When cardiac rhabdomyomas are detected in a fetus, both the perinatal outcome dominated by tumor space-occupying and the long-term prognosis related to TSC are critical, but tricky. We try to assess perinatal risk and prenatal diagnosis of TSC with CR as the initial symptom through a multidisciplinary approach, which opens up the possibility of relatively reliable assessment for numerous TSC VUS.
The influence of tumor on perinatal outcome
The influence of space-occupying is the determinant of the perinatal outcome. Outflow tract obstruction is not rare occurred in approximately 10% of our subjects. But severe hemodynamics disorders were scattered partly because multiple CRs, which accounted for 70% in this study, tended to be smaller [9] and chamber located [10]. Although 83.7% of tumors increased by the stimulation of maternal estrogen prenatally [11], the increment reduced after the 32nd week [12]. So the masses were at their most significant in the mid/last trimester for which required more rigorous and careful monitoring. The earliest ultrasonic observable time of CR was reported at 15 weeks of GA [13], but the exact time of tumorigenesis is difficult to deduce. Case 6 provided some clues for early events in tumor formation. We conjectured that early-onset CRs might be manifested as diffuse ventricular hyperechoic and could affect the normal remodeling process in the heart. Thus, the diagnosis of TSC in some complex cardiac structural malformation fetuses might be underestimated or ignored. Although the proliferation of CRs seems self-limiting and independent of TSC gene alteration, the tumorigenesis might result in gene instability, such as a loss of heterozygosity or a second variaous [14]. Supporting evidence comes from targeted-deep sequencing in TSC-associated hamartoma samples, roughly 2/3 of hamartomas from TSC individuals harbored two TSC hits [15]. We tested CR tissues in two aborted TSC fetuses, and both the mutations were consistent with their amniotic fluid results (data was not shown), suggested a germline origin of mutations. It reported that the total mutational burden of TSC lesions was low and suggested a low mitotic index [15], which consistent with the slow-growing nature. Anyway, strengthen echocardiography monitoring and reassess tumors in middle pregnancy is indispensable.
Some postpartum study reveals that up to 23% children with CRs may present arrhythmia [16], which may be related to abnormal conduction of electrical excitement in tumor tissues near the atrioventricular junction [17]. Tetsuya et al. reviewed 20 fetuses with CRs and speculated that, regardless of number and location, tumor diameter > 30 mm is associated with postnatal arrhythmia [18]. We observed no arrhythmia in all the subjects including the one who suffered a single huge tumor located in LV exceeded 30 mm.
The influence of tumor-related TSC on long-term prognosis
The influence of CR-related TSC is a decisive factor for long-term prognosis although it is really a hot potato. NGS facilitates the detection process but leave the problem to variants recognition. We amplified the detection yield to 73.8% (including mutations of P, LP and VUS), which was much higher than ten years before [19]. Chen, et al had identified TSC1/TSC2 P/LP variants in 69.8% fetus CR which seemed more instructive. We speculated it was a selection bias for inclusion criteria: over 90% of the fetal CR included in our study were accidental findings during ultrasound screening in phenotypically normal or presumed normal individual. Although the proportion of multiple CRs here was 10% less than Chen’ paper, both of our study considered that the correlation between CR and TSC is strong regardless of the presence of single or multiple tumors. Rhabdomyomas could not be fully confirmed pathologically until fetus tissues were obtained. Therefore, some other benign cardiac tumors, such as teratomas, fibroids, hemangiomas, or hamartomas, might pulled down the variants detection rate. In order to avoid this selection bias as much as possible, we differentiate these tumors by rigorous fetal echocardiogram selection according to various characteristics [20] before inclusion. Teratoma mostly presents as a single mixed mass in the pericardial cavity, often with pericardial effusion. Fibroid and hemangioma were often higher in size and grew fast, with or without calcification. Myxomas may attach to the atria and swing with the rhythm; also, malignant tumors are rare [20].
Family phenotype collection is in the top priority of genetic counseling. About 32% of mutations identified in this cohort were inherited from parents without typical phenotype. Clinical expression is hugely variable. Some subtle signs, mostly the skin manifestations, were easily overlooked or misdiagnosed until 3–6 outpatient visits to get a preliminary diagnosis in China. A couple who claimed to be “totally healthy” also need double check and confirmation of suspicious phenotype. p20 and p23 had both suffered from “acne” for decades, which was actually the atypical facial angiofibromas for TSC. (Fig. 3de). Mosaicism is another essential factor affecting phenotype, especially of mild cases with NVI by conventional testing, which according for 7.5% of patients with a clinical TSC diagnosis [21]. Variant allele fraction (VAF) in the blood positively correlated with the number of major features [21–22]. Interestingly, low-level mosaicism (0–10%, median 1.7%), which likely arising from a later postzygotic variant, had a milder and distinct clinical phenotype in comparison with other TSC series, with similar facial angiofibromas (92%) and kidney angiomyolipomas (83%) [23]. VAF of the p61 husband was 12%, he showed an unnoticeable asymmetric facial angiofibroma (Fig. 3a). However, for p61 infant who carried the heterozygous variant, not only facial lesions but also moderate mental retardation and epilepsy presented at 6 months. The difference in performance is not only closely related to the mosaic level, but may also be affected by factors such as the structure of the variant, the tissue distribution and the penetrance rate. Cutaneous signs are the most common signs in TSC phenotype series, vice versa, the variant of TSC1/TSC2 was the most frequently to detect in skin lesions [24]. Thus, dermatological consultations or even skin biopsy sequencing are necessary for asymptomatic parents whose variant are accidentally discovered by NGS-trio to confirm the diagnosis, especially when encountering a VUS. The facial phenotype is a useful message for mosaic TSC, and may be the only clue for pathogenicity assessment with an uncertain variation.
Penetrance is high but not absolutely, which is also influences the assessment of pathogenicity of VUS. The TSC2 R462H substitution has been recorded in individuals affected with TSC by Leiden Open-source Variation Database (http://www.LOVD.nl/) and been classified as likely pathogenic. In vitro experimental studies shown that this missense change has an effect on TORC1 activity and interferes with the formation of TSC1-TSC2 complex, and results in an unstable protein and impairs protein function [25]. This mutant is not recorded in population databases (rs45494392, ExAC no frequency) but uniquely presented in fetal 50 and his/her clinically normal sibling, father, and grandfather. The fetal was negative for neuroimaging findings during the whole pregnancy. All index points to benign prognosis, which had been confirmed in following up. Integrated with genetic information in trios, and imaging results may enable early implementation of further diagnostic investigations, perinatal surveillance and family counseling, especially for couples with no signs of TSC phenotype on clinical examination.
For post-test analysis in the NGS era, it was common for phenotypic fetuses to show reported/novel VUS or NVI despite over 10000 variants that have been recorded in these two genes. Numerous VUS intensified the pathogenesis uncertainty and anxiety in pregnant women. Some other angles of view needed to be introduced to aid further evaluation. Pathological changes in the CNS are observed in almost all TSC individuals, and approximately 80–90% of them present cortical tubers and/or SENs, and seems equally familiar in TSC1 and TSC2 pathogenic mutation [24]. However, MRI is not a routine inspection. Neurological manifestations would be easily overlooked and mostly be checked after CR detection during the second-trimester scanning when the pathological changes in CNS might have gone a long way. Saada J et al. [13] indicate that the characteristic cerebral lesions of TSC may form as early as 10–20 weeks of embryonic development. As soon as 2007, Mühler MR et al. had preliminary explored the application of fetal cerebral MRI in sonographically proven CR [26]. Here, we disclosed that neuroimaging finding is a much more specificity indication of genetically TSC and was proportional to poor prognosis. TSC2 mutations seemed more likely to have neuropathy than TSC1 in prenatal. Combination with fetal MRI, neurological lesions were found in 2 NVI cases and 5 TSC2 VUS cases, for whom clinically diagnosis of TSC was made. Exceptionally, in one case with VUS, neurological symptoms occurred, and brain damage originated four months after birth. Among the seven newborns with VUS, 2 (28.6%) developed epilepsy and had neuroimaging findings with following-up.
Benefit from prenatal MRI scanning, some VUS, or even NVI turned to be potentially clinically actionable. But sometimes, lesions might be later onset and are too tiny to find at the time of examination, wherefore any possible sign of occupation requires careful attention. p38 was identified as a de novo TSC2 LP mutation, in 35 GA, only right ventriculomegaly had been sawing, but SENs progressed soon in infancy. Both case37 and case 61 identified mutation of TSC2 and had normal brain MRI at 24 GA and 22GA, respectively, while occurred epilepsy and observed SENs afterbirth. It was impossible to know the exact time of the occurrence of SENs as only brain MRI scanning was performed during the whole pregnancy. So, regular brain MRI monitoring cannot be ignored when the CR fetuses are carrying a P/LP/VUS mutation of TSC genes.
It had been recognized that high-yield mutation detection methods would also help to reduce uncertainty and anxiety in the significant proportion of individuals and families for whom existing diagnostic methods are not informative. In a series of 38 TSC NVI individuals, 2 (6%) mosaic mutations and 5 (13%) heterozygous mutations that had been missed by other mutation detection methods were identified using exon-specific ultra-deep sequencing [21]. As epigenetic silencing having been demonstrated rare in TSC1 and TSC2 [15], possible other reasons for the inability to detect mutations in TSC NVI individuals was: gonadal chimerism, low-level somatic mosaicism, mutations hiding in deep intronic and regulatory regions or mutations to other as yet unidentified genes may cause TSC. For this consideration, the residual risk of TSC in the NVI subset was approximately 5–10% varies between platform [21].
Besides, the possibility of parental gonadal mosaicism cannot be debarred in a typical CR fetus with more than one proband sibling. The reported incidence of germ-line mosaicism range from 2–5% [27]. Case 33 here was a family had had two fetal with CR(s). For this pregnancy, two missense variations of TSC2 were detected but absent in both parents. Although both of the mutations were pathogenic ambiguous, it was highly suspected germ-line mosaicism in the couple. It implied that the recurrence risk of unaffected parents who have had an affected child would increase because of germline mosaicism, and families with high suspicion of this causing should be advised to prenatal diagnosis for each pregnancy.