Ultrasound Obstet Gynecol 2014; 44: 188–196 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.13243
Accuracy of neurosonography and MRI in clinical management of fetuses referred with central nervous system abnormalities D. PALADINI*†, M. QUARANTELLI‡, G. SGLAVO*, G. PASTORE*, A. CAVALLARO*, M. R. D’ARMIENTO§, M. SALVATORE‡ and C. NAPPI¶ *Fetal Medicine and Cardiology Unit, Department of Gynecology and Obstetrics, University Federico II of Naples, Naples, Italy; †Fetal Medicine and Surgery Unit, Giannina Gaslini Institute, Genoa, Italy; ‡Biostructure and Bioimaging Institute, National Research Council, Naples, Italy; §Department of Pathology, University Federico II of Naples, Naples, Italy; ¶Department of Gynecology and Obstetrics, University Federico II of Naples, Naples, Italy
K E Y W O R D S: 3D ultrasound; CNS; fetus; magnetic resonance imaging; space-occupying lesion
A B S T R AC T Objective To assess the accuracy of expert neurosonography (two- and three-dimensional NSG) in the characterization of major fetal central nervous system (CNS) anomalies seen at a tertiary referral center and to report the differential clinical usefulness of magnetic resonance imaging (MRI) used as a second-line diagnostic procedure in the same cohort. Methods This was a retrospective analysis of all 773 fetuses with confirmed CNS abnormalities referred to our center between 2005 and 2012. The following variables were analyzed: gestational age at NSG and MRI, NSG and MRI diagnoses, indication for MRI (confirmation of NSG findings; diagnostic doubt; search for possible additional brain anomalies), association with other malformations, diagnostic accuracy of NSG vs MRI (no additional clinical value for either MRI or NSG; additional information with clinical/prognostic significance on MRI relative to NSG; additional information with clinical/prognostic significance on NSG relative to MRI, NSG and MRI concordant but incorrect) and final diagnosis, which was made at autopsy or postnatal MRI/surgery. Results CNS malformations were associated with other anomalies in 372/773 (48.1%) cases and were isolated in the remaining 401 (51.9%) cases. NSG alone was able to establish the diagnosis in 647/773 (83.7%) cases. MRI was performed in 126 (16.3%) cases. The indication for MRI was: confirmation of NSG diagnosis in 59 (46.8%) cases; diagnostic query (in the case of inconclusive or uncertain finding on NSG) in 20 (15.9%) cases; search for possible additional brain anomalies in 47 (37.3%) cases. NSG and MRI were concordant and correct in 109/126
(86.5%) cases. Clinically relevant findings were evident on MRI alone in 10/126 (7.9%) cases (1.3% of the whole population) and on NSG alone in 6/126 (4.8%) cases; in all six of these cases, MRI had been performed at < 24 weeks of gestation. In one case, both NSG and MRI diagnoses were incorrect. The main type of malformation in which MRI played an important diagnostic role was space-occupying lesions, MRI identifying clinically relevant findings in 42.9% (3/7) of these cases. Conclusions (1) In a tertiary referral center with good NSG expertise in the assessment of fetal CNS malformations, MRI is likely to be of help in a limited proportion of cases; (2) MRI is more useful after 24 weeks of gestation; (3) the lesions whose diagnosis is most likely to benefit from MRI are gross space-occupying lesions. Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
I N T RO D U C T I O N Central nervous system (CNS) malformations, together with congenital heart disease, represent the group of congenital anomalies most difficult to characterize prenatally. For this reason, new imaging modalities are being used increasingly in the study of normal and abnormal development of the fetal brain: in addition to two-dimensional (2D) ultrasound (US), three-dimensional (3D) US and magnetic resonance imaging (MRI) are currently employed in the diagnosis of fetal CNS abnormalities, though with different indications and results1 – 6 . A particular area of controversy is the clinical usefulness of MRI relative to that of or as an adjunct to neurosonography (NSG), the latter being defined as US examination of the fetal brain performed by an experienced sonologist
Correspondence to: Prof. D. Paladini, Viale Teano 61 – 16147 Genoa, Italy (email: [email protected]) Accepted: 22 October 2013
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
ORIGINAL PAPER
MRI in management of fetal CNS malformations using a multiplanar, possibly transvaginal approach, as reported in the recently published guidelines of the International Society of Ultrasound in Obstetrics & Gynecology (ISUOG)6 . In fact, the yield of ‘clinically relevant information’ provided by MRI in comparison with that of 2D or 3D NSG ranges from 7% to 40%1 – 5,7 – 9 . In our center, the management protocol for fetuses with CNS malformations has always been multimodal, with detailed NSG being performed in all fetuses with suspected brain abnormalities that are referred to our unit; MRI is only performed as a second-line diagnostic procedure in selected cases. In this retrospective analysis, we describe our experience in the clinical management of 773 fetuses with suspected brain anomalies referred to our unit for diagnosis and management over an 8-year period (January 2005–December 2012). The objectives of this study were: to assess the accuracy of expert NSG (2D and 3D US) in the characterization of major fetal CNS anomalies; and to report the differential clinical usefulness of MRI as a second-line diagnostic procedure in the same cohort.
METHODS During the 8-year study period from 2005 to 2012, there were 834 cases of suspected fetal CNS abnormalities referred to our unit for final diagnosis and management. Of these, 61 were excluded from this analysis because they were lost to follow-up or because an autopsy was not available, leaving the sample population of 773 fetuses with CNS anomalies. If no brain anomaly was detected we did not take into account MRI performed for other indications, such as in monochorionic twins undergoing laser coagulation of placental anastomoses for twin-to-twin transfusion syndrome, or in monochorionic or dichorionic twin survivors after cord occlusion or spontaneous intrauterine demise. For all cases, the following variables were retrieved from the computerized database and analyzed: gestational age at NSG and MRI, US diagnosis, indication for MRI, MRI diagnosis, associated anomalies and final diagnosis. Malformations occurring in the study population are summarized in Table 1. In fetuses with multiple brain malformations, classification was according to organogenesis, each case being defined by the malformation that would have developed first. For example, fetuses with holoprosencephaly and agenesis of the corpus callosum were considered as having holoprosencephaly; those with anencephaly and holoprosencephaly were considered as having anencephaly; those with Dandy–Walker malformation and agenesis of the corpus callosum were considered to have Dandy–Walker malformation. In addition, to facilitate the analysis, malformations were categorized into six groups: 1) anomalies of the corpus callosum and cavum septi pellucidi; 2) anomalies of the posterior fossa; 3) primary ventriculomegaly (including mild (borderline) ventriculomegaly); 4) intracranial hemorrhage; 5) space-occupying lesions (arachnoid cysts, tumors,
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
189
arteriovenous malformations); 6) others (neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and other anomalies). The first step in our diagnostic protocol for fetal CNS anomalies includes 2D and 3D NSG. Only if the expert neurosonologist deems it necessary or advisable to perform MRI is this proposed to the parents and, if they consent, this is scheduled as a second-level test. Expert 2D and 3D NSG was performed using a transvaginal approach in all cases of vertex presentation, whether this was spontaneous or after external version; only in the case of failed version and persistent breech presentation was the scan performed transabdominally. High-end 3D equipment (GE E8 and Voluson 730 Expert US machines (GE Medical Systems, Zipf, Austria), equipped with 5–9-MHz volumetric transvaginal transducers and a 4–8-MHz volumetric convex transducer, and the former with a 6–12-MHz volumetric transvaginal transducer) was used for all examinations. 3D volumes were acquired in more than 95% of cases and analyzed offline with dedicated software (4D View, GE Medical Systems). The examination protocol followed that described in the ISUOG guidelines for US assessment of the fetal brain6 . All examinations were performed by a single experienced operator (D.P.). In cases which did not undergo termination of pregnancy (TOP), follow-up scans were performed at 3–4-week intervals. When comparing the diagnostic accuracy of NSG with that of MRI, only findings detected prior to MRI were taken into account. Fetal MRI studies were performed with a Philips Achieva 1.5 T system (amplitude on axis, 30 mT/m; slew rate on axis, 120 T/m/s gradients, equipped with a four-channel phased-array body coil; Philips Healthcare, Andover, MA, USA), using Single Shot Fast Spin Echo sequences to produce high-resolution T2-weighted images (echo time (TE), 120 ms; echo train length (ETL), 96; slice thickness, 3–4 mm), fast-gradient-echo (GrE) sequences to obtain axial T1-weighted images (repetition time (TR)/TE, 154/4.6 ms; flip angle (FA), 80∘ ; slice thickness, 5 mm), and echo planar imaging (EPI) for diffusion-weighted imaging (DWI) (TR/TE, 2000/90 ms; Max B-factor, 600; slice thickness, 5 mm; with calculation of trace and apparent diffusion coefficient (ADC) maps). All sequences were acquired without breath-hold. In cases of suspected hemorrhagic lesions, besides T1-weighted sequences, the B0 image of the EPI-DWI sequence was also used. Neither maternal sedation nor fetal curarization was used during the examination. The expert neuroradiologist (M.Q.) who performed MRI was aware of the NSG findings and of any specific queries resulting from the neurosonographic examination (see below). The indications for MRI fell into three categories: 1) confirmation of NSG diagnosis; 2) diagnostic query (in case of inconclusive or uncertain findings on NSG); 3) search for possible additional brain anomalies, if the final diagnosis at NSG implied the possibility of additional abnormalities not always diagnosable by US (e.g. heterotopia, cortical lesions). In general, MRI was carried out at
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
190
Table 1 Categorization of 773 malformations of the central nervous system diagnosed at our referral center in the period 2005–2012, according to association with other congenital anomalies and termination of pregnancy (TOP) Type of malformation Holoprosencephaly Neural tube defect Anencephaly Cephalocele Open spina bifida Closed spina bifida Posterior fossa malformation Dandy–Walker malformation Inferior vermian hypoplasia Blake’s pouch cyst Megacisterna magna Other Corpus callosal dysraphism Ventriculomegaly Cavum septi pellucidi abnormality Intracranial hemorrhage Schizencephaly Arteriovenous malformation Hydranencephaly Microcephaly Neuronal migration disorder Tumor Arachnoid cyst Craniosynostosis Infection Cytomegalovirus Toxoplasmosis Diastematomyelia Total
Isolated
Associated
TOP
17 (33.3) 118 (73.7) 31 12 72 3 73 (43.9) 14 12 21 14 12† 37 (35.9) 80 (49.1) 8 (80.0) 25 (100) 1 (14.2) 6 (85.8) 5 (83.3) 6 (35.3) 2 (12.5) 8 (100) 4 (33.3) 6 (54.5)
34 (66.7) 42 (26.3) 4 16 22 — 93 (56.1) 38 35 13 7 — 66 (64.1) 83 (50.9) 2 (20.0) — 6 (85.8) 1 (14.2) 1 (16.7) 11 (64.7) 14 (87.5) — 8 (66.7) 5 (45.5)
50 (98)* 150 (94) 33 (94)* 27 (96) 89 (95) 1 (33) 115 (69) 50 (96) 45 (96) 14 (41) 6 (29)
1 (25.0) — 4 (80.0) 401 (51.9)
3 (75.0) 2 (100) 1 (20.0) 372 (48.1)
3 (75) — 4 (80) 560 (72.4)
97 (94) 129 (79) 9 (90) 8 (32) 7 (100) 2 (29) 5 (83) 9 (53) 15 (94)‡ 1 (12) 1 (8) 5 (45)
Data are given as n (%). *Includes twin pregnancy. †Chiari I, hemorrhage, asymmetrical hemispheres. ‡Note this group does not include cases associated with corpus callosal agenesis and other major brain anomalies which have been included in other categories.
28–30 gestational weeks for the latter indication, whereas it was performed within 1 week of the initial NSG diagnosis for the other two indications. After initial MRI at < 24 weeks, follow-up MRI was performed at 28–30 weeks if necessary. With respect to the final diagnosis, made at autopsy or postnatal MRI or surgery, the following diagnostic categories were defined: 1. No additional value of MRI or NSG 2. NSG–MRI discordance; NSG correct, i.e. additional value of NSG 3. NSG–MRI discordance; MRI correct, i.e. additional value of MRI 4. NSG and MRI concordant but incorrect The first category included both cases in which MRI confirmed the NSG diagnosis, and cases in which additional but clinically irrelevant findings were evident on either modality. Additional information provided by NSG or MRI in a particular case was considered clinically relevant if it would affect the diagnosis and, consequently, change the prognosis significantly, according to current knowledge. Over the years, alobar/semilobar holoprosencephaly and exencephaly/anencephaly were diagnosed in a significant number of fetuses at 11–14 weeks. In
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
these cases, an autopsy was not performed, but we assumed that the NSG diagnosis was confirmative per se and, consequently, these cases were included in the analysis. It should also be borne in mind that, although in some cases (e.g. agenesis of the corpus callosum), MRI was recommended routinely, this may have been declined by the parents if it would not modify the clinical management: for example, if the diagnosis was made before 24 weeks’ gestation and the amount of information provided on the basis of the NSG findings was considered sufficient to opt for termination of pregnancy.
R E S U LT S Among the study population of 773 fetuses with CNS anomalies, these were isolated in 401 (51.9%) cases and there were other associated anomalies in 372 (48.1%) cases (Table 1). The mean gestational age at NSG was 21 (range, 13–38) weeks for the whole group, 20 (range, 13–38) weeks for cases undergoing NSG only and 24 (range, 17–36) weeks for cases undergoing both NSG and MRI. NSG alone was able to establish the diagnosis in 647/773 (83.7%) cases. This group included 134 cases in which an MRI was requested but not performed, either because the patient declined the exam (n = 127; opted
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations directly for TOP) or because claustrophobia necessitated termination of the exam before it was complete (n = 7). In this group, the primary diagnosis was correct in all cases, but in 45/647 (7%) cases autopsy revealed additional brain malformations: mainly subcortical/subependymal nodular heterotopia, gyration abnormalities and hemorrhage, found in cases of agenesis of the corpus callosum, Dandy–Walker malformation and ventriculomegaly. MRI was performed in 126 (16.3%) cases (Table 2). In this group, the mean gestational age at NSG was 24 (range, 17–36) weeks, with 73/126 (57.9%) cases being evaluated for the first time before 24 weeks. The mean gestational age at MRI was 27 (range, 21–36) weeks, with two-thirds (85/126, 67.5%) of the cases being evaluated after 24 weeks. The indication for MRI was: confirmation of the NSG diagnosis in 59 (46.8%) cases; diagnostic query (when findings on NSG were inconclusive or uncertain) in 20 (15.9%) cases; search for possible additional brain anomalies in 47 (37.3%) cases. In general, MRI was more likely to be performed for midline anomalies, including those of the corpus callosum, cavum septi pellucidi and posterior fossa, for ventriculomegaly and for space-occupying lesions than it was for open neural tube defects and ventral induction defects (Table 2). With respect to the final diagnoses, NSG and MRI were concordant and correct in 109/126 (86.5%) cases. Additional clinically relevant findings were evident on MRI in 10 (7.9%) cases (Table 3) and on NSG in six (4.8%) cases (Table 4). In one case, both NSG and MRI were incorrect. This was a case of atypical frontoethmoidal cephalocele in which both NSG and MRI identified a solid tumor-like lesion rather than the neural tube defect (Figure 1). In fact, the lesion could be demonstrated adequately on retrospective analysis of MRI slices, and similarly it could be suspected on retrospective analysis of 3D-US volumes; however, we considered this to be an incorrect diagnosis for both techniques, although it was the physician who misinterpreted the images at the time of diagnosis. The same applies to the case of lissencephaly missed prospectively on NSG and recognized later on retrospective review of the case. Assessing final diagnosis with respect to the indication for MRI (Table 5 and Figure 2) suggested that MRI had a very limited contribution to the final diagnosis if it had been requested as confirmation of the NSG finding or to look for additional subtle abnormalities that are more evident at MRI than at NSG; it seemed more helpful in cases of diagnostic doubt, although in this category there was also the highest rate of incorrect diagnoses for MRI. Analyzing final diagnosis according to malformation category (Table 6 and Figure 3) indicated that the only category in which MRI played an important diagnostic role was space-occupying lesions, followed by ‘other’ and ‘posterior fossa’ categories. A detailed analysis of the cases in which NSG or MRI was misleading, missing the final diagnosis, revealed that the diagnosis had been performed in the second trimester (Table 4). Excluding cases undergoing MRI < 24 weeks
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
191 Table 2 Categorization of 126 fetuses with central nervous system malformations undergoing magnetic resonance imaging (MRI), according to type of anomaly among the 773 cases assessed in the study period Type of anomaly Holoprosencephaly Alobar Semilobar Lobar Middle interhemispheric Neural tube defect Anencephaly Cephalocele Open spina bifida Closed spina bifida Posterior fossa malformation Dandy–Walker malformation Inferior vermian hypoplasia Blake’s pouch cyst Megacisterna magna Other Corpus callosal dysraphism Agenesis Hypoplasia Thick corpus callosum Ventriculomegaly Mild (borderline) Severe Cavum septi pellucidi abnormality Intracranial hemorrhage Schizencephaly Arteriovenous malformation Hydranencephaly Microcephaly Neuronal migration disorder Tumors Arachnoid cyst Craniosynostosis Infection Cytomegalovirus Toxoplasmosis Diastematomyelia Total
US
MRI
51 32 15 3 1 160 35 28 94 3 166 52 47 34 21 12 103 95 5 3 163 44 119 10 25 7 7 6 17 16 8 12 11 6 4 (8) 2 (14) 5 773
2 (3.9) — — 1 1 2 (1.2) — — — 2 31 (18.7) 2 6 10 4 9 16 (15.5) 12 4 — 41 (25.1) 30 11 7 (70) 7 (28) 1 — — — 5 3 5 2 3 2 1 1 126 (16.3)
Data are given as n or n (%).
left only one case in which NSG performed better than did MRI (Table 4, Case 2), while MRI added clinically relevant information in 7/85 (8.2%) cases.
DISCUSSION In a population referred to a tertiary center for diagnosis and management of CNS abnormalities, it was deemed worthwhile to request MRI as a second-level diagnostic test in 260 of 773 (33.6%) cases, though eventually it was performed in only 16.3% (126/773). Excluding the seven cases in which MRI had to be interrupted due to maternal claustrophobia, patients often declined undergoing MRI because, in fetuses with multiple malformations, which accounted for half of the series (48.1%; 372/773), as well as in those with severe albeit isolated brain malformations, the prognosis was felt to be so poor that the decision to opt
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
192
Table 3 Characteristics of the 10 cases with central nervous system malformations in which magnetic resonance imaging (MRI) provided clinically relevant information additional to that obtained by NSG
Case
Indication
1 2 3 4 5 6 7 8 9
Confirmation* Confirmation* Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Search addtn. anom.‡
10
Search addtn. anom.‡
GA at MRI (weeks)
NSG diagnosis
MRI (and final) diagnosis
23 31 32 22 30 28 22 23 30
Chiari I malformation Pericallosal lipoma Arachnoid cyst Arachnoid cyst or schizencephaly Ventriculomegaly Ventriculomegaly Megacisterna magna or arachnoid cyst Mildly dysmorphic cerebellum Triventricular hydrocephalus
32
Ventriculomegaly, subependymal heterotopia?
Chiari I + craniosynostosis Pericallosal lipoma + lissencephaly§ Arachnoid cyst + pACC + delayed myelination Arachnoid cyst + closed lip schizencephaly Ventriculomegaly + abn. sulcation¶ Ventriculomegaly + abn. sulcation¶ Arachnoid cyst Cerebellar hemorrhagic necrosis Triventricular hydrocephalus + tumor in cisterna ambiens (pinealoblastoma) Ventriculomegaly + severe edema
*Confirmation of NSG diagnosis. †Diagnostic query in case of inconclusive or uncertain finding on NSG. ‡Search for additional brain anomalies. §Evident on retrospective evaluation of NSG volumes. ¶Two full siblings. Abn., abnormal; GA, gestational age at MRI; pACC, partial agenesis of the corpus callosum. Table 4 Characteristics of the six cases with central nervous system malformations in which neurosonography (NSG) provided clinically relevant information additional to that obtained by magnetic resonance imaging (MRI)
Case
Indication
1 2 3 4 5 6
Confirmation† Confirmation† Confirmation† Diagnostic query‡ Diagnostic query‡ Diagnostic query‡
GA at MRI* (weeks) 22 23 21 22 22 21
NSG (and final) diagnosis
MRI diagnosis
Subdural diffuse hemorrhage§ Partial agenesis of CSP Cerebellar hypoplasia Lobar holoprosencephaly Lissencephaly?** Inferior vermian hypoplasia
No hemorrage, ample subdural spaces Normal CSP¶ Cerebellar diameters at 5th centile pACC Normal†† Blake’s pouch cyst
*MRI was performed within 1 week following neurosonography, unless indicated. †Confirmation of NSG diagnosis. ‡Diagnostic query in case of inconclusive or uncertain finding on NSG. §Patient with renal hemorrhage subsequently diagnosed with fetomaternal alloimmune thrombocytopenia. ¶MRI performed at 28 gestational weeks. **Very difficult examination due to premature ossification of the cranial sutures. ††Diffuse atrophy + lissencephaly of frontal lobes on follow-up MRI at 29 weeks. CSP, cavum septi pellucidi; GA, gestational age; pACC, partial agenesis of the corpus callosum.
for TOP was taken before considering the possible clinical relevance of MRI. From this point onwards, all figures and percentages discussed refer only to the 126 cases undergoing both NSG and MRI; and, in order to appreciate the role of MRI in the diagnosis of fetal CNS malformations, the reader should bear in mind that this discussion focuses only on 16.3% of the whole series of 773 cases undergoing expert evaluation at the referral center. With respect to the diagnostic performance of NSG and MRI, both proved highly accurate overall (91.3% (115/126) for NSG and 94.4% (119/126) for MRI, with the accuracy increasing if both were employed because the incorrect diagnoses at NSG differed from those at MRI (Tables 3 and 4). However, before 24 gestational weeks (73 cases), the clinical relevance of MRI in comparison with NSG was lower (93.1% accuracy for NSG and 90.4% for MRI). Overall, MRI provided clinically relevant information missed by NSG in 7.9% (10/126) of cases (Table 3), while the reverse occurred in six cases (4.8%, Table 4). These data compare favorably with only some of those reported in the literature. Malinger et al.9 reported additional clinically relevant information in 7.1% (3/42) of cases undergoing MRI over a period of
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
2 years. However, in their center, there are > 400 referrals per year (pers. comm.). It can be considered, therefore, that in tertiary referral centers where expert NSG is available, the percentage of fetuses with suspected brain abnormalities in which MRI is likely to be helpful is in the range of 5–10%, which agrees well with the 7.9% of our study. Altogether these data (ours and those of Malinger et al.9 ) are rather different from those reported in other papers, in which either MRI was used as a back-up in the absence of an expert neurosonographer8 or it was employed to make some diagnoses that could probably have been made with NSG alone7 . In these papers, the apparent clinical usefulness of MRI was 31%8 and 48%7 , respectively. However, we share the opinion of Malinger et al. that expert NSG is capable of characterizing most CNS abnormalities and that MRI is needed in selected cases with specific indications or queries10 . With this multimodal approach, in our study, MRI added clinically relevant information in 7.9% of cases undergoing expert NSG and MRI, which, depending upon the sonologist’s experience and the spectrum of CNS anomalies, may represent 10–20% of all fetuses with CNS malformations referred to a tertiary center.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations
193
Figure 1 In this case with a final diagnosis after delivery of fetal anterior frontoethmoidal cephalocele (indicated by arrow in all figure parts), both neurosonography at 23 gestational weeks and magnetic resonance imaging (MRI) at 28 weeks incorrectly diagnosed frontoethmoidal or orbital tumor. (a,b) Two-dimensional ultrasound (US): (a) axial view of the orbits; (b) profile modified view. (c) Three-dimensional US: surface-rendering of the fetal face. (d) MR T2-weighted image: profile modified plane. (e) Neonate at delivery, with cephalocele indicated (arrow). On retrospective evaluation, the lesion could be diagnosed correctly on MRI and could perhaps be suspected on US. Table 5 Indication for magnetic resonance imaging (MRI) according to diagnostic category relative to final diagnosis in 125 fetuses* with central nervous system malformations undergoing neurosonography (NSG) and MRI Final diagnosis¶ Indication for MRI
No add. value of MRI
Confirmation† Diagnostic query‡ Search addtn. anom.§ Total
54 (91.5) 11 (57.9) 44 (93.6) 109 (87.2)
Add. value of NSG 3 (5.1) 3 (15.8) 0 6 (4.8)
Add. value of MRI 2 (3.4) 5 (26.3) 3 (6.4) 10 (8.0)
Total 59 (47.2) 19 (15.2) 47 (37.6) 125 (100.0)
Data are given as n (%). *In one of the 126 cases (shown in Figure 1), interpretation of both NSG and MRI was incorrect. †Confirmation of NSG diagnosis. ‡Diagnostic query in case of inconclusive or uncertain finding on NSG. §Search for additional brain anomalies. ¶Final diagnoses: No additional value of MRI; NSG–MRI discordant, NSG correct (i.e. additional value of NSG); NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Add., additional.
Considering the results according to indication (Table 5, Figure 2), MRI proved most useful in comparison to NSG in the 19 cases referred for ‘diagnostic doubt’, with 26.3% (5/19) of these cases benefiting from the addition of MRI. It was in this same subset of cases, however, that NSG proved most useful in comparison to MRI, there
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
being 15.8% (3/19) of these cases in which NSG was more informative than MRI. It is noteworthy that the clinical usefulness of MRI was rather low both in cases referred for ‘confirmation of NSG diagnosis’ (2/59; 3.4%) and in those referred to ‘search for possible additional brain anomalies’ (3/47; 6.4%). The explanations for these
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
194
60
40
50 Number of cases
Number of cases
30
40 30 20
20
10
10 0
Confirmation*
Diagnostic query†
Search addtn. anom.‡
0
CC & CSP
Indication for MRI
Figure 2 Distribution of the 125 cases undergoing expert neurosonography (NSG) and MRI according to indication for MRI (excluding the wrong diagnosis shown in Figure 1, in which both techniques failed to demonstrate the cephalocele). Diagnostic categories relative to final diagnosis are indicated: , no additional value of MRI; , NSG–MRI discordant, NSG correct (i.e. additional value of NSG); , NSG–MRI discordant, MRI correct (i.e. additional value of MRI). *Confirmation of NSG diagnosis. †Diagnostic query in case of inconclusive or uncertain finding on NSG. ‡Search for additional brain anomalies.
two results are quite different, however. With respect to confirmation of NSG diagnosis, it is likely that the higher resolution of NSG made the difference, particularly as in this category about 40% of the cases were assessed with both techniques before 24 weeks. With respect to the search for additional anomalies, the failure to demonstrate usefulness of MRI relative to NSG was likely due to the small number of cases assessed (n = 47), such subtle, additional anomalies as heterotopia or cortical lesions being quite rare. Regarding assessment of the results according to type of lesion (Table 6 and Figure 3), it seems that MRI makes a considerable difference in those cases in which large
PF
Ventriculo. ICH
SOL
Other
Malformation category
Figure 3 Distribution of the 125 cases undergoing expert neurosonography (NSG) and MRI according to malformation category (excluding the wrong diagnosis shown in Figure 1, in which both techniques failed to demonstrate the cephalocele). Diagnostic categories relative to final diagnosis are indicated: , no additional value of MRI; , NSG–MRI discordant, NSG correct (i.e. additional value of NSG); , NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Malformations were categorized as follows: corpus callosum and cavum septi pellucidi anomalies (CC & CSP); posterior fossa anomalies (PF); primary ventriculomegaly (including mild (borderline) ventriculomegaly) (Ventriculo.); intracranial hemorrhage (ICH); space-occupying lesions (arachnoid cysts, tumors, arteriovenous malformations) (SOL); and other anomalies (neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and others) (Other).
lesions lead to significant distortion of brain anatomy: in 3/7 (42.9%) cases in this subgroup, MRI provided additional information that NSG had missed. Among more conventional pathologies, the posterior fossa seems the next most likely to benefit from MRI, as long as it is performed after 24 weeks. Finally, examining the cases with discordant NSG and MRI diagnoses, in the 10 cases in which MRI provided
Table 6 Malformation category according to diagnostic category relative to final diagnosis in 125 fetuses* with central nervous system malformations undergoing neurosonography (NSG) and magnetic resonance imaging (MRI) Final diagnosis¶ Malformation category
No add. value of MRI
CC and CSP anomalies Posterior fossa anomalies Primary ventriculomegaly† Intracranial hemorrhage Space-occupying lesions‡ Other malformations§ Total
22 (95.7) 26 (83.9) 39 (95.1) 6 (85.7) 4 (57.1) 12 (75.0) 109 (87.2)
Add. value of NSG 1 (4.3) 2 (6.5) 0 1 (14.3) 0 2 (12.5) 6 (4.8)
Add. value of MRI 0 3 (9.7) 2 (4.9) 0 3 (42.9) 2 (12.5) 10 (8.0)
Total 23 (18.4) 31 (24.8) 41 (32.8) 7 (5.6) 7 (5.6) 16 (12.8) 125 (100.0)
*In one of the 126 cases (shown in Figure 1), interpretation of both NSG and MRI was incorrect. †Including mild (borderline) ventriculomegaly. ‡Arachnoid cysts, tumors, arteriovenous malformations. §Neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and other anomalies. ¶Final diagnoses: No additional value of MRI; NSG–MRI discordant, NSG correct (i.e. additional value of NSG); NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Add., additional; CC, corpus callosum; CSP, cavum septi pellucidi.
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations
195
Figure 4 Imaging in the case that was false-negative at three-dimensional (3D) neurosonography. In this 31-week fetus, the following anomalies were detected on ultrasound: pre-auricular tags (a,d, arrowhead); unilateral cleft lip (b,e, arrow), pericallosal lipoma (c, arrow). Magnetic resonance imaging, performed 2 days later, showed gyral abnormalities consistent with lissencephaly (f, arrows). On retrospective review of 3D volume datasets, complete absence of the cingulate gyrus and a complete smooth cortical surface were noted (c, arrowhead), a finding overlooked at initial examination, probably due to the presence of multiple anomalies.
a clinically relevant additional or discordant piece of information, the 3D volume datasets were reevaluated retrospectively in an attempt to correctly identify the missed or incorrect NSG diagnoses. Only one such case was found: a fetus with multiple malformations assessed at 31 weeks of gestation. The fetus had a dysmorphic face, with pre-auricular tags, wrinkled ears, unilateral cleft lip, a pericallosal lipoma and lissencephaly (Figure 4). The latter finding had been missed initially at NSG; however, on retrospective reassessment of the volume the gyration abnormality was evident. In this case, the likely explanation is that the attention of the operator was diverted by the lipoma and the other non-CNS malformations. Considering, on the other hand, the six cases in which MRI proved incorrect (Table 4), it is evident that they underwent MRI before 24 gestational weeks; evidently, MRI before 24–25 weeks should be interpreted with caution. This study had some limitations. First, despite the fact that MRI is known to have relatively low accuracy when performed at mid-gestation, almost one third (41/126; 32.5%) of cases were evaluated before 24 weeks, due to the Italian law for TOP. This may have led
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
to an underestimation of the clinical usefulness of MRI. Second, regarding the analysis by subgroup, in some cases the numbers were so small as to limit significantly the relevance of any finding; for example, in the breakdown according to type of malformation (Table 6 and Figure 3). We conclude that: (1) in a tertiary referral center with expertise in the US assessment of fetal CNS, MRI is likely to be of help in a limited proportion of cases, being performed in 16% of the whole fetal population with CNS anomalies and proving clinically useful in 7.9% of this subgroup (1.3% of the whole population); (2) it is probably worth restricting the use of MRI, performing it only when: a) requested due to diagnostic doubt following NSG, b) only after 24 weeks of gestation, and c) mainly in the investigation of gross space-occupying lesions, rare anomalies and, to a lesser extent, posterior fossa and corpus callosal anomalies. For other indications (confirmation of NSG findings and search for additional subtle abnormalities) and pathologies (e.g. moderate ventriculomegaly, holoprosencephaly, open neural tube defects) the use of MRI should be weighed against the availability of economic and human resources.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
196
REF EREN CES 1. Simon EM, Goldstein RB, Coakley FV, Filly RA, Broderick KC, Musci TJ, Barkovich AJ. Fast MR imaging of fetal CNS anomalies in utero. Am J Neuroradiol 2000; 21: 1688–1698. 2. Peruzzi P, Corbitt RJ, Raffel C. Magnetic resonance imaging versus ultrasonography for the in utero evaluation of central nervous system anomalies. J Neurosurg Ped 2010; 6: 340–345. 3. Messing-Jünger MM, Röhrig A, Stressig R, Schaper J, Turowski B, Blondin D. Fetal MRI of the central nervous system: clinical relevance. Childs Nerv Syst 2009; 25: 165–171. 4. Kubik-Huch RA, Huisman TA, Wisser J, Gottstein-Aalame N, Debatin JF, Seifert B, Ladd ME, Stallmach T, Marincek B. Ultrafast MR imaging of the fetus. AJR Am J Roentgenol 2000; 174: 1599–1606. 5. Griffiths PD, Porteous M, Mason G, Russell S, Morris J, Fanou M, Reeves MJ. The use of in utero MRI to supplement ultrasound in the foetus at high risk of developmental brain or spine abnormality. Br J Radiol 2012; 85: e1038–e1045.
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
Paladini et al. 6. International Society of Ultrasound in Obstetrics & Gynecology (ISUOG) Education Committee. Sonographic examination of the fetal central nervous system: guidelines for performing the ‘basic examination’ and the ‘fetal neurosonogram’. Ultrasound Obstet Gynecol 2007; 29: 109–116. 7. Whitby EH, Paley MNJ, Sprigg A, Rutter S, Davies NP, Wilkinson ID; Griffiths PD. Comparison of ultrasound and magnetic resonance imaging in 100 singleton pregnancies with suspected brain abnormalities. BJOG 2004; 111: 784–792. 8. Levine D, Barnes PD, Robertson RR, Wong G, Mehta TS. Fast MR imaging of fetal central nervous system abnormalities. Radiology 2003; 229: 51–61. 9. Malinger G, Ben Sira L, Lev D, Ben-Aroya Z, Kidron D, Lerman-Sagie T. Fetal brain imaging: a comparison between magnetic resonance imaging and dedicated neurosonography. Ultrasound Obstet Gynecol 2004; 23: 333–340. 10. Malinger G, Lev D, Lerman-Sagie T. Is fetal magnetic resonance imaging superior to neurosonography for detection of brain anomalies? Editorial. Ultrasound Obstet Gynecol 2002; 20: 317–321.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Accuracy of neurosonography and MRI in clinical management of fetuses referred with central nervous system abnormalities D. PALADINI*†, M. QUARANTELLI‡, G. SGLAVO*, G. PASTORE*, A. CAVALLARO*, M. R. D’ARMIENTO§, M. SALVATORE‡ and C. NAPPI¶ *Fetal Medicine and Cardiology Unit, Department of Gynecology and Obstetrics, University Federico II of Naples, Naples, Italy; †Fetal Medicine and Surgery Unit, Giannina Gaslini Institute, Genoa, Italy; ‡Biostructure and Bioimaging Institute, National Research Council, Naples, Italy; §Department of Pathology, University Federico II of Naples, Naples, Italy; ¶Department of Gynecology and Obstetrics, University Federico II of Naples, Naples, Italy
K E Y W O R D S: 3D ultrasound; CNS; fetus; magnetic resonance imaging; space-occupying lesion
A B S T R AC T Objective To assess the accuracy of expert neurosonography (two- and three-dimensional NSG) in the characterization of major fetal central nervous system (CNS) anomalies seen at a tertiary referral center and to report the differential clinical usefulness of magnetic resonance imaging (MRI) used as a second-line diagnostic procedure in the same cohort. Methods This was a retrospective analysis of all 773 fetuses with confirmed CNS abnormalities referred to our center between 2005 and 2012. The following variables were analyzed: gestational age at NSG and MRI, NSG and MRI diagnoses, indication for MRI (confirmation of NSG findings; diagnostic doubt; search for possible additional brain anomalies), association with other malformations, diagnostic accuracy of NSG vs MRI (no additional clinical value for either MRI or NSG; additional information with clinical/prognostic significance on MRI relative to NSG; additional information with clinical/prognostic significance on NSG relative to MRI, NSG and MRI concordant but incorrect) and final diagnosis, which was made at autopsy or postnatal MRI/surgery. Results CNS malformations were associated with other anomalies in 372/773 (48.1%) cases and were isolated in the remaining 401 (51.9%) cases. NSG alone was able to establish the diagnosis in 647/773 (83.7%) cases. MRI was performed in 126 (16.3%) cases. The indication for MRI was: confirmation of NSG diagnosis in 59 (46.8%) cases; diagnostic query (in the case of inconclusive or uncertain finding on NSG) in 20 (15.9%) cases; search for possible additional brain anomalies in 47 (37.3%) cases. NSG and MRI were concordant and correct in 109/126
(86.5%) cases. Clinically relevant findings were evident on MRI alone in 10/126 (7.9%) cases (1.3% of the whole population) and on NSG alone in 6/126 (4.8%) cases; in all six of these cases, MRI had been performed at < 24 weeks of gestation. In one case, both NSG and MRI diagnoses were incorrect. The main type of malformation in which MRI played an important diagnostic role was space-occupying lesions, MRI identifying clinically relevant findings in 42.9% (3/7) of these cases. Conclusions (1) In a tertiary referral center with good NSG expertise in the assessment of fetal CNS malformations, MRI is likely to be of help in a limited proportion of cases; (2) MRI is more useful after 24 weeks of gestation; (3) the lesions whose diagnosis is most likely to benefit from MRI are gross space-occupying lesions. Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
I N T RO D U C T I O N Central nervous system (CNS) malformations, together with congenital heart disease, represent the group of congenital anomalies most difficult to characterize prenatally. For this reason, new imaging modalities are being used increasingly in the study of normal and abnormal development of the fetal brain: in addition to two-dimensional (2D) ultrasound (US), three-dimensional (3D) US and magnetic resonance imaging (MRI) are currently employed in the diagnosis of fetal CNS abnormalities, though with different indications and results1 – 6 . A particular area of controversy is the clinical usefulness of MRI relative to that of or as an adjunct to neurosonography (NSG), the latter being defined as US examination of the fetal brain performed by an experienced sonologist
Correspondence to: Prof. D. Paladini, Viale Teano 61 – 16147 Genoa, Italy (email: [email protected]) Accepted: 22 October 2013
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
ORIGINAL PAPER
MRI in management of fetal CNS malformations using a multiplanar, possibly transvaginal approach, as reported in the recently published guidelines of the International Society of Ultrasound in Obstetrics & Gynecology (ISUOG)6 . In fact, the yield of ‘clinically relevant information’ provided by MRI in comparison with that of 2D or 3D NSG ranges from 7% to 40%1 – 5,7 – 9 . In our center, the management protocol for fetuses with CNS malformations has always been multimodal, with detailed NSG being performed in all fetuses with suspected brain abnormalities that are referred to our unit; MRI is only performed as a second-line diagnostic procedure in selected cases. In this retrospective analysis, we describe our experience in the clinical management of 773 fetuses with suspected brain anomalies referred to our unit for diagnosis and management over an 8-year period (January 2005–December 2012). The objectives of this study were: to assess the accuracy of expert NSG (2D and 3D US) in the characterization of major fetal CNS anomalies; and to report the differential clinical usefulness of MRI as a second-line diagnostic procedure in the same cohort.
METHODS During the 8-year study period from 2005 to 2012, there were 834 cases of suspected fetal CNS abnormalities referred to our unit for final diagnosis and management. Of these, 61 were excluded from this analysis because they were lost to follow-up or because an autopsy was not available, leaving the sample population of 773 fetuses with CNS anomalies. If no brain anomaly was detected we did not take into account MRI performed for other indications, such as in monochorionic twins undergoing laser coagulation of placental anastomoses for twin-to-twin transfusion syndrome, or in monochorionic or dichorionic twin survivors after cord occlusion or spontaneous intrauterine demise. For all cases, the following variables were retrieved from the computerized database and analyzed: gestational age at NSG and MRI, US diagnosis, indication for MRI, MRI diagnosis, associated anomalies and final diagnosis. Malformations occurring in the study population are summarized in Table 1. In fetuses with multiple brain malformations, classification was according to organogenesis, each case being defined by the malformation that would have developed first. For example, fetuses with holoprosencephaly and agenesis of the corpus callosum were considered as having holoprosencephaly; those with anencephaly and holoprosencephaly were considered as having anencephaly; those with Dandy–Walker malformation and agenesis of the corpus callosum were considered to have Dandy–Walker malformation. In addition, to facilitate the analysis, malformations were categorized into six groups: 1) anomalies of the corpus callosum and cavum septi pellucidi; 2) anomalies of the posterior fossa; 3) primary ventriculomegaly (including mild (borderline) ventriculomegaly); 4) intracranial hemorrhage; 5) space-occupying lesions (arachnoid cysts, tumors,
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
189
arteriovenous malformations); 6) others (neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and other anomalies). The first step in our diagnostic protocol for fetal CNS anomalies includes 2D and 3D NSG. Only if the expert neurosonologist deems it necessary or advisable to perform MRI is this proposed to the parents and, if they consent, this is scheduled as a second-level test. Expert 2D and 3D NSG was performed using a transvaginal approach in all cases of vertex presentation, whether this was spontaneous or after external version; only in the case of failed version and persistent breech presentation was the scan performed transabdominally. High-end 3D equipment (GE E8 and Voluson 730 Expert US machines (GE Medical Systems, Zipf, Austria), equipped with 5–9-MHz volumetric transvaginal transducers and a 4–8-MHz volumetric convex transducer, and the former with a 6–12-MHz volumetric transvaginal transducer) was used for all examinations. 3D volumes were acquired in more than 95% of cases and analyzed offline with dedicated software (4D View, GE Medical Systems). The examination protocol followed that described in the ISUOG guidelines for US assessment of the fetal brain6 . All examinations were performed by a single experienced operator (D.P.). In cases which did not undergo termination of pregnancy (TOP), follow-up scans were performed at 3–4-week intervals. When comparing the diagnostic accuracy of NSG with that of MRI, only findings detected prior to MRI were taken into account. Fetal MRI studies were performed with a Philips Achieva 1.5 T system (amplitude on axis, 30 mT/m; slew rate on axis, 120 T/m/s gradients, equipped with a four-channel phased-array body coil; Philips Healthcare, Andover, MA, USA), using Single Shot Fast Spin Echo sequences to produce high-resolution T2-weighted images (echo time (TE), 120 ms; echo train length (ETL), 96; slice thickness, 3–4 mm), fast-gradient-echo (GrE) sequences to obtain axial T1-weighted images (repetition time (TR)/TE, 154/4.6 ms; flip angle (FA), 80∘ ; slice thickness, 5 mm), and echo planar imaging (EPI) for diffusion-weighted imaging (DWI) (TR/TE, 2000/90 ms; Max B-factor, 600; slice thickness, 5 mm; with calculation of trace and apparent diffusion coefficient (ADC) maps). All sequences were acquired without breath-hold. In cases of suspected hemorrhagic lesions, besides T1-weighted sequences, the B0 image of the EPI-DWI sequence was also used. Neither maternal sedation nor fetal curarization was used during the examination. The expert neuroradiologist (M.Q.) who performed MRI was aware of the NSG findings and of any specific queries resulting from the neurosonographic examination (see below). The indications for MRI fell into three categories: 1) confirmation of NSG diagnosis; 2) diagnostic query (in case of inconclusive or uncertain findings on NSG); 3) search for possible additional brain anomalies, if the final diagnosis at NSG implied the possibility of additional abnormalities not always diagnosable by US (e.g. heterotopia, cortical lesions). In general, MRI was carried out at
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
190
Table 1 Categorization of 773 malformations of the central nervous system diagnosed at our referral center in the period 2005–2012, according to association with other congenital anomalies and termination of pregnancy (TOP) Type of malformation Holoprosencephaly Neural tube defect Anencephaly Cephalocele Open spina bifida Closed spina bifida Posterior fossa malformation Dandy–Walker malformation Inferior vermian hypoplasia Blake’s pouch cyst Megacisterna magna Other Corpus callosal dysraphism Ventriculomegaly Cavum septi pellucidi abnormality Intracranial hemorrhage Schizencephaly Arteriovenous malformation Hydranencephaly Microcephaly Neuronal migration disorder Tumor Arachnoid cyst Craniosynostosis Infection Cytomegalovirus Toxoplasmosis Diastematomyelia Total
Isolated
Associated
TOP
17 (33.3) 118 (73.7) 31 12 72 3 73 (43.9) 14 12 21 14 12† 37 (35.9) 80 (49.1) 8 (80.0) 25 (100) 1 (14.2) 6 (85.8) 5 (83.3) 6 (35.3) 2 (12.5) 8 (100) 4 (33.3) 6 (54.5)
34 (66.7) 42 (26.3) 4 16 22 — 93 (56.1) 38 35 13 7 — 66 (64.1) 83 (50.9) 2 (20.0) — 6 (85.8) 1 (14.2) 1 (16.7) 11 (64.7) 14 (87.5) — 8 (66.7) 5 (45.5)
50 (98)* 150 (94) 33 (94)* 27 (96) 89 (95) 1 (33) 115 (69) 50 (96) 45 (96) 14 (41) 6 (29)
1 (25.0) — 4 (80.0) 401 (51.9)
3 (75.0) 2 (100) 1 (20.0) 372 (48.1)
3 (75) — 4 (80) 560 (72.4)
97 (94) 129 (79) 9 (90) 8 (32) 7 (100) 2 (29) 5 (83) 9 (53) 15 (94)‡ 1 (12) 1 (8) 5 (45)
Data are given as n (%). *Includes twin pregnancy. †Chiari I, hemorrhage, asymmetrical hemispheres. ‡Note this group does not include cases associated with corpus callosal agenesis and other major brain anomalies which have been included in other categories.
28–30 gestational weeks for the latter indication, whereas it was performed within 1 week of the initial NSG diagnosis for the other two indications. After initial MRI at < 24 weeks, follow-up MRI was performed at 28–30 weeks if necessary. With respect to the final diagnosis, made at autopsy or postnatal MRI or surgery, the following diagnostic categories were defined: 1. No additional value of MRI or NSG 2. NSG–MRI discordance; NSG correct, i.e. additional value of NSG 3. NSG–MRI discordance; MRI correct, i.e. additional value of MRI 4. NSG and MRI concordant but incorrect The first category included both cases in which MRI confirmed the NSG diagnosis, and cases in which additional but clinically irrelevant findings were evident on either modality. Additional information provided by NSG or MRI in a particular case was considered clinically relevant if it would affect the diagnosis and, consequently, change the prognosis significantly, according to current knowledge. Over the years, alobar/semilobar holoprosencephaly and exencephaly/anencephaly were diagnosed in a significant number of fetuses at 11–14 weeks. In
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
these cases, an autopsy was not performed, but we assumed that the NSG diagnosis was confirmative per se and, consequently, these cases were included in the analysis. It should also be borne in mind that, although in some cases (e.g. agenesis of the corpus callosum), MRI was recommended routinely, this may have been declined by the parents if it would not modify the clinical management: for example, if the diagnosis was made before 24 weeks’ gestation and the amount of information provided on the basis of the NSG findings was considered sufficient to opt for termination of pregnancy.
R E S U LT S Among the study population of 773 fetuses with CNS anomalies, these were isolated in 401 (51.9%) cases and there were other associated anomalies in 372 (48.1%) cases (Table 1). The mean gestational age at NSG was 21 (range, 13–38) weeks for the whole group, 20 (range, 13–38) weeks for cases undergoing NSG only and 24 (range, 17–36) weeks for cases undergoing both NSG and MRI. NSG alone was able to establish the diagnosis in 647/773 (83.7%) cases. This group included 134 cases in which an MRI was requested but not performed, either because the patient declined the exam (n = 127; opted
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations directly for TOP) or because claustrophobia necessitated termination of the exam before it was complete (n = 7). In this group, the primary diagnosis was correct in all cases, but in 45/647 (7%) cases autopsy revealed additional brain malformations: mainly subcortical/subependymal nodular heterotopia, gyration abnormalities and hemorrhage, found in cases of agenesis of the corpus callosum, Dandy–Walker malformation and ventriculomegaly. MRI was performed in 126 (16.3%) cases (Table 2). In this group, the mean gestational age at NSG was 24 (range, 17–36) weeks, with 73/126 (57.9%) cases being evaluated for the first time before 24 weeks. The mean gestational age at MRI was 27 (range, 21–36) weeks, with two-thirds (85/126, 67.5%) of the cases being evaluated after 24 weeks. The indication for MRI was: confirmation of the NSG diagnosis in 59 (46.8%) cases; diagnostic query (when findings on NSG were inconclusive or uncertain) in 20 (15.9%) cases; search for possible additional brain anomalies in 47 (37.3%) cases. In general, MRI was more likely to be performed for midline anomalies, including those of the corpus callosum, cavum septi pellucidi and posterior fossa, for ventriculomegaly and for space-occupying lesions than it was for open neural tube defects and ventral induction defects (Table 2). With respect to the final diagnoses, NSG and MRI were concordant and correct in 109/126 (86.5%) cases. Additional clinically relevant findings were evident on MRI in 10 (7.9%) cases (Table 3) and on NSG in six (4.8%) cases (Table 4). In one case, both NSG and MRI were incorrect. This was a case of atypical frontoethmoidal cephalocele in which both NSG and MRI identified a solid tumor-like lesion rather than the neural tube defect (Figure 1). In fact, the lesion could be demonstrated adequately on retrospective analysis of MRI slices, and similarly it could be suspected on retrospective analysis of 3D-US volumes; however, we considered this to be an incorrect diagnosis for both techniques, although it was the physician who misinterpreted the images at the time of diagnosis. The same applies to the case of lissencephaly missed prospectively on NSG and recognized later on retrospective review of the case. Assessing final diagnosis with respect to the indication for MRI (Table 5 and Figure 2) suggested that MRI had a very limited contribution to the final diagnosis if it had been requested as confirmation of the NSG finding or to look for additional subtle abnormalities that are more evident at MRI than at NSG; it seemed more helpful in cases of diagnostic doubt, although in this category there was also the highest rate of incorrect diagnoses for MRI. Analyzing final diagnosis according to malformation category (Table 6 and Figure 3) indicated that the only category in which MRI played an important diagnostic role was space-occupying lesions, followed by ‘other’ and ‘posterior fossa’ categories. A detailed analysis of the cases in which NSG or MRI was misleading, missing the final diagnosis, revealed that the diagnosis had been performed in the second trimester (Table 4). Excluding cases undergoing MRI < 24 weeks
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
191 Table 2 Categorization of 126 fetuses with central nervous system malformations undergoing magnetic resonance imaging (MRI), according to type of anomaly among the 773 cases assessed in the study period Type of anomaly Holoprosencephaly Alobar Semilobar Lobar Middle interhemispheric Neural tube defect Anencephaly Cephalocele Open spina bifida Closed spina bifida Posterior fossa malformation Dandy–Walker malformation Inferior vermian hypoplasia Blake’s pouch cyst Megacisterna magna Other Corpus callosal dysraphism Agenesis Hypoplasia Thick corpus callosum Ventriculomegaly Mild (borderline) Severe Cavum septi pellucidi abnormality Intracranial hemorrhage Schizencephaly Arteriovenous malformation Hydranencephaly Microcephaly Neuronal migration disorder Tumors Arachnoid cyst Craniosynostosis Infection Cytomegalovirus Toxoplasmosis Diastematomyelia Total
US
MRI
51 32 15 3 1 160 35 28 94 3 166 52 47 34 21 12 103 95 5 3 163 44 119 10 25 7 7 6 17 16 8 12 11 6 4 (8) 2 (14) 5 773
2 (3.9) — — 1 1 2 (1.2) — — — 2 31 (18.7) 2 6 10 4 9 16 (15.5) 12 4 — 41 (25.1) 30 11 7 (70) 7 (28) 1 — — — 5 3 5 2 3 2 1 1 126 (16.3)
Data are given as n or n (%).
left only one case in which NSG performed better than did MRI (Table 4, Case 2), while MRI added clinically relevant information in 7/85 (8.2%) cases.
DISCUSSION In a population referred to a tertiary center for diagnosis and management of CNS abnormalities, it was deemed worthwhile to request MRI as a second-level diagnostic test in 260 of 773 (33.6%) cases, though eventually it was performed in only 16.3% (126/773). Excluding the seven cases in which MRI had to be interrupted due to maternal claustrophobia, patients often declined undergoing MRI because, in fetuses with multiple malformations, which accounted for half of the series (48.1%; 372/773), as well as in those with severe albeit isolated brain malformations, the prognosis was felt to be so poor that the decision to opt
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
192
Table 3 Characteristics of the 10 cases with central nervous system malformations in which magnetic resonance imaging (MRI) provided clinically relevant information additional to that obtained by NSG
Case
Indication
1 2 3 4 5 6 7 8 9
Confirmation* Confirmation* Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Diagnostic query† Search addtn. anom.‡
10
Search addtn. anom.‡
GA at MRI (weeks)
NSG diagnosis
MRI (and final) diagnosis
23 31 32 22 30 28 22 23 30
Chiari I malformation Pericallosal lipoma Arachnoid cyst Arachnoid cyst or schizencephaly Ventriculomegaly Ventriculomegaly Megacisterna magna or arachnoid cyst Mildly dysmorphic cerebellum Triventricular hydrocephalus
32
Ventriculomegaly, subependymal heterotopia?
Chiari I + craniosynostosis Pericallosal lipoma + lissencephaly§ Arachnoid cyst + pACC + delayed myelination Arachnoid cyst + closed lip schizencephaly Ventriculomegaly + abn. sulcation¶ Ventriculomegaly + abn. sulcation¶ Arachnoid cyst Cerebellar hemorrhagic necrosis Triventricular hydrocephalus + tumor in cisterna ambiens (pinealoblastoma) Ventriculomegaly + severe edema
*Confirmation of NSG diagnosis. †Diagnostic query in case of inconclusive or uncertain finding on NSG. ‡Search for additional brain anomalies. §Evident on retrospective evaluation of NSG volumes. ¶Two full siblings. Abn., abnormal; GA, gestational age at MRI; pACC, partial agenesis of the corpus callosum. Table 4 Characteristics of the six cases with central nervous system malformations in which neurosonography (NSG) provided clinically relevant information additional to that obtained by magnetic resonance imaging (MRI)
Case
Indication
1 2 3 4 5 6
Confirmation† Confirmation† Confirmation† Diagnostic query‡ Diagnostic query‡ Diagnostic query‡
GA at MRI* (weeks) 22 23 21 22 22 21
NSG (and final) diagnosis
MRI diagnosis
Subdural diffuse hemorrhage§ Partial agenesis of CSP Cerebellar hypoplasia Lobar holoprosencephaly Lissencephaly?** Inferior vermian hypoplasia
No hemorrage, ample subdural spaces Normal CSP¶ Cerebellar diameters at 5th centile pACC Normal†† Blake’s pouch cyst
*MRI was performed within 1 week following neurosonography, unless indicated. †Confirmation of NSG diagnosis. ‡Diagnostic query in case of inconclusive or uncertain finding on NSG. §Patient with renal hemorrhage subsequently diagnosed with fetomaternal alloimmune thrombocytopenia. ¶MRI performed at 28 gestational weeks. **Very difficult examination due to premature ossification of the cranial sutures. ††Diffuse atrophy + lissencephaly of frontal lobes on follow-up MRI at 29 weeks. CSP, cavum septi pellucidi; GA, gestational age; pACC, partial agenesis of the corpus callosum.
for TOP was taken before considering the possible clinical relevance of MRI. From this point onwards, all figures and percentages discussed refer only to the 126 cases undergoing both NSG and MRI; and, in order to appreciate the role of MRI in the diagnosis of fetal CNS malformations, the reader should bear in mind that this discussion focuses only on 16.3% of the whole series of 773 cases undergoing expert evaluation at the referral center. With respect to the diagnostic performance of NSG and MRI, both proved highly accurate overall (91.3% (115/126) for NSG and 94.4% (119/126) for MRI, with the accuracy increasing if both were employed because the incorrect diagnoses at NSG differed from those at MRI (Tables 3 and 4). However, before 24 gestational weeks (73 cases), the clinical relevance of MRI in comparison with NSG was lower (93.1% accuracy for NSG and 90.4% for MRI). Overall, MRI provided clinically relevant information missed by NSG in 7.9% (10/126) of cases (Table 3), while the reverse occurred in six cases (4.8%, Table 4). These data compare favorably with only some of those reported in the literature. Malinger et al.9 reported additional clinically relevant information in 7.1% (3/42) of cases undergoing MRI over a period of
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
2 years. However, in their center, there are > 400 referrals per year (pers. comm.). It can be considered, therefore, that in tertiary referral centers where expert NSG is available, the percentage of fetuses with suspected brain abnormalities in which MRI is likely to be helpful is in the range of 5–10%, which agrees well with the 7.9% of our study. Altogether these data (ours and those of Malinger et al.9 ) are rather different from those reported in other papers, in which either MRI was used as a back-up in the absence of an expert neurosonographer8 or it was employed to make some diagnoses that could probably have been made with NSG alone7 . In these papers, the apparent clinical usefulness of MRI was 31%8 and 48%7 , respectively. However, we share the opinion of Malinger et al. that expert NSG is capable of characterizing most CNS abnormalities and that MRI is needed in selected cases with specific indications or queries10 . With this multimodal approach, in our study, MRI added clinically relevant information in 7.9% of cases undergoing expert NSG and MRI, which, depending upon the sonologist’s experience and the spectrum of CNS anomalies, may represent 10–20% of all fetuses with CNS malformations referred to a tertiary center.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations
193
Figure 1 In this case with a final diagnosis after delivery of fetal anterior frontoethmoidal cephalocele (indicated by arrow in all figure parts), both neurosonography at 23 gestational weeks and magnetic resonance imaging (MRI) at 28 weeks incorrectly diagnosed frontoethmoidal or orbital tumor. (a,b) Two-dimensional ultrasound (US): (a) axial view of the orbits; (b) profile modified view. (c) Three-dimensional US: surface-rendering of the fetal face. (d) MR T2-weighted image: profile modified plane. (e) Neonate at delivery, with cephalocele indicated (arrow). On retrospective evaluation, the lesion could be diagnosed correctly on MRI and could perhaps be suspected on US. Table 5 Indication for magnetic resonance imaging (MRI) according to diagnostic category relative to final diagnosis in 125 fetuses* with central nervous system malformations undergoing neurosonography (NSG) and MRI Final diagnosis¶ Indication for MRI
No add. value of MRI
Confirmation† Diagnostic query‡ Search addtn. anom.§ Total
54 (91.5) 11 (57.9) 44 (93.6) 109 (87.2)
Add. value of NSG 3 (5.1) 3 (15.8) 0 6 (4.8)
Add. value of MRI 2 (3.4) 5 (26.3) 3 (6.4) 10 (8.0)
Total 59 (47.2) 19 (15.2) 47 (37.6) 125 (100.0)
Data are given as n (%). *In one of the 126 cases (shown in Figure 1), interpretation of both NSG and MRI was incorrect. †Confirmation of NSG diagnosis. ‡Diagnostic query in case of inconclusive or uncertain finding on NSG. §Search for additional brain anomalies. ¶Final diagnoses: No additional value of MRI; NSG–MRI discordant, NSG correct (i.e. additional value of NSG); NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Add., additional.
Considering the results according to indication (Table 5, Figure 2), MRI proved most useful in comparison to NSG in the 19 cases referred for ‘diagnostic doubt’, with 26.3% (5/19) of these cases benefiting from the addition of MRI. It was in this same subset of cases, however, that NSG proved most useful in comparison to MRI, there
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
being 15.8% (3/19) of these cases in which NSG was more informative than MRI. It is noteworthy that the clinical usefulness of MRI was rather low both in cases referred for ‘confirmation of NSG diagnosis’ (2/59; 3.4%) and in those referred to ‘search for possible additional brain anomalies’ (3/47; 6.4%). The explanations for these
Ultrasound Obstet Gynecol 2014; 44: 188–196.
Paladini et al.
194
60
40
50 Number of cases
Number of cases
30
40 30 20
20
10
10 0
Confirmation*
Diagnostic query†
Search addtn. anom.‡
0
CC & CSP
Indication for MRI
Figure 2 Distribution of the 125 cases undergoing expert neurosonography (NSG) and MRI according to indication for MRI (excluding the wrong diagnosis shown in Figure 1, in which both techniques failed to demonstrate the cephalocele). Diagnostic categories relative to final diagnosis are indicated: , no additional value of MRI; , NSG–MRI discordant, NSG correct (i.e. additional value of NSG); , NSG–MRI discordant, MRI correct (i.e. additional value of MRI). *Confirmation of NSG diagnosis. †Diagnostic query in case of inconclusive or uncertain finding on NSG. ‡Search for additional brain anomalies.
two results are quite different, however. With respect to confirmation of NSG diagnosis, it is likely that the higher resolution of NSG made the difference, particularly as in this category about 40% of the cases were assessed with both techniques before 24 weeks. With respect to the search for additional anomalies, the failure to demonstrate usefulness of MRI relative to NSG was likely due to the small number of cases assessed (n = 47), such subtle, additional anomalies as heterotopia or cortical lesions being quite rare. Regarding assessment of the results according to type of lesion (Table 6 and Figure 3), it seems that MRI makes a considerable difference in those cases in which large
PF
Ventriculo. ICH
SOL
Other
Malformation category
Figure 3 Distribution of the 125 cases undergoing expert neurosonography (NSG) and MRI according to malformation category (excluding the wrong diagnosis shown in Figure 1, in which both techniques failed to demonstrate the cephalocele). Diagnostic categories relative to final diagnosis are indicated: , no additional value of MRI; , NSG–MRI discordant, NSG correct (i.e. additional value of NSG); , NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Malformations were categorized as follows: corpus callosum and cavum septi pellucidi anomalies (CC & CSP); posterior fossa anomalies (PF); primary ventriculomegaly (including mild (borderline) ventriculomegaly) (Ventriculo.); intracranial hemorrhage (ICH); space-occupying lesions (arachnoid cysts, tumors, arteriovenous malformations) (SOL); and other anomalies (neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and others) (Other).
lesions lead to significant distortion of brain anatomy: in 3/7 (42.9%) cases in this subgroup, MRI provided additional information that NSG had missed. Among more conventional pathologies, the posterior fossa seems the next most likely to benefit from MRI, as long as it is performed after 24 weeks. Finally, examining the cases with discordant NSG and MRI diagnoses, in the 10 cases in which MRI provided
Table 6 Malformation category according to diagnostic category relative to final diagnosis in 125 fetuses* with central nervous system malformations undergoing neurosonography (NSG) and magnetic resonance imaging (MRI) Final diagnosis¶ Malformation category
No add. value of MRI
CC and CSP anomalies Posterior fossa anomalies Primary ventriculomegaly† Intracranial hemorrhage Space-occupying lesions‡ Other malformations§ Total
22 (95.7) 26 (83.9) 39 (95.1) 6 (85.7) 4 (57.1) 12 (75.0) 109 (87.2)
Add. value of NSG 1 (4.3) 2 (6.5) 0 1 (14.3) 0 2 (12.5) 6 (4.8)
Add. value of MRI 0 3 (9.7) 2 (4.9) 0 3 (42.9) 2 (12.5) 10 (8.0)
Total 23 (18.4) 31 (24.8) 41 (32.8) 7 (5.6) 7 (5.6) 16 (12.8) 125 (100.0)
*In one of the 126 cases (shown in Figure 1), interpretation of both NSG and MRI was incorrect. †Including mild (borderline) ventriculomegaly. ‡Arachnoid cysts, tumors, arteriovenous malformations. §Neuronal migration and proliferation anomalies, holoprosencephaly, neural tube defects, craniosynostoses, spinal and other anomalies. ¶Final diagnoses: No additional value of MRI; NSG–MRI discordant, NSG correct (i.e. additional value of NSG); NSG–MRI discordant, MRI correct (i.e. additional value of MRI). Add., additional; CC, corpus callosum; CSP, cavum septi pellucidi.
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
MRI in management of fetal CNS malformations
195
Figure 4 Imaging in the case that was false-negative at three-dimensional (3D) neurosonography. In this 31-week fetus, the following anomalies were detected on ultrasound: pre-auricular tags (a,d, arrowhead); unilateral cleft lip (b,e, arrow), pericallosal lipoma (c, arrow). Magnetic resonance imaging, performed 2 days later, showed gyral abnormalities consistent with lissencephaly (f, arrows). On retrospective review of 3D volume datasets, complete absence of the cingulate gyrus and a complete smooth cortical surface were noted (c, arrowhead), a finding overlooked at initial examination, probably due to the presence of multiple anomalies.
a clinically relevant additional or discordant piece of information, the 3D volume datasets were reevaluated retrospectively in an attempt to correctly identify the missed or incorrect NSG diagnoses. Only one such case was found: a fetus with multiple malformations assessed at 31 weeks of gestation. The fetus had a dysmorphic face, with pre-auricular tags, wrinkled ears, unilateral cleft lip, a pericallosal lipoma and lissencephaly (Figure 4). The latter finding had been missed initially at NSG; however, on retrospective reassessment of the volume the gyration abnormality was evident. In this case, the likely explanation is that the attention of the operator was diverted by the lipoma and the other non-CNS malformations. Considering, on the other hand, the six cases in which MRI proved incorrect (Table 4), it is evident that they underwent MRI before 24 gestational weeks; evidently, MRI before 24–25 weeks should be interpreted with caution. This study had some limitations. First, despite the fact that MRI is known to have relatively low accuracy when performed at mid-gestation, almost one third (41/126; 32.5%) of cases were evaluated before 24 weeks, due to the Italian law for TOP. This may have led
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
to an underestimation of the clinical usefulness of MRI. Second, regarding the analysis by subgroup, in some cases the numbers were so small as to limit significantly the relevance of any finding; for example, in the breakdown according to type of malformation (Table 6 and Figure 3). We conclude that: (1) in a tertiary referral center with expertise in the US assessment of fetal CNS, MRI is likely to be of help in a limited proportion of cases, being performed in 16% of the whole fetal population with CNS anomalies and proving clinically useful in 7.9% of this subgroup (1.3% of the whole population); (2) it is probably worth restricting the use of MRI, performing it only when: a) requested due to diagnostic doubt following NSG, b) only after 24 weeks of gestation, and c) mainly in the investigation of gross space-occupying lesions, rare anomalies and, to a lesser extent, posterior fossa and corpus callosal anomalies. For other indications (confirmation of NSG findings and search for additional subtle abnormalities) and pathologies (e.g. moderate ventriculomegaly, holoprosencephaly, open neural tube defects) the use of MRI should be weighed against the availability of economic and human resources.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
196
REF EREN CES 1. Simon EM, Goldstein RB, Coakley FV, Filly RA, Broderick KC, Musci TJ, Barkovich AJ. Fast MR imaging of fetal CNS anomalies in utero. Am J Neuroradiol 2000; 21: 1688–1698. 2. Peruzzi P, Corbitt RJ, Raffel C. Magnetic resonance imaging versus ultrasonography for the in utero evaluation of central nervous system anomalies. J Neurosurg Ped 2010; 6: 340–345. 3. Messing-Jünger MM, Röhrig A, Stressig R, Schaper J, Turowski B, Blondin D. Fetal MRI of the central nervous system: clinical relevance. Childs Nerv Syst 2009; 25: 165–171. 4. Kubik-Huch RA, Huisman TA, Wisser J, Gottstein-Aalame N, Debatin JF, Seifert B, Ladd ME, Stallmach T, Marincek B. Ultrafast MR imaging of the fetus. AJR Am J Roentgenol 2000; 174: 1599–1606. 5. Griffiths PD, Porteous M, Mason G, Russell S, Morris J, Fanou M, Reeves MJ. The use of in utero MRI to supplement ultrasound in the foetus at high risk of developmental brain or spine abnormality. Br J Radiol 2012; 85: e1038–e1045.
Copyright © 2013 ISUOG. Published by John Wiley & Sons Ltd.
Paladini et al. 6. International Society of Ultrasound in Obstetrics & Gynecology (ISUOG) Education Committee. Sonographic examination of the fetal central nervous system: guidelines for performing the ‘basic examination’ and the ‘fetal neurosonogram’. Ultrasound Obstet Gynecol 2007; 29: 109–116. 7. Whitby EH, Paley MNJ, Sprigg A, Rutter S, Davies NP, Wilkinson ID; Griffiths PD. Comparison of ultrasound and magnetic resonance imaging in 100 singleton pregnancies with suspected brain abnormalities. BJOG 2004; 111: 784–792. 8. Levine D, Barnes PD, Robertson RR, Wong G, Mehta TS. Fast MR imaging of fetal central nervous system abnormalities. Radiology 2003; 229: 51–61. 9. Malinger G, Ben Sira L, Lev D, Ben-Aroya Z, Kidron D, Lerman-Sagie T. Fetal brain imaging: a comparison between magnetic resonance imaging and dedicated neurosonography. Ultrasound Obstet Gynecol 2004; 23: 333–340. 10. Malinger G, Lev D, Lerman-Sagie T. Is fetal magnetic resonance imaging superior to neurosonography for detection of brain anomalies? Editorial. Ultrasound Obstet Gynecol 2002; 20: 317–321.
Ultrasound Obstet Gynecol 2014; 44: 188–196.
