Article Text

PDF

Characterisation of the host inflammatory response to Staphylococcus epidermidis in neonatal whole blood
  1. C Härtel1,
  2. I Osthues1,
  3. J Rupp2,
  4. B Haase1,
  5. K Röder3,
  6. W Göpel1,
  7. E Herting1,
  8. C Schultz1
  1. 1
    Department of Paediatrics, University of Lübeck, Lübeck, Germany
  2. 2
    Institute of Medical Microbiology and Hygiene, University of Lübeck, Lübeck, Germany
  3. 3
    Department of Obstetrics and Gynaecology, University of Lübeck, Lübeck, Germany
  1. Dr C Härtel, Department of Paediatrics, University of Lübeck, Ratzeburger Allee 160, D-23538 Lübeck, Germany; haertel{at}paedia.ukl.mu-luebeck.de

Abstract

Background: Coagulase-negative staphylococci (CoNS) are the most prevalent pathogens causing late-onset sepsis, and gestational age is the most important risk factor for these infections.

Objective: To characterise innate immune responses to S epidermidis by assessment of whole blood in vitro cytokine production in a large group of preterm and term infants.

Results: The S epidermidis-induced in vitro production of proinflammatory cytokines such as intracytoplasmic interleukin (IL) 6 and tumour necrosis factor α in cord blood samples was found to be dependent on gestational age (R = 0.279, 95% CI 0.10 to 0.44, p = 0.002; R = 0.251, 95% CI 0.07 to 0.41, p = 0.005, respectively; n = 121). In contrast, the production of anti-inflammatory cytokines such as IL10 and transforming growth factor β was not associated with gestational age. When different stimulation strategies were compared, a strong correlation was noted for cytokine responses after lipopolysaccharide and S epidermidis exposure—that is, IL6 (R = 0.431, 95% CI 0.29 to 0.55, p<0.001, n = 161) and IL10 (R = 0.332, 95% CI 0.18 to 0.47, p<0.001, n = 161). In addition, a lower IL6 production was found in supernatants of whole blood cultures infected with a clinically isolated IcaABD-positive (biofilm production) strain compared with a control IcaABD-negative ATCC strain (p = 0.009).

Conclusions: These in vitro data suggest that proinflammatory responses to S epidermidis are dependent on gestational age in preterm infants, whereas the counteracting anti-inflammatory response to S epidermidis may not be directly related to gestational age. Individual host factors may have a role as well as bacterial determinants, such as biofilm production. Further studies are encouraged to investigate the different aspects of innate immune responses to CoNS in vivo.

Statistics from Altmetric.com

Advances in perinatal and neonatal care have improved survival rates of preterm infants, but as a consequence of intensive care procedures and prolonged treatment in neonatal intensive care units, the number of infants at risk for invasive infections has increased.1 Simultaneously, the aetiological pattern of neonatal sepsis has changed from highly virulent bacteria, such as group B streptococci and Escherichia coli, to a pattern dominated by low-virulence bacteria causing nosocomial, late-onset infections. Among all pathogens, coagulase-negative staphylococci (CoNS),—for example, Staphylococcus epidermidis (S epidermidis), are the most common in preterm infants.1 Epidemiological studies have clearly demonstrated that gestational age is the most important clinical risk factor for S epidermidis infections in newborn infants.14 However, the underlying mechanisms for this finding have not been elucidated in great detail so far.

The immaturity of the skin as first-line physical and immunological defence has been considered to be crucial, as the skin of preterm babies is fragile, and small breaks in the skin integrity may become access sites for bacteria. A recent study proposed that erythema toxicum neonatorum, which is a common phenomenon in newborns within the first days of life, might reflect an innate immune response to colonisation and penetration of CoNS bacteria via hair shafts into the dermis.5 This points to the important role of innate immune responses for a balanced transition from the normally sterile intrauterine environment to the antigen-rich outside environment. Furthermore, prolonged need of indwelling vascular catheters, mechanical ventilation and antimicrobial treatment on the neonatal intensive care unit might contribute to the finding that extremely preterm infants are most vulnerable for CoNS infection.1 2

Although S epidermidis infections are rarely fatal, with a case mortality rate <2%,4 they may cause significant long-term morbidity among preterm infants.14 6 The clinical and laboratory signs of S epidermidis sepsis are non-specific, and therefore, microbiological evidence of the same strain in sequential blood cultures withdrawn from different sites (although rarely done in practice) is considered the most reliable way for discrimination between true inducers of sepsis and contaminants.7 S epidermidis strains often display multiresistance to antibiotics and can form an adherent multilayered biofilm on polymer surfaces (eg, indwelling devices) which is considered to be the main virulence factor.8 However, less is known about how certain virulence factors influence clinical outcome of CoNS infections.911

The immaturity of the neonatal immune system, especially in preterm infants, has been assumed to be responsible for the high susceptibility to neonatal infections.2 1214 Given the high prevalence of complications secondary to late-onset infections including white matter disease and bronchopulmonary dysplasia,1521 continuing research is required to characterise the host inflammatory responses to predominant pathogens in preterm infants such as S epidermidis.22 This study aimed at characterising cytokine responses to S epidermidis in preterm and term neonates immediately after delivery.

MATERIALS AND METHODS

In vitro study I: Evaluation of the “S epidermidis in vitro sepsis model”

To investigate the feasibility of the “S epidermidis in vitro sepsis model”, 11 healthy adult volunteers (blood donors from the local blood bank; Institute of Immunology and Transfusion Medicine University of Luebeck, Director: Professor Dr H Kirchner) were recruited and served as controls in the in vitro study I after informed consent. Cord blood samples of 14 term infants born in the perinatal unit of the University of Lübeck, Germany, Department of Obstetrics and Gynaecology (Director: Professor Dr K Diedrich) in July 2007 were obtained after parents’ informed consent. Whole blood samples of each individual term infant or adult were stimulated at the same time with 1 colony-forming unit/white blood cell (CFU/WBC) and 10 CFU/WBC S epidermidis, respectively.

In vitro study II: Evaluation of cytokine responses to S epidermidis dependent on gestational age

We performed a prospective study recruiting a single-centre hospital cohort of 115 preterm (September 2005 until December 2006) and 127 term infants (September 2005 until February 2006) born in the perinatal unit of the University of Lübeck, Germany, Department of Obstetrics and Gynaecology.

Inclusion criteria

Inborn preterm infants born >24 weeks’ gestational age, healthy term infants with unremarkable perinatal history were included.

Exclusion criteria

Amniotic infection, histological chorioamnionitis as cause of preterm delivery, early-onset sepsis, lethal abnormalities comprised the exclusion criteria.

Whole blood samples of individual infants were stimulated with lipopolysaccharide (LPS) and 1 CFU/WBC S epidermidis, respectively.

In vitro study III: Investigation of strain differences and their impact on cytokine responses

Whole blood samples of 11 healthy term infants were stimulated at the same time for 24 hours with 1 CFU S epidermidis /WBC control ATCC 12228 strain, strain 1 and strain 2, as indicated.

Strain 1 94B080—genotype A according to the classification introduced by Björkqvist et al.9 IcaABD positive (biofilm production), MecA positive, isolated from a preterm infant with clinical sepsis, gestational age 26+0 weeks, birth weight 990 g.

Strain 2 94B575—genotype “single isolate” according to the classification introduced by Björkqvist et al.9 IcaABD and MecA negative, isolated from a preterm infant with clinical sepsis, gestational age 31+0 weeks, birth weight 1195 g.

Approval of the institutional review board at University of Lübeck was given for all study parts.

Material

Heparinised venous cord blood samples were obtained after informed parental consent from healthy term infants and preterm infants immediately after delivery. Healthy adult volunteers served as a control in in vitro study I after informed consent. We performed an ex vivo model of S epidermidis sepsis before any postnatal treatment, in which S epidermidis was added as full pathogen to whole cord blood cultures under defined conditions. In parallel, cord blood samples of preterm and term neonates were stimulated with LPS, the critical virulence factor of Gram-negative bacteria, which served as control for initiation of immune responses. Cytokines were analysed directly at the cell level by flow cytometry or assayed for their secretion into culture supernatants.

Whole blood culture

All blood samples were collected in lithium–heparin tubes (Sarstedt, Nürnbrecht, Germany) and were stored at room temperature for no longer than 24 hours before processing. For assessment of cytokine secretion into culture supernatants, whole blood specimens were suspended in RPMI 1640 supplemented with 1% penicillin/streptomycin, 2 mM glutamine, 1 mM pyruvate and non-essential amino acids (Seromed Biochrome) at a concentration of 5×106 leucocytes/ml and stimulated with 30 ng/ml LPS or S epidermidis (as indicated for each study) for 24 hours. Unstimulated controls were added to each experiment. After incubation, 2 ml of the supernatant was withdrawn for EIA and frozen at −80°C for analysis of cytokine production. EIA analysis was performed with human Quantikine IL6, IL10 and transforming growth factor β (TGFβ) ELISA kits (R&D, Wiesbaden, Germany) according to manufacturers instructions.

Assessment of intracellular cytokine production in CD14+ cells

Heparinised whole blood was suspended in RPMI 1640 supplemented with 1% penicillin/streptomycin, 2 mM glutamine, 1 mM pyruvate and non-essential amino acids at a concentration of 5×106 leucocytes/ml. To induce intracytoplasmic production of proinflammatory cytokines in monocytes, whole blood cultures were stimulated with 30 ng/ml LPS or S epidermidis for 4 hours. Cells were exposed to 3 μM monensin (Sigma, Deisenhofen, Germany) during the whole stimulation period, followed by fixation with 4% paraformaldehyde (Riedel de Haen, Seelze, Germany). An unstimulated control was added in each experiment.

For intracellular staining, cells were washed in Hanks’s buffered salt solution (HBSS) and resuspended in a buffer consisting of HBSS, 0.1% saponin (Riedel de Haen, Selze, Germany) and 0.01 M HEPES buffer (Seromed Biochrome, Berlin, Germany). Cells (200 μl aliquots) were added to tubes containing 0.5 μg/10 μl of monoclonal antibodies (mAbs; BD Pharmingen, Heidelberg, Germany) against CD14 (M5E2, phycoerythrin-conjugated), IL6 (MQ2-13A5, FITC-conjugated) and tumour necrosis factor α (TNFα) (mAb11, FITC-conjugated), IL1α (364-3B3-14, phycoerythrin-conjugated). Preincubation with a surplus of unconjugated anticytokine mAbs (5 μg/10 μl; Pharmingen) served as a negative control for intracellular staining to each sample. Isotype-specific antibodies were used to detect irrelevant specificity for surface molecule staining.

Flow cytometry

Flow cytometric analysis was performed on BD FACS Canto (Pharmingen). A total of at least 2000 CD14+ cells were acquired from each sample. Dead cells were excluded by forward and side scatter gating. Thresholds were set according to the unconjugated anticytokine mAb control. Positive cells <2% were allowed beyond the statistical marker. Data were expressed as a percentage of CD14+ cells.

Statistical analysis

Statistical differences between groups were tested for paired data by the Wilcoxon test (two-tailed). The Mann–Whitney U-test was applied for statistical analysis of differences between neonates and adults. For analysis of correlations, the Spearman rs test was used. The level of significance was defined as p<0.05 in single comparisons. Statistical analyses were performed using SPSS 13.0 statistical software (SPSS Inc, Chicago, USA).

RESULTS

In vitro-study I: Cytokine expression is inducible in neonatal cord blood after full bacteria stimulation with S epidermidis

Study group

Peripheral venous blood samples were obtained from 11 healthy adult volunteers (median age 44 years, 95% CI 38.7 to 50.9 years; three female, eight male) who served as controls. Cord blood samples were collected from 14 healthy term infants (median gestational age 39 weeks, 95% CI 37.5 to 40.5 weeks; six female, eight male; 3/14 born by Caesarean section delivery).

Feasibility of whole blood cytokine induction after S epidermidis exposure

After 4 hours of incubation, a dose-dependent induction of intracytoplasmic IL6 production was noted in term neonates (1 CFU/WBC: median 12.6 (95% CI 10.9 to 31.9)% vs 10 CFU/WBC: median 43.6 (95% CI 29.3 to 49.4)%; p = 0.025; n = 14) which was comparable to stimulated IL6 responses in adults (1 CFU/WBC: median 30.1 (95% CI 8.9 to 34.9)% vs 10 CFU/WBC: median 41.2 (95% CI 19.0 to 51.5)%; p = 0.005; n = 11). Induction of intracytoplasmic TNFα was not dose dependent in term neonates (1 CFU/WBC: median 4.5 (95% CI 2.8 to 11.9)% vs 10 CFU/WBC: median 6.2 (95% CI 4.3 to 20.5)%; p = 0.12; n = 13) in contrast to adult controls (1 CFU/WBC: median 5.6 (95% CI 2.5 to 13.2)% vs 10 CFU/WBC: median 16.7 (95% CI 5.9 to 23.8)%; p = 0.008; n = 10; fig 1). No statistical differences were noted between adults and term neonates. Infection with S epidermidis was also found to stimulate the intracytoplasmic production of other proinflammatory cytokines, such as IL1β and IL8 (data not shown).

Figure 1 Term infants display an inflammatory response to S epidermidis comparable to that of adults. Whole blood samples of healthy adult controls (black symbols) and cord blood samples of healthy neonates (white symbols) were stimulated at the same time for 4 hours with 0, 1 or 10 colony-forming units (CFU) S epidermidis/white blood cell as indicated. Data are presented as each symbol corresponding to an individual’s percentage of interleukin (IL)6 (fig 1A) or tumour necrosis factor α (TNFα) (fig 1B) producing CD14+ cells (intracytoplasmic assessment by flow cytometry). Horizontal bars indicate the median value of all tested subjects. No statistical differences were noted between infants and adults (Mann–Whitney U-test).

In vitro study II: Characterisation of S epidermidis-induced cytokine responses in preterm and term infants

Study group

One hundred and fifty preterm infants fulfilled the inclusion criteria in the specified period of time. For 33 preterm infants their parents were not approached for consent, and for two preterm infants consent was not obtained from parents owing to language barriers. Table 1 shows the clinical data of the 115 preterm infants enrolled in this study. Eight preterm infants were excluded from analysis owing to clinical and histopathological evidence of amniotic infection/chorioamnionitis (n = 7) or blood-culture proven early-onset sepsis (n = 1). Term controls (n = 127) had a mean (SD) (range) birth weight of 3463 (473) (2250–4640) g, a mean (SD) (range) gestational age of 39.1 (1.55) (37.0–43.0) weeks, 53.5% were male and 29.7% were delivered by Caesarean section.

Table 1 Study group of 115 preterm infants

Proinflammatory neonatal cytokine responses to S epidermidis exposure in vitro are correlated with gestational age

At the single-cell level, the intracytoplasmic IL6 (R = 0.279, 95% CI 0.10 to 0.44; p = 0.002; n = 121) and TNFα (R = 0.251, 95% CI 0.07 to 0.41; p = 0.005; n = 121) response of neonatal cord blood CD14+ cells to infection with 1 CFU S epidermidis/WBC significantly correlated with gestational age (fig 2). The association with gestational age was found to be similar to proinflammatory intracytoplasmic cytokine production upon LPS exposure (IL6: R = 0.371, 95% CI 0.24 to 0.49; p<0.001; n = 193 and TNFα R = 0.344, 95% CI 0.21 to 0.47’ p<0.001; n = 193). In addition, a strong correlation between different proinflammatory cytokines produced in the cytoplasm after stimulation with S epidermidis (IL6 vs IL1β, R = 0.76, 95% CI 0.64 to 0.84; p<0.001; n = 88; IL6 vs TNFα, R = 0.66, 95% CI 0.53 to 0.76; p<0.001; n = 120) was noted at the single-cell level.

Figure 2 The proinflammatory cytokine response to S epidermidis is dependent on gestational age. Whole blood cultures were stimulated for 4 hours with 1 colony-forming unit/white blood cell (CFU/WBC) S epidermidis to induce IL6 (fig 2A) and TNFα synthesis (fig 2B) on a single-cell level in CD14+ cells (intracytoplasmic production assessed by flow cytometry). Data are expressed as percentage of cytokine-positive cells and plotted as individual points on the graph. (Full datasets of intracytoplasmic cytokine production were only available from 121 infants.) All data were included to estimate the non-parametric correlation between age and cytokine expression using the Spearman’s rs correlation (R, rank order correlation coefficient; two-sided level of significance is indicated as p value; SPSS version 13.0).

Anti-inflammatory cytokine responses to S epidermidis are independent of gestational age

We also investigated the cytokine production in stimulated whole blood culture supernatants. Along the secretory pathway, the amount of produced proinflammatory cytokine IL6 in supernatants of S epidermidis-stimulated whole blood cultures was shown to correlate with gestational age. Interestingly, the S epidermidis-induced production of anti-inflammatory cytokines such as IL10 and TGFβ was not associated with gestational age (table 2). Indicating an individually regulated cytokine balance upon S epidermidis exposure, the anti-inflammatory IL10 production was not correlated with proinflammatory IL6 production (R = 0.153, 95% CI −0.002 to 0.3, p = 0.06, n = 161). Upon LPS exposure, however, not only proinflammatory but also anti-inflammatory innate immune responses were found to be dependent on gestational age, and a correlation was noted for IL10 and IL6 responses (R = 0.266, 95% CI 0.14 to 0.39; p<0.001; n = 223). No gender differences were observed for pro- and anti-inflammatory cytokine responses after S epidermidis exposure.

Table 2 Secretory cytokine responses into whole blood culture supernatants and correlation with gestational age

Individual differences in cytokine responses to LPS and S epidermidis

In comparison of the two different stimulation strategies, we could show a correlation between cytokine responses after S epidermidis and LPS exposure with regard to protein secretion in supernatants (IL6: R = 0.431, 95% CI 0.29 to 0.55; p<0.001; n = 161; IL10: R = 0.332, 95% CI 0.18 to 0.47; p<0.001; n = 161). This observation holds also true for intracytoplasmic production of IL6 (R = 0.423, 95% CI 0.26 to 0.57; p<0.001; n = 120) and TNFα (R = 0.299, 95% CI 0.12 to 0.46; p<0.001; n = 120) at the single-cell level.

What is already known on this topic

  • Coagulase-negative staphylococci (CoNS), such as Staphylococcus epidermidis, are the most prevalent pathogens causing late-onset sepsis.

  • Gestational age is the most important clinical risk factor for these infections; preterm infants have increased susceptibility.

In vitro study III: Role of different clinical S epidermidis strains for the induction of proinflammatory cytokine production

The influence of different S epidermidis strains isolated from preterm infants with blood-culture proven clinical sepsis on the host inflammatory response of 11 term neonates (median gestational age 38.5 weeks, 95% CI 37.3 to 40.2 weeks; five female, six male; 3/11 born by Caesarean section delivery) was investigated in vitro. We found a significantly lower IL6 secretion in supernatants of whole blood cultures infected with biofilm-producing strain 1 (IcaABD positive for biofilm production, MecA positive for methicillin/aminoglycoside resistance) compared with the biofilm non-producing strain ATCC strain 12228 (ATCC control: median IL6: 9725 (95% CI 6323 to 16368) pg/ml vs strain 1: median IL6: 6901 (95% CI 3408 to 13922) pg/ml (p = 0.009, 95% CI 0.007 to 0.011) while no differences were noted for comparison with IcaABD negative strain 2 (median IL6: 7837 (95% CI 5244 to 14108; fig 3).

Figure 3 The role of different clinical S epidermidis strains for the induction of proinflammatory cytokine production. Whole blood samples of 11 healthy term infants were stimulated for 24 hours with 1 colony-forming unit (CFU) S epidermidis /white blood cell control ATCC 12228 strain, strains 1 and 2, as indicated. Each symbol corresponds to an individual subject’s IL6 protein secretion in the supernatant. Horizontal bars indicate the median of all tested subjects. Statistical analysis was performed with the Wilcoxon test, a p value of <0.05 was regarded as significant.

What this study adds

  • In this in vitro model, proinflammatory responses to S epidermidis were shown to be dependent on gestational age in preterm infants, while the counteracting anti-inflammatory response to S epidermidis was not associated with gestational age.

  • Individual host and bacterial factors may be important as shown by the correlation between cytokine responses to lipopolysaccharide and S epidermidis and the impact of strain differences on cytokine responses.

DISCUSSION

To our knowledge, this is the first investigation of S epidermidis-induced cytokine production in vitro in neonates. Previous studies have been limited to adults, demonstrating stimulation of TNFα, IL12, IL1β and IL6 expression in human mononuclear leucocytes exposed to whole, heat-inactivated CoNS23 24 or in whole blood stimulated with cell-free CoNS supernatants.25

Our data clearly indicate that the neonatal in vitro proinflammatory response to S epidermidis is dependent on gestational age in preterm infants. Term infants can initiate immune responses to S epidermidis comparable to those of adults. Interestingly, the S epidermidis-induced production of anti-inflammatory cytokines such as IL10 and TGFβ was not associated with gestational age. This finding suggests that proinflammatory immune responses develop gradually with gestational age, which leaves the extremely preterm infant susceptible to infection.2 Anti-inflammatory responses, however, are regulated on an individual level, and significantly influenced by the intrauterine cytokine environment (which is mainly polarised to TH2-cytokines) and contact with different microbial antigens—for example, Gram-positive or Gram-negative bacteria. Upon LPS exposure, we found a correlation of IL10 and TGFβ responses with gestational age (and thus confirmed our previous data26). In contrast, stimulation of anti-inflammatory cytokine production with S epidermidis is not dependent on gestational age, suggesting that the pathogenesis of CoNS infection and their harmful long-term effects may include different key elements of innate immune responses and result in a disturbed balance between pro- and counteracting anti-inflammatory responses. We further observed that neonates with the ability to mount a strong proinflammatory response to LPS which is produced primarily by Gram-negative pathogens, are also likely to produce high titres of proinflammatory cytokines after infection with S epidermidis. This correlation between cytokine responses to LPS and S epidermidis was also shown for the anti-inflammatory IL10 production, suggesting that there may be individual host variation in the strength of the immune response to different pathogens.

To our knowledge, our study provides the first large-scale data on neonatal in vitro cytokine responses to S epidermidis in an ex vivo sepsis model. The observation of comparable immune responses between term neonates and adults in our study may explain why S epidermidis sepsis is rarely seen in term neonates. In addition, the results point to the clinical relevance of immunological immaturity besides fragile skin leading to increased susceptibility to S epidermidis infections in very preterm infants. More data are needed to indicate whether extremely preterm infants can mount a proinflammatory response to CoNS bacteria at all. Furthermore, knowledge on innate cytokine immune responses to CoNS infections might contribute to a better understanding of the pathogenesis and lead to improved prevention and treatment strategies.27

There are some limitations to this study. First, the mean gestational age of 31.6 weeks in the cohort of preterm infants only partially reflects the main group of preterm infants at risk for S epidermidis infections—those infants born with a gestational age <28 weeks. Second, considering the inflammatory response to S epidermidis, several potential confounders beside clinical risk measures—for example, gestational age, should be considered. In particular, bacterial virulence factors, such as biofilm production and methicillin/aminoglycoside resistance may have an important role.28 Klingenberg et al recently reported an increased C reactive protein (CRP) response in methicillin/aminoglycoside resistant CoNS strains compared with susceptible isolates, and in contrast, lower CRP levels in biofilm-positive CoNS isolates compared with biofilm-negative CoNS strains.28 In our whole blood setting, we found a decreased IL6 secretion in supernatants of whole blood cultures infected with a biofilm producing, methicillin/aminoglycoside resistant strain compared with control ATCC 12228 strain. This may be in line with a proposed model that S epidermidis may evade the host’s proinflammatory response by masking with a biofilm multilayer.8

In summary, we demonstrated that preterm infants have a reduced proinflammatory cytokine response to S epidermidis, while anti-inflammatory responses were independent of gestational age. These data, in part, explain the high vulnerability of preterm infants to these infections. We also found an individually regulated immune response indicated by some infants who were able to produce sufficient amounts of cytokines to S epidermidis and LPS independently of gestational age. These finding should stimulate further detailed analysis of the individual genetic background in this context. Moreover, we could show the relevance of virulence of S epidermidis with a lower cytokine response in biofilm-producing, antibiotic resistant strains indicating an evasion from the host’s immune response by masking with a biofilm multilayer. Further studies are encouraged to investigate the different aspects of innate immune responses to CoNS in vivo.

Acknowledgments

The work was part of dissertational theses of IO and BH. This work was supported by “Sanitätsrat Dr Emil-Alexander-Hübner-und-Gemahlin-Stiftung”, “Possehl-Stiftung Lübeck”, “Else-Bluhme und Friedrich-Jebsen-Stiftung” and two research grants of the University of Lübeck (to CH, A14/2005; A44/2007). For comparison of inflammatory responses to different clinical isolates, we used clinical S epidermidis strains isolated from preterm infants which were phenotypically/genotypically characterised and kindly provided by Dr Maria Björkqvist and Professor Jens Schollin, Department of Paediatrics, University of Orebro, Sweden. We thank Dr rer. biol. hum. Inke Koenig, Institute for Medical Biometrics and Statistics, University of Luebeck, for her statistical assistance.

REFERENCES

View Abstract

Footnotes

  • Competing interests: None declared.

  • Ethics approval: Ethics approval was obtained.

  • Patient consent: Informed consent was given.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.