Article Text
PDF
Abstract
Background Torcular dural sinus malformations (tDSMs) are described as slow flow dural arteriovenous fistulae with frequently poor outcomes in the neuroangiographic literature, but other etiologies have been proposed in the obstetric literature, where outcomes have been more favorable.
Objective To review tDSMs reported in the literature of multiple specialties for features that support a common etiology, and to identify key prognostic factors, with an emphasis on tDSM trajectory highlighted in part I.
Methods Analysis of imaging features and clinical outcome for 77 prenatal and 22 postnatal tDSMs reported in 37 papers from the literature.
Results In addition to large venous lakes, 36% of prenatal and 96% of postnatal tDSMs had evidence of arterialization, where specifically assessed. For fetal cases, where there was an observable natural history, 97% underwent a spontaneous decrease—13% after an initial increase and only 1 case with subsequent enlargement after a decrease. Prenatal cases had 83% survival (62% with a favorable outcome) whereas postnatal cases had 59% survival (29% favorable). In addition to a postnatal diagnosis, unfavorable features included ventriculomegaly, parenchymal injury, arterialization, and need for intervention. Favorable features included decreasing tDSM size, presence of clot, and increasing clot percentage.
Conclusions Neonatal and fetal tDSMs have overlapping imaging appearances, suggesting a common etiology, where neonatal tDSMs represent those rare fetal tDSMs that do not undergo spontaneous regression and have a propensity for worse outcomes. Decrease in tDSM size is a critical observation when managing a tDSM because it is generally irreversible and associated with a favorable outcome.
- pediatrics
- fistula
- vascular malformation
- mri
- angiography
Statistics from Altmetric.com
Introduction
Dural sinus malformations (DSMs) are rare vascular malformations that manifest in fetuses and infants, characterized by venous lakes at the involved dural sinuses. In the neuroangiographic and neurosurgical literature, DSMs are seen in association with low flow arteriovenous fistulae (AVFs) and historically have been reported to have a poor outcome. This is particularly so for the more severely affected torcular DSM (tDSM) subgroup, where up to 71% of patients have been reported to have severe neurologic impairment or death.1
A morphologically identical lesion to the tDSM has been reported in the fetal imaging literature, but the inability to directly demonstrate the angioarchitecture of these lesions has led to alternative etiologies being advanced for this population (eg, primary dural sinus dysplasia or thrombosis followed by dural sinus dilation).2–13 Some autopsy papers have also reported extraluminal localization of these fetal tDSMs.14 15 In distinction to the neuroangiographic literature, the outcome for these fetal tDSMs has been favorable in as many as 75–90% of cases.8 10 16
These differences in formulation and outcome cast doubt as to whether tDSMs seen in utero and postnatally represent the same entity. They also raise questions about how best to counsel late gestation pregnancies affected with a tDSM and how best to follow such pregnancies.
In part I of this article,17we examined a local cohort of tDSM cases and found imaging commonalities across the fetal–neonatal age spectrum. For example, we described a fetal tDSM with directly visualized arteriovenous communications in utero, as well as postnatal tDSMs where AVFs were undetectable even with catheter angiography. We also showed that tDSM involution in our cohort was an irreversible process and one that may occur despite initial enlargement.
In part II, we have conducted an analysis of the fragmented multidisciplinary literature on tDSMs to highlight prognostic factors identifiable from other cohorts and assess these in light of the prognosticators in our cohort. We have synthesized these data to offer a plausible model for tDSM evolution across fetal–neonatal life and provided a rational approach to following tDSM pregnancies and performing interventions when necessary.
Methods
Definition of tDSM
As with part I, we focused on tDSMs, defined as an intradural mass (fluid and/or clot) centered at the torcular, regardless of whether it was associated with demonstrable thrombus or arterialized flow (see discussion in part I for further details).17
Literature review
PubMed was queried using the terms ‘dural sinus malformation,’ ‘fetal sinus thrombosis,’ and ‘torcular hematoma.’ Recursive search of the citations from the resulting papers was performed until new papers describing tDSMs were exhausted. Cases described in more than one publication were included for review only once.
This process yielded 37 papers, detailing 99 tDSMs that fitt our criteria, with the majority (n=22) being single case reports (see online supplementary table 1): the number of subjects per paper (and in parentheses, the number of papers with that number of subjects) was 13 (1), 8 (2), 6 (1), 4 (2), 3 (4), 2 (4), and 1 (22). Although there was variation in the type of data available for each literature case, variables of interest were analyzed in all cases where the data were provided. To analyze the natural history of tDSMs, terminated pregnancies and cases with a surgical/endovascular intervention directed at the tDSM were excluded; cases where medical therapies were administered (eg, platelet transfusion) and/or ancillary surgical procedures performed (eg, ventriculostomy placement) were included. When assessing for parenchymal injury, any gliosis/encephalomalacia (supratentorial or infratentorial) or hemorrhage (germinal matrix or in the substance of the parenchyma) was scored as a parenchymal injury. Arterialization was assessed as positive in those cases with direct arteriovenous communication demonstrated by MRI/MR angiography or catheter angiography. Arterialization was also scored as positive for cases where venous reflux was noted (two cases), a carotid bruit was noted (one case), or high flow was observed on Doppler ultrasound (two cases). The evidence for arterialization is indicated in the online supplementary table 1, column U.
Supplementary file 1
Clinical outcome
As clinical outcome was not reported in a consistent manner, we used a simplified three tiered outcome scale for all cases. Our three tiered outcome scale (and the corresponding score from Barbosa et al 1) is:
Favorable (minimal to no neurologic deficit) (Barbosa 3–5)
Unfavorable/uncertain (moderate to severe deficit or unclear outcome, but certainly neither dead nor only minimally injured) (Barbosa 1, 2, or unspecified)
Death (Barbosa 0)
Results
Literature analysis
Despite inconsistent reporting of prognostic variables in prior reports, useful data were gleaned from the literature.
Demographics
Of 99 literature subjects, ranging in age at diagnosis from 18 weeks' gestational age to 4 years postnatal age, 77 patients were diagnosed prenatally and 22 postnatally. Excluding four prenatal subjects for whom age at detection was not stated, the mean age of detection for the other 73 prenatal cases was 26 weeks' GA (±5 weeks). The mean age of detection for the 22 postnatal cases was 69 weeks PMA (±48 weeks). Of the 76 subjects for whom sex was indicated, 54% were female (table 1).
Characteristics of the 99 cases in the literature of torcular dural sinus malformation
Imaging characteristics
Because the original images are not readily available and size variables are inconsistently reported in the literature, we do not have a good estimate for presentation size of the literature tDSMs. Evidence of arterialization was present in 34 of 58 cases where arterialization was specifically evaluated (57%). Thirteen of 36 prenatal cases demonstrated arterialization (36%) whereas 21 of 22 postnatal cases did (96%) (supplementary table 3).
Supplementary file 2
Evolution over time
Non-terminated fetal tDSMs spontaneously involuted in 38 of 39 cases (97%), and 5 tDSMs initially increased in size prior to spontaneous involution (supplementary table 4). Only one case could be found describing a decrease in tDSM size followed by subsequent enlargement.8 Equivalent natural history data on postnatal tDSMs are lacking as virtually all of these subjects in the literature underwent an intervention.
tDSM thrombosis
Of 59 postnatal and non-terminated prenatal tDSM cases, 52 (82%) had thrombus at some point (supplementary table 5). Like the decrease in size of the tDSM itself, the decrease in clot size also appeared generally to be an irreversible process, with 31 of 32 observation only prenatal cases that described clot evolution having involution (9 increased before decreasing, and only 1 increased following an observed decrease in size) (supplementary table 6). Equivalent natural history data in postnatal tDSMs are not available, as only one case without intervention and with thrombus was identified in the literature.
Delivery
Delivery management is often debated in cases of prenatal tDSMs, specifically whether the pressure of a vaginal delivery will have negative impact on the clinical course of the tDSM (eg, hemorrhage or venous infarction).2 In the cases documented in the literature, 13 were delivered vaginally: 3 by induction, 10 spontaneously. Seven of the vaginally delivered cases had complete or near resolution of the tDSM prior to delivery, and two did not comment on the size of the lesion at birth. Of the four remaining tDSMs with vaginal delivery, there was one stillbirth, one death in infancy, and two grade A outcomes. It is difficult to make generalizable statements about vaginal delivery with such limited data, although vaginal delivery itself does not appear directly responsible for morbidity in the four reported cases.
Interventions and treatment
Interventions specifically aimed at managing the tDSM were described in 14 prenatally and 18 postnatally diagnosed cases. The majority of interventions were embolization or surgery (27 total, 11/14 prenatal diagnosis and 16/18 postnatal diagnosis) (supplementary table 9).
Clinical outcomes
Of the 77 cases in the prenatal literature cohort, 17 were electively terminated, leaving 60 prenatally diagnosed subjects with ascertainable postnatal outcomes: 83% of these prenatal diagnoses had short term survival (ie, during the study period) and 62% had a grade A outcome for the reported study duration. By comparison, the postnatal cohort fared significantly worse, with an overall survival of 59%, 29% with grade A outcomes (supplementary table 2). This discordance is unlikely to be wholly explained by the elective terminations removed from the outcome analysis as the stated reason for the 17 terminations were not generally related to the status of the tDSM itself: reasons included cardiac failure (1), progressive ventriculomegaly (1), associated aneuploidy (1), enlargement of the tDSM (1), enlargement of clot (1), interpretation of other pathology such as Dandy–Walker malformation (2), and prognostic uncertainty/unstated (9).
Favorable prognostic factors included absence of evidence for arterialization (23% grade A with arterialization vs 65% without), decrease in tDSM size (79% grade A, including those who initially increased), presence of clot (65% grade A with vs 29% without), documented increase in clot size/percentage (89/83% grade A), absence of ventriculomegaly (40% grade A with ventriculomegaly vs 80% without), and absence of parenchymal injury (40% grade A with injury vs 60% grade A without). These findings are summarized in supplementary tables 3–8. Finally, in contrast with our own experience, in the literature, invasive interventions were associated with poor outcomes, with only 36% of the prenatal and 31% of the postnatal cases with interventions having grade A outcomes (supplementary table 9); these patients were extremely ill (ie, 16 of 18 postnatal intervention patients had congestive failure, seizure, signs of venous hypertension such as dilated scalp veins, or ventriculomegaly). The discordance in outcome of untreated prenatal tDSM diagnoses (67% grade A) and treated prenatal tDSM diagnoses (36%) may be affected by intrinsically worse natural histories for the treated prenatal population, but it is worth noting that only 14 of 58 prenatal diagnoses born live required intervention as opposed to 16 of 22 postnatal diagnoses.
DISCUSSION
tDSM: one or several entities/pathophysiologies?
Blood containing intradural masses at the torcular have been characterized as intradural thrombus in some pathology series,14 15 as a form of isolated in utero dural sinus thrombosis in some of the prenatal literature,2–13 and as low flow dural AVFs (dAVFs) in the neuroradiology/angiographic literature.1 16 18–24 These differences in formulation naturally lead to differences in approach to assessing these lesions (eg, emphasis on arterialization), and also to differences in pathophysiologic explanations (eg, dysplastic sinus or arterialized sinus as the primary event vs spontaneous clot formation leading to sinus expansion). As mentioned earlier, the literature on tDSMs also reaches contradictory conclusions regarding prognosis.
Using a standardized definition for a tDSM, we find that the radiographic appearance of lesions described within each specialty’s literature appear morphologically indistinguishable. Through analysis of our own series in part I and review of the literature in part II, we suggest that many, if not all, of the in utero tDSMs represent ongoing dAVFs or sequelae of prior dAVFs. tDSMs can be viewed as an evolutionary continuum, from frank dAVF → remodeling to tDSM → stagnant flow or spontaneous thrombosis in the tDSM leading to occlusion of the dAVF → tDSM involution and progressive thrombosis, followed by recanalization or re-routing of flow (figure 1).
Proposed model for torcular dural sinus malformation (tDSM) development and spontaneous thrombosis (good outcome) versus persistent arterialization (risk of poor outcome). dAVF, dural arteriovenous fistula.
Evidence supporting this interpretation includes the following. First, we find that fetal tDSMs can have identical angioarchitecture to the originally described postnatal lesions (eg, middle meningeal artery (MMA) to torcular in case No 1 of part I),17 and conversely some postnatal tDSMs can lack detectable arteriovenous communication, as is the case for most fetal cases (case No 12 of part I). Moreover, although normal transverse sinus pulsatility in fetuses can confound interpretation,25 13 prenatally diagnosed tDSMs have been reported to have underlying arterialization, with three demonstrating clearly elevated blood flow on fetal Doppler ultrasound24 26 27 and the rest potentially being too low velocity for Doppler detection. Thus there is overlap in the angioarchitecture as well as in the morphology of the prenatal and postnatal tDSMs. Second, while there has been observation of arterialization preceding torcular remodeling and tDSM dilation preceding thrombosis,16 22 there is no compelling evidence from our cases or from the reviewed literature of clot preceding dural sinus remodeling or preceding dAVF formation (as hypothesized in adults).28 In this regard, it should be noted that there is a normal developmental ballooning of the torcular and proximal transverse sinus in the fourth–fifth month of gestation, then followed by a decrement in size.29 Thus it is unclear whether the torcular venous lake that characterizes a DSM results from lesional fast flow at the particular developmental stage where the posterior sinuses are ballooned, or instead results from subsequent remodeling and enlargement of what had already become a smaller torcular. Third, the better outcome in cases of thrombosed tDSMs is potentially explained by clot tamping the dAVF flow, allowing for healing of the arteriovenous communication. In some instances, the clot appears to incorporate in the tDSM wall or extraluminal space, as indicated by calcification in resolved tDSM cases, explaining reports of extraluminal clot.
Thus there is some evidence to support a common pathophysiologic mechanism behind tDSMs in both prenatal and postnatal life. From this perspective, the ongoing presence of dAVF in 21 of 22 postnatally diagnosed tDSMs in the literature and 2 of 3 postnatal cases in our cohort could represent failure to spontaneously involute in utero. By this logic, the worse outcomes in the postnatal population could be explained as a form of ascertainment bias, where the generally favorable involution of the fetal tDSM does not occur. Also, cases diagnosed in utero are incidental while postnatal diagnoses are generally symptomatic, potentially contributing to different outcomes based on age at diagnosis.
Alternative explanations cannot be entirely excluded—namely, multiple etiologies, primary venous dysplasia, or a genetic etiology.30 In particular, one cannot exclude that some tDSMs have aberrant physiologic coalescence/ballooning of the dural sinuses or an interaction of physiologic dural sinus development with transient dAVFs (ie, a spectrum between dural sinus dysplasia and dAVF). However, the data suggest that a substantial number of tDSMs have underlying arterialization and that underlying dural sinus thrombosis is probably not the inciting event.
Prognostic variables
Assessment of potential prognostic variables in tDSMs is difficult, considering the scattered small case series that comprise the majority of the reported literature, differing durations of follow-up, and the different points of emphasis (and thus reported variables) in prior work. Nonetheless, several important lessons can be drawn.
First, as others have concluded,2 4 6 8 16 22 26 31 we find that ventriculomegaly and parenchymal injury are adverse factors in the outcome of tDSM. Second, while not true in our own cases, the literature suggests that the need for intervention can be a harbinger of bad outcome, as it may reflect clinically significant cardiac failure or parenchymal injury and thus an unfavorable natural history. Although we postulate that all tDSMs may initially have dAVFs as the inciting cause, the data suggest that documented arterialization is cause for concern, possibly because it indicates that the tDSM is not involuting and instead persists in a state of high enough flow for detection on routine imaging. Surprisingly, we find, both in the literature review and in our own series, that clot formation and even clot enlargement can be a favorable factor, possibly because it represents a step towards involution of the tDSM.16 22
One of the most important prognostic variables is the size trajectory of the tDSM. As illustrated in our own case series (part I)17 and also suggested in the literature, decrease in size of a tDSM appears generally to be an irreversible process, with documentation of only a single case where a tDSM initially decreased in size and then grew. On the other hand, initial growth of tDSM is not necessarily a harbinger of poor outcome, as several of our own cases and several cases from the literature transiently grew before involuting. The obvious implication of this observation is that a tDSM encountered in fetal life requires short term follow-up until either the lesion begins to shrink or the lesion evolves in a manner which will clearly have adverse neurodevelopmental consequences (eg, parenchymal hemorrhage or infarct). Relatively large tDSMs may therefore still evolve favorably, if no parenchymal injury has occurred.
Treatment and counseling
Given the relative rarity of prenatal tDSMs, there is no consensus regarding how to counsel families who face this diagnosis, likely explaining the high proportion of cases in which termination has been chosen (ie, the prognostic uncertainty in 9 of 17 terminated pregnancies described earlier). However, based on our literature review and on our own experience, it is manifestly clear that the natural history of prenatal tDSMs is not as dire as has been thought. Optimism is especially warranted in cases where the tDSM has begun involution during fetal life, with no parenchymal injury.
While in the best case, no treatment will be needed after delivery, should signs of venous hypertension persist or develop at birth, our approach is to pursue early endovascular obliteration of sufficient extent to arrest the concerning signs (usually hydrocephalus or parenchymal congestion). In our series, this approach has resulted in grade A outcome in all treated patients; these outcomes are better than reported in prior literature, potentially reflecting progressive improvement in endovascular techniques or our earlier intervention.
Although thrombosis may be part of the involution of the tDSM, anticoagulation can play an important role in some cases, during periods of rapid progression of thrombosis and growth of the tDSM, including cases with abrupt changes in hemodynamics associated with intervention. This concept is well illustrated by case No 10 from part I,17 where embolization was followed by increased clot burden, and careful optimization of anticoagulation contributed to a good patient outcome.
Regarding mode of delivery, there are insufficient data to recommend either vaginal delivery or cesarean section. The vein of Galen literature suggests that vaginal delivery in the face of venous hypertension is not intrinsically unsafe,32 but whether the superficial location of tDSMs warrants greater caution than galenic malformations requires future study. Nonetheless, we have not found a case in which vaginal delivery itself clearly resulted in morbidity.
Conclusion
We have reviewed tDSMs seen at our own center as well as those in the published literature. These data support the hypothesis that the different entities ascribed to tDSMs in the prenatal imaging and neuroangiographic literature share a common pathophysiological mechanism, in many instances beginning with a fetal dAVF and typically ending in utero with spontaneous thrombosis of the tDSM followed by healing of the dural sinus. In addition to confirming adverse outcomes associated with parenchymal injury and hydrocephalus, we have provided an evidence based guide for clinical practice by highlighting the importance of serial imaging to evaluate tDSMs identified in utero, particularly for detecting tDSM involution, which seems to be generally irreversible. We also propose that the worse outcomes reported with postnatally or perinatally diagnosed tDSM, compared with a prenatal diagnosis, may be explained by ascertainment bias related to subjects who have more persistent or otherwise poorly involuting tDSMs, and thus does not generalize to all or most tDSMs diagnosed in utero. Finally, we highlighted important criteria that should be considered when judging the natural history and prognosis of tDSMs, both in routine clinical work as well as in future research: overall tDSM size, clot size, percentage of tDSM occupied by clot, arterialization (including peak velocities and resistive indices by Doppler ultrasound, echocardiography correlation), consistent clinical outcome reporting, parenchymal changes, and hydrocephalus.
References
Footnotes
Contributors EY and AS abstracted literature cases, analyzed the outcomes by patient variables, and prepared the manuscript. DBO conceived of and supervised the study. All authors have contributed to the interpretation of the literature and drafting of the manuscript. All authors have approved the final version.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.