0.67	4.52 ± 0.52
100	4.30 ± 0.50	3.95 ± 0.39
150	4.01 ± 0.45	3.58 ± 0.30
200	3.94 ± 0.42	3.54 ± 0.33
Plasma (mPas)
500	1.28 ± 0.12	1.14 ± 0.08

Mean ± SD.

Fig.5

Whole blood viscosity values measured by Hevimet 40 and Brookfield DV-III Ultra viscometers (37°C).

Fig.6

Bland-Altman plots. Dotted lines represent 95% confidence interval. Dashed line represents mean bias.

5 Discussion

Due to the structure of the cone-plate system – as it was expected – some time is required to reach a steady torque value after a shear rate change. According to our data – depending on the magnitude of this change – around 10 seconds is required before viscosity values can be obtained. This finding should be taken into consideration when an automated measuring algorithm is created. Measuring viscosity at several shear rates will increase the measurement time significantly.

Different device temperatures had no statistically significant effect on viscosity values, presumably due to the small sample size. It has to be emphasized that this study only examined the 20–40°C temperature range (which is most likely in clinical conditions), therefore if greater temperature ranges are necessary, a new test needs to be done. Several factors can alter viscosity values during temperature changes: (1) thermal expansion of sample, (2) thermal expansion of chamber components, (3) sedimentation of RBCs during the time until the new temperature is reached. Thermal expansion can change the distance between the cone and the plate, thus altering chamber geometry. Before this experiment we cooled down the empty instrument to 10°C, set the distance and heated up to 37°C. At 37°C the micrometer adjusting ring had to be turned one scale to achieve the original distance (distance increased due to heating). According to the user’s manual, one scale division is equivalent to 0.0005 inch movement of the plate relative to the cone. Heating up the instrument from 20°C to 40°C took around 21 minutes (in the opposite direction it is about 30 minutes), after which time RBC sedimentation must be taken into consideration, which can be reduced by continuous shear. Although no significant changes were observed, it is recommended to calibrate the device after a temperature change.

PV and WBV presented in Fig. 2 demonstrate good reproducibility over a wide range of shear rates. In 3 cases (out of 7) there was a negative correlation between shear rate and CV. Originating from the instrument’s design, it is more accurate at higher shear rates or in case of higher viscosity samples (when torque value is higher). The accuracy of measurements depends on several variables, e.g. sample handling, device setting and environmental conditions (e.g. vibration, dust, temperature, humidity, etc.), thus their precise control is inevitable to conduct valid measurements.

Discrepancies were found in WBV and PV values obtained by the two viscometers. Antonova et al. reported lower PV values measured by a capillary viscometer compared to a Couette type one [17]. Wang et al. measured lower WBV and PV with a capillary viscometer compared to a Couette type viscometer. The relative difference was found greater at lower shear rates [18]. On the other hand, Marinakis et al. obtained smaller WBV values with a cone-plate viscometer compared to a capillary type one and the relative difference was greater at lower shear rates [19]. These findings suggest the WBV and PV values could vary depending on the method applied.

The discrepancies in the values may originate from the different measuring technique and the artifacts – arising from plasma proteins and RBC sedimentation – affecting the two instruments differently. In the Hevimet capillary viscometer the sample is exposed to a range of shear rates making it hard to define a single shear rate at which the particular viscosity is measured, moreover the pressure gradient and blood flow will decrease over time. From the gained data the software will define apparent viscosity values based on the Casson’s equation.

In a cone-plate system the entire sample is exposed to the same shear rate, while viscosity is directly measured. Measurements in a cone-plate geometry can be affected by surface film artifacts and RBC sedimentation. The plasma proteins are surfactants and form a semi-rigid protein layer at fluid-air interfaces. The extra force from this layer can transmit significant torque and cause false high viscosity values. At higher shear rates – where shear stress is high – surface film tension causes negligible extra force, but at low shear rates this tension force becomes more prominent compared to the total shear stress. In clinical conditions with elevated plasma proteins (e.g. paraproteinemia) surface tension artifact might have a greater impact on the measurements. The sedimentation of RBCs also has a high influence on WBV measurements: at very low (several 1/s) shear rates the settling away of RBCs from the upper surface in a cone-pale system causes false viscosity readings. This phenomenon can be avoided by continuous shear during incubation period and pre-shear before each measurement.

WBV and PV were stable at room temperature up to 8 hours and at 4°C up to 48 hours allowing to postpone the measurements. The significant difference in WBV was detected at 37°C after 3 hours which result meets a previous finding [16]. This has to be taken into consideration when designing studies with long incubation times.

6 Conclusions

Installing a new device is always very challenging, beneath the general aspects, device specific problems have to be addressed. Our results indicate that the rotational viscometer has a good reproducibility. The torque stability – depending on the magnitude of shear rate change – requires around 10 seconds, which needs to be taken into consideration. The device is able to measure accurate viscosity between clinically relevant temperatures and measures slightly lower values compared to our tested instrument. Samples can be stored up to 48 hours without affecting measured values, but storage at 37°C is not recommended for several hours.

Footnotes

Acknowledgments

The present scientific contribution is dedicated to the 650th anniversary of the foundation of the University of Pecs, Hungary.

References

Toth

, Szukits

, Varady

, Sandor

, Rabai

, Papp

, et al. Hemorheological parameters in coronary artery disease detected by multi-slice CT. Korea-Aust Rheol J. 2014;26(2):229–35. doi:10.1007/s13367-014-0026-0.

Koenig

, Sund

, Filipiak

, Lowel

, Ernst

. Plasma viscosity and the risk of coronary heart disease: Results from the MONICA-Augsburg Cohort Study, 1984 to 1992. Arterioscler Thromb Vasc Biol. 1998;18, 768–72.

Kannel

, D’Agostino

, Belanger

. Fibrinogen, cigarette smoking, and risk of cardiovascular disease: Insights from the Framingham Study. Am Heart J. 1987;113, 1006–10.

Lee

, Mowbray

, Lowe

, Rumley

, Fowkes

, Allan

. Blood viscosity and elevated carotid intima-media thickness in men and women: The Edinburgh Artery Study. Circulation. 1998;97, 1467–73.

Lowe

, Fowkes

, Dawes

, Donnan

, Lennie

, Housley

. Blood viscosity, fibrinogen, and activation of coagulation and leukocytes in peripheral arterial disease and the normal population in the Edinburgh Artery Study. Circulation. 1993;87, 1915–20.

Koenig

, Sund

, Filipiak

, Doring

, Lowel

, Ernst

. Plasma viscosity and the risk of coronary heart disease: Results from the MONICA-Augsburg Cohort Study, 1984 to 1992. Arterioscler Thromb Vasc Biol. 1998;18, 768–72.

, Hennekens

, Ridker

, Stampfer

. A prospective study of fibrinogen and risk of myocardial infarction in the Physicians’ Health Study. J Am Coll Cardiol. 1999;33, 1347–52.

Sweetnam

, Thomas

, Yarnell

JWG

, Beswick

, Baker

, Elwood

. Fibrinogen, viscosity and the 10-year incidence of ischemic heart disease: The Caerphilly and Speedwell Studies. Eur Heart J. 1997;17, 1814–20.

Yarnell

, Baker

, Sweetnam

, Bainton

, O’Brien

, Whitehead

, et al. Fibrinogen, viscosity, and white blood cell count are major risk factors for ischemic heart disease. The Caerphilly and Speedwell collaborative heart disease studies. Circulation. 1991;83, 836–44.

10.

Toth

, Kesmarky

. Clinical Significance of Hemorheological Alterations. In: Baskurt

, Hardeman

, Rampling

, Meiselman

, editors. Handbook of Hemorheology and Hemodynamics. Amsterdam: IOS Press 2007. p. 392–432 .

11.

Baskurt

, Levi

, Caglayan

, Dikmenoglu

, Ucer

, Guner

, et al. The role of hemorheological factors in the coronary circulation. Clin Hemorheol. 1991;(11):121–7.

12.

Baskurt

, Meiselman

. In Vivo Hemorheology. In: Baskurt

, Hardeman

, Rampling

, Meiselman

, editors. Handbook of Hemorheology and Hemodynamics. Amsterdam: IOS Press; 2007. p. 322–38 .

13.

Cabrales

, Tsai

. Plasma viscosity regulates systemic and microvascular perfusion during acute extreme anemic conditions. Am J Physiol Heart Circ Physiol. 2006;(291): H2445–52.

14.

Chien

. Clinical rheology in cardiovascular diease. Bibl Anat. 1977;(16): 472–4.

15.

Hardeman

, Goedhart

, Shin

. Methods in Hemorheology. In: Baskurt

, Hardeman

, Rampling

, Meiselman

, editors. Amsterdam: IOS Press; 2007. p. 322–38 .

16.

Alexy

, Wenby

, Pais

, Goldstein

, Hogenauer

, Meiselman

. An automated tube-type blood viscometer: Validation studies. Biorheology. 2005;(42): 237–47.

17.

Antonova

, Velcheva

, Boeva

, Marinova

Tsv

. Comparative plasma viscosity measurements in healthy subjects. Clin Hemorheol Microcirc. 1995;(15):45–52.

18.

Wang

, Liao

, Stoltz

. A new simple cone-plate viscometer for hemorheology. Clin Hemorheol Microcirc. 1998;(19):25–31.

19.

Marinakis

, Barbenel

, Tsangaris

. A new capillary viscometer for small samples of whole blood. Proc Inst Mech Eng H. 2002;(16):385–92.