Advertisement

Pediatric Radiology

, Volume 48, Issue 6, pp 858–864 | Cite as

Do we need gadolinium-based contrast medium for brain magnetic resonance imaging in children?

  • Dennis Dünger
  • Matthias Krause
  • Daniel Gräfe
  • Andreas Merkenschlager
  • Christian Roth
  • Ina Sorge
Original Article

Abstract

Background

Brain imaging is the most common examination in pediatric magnetic resonance imaging (MRI), often combined with the use of a gadolinium-based contrast medium. The application of gadolinium-based contrast medium poses some risk. There is limited evidence of the benefits of contrast medium in pediatric brain imaging.

Objective

To assess the diagnostic gain of contrast-enhanced sequences in brain MRI when the unenhanced sequences are normal.

Materials and methods

We retrospectively assessed 6,683 brain MR examinations using contrast medium in children younger than 16 years in the pediatric radiology department of the University Hospital Leipzig to determine whether contrast-enhanced sequences delivered additional, clinically relevant information to pre-contrast sequences. All examinations were executed using a 1.5-T or a 3-T system.

Results

In 8 of 3,003 (95% confidence interval 0.12–0.52%) unenhanced normal brain examinations, a relevant additional finding was detected when contrast medium was administered. Contrast enhancement led to a change in diagnosis in only one of these cases.

Conclusion

Children with a normal pre-contrast brain MRI rarely benefit from contrast medium application. Comparing these results to the risks and disadvantages of a routine gadolinium application, there is substantiated numerical evidence for avoiding routine administration of gadolinium in a pre-contrast normal MRI examination.

Keywords

Brain Children Contrast medium Diagnostic value Gadolinium Magnetic resonance imaging 

Introduction

Gadolinium has been used since 1988 for MRI brain scans for contrast enhancement of the vessels and to provide evidence of impaired vessel wall integrity [1]. The latter is of high diagnostic value for inflammation and tumor diagnostics and is therefore a standard part of brain scans in children at most hospitals.

In 2006 the first reports of potentially fatal long-term effects of gadolinium application linked to the appearance of nephrogenic systemic fibrosis came as a surprise. This disease is caused by gadolinium-based contrast medium in patients with renal insufficiency [2, 3, 4, 5, 6]. Todd and Kay [3] suggested the term gadolinium-induced fibrosis. Though rare, the sometimes fatal course of this disease led to regulatory warnings for gadolinium-based contrast medium in 2006. Subsequently, consequences of gadolinium administration were also detected in patients with intact kidney function. Depositions of gadolinium were found primarily in the pallidum and dentate nucleus [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19]. Clinical consequences have not been clarified, but considering the rare cases of nephrogenic systemic fibrosis, long-term effects from gadolinium molecules in the brain cannot be ruled out, especially in pediatrics, with many decades of possible incubation. Therefore every use of gadolinium-based contrast medium should follow a reasonable indication with risk–benefit ratio.

Because scientific evidence concerning this subject is lacking, we assessed the diagnostic benefit of gadolinium administration in children after a normal unenhanced brain MRI scan.

Materials and methods

We retrospectively re-evaluated all brain MRI scans from children younger than 17 years performed between July 2005 and October 2016 in the pediatric radiology department of the University Hospital Leipzig (Table 1). The institution has a wide orientation typical for a German university hospital (1,500 beds) and a diverse pediatric neurological and oncological patient collective (comparable with a tertiary-care hospital). The MRI scans were conducted either on a 1.5-T system (Intera; Philips, Amsterdam, the Netherlands) or a 3-T system (Trio Tim; Siemens, Erlangen, Germany).
Table 1

Age distribution of the patients (n=6,683)

Age

0–28 days

29 days to 1 year

2–3 years

4–5 years

6–16 years

Number of patients

59

908

836

774

4,106

All MRI studies were re-assessed by two pediatric radiologists (IS, WH), with 13 years and 20 years of MRI experience, in a consensus procedure. Initially we recorded all pediatric brain examinations (n=10,474) and then we assessed those with contrast medium application (n=7,248) separately. We excluded brain examinations of the pituitary gland and those for non-central-nervous-system indications.

We evaluated the following set of sequences as a minimum: axial T2-weighted turbo spin echo, axial fluid-attenuation inversion recovery, axial T1-weighted turbo spin echo or 3-D gradient echo before and after application of contrast medium. A diffusion-weighted sequence was available in 43% of the examinations.

In 31% (n=3,226) of the brain scans, no contrast medium was applied for medical indications (e.g., examination of ventricular width for hydrocephalus). Of the 7,248 contrast-enhanced brain MRI scans, only 6,683 (92%) met the inclusion criteria (Table 2).
Table 2

Inclusion and exclusion criteria

Inclusion criteria

Exclusion criteria

<17 years

≥17 years

Brain-related indication

1. Non-brain-related indications (for instance orbital or skull MRI)

2. Examinations of the pituitary gland

Availability of every of the following sequences: axial T2-TSE, T1-TSE (alternatively 3D-GRE) before and after contrast medium

One of the following sequences missing: axial T2-TSE, T1-TSE (alternatively 3-D GRE) before and after contrast medium

GRE gradient recalled echo, MRI magnetic resonance imaging, TSE turbo spin echo

Gadolinium preparations employed in these examinations were Magnevist (Bayer, Leverkusen, Germany), Omniscan (GE Healthcare, Chalfont St. Giles, UK) and MultiHance (Bracco, Milan, Italy), all linear contrast media; and Dotarem (Guerbet, Villepointe, France) and Gadovist (Bayer, Leverkusen, Germany), both macrocyclic contrast media at a dose of 0.1 mmol/kg body weight.

The clinical data were part of the routine diagnostics. Because the study was retrospective in design and did not examine patients’ personal information, the evaluation was ethically uncritical and did not need an explicit vote of the local ethics committee.

We categorized the indications for brain imaging into 12 groups (Fig. 1). Neurological indications not categorized as “seizures” were summarized under “neurology,” such as hemiparesis, paraesthesia, dizziness and ataxia. Psychiatric indications were those such as anorexia, psychoses, behavioral disorders, personality disorders and panic attacks. Indications could be assigned to multiple categories.
Fig. 1

Clinical reasons for brain MRI (percentages)

Three-fourths of the examinations were classified as either suspected or known brain tumor (29.5%), clinically proven neurological findings (24.8%) or seizures (21.2%). The residual indications, in descending order, were developmental disorder or malformation (13%), headaches (10.2%), brain pressure (6.1%), psychiatry (5.4%), inflammation (4.3%), dilation of liquor spaces (2.6%), vascular malformation (2.4%), trauma (1.1%) and other (0.6%) (Fig. 1).

For each MR examination, we determined whether abnormal findings were detected before contrast medium administration. Subsequently, we evaluated diagnostic benefit of additional post-contrast sequences. A finding in native MRI sequences was classified as abnormal when artifacts had been ruled out and possible pathological implications were recorded, namely signal alterations of white or gray matter, ventricular enlargement, structural cerebral disorders and — if conducted — diffusion disorders.

Post-contrast MRI findings were designated as “relevant additional findings” when contrast enhancement yielded additional or divergent diagnoses. The remaining additional findings were classified as “irrelevant coincidental findings.” Abnormal findings in unenhanced MRI were grouped according to the algorithm shown in Fig. 2.
Fig. 2

Algorithm for further classification of the MRI findings. CM contrast medium

We analyzed statistical relations between relevant and irrelevant additional findings. We estimated the likelihood of contrast medium administration adding diagnostic benefit to an otherwise normal brain.

Results

Normal unenhanced MRI

In almost half of the examinations (3,003 of 6,683; 44.9%), no abnormality was found before administration of gadolinium-based contrast medium. In 2,855 of these 3,003 cases (95.1%) no pathological finding was detected even after contrast medium administration. Irrelevant coincidental findings were registered in 140 of these 3,003 patients (4.7%), of which developmental venous anomalies represented the most frequent pathology. Eight of the 3,003 examinations (0.3%, 95% confidence interval 0.12–0.52%), revealed a relevant additional finding after gadolinium application, all of them pathological meningeal enhancement (Fig. 3). In other words, administration of contrast medium in examinations without pathological findings in unenhanced sequences has an estimated likelihood of 0.12–0.52% of adding any relevant information in cohorts similar to the one studied here.
Fig. 3

Example of additional findings after application of contrast medium: post-contrast-medium extensive meningeal enhancement in a boy (3 weeks) with meningitis (1.5 T, T1-weighted spin-echo transverse section MRI, repetition time [TR]/echo time [TE]=696/15.5 ms, slice thickness 3 mm)

In the eight normal unenhanced MRI scans that demonstrated pathological meningeal enhancement after application of contrast medium, two had been scanned at 3 T and six at 1.5 T (Table 3).
Table 3

Cases with normal unenhanced MRI and pathological meningeal enhancement after application of contrast medium

Age

Indications

Contrast medium

Field strength (T)

Findings

Diagnostic impact

3 weeks

Meningitis (E. coli)

2.5 mL Omniscan

1.5

Meningeal enhancement

None

10 months

Meningitis (pneumococci)

1 mL MultiHance

1.5

Meningeal enhancement

None

4 years

Meningitis (Haemophilus influenzae)

2 mL Gadovist

3

Meningeal enhancement

None

15 years

Meningitis (neuroborreliosis)

6 mL Gadovist

1.5

Meningeal enhancement

None

11 years

Headaches, dizziness and nausea

3 mL Gadovist

1.5

Meningeal enhancement

None

13 years

Headaches, dizziness, vomiting

12 mL Omniscan

1.5

Meningeal enhancement

New diagnosis: viral meningitis (enterovirus)

15 years

Fever and fracture of the temporal bone

5 mL Gadovist

1.5

Meningeal enhancement

None

15 years

Burkitt lymphoma

6.5 mL Gadovist

3

Meningeal enhancement

None

Abnormal unenhanced MRI

In 3,680 of 6,683 examinations (55.1%) an abnormality was found before administration of gadolinium-based contrast medium. Out of these 3,680 children, contrast enhancement application in 2,511 (68.2%) did not result in an abnormal enhancement in any brain area. In the other 1,169 scans (31.8%), enhancement was detected in the area already described as abnormal in the pre-contrast sequences. An “irrelevant coincidental finding” affecting previously normal regions was determined in 102/3,680 (2.8%) cases, with developmental venous anomalies representing the most frequent entity. However, in 297/3,680 scans (8.1%, 95% confidence interval 7.2–9.0%), contrast medium application disclosed one or several “relevant additional findings” in otherwise normal brain regions. These relevant additional findings were, among others, meningeal contrast-medium enhancement and additional information about the tumor spreading or an impaired blood–brain barrier (Fig. 4).
Fig. 4

Example of additional findings after application of contrast medium: slightly intensified contrast-medium enhancement of the meninges in a girl (13 years) with headaches and dizziness and diagnosis of meningitis (1.5 T, T1-weighted spin-echo coronal MRI section, TR/TE 500/7.7 ms, slice thickness 5 mm). TE echo time, TR repetition time

Additional findings were found most frequently in the setting of specific anomalies in pre-contrast sequences as follows: tumor (76.4%), any pathology in MRI correlating to neurological pathologies (8.8%), seizures (6.4%), inflammation (5.7%) and signs of elevated intracranial pressure (5.1%).

Discussion

To the best of our knowledge, no study with a comparable number of patients has analyzed the diagnostic benefit of contrast medium administration in pediatric cerebral MRI. Every study dedicated to this subject has involved a significantly lower number of patients [20, 21, 22, 23, 24]. Moreover, most studies were at least 10 years of age, and several authors dealt exclusively with children younger than 2 years. Petrou et al. [23] and Foerster et al. [24] evaluated the benefits of contrast medium administration in that age group, labeling the additional results as not helpful, helpful or even essential. Petrou et al. retrospectively examined 437 children with seizures. Contrast-medium administration was regarded to be helpful in 5.9% of the children and even essential for diagnosis in 1.8%, all of them presenting with a history of infection. Therefore, the authors recommended contrast-medium application when infection or tumor is suspected.

Foerster et al. [24] focused retrospectively on contrast-enhanced brain MRIs of developmentally delayed children. In 107 of 170 children, when developmental delay represented the primary indication for brain MRI, contrast-medium administration proved neither helpful nor essential. In the remaining children (63 of 170) with developmental delay as secondary diagnosis, contrast-medium application was assessed as helpful in 11% but did not prove to be essential in any child. The majority of these cases were associated with neoplasia or infections. The authors argued that a contrast-medium application might pose a benefit only in the setting of suspected tumor or brain/meningeal infection. Eldevik and Brunberg [22] prospectively evaluated brain MR images from 125 children younger than 2 years, before and after contrast-medium administration. In four cases gadolinium administration after an already abnormal pre-contrast MRI was found to be essential for final diagnosis. In all children with unenhanced normal MRI (45 of 125) contrast-medium application did not result in a pathological enhancement. Thus, the authors summarized that contrast medium might not be necessary in children with normal results in unenhanced brain MRI.

The studies by Elster and Rieser [20] and Baierl et al. [21] are the most similar to our study with regard to the cohort. Elster et al. included 65 children and adolescents between 0 and 18 years of age. Evaluating prospectively pre- and post-contrast brain MR images, gadolinium enhancement was extremely helpful in 4 of 65 (6%) patients (all with tumor disease) and moderately helpful in 4 (6%) patients. However, in all of these lesions unenhanced MRI already disclosed abnormal findings. Patients with normal unenhanced MRIs did not prove any diagnostic benefit from contrast medium application. In concordance with previous studies the authors concluded that application of gadolinium-based contrast media as a matter of routine is not justified, yet after abnormal findings in unenhanced MRI and in the setting of suspected tumor or inflammatory brain disease, contrast medium should be employed.

Baierl et al. [21] examined 40 children and adolescents between 1 year and 16 years of age with suspected neoplasia, infection or vascular diseases. Twelve of 40 (30%) children in this cohort had normal unenhanced MRI. In 2 of these 12 cases (17%) the application of gadolinium led to a tumor diagnosis. However, it should be considered that on the one hand, the authors only examined children with the suspected diagnosis, for which a contrast-medium administration had also already been found to be helpful by other studies. On the other hand, the study dates to 1990 and it has to be expected that contemporary unenhanced sequences are of significantly better quality because of their higher magnetic flux density (tesla) and higher spatial resolution. Despite that, this study recommended that contrast medium should be applied in cases of abnormal unenhanced MR images or for suspicion of tumor or infection.

Normal unenhanced MRI

A relevant additional finding was found only in 8 of 3,003 of our patients (0.3%) after gadolinium application, all of them presenting as pathological meningeal enhancement. With a negative predictive value of 0.997, the probability of missing a relevant pathological meningeal enhancement in the case of a normal unenhanced MRI is approximately 3 per mill. In order to evaluate the impact for diagnosis and treatment of the finding in our 8 patients with pathological meningeal enhancement, we analyzed every case in detail (Table 3).

In four children bacterial meningitis was causative for the pathological meningeal enhancement (meningitis from E. coli, Haemophilus influenzae, pneumococci, neuroborreliosis). The enhancement could either be interpreted as a result of inflammation or as a sequela of a previous lumbar puncture and was not regarded as an unexpected finding.

In one child with nonspecific neurological complaints (headaches, dizziness, vomiting) the pathological meningeal enhancement finally led to the diagnosis of viral meningitis from enterovirus (proved by lumbar puncture).

In another child the MRI was conducted after a traffic accident. Because of fever in conjunction with a fracture of the temporal bone, the MRI was performed to rule out intracranial inflammation. The meningeal enhancement in this case can be interpreted as meningeal irritation after trauma.

In one child with Burkitt lymphoma a lumbar puncture was conducted before the MRI, implying the assumption that this enhancement might be as equal of this procedure.

In one child presenting with headaches, dizziness and nausea the pathological meningeal enhancement was also interpreted as a result of suspected meningitis. However, lumbar puncture failed to confirm this diagnosis. At follow-up examinations no meningeal pathology was found. Therefore, the meningeal contrast enhancement remains nonspecific without probable impact.

Abnormal unenhanced MRI

In 3,680 (55.1%) of our patients, there were anomalies in unenhanced examinations. In these cases contrast medium served to further characterize lesions marked as abnormal even in unenhanced MRI.

There is a consensus that in the presence of pathological results in unenhanced MRI, contrast medium should be administered to further characterize these lesions. It improves the exact description of the overall appearance, the borders and vascularization [25]. Thus contrast-enhanced sequences are a part of diagnostic criteria in several diseases, e.g., autoimmune diseases. In our study, 8.1% of cases had additional findings after contrast-medium administration located beyond the abnormal areas already revealed in pre-contrast sequences. In three quarters of these children, MRIs were performed for (suspected or known) neoplasia. In tumor diagnostics the benefit of contrast-enhanced images has been proved in several studies [20, 21, 22, 23, 24]. Therefore, gadolinium-based contrast medium should be administered in adults and children when a tumor is suspected. However, absence of unenhanced abnormal lesions is also of diagnostic value. Eldevik and Brunberg [22] found that in 22 of 125 children younger than 2 years a negative enhancement provided helpful diagnostic information.

Our study has several limitations, including lack of clinical information. Many children got an MRI examination with contrast medium for several indications. Furthermore, all data were gathered from only one institution, the University Hospital in Leipzig, making the results comparable only to children’s hospitals of the same size and structure.

Conclusion

Our study demonstrates that contrast-medium administration in cases of normal unenhanced MRI in children does not provide additional information in 99.7%; in 0.3% only a pathological meningeal enhancement was revealed after application of contrast medium. The results of our study argue against contrast-medium administration as a matter of routine in MR brain scanning. This provides evidence that gadolinium-based contrast-enhanced MR should be reserved for further characterization of equivocal lesions that were detected in the non-contrast scan and in selected individual cases of a tumor or infection for exclusion of meningeal involvement.

Notes

Compliance with ethical standards

Conflicts of interest

None

References

  1. 1.
    Kanal E (2016) Gadolinium based contrast agents (GBCA): safety overview after 3 decades of clinical experience. Magn Reson Imaging 34:1341–1345CrossRefPubMedGoogle Scholar
  2. 2.
    Collidge TA, Thomson PC, Mark PB et al (2007) Gadolinium-enhanced MR imaging and nephrogenic systemic fibrosis: retrospective study of a renal replacement therapy cohort. Radiology 245:168–175CrossRefPubMedGoogle Scholar
  3. 3.
    Todd DJ, Kay J (2016) Gadolinium-induced fibrosis. Annu Rev Med 67:273–291CrossRefPubMedGoogle Scholar
  4. 4.
    Todd DJ, Kagan A, Chibnik LB et al (2007) Cutaneous changes of nephrogenic systemic fibrosis: predictor of early mortality and association with gadolinium exposure. Arthritis Rheum 56:3433–3441CrossRefPubMedGoogle Scholar
  5. 5.
    Grobner T (2006) Gadolinium -- a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol Dial Transplant 21:1104–1108CrossRefPubMedGoogle Scholar
  6. 6.
    Marckmann P, Skov L, Rossen K et al (2006) Nephrogenic systemic fibrosis: suspected causative role of gadodiamide used for contrast-enhanced magnetic resonance imaging. J Am Soc Nephrol 17:2359–2362CrossRefPubMedGoogle Scholar
  7. 7.
    Quattrocchi CC, Mallio CA, Errante Y et al (2015) Gadodiamide and dentate nucleus T1 hyperintensity in patients with meningioma evaluated by multiple follow-up contrast-enhanced magnetic resonance examinations with no systemic interval therapy. Investig Radiol 50:470–472CrossRefGoogle Scholar
  8. 8.
    Radbruch A, Weberling LD, Kieslich PJ et al (2015) Gadolinium retention in the dentate nucleus and globus pallidus is dependent on the class of contrast agent. Radiology 275:783–791CrossRefPubMedGoogle Scholar
  9. 9.
    Kanda T, Osawa M, Oba H et al (2015) High signal intensity in dentate nucleus on unenhanced T1-weighted MR images: association with linear versus macrocyclic gadolinium chelate administration. Radiology 275:803–809CrossRefPubMedGoogle Scholar
  10. 10.
    Murata N, Gonzalez-Cuyar LF, Murata K et al (2016) Macrocyclic and other non-group 1 gadolinium contrast agents deposit low levels of gadolinium in brain and bone tissue: preliminary results from 9 patients with normal renal function. Investig Radiol 51:447–453CrossRefGoogle Scholar
  11. 11.
    White GW, Gibby WA, Tweedle MF (2006) Comparison of Gd(DTPA-BMA) (Omniscan) versus Gd(HP-DO3A) (ProHance) relative to gadolinium retention in human bone tissue by inductively coupled plasma mass spectroscopy. Investig Radiol 41:272–278CrossRefGoogle Scholar
  12. 12.
    Weberling LD, Kieslich PJ, Kickingereder P et al (2015) Increased signal intensity in the dentate nucleus on unenhanced T1-weighted images after gadobenate dimeglumine administration. Investig Radiol 50:743–748CrossRefGoogle Scholar
  13. 13.
    Zhang Y, Cao Y, Shih GL et al (2017) Extent of signal hyperintensity on unenhanced T1-weighted brain MR images after more than 35 administrations of linear gadolinium-based contrast agents. Radiology 282:516–525CrossRefPubMedGoogle Scholar
  14. 14.
    Miller JH, Hu HH, Pokorney A et al (2015) MRI brain signal intensity changes of a child during the course of 35 gadolinium contrast examinations. Pediatrics 136:e1637–e1640Google Scholar
  15. 15.
    Roberts DR, Holden KR (2016) Progressive increase of T1 signal intensity in the dentate nucleus and globus pallidus on unenhanced T1-weighted MR images in the pediatric brain exposed to multiple doses of gadolinium contrast. Brain Dev 38:331–336Google Scholar
  16. 16.
    McDonald RJ, McDonald JS, Kallmes DF et al (2015) Intracranial gadolinium deposition after contrast-enhanced MR imaging. Radiology 275:772–782CrossRefPubMedGoogle Scholar
  17. 17.
    Flood TF, Stence NV, Maloney JA et al (2017) Pediatric brain: repeated exposure to linear gadolinium-based contrast material is associated with increased signal intensity at unenhanced T1-weighted MR imaging. Radiology 282:222–228CrossRefPubMedGoogle Scholar
  18. 18.
    Roberts DR, Chatterjee AR, Yazdani M et al (2016) Pediatric patients demonstrate progressive T1-weighted hyperintensity in the dentate nucleus following multiple doses of gadolinium-based contrast agent. AJNR Am J Neuroradiol 37:2340–2347CrossRefPubMedPubMedCentralGoogle Scholar
  19. 19.
    Hu HH, Pokorney A, Towbin RB, Miller JH (2016) Increased signal intensities in the dentate nucleus and globus pallidus on unenhanced T1-weighted images: evidence in children undergoing multiple gadolinium MRI exams. Pediatr Radiol 46:1590–1598Google Scholar
  20. 20.
    Elster AD, Rieser GD (1989) Gd-DTPA-enhanced cranial MR imaging in children: initial clinical experience and recommendations for its use. AJR Am J Roentgenol 153:1265–1268CrossRefPubMedGoogle Scholar
  21. 21.
    Baierl P, Muhlsteffen A, Haustein J et al (1990) Comparison of plain and Gd-DTPA-enhanced MR-imaging in children. Pediatr Radiol 20:515–519CrossRefPubMedGoogle Scholar
  22. 22.
    Eldevik OP, Brunberg JA (1994) Gadopentetate dimeglumine-enhanced MR of the brain: clinical utility and safety in patients younger than two years of age. AJNR Am J Neuroradiol 15:1001–1008PubMedGoogle Scholar
  23. 23.
    Petrou M, Foerster B, Maly PV et al (2007) Added utility of gadolinium in the magnetic resonance imaging (MRI) workup of seizures in children younger than 2 years. J Child Neurol 22:200–203CrossRefPubMedGoogle Scholar
  24. 24.
    Foerster BR, Ksar J, Petrou M et al (2006) Value of gadolinium in brain MRI examinations for developmental delay. Pediatr Neurol 35:126–130CrossRefPubMedGoogle Scholar
  25. 25.
    Gutierrez JE, Rosenberg M, Seemann J et al (2015) Safety and efficacy of gadobutrol for contrast-enhanced magnetic resonance imaging of the central nervous system: results from a multicenter, double-blind, randomized, comparator study. Magn Reson Insights 8:1–10PubMedPubMedCentralGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany 2017

Authors and Affiliations

  1. 1.Department of Paediatric RadiologyUniversity LeipzigLeipzigGermany
  2. 2.Department of Neurosurgery/Paediatric NeurosurgeryUniversity LeipzigLeipzigGermany
  3. 3.Department of Woman and Child Health, Hospital for Children & Adolescents, University LeipzigLeipzigGermany

Personalised recommendations