Pronounced variation in bile acid synthesis in humans is related to gender, hypertriglyceridaemia and circulating levels of fibroblast growth factor 19


  • C. Gälman,

    1. From the Department of Endocrinology, Metabolism & Diabetes
    2. Center for Biosciences, Department of Medicine, Karolinska Institutet at Karolinska University Hospital, Huddinge, Stockholm, Sweden
    Search for more papers by this author
  • B. Angelin,

    1. From the Department of Endocrinology, Metabolism & Diabetes
    2. Center for Biosciences, Department of Medicine, Karolinska Institutet at Karolinska University Hospital, Huddinge, Stockholm, Sweden
    Search for more papers by this author
  • M. Rudling

    1. From the Department of Endocrinology, Metabolism & Diabetes
    2. Center for Biosciences, Department of Medicine, Karolinska Institutet at Karolinska University Hospital, Huddinge, Stockholm, Sweden
    Search for more papers by this author

Mats Rudling, MD, PhD, Department of Endocrinology, Metabolism & Diabetes, C2-94, Karolinska University Hospital at Huddinge, S-141 86 Stockholm, Sweden. (fax: +468-711-07-10; e-mail:


Abstract.  Gälman C, Angelin B, Rudling M. (Karolinska Institutet at Karolinska University Hospital, Stockholm). Pronounced variation in bile acid synthesis in humans is related to gender, hypertriglyceridaemia and circulating levels of fibroblast growth factor 19 (Rapid Communication). J Intern Med 2011; doi:10.1111/j.1365-2796.2011.02466.x

Background.  Bile acid (BA) synthesis is essential in cholesterol and lipid homoeostasis.

Methods.  Serum samples from 435 normal and 23 cholecystectomized subjects were obtained after overnight fasting and assayed for markers of BA and cholesterol synthesis, as well as cholesterol absorption. We determined whether BA synthesis was related to fibroblast growth factor 19 (FGF19; a circulating metabolic regulator that is thought to inhibit BA synthesis), gender, age and serum lipids.

Results.  Bile acid synthesis varied more than 9-fold in normal individuals and was 29% higher in men than in women. Whilst low-density lipoprotein cholesterol increased with age, BA and cholesterol synthesis were stable. BA production was positively correlated with serum triglycerides (TGs), and 35% of individuals with a high level (>95th percentile) of BA synthesis had hypertriglyceridaemia (HTG) (>95th percentile). Serum FGF19 levels varied by 7-fold in normal individuals and were related inversely to BA synthesis but were not related to gender, plasma lipids or history of cholecystectomy.

Conclusions.  Bile acid synthesis has a wide inter-individual variation, is lower in women than in men and is correlated positively with serum TGs. High BA production is frequently linked to HTG. Age-related hypercholesterolaemia is not associated with changes in BA or cholesterol production, nor to an increase in cholesterol absorption. In humans, the circulating level of FGF19 may regulate hepatic BA production under fasting conditions.


Bile acids (BAs) are synthesized from cholesterol in the liver, stored in the gallbladder and released into the upper small intestine in response to ingestion of food, thereby facilitating the absorption of lipophilic nutrients [1]. Because of efficient subsequent absorption in the lower small intestine, the circulating pool of BAs is maintained despite cycling between four and 10 times daily, with a net excretion of about 30–40% per day. This loss of BAs is compensated by de novo synthesis and, together with the direct secretion of cholesterol into bile, comprises the major pathway for cholesterol elimination from the body. In the liver, the synthesis of BAs is regulated by end-product inhibition by BAs returning in the portal vein, which interact with the hepatic nuclear farnesoid X receptors (FXRs) [1, 2], initiating events that reduce the transcription and activity of the rate-limiting enzyme in BA synthesis, cholesterol 7α-hydroxylase (CYP7A1). The hepatic production of BAs may also be controlled by signalling from the gut: during their transintestinal flux, BAs elicit the synthesis and release of fibroblast growth factor (FGF) 19 (or its orthologue in the mouse, FGF15) through interaction with intestinal FXRs [3, 4]. Circulating FGF15/19 in turn bind to their receptors in the liver and suppresses BA synthesis [3]. Strong correlative evidence indicates that circulating FGF19 may be an important regulator of hepatic BA synthesis in humans [4].

Disturbances of BA production and cholesterol elimination are involved in the pathogenesis of several clinical conditions including dyslipidaemia and atherosclerosis, malabsorption, and gallstone disease [5]. Interference with the enterohepatic circulation of BAs has been shown to result in powerful therapeutic effects in these conditions [5]. In addition, recent studies in mouse models indicate that BAs may also serve as ligands of specific BA receptors regulating lipid and energy metabolism [2, 6, 7]. However, there are pronounced species differences in BA metabolism between mice and humans [1], and our understanding of how BA synthesis varies under normal circumstances in humans is still incomplete. This is partly because previous studies have used complicated methods, thereby limiting the investigation of larger series of subjects. The serum level of circulating 7α-hydroxy-4-cholestene-3-one (C4), an intermediate in BA synthesis, has been shown to be a reliable marker of the activity of CYP7A1 and BA synthesis [8–10]. Using this marker, we have shown in humans that BA synthesis has a unique diurnal rhythm [11], that it is modulated by treatment with BAs and BA-binding resins [4], that it is increased in cholesterol gallstone disease [12] and that it may be regulated by levels of circulating FGF19 [4].

The aims of this study were (i) to describe the normal variation and range of BA synthesis in a healthy population, (ii) to determine whether the synthesis of BAs is influenced by age or gender, (iii) to characterize the relationship between BA synthesis and fasting serum levels of triglycerides (TGs), (iv) to describe the normal variation in FGF19 levels and to establish whether they relate to BA synthesis and (v) to evaluate how BA synthesis relates to cholesterol synthesis and to the absorption of dietary cholesterol. Our results show that BA synthesis has a pronounced inter-individual variation, is higher in men than in women, and is unaltered with ageing. Elevated BA production is commonly associated with hypertriglyceridaemia (HTG). Furthermore, overnight fasting serum levels of FGF19 have a wide inter-individual variation and are correlated inversely with BA synthesis. Although BA synthesis is increased in subjects who have undergone cholecystectomy, their serum levels of FGF19 are normal.

Materials and methods

Subjects and study design

A total of 495 subjectively healthy volunteers were initially recruited for this investigation. Subjects with laboratory evidence or a history of hepatic, kidney or intestinal disease, and those taking drugs known to alter lipid metabolism, were excluded. The final ‘normal’ population comprised 435 individuals (222 women and 213 men) with an age range from 20 to 89 years; baseline data are shown in Table 1. An additional 23 subjects (18 women and 5 men) who had a history of cholecystectomy were evaluated separately.

Table 1. Baseline data for healthy normal subjects and those with previous cholecystectomy
 AllFemale subjectsMale subjectsCholecystectomised female subjects
  1. Significance of differences (Mann–Whitney U test): male vs. female subjects: A, < 0.0001; B, < 0.001; C, < 0.003; D, < 0.006; and cholecystectomised vs. normal female subjects: a, < 0.0001; b, = 0.0003; c, < 0.004, d, < 0.02, e, < 0.04.

  2. Data are given as mean ± SD, or median (5th, 95th percentile).

Age (years)49 ± 1749 ± 1749 ± 1764 ± 12b
BMI (kg m−2)24.9 ± 3.724.3 ± 3.925.5 ± 3.4A27 ± 3.9c
Glucose (mmol L−1)5.1 ± 0.805.1 ± 1.05.2 ± 0.55.2 ± 0.46
Total cholesterol (mmol L−1)5.3 ± 1.15.4 ± 1.15.3 ± 1.15.8 ± 0.90
LDL cholesterol (mmol L−1)3.3 ± 1.03.3 ± 1.03.3 ± 0.903.7 ± 0.78
HDL cholesterol (mmol L−1)1.5 ± 0.401.6 ± 0.381.4 ± 0.36A1.6 ± 0.32
Total TG (mmol L−1)1.18 ± 0.72
1.00 (0.50, 2.4)
1.04 ± 0.54
1.0 (0.5, 2.1)
1.32 ± 0.84A
1.1 (0.5, 2.7)
1.27 ± 0.70
1.1 (0.40, 3.4)
C4 (ng mL−1)16.8 ± 13.1
13.4 (4.0, 41)
14.8 ± 10.0
12.8 (3.5, 36)
19 ± 15.5C
14.5 (4.2, 46)
36.8 ± 14.9a
33.9 (4.3, 7.8)
C4c (mg mol−1)3.2 ± 2.4
2.5 (0.80, 7.5)
2.8 ± 1.8
2.3 (0.75, 6.1)
3.6 ± 2.9B
2.9 (0.90, 8.0)
6.4 ± 2.8a
6.4 (2.9, 13.2)
FGF19 (pg mL−1)146 ± 88
121 (49, 343)
144 ± 88
129 (51, 342)
147 ± 89
124 (47, 344)
148 ± 105
106 (35, 377)
Lathosterol (ng mL−1)1432 ± 602
1328 (657, 2534)
1353 ± 577
1243 (642, 2523)
1509 ± 620D
1424 (670, 2558)
1785 ± 771d
1634 (607, 3493)
Lathosterol/c (mg mol−1)269 ± 100
254 (137, 476)
254 ± 94
244 (120, 479)
285 ± 103B
270 (147, 475)
309 ± 133c
267 (112, 592)
Campesterol/c (mg mol−1)808 ± 413
731 (299, 1606)
804 ± 423
755 (294,1565)
810 ± 403
700 (301,1641)
690 ± 502
506 (225, 2324)
Campesterol/lathosterol3.6 ± 2.8
2.8 (0.90, 9.5)
3.7 ± 2.7
3.0 (0.94, 9.5)
3.5 ± 2.9
2.7 (0.86, 9.0)
3.0 ± 3.8e
2.2 (0.55, 17.2)

Blood samples were taken in the morning between 08.30 and 10.00 after overnight fasting. All subjects gave informed consent to participate in the study, which had been approved by the Ethics Committee of the Karolinska Institutet.


C4 concentration in serum was determined in duplicate by high-performance liquid chromatography using 7β-hydroxy-4-cholesten-3-one as internal standard [9]. The values were normalized for total cholesterol [11] and expressed as milligrams of C4 per mole of cholesterol (C4c).

Unesterified lathosterol levels, reflecting whole-body cholesterol synthesis [13], were assayed by isotope dilution mass spectrometry after the addition of deuterium-labelled internal standard as described previously [14]. Lathosterol levels were corrected for total cholesterol (lathosterol/c) [13].

Campesterol was analysed in duplicate by gas chromatography-mass spectrometry using D5-campesterol as internal standard [15]. The ratio between campesterol and lathosterol was calculated as an indicator of the absorption of dietary cholesterol.

Fibroblast growth factor 19 in serum was assayed in duplicate using a sandwich enzyme-linked immunosorbent assay (ELISA; FGF19 Quantikine ELISA kit, cat. no. DF1900; R&D Systems, Minneapolis, MN, USA) [4].

Total and high-density lipoprotein (HDL) cholesterol, TGs, aspartate aminotransferase, alanine aminotransferase and glucose were determined using standard clinical chemistry techniques. Low-density lipoprotein (LDL) cholesterol was calculated according to Friedewald et al. [16].


When two groups were compared, the Mann–Whitney U test was used. In Fig. 1, multiple groups were compared using the Bonferroni test. Correlations were evaluated by Spearman rank correlation. Frequency differences were assessed by chi-squared analysis. GraphPad Prism version 5 was used (GraphPad Software, San Diego, CA, USA.

Figure 1.

(a) Plasma levels of total cholesterol and low-density lipoprotein (LDL) cholesterol increase with age. Serum total, LDL and HDL cholesterol in relation to age in 435 healthy volunteers (222 women and 213 men). Error bars indicate SD; *P < 0.01, Bonferroni test. Subjects were divided into age quintiles, which are plotted on the graph as their means. Quintiles spanned 20–34, 35–49, 50–62 and 63–85 years. (b) C4c in serum has a skewed distribution. Frequency distribution of C4c in 435 normal subjects; inset graph shows the distribution after logarithmic transformation. (c) Lack of relationship between age and serum levels of C4c. Data from 222 women (red circles) and 213 men (black circles). (d) Correlation between C4c and total serum triglycerides in 435 normal subjects. Rs, Spearman’s rank correlation coefficient.


Distribution of serum lipoproteins

In the 435 normal subjects, there was a clear age-related increase in total and LDL cholesterol in both genders (Fig. 1a). Total and LDL cholesterol tended to be lower in women than in men, whereas HDL cholesterol was significantly higher in women. In addition, HDL cholesterol was stable with age. Serum TG levels had a skewed distribution and were significantly higher in men (Table 1).

Circulating C4c in normal subjects after overnight fasting

The distribution of serum C4c in normal subjects was skewed (Fig. 1b) and varied 9.4-fold between the 5th and 95th percentiles. Values were normalized after log transformation (inset, Fig. 1b). C4c levels were 29% higher in men than women (Table 1) and remained unaltered with ageing in both genders (Fig. 1c). There were weak correlations between C4c levels and body mass index (BMI; Rs = +0.19) and body weight (Rs = +0.17; < 0.0001 for both).

For all 435 subjects, there was a correlation between C4c and TGs (Rs = +0.31, Fig. 1d). It is evident that whereas the majority of subjects with HTG had C4c levels within the normal range, there was an increase in the proportion of individuals with HTG amongst subjects with high C4c levels. Thus, amongst those with C4c > 7.5 mg mol−1 (95th percentile), 35% had TGs > 2.4 mmol L−1 (95th percentile) and this distribution was significant at < 0.001 (χ2 = 27.13). Plasma glucose levels were not related to C4c levels (Rs = +0.080).

Circulating levels of FGF19

Fasting FGF19 levels displayed a 7-fold inter-individual variation, with a skewed distribution that was normalized after log transformation (Fig. 2a). Also, similar to C4c, FGF19 levels remained stable with increasing age (data not shown) but, in contrast to C4c, FGF19 levels were not different between men and women. In agreement with the notion that FGF19 may regulate BA synthesis, there was a negative correlation between fasting levels of FGF19 and C4c (Rs = −0.28; Fig. 2b). However, there were no correlations between FGF19 and BMI (Rs = −0.098), plasma TGs (Rs = −0.027) or plasma glucose (Rs = −0.041).

Figure 2.

(a) The distribution of serum fibroblast growth factor 19 (FGF19) in normal subjects is skewed but normalized after logarithmic transformation (inset graph). (b) Inverse correlation between C4c and serum FGF19 in normal subjects. Inset graph shows the same correlation for the 22 cholecystectomised subjects. Rs, Spearman’s rank correlation coefficient.

Serum levels of lathosterol/c

The levels of this marker of total body cholesterol synthesis [13] showed a 3.5-fold variation as well as a skewed distribution. Serum lathosterol/c levels were 14% higher in men than in women (Table 1). There was no change in lathosterol/c with increasing age. C4c and lathosterol/c levels were positively correlated (Rs = +0.27, Fig. 3a), supporting a relationship between BA synthesis and cholesterol production.

Figure 3.

(a) Correlation between bile acid (BA) synthesis and cholesterol synthesis measured as C4c and lathosterol/c, respectively, in 435 normal subjects. (b) Correlation between BA synthesis and cholesterol absorption as estimated from the camposterol/lathosterol ratio. Rs, Spearman’s rank correlation coefficient.

Serum campesterol and campesterol/lathosterol ratio

Campesterol levels corrected for total cholesterol (campesterol/c), which reflect the intestinal uptake of dietary sterol [17], were not increased and in fact were reduced with increasing age (Rs = −0.30, < 0.0001). In agreement with this finding, the ratio of campesterol/lathosterol also decreased with increasing age (Rs = −0.21, = 0.0023). There were negative correlations between C4c and campesterol/c (Rs = −0.11, < 0.03) and between C4c and the campesterol/lathosterol ratio (Rs = −0.22, = 0.0016; Fig. 3b).

Influence of cholecystectomy

In agreement with previous findings [18], the 23 healthy subjects who had undergone cholecystectomy had 88% higher C4c levels compared to those with an intact gallbladder (< 0.0001). This was most evident for the female subgroup (= 18) when compared to the 222 female controls (+129%; < 0.0001; Table 1). There was no difference in FGF19 or lathosterol/c levels between normal women with and without a gallbladder, whereas dietary cholesterol absorption, as evaluated by the camposterol/lathosterol ratio, was slightly reduced in those who had had a cholecystectomy (−19%; < 0.04). Within the entire group of cholecystectomized subjects, there was a significant inverse relation between C4c and FGF19 (Rs = −0.43; < 0.05, Fig. 2b, inset).


In this study, we determined a set of markers of cholesterol metabolism in normal subjects, obtaining a view at a single time-point under standardized conditions. In addition, we explored the potential role of the recently described circulating metabolic regulator FGF19. Several important novel findings of potential physiological relevance were identified.

First, we demonstrated that BA synthesis, as evaluated by C4c levels in serum, has a pronounced inter-individual variation (∼9-fold) in normal human subjects under fasting conditions. Such a remarkably broad variation is, however, in line with previous measurements of the synthesis of BAs in normal subjects, using isotope kinetics and faecal balance techniques which reflect mean BA production over several days [19–21], as well as using assays of liver CYP7A1 activity which gives an indication of production at one time-point [22]. However, the small size of these previous studies does not permit any valid conclusions about or evaluation of possible contributory factors. Furthermore, previous reports on fasting levels of C4 in normal subjects [23, 24] are compatible with our results. Although some of the variation in BA synthesis between individuals may be explained by factors such as HTG, gender and body weight, it is clear that a substantial fraction remains to be explained. To what degree this is related to genetic and environmental factors, including influence of diet and intestinal microflora, is an important area for further research. The variations in C4c levels were linked to somewhat less pronounced (∼5-fold) variations in lathosterol/c levels. This indicates that the variation in BA production is generally met by corresponding adjustments in cholesterol synthesis, and further underscores the powerful way that changes in BA metabolism can influence cholesterol balance in humans.

Second, we were able to establish a clear effect of gender on BA synthesis, with men having ∼30% higher C4c levels than women. Previous studies have shown that the BA pool size is ∼30% greater in men [25], and this gender-related difference may thus be explained by a difference in BA production. Also, cholesterol synthesis, as evaluated by serum lathosterol/c levels, was higher (∼15%) in men than in women. Although we cannot explain the mechanism behind this interesting gender effect, the absence of changes in C4c levels with age in women means that the proposed involvement of oestrogens is unlikely. Lower levels of BA synthesis and pool size in women may be of relevance for their higher propensity than men to develop cholesterol gallstones [12]. Of note, there is an opposite gender difference in BA synthesis in mice, with males having a lower BA pool size and production rate [26, 27] and an increased susceptibility to gallstone formation [28, 29].

Third, we did not find any evidence for a reduced synthesis of BAs with increasing age. This is in contrast to previous findings from studies using isotope techniques in smaller numbers of subjects [20, 30]. As we did observe the expected rise in LDL cholesterol with age in our subjects (Fig. 1a), the present results indicate that, in humans, the age-dependent decrease in clearance of plasma LDL [31] is not likely to be the result of a reduction in BA synthesis.

Fourth, there was a positive correlation between the level of BA synthesis measured as C4c and total serum TGs. This finding in normal subjects is in agreement with previous results in patients with hyperlipidaemia, showing that the production of VLDL TGs relates to the synthesis of BAs both in basal conditions and during perturbations of BA synthesis through treatment with BAs or BA-binding resins [32, 33]. Although the exact mechanism(s) responsible for this is not known in humans, variation in hepatic exposure to BAs probably influences TG synthesis in the liver through FXR-mediated regulation of sterol regulatory element-binding protein-1c (SREBP-1c) [7] Most individuals with HTG had normal levels of BA production, but the fact that there was a clear subgroup of subjects with markedly elevated BA synthesis and with an increased prevalence of HTG (Fig. 1d) is of interest. Patients with monogenic familial HTG with deficient BA absorption and increased BA synthesis have been identified [34, 35] and it should now be possible to utilize measurements of C4c for further phenotypic and genetic exploration.

Fifth, an important novel aspect of this study was our analysis of the potential role of circulating FGF19 as a regulator of hepatic BA synthesis in normal humans. Fasting plasma levels of FGF19 increase in response to administration of the primary BA chenodeoxycholic acid, whereas they are reduced during treatment with the BA-binding resin cholestyramine [4]. These findings support the hypothesis that circulating FGF19 is generated by enterocytes sensing the transintestinal flux of BAs. Further support for this has been gained from the fact that serum FGF19 has a diurnal rhythm with postprandial peaks that disappear during fasting [4].

Fasting levels of FGF19 also had a pronounced inter-individual variation (∼7-fold) in healthy humans. The presence of an inverse relationship between fasting serum levels of FGF19 and C4c would be compatible with the prevailing hypothesis that FGF19 can exert feedback inhibition on CYP7A1 in humans under basal conditions, although it is also clear that the relationship does not explain all of the variation in BA synthesis (Fig. 2b). Whether circulating FGF19 exerts direct effects on hepatic BA synthesis in humans is still not fully established, and further testing of this possibility would require human studies of FGF19 infusion. The amounts of synthetic FGF19 (0.1–1 mg kg−1) that have been administered to mice to establish metabolic effects, including suppression of BA synthesis, are extremely high [3, 36–38] and result in plasma concentrations that are 50- to 250-fold higher (∼26 000 pg mL−1) [38] than those observed in humans, even during treatment with exogenous BAs [4].

In addition to their effects on BA synthesis, high doses of FGF19 have been reported to exert powerful metabolic effects in animal models of obesity, dyslipidaemia and diabetes [36–39]. However, it is unclear whether FGF19 may induce metabolic rate or reduce plasma glucose and TGs in humans. The lack of correlation between FGF19 and plasma TGs or glucose observed in the present work would suggest that, under basal conditions, it is unlikely that FGF19 is an important regulator of these metabolites in humans. Further studies are needed to explore how genetic and environmental factors govern FGF19 levels, and to what extent they may physiologically regulate BA synthesis and metabolism of lipids and glucose.

Although there was a negative relationship between FGF19 and C4c levels in normal subjects, there were three situations in which there was a clear lack of association. First, in contrast to serum C4c levels, serum FGF19 levels did not show any gender difference. We cannot explain this interesting finding at present, but it may relate to differences in sensitivity to transintestinal BA flux or different modes of action of BAs or FGF19 in the liver in men and women. Second, that FGF19 levels were not related to serum TGs and were not reduced in individuals with HTG again indicates that there may be different signalling pathways for the hepatic effects of BAs and FGF19 in humans. The third intriguing observation was that in the subjects who underwent cholecystectomy, in whom BA synthesis was doubled, the FGF19 levels were not different from those in subjects with intact gallbladders. However, similar to the normal subjects, there was an inverse relationship between C4c and FGF19 levels within the cholecystectomy group (Fig. 2b, inset). This may indicate that the induced BA synthesis in these subjects, several years after cholecystectomy, is not driven by reduced FGF19 levels, but that FGF19 still takes part in the physiological regulation of BA synthesis in this situation. This is in contrast to the situation during chronic cholestyramine treatment where FGF19 levels are reduced [4] and may indicate that subjects without a gallbladder, despite a reduced BA pool size, maintain an increased enterohepatic circulation of BAs [40, 41], thus stabilizing FXR signalling in the gut. Again, further studies are required to understand the relative contribution of intestinal FGF19 in the complex regulation of BA metabolism.

Finally, we analysed the possible influence of inter-individual variation in dietary cholesterol absorption on cholesterol and BA metabolism. Serum levels of the plant sterol campesterol and the ratio between campesterol and lathosterol levels have been established as markers of cholesterol uptake [17]. We tested the hypothesis that an increase in cholesterol absorption could explain the observed increase in LDL cholesterol with age. This was clearly not the case, however, and the fact that instead both markers of cholesterol absorption were reduced with increasing age could be taken as an indication that the relative uptake of dietary cholesterol is lowered during ageing. However, another possibility that may be more relevant is that the increase in biliary secretion of cholesterol that occurs with ageing [20] competes with dietary plant sterols for absorption [17], resulting in lowered plasma levels of the latter. Further studies will be necessary to more fully understand how disturbances in BA metabolism may influence absorption and biliary secretion of cholesterol, and eventually contribute to the pathogenesis of cholesterol gallstone disease.

In conclusion, we have characterized the regulation of BA synthesis in a substantial number of healthy subjects. BA synthesis has a wide inter-individual variation, is higher in men than in women and is clearly linked to TG levels. The serum level of FGF19, which presumably mirrors the transintestinal BA flux, also displays a pronounced inter-individual variation and may be involved in the physiological regulation of BA synthesis in humans. Whilst our findings could be important in explaining both the increased propensity for gallstone formation and the lower prevalence of HTG in women than in men, further studies should now be undertaken to elucidate how genetic, dietary and microbial factors interact to determine the capacity of each individual to eliminate cholesterol as BAs, and how this may influence the susceptibility to disease.

Conflict of interest statement

No conflict of interest to declare.


We thank Ingela Arvidsson and Lisbet Benthin for expert technical assistance. This work was supported by grants from the Swedish Research Council, the Swedish Foundation for Strategic Research, the Grönberg and the Swedish Heart-Lung Foundations, “Förenade Liv” Mutual Group Life Insurance Co., the Foundation of Old Female Servants, Stockholm City Council (ALF), AstraZeneca Future Forum, the Swedish Diabetes Foundation and the Cardiovascular Program, Karolinska Institutet/Stockholm City Council.