1
Pregnancy and the rise of the CDT levels, do we need to update
the cut-off values?
Versie: Eerste versie /eindversie (doorhalen wat niet van toepassing is)
Naam student: Laurens Haverkate studentnummer: 11338636
Opdracht, versie: Scriptie Inleverdatum: 23-10-2020 Aantal woorden: 5502
2
Pregnancy and the rise of the CDT levels, do we need to update the cut-off values? Abstract
Extensive consumption of alcohol during pregnancy can lead to severe complications for the unborn child. Therefore an objective unbiased screening method is necessary to detect excessive intake of alcohol during pregnancy. Carbohydrate-deficient Transferrin (CDT) levels in serum have become a common biomarker for chronic alcohol abuse. However, several studies showed that CDT levels might be elevated during pregnancy, for reasons unrelated to alcohol intake. The aim of this study is to investigate the changes in CDT values during pregnancy and to determine accurate, trimester dependent reference values. To achieve this, 439 serum samples of 147 pregnant women that were followed during pregnancy and post-partum were analysed by the high-performance liquid chromatography (HPLC) and N-Latex immunophelometric assay. From these measurements, new reference values were calculated. This study shows that there is a trimester-dependent increase of CDT levels for both the HPLC and N-Latex immunoassay. The estimated reference values for CDT analysis were 1.52%, 1.94%, 2.04% and 1.37% for trimester 1, 2 , 3 and four weeks post-partum for the HPLC and 1.49%, 1.76%, 1.75% and 1.44% for the N-Latex immunoassay. This study shows that CDT levels clearly increase during pregnancy and that there is a need for higher cut-off values to prevent wrongful accusations of chronic alcohol abuse.
Introduction
Extensive consumption of alcohol during pregnancy can lead to several health complications for the unborn child. A daily disproportionate consumptionof alcohol has major implications on the development of the unborn child and could lead to severe disorders, such as Fetal Alcohol Syndrome (FAS), miscarriage, intrauterine fetal death and limited development of cognitive and behavioural characteristics (Gupta et
al.¸ 2016). As such it is important to refrain of alcohol during pregnancy. Considering
3
drinking problems might not always share this information with healthcare providers. Therefore, an objective unbiased screening method to detect excessive intake of alcohol independent of the sincerity of the patient is needed.
Analysis methods for recently consumed alcohol comprise the detection of alcohol in blood and/or breath, in which ethanol can be detected for +/- 12 hours (Helander et al., 2002) and ethylglucoride (EtG), an alcohol breakdown product which can remain elevated in urine for up to five days after drinking (Wurst et al., 2000). In order to detect prolonged excessive ethanol intake, analysis of Carbohydrate-deficient Transferrin (CDT) in serum has become the most common biomarker for detection of chronic ethanol abuse (Howlett et al.¸ 2018). CDT is defined as the ratio of monosialo- and disialo-transferrin (one or two sialic acids) to the total amount of transferrin and it is known that a daily intake of 50-80 gram of ethanol for two weeks alters the composition of the glycosylation-sites of Transferrin by elevating the CDT levels (Stibler et al., 1991). CDT levels can remain elevated for 2-3 weeks, depending on the amount of consumed alcohol and the period of time (Bortolotti et al., 2018) In more extreme cases it lasts for approximately a month before the CDT levels are restored to normal (Hock et al., 2005).
Transferrin, a glycoprotein for iron transport, contains two possible positions for glycosylation. These sites contain two N-linked oligosaccharide chains that can be bound by one or more sialic acid residues which results in six isoforms of transferrin. The formation of Acetaldehyde out of alcohol leads to a higher hydrolysis of sialic acid and less binding of carbohydrate chains, because sialyltransferase is inhibited (Stibler
et al.¸1991). Therefore, chronic alcohol intake is associated with an increase in
carbohydrate deficient transferrin isoforms, which makes CDT a key factor in identification of alcoholics and in the subsequent procedures such as driving license restrictions (NVKC, 2015).
Because of these severe consequences of CDT analysis, it is important that the results of CDT analysis are clear and reliable to prevent incorrect accusations of chronic ethanol abuse. The current cut-off values are determined for the general population,
4
however previous reports suggest that these might not be suitable for pregnant women (Bortolotti et al., 2020). Several studies described the physiological alterations in pregnant women in whom increasing CDT values could lead to false-positive identification of chronic alcohol abuse (Kenan et al., 2011; Bianchi et al., 2011). While the intake of alcohol during pregnancy is highly discouraged, alcohol consumption by pregnant women is still a problem in several countries, for instance in Europe almost 16% of the resident women consume alcohol during pregnancy (Mårdby et al., 2017) and in the United States and in Canada 10%-15% of the pregnant women consumed ethanol during pregnancy (Popova et al., 2017). Moreover, the study of Popova et al. (2017) stated that 3% of the women were considered to partake in binge drinking during pregnancy. These previous reports show that an objective parameter for alcohol is necessary for pregnant women. However, this parameter should be unbiased by physiological alterations during pregnancy to prevent incorrect accusations of chronic alcohol consumption.
The aim of this study is to investigate the changes in CDT values during pregnancy and to determine accurate, trimester dependent reference values. Therefore, plasma samples of 147 pregnant women will be analyzed by two analytically distinct experimental approaches, the High-performance liquid chromatography (HPLC) and the N-Latex CDT immunonephelometric assay. The HPLC method presented by Helander et al. (2003) is considered as the reference method (Helander et al., 2003; Delanghe et al.¸ 2007; Schellenberg et al., 2016).
Methods
Subjects and samples
Blood serum was collected from 147 during pregnancy in the first, second and third trimester and one month post-partum. The CDT levels of the samples taken one month post-partum are considered to represent the normal non-pregnant condition. Samples were collected by the Amphia Hopsital in Breda and analyzed at the Meander Medical Centre in Amersfoort. The sera were left-overs from previous research. Samples were
5
stored at -70 ◦C until analysis. The participating women were considered as healthy and their gestation as normal.
Methods
In this study the High-performance liquid chromatography (HPLC) protocol of the Meander Medical Centre was used and this is based on the HPLC as described by Helander et al. (2003). Before performing an analysis with the HPLC, 100 µl of each sample was incubated for at least 4 hours after saturation with 10 µl FeNTA and addition of 10 µl of a lipid precipitation mix, consisting of dextran sulfate (20 g/L) and sodium chloride (1 mol/L). Subsequently, the samples were centrifuged (10 minutes at 14000 rpm) and 100 µl of the supernatant was brought into 400 µl Aquadest, which was centrifuged again for 10 minutes at 14000 rpm. Afterwards, 450 ul of this supernatant was pipetted in a microvial for analysis. The liquid flow was 1 ml/min. The HPLC separates the distinct transferrin isoforms by using anion exchange chromatography. Subsequently, the absorbance was measured at 470 nm.
The used set-up comprised an HPLC device from Shimadzu outfitted with an LC-20AT prominence pump including a low-pressure gradient control valve FCV-10ALvp, a DGU-14A degasser and a SIL-20AC autosampler. The used UV detector was a UV-vis detector SPD-20AV prominence. To separate the distinct transferrin glycoforms an Source® 15Q PE 4.6/100 anion-exchange chromatography column (GE Healthcare, Uppsala, Sweden) was used.
Each run included three controls: two control samples for laboratory precision (Clinchek-control serum, Recipe; via AKSA Medical), so-called Clinchek 1 (%CDT range 1.06 – 1.76) and Clinchek 2 (%CDT range 2.92 – 4.39) and a pool of sera leftovers of CDT analysis of the laboratory of the Meander Medical Centre (expected %CDT range 1.62%-1.82%).
For the HPLC assay CDT is defined as the fraction of disialo-transferrin (as percent of the total amount of transferrin). The CDT cut-off value for defining elevated %CDT values is 1,7%.
6
The chromatographs (N = 431) were visually inspected by using Shimadzu Class-agent software for aberrant HPLC chromatographs such as Transferrin variants, di-tribridging, icteric and hemolytic samples.
The second way to determine the %CDT levels of the serum samples was by performing a N-Latex immunophelometric assay on the Atellica Neph 630 nephelometer (Siemens Healthineers). The used protocol of the Meander Medical Centre is based on the general Manufacture factional settings as provided by Siemens Healthineers and as described by Delanghe et al. (2007). At least 250 ul serum was necessary for analysis. Furthermore, transferrin levels were measured by the N-Latex immunoassay. CDT for the N-latex immunophelomtric assay is defined as the sum of asialo-, monosialo- and disialotransferrin (relative to the total amount of transferrin). In this assay a monoclonal antibody is used that is designed to bind specifically the transferrin glycoforms without 1 or 2 complete N-glycans and glycoforms with empty glycosylation sites, in other words, the asialo-, monosialo- and disialotransferrin glycoforms. To enable a better comparison with the results of the HPLC, which defines the DST fraction as CDT, and by using harmonization procedures from the International Federation of Clinical Chemistry (IFCC), the output of the N-latex immunoassay is converted by using the following formula, %CDT/IFCC = (N-Latex %CDT - 0,97)/0,7 as reported by Delanghe et al. (2007). The cut-off value for elevated CDT levels for the N-Latex immunoassay is 1.7% The used reagents and controls are taken from the standard N-Latex CDT kit (Siemens Healthineers) and in addition a serum pool, consisting of leftover sera from regular CDT analyses from the laboratory of the Meander Medical Centre, was used as extra control (expected %CDT range 1.80%-2.20%).
Statistical analysis
The outcome of both experimental approaches was analyzed using Statistical Package for the Social Sciences (SPSS) version 24. The distribution of each trimester was
7
examined using the Shapiro-Wilk test (significance at p< 0.05), because this test is more powerful than the Kolmogorov-Smirnov test (Steinskog et al. 2007; Ghasemi & Zahediasl, 2012). The homogenity of the Variances was tested with the Levene’s test. Since the pregnant women were followed during pregnancy, the samples of each trimester and post-partum were considered dependent and therefore to analyze the differences between the trimesters for CDT a one-way ANOVA and a Tukey’s post-hoc test were performed at a significance level of p < 0.05 (Lee & Lee, 2018). Extra attention was given to a subgroup of patients with a complete collection of samples (samples present for all periods under investigation). To analyze the differences between the trimesters of this group, the results were analyzed by using an one-way ANOVA and a Tukey’s post-hoc test for the HPLC output and a Sign test for the N-Latex immunoassay output (significance at p < 0.05).
Finally, the cut-off values were determined. To do so, the upper limit of the 95% confidence interval, in other words the 97.5th percentile, was calculated following the
formula: µ + 1.96 * SD, whereby µ is the mean and SD the standard deviation of the corresponding period.
Results
From 147 patients 123, 120, 100 and 96 serum samples from trimester 1, 2, 3 and post-partum respectively were available for CDT analysis. Visual control of the output of the HPLC and the N-Latex immunoassay led to exclusion of 6 samples, because of Transferrin variants (N = 5). An extraordinary high CDT level (%CDT = 3.13 as measured by the HPLC; %CDT = 2.54 as measured by the N-Latex immunoassay) was observed in one sample with both assays in trimester 2 and since this was the only sample of this patient, this sample was excluded, because it raised suspicion of chronic alcohol consumption. Moreover, after data analysis it was considered an outlier (the data and figures can be found in the supplemental data). One sample of trimester 1 did not contain enough serum for both experiments and was only analyzed in the HPLC. In the end, the data consisted of 122, 117, 99 and 95 samples from trimester 1,
8
2, 3 and post-partum for the HPLC and 122, 117, 99 and 95 samples for the N-Latex immunoassay.
A summary of the results of the statistical analysis can be found in table 1. The Shapiro-Wilk test showed that the Trimester 1, 2, 3 and the Post-Partum samples were normally distributed for both methods at a significance of p<0.05. In the supplemental data can be found that the assumption of equal variances was not violated (significance at p<0.05).
Table 1. Descriptive statistical analysis of CDT levels in pregnant women in trimester 1, 2,3 and 4 weeks post-partum, measured by 1a. a high-performance liquid chromatography (HPLC) and 1b. a N-Latex immunoassay.
1a. High-Performance liquid chromatography (HPLC)
HPLC Trimester 1 Trimester 2 Trimester 3 Post-partum
Sample size 122 117 99 95
Mean 1.11 1.5421 1.6243 1.0376
95% Confidence Interval for Mean 1.0728 1.5952 1.5824 1.0035
1.1472 1.5790 1.6663 1.0716 Median 1.15 1.55 1.6400 1.0500 Variance 0.043 0.041 0.044 0.028 Std. Deviation 0.20755 0.20151 0.21015 0.16713 Minimum 0.50 0.96 1.05 0.53 Maximum 1.59 1.94 2.11 1.53 Range 1.09 0.98 1.06 1.00 Interquartile Range 0.27 0.29 0.28 0.24 Skewness -0.439 -0.439 -0.244 -0.094 Kurtosis 0.275 -0.152 0.069 0.797
9 1b. N-Latex immunoassay
N-Latex immunoassay Trimester 1 Trimester 2 Trimester 3 Post-partum
Sample size 121 117 99 95
Mean 1.0429 1.2470 1.2165 0.9860
95% Confidence Interval for Mean 1.0015 1.1995 1.1620 0.9371
1.0843 1.2945 1.2709 1.0349 Median 1.0600 1.21 1.21 0.96 Variance 0.053 0.067 0.074 0.058 Std. Deviation 0.22982 0.25954 0.27291 0.24010 Minimum 0.51 0.43 0.54 0.44 Maximum 1.69 1.89 1.92 1.77 Range 1.18 1.46 1.38 1.33 Interquartile Range 0.31 0.39 0.37 0.30 Skewness 0.145 0.159 0.188 0.397 Kurtosis 0.014 0.121 -0.004 0.811
Shapiro-Wilk test for normal distribution P = 0.694 P = 0.231 P = 0.856 P = 0.238
Table 2 shows an overview of the results of the one-way ANOVA and Tukey’s post-hoc test of the HPLC and the N-Latex immunoassay (significance at p < 0.05).
During pregnancy the CDT values increase strongly compared to the post-partum value and the CDT values in trimester 2 and 3 are significantly higher than those post-partum (see table 2 and figure 1) with both assays. In trimester 1 the increase was only modest and significant only for the HPLC assay.
Several samples are approaching or even exceeding the cut-off values for both experimental approaches. These results are visualized in figure 1 as histograms (1a; 1b) and box plots (1c; 1d). In these figures, the red line at the histograms represent the normal distribution and in the boxplot the horizontal line inside each box represents the median, the box frame represents the interquartile range (IQR), the vertical lines represent the range of the dataset, except the outliers and the red dotted line represents
10
the current CDT cut-off value of 1.7%. Outliers are defined as the function of 1.5 of the IQR.
Table 2. Results of testing differences in CDT levels between trimesters of pregnancy and post-partum for significance by performing an one-way ANOVA – Tukey Post-Hoc test. 2a. High-performance liquid chromatography (HPLC) and 2b. N-Latex immunoassay. The N-latex immunoassay output is corrected to IFCC values.
One-way ANOVA – Tukey Post-Hoc test HPLC N-Latex immunoassay
Trimester 1 Trimester 2 P < 0.001 P < 0.001 Trimester 3 P < 0.001 P < 0.001 Post-partum P = 0.039 P = 0.348 Trimester 2 Trimester 1 P < 0.001 P < 0.001 Trimester 3 P= 0.013 P = 0.809 Post-partum P < 0.001 P < 0.001 Trimester 3 Trimester 1 P < 0.001 P < 0.001 Trimester 2 P = 0.013 P = 0.809 Post-partum P < 0.001 P < 0.001 Post-partum Trimester 1 P = 0.039 P = 0.348 Trimester 2 P < 0.001 P < 0.001 Trimester 3 P < 0.001 P < 0.001
11
1a.
1b.
12
Figure 1. CDT levels of each trimester of pregnancy and four weeks post-partum measured by a high-performance liquid chromatography (HPLC) and N-Latex immunoassay visualized in Histograms (1a. HPLC; 1b. N-Latex immunoassay) and box-plots (1c. HPLC; 1d. N-Latex immunoassay). The red line at the histograms represent the normal distribution and in the boxplot the horizontal line inside each box represents the median, the box frame represents the interquartile range (IQR), the vertical lines represent the range of the dataset, except the outliersOutliers are defined as the function of 1.5 of the IQR. The red dotted lines represent the current CDT cut-off value of 1.7%. The p-values are stated in table 2.
The results of analysis of the subpopulation of patients with four available samples (N = 53 for HPLC; N = 52 for N-Latex immunoassay), analyzed with an one-way ANOVA and Tukey’s post-hoc test for the HPLC results and a Sign test for the N-Latex immunoassay, can be found in Table 3 and are visualized in figure 2, 3 and 4 and as can be seen CDT levels are elevated in trimester 1, 2 and 3 compared to post-partum. In trimester 2 and 3 for both the HPLC and the N-Latex immunoassay the rise of the CDT levels was significant. Trimester 1 was not significantly elevated in both experimental approaches.
Table 3. Results of testing differences in CDT levels between trimesters of pregnancy and post-partum for significance by performing an one-way ANOVA – Tukey Post-Hoc test for the 3a. High-performance liquid chromatography (HPLC; N = 53) and a Sign test for the 3b. N-Latex immunoassay output (N = 52). This group receives extra attention since it possible to analyze the results during all distinct periods.
One-way ANOVA – Tukey Post-Hoc test Sign test (HPLC) (N-Latex) HPLC N-Latex immmunoassay Trimester 1 Trimester 2 P < 0.001 P < 0.001 Trimester 3 P < 0.001 P < 0.001 Post-partum P = 0.125 P = 0.263 Trimester 2 Trimester 1 P < 0.001 P < 0.001
13 2d. Trimester 3 P= 0.312 P = 0.038 Post-partum P < 0.001 P < 0.001 Trimester 3 Trimester 1 P < 0.001 P < 0.001 Trimester 2 P = 0.312 P = 0.038 Post-partum P < 0.001 P < 0.001 Post-partum Trimester 1 P = 0.125 P = 0.263 Trimester 2 P < 0.001 P < 0.001 Trimester 3 P < 0.001 P < 0.001 2a. 2b. 2c.
14
Figure 2. CDT levels of each trimester of pregnancy and four weeks post-partum measured by a high-performance liquid chromatography (HPLC) and N-Latex immunoassay visualized in Histograms (2a. HPLC; 2b. N-Latex immunoassay) and box-plots (2c. HPLC; 2d. N-Latex immunoassay). The red line at the histograms represent the normal distribution and in the boxplot the horizontal line inside each box represents the median, the box frame represents the interquartile range (IQR), the vertical lines represent the range of the dataset, except the outliers and the red dotted line represents the current CDT cut-off value of 1.7%. Outliers are defined as the function of 1.5 of the IQR. This group receives extra attention since it possible to analyze the results during all distinct periods. P-values are stated in table 3.
Figure 3. CDT levels of each trimester of pregnancy and four weeks post-partum measured by a high-performance liquid chromatography (HPLC). 53 women provided serum samples for CDT analysis at every trimester and four weeks post-partum. The numbers 1, 2, 3 and 4 on the X-axis, represent trimester 1, 2 3 and post-partum.
15
Figure 4. CDT levels of each trimester of pregnancy and four weeks post-partum measured by a N-Latex immunoassay. 53 women provided serum samples for CDT analysis at every trimester and four weeks post-partum. The numbers 1, 2, 3 and 4 on the X-axis, represent trimester 1, 2 3 and post-partum.
Finally, trimester-dependent cut-off values were determined. The results of these calculations can be found in table 4.
In the Netherlands the cut-off values used to determine excessive alcohol abuse take the inter-laboratory and biological variation into account. Specifically, for the Dutch situation an extra section is added for the Dutch approach.
16
Table 4. Descriptive statistical analysis to determine CDT cut-off values for trimester 1, 2,3 and four weeks post-partum (5a. HPLC; 5b. N-Latex immunoassay). The upper limit represents the 97.5th percentile and is calculated with the formula: µ + 1.96 * SD, whereby µ is the mean and SD the standard deviation of the corresponding period.
4a. High-performance liquid chromatography (HPLC)
HPLC Trimester 1 Trimester 2 Trimester
3
Post-partum
97.5th percentile 1.52% 1.94% 2.04% 1.37%
HPLC reference values determined by Bortolotti et al. (2020)
1.45% 2.01% 2.05% -
Current cut-off value 1.7% 1.7% 1.7% 1.7%
Estimated cut-off value 1.7% 1.94% 2.04% 1.7%
4b. N-Latex immunoassay
HPLC Trimester 1 Trimester 2 Trimester 3 Post-partum
97.5th percentile 1.49% 1.76% 1.75% 1.44%
Current cut-off value 1.7% 1.7% 1.7% 1.7%
17
The CDT reference values in the Netherlands
In the Netherlands, a different approach is used to determine the off value of elevated CDT. The international cut-off value for elevated CDT of 1.7% is used, however a critical value is added for compensation of interlaboratory and biological variation. Therefore, the cut-off value for elevated CDT in the Netherlands is 2.0%.
The formula for the cut-off value in the Netherlands is 97,5th percentile + critical difference. (NVKC, 2015) The upper
limit of the 95% confidence interval, in other words the 97.5th percentile, was determined following the formula: µ + 1.96
* SD, whereby µ is the mean and SD the standard deviation of the corresponding period. The formula for the critical difference is: µ1 – µ0 = µ1 – µ0 = z * √[2] * √ [CVa,b^2 + CVb,w^2] * GW/100, where z is the amount of standard deviations for the chosen interval (1.65 for the 95% confidence interval) CVa,b the analytical variation between laboratories (for HPLC, EQUALIS = 8.8%, for N-Latex, Equalis = 6.6%), CVb,w the biological variance (4.7%; Helander et al.; 2003) and GW is the involved limit-value, in this case the output of the 97.5th percentile. (NVKC, 2015) The results of these
calculations can be found in table 6.?---
Table 5. Descriptive statistical analysis to determine CDT cut-off values for trimester 1, 2,3 and four weeks post-partum (5a. HPLC; 5b. N-Latex immunoassay). The upper limit represents the 97.5th percentile and is calculated with the formula: µ + 1.96 * SD,
whereby µ is the mean and SD the standard deviation of the corresponding period. The critical value is calculated with the formula µ1 – µ0 = µ1 – µ0 = z * √[2] * √ [CVa,b^2 + CVb,w^2] * GW/100 in which z represents the amount of standard deviations for the chosen interval (1.65 for the 95% confidence interval) CVa,b the analytical variation between laboratories (for HPLC, EQUALIS = 8.8%, for N-Latex, Equalis = 6.6%), CVb,w the biological variance (4.7%) and GW is the involved limit-value, in this case the output of the 97.5th percentile.
5a. High-performance liquid chromatography (HPLC)
HPLC Trimester 1 Trimester 2 Trimester 3 Post-partum
97.5th percentile 1.52% 1.94% 2.04% 1.37%
Critical difference 0.35% 0.45% 0.47% 0.32%
Current cut-off value 2.0% 2.0% 2.0% 2.0%
New cut-off value 2.0% (1.87%*) 2.39% 2.51% 2.0% (1.69%*)
5b. N-Latex immunoassay
HPLC Trimester 1 Trimester 2 Trimester 3 Post-partum
97.5th percentile 1.49% 1.76% 2.75% 1.44%
Critical difference 0.28% 0.33% 0.33% 0.27%
Current cut-off value 2.6% 2.6% 2.6% 2.6%
New cut-off value 2.6% (1.78%*) 2.6% (2.09%*) 2.6% (2.08%*) 2.6% (1.71%*)
18
Discussion
Extensive consumption of alcohol during pregnancy can lead to severe complications for the unborn child (Gupta et al.¸ 2016). Therefore an objective unbiased screening method is necessary to detect excessive intake of alcohol during pregnancy. There are several possible biomarkers to detect alcohol intake, however for prolonged alcohol consumption CDT turns out to be a suited and accurate biomarker, since it remains elevated for 2-3 weeks after excessive alcohol consumption, enabling long-term detection of chronic alcohol abuse (Howlett et al.¸ 2018; Stibler et al., 1991; Bortolotti et
al., 2018). Yet, it is observed that pregnant women may already have elevated CDT
levels, unrelated to alcohol intake (Kenan et al., 2011; Bianchi et al., 2011; Bortolotti et
al., 2020).
The aim of this study was to examine the CDT levels during pregnancy and to determine new cut-off values for this population. 439 serum samples of 147 pregnant women collected during the three trimesters of pregnancy and four weeks post-partum were analysed by high-performance liquid chromatography (HPLC) and the N-latex Immunoassay. A significant increase in CDT-levels was found with the HPLC for trimester 1, 2 and 3 compared to the post-partum samples. In trimester 2 and 3, the CDT levels were approaching and exceeding the cut-off value of 1.7%. To be more specific, 27.4% of the samples in trimester 2 and 40.4% in trimester 3 had a %CDT of 1.7% or higher. No elevated samples were observed in trimester 1 and post-partum. For the N-latex immunoassay similar results were observed, although the rise of CDT levels in trimester 1 was not significant. Moreover, 4.3% of the samples in trimester 2, 5.1% in trimester 3 and 1.1% in post-partum had a %CDT of 1.7% or higher. Furthermore, transferrin levels increased significantly in trimester 2 and 3 as measured by the N-Latex immunoassay (supplemental data). This study confirms that CDT levels are rising over time during pregnancy and that the current cut-off value of 1.7% is not justified for trimester 2 and 3 and could lead to a false-positive identification of extensive alcohol consumption. The estimated reference values for CDT analysis were 1.52%, 1.94%, 2.04% and 1.37% for trimester 1, 2 , 3 and four weeks post-partum for the
19
HPLC and 1.49%, 1.76%, 1.75% and 1.44% for the N-Latex immunoassay. Considering the harmonisation of the CDT assays, these results could be combined towards a general CDT cut-off values during pregnancy: 1.7% for trimester 1, 1.92% for trimester 2, 2.04% for trimester 3 and 1.7% for four weeks post-partum. Note that such an approach may not be justified because of the differences between the N-latex and HPLC assay, which remain present after harmonisation (Jacobs et al., 2020).
The elevation of CDT levels during pregnancy was expected, since in previous research of Kenan et al. (2011), Bianchi et al. (2011) and Bortolotti et al. (2020) increased CDT levels in pregnant women were observed over time. Noteworthy, the previously mentioned studies all used the HPLC for determining the CDT levels. Only one study was found that performed N-Latex immunoassay for serum samples of women during pregnancy, but this study of Niemelä et al. (2016) showed only a significant increase of the CDT levels for women that consumed alcohol during pregnancy. No elevated CDT levels were found in women who refrained from drinking alcohol during pregnancy or in abstainers of alcohol (Niemelä et al., 2016). Our study used two distinct CDT analysis methods, has a large sample size and provides an overview of the CDT levels in the same pregnant women over time. The study of Bortolotti et al. (2020) is interesting, because this study has a comparable aim and has a similar sample size, in contrast to the smaller sample sizes of Kenan et al. (2011) and Bianchi et al. (2011). However, the trend of elevation of CDT levels during pregnancy and the mean CDT ratio of each study is similar or close to the observed results of this study. Kenan et al. (2011) found mean %CDT levels of 1.07 and 1.61 in trimester 1 and 3, Bianchi et al. (2011) found %CDT levels of 1.01, 1.30 and 1.53 in trimester 1, 2 and 3 and Bortolotti et
al. (2020) found %CDT levels of 1.07, 1.43 and 1.59 for trimester 1, 2 and 3, all measured
by the HPLC. Importantly, in contrast to our approach of using the post-partum levels as baseline CDT values, Bortolotti et al. (2020) used a reference group. In our approach we assumed that CDT levels would have returned to baseline values in one month post-partum. Indeed, Bortolotti et al. (2018) and Hock et al. (2005) have shown that CDT values return to normal post-partum. The excellent agreement of our post-partum
20
results with results of the control group of Bortolotti et al. (2020) supports this conclusion. The reference values estimated with a HPLC by Bortolotti (1.45%, 2.01% and 2.05% for trimester 1, 2 and 3) are also very similar to the reference values of the HPLC determined in this study. The differences for each trimester is 0.07 %CDT or less. Overall, our study, in combination with those described in literature demonstrated that the current CDT cut-off value is not adequate for pregnant women and that a revision of the reference values is needed in this group.
The mechanism leading to elevated CDT levels during pregnancy is not clear. However, this study also observed rising transferrin levels during pregnancy (supplemental data) and previous research showed that human serum sialic acids are dropping during pregnancy (Rajan et al.; 1983). The rising transferrin levels combined with the dropping serum sialic acids levels may indicate an enhanced production of transferrin with short carbohydrate chains. In general, rising transferrin levels during pregnancy could result in altered levels of isoforms. This is supported by a study by Rajan et al. (1983) which also observed that the sialyltransferase levels do not increase during pregnancy. These observations suggest that a possible disbalance of available transferrin, sialyltransferase and sialic acids could lead to reduced glycosylation of transferrin.
Noteworthy is the difference between the HPLC and the N-Latex immunoassay, despite harmonisation and standardisation procedures by the IFCC. This was already observed in a study of Jacobs et al. (2020), which observed more variation between the experimental approaches at low and high CDT levels.The results of this study likely support this observation, since the discrepancy between the two analysis methods is larger in trimester 2 and 3 compared to trimester 1 and post-partum. However, an analysis of these different outputs is not performed in this study because a direct method comparison was not the aim of this study. It is also remarkable that a study of Niemelä et al. (2016) also showed that CDT levels did not increase during pregnancy when measured with the N-Latex immunoassay. Although it remains unclear where this variation comes from, it is important to realize that these detecting methods are
21
fundamentally different. The HPLC measures all glycoforms, including CDT (as disialotransferrin) and the N-Latex immunophelometric assay detects the lack of binding of monoclonal antibodies without one or two N-glycans. The HPLC is a very robust method and it is possible that an immunoassay is susceptible to interference, especially during physiological alterations of the blood composition during pregnancy. The observed discrepancy between these two analysis methods in this study, the study of Jacobs et al. (2020) and in the study of Niemelä et al. (2016), suggest that there might be need for a revision of the harmonisation procedures, for example the formula for the correction of the N-Latex immunoassay output to IFCC-CDT values.
In addition, the distinct trends of both analysis methods is noteworthy. As can be seen in figure 3 and 4, the CDT levels of many samples analysed with the N-Latex immunoassay are decreasing after trimester 2, while the CDT levels of samples measured by the HPLC are dropping after trimester 3. In general, detecting a visual trend in the results of the N-Latex immunoassay is difficult. Another remarkable outcome of this study is the variation in the p-values of the Shapiro-Wilk test for normal distribution between the trimesters: the p-values of, for example, the HPLC trimester 3 and post-partum are higher than the p-values of HPLC trimester 1 and 2 and also variation is seen in the skewness and the kurtosis. A possible explanation for this could be that the rate of physiological alterations in women is probably higher at the beginning of gestation leading to more variation.
The number of studies investigating CDT levels during pregnancy is small, while CDT analysis plays a major role in the detection of chronic alcohol abuse. However, the results of previous studies are mostly consistent in showing elevation of the CDT levels during pregnancy. Although in most countries alcohol consumption during pregnancy is discouraged, it is important to have an objective and unbiased detecting method, but also to prevent false-positive accusations. Moreover, early detection of chronic ethanol abuse could be important because of the severe implications for the foetus. This study provided a better insight in the CDT levels during pregnancy and
22
showed that the current cut-off value for elevated CDT is not sufficient for pregnant women and that there is need for trimester dependent cut-off values. Further study could assess other CDT analysis methods, like capillary electrophoresis. It is also recommended to acquire an improved understanding of CDT levels during pregnancy. Additional studies will be needed to examine the possible mechanism that is causing the elevated CDT levels. Suggestions for these studies are examining serum sialyltransferase and sialic acids levels during pregnancy. This study shows that the CDT levels are clearly increased during pregnancy and that there is a need for higher cut-off values to prevent false positive detections of chronic alcohol abuse. Moreover, this study demonstrates the importance of critically assessing the application of general cut-off values and guidelines for specific subgroups. Indeed, for pregnant women, the use of higher, trimester dependent CDT cut off values contributes to improvements in the CDT policy and the corresponding judicial and medical processes to prevent injustice.
Literatuurlijst
1. Gupta, K. K., Gupta, V. K., & Shirasaka, T. (2016). An Update on Fetal Alcohol Syndrome-Pathogenesis, Risks, and Treatment. Alcoholism, clinical and experimental research, 40(8), 1594–1602. https://doi.org/10.1111/acer.13135 2. Helander, A. and Eriksson, C.J.P. (2002), Laboratory Tests for Acute Alcohol
Consumption: Results of the WHO/ISBRA Study on State and Trait Markers of Alcohol Use and Dependence. Alcoholism: Clinical and Experimental Research, 26: 1070-1077. doi:10.1111/j.1530-0277.2002.tb02641.x
3. Friedrich Martin Wurst, Christoph Kempter, Joerg Metzger, Stephan Seidl, Andreas Alt (2000), Ethyl glucuronide: a marker of recent alcohol consumption with clinical and forensic implications, Alcohol, Volume 20, Issue 2, Pages 111-116, ISSN 0741-8329, https://doi.org/10.1016/S0741-8329(99)00076-2.
4. Howlett, H., Mackenzie, S., Gray, W. K., Rankin, J., Nixon, L., Richardson, A., Strehle, E. M., & Brown, N. W. (2018). Assessing prevalence of alcohol consumption in early pregnancy: Self-report compared to blood biomarker analysis. European journal of medical genetics, 61(9), 531–538. https://doi.org/10.1016/j.ejmg.2018.05.009
5. Stibler H. (1991). Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clinical chemistry, 37(12), 2029–2037
23
6. Bortolotti, F., Sorio, D., Bertaso, A., & Tagliaro, F. (2018). Analytical and diagnostic aspects of carbohydrate deficient transferrin (CDT): A critical review over years 2007-2017. Journal of pharmaceutical and biomedical analysis, 147, 2–12. https://doi.org/10.1016/j.jpba.2017.09.006
7. Hock, B., Schwarz, M., Domke, I., Grunert, V.P., Wuertemberger, M., Schiemann, U., Horster, S., Limmer, C., Stecker, G. and Soyka, M. (2005), Validity of carbohydrate‐deficient transferrin (%CDT), γ‐glutamyltransferase (γ‐GT) and mean corpuscular erythrocyte volume (MCV) as biomarkers for chronic alcohol abuse: a study in patients with alcohol dependence and liver disorders of non‐alcoholic and alcoholic origin. Addiction, 100: 1477-1486. doi:10.1111/j.1360-0443.2005.01216.x
8. Richtlijn van de Nederlandse Vereniging voor Klinische Chemie en Laboratoriumgeneeskunde; Geschiktheid van CDT analysemethoden ten behoeve van onderzoek naar chronisch overmatig alcohol gebruik, opgesteld door de werkgroep CDT van de NVKC, Richtlijn geaccordeerd door voorzitter NVKC dd 15 oktober 2008 en geactualiseerd 20 december 2012, herzien december 2015.
9. Bortolotti, F., Raffaelli, R., Di Simone, N., Semprebon, M., Mirandola, M., Simonetto, C., De Marchi, F., Trevisan, M. T., Carli, G., Dorizzi, R. M., Scambia, G., Franchi, M., & Tagliaro, F. (2020). CDT reference values for monitoring chronic alcohol abuse in pregnancy. Clinica chimica acta; international journal of clinical chemistry, 507, 156–160. https://doi.org/10.1016/j.cca.2020.04.014 10. Kenan, N., Larsson, A., Axelsson, O., & Helander, A. (2011). Changes in
transferrin glycosylation during pregnancy may lead to false-positive carbohydrate-deficient transferrin (CDT) results in testing for riskful alcohol consumption. Clinica chimica acta; international journal of clinical chemistry, 412(1-2), 129–133. https://doi.org/10.1016/j.cca.2010.09.022
11. Bianchi, V., Ivaldi, A., Raspagni, A., Arfini, C., & Vidali, M. (2011). Pregnancy and variations of carbohydrate-deficient transferrin levels measured by the candidate reference HPLC method. Alcohol and alcoholism (Oxford, Oxfordshire), 46(2), 123–127. https://doi.org/10.1093/alcalc/agq092
12. Mårdby, A. C., Lupattelli, A., Hensing, G., & Nordeng, H. (2017). Consumption of alcohol during pregnancy-A multinational European study. Women and birth : journal of the Australian College of Midwives, 30(4), e207–e213. https://doi.org/10.1016/j.wombi.2017.01.003
13. Popova, S., Lange, S., Probst, C., Parunashvili, N., & Rehm, J. (2017). Prevalence of alcohol consumption during pregnancy and Fetal Alcohol Spectrum Disorders among the general and Aboriginal populations in Canada and the United States. European journal of medical genetics, 60(1), 32–48. https://doi.org/10.1016/j.ejmg.2016.09.010
14. Helander, A., Husa, A., & Jeppsson, J. O. (2003). Improved HPLC method for carbohydrate-deficient transferrin in serum. Clinical chemistry, 49(11), 1881– 1890. https://doi.org/10.1373/clinchem.2003.023341
24
15. Delanghe, J. R., Helander, A., Wielders, J. P., Pekelharing, J. M., Roth, H. J., Schellenberg, F., Born, C., Yagmur, E., Gentzer, W., & Althaus, H. (2007). Development and multicenter evaluation of the N latex CDT direct immunonephelometric assay for serum carbohydrate-deficient transferrin.
Clinical chemistry, 53(6), 1115–1121. https://doi.org/10.1373/clinchem.2006.084459
16. Schellenberg, F., Wielders, J., Anton, R., Bianchi, V., Deenmamode, J., Weykamp, C., Whitfield, J., Jeppsson, J. O., & Helander, A. (2017). IFCC approved HPLC reference measurement procedure for the alcohol consumption biomarker carbohydrate-deficient transferrin (CDT): Its validation and use. Clinica chimica acta; international journal of clinical chemistry, 465, 91–100. https://doi.org/10.1016/j.cca.2016.12.022
17. Steinskog, D. J., D. B. Tjøstheim, and N. G. Kvamstø, 2007: A Cautionary Note on the Use of the Kolmogorov–Smirnov Test for Normality. Mon. Wea. Rev., 135, 1151–1157, https://doi.org/10.1175/MWR3326.1.
18. Ghasemi, A., & Zahediasl, S. (2012). Normality tests for statistical analysis: a guide for non-statisticians. International journal of endocrinology and metabolism, 10(2), 486–489. https://doi.org/10.5812/ijem.3505
19. Lee, S., & Lee, D. K. (2018). What is the proper way to apply the multiple comparison test?. Korean journal of anesthesiology, 71(5), 353–360. https://doi.org/10.4097/kja.d.18.00242
20. Jacobs, L., Te Stroet, R. M., & Demir, A. Y. (2020). Evaluation of carbohydrate-deficient transferrin measurements on the V8 capillary electrophoresis system and comparison with the IFCC approved HPLC reference method and N-Latex immunonephelometric assay. Clinical chemistry and laboratory medicine, /j/cclm.ahead-of-print/cclm-2020-0545/cclm-2020-0545.xml. Advance online publication. https://doi.org/10.1515/cclm-2020-0545
21. Niemelä, O., Niemelä, S., Ritvanen, A., Gissler, M., Bloigu, A., Vääräsmäki, M., Kajantie, E., Werler, M.M. and Surcel, H.‐M. (2016), Assays of Gamma‐ Glutamyl Transferase and Carbohydrate‐Deficient Transferrin Combination from Maternal Serum Improve the Detection of Prenatal Alcohol Exposure. Alcohol Clin Exp Res, 40: 2385-2393. doi:10.1111/acer.13207
22. Rajan, R., Sheth, A. R., & Rao, S. S. (1983). Sialic acid, sialyltransferase and neuraminidase levels in maternal plasma, urine and lymphocytes during pregnancy and post-partum period--a longitudinal study in women. European journal of obstetrics, gynecology, and reproductive biology, 16(1), 37–46. https://doi.org/10.1016/0028-2243(83)90218-6
25
1c. 1d.
Supplemental data
Figure 1. Normal Q-Q plots and boxplots of trimester 2 of the HPLC analysis of a sample with (1a; 1c) and without (1b; 1d) a remarkable high %CDT level of 3.13 (sample id is 140). This sample was excluded, because of suspicion of chronic alcohol consumption and because it turned out to be an outlier. The red line at the histograms represent the normal distribution and in the boxplot the horizontal line inside each box represents the median, the box frame represents the interquartile range (IQR), the vertical lines represent the range of the dataset, except the outliers. Outliers are defined as the function of 1.5 of the IQR.
26
Normality plots of each trimester of each detecting method
27 HPLC Trimester 2
28 HPLC post-partum
29 N-Latex immunoassay trimester 1
30 N-Latex immunoassay trimester 3
31 N-Latex immunoassay post-partum
Statistical Analysis:
Test of homogenity of Variances Levene statistic
df1 df2 significance
HPLC 2.581 3 429 0.053
32
Descriptives
HPLC
N Mean Std. Deviation Std. Error
95% Confidence Interval for Mean
Minimum Maximum
Lower Bound Upper Bound
1 122 1,1100 ,20755 ,01879 1,0728 1,1472 ,50 1,59 2 117 1,5421 ,20151 ,01863 1,5052 1,5790 ,96 1,94 3 99 1,6243 ,21015 ,02112 1,5824 1,6663 1,05 2,11 4 95 1,0376 ,16713 ,01715 1,0035 1,0716 ,53 1,53 Total 433 1,3285 ,32182 ,01547 1,2981 1,3589 ,50 2,11 Multiple Comparisons Dependent Variable: HPLC Tukey HSD
(I) Period (J) Period
Mean Difference
(I-J) Std. Error Sig.
95% Confidence Interval Lower Bound Upper Bound
1 2 -,43205* ,02566 ,000 -,4982 -,3659 3 -,51434* ,02683 ,000 -,5835 -,4451 4 ,07242* ,02714 ,039 ,0024 ,1424 2 1 ,43205* ,02566 ,000 ,3659 ,4982 3 -,08229* ,02708 ,013 -,1521 -,0124 4 ,50447* ,02739 ,000 ,4338 ,5751 3 1 ,51434* ,02683 ,000 ,4451 ,5835 2 ,08229* ,02708 ,013 ,0124 ,1521 4 ,58676* ,02849 ,000 ,5133 ,6602 4 1 -,07242* ,02714 ,039 -,1424 -,0024 2 -,50447* ,02739 ,000 -,5751 -,4338 3 -,58676* ,02849 ,000 -,6602 -,5133
*. The mean difference is significant at the 0.05 level.
ANOVA
HPLC
Sum of Squares df Mean Square F Sig.
Between Groups 27,865 3 9,288 236,115 ,000
Within Groups 16,876 429 ,039
33
Descriptives
Latex
N Mean Std. Deviation Std. Error
95% Confidence Interval for Mean
Minimum Maximum
Lower Bound Upper Bound
1 121 1,0429 ,22982 ,02089 1,0015 1,0843 ,51 1,69 2 117 1,2470 ,25954 ,02399 1,1995 1,2945 ,43 1,89 3 99 1,2165 ,27291 ,02743 1,1620 1,2709 ,54 1,92 4 95 ,9860 ,24010 ,02463 ,9371 1,0349 ,44 1,77 Total 432 1,1254 ,27287 ,01313 1,0996 1,1512 ,43 1,92 ANOVA Latex
Sum of Squares df Mean Square F Sig.
Between Groups 5,221 3 1,740 27,721 ,000
Within Groups 26,870 428 ,063
Total 32,091 431
Multiple Comparisons
Dependent Variable: Latex Tukey HSD
(I) Period (J) Period
Mean Difference
(I-J) Std. Error Sig.
95% Confidence Interval Lower Bound Upper Bound
1 2 -,20412* ,03249 ,000 -,2879 -,1203 3 -,17357* ,03396 ,000 -,2611 -,0860 4 ,05689 ,03435 ,348 -,0317 ,1455 2 1 ,20412* ,03249 ,000 ,1203 ,2879 3 ,03054 ,03422 ,809 -,0577 ,1188 4 ,26101* ,03460 ,000 ,1718 ,3503 3 1 ,17357* ,03396 ,000 ,0860 ,2611 2 -,03054 ,03422 ,809 -,1188 ,0577 4 ,23046* ,03599 ,000 ,1377 ,3233 4 1 -,05689 ,03435 ,348 -,1455 ,0317 2 -,26101* ,03460 ,000 -,3503 -,1718 3 -,23046* ,03599 ,000 -,3233 -,1377
34
Transferrin levels measured by the N-Latex immunoassay
Transferrin levels were examined and it turned out that only trimester 2 was accordant a normal distribution (p = 0.655). In contrary, the Shapiro-Wilk test showed that trimester 1, 3 and post-partum did not correspond to a normal distribution, because p<0.05. Therefore, a non-parametric test was needed for examination of the increasing transferrin levels. A Sign test was applied to analyze the increase of the transferrin concentrations and as can be found in table 6, transferrin levels are. This is visually supported by figure 6.
Table 7. Results of testing differences in transferrin levels between trimesters of pregnancy and post-partum for significance by performing a Sign test.
Tri 1 – tri2 Tri 1 – tri 3 Tri 1 - PP Tri 2 – tri 3 Tri 2 - PP Tri 3 - PP P-value P<0.001 P < 0.001 0.094 P<0.001 P < 0.001 P < 0.001
Figure 7. Transferrin levels of each trimester of pregnancy and four weeks post-partum measured by a N-Latex immunoassay visualized in a Histogram (3a) and a box-plot (3b) The red line at the histograms represent the normal distribution and in the boxplot the horizontal line inside each box represents the median, the box frame represents the interquartile range (IQR), the vertical lines represent the range of the dataset. Outliers are defined as the function of 1.5 of the IQR. Palatino Linotype
