Prenatal MR Imaging Phenotype of Fetuses with Tuberous Sclerosis: An Institutional Case Series and Literature Review

DISCUSSION

The current review, including 4 cases from our own institution, establishes that a cardiac tumor was present at the time of iuMR in all except 3 of 51 cases of proved TSC, and in 2 of these 3 cases, the fetus’s father had known TSC. This finding highlights the importance of a careful search for CR in fetuses with cranial abnormalities suggestive of TSC because CRs substantially increase the pretest probability of TSC and their absence should raise questions about the diagnosis, especially in a fetus with no family history.

Similarly, when a unilateral foramen of Monro nodule or mass, essentially synonymous with unilateral GE enlargement, is seen as an isolated abnormality, or with only ventriculomegaly, TSC should be considered if a CR is present. Unilateral germinal matrix hemorrhage can mimic this appearance; consequently, routine performance of susceptibility-weighted imaging and fetal echocardiography can help avoid overdiagnosis of TSC, with resulting inappropriate prognostic and recurrence counseling in this situation.

The incidence of CR in patients with TSC is 50%–80%,²² with a recent study suggesting that only 22% of a cohort of patients with TSC had CR diagnosed prenatally. Moreover, CRs can appear late in pregnancy, most often after the midtrimester screening examination is performed.²³ Thus, it is possible that a substantial number of fetuses with TSC will not have this diagnosis suggested prenatally by the presence of CR. While intracranial manifestations of TSC on iuMR might, therefore, be the first indication of the disorder due to iuMR being more sensitive and specific than prenatal US for intracranial pathology,²⁴ this was not borne out by our literature review. This may, however, reflect selection bias in that reported cases of TSC on iuMR have been referred for MR imaging due to the identification of CR either at the midtrimester screening examination or in the third trimester as an “incidental” finding when growth, placental position, or other surveillance has occurred.

We have identified that among fetuses with proved TSC who have had iuMR, the likelihood of abnormal cranial findings varies significantly with GA. Fetuses who have abnormal findings were, on average, at 30 weeks’ gestation compared with 25 weeks for fetuses with no abnormality. Interestingly, most fetuses with TSC, when the diagnosis was made at 24 weeks or earlier in gestation, demonstrated a dominant hemispheric mass, whereas only 1 fetus (11% of this early-gestation cohort) had subependymal nodules as their sole imaging manifestation. This finding makes intuitive sense because more severe “masslike” lesions would be more likely to be detected at the midtrimester screening US examination, typically performed between 19 and 21 weeks’ gestation, and would manifest as a midline shift. Such masslike presentations at screening US can be misinterpreted as hemorrhage and underscore the importance of careful tertiary echocardiography and T2*-weighted iuMR for further fetal evaluation.

While association of cranial abnormalities with increasing gestation might suggest that it is better to wait to perform iuMR, 9/42 (23%) fetuses with abnormal cranial MR imaging findings were 19–24 weeks’ gestation when MR imaging was performed. This timing is potentially important in jurisdictions where GA-related limitations exist on pregnancy termination. It may also be salient to the commencement of intrauterine therapy with rapamycin. When genomic testing is either unavailable or not turned around quickly enough to be clinically useful regarding counseling and therapeutic decision making, earlier diagnosis may also be important. However, “early” (<25 gestational weeks) MR imaging may be associated with a higher rate of negative findings; thus, it may be a more costly strategy if a second iuMR is performed later in pregnancy after initial negative examination.

Improvements in ultrasound equipment during the past decade and in particular the evolution of the tertiary neurosonogram using a transvaginal approach and, when possible, the sonographic window of the anterior fontanelle have dramatically improved the quality of fetal brain imaging with US. In addition, the spatial resolution of US is superior to that of MR imaging, theoretically making subependymal nodule detection easier with US than with MR imaging. However, tertiary neurosonography remains highly operator-, patient-, and fetal position–dependent. Whether such optimized US can detect subependymal nodules in practice in fetuses at a high demographic risk of TSC when these are not demonstrable with iuMR is a question outside of the scope of the current review.

Finally, by aggregating data from multiple case reports and case series, the current study highlights some less common presentations of TSC, in particular hemimegalencephaly and megalencephaly. Because other genetic disorders, in particular other mTOR pathogenic variants, such as PIK3CA-related overgrowth spectrum disorders, are the underlying pathogenic variants in most cases of megalencephaly or hemimegalencephaly and these are almost invariably de novo mutations or, much less often, the result of germline mosaicism, distinguishing TSC-related hemimegalencephaly/megalencephaly from these more common non-TSC causes is very important. It is theorized that because activation of TSC1 and TSC2 is driven by upstream signals of the PIK3/AKT pathway, the loss-of-function pathogenic variants of TSC1 or TSC2 that result in clinical TSC could induce an imbalance in PIK3/AKT pathway functioning, producing cerebral hemispheric overgrowth.¹⁰ Hence, careful examination and re-examination of the fetal heart with tertiary fetal echocardiography are recommended in the presence of hemimegalencephaly/megalencephaly in the fetus. These will help direct genomic testing and prevent inappropriate advice regarding the utility of such testing or the likelihood of recurrence in subsequent pregnancies.

Our study has limitations, mainly due to potential selection bias in published cases driven by a prenatal US diagnosis of CR being the trigger to refer a patient for iuMR. Because it is recognized that CR may not be detectable with US until the third trimester of pregnancy and frequently not until the postnatal period, the cases included in this review are likely to provide a biased impression regarding the following:

• The likelihood of TSC being the cause for fetal cerebral abnormalities when there is no CR or family history (because the pregnancy may be terminated without a genetic diagnosis or postmortem examination)

• How early in gestation the cerebral lesions of TSC can be diagnosed with MR imaging in fetuses with high pretest risk

• How often iuMR findings are negative for cerebral lesions in fetuses with high pretest risk of TSC (because of family history of the condition) and at what gestational age, because cases with negative findings are less likely to be published.

A prospective cohort study of serial MR imaging in fetuses who are the offspring or siblings of individuals with TSC would provide a more accurate estimate of these important issues to help inform clinical practice. However, pre-implantation genetic diagnosis and/or gene panel testing for pathogenic TSC1 and TSC2 variants during early pregnancy may become more feasible, economical, and acceptable to patients than second trimester iuMR screening when the fetus has a high risk of the condition due to an affected sibling or parent. In addition, only a small minority of fetuses reported in this study were imaged using a 3T scanner; the reduced signal-to-noise and thus the resolution of iuMR at 1.5T may reduce the sensitivity for cerebral abnormalities in fetuses with TSC.

A further limitation of the current study is due to variable terminology used to describe iuMR imaging findings and, in particular, the general lack of routine reporting of cranial biometry. Specifically, when iuMR was said have normal findings but it was unclear whether fetal brain biometry had actually been performed, it is possible that megalencephaly with a “normal”-appearing brain may have been underrecognized. Moreover, the range of terms used to describe a mass, nodule, or swelling in the region of the foramen of Monro or GE of a fetus on MR imaging may have made the distinction among a foramen of Monro “mass,” a small hemispheric mass, and subependymal nodule in the region of the foramen of Monro an artificial one. Hence, we may have misclassified abnormalities in this region when relying on descriptive terms used in the report, mainly because their imaging appearances can overlap.