Introduction

Long QT syndrome (LQTS) and its signature rhythm, Torsades de Pointes (TdP), have been diagnosed in utero by magnetocardiography (MCG) and, in a few cases, successfully treated before birth.^1–6 The findings of a prolonged corrected QT (QTc) interval and characteristic polymorphic ventricular tachycardia are well described in case reports, yet the accuracy of MCG for diagnosing LQTS in a sizable population of at-risk fetuses has not been studied. In addition, little is known of the electrophysiology and natural history of LQTS before birth, how repolarization characteristics and LQTS rhythms compare before and after birth, or whether there are electrophysiological harbingers of severe prenatal LQTS phenotypes.

Clinical Perspective on p 2191

The purposes of this study were 2-fold. First, in a large cohort of fetuses at risk for LQTS we tested the ability of fetal MCG to distinguish fetuses that had LQTS from those that did not. Second, we compared the cardiac intervals, rhythms, and repolarization characteristics before birth (by MCG) and after birth (by ECG) in the same subjects. We expected that subjects with the longest QTc intervals would manifest the most severe LQTS rhythm phenotypes.

Methods

Subject Cohort

A cohort of 30 subjects was recruited from subjects referred for fetal MCG to the Biomagnetism Laboratories in the Department of Medical Physics at the University of Wisconsin – Madison (n=23), the Department of Pediatrics at the University of Tsukuba, Tsukuba, Japan (n=6) and the Department of Pediatric Cardiology at the National Cerebral and Cardiovascular Center, Osaka, Japan (n=1) from 1996 to 2012. Inclusion criteria were (1) a positive family history of LQTS or unexplained sudden death of a sibling during infancy or childhood or (2) findings of an LQTS rhythm: low fetal heart rate (fHR), defined as heart rate ≤3^rd percentile for gestational age⁷; ventricular tachycardia (VT; atrioventricular [AV] dissociation and faster ventricular than atrial rate); or anti-Ro/La antibody-negative second-degree atrioventricular block (2°AV block) with a structurally normal heart diagnosed by fetal echocardiography. We reviewed the pre- and postnatal medical records, including results of fetal echocardiograms, postnatal ECG, and postnatal genetic testing. The research was approved by the Institutional Review Boards at the participating centers.

Fetal MCG Evaluation

Fetal MCGs were performed using a 37-channel biomagnetometer (Magnes, 4D Neuroimaging, Inc.) at the University of Wisconsin-Madison and 64-channel biomagnetometers (MC-6400, Hitachi Ltd.) at both the University of Tsukuba and the National Cerebral and Cardiovascular Center in Osaka. All of the recordings were made in a magnetically shielded room. For each subject, continuous MCG data were collected in 5- to 10-minute runs over a period of 10 minutes to 1.5 hour. The probe was positioned on the maternal abdomen, proximal to the fetal heart. Using autocorrelation to detect QRS complexes, ventricular fHR tracings were computed from the RR intervals, and actograms (tracings of fetal activity derived from movement-related changes in signal amplitude) were derived from the instantaneous QRS amplitudes.⁸ Approximately 50 consecutive fetal complexes were averaged during periods of fetal quiescence to increase the signal-to-noise ratio, before measurement of the fetal (f) cardiac time intervals—fQRS, fQT, and fQTc.

Actocardiograms were examined to identify movement-related changes in fetal rhythm and fHR (ie, fHR reactivity). Reactivity was assessed in fetuses at ≥ 28 weeks’ gestation; before this time it cannot be reliably assessed. At 28–32 weeks, normal reactivity is defined as fHR accelerations ≥10 bpm, lasting ≥10 s that occur in association with fetal movement. After 32 weeks’, the accelerations increase in amplitude and duration to >15 bpm and >15 s, respectively.⁹

Fetal rhythm was assessed from continuous rhythm recordings, after removal of maternal interference. We defined 2° atrioventricular block as a regular P-P interval with conduction of every other atrial beat, and TdP as paroxysms of a wide complex rhythm with a variable QRS morphology that exceeded the baseline ventricular rate. We described the QRS morphology as narrow (< 95^th % for gestational age) or wide (> 95^th %).¹⁰ Because P-waves were not always visible on the tracings, we made the assumption that a rhythm with a regular R-R interval, a narrow QRS, and a rate of 90–160 bpm was sinus rhythm. We also inspected the recordings for beat-to-beat changes in QRS and T-wave morphology (ie, alternans).

Postnatal Evaluation: ECG, Rhythm, and Genetic Testing

After birth, neonatal (n) cardiac intervals (nQRS, nQT, and nQTc), rhythm, and T-wave characteristics were assessed by 12-lead ECG performed within the first 48 hours of life. Subjects were tested for mutations in LQTS-susceptibility genes (LQT1, 2, 3, 5, and 8) by commercial laboratories (Familion, Transgenomic Inc, New Haven, CT; GeneDx, Gaithersburg, MD). If the family history was positive for LQTS, we tested the newborn only for the family mutation. Similarly, if a newborn was found to have an LQTS mutation, first degree relatives were evaluated by ECG and, when appropriate, genetic testing. LQTS was diagnosed by the presence of a mutation in an LQTS-susceptibility gene or, in the absence of such a mutation, the clinical findings of TdP or neonatal QTc > 490 ms. We compared the cardiac intervals, rhythm, and repolarization characteristics between individuals before and after birth, and between different LQTS mutation types. Finally, we sought fMCG characteristics predictive of fetal/neonatal TdP.

Statistical Analysis

The performance of fQTc and fQT for detection of LQTS were assessed using receiver operating characteristic curves. The receiver operating characteristic curves were computed 2 ways: first using data from all sessions and second using data from just the first session of each subject, to remove the effects of within-subject correlation. We reported descriptive statistics (mean±SD) for continuous data. Within subjects, differences between fetal and neonatal cardiac time intervals and heart rate were tested by dependent t test. The degree of correlation between QTc determined by fMCG and postnatal ECG was assessed by linear correlation analysis. Differences between LQTS and non-LQTS subjects were tested by t test. For subjects with multiple sessions only the first session was used. Because of the small sample size of subjects with TdP, nonparametric statistics (Mann–Whitney test and Kruskal–Wallis test) were used to compare differences in continuous variables for T-wave alternans (TWA), QTc, and LQTS rhythms to determine which fMCG features predicted fetal/neonatal TdP. A 2-tailed P value of ≤0.05 was considered statistically significant. Analyses were performed with SPSS software (release 19.0, SPSS-IBM, Chicago, IL).

Results

Subject Cohort

The subject cohort is shown in the Table, which divides the subjects into 2 groups based on LQTS diagnosis. Between 1996 and 2012, 29 pregnant women during 30 pregnancies were referred for 36 fMCG studies (4 subjects had 2 studies, 1 subject had 3) at 19–38 (mean 29.5±5.2) weeks’ gestation because of risk factors for fetal LQTS. One mother was evaluated for 2 pregnancies that were counted separately (#4 and #20). Twenty subjects were studied because of a family history of either a known LQTS gene (n=17) or the neonatal/childhood sudden death of a sibling (n=3), and were referred at 19.1±4.8 weeks. Of these, one had 2° AV block and one had premature ventricular contractions (PVCs) at the time of referral. The other 10 fetuses were studied because of suspicious rhythms—low fHR in sinus rhythm (n=6), 2° AV block (n=1), VT (n=1), 2° AV block + VT (n=2)—and were referred at 29.8±5.2 weeks.

In the Table, Group 1 consists of the 21 fetuses found to have LQTS: 17 with a mutation in an LQTS susceptibility gene, 1 mutation negative subject with nQTc of 540 ms, and 3 that declined genetic testing, but had nQTc> 490 ms and a family history of LQTS. The QTc intervals of the latter 4 subjects remained prolonged (>490 ms) for months to years after birth (data not shown). Remarkably, only 57% (12) of these fetuses had a family history of LQTS (10 maternal, 2 paternal). Group 2 consists of 9 fetuses found not to have LQTS, based on absence of mutation and normal postnatal QTc. Of these, 8 were referred because of a family history of LQTS or the sudden death of a sibling during infancy or childhood, and 1 had low fHR in sinus rhythm.

Fetal MCG

Fetal Cardiac Intervals and fHR Reactivity

The fQTc intervals are indicated on the averaged waveforms in Figure 1A and 1B. As expected, fQTc was much longer in Group 1 (565±65 ms) than in Group 2 (421±54 ms; P<0.0001; Table). Furthermore, the 4 subjects with fetal TdP had QTc 628–692 ms. Their mean QTc (660±31 ms) was significantly longer than that of the Group 1 subjects without TdP (549±55ms; P= 0.003). A notable feature of the data of the Group 1 subjects (Figure 1) is the late-peaking morphology of T-waves, which show T-wave humps in some subjects. The QT_peak/QT ratio was 0.88±0.05 for Group 1 and 0.73±0.12 (p<0.0001) for Group 2, where QT_peak is the interval from the beginning of the QRS complex to the peak of the T-wave.

• Download figure
• Download PowerPoint

The accuracy of fQTc for detection of LQTS was assessed using a receiver operating characteristic curve (not shown). Using data from all sessions, the best performance, corresponding to the point closest to the upper left corner of the graph, was obtained for a threshold of 460< QTc<520 ms, which yielded sensitivity 89% and specificity 89%. Using data from just the first session of each subject, the optimal threshold was unchanged and yielded sensitivity 86% and specificity 89%.

Baseline fHR in sinus rhythm was significantly lower in Group 1 (range 90–144 bpm, mean 118.9±13.3 bpm) than in Group 2 (range 125–147 bpm, mean 141.0±9.4; P<0.0001; Figure 2). FHR reactivity, which cannot be reliably assessed before 28 weeks gestation, was low in 7 of 16 sessions of the Group 1 subjects and in 0 of 3 of the Group 2 subjects studied after 28 weeks gestation (Table).

• Download figure
• Download PowerPoint

Fetal Rhythms

The 21 subjects in Group 1 showed VT or VT + 2° AV block (n=4), isolated 2° AV block (n=2), normal fHR in sinus rhythm (n=2), and isolated low fHR in sinus rhythm (n=13; Table). Of the 13 with isolated low fHR in sinus rhythm, only 2 had fHR≤110 bpm, the commonly accepted obstetrical definition of fetal bradycardia. For the 4 subjects showing fetal TdP, the chronology of the various rhythms during the fMCG recording session is shown in Figure 3. The percent time in each rhythm was in the range 21–68% for sinus rhythm, 0.9–31% for 2°AV block, and 22–79% for TdP. The duration of TdP salvos (n=111) was in the range 1–462 s (mean 39 s). The ventricular rate during TdP was in the range 160–300 bpm and varied by as much as 30–70 bpm from episode to episode within the same subject.

• Download figure
• Download PowerPoint

Patterns of Initiation and Termination of Torsades de Pointes

We observed complex and myriad rhythm patterns at initiation and termination of TdP (Figure 4). The most complex and varied were seen in subject #15 (SCN5A R1623Q). In both SCN5A and KCNH2 subjects, episodes of TdP were initiated by a regular wide complex or aberrantly conducted rhythm, often involving QRS alternans or T-wave alternans, with a rate similar to narrow complex sinus rhythm (cycle length 418–488; Figure 4A–4C). TdP also followed a period of 2° AV block (Figure 4D). What was notably rare was the typical long–short sequence (ventricular ectopy) that initiates TdP seen after birth.

• Download figure
• Download PowerPoint

Rhythms occurring after spontaneous termination of TdP are shown in Figure 4E–4I. These rhythms included transient 2° AV block followed by sinus rhythm with QRS alternans Figure 4E; 3° AV block (Figure 4F), pauses or narrow complex bradycardia with or without T-wave alternans (Figure 4G–4I). Ventricular fibrillation was not observed during any of the recordings.

Repolarization Characteristics

Episodes of macroscopic TWA conforming to an ABAB patterns were seen in 6 subjects (Figure 5). Discordance of T/QRS and large, late-peaking T-waves are prominent in these data. Subject #10 (Figure 5C) showed relatively mild TWA and did not have TdP. The other 5 showed prominent TWA and also QRS alternans. Importantly, they all had worse outcome: subjects #11, 15, 16, and 18 had fetal or neonatal TdP; subject #17 was delivered early because of new-onset multiform PVCs and R on T phenomenon diagnosed solely by fMCG. In this case the family history predicted significant risk of unstable arrhythmia. After preterm delivery, the infant was started on, and responded to, β-blocker therapy. The most unusual tracing was seen in the 21-week fetus with a de novo SCN5A L409P mutation (subject #16; Figure 5B).¹¹ In this tracing, the T-wave is not well defined, and instead repolarization appears to occur continuously throughout the cardiac cycle. In contrast to adult TWA, the episodes of fetal TWA were typically brief and transient, often lasting 10 s or less. In addition, they were not strongly associated with elevated fHR. They could occur before or during fHR accelerations or during periods of quiescence.

• Download figure
• Download PowerPoint

An additional repolarization anomaly observed in 7 (33%) subjects in Group 1 was a marked increase in the T/QRS amplitude ratio during fetal movement (Figure 6). In normal subjects, fetal movement commonly alters the amplitude of the individual waveform components, but typically the ratio of the amplitudes shows relatively small changes and is equally likely to increase or decrease. In these 7 subjects, however, the ratio showed a consistent increase by a factor of 2 or more (Figure 6A). These subjects also showed little or no fHR reactivity (Figure 6C).

• Download figure
• Download PowerPoint

Fetal Echocardiography, Fetal MCG, and Clinical Management

Fetal echocardiography and fMCG were usually concordant (Table), but the additional information from fMCG was significant in several cases. In 2 subjects, fMCG revealed rhythms not seen during serial echocardiograms: subject #18 (maternal KCNH2 mutation) had short runs of TdP that prompted substitution of nadolol with high dose propranolol and magnesium, and subject #17 (paternal KNCH2 mutation) was delivered because of multiform PVCs and R on T phenomenon (Figure 5E). Also, the definitive diagnosis of TdP in fetuses with VT allowed targeted prenatal treatment, including transplacental treatment with intravenous lidocaine and magnesium or oral β-adrenergic blocking agents that restored sinus rhythm (subject #6, 15)^1,3 or reduced the frequency of TdP episodes (subject #18), thereby postponing delivery until fetal maturity. Lastly, rhythm and QTc duration derived from fMCG was used to risk-stratify prenatal and postnatal care: fetuses with normal fQTc were usually delivered at their referring hospitals; fetuses with prolonged fQTc were delivered at cardiac centers of excellence.

Postnatal Evaluation

ECG Cardiac Intervals and Rhythm

Of the 30 subjects, 29 had postnatal ECGs; 1 pregnancy (fetus #16) was electively interrupted. The values of QTc determined by fMCG and postnatal ECG were highly correlated (n=29, r²= 0.494, p<0.0001). Prenatal fMCG and postnatal ECG were concordant for assessment of QTc prolongation with the exception of subject #22. This subject was referred for fMCG at 31 weeks’ because of fetal bradycardia (fHR 104–125 bpm) and was found to have a prolonged fQTc (524 ms), accompanied by a late-peaking T-wave characteristic of fetal LQTS. Postnatally, however, the subject showed sinus bradycardia, a normal nQTc (424 ms), no LQTS mutation, and no Tdp or 2° AV block. Excluding this subject, there were no substantial differences between mean fetal and neonatal QRS duration (data not shown), mean fetal and neonatal heart rates (118.9±13.3 vs. 117.2±13.5, P=0.67), or mean fetal and neonatal QTc duration (565.3±65.3 vs. 540.7±53.4; P=0.18). Pre- and postnatal rhythm concordance was also high. Nearly all fetuses with isolated low fHR in sinus rhythm and all liveborn infants with fetal TdP had the same rhythm after birth. The 1 exception was Subject #11, who had a brief episode of 2° AV block in utero, but had both 2° AV block and TdP after birth.¹²

Genetic Testing of Subjects and Family Members

Genetic testing was performed on 27 of 30 neonatal subjects (Table). All 9 subjects in Group 2 (normal fetal and neonatal QTc) were tested and none had an LQTS mutation. Of the 21 Group 1 subjects, 17 had a positive mutation: 9 had KCNQ1 mutations, 5 had KCNH2 mutations, 2 had SCN5A mutations, and 1 had a CALM2 mutation. One subject (#10) did not have a mutation in a known LQTS susceptibility gene, and 3 subjects were not tested.

If the fetus was a proband (n=7), ECG screening for LQTS was recommended to first-degree relatives. Parents of 1 subject refused to be tested because of insurability concerns, members of 3 families had normal ECGs, and 6 members of 3 families had abnormal ECGs. These asymptomatic family members (1 father, 2 mothers, and 3 siblings) had the same mutation as that of the proband.

Genotype/Phenotype Correlations

No significant differences in fetal HR or fetal QTc duration could be detected between LQTS mutation type; however, the numbers were too small to draw meaningful conclusions. Associations between the most severe phenotypes—QTc≥620 ms, TdP, and TWA—and specific mutations were not evident; however, these phenotypes were not seen in any fetal subject with a KCNQ1 mutation. This is of interest because TWA was first noticed in an LQT1 subject.¹³

Discussion

This is the first study to document the electrophysiological characteristics of fetal LQTS and to evaluate the accuracy of fMCG for identifying congenital LQTS in a sizable population of at-risk fetuses. The main findings were as follows: (1) Using a threshold of QTc> 490 ms, fMCG was able to identify the fetuses with LQTS with high accuracy (89% sensitivity, 89% specificity). This threshold value happens to correspond to the upper bound of the 95% prediction interval of QTc in normal fetuses.¹⁴ (2) Perinatal TdP was associated with markedly prolonged QTc (≥620ms) and was accompanied by multiple, uncommon rhythms: 2° atrioventricular block, TWA, and QRS alternans; however, TdP did not degenerate into ventricular fibrillation, even when intermittent TdP was present for days or even weeks. (3) Fetal TdP sometimes initiated with a long–short pattern, but most often it initiated with an aberrantly conducted beat with little or no cycle lengthening. The predominance of this initiation pattern may be unique to the fetus. Below we discuss these findings and consider their clinical implications.

As verified by genetic testing, the sensitivity and specificity of fMCG for the diagnosis of LQTS in utero was high. The concordance between fMCG and postnatal ECG was also high, both for QRS and QTc measurements and for rhythm assessment. In addition to fQTc prolongation, another sentinel marker of fetal LQTS was the late-peaking T-wave morphology, and the occasional appearance of T-wave humps. In addition, some LQTS subjects showed a pronounced increase in T-wave amplitude in association with fetal movement, which we speculate may mimic a stress test. These distinctive characteristics were especially helpful at early gestational ages when fetal T-waves tend to be flat.

Our study demonstrates that QTc prolongation, the defining characteristic of LQTS, manifests early in life—just how early remains unknown, but our data support the inference that it is commonly present after mid-gestation. One LQTS subject (#21), however, had a normal fQTc with normal T-wave morphology in the first session at 27 weeks’ and a prolonged fQTc with late-peaking T-wave morphology in 2 subsequent sessions at 29 and 34 weeks. This suggests that in some subjects QTc prolongation does not develop until later in pregnancy, and that a single screening at an early gestational age may fail to detect a small percentage of cases of LQTS.

The prognostic value of fQTc assessment was demonstrated by the finding that of the 6 fetuses with longest QTc, 4 (QTc≥ 620 ms) had fetal TdP, 1 (QTc= 633 ms) had neonatal TdP, and 1 (QTc= 610 ms) was delivered at 34 weeks after fMCG revealed the ominous findings of multiform PVCs and multiple episodes of R on T phenomenon.

Our fHR data agree with those of Mitchell and co-workers, who showed that LQTS is strongly associated with fHR< 3^rd%. In addition, we studied fetuses with a family history of LQTS that do not carry the mutation and showed that they have fHR ≥ 3^rd%. Along with low baseline heart rate, the prevalence of nonreactive heart rate patterns in the LQTS fetuses documented in this study is compatible with lower-than-normal right sympathetic cardiac activity or a blunted response to sympathetic drive, as seen postnatally.¹⁵ Sympathetic imbalance is believed to play a critical role in triggering lethal arrhythmias in LQTS.¹⁶

Among the most striking rhythm patterns were the episodes of TWA, which commonly showed large, late peaking T-waves, QRS-T discordance, and QRS alternans. TWA is a critically important marker of cardiac instability and severe disease.¹³ QRS alternans, however, is not considered an indicator of cardiac instability, nor is it normally associated with congenital LQTS. In our subjects it occurs because extreme QTc prolongation in conjunction with the higher heart rate of the fetus results in substantial overlap of ventricular depolarization and repolarization, as exemplified by R on T phenomenon, which in the presence of TWA produces alternation in QRS morphology as a result of aberrancy. The combination of prominent TWA and QRS alternans was strongly associated with a poorer prognosis. This is not surprising because QRS alternans occurs in the most severe cases when QTc is markedly prolonged and TWA is present. In addition, however, aberrancy associated with QRS alternans appears to be an important mechanism of initiation of TdP.

Unlike what is commonly seen in the postnatal patient, most of the episodes of TdP were not pause-dependent. TdP usually occurs after a pause that delays repolarization. This can induce early afterdepolarizations and enable the formation of a PVC that initiates TdP, resulting in the long–short pattern that is a hallmark of TdP. In our subjects, however, most of the initiations involved various forms of aberrantly conducted beats with little or no cycle length changes. We therefore speculate that reentry, facilitated by bundle-branch block, was an important mechanism of initiation. Its predominance in the fetus may result from the prevalence of aberrancy, as exemplified by QRS alternans, relative to what is seen postnatally. In addition, initiation by the long-short pattern may be hampered by the strong rate dependence of early after depolarizations, which are less likely to reach critical threshold at the higher heart rates of the fetus.

Another unexpected finding was that ventricular fibrillation was not seen, even when TdP recurred often, was present continuously for many minutes, or occurred intermittently for days and even weeks. Postnatally, it is commonly observed that TdP degenerates into ventricular fibrillation if it persists for more than several minutes. We speculate that the fetus does not develop severe myocardial hypoxia because the maternal/fetal circulation is protective or is better adapted to tolerate hypoxia because of the lower ambient pO₂ in utero. Although we did not see ventricular fibrillation, 2 fetuses (gestational ages 21 and 31 weeks) with SCN5A mutations developed heart failure. Ventricular fibrillation could still occur unobserved, because lethal rhythms are exceedingly brief. Without molecular autopsy, the resultant deaths would be classified as stillbirths or sudden infant death syndrome (SIDs). A recent study of stillborn infants showed a 3.3% incidence of mutations in LQTS susceptibility genes and an 8.8% incidence of genetic variants that lead to dysfunctional LQTS-associated ion channels in vitro,¹⁷ supporting the hypothesis that LQTS can cause stillbirths as well as sudden infant death syndrome.¹⁸

Fetal MCG results altered antenatal management of the mother and fetus, the timing and location of delivery, and the anticipation of postnatal management, regardless of the family history and independent of LQTS rhythm. Mothers of fetuses with suspected LQTS did not receive QT prolonging medications such as oxytocin, erythromycin, or ondansetron during pregnancy, labor, or delivery. Such management may have averted harm not only to the fetus but also to the pregnant mother unexpectedly found to have LQTS after undergoing ECG screening because of her fetus’s MCG results. MCG also enabled definitive diagnosis of TdP, despite its transient and variable nature, which prompted appropriate, effective in utero pharmacologic therapy. Importantly, TdP was detected in 18% of subjects with an LQTS family history, but negative echocardiographic surveillance before fMCG evaluation. These subjects were successfully delivered at cardiac centers of excellence.

Finally, LQTS was diagnosed in 5 subjects with an isolated finding of fHR < 3^rd% and in 6 unsuspecting members of 3 fMCG-diagnosed LQTS families. These results suggest that, in addition to a positive family history or findings of complex rhythm in utero, fetuses with low-for-gestational-age heart rates should be monitored for the development of LQTS rhythms and considered for fMCG screening.

In conclusion, QTc assessment by fMCG was shown to have diagnostic value for identification of fetuses with LQTS and prognostic value for prediction of a severe phenotype. In addition, fMCG was invaluable for the detection and definitive diagnosis of TdP. FMCG findings not only prompted successful in utero pharmacological treatment to restore sinus rhythm and postpone the delivery of a premature fetus, but also guided anticipatory neonatal care. The ability to recognize LQTS in the fetus and to effectively treat TdP in utero can be lifesaving.

Footnotes