Human serum transferrin glycosylation pattern : population differences, analytical methodology and application as a biomarker for testing of alcohol abuse and CDG

(1)

From the Division of Alcohol and Drug Dependence Research, Department of Clinical Neuroscience,

Karolinska Institutet, Stockholm, Sweden

HUMAN SERUM TRANSFERRIN GLYCOSYLATION PATTERN

– POPULATION DIFFERENCES, ANALYTICAL METHODOLOGY AND APPLICATION AS BIOMARKER FOR TESTING OF ALCOHOL

ABUSE AND CDG

Jonas P Bergström

Stockholm 2007

(2)

(3)

All previously published papers were reproduced with permission from the publisher.

Published by Karolinska Institutet. Printed by US-AB

(4)

“Anhängare av positivismen får hävda vad de vill, det finns människor till allt har jag börjat förstå.”

- Anonym, Internet

”Att slåss med troll, befria prinsessor och döda varulvar, det är att leva.”

- August Strindberg

(5)

(6)

(7)

ABSTRACT

Alcohol use and abuse is a major social and economic problem in many societies.

Carbohydrate-deficient transferrin (CDT) is a widely used and highly specific biochemical alcohol marker of prolonged alcohol abuse. Increased knowledge about the clinical characteristics of CDT may lead to better possibilities of employing CDT as a biochemical alcohol marker of risky and heavy alcohol consumption and thus creating better opportunities for prevention and intervention of alcohol dependence. The aim of this thesis was to contribute to this knowledge by investigating the properties of CDT in different populations and evaluating the sensitivity and specificity of different analytical methodologies for analysis of CDT.

Serum samples (n=1387) from subjects originating from five different countries were analyzed with a HPLC candidate reference method for CDT. In non-drinkers there were minimal differences in the serum transferrin glycoform pattern with respect to different ethnicity, gender, age and BMI. When evaluating disialotransferrin, the primary glycoform in CDT, with respect to the same categories, no clinically significant differences were detected. Furthermore the overall test accuracy for identification of heavy drinkers (>210 g ethanol/week for men and >140 g ethanol/week for women) showed no gender difference.

Serum samples (n=178) were analyzed from subjects with clinical or pharmacological factors previously reported to cause false-positive CDT levels. Only ~5% showed a relative disialotransferrin level exceeding the upper limit for the reference interval, leading to a conclusion that earlier reports on reasons for false positive CDT values are linked with the methodology used rather than with true physiological influences.

When compared with a HPLC candidate reference method, the Bio-Rad %CDT HPLC test and the CEofix™ CDT assay proved to be appropriate for confirmatory and routine

%CDT testing, showing an overall good correlation and agreement.

The HPLC candidate reference method could readily be used for preliminary diagnosis of CDG and for assignment of cases to either CDG-I or CDG-II.

(8)

(9)

LIST OF PUBLICATIONS

I. Bergström JP, Helander A.

Influence of alcohol use, ethnicity, age, gender, BMI and smoking on the serum transferrin glycoform pattern: implications for the use of carbohydrate- deficient transferrin (CDT) as alcohol biomarker. Clin Chim Acta 2007;In press.

II. Bergström JP, Helander A.

Clinical characteristics of carbohydrate-deficient transferrin (CDT) measured by HPLC: sensitivity, specificity, gender differences and relationship with other markers of prolonged alcohol abuse. Submitted 2007.

III. Bergström JP, Helander A.

HPLC evaluation of clinical and pharmacological factors reported to cause false-positive carbohydrate-deficient transferrin (CDT) levels. Submitted 2007.

IV. Helander A, Bergström JP.

Determination of carbohydrate-deficient transferrin in human serum using the Bio-Rad %CDT by HPLC test. Clin Chim Acta 2006;371:187-190.

V. Helander A, Wielders JPM, te Stroet R, Bergström JP.

Comparison of HPLC and capillary electrophoresis for confirmatory testing of the alcohol misuse marker carbohydrate-deficient transferrin. Clin Chem 2005;51:1528-1531.

VI. Helander A, Bergström JP, Freeze H.

Testing for congenital disorders of glycosylation by HPLC measurement of serum transferrin glycoforms. Clin Chem 2004;50:954-958.

The original articles (I, IV, V and VI) have been printed with permission from the publishers.

(10)

LIST OF ABBREVIATIONS

5-HIAA 5-Hydroxyindole-3-Acetic Acid 5-HTOL 5-Hydroxytryptophol

AED Antiepileptic Drug

ALT Alanine Aminotransferase

AST Aspartate Aminotransferase

AUC Area Under the Curve

BMI Body Mass Index

CDG Congenital Disorders of Glycosylation

CDT Carbohydrate-Deficient Transferrin

CE Capillary Electrophoresis

CRP C-Reactive Protein

CV Coefficient of Variation

CF Cystic Fibrosis

DST Disialotransferrin

EtG Ethyl Glucuronide

EtS Ethyl Sulfate

GC-MS Gas Chromatography-Mass Spectrometry

GGT Gamma Glutamyltransferase

HPLC High-Performance Liquid Chromatography

FeNTA Ferric Nitrilotriacetic Acid

ID Inner Diameter

IFCC International Federation of Clinical Chemistry and Laboratory Medicine

IEF Isoelectric Focusing

ISBRA International Society for Biomedical Research on Alcoholism LC-MS Liquid Chromatography-Mass Spectrometry

LC-MS/MS Liquid Chromatography-Mass Spectrometry/Mass Spectrometry

LOD Limit of Detection

LOQ Limit of Quantification

NIAAA National Institute on Alcohol Abuse and Alcoholism PEth Phosphatidylethanol

RIA Radioimmunoassay

ROC Receiver-Operating Characteristic

U/L Units/Liter

WHO World Health Organization

(12)

(13)

1 INTRODUCTION

1.1 BIOCHEMICAL ALCOHOL MARKERS

Alcoholic beverages, and the pleasures and problems they cause, have been known to mankind since the beginning of recorded history. The use and abuse of alcohol is the cause of immense costs in most of the western societies, with a rather big proportion of its citizens being alcohol dependent or consuming harmful amounts of alcohol (1).

Therefore, screening for alcohol-related problems is an important task to detect alcohol dependence or risky alcohol consumption behavior, and is usually accomplished by structured interviews and/or laboratory tests. The most widely employed alcohol questionnaires over the last years are the Alcohol Use Disorder Identification Test (AUDIT) (2-4) and CAGE (5, 6), both of them relying on the patient’s self-reported alcohol consumption and the associated problems. The self-reported amount of alcohol consumed is however known to be frequently underreported, therefore leading to possible under-diagnosis of alcohol related problems (7). Laboratory tests, or biochemical markers of alcohol use and abuse, offers more objective methods to monitor or detect excessive alcohol consumption.

1.1.1 Markers of acute alcohol consumption

The biochemical markers of acute alcohol consumption are laboratory tests developed in order to detect single intakes of alcohol, typically in the last 0-48 hours prior to sampling, depending on the sensitivity of the test and the amount of alcohol consumed.

These markers provide a powerful tool within areas such as forensic medicine, criminal applications, occupational medicine, and can also be used for confirmation of abstinence in outpatient treatment (8-10).

1.1.1.1 Ethanol

After alcohol consumption, ethanol can be determined in breath or body fluids (11).

The major drawback with ethanol as a biochemical marker is that the human body rapidly metabolizes even large amounts of ethanol in typically <12 hours, making the time range for detection very small (12). Another drawback is that the ethanol concentration does not provide information about hazardous drinking patterns and eventual alcohol problems.

(14)

2

1.1.1.2 EtG and EtS

Ethyl glucuronide (EtG) and ethyl sulfate (EtS) are water-soluble, stable, conjugated direct metabolites of ethanol (13, 14). EtG is formed from a minor part of the ingested ethanol (<0.1%) via reaction with uridine-5-diphospho-β-glucuronic acid (15). EtS is also formed from a minor part of the ingested ethanol (<0.1%) after reaction between ethanol and sulfate by sulfotransferase (16). Both EtG and EtS are excreted in the urine and, sharing approximately the same elimination profiles, are washed out from the body at a much slower rate than ethanol itself, and can often be detected days after alcohol consumption (17, 18). Therefore EtG and EtS offer an extended window for assessment of single ethanol intake. Urinary EtG and EtS can today be determined with a range of analytical methodology (14, 19-21).

1.1.1.3 5-HTOL/5-HIAA

During normal conditions, the major part of all serotonin is metabolized to 5- hydroxyindole-3-acetic acid (5-HIAA) with only a smaller part (<1%) forming into 5- hydroxytryptophol (5-HTOL). However, alcohol consumption results in a shift towards an increased formation of 5-HTOL, and consequently an elevated ratio of 5-HTOL/5- HIAA can be detected in urine for several hours after all ethanol is cleared from the body. This makes it possible to detect a single alcohol intake for a much longer period of time (22-24). 5-HTOL and 5-HIAA have traditionally been analyzed by gas chromatography-mass spectrometry (GC-MS) and high-performance liquid chromatography (HPLC), respectively (25, 26), but lately there has been suggestions to replace 5-HTOL as target analyte and instead analyze 5-HTOL glucuronide (GTOL) together with 5-HIAA with a highly sensitive and specific LC-MS/MS method (27).

1.1.2 Markers of chronic alcohol consumption

Biochemical markers of chronic alcohol consumption are laboratory tests that can be used for identification of a chronic, sustained consumption that has been going on for typically over a week (28). These markers are currently in use in areas such as traffic medicine, occupational medicine and in various clinical settings as indicators of a chronic alcohol misuse (29). The marker in focus in this thesis, CDT, is presented in detail in section 1.2.

(15)

1.1.2.1 GGT

Gamma-glutamyltransferase is an intracellular liver enzyme that in response to acute hepatocellular damage, e.g. following prolonged alcohol abuse, can leak into the blood.

The reported alcohol consumption needed for an elevated serum level of GGT have shown great variations. Still, it is today maybe the most used laboratory marker for chronic alcohol consumption (30, 31). The specificity of GGT for chronic alcohol abuse is very poor. Some of the reported factors that cause an elevated GGT level are smoking, male gender, obesity, age, non-alcoholic liver diseases, medication and diabetes (28, 32, 33). The half-life of GGT is between 2-3 weeks (34).

1.1.2.2 AST and ALT

Elevated serum concentrations of the liver enzymes aspartate aminotransferase (AST) and alanine aminotransferase (ALT) are indicators of non-specific liver dysfunction, and concentrations are frequently heightened also in alcoholic patients (35, 36). These liver enzymes leaks into the blood following hepatocellular damage in the same way as GGT. The sensitivity and specificity for identification of alcohol misuse are in most cases reported low or moderate for both AST and ALT (37). An elevated AST level can also arise from non-hepatic sites (e.g., heart and muscles), and from conditions such as non-alcoholic liver disease, myocardial infarction and skeletal muscle trauma (38). The half-life of AST and ALT is reported to be 2-3 weeks (37).

1.1.2.3 PEth

Phosphatidylethanol (PEth) is an abnormal phospholipid formed exclusively by the action of phospholipase D in the presence of ethanol and is considered a promising new biochemical marker of chronic alcohol misuse (39, 40). PEth has a half-life of ~4 days, making it possible to detect PEth in blood of chronic alcohol users up to 3 weeks after alcohol withdrawal. The amount of PEth in blood is reported to be highly correlated with the past alcohol intake in alcohol abusers (41). The sensitivities and specificities that are reported so far are high, but this biomarker is in need of further evaluation within different clinical settings. The current methods for analysis of PEth include HPLC with evaporative light scattering detection (42, 43) and LC-MS (44).

1.2 CARBOHYDRATE-DEFICIENT TRANSFERRIN (CDT)

Transferrin is a glycoprotein mainly synthesized in the hepatocytes(45). It is the major Fe³⁺-transport protein in the body, with a normal serum concentration range of 1.9-3.3

(16)

4

g/L (46). Under prolonged heavy alcohol consumption, the microheterogeneity of the glycoform pattern of transferrin changes towards a higher proportion of so called carbohydrate-deficient transferrin (CDT), making it useful as a laboratory marker of sustained alcohol abuse (47, 48). This discovery was first reported in 1976 by Stibler and Kjellin, after studies on the transferrin glycoform pattern in cerebrospinal fluid from alcoholics (49).

1.2.1 Transferrin structure

Transferrin consists of three different sub-structural domains: a single polypeptide chain with 679 amino acids, two Fe³⁺ ion-binding sites, one within the N-terminal domain and one within the C-terminal domain, and two N-linked complex oligosaccharide chains (Figure 1) (45). There are many different polypeptide chain variants reported (~40) (50). The most common polypeptide chain is the homozygous transferrin C variant, with the subtype transferrin C1 being the most frequent occurring in Caucasians (>95%). Heterozygous transferrin BC, CD and other variants seldom show a prevalence over 1% of the population. Another microheterogeneity is the varying iron load of transferrin in blood. With a normal iron saturation of transferrin in blood (~30%) there are transferrin molecules with none, one or two Fe³⁺ions present simultaneously. The microheterogeneity between the oligosaccharide chains is complex: the chains can be biantennary, triantennary or even tetraantennary, each antennary usually terminating in a sialic acid residue, the total number of residues traditionally giving name to the glycoform (51).

N-Acetylglucosamine Mannose

Galactose Sialic Acid

Fe³⁺ Fe³⁺

N C

N-Acetylglucosamine Mannose

Galactose Sialic Acid

Fe³⁺ Fe³⁺

N C

Figure 1. Detailed structure of the main transferrin glycoform tetrasialotransferrin that normally represents ~80% of the total serum transferrin in healthy adults. The line between N and C represents the single polypeptide chain.

(17)

For example, the most common transferrin glycoform in healthy adults, tetrasialotransferrin (~80%; Figure 1), consists of two disialylated N-linked glycans, e.g. a total of four terminal sialic acid residues, hence its structural name. Other normally occurring transferrin glycoforms in healthy adults are disialotransferrin (<2%), trisialotransferrin (~4%), pentasialotransferrin (~14%) and hexasialotransferrin (~1%) (52-54).

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Trisialotransferrin ~4% Pentasialotransferrin ~14% Hexasialotransferrin ~1%

Asialotransferrin ( alcohol)

Monosialotransferrin ( with high trisialotransferrin)

Disialotransferrin ~1%

( alcohol) Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Fe³⁺ Fe³⁺

N C

Trisialotransferrin ~4% Pentasialotransferrin ~14% Hexasialotransferrin ~1%

Asialotransferrin ( alcohol)

Monosialotransferrin ( with high trisialotransferrin)

Disialotransferrin ~1%

( alcohol)

Figure 2. Detailed structure of the most common minor transferrin glycoforms and their relative occurrence in serum from healthy adults. Prolonged alcohol abuse elevates the relative level of disialotransferrin. In subjects with high levels of disialotransferrin, sometimes asialotransferrin can be detected. Monosialotransferrin can often be seen in individuals with a genetically high trisialotransferrin level.

Under prolonged alcohol abuse, the glycoform disialotransferrin increases relatively to the total transferrin, and in subjects with high levels of disialotransferrin, sometimes asialotransferrin can be detected (Figure 2) (55-57). Monosialotransferrin can sometimes be detected in subjects with genetically very high levels of trisialotransferrin (57).

1.2.2 CDT pathomechanisms

The detailed pathomechanisms behind the increase of CDT during prolonged alcohol consumption are not fully evaluated yet. There is however some evidence that ethanol or its metabolite acetaldehyde interferes in the transferrin N-glycan chain synthesis in the Golgi apparatus. The activities of the glycoprotein glycosyltransferases galactosyltransferase and N-acetylglucosaminyltransferase were lower in serum from alcoholics than in subjects from a healthy control group (58). In alcohol fed rats,

(18)

6

ethanol lowers the levels of sialyltransferase mRNA, and consequently lowers the sialyltransferase activity (59). In another study on alcohol fed rats, an increase of sialidase activity in liver plasma was seen (60). Accordingly with these studies, ethanol or maybe more likely acetaldehyde seems to hamper glycoprotein glycosyltransferase activity in general, which in combination with an increased sialidase activity give rise to increased levels of CDT .

1.2.3 CDT sensitivity

The sensitivity of CDT as a marker of chronic alcohol abuse is reported with great variation depending on factors such as study population, mean daily alcohol intake and drinking patterns (47, 61-63). When reviewed, a sensitivity of 30-50% for women and 50-70% for men seemed to be of average (48). The amounts of alcohol reported to increase the CDT value over the reference interval for social drinkers have been debated, but it seems that at least 50-80 grams of ethanol per day for 2 weeks or longer is necessary to produce elevated CDT concentrations (64-66). Factors that have been reported to affect the sensitivity of CDT include age, gender, drinking patterns, body mass, hypertension and smoking, to name a few (67-70). It is however unclear how much the often aged and less specific analysis methodology (e.g., CDTect™) used in many reports on this issue have influenced the reliability of the results (71-74). The half-life of CDT is reported to be 1.5-2 weeks, and following alcohol abstinence a normalization of the CDT level occurs within approximately 4 weeks (47).

1.2.4 CDT specificity

The single biggest advantage of using CDT as a marker of chronic alcohol abuse is that of all laboratory tests available CDT is the most specific (48). That said, there have over the years been numerous reports published on clinical conditions or other factors that could produce false-positive (i.e., non alcohol related) CDT results (70). These reports identify, among other things, genetic transferrin variants, congenital disorders of glycosylation (CDG), different liver diseases, iron deficiency, haemochromatosis, hypertension, cystic fibrosis (CF), various medication and sepsis to be factors that elevate CDT values (75-83). However, these findings are often based on very small subject groups or with a methodology for CDT analysis that is now out of date (e.g., CDTect), making it hard to estimate the true impact of the different clinical factors investigated on CDT levels.

(19)

1.2.5 Congenital disorders of glycosylation (CDG)

Congenital disorders of glycosylation (CDG) are a group of rare hereditary diseases characterized by defects in the synthesis of the glycan moieties of glycoproteins or other glycoconjugates caused by mutations in the genes coding for enzymes involved in the glycoprotein synthesis (84, 85). The two main types of protein glycosylation are N- glycosylation and O-glycosylation, where the latter will not be further discussed here.

In general, N-glycosylation consists of an assembly pathway (in cytosol and endoplasmatic reticulum) followed by a processing pathway (in endoplasmatic reticulum and Golgi). The two main types of N-glycosylation CDG are defects in glycoprotein synthesis during the assembly pathway (CDG-I) or during the processing pathway (CDG-II). These two main groups are further sub-categorized, so that each defective gene correspond to one specific CDG (e.g. CDG-Ia, CDG-Ib, CDG-IIa etc.), and today there are ~10 CDG-I and ~5 CDG-II categorized (86). N-Glycosylation defects affects primarily the nervous system with the clinical expression including psychomotor, growth and mental retardation, and the effect varies between being extremely severe to very mild. Since the clinical symptoms often are unspecific, CDG are probably under-diagnosed. The main laboratory tool used to identify and categorize N-glycosylation CDG has traditionally been isoelectric focusing (IEF) of the transferrin glycoform pattern (87), but mass spectrometry techniques for identification of CDG are also available (88).

1.2.6 Analysis of CDT

CDT was originally defined as the three transferrin glycoforms with a pI > 5.7 after IEF, i.e. asialo-, monosialo- and disialotransferrin (47). IEF was the first reference method for serum transferrin glycoform analysis, and is still used today as a reference method for identification of genetic transferrin variants and CDG, much because of its high selectivity. The major drawback with IEF is difficulties with quantitative CDT analysis (89).

Jeppsson et al. developed the first HPLC method for identification and quantification of transferrin glycoforms in 1993 (100), and it has later been followed by commercial HPLC kits (101). Some of the general advantages with HPLC methods for CDT analysis is the ability to identify genetic transferrin variants, the specificity (i.e., only disialotransferrin is measured) and that HPLC is a well known methodology in many

(20)

8

clinical laboratories. Later Helander et al. developed an improved HPLC method that is now a suggested candidate reference method for CDT analysis (102).

Over the years there have been several capillary electrophoresis (CE) and capillary zone electrophoresis methods available for analysis of CDT, with gradually increasing sensitivity and selectivity, and lately some methods have been commercially available for routine use (90-92).

The immunoassay CDTect RIA (Pharmacia & Upjohn, Uppsala, Sweden) was the first commercial test kit for CDT analysis and was launched 1993 (93, 94). This method, and later its follower the %CDT immunoassay (Axis-Shield ASA, Oslo, Norway) have both been available in a number of different variants for different applications (details on methodology in section 3.3.3 and 3.3.4) (95-98). Notably, both are measuring the sum of asialo-, monosialo-, disialo- and sometimes a fraction of trisialotransferrin making them unspecific and not capable of identifying genetic transferrin variants that are known to produce false-high or false-low CDT values (99).

Recently the first direct immunoassay for CDT, N Latex CDT (Dade Behring, Marburg, Germany), became available. Measurement with N Latex CDT is based on a monoclonal antibody that recognizes transferrin glycoforms that lack one or both of the complete N-glycans (i.e., asialo-, monosialo- and disialotransferrin) (103). The difference compared with other methods that measures the sum of asialo-, monosialo- and disialotransferrin is that the monoclonal antibody used in N Latex CDT does not discriminate between different genetic transferrin variants, and thus genetic variants do not interfere with measurements, making it a more specific method.

During the last years, reports on analysis of transferrin with mass spectrometry have begun to be published, providing valuable information on the structural microheterogeneity of the transferrin glycans (104-106).

(21)

2 GENERAL AIMS OF THE THESIS

• To evaluate any baseline differences in the transferrin glycoform pattern in relation to ethnicity, age, gender, body mass index and smoking in a large study population from five different countries.

• Within the same study population evaluate the sensitivity and specificity of serum disialotransferrin as a biochemical marker for chronic alcohol abuse in relation to ethnicity, gender, age and body mass index, and to compare the performance of disialotransferrin with the traditional alcohol biomarkers GGT and AST.

• Evaluate clinical and pharmacological factors previously reported to cause false-positive CDT levels.

• Evaluate two potential confirmatory and routine methods for CDT testing, the Bio-Rad %CDT by HPLC test and the CEofix™ CDT assay.

• To study the usefulness of a HPLC candidate reference method for the detection and preliminary diagnosis of CDG.

(22)

10

3 MATERIALS AND METHODS

3.1 STUDY POPULATIONS

The serum samples analyzed in Paper I and II were all from subjects participating in the WHO/ISBRA Study on State and Trait Markers of Alcohol Use and Dependence, an international multicenter study (107). The serum samples analyzed in Paper III, IV and V were anonymous leftover volumes from the Karolinska University Hospital or (in Paper V) also from Meander Medical Center, Amersfoort, The Netherlands. In Paper VI, the serum samples from subjects with different CDG types were from The Burnham Institute, La Jolla, California, USA, all other serum samples were de- identified leftover volumes from the Karolinska University Hospital.

3.1.1 The WHO/ISBRA Collaborative Project (Paper I-II)

The WHO/ISBRA Study on State and Trait Markers of Alcohol Use and Dependence was established in 1988. The aim with the study was to assess and compare markers of recent alcohol use and also of the trait of alcohol dependence in a multicenter trial (107). The study group consisted of representatives from the World Health Organization (WHO), the International Society for Biomedical Research on Alcoholism (ISBRA) and the National Institute on Alcohol Abuse and Alcoholism (NIAAA). The samples used within this study were collected from both the community and from alcohol treatment services in 5 countries: Australia (Sydney), Brazil (São Paolo), Canada (Montreal), Finland (Helsinki) and Japan (Sapporo). A total of 1863 subjects aged over 18 were recruited, 67% of the subjects men and 33% women. In Australia and Finland, only men were recruited. All participants were extensively characterized for socio-demographic, health and lifestyle factors, and thoroughly interviewed about their drinking habits by trained researched staff using the structured WHO/ISBRA Interview Schedule (108). Based on this information, each subject was classified as either “non-drinker” = totally abstinent, or one who drinks alcohol on no more 6 special occasions per year (e.g., birthdays), and no more than 15 g ethanol on each occasion; “light/moderate drinker” = drinks at least once per month but <210 g ethanol/week for men and < 140 g ethanol/week for women, but no past treatment for alcohol-related problems; “heavy drinker” = drinks > 210 g ethanol/week for men and

> 140 g ethanol/week for women, but no past treatment for alcohol-related problems; or

“under alcohol treatment” = a person currently receiving treatment for alcohol

(23)

Table 1. Demographic data on the country of origin, gender, age, ethnicity and drinking status of the WHO/ISBRA collaborative project study population used in Paper I and II.

Australia Brazil Canada Finland Japan Total

Serum samples, n 239 385 432 200 131 1387

Gender, n (%)

Male 239 (100%) 211 (54.8%) 231 (53.5%) 200 (100%) 65 (49.6%) 946 (68.2%) Female 0 174 (45.2%) 201 (46.5%) 0 66 (50.4%) 441 (31.8%)

Age, years

Mean ± SD 37.0 ± 13.1 35.5 ± 11.7 36.8 ± 12.0 39.2 ± 11.4 37.5 ± 11.3 36.9 ± 12.0

Median 36 33 36 38 35 35

Range 18-65 18-60 18-60 18-60 21-59 18-65

Age <31 (%) 87 (36.4%) 161 (41.8%) 156 (36.1%) 57 (28.5%) 45 (34.4%) 506 (36.5%) Age 31-40 51 (21.3%) 95 (24.7%) 95 (22.0%) 53 (26.5%) 34 (26.0%) 328 (23.6%) Age 41-50 51 (21.3%) 73 (19.0%) 110 (25.5%) 52 (26.0%) 30 (22.9%) 316 (22.8%) Age >50 50 (20.9%) 56 (14.5%) 71 (16.4%) 38 (19.0%) 22 (16.8%) 237 (17.1%)

Ethnicity, n (%)

White 206 (86.2%) 268 (69.6%) 382 (88.4%) 198 (99.0%) 1 (0.8%) 1055 (76.0%) Asian/Indian 24 (10.0%) 12 (3.1%) 23 (5.3%) 0 129 (98.5%) 188 (13.6%)

Black 5 (2.1%) 58 (15.1%) 9 (2.1%) 0 0 72 (5.2%)

Pacific 1 (0.4%) 0 2 (0.5%) 0 0 3 (0.2%)

American Indian 0 1 (0.3%) 2 (0.5%) 0 1 (0.8%) 4 (0.3%) Other/Unknown 3 (1.3%) 46 (11.9%) 14 (3.2%) 2 (1.0%) 0 65 (4.7%)

Drinking status, n (%)

Non-drinker 65 (27.2%) 90 (23.4%) 132 (30.6%) 61 (30.5%) 12 (9.2%) 360 (25.9%) Light drinker 97 (40.6%) 180 (46.8%) 189 (43.8%) 117 (58.5%) 106 (80.9%) 689 (49.7%) Heavy drinker 77 (32.2%) 115 (29.9%) 111 (25.7%) 22 (11.0%) 13 (9.9%) 338 (24.4%)

dependence. However, in Paper I and II, the samples from the category with subjects undergoing alcohol treatment were excluded from determination of CDT with the HPLC candidate reference method, since these samples did not meet the criteria of the aims of the studies. In the remaining three drinking categories some serum samples were excluded because of too small sample volumes. In the end, serum samples from a total of 1387 subjects were analyzed with the HPLC candidate reference method for CDT (102, 109). Detailed information about the study population concerning gender, age, ethnicity and drinking status is shown in Table 1. Based on the HPLC transferrin pattern, 1362 (98%) subjects were indicated to be of transferrin C phenotype (a further sub-classification into C1, C2 and C3 was not performed). The other 25 samples were of different genetic variants and were excluded from the calculations, due to overlapping peaks causing unreliable quantification of individual glycoforms (57).

(24)

12

3.1.2 Clinical samples (Paper III-V)

The serum samples in these studies were all anonymous leftover volumes from different departments at the Karolinska University Hospital, with the exception of 37 serum samples in Paper V that were collected at the Meander Medical Center, Amersfoort, The Netherlands. In Paper III serum samples were collected from subjects with various clinical conditions or undergoing medication with drugs previously reported to cause non-alcohol related elevations of CDT. The samples were from subjects with end-stage liver disease (n=50), diabetes mellitus type 2 (n=46), an elevated C-reactive protein (CRP) level (n=15), medication with enzyme or non- enzyme inducing antiepileptic drugs (AED; n=43), and cystic fibrosis (CF; n=24). In Paper IV the serum samples used were two human serum pools (containing ~1.3% and

~2.6% disialotransferrin, respectively, by HPLC), 150 surplus clinical sera selected from routine samples pool in order to cover the entire measurement range from low/normal to highly elevated disialotransferrin values (0.7-22% by HPLC), as well as 18 serum samples with genetic transferrin variants. In Paper V, 42 anonymous surplus sera were collected from the routine samples pool to cover the measuring range from low/normal to highly elevated %CDT levels (1.3-24.2% by %CDT immunoassay), as well as some genetic transferrin variants. In addition, 37 serum samples for this study were collected at the Meander Medical Center, Amersfoort, The Netherlands, as mentioned above.

3.1.3 CDG samples (Paper VI)

In this study, serum samples from 9 patients with biochemically and/or genetically confirmed CDG type I (a, b and g subtypes) and 4 with undefined CDG type IIx defects were obtained from The Burnham Institute, La Jolla, California, USA. Serum samples used for comparison from 42 children and adolescents (0-3 weeks, n=13; 1-9 months, n=11; 1-18 years, n=18), 132 adult social drinkers and 74 chronic alcohol misusers were all randomly selected leftover volumes from routine samples at the Karolinska University Hospital (110).

3.2 HPLC CANDIDATE REFERENCE METHOD (PAPER I-VI)

The HPLC candidate reference method is an improved HPLC method for measurement of CDT in serum based on anion-exchange chromatographic separation of the different transferrin glycoforms followed by photometric detection (Figure 3) (102). The method

(25)

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

d

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Disialotransferrin Trisialotransferrin Tetrasialotransferrin Pentasialotransferrin Hexasialotransferrin

Absorbancy(mAU)

1.0

0.0

a

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

Asialotransferrin

b

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

Monosialotransferrin

c

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

d

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

a

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

Asialotransferrin

b

Time (min)

5 10 15 20 25

0.2 0.4 0.6 0.8

Absorbancy(mAU)

1.0

0.0

Monosialotransferrin

c

Figure 3. Shown in (a) is a control serum sample from a light drinker with the predominant transferrin C homozygote variant; (b), transferrin C serum from a heavy drinker with increased disialotransferrin (3.2% of total AUC for transferrin) and detectable (~0.4%) asialotransferrin; (c), transferrin C serum from a person with high trisialotransferrin (8.4%) and a measurable (~0.5%) monosialotransferrin; (d), transferrin BC heterozygote, a genetic transferrin variant. Figure edited from (102).

is based on earlier HPLC methodology developed by Jeppson et al. (100). The primary target molecule for this method is disialotransferrin, as recommended by the IFCC working group on CDT standardization (111). Transferrin was first iron-saturated by mixing 10-100 µL (depending on the available volume) of serum at a volume ratio of 5:1 with FeNTA (final concentration, 1.7 mmol/L), a well-known transferrin iron donor (112). Lipoproteins were precipitated by mixing the iron-saturated sample 6:1 (by volume) with dextran sulfate and CaCl2 (1.4 mg/L and 70 mmol/L, respectively). The samples were mixed gently and left in the cold (~5 ºC) 30-60 minutes, and then centrifuged at 3500g for 10 minutes. The clear supernatant was diluted fivefold with water and then transferred to glass HPLC vials. If available, the injection volume was 200 µL. The HPLC system consisted of an Agilent 1100 Series Liquid Chromatograph, equipped with a quarternary pump and degasser, thermostated autosampler (4 ºC) and column compartment (25 ºC), a G1365B multiple wavelength detector and ChemStation software. The transferrin glycoforms were separated by use of a SOURCE^® 15Q PE 4.6/100 anion-exchange chromatography column (Amersham Biosciences, Uppsala, Sweden) with a linear salt gradient elution at a flow rate of 1

(26)

14

mL/min. Quantification of the different transferrin glycoforms relied on the selective absorbance of the iron-transferrin complex at 470 nm. Baseline integration was used for all transferrin peaks, typically from monosialotransferrin (if visible) or disialotransferrin to hexasialotransferrin, with asialotransferrin (if visible) integrated separately. With this method, the relative amount of any single glycoform (e.g., disialotransferrin) or combination of glycoforms to total transferrin (all quantified glycoforms) are measured in terms of the relative area under the curve (%AUC). The typical LOD and LOQ for the transferrin glycoforms were ~0.05% and 0.10%, respectively, of total serum transferrin in the normal transferrin concentration range (reference interval 1.9-3.3 g/L) (46). The intra- and inter-assay imprecision (CV) determined for serum samples containing normal and elevated disialotransferrin (range 1.0-5.6%) are below 5% (102).

3.3 OTHER METHODS FOR CDT ANALYSIS

Apart from the HPLC candidate reference method described above, which served as the principal analysis method during the work with this thesis, other methods for analysis of CDT were also used; Bio-Rad %CDT by HPLC Reagent Kit (Paper IV) (101), the CEofix™ CDT Assay for CE analysis (Paper V) (113), the %CDT immunoassay (Paper V) (114) and the CDTect™ radioimmunoassay (RIA) (94) method (Paper II).

3.3.1 Bio-Rad %CDT by HPLC Reagent Kit (Paper IV)

Bio-Rad %CDT by HPLC Reagent Kit (Bio-Rad, Munich, Germany) is a new commercial application for HPLC that allows separation of asialo-, disialo-, trisialo, tetrasialo- and pentasialotransferrin in serum within ~6 min (total analysis time for 1 sample is ~10 min) (101). The method measures individual glycoforms in proportion to total transferrin using baseline integration. The samples were prepared following the manufacturers instructions for the reagent kit. The transferrin glycoforms were separated on a gradient HPLC system with anion-exchange cartridge (guard, 5*4.6 mm ID; analytical, 30*4 mm ID), followed by specific measurement of the iron-transferrin complex at 460 nm. With this HPLC method monosialo- and disialotransferrin are co- eluating, thus CDT is defined as the sum of asialo-, monosialo- and disialotransferrin, and according to the manufacturer’s instructions the upper 95% confidence limit (mean + 2 SD) for this kit is 1.7%.

(27)

3.3.2 CEofix™, CE, (Paper V)

The CEofix™ CDT Assay (Analis, Namur, Belgium) is a commercial CE method for separation of individual transferrin glycoforms and for determination of CDT (92, 115, 116). The serum samples were first iron saturated. The fused capillary had an ID of 50 µm with a total length of 50 cm and had 40 cm to the detector. Before separation, the capillary is dynamically coated by rinsing the capillary with an initiator buffer containing a polycation and with the separation buffer containing a polyanion, creating a dynamic double coating. Electrophoresis was carried out with an overall separation time of ~6 min with ultraviolet detection at 214 nm on a Beckman Coulter P/ACE 5000, according to the manufacturer’s instructions. The relative amounts of single transferrin glycoforms were calculated from peak areas by valley-to-valley integration.

All analysis with CEofix™ was performed at the Department of Clinical Chemistry, Meander Medical Center, Amersfoort, The Netherlands (117).

3.3.3 %CDT immunoassay (Paper V)

The %CDT immunoassay (Axis-Shield ASA, Oslo, Norway) is a heterogeneous immunoassay with column separation followed by turbidimetric measurement that measures CDT, i.e. the sum of asialo-, monosialo-, disialo-, and a portion (~50%) of trisialotransferrin as the relative amount to the total transferrin (114) . The serum transferrin in the sample was saturated with Fe³⁺ and applied to an ion-exchange column where the different glycoforms are separated due to differences in charge. The CDT glycoforms (as defined above) were eluted and determined by turbidimetric measurement after formation of an immune complex with anti-transferrin antibodies.

The total transferrin content in the sample was determined separately using the same anti-transferrin antibodies, and the %CDT concentration could then be calculated as the ratio between CDT and total transferrin. According to the manufacturer the cut-off is 2.6%.

3.3.4 CDTect™ (Paper II)

The CDTect RIA method (Pharmacia & Upjohn, Uppsala, Sweden) is based on column separation followed by a double antibody RIA procedure measuring CDT (94), i.e. the sum of asialo-, monosialo, part of disialo-, and traces of trisialotransferrin (71). The serum transferrin in the sample was iron saturated with ferric citrate solution, and elution buffer was added. An aliquot of the sample-buffer mixture was applied to the anion-exchange microcolumns, and the CDT glycoforms (as defined above) were after

(28)

16

separation obtained in the column effluxes. Finally, the CDT from the eluate was quantified with a double antibody RIA procedure and expressed as an absolute value in units/liter (U/L). According to the manufacturer the cut-off value is 20 U/L for men and 26 U/L for women. Analysis with CDTect™ of the serum samples in this study was already performed at an earlier point at the Alcohol Laboratory, Karolinska University Hospital, Stockholm, Sweden, when the work with Paper II began.

3.4 ANALYSIS OF GGT AND AST (PAPER II)

Gamma glutamyltransferase (GGT) and Aspartate Aminotransferase (AST) were assayed by reflectance spectrophotometry using a Vitros 250 Analyser (Ortho Clinical Diagnostics, Rochester, NY). Analysis of plasma samples for GGT and AST was already performed at an earlier point at the laboratories of ALKO and KTL, Helsinki, Finland, when the work with Paper II began (118).

3.5 STATISTICS

Statistical calculations were performed using the Student-Newman-Keuls test for pairwise comparisons, a T-test (parametric) when the examined groups showed a Gaussian distribution or Wilcoxon (non-parametric) if not. For statistical analysis of correlations, Pearson´s correlation coefficient (parametric) or Spearman´s coefficient of rank correlation (nonparametric) were used. All statistical analysis was performed with MedCalc statistical software.

(29)

4 RESULTS

4.1 PAPER I

The relative amounts of serum transferrin glycoforms in non-drinkers (n=358), light/moderate drinkers (n=677) and heavy drinkers (n=327) within the WHO/ISBRA Study on State and Trait Markers of Alcohol Use and Dependence are found in Table 2.

The major differences in the transferrin glycoform pattern between the three drinking categories were the higher levels of disialo- (P<0.0001) and trisialotransferrin (P<0.001) in light/moderate drinkers and heavy drinkers compared with non-drinkers.

Also, there were no detectable amounts of asialotransferrin in non-drinkers, but in 2.2%

of the light/moderate drinkers and in 18.3% of the heavy drinkers.

Table 2. Distribution of serum transferrin glycoforms in different drinking categories.

Non-drinkers Light/Moderate drinkers Heavy drinkers

Serum samples, n 358 677 327

Transferrin glycoform

Unknown peak 0.63 ± 0.65 (n = 36) 0.39 ± 0.32 (n = 39) 0.65 ± 0.47 (n = 13) 0.47, 0.11-3.26* 0.22, 0.10-1.16 0.46, 0.11-1.53 Asialotransferrin (n = 0) 0.40 ± 0.23 (n = 15) 0.53 ± 0.48 (n = 60)

0.30, 0.16-0.90 0.39, 0.10-2.43 Monosialotransferrin 0.17 ± 0.06 (n = 200) 0.18 ± 0.09 (n = 323) 0.22 ± 0.12 (n = 197)

0.16, 0.10-0-58 0.16, 0.10-0.69 0.19, 0.10-0.90 Disialotransferrin 1.14 ± 0.19 1.34 ± 0.55 2.25 ± 1.57

1.14, 0.54-1.87 1.22, 0.50-5.87 1.72, 0.86-10.6 Trisialotransferrin 4.02 ± 1.10 4.38 ± 1.35 4.89 ± 1.46

3.89, 1.78-10.0 4.16, 1.45-10.4 4.64, 2.05-12.7 Tetrasialotransferrin 79.9 ± 1.75 79.7 ± 2.04 78.1 ± 3.23

80.1, 74.0-85.5 80.0, 68.3-84.7 78.7, 61.1-84.4 Pentasialotransferrin 14.0 ± 1.48 13.8 ± 1.64 13.8 ± 1.99

13.9, 9.52-19.4 13.7, 9.27-26.7 13.7, 9.37-22.2 Hexasialotransferrin 0.73 ± 0.36 (n = 348) 0.64 ± 0.35 (n = 646) 0.75 ± 0.45 (n = 311)

0.67, 0.11-2.72 0.59, 0.10-3.15 0.68, 0.11-2.80

*Mean ± SD, median, range.

(30)

18

When comparing the correlations between the serum transferrin glycoforms for all samples, the strongest positive correlation (r=0.80) was obtained between disialo- and asialotransferrin, whereas disialo- and asialotansferrin were negatively associated with tetrasialotransferrin. These associations were dependent on the alcohol consumption level. The level of trisialotransferrin was positively correlated with monosialotransferrin but this association was indicated to be unrelated to the alcohol consumption level.

When the relative amounts of serum transferrin glycoforms in non-drinkers from the five different countries in the WHO/ISBRA project were compared, it could be seen that the general differences were very small (Table 3). The only statistically significant differences (P<0.05) observed were the higher mean levels of trisialotransferrin in the Finnish and Japanese subjects compared with Australians, Brazilians and Canadians.

When the relative amounts of serum transferrin glycoforms for the non-drinkers of different ethnic origin were compared, the general differences between the transferrin glycoform levels were also found to be very small. In this case, the only statistically significant difference observed was a lower trisialotransferrin level in Asian/Indians compared with whites (P<0.05).

Compared with female non-drinkers (n=117), male non-drinkers (n=241) showed small but statistically significant higher levels of tetrasialotransferrin (mean 79.5% vs. 80.2%

respectively) and lower levels of pentasilaotransferrin (mean 14.4% vs. 13.9%). Most important for CDT testing was that there were no gender associated statistically significant difference between the relative amounts of disialotransferrin (mean 1.16%

vs. 1.13%). When all non-drinkers were divided into four age groups (<31, 31-40, 41- 50, and >50 years; n=66-128/group) the only statistically significant difference between the levels of transferrin glycoforms were the slightly lower tetrasialotransferrin level in those aged 41-50 years compared with those <31 years. When all non-drinkers were subdivided based on their BMI (<20, 20-24.9, 25-30 or >30). Those with a BMI >30 (obese individuals) showed a significantly (P<0.05) higher mean value for disialotransferrin (mean 1.22%) compared with the other subgroups (range 1.11- 1.12%). Other small but statistically significant differences were the higher levels of monosialotransferrin (mean 0.20% vs. 0.14-0.17%), trisialotransferrin (4.53% vs. 3.39- 4.07%) and lower levels of hexasialotransferrin (0.57% vs. 0.74-0.77%) for those with a BMI <20 compared with the other BMI subgroups.

(31)

Table 3. Distribution of serum transferrin glycoforms in non-drinkers from different study sites.

(32)

20

Significantly (P<0.05) higher disialotransferrin levels were found in smokers compared with non-smokers for all 3 drinking categories. In non-drinkers, the disialotransferrin mean value was 1.13% compared with 1.20% for smokers, but between the other drinking categories the differences were higher. However, in the light/moderate and heavy drinkers, the higher disialotransferrin levels in smokers could largely be explained by a higher alcohol consumption level compared with non-smokers.

4.2 PAPER II

The relative levels of different serum transferrin glycoforms in samples from the WHO/ISBRA Study on State and Trait Markers of Alcohol Use and Dependence were determined. There was a statistically significant difference between the relative amounts of disialotransferrin (%DST) in men and women in both light/moderate drinkers (mean 1.40% vs. 1.23%) and heavy drinkers (2.40% vs. 1.75%). When the whole sample instead was divided into four age groups (<31, 31-40, 41-50 and >50 years), the only significant (P<0.05) difference in %DST was that heavy drinkers aged 41-50 years showed higher levels (mean 2.80%) compared with those aged <31 years (2.08%) and >50 years (2.02%). Notably, the self reported average alcohol consumption level during the last month was similar for all age subgroups (range for means 74-95 g/day). When the whole sample was grouped based on body mass index (BMI <20, 20-25, >25-30, >30), the only statistically significant (P<0.05) difference was that heavy drinkers with a normal BMI of 20-25 showed higher %DST values (mean 2.66%) compared with the other groups (range for means 1.53-2.05%), and again this was with rather similar alcohol consumption levels for all BMI subgroups (range for means 60-95 g/day).

Correlations of %DST, CDTect, GGT and AST with self-reported mean daily alcohol consumption in the month prior to blood sampling are given in Table 4. For all drinkers combined, and for light/moderate drinkers of both genders, and also for male heavy drinkers, the strongest correlation with the self-reported alcohol intake was found for

%DST. In female heavy drinkers GGT was the strongest correlate with mean alcohol intake.

(33)

Table 4. Correlations between self-reported mean daily alcohol consumption and serum levels of %disialotransferrin (%DST), CDT by CDTect, gamma-glutamyltransferase (GGT) and aspartate aminotransferase (AST).

Self-reported mean daily alcohol consumption

Light/Moderate drinkers Heavy drinkers All drinkers

Men Women Men Women

%DST 0.47***

(n = 436)

0.27***

(n = 241)

0.30***

(n = 250)

0.24*

(n = 77)

0.59***

(n = 1004) CDTect 0.32***

(n = 444)

0.18**

(n = 245)

0.22***

(n = 259)

0.22 (n = 79)

0.31***

(n = 1027) GGT 0.16*

(n = 442)

0.06 (n = 243)

0.13*

(n = 258)

0.31***

(n = 78)

0.34***

(n = 1021) AST 0.14**

(n = 442)

0.00 (n = 243)

0.20*

(n = 258)

0.26*

(n = 78)

0.25***

(n = 1021)

*** P<0.0001, **P<0.01, *P<0.05

Figure 4. Receiver-operating characteristic (ROC) analysis was used to distinguish female and male heavy drinkers from the combination of non-drinkers and light/moderate drinkers by measurement of serum %disialotransferrin. ROC analysis was also used analysing the CDTect and GGT values for non-drinkers and light/moderate drinkers compared with heavy drinkers.

0 20 40 60 80 100

80

60

40

20

0

Sensitivity (%)

%DST Men %DST Women GGT CDTect

100-Specificity (%)

(34)

22

For evaluation of the overall test accuracy of %DST as an alcohol biomarker and for comparision with CDTect, GGT and AST, receiver-operating characteristics (ROC) analysis was performed (Figure 4). The area under the ROC curves (AUC) for the combination of the non-drinkers and light/moderate drinkers in comparison with heavy drinkers was not significantly different between men (AUC 0.83) and women (0.82).

The AUC for %DST was significantly (P<0.001) higher than for CDTect (0.68) and GGT (0.69). The sensitivities and specificities of serum %DST for “heavy drinking”, according to the WHO/ISBRA Interview Schedule classification, at different threshold limits are shown in Table 5 for all subjects combined and for men and women separately. At any potential cut-off, women showed lower sensitivity but higher specificity compared with men.

Table 5. Comparisons of sensitivities and specificities of serum %disialotransferrin (%DST) for

“heavy drinking” at different cut-off limits for all subjects combined and for men and women separately.

Non-drinkers + Light/Moderate drinkers vs. Heavy drinkers

All subjects Men Women

%DST cut-off Sensitivity Specificity Sensitivity Specificity Sensitivity Specificity

1.5 66.1 87.0 68.8 84.5 57.1 91.6

1.6 59.6 90.7 63.2 88.5 48.1 95.0

1.7 51.1 92.8 55.6 90.7 35.1 96.6

1.8 44.6 94.4 48.8 92.9 31.2 97.5

1.9 38.2 95.2 42.8 93.6 23.4 98.0

2.0 36.1 95.9 40.8 94.4 20.8 98.9