Clinical and molecular genetic risk determinants in adult long QT syndrome type 1 and 2 patients

The study population was drawn from the Finnish Inherited Arrhythmic Disorder Research Registry established in 1991 and comprising over 4000 molecularly tested subjects. The inclusion criteria were 1) genetically confirmed KCNQ1 or KCNH2 mutation, or genetically confirmed non-carrier status of the family-specific LQTS mutation, 2) and the age of more than 18 years at follow-up end. A questionnaire (Additional file 1) was sent to the study subjects and collected data included occurrence of syncope, setting in which syncope occurred, and data regarding β-blocker therapy. Compliance was defined as forgetting or not taking medication once a month or more often. The decision whether to initiate β-blocker therapy was made by the treating physician upon establishment of the diagnosis.

Data of all deaths during the follow-up were obtained from Statistics Finland by means of social security number search. ICD and pacemaker implantations, and left cardiac sympathetic denervations (LCSD) were identified using the Finnish Hospital Discharge Register (National Institute for Health and Welfare). Medical records were acquired for patients who had device therapy, underwent LCSD, suffered an aborted cardiac arrest (ACA), or died. Collected ICD data included implantation indications, complications, revisions, and ICD discharges. Autopsy documents of patients who died during the follow-up were evaluated. The study was approved by the Ethical Review Committee of Helsinki University Hospital, and a written informed consent was obtained from the study subjects. The Ministry of Social Affairs and Health consented for the participation of deceased subjects.

The follow-up study started from the age of 18 years and ended when the subject 1) returned the questionnaire, 2) turned 40 years [to avoid the effect of acquired cardiac disease] or 3) was deceased, which ever occurred first. The end point for statistical analyses was cardiac event comprising LQTS-related syncope, ACA, appropriate ICD shock, or SCD. LQTS-related syncope was defined as a transient loss of consciousness that was abrupt in onset and offset, and triggered by one of the following factors: swimming, other sports, loud noise, or startle, to avoid inclusion of vasovagal events. [9] Resuscitation events that required external defibrillation were defined as ACA. A death was regarded as being SCD if it was abrupt in onset without evident cause if witnessed, or was not explained by any other cause if it occurred in an unwitnessed setting such as sleep.

Direct DNA sequencing and restriction enzyme assays were used in identification of KCNQ1 and KCNH2 mutations as previously described. [20, 21] Mutations were categorized by mutation type as missense or non-missense (nonsense, frameshift, splice site, insertion or deletion) mutations, and by their location as described previously. [10, 11] Patients carrying more than one LQTS mutation (n = 7) were excluded from the comparison of the clinical characteristics and the multivariate risk analyses, but were included in the sections depicting device therapy. The specific single mutations included in the final study population are detailed in Additional file 2: Table S1. LQT1 Finnish founder (FF) mutations KCNQ1 G589D and KCNQ1 c.1129-2A > G, and LQT2 FF mutations KCNH2 R176W and KCNH2 L552S were combined to form the FF mutation population for LQT1 and LQT2, respectively. Non-carrier family members of the familial KCNQ1 and KCNH2 mutations served as the comparison group. All study subjects were of Finnish origin.

Clinical characteristics were analyzed using chi-square and Fisher’s exact tests for categorical, and Wilcoxon rank-sum and Kruskal-Wallis one-way ANOVA tests for continuous variables. Kaplan-Meier methods were used to depict the cumulative incidence rate (=cumulative probability) of first cardiac event after the age of 18 years by genotype, gender, QTc interval, and mutation. The QTc cut-offs used were based on previous LQTS studies. [4, 7, 18] The significance of the differences was tested by the log-rank test. Multivariate Cox proportional hazards regression models were used to evaluate the independent contribution of genetic and clinical risk factors to first cardiac event after 18 years of age. Survival was also evaluated by assessing incidence rates of first cardiac events per person-years. All cumulative incidence graphs, log-rank tests, and primary Cox regression and incidence rate analyses were censored at the initiation of β-blocker medication. Secondary Cox regression and incidence rate analyses (Medical Treatment paragraph) were carried out including the follow-up time with time-dependent β-blocker medication. No violation of the proportional hazards assumption was detected as tested by log-log graphs. A separate QTc missing covariate was used for mutation carriers whose QTc data were unavailable (n = 35). No statistically significant interactions were discovered in interaction term analyses. All Cox regression models were adjusted for gender, QTc duration, cardiac events before the age of 18, and family membership using robust sandwich estimators. Statistical analyses were carried out using SPSS version 22. A 2-sided p-value ≤0.05 was interpreted as statistically significant.

The characteristics of the patients with a single mutation are shown in Table 1 and with more than one mutation in Additional file 2: Table S2. Characteristics of the non-carrier relatives are presented in Additional file 2: Table S3. Altogether seven of the ten deaths in mutation carriers, and none of the four deaths in non-carrier relatives were arrhythmia-related. Among mutation carriers seven (1%) suffered a SCD and eight (1%) at least one ACA. In LQT1 2% and in LQT2 3% of the cardiac events were fatal. Female mutation carriers were more often probands (26% vs 15%, p = 0.001), and had a longer QTc (473 vs 454 ms, p < 0.001) as compared with males. Altogether, QTc duration ≥500 ms was measured in 132 (15%) mutation carriers. One family had 7, three families had 3, and the remaining families had 0–2 cardiac events.

LQT2 genotype was associated with a higher risk of cardiac events in comparison to LQT1 (cumulative probability 18% vs 11%, p = 0.01; HR = 2.1, p = 0.002, Table 2). Both LQT1 and LQT2 females were more often symptomatic than males (cumulative rate 16% vs 3%, p < 0.001, for LQT1; and 23% vs 8%, p = 0.01, for LQT2, Figure 1), with a hazard ratio of 3.2 for the female versus male comparison (p < 0.001). The risk was distinctly higher in patients who were symptomatic before the age of 18 years (cumulative rate 52% vs 9%, p < 0.001; HR = 5.93, p < 0.001). QTc duration ≥500 ms increased the risk 2.7-fold compared to QTc < 470 ms (p = 0.001). We repeated the analyses after excluding FF mutation carriers, and the results regarding genotype, gender, symptoms before age 18, and QTc duration were similar.

The risk of cardiac events was higher in mutation carriers than in non-carrier relatives even after adjustment for QTc, gender, and cardiac events before age 18 (Table 3). In comparison with non-carrier relatives with QTc < 440 ms, mutation carriers with QTc < 440 and ≥ 440 ms had a 4.2-fold (p = 0.01) and an 11.1-fold (p < 0.001) risk, respectively, of suffering a cardiac event.

Upon pairwise comparison, KCNQ1 D317N mutation carriers showed a higher risk of cardiac events than G589D, c.1129-2G > A or other KCNQ1 mutation carriers (cumulative probability 40%, 10%, 11% and 14%, respectively, p = 0.002–0.047, Figure 2; HR = 3.0–3.9, p < 0.001–0.03). Risk or rate of events did not differ between G589D, c.1129-2G > A and other KCNQ1 mutation carriers. Among LQT2 patients the carriers of the c.453delC, L552S or R176W mutation had fewer cardiac events than carriers of other KCNH2 mutations (cumulative rate 5%, 13%, 9% and 43%, respectively, p < 0.001–0.013, Figure 3; HR = 0.11–0.23, p < 0.001).

An analysis of the effect of the mutation site (Table 1S) indicated that the cumulative rate of cardiac events was higher in patients carrying a missense KCNQ1 non-FF mutation in pore-loop than in non-pore-loop region (37% vs 7%, p = 0.01). In a pairwise comparison, pore-loop mutation carriers also had a higher rate of events than carriers of a KCNQ1 G589D or c.1129-2G > A mutation (p = 0.02–0.001), but there was no difference between non-pore-loop and FF mutation carriers (p = 0.48–0.53). A similar analysis of the LQT2 patients showed that among missense KCNH2 non-FF mutation carriers there was a tendency of higher event rate in patients with a non-pore-loop than a pore-loop mutation (58% vs 28%, p = 0.14). Non-pore-loop mutation carriers had a higher event rate than KCNH2 L552S or R176W mutation carriers (p < 0.001), and the event rate was similar between carriers of a pore-loop and a FF mutation (p = 0.054–0.15).

The rate of cardiac events was the same regardless of whether the mutation had been inherited from mother or father as tested separately among all KCNQ1, KCNQ1 G589D, and all KCNH2 mutation carriers (data were available for 459 KCNQ1 and 150 KCNH2 mutation carriers).

β-blocker medication was initiated to 249 mutation carriers at a mean age of 22.3 and 22.7 years in LQT1 and LQT2 patients, respectively. Use of medication was more common in non-FF than FF patients (45% vs 25%, p < 0.001), and in females than males (34% vs 20%, p < 0.001). Altogether 27 patients suffered a cardiac event during the medication, including SCD in four patients. Breakthrough events were more common in non-FF than FF mutation carriers (16% vs 7%, p = 0.04). In patients to whom β-blocker was prescribed, the medication was associated with 60–81% reduction in the risk of first cardiac event at the age of 18–40 years (p < 0.001, Table 4).

Treatment with psychotropic agents was equally common in mutation carriers and non-carrier relatives (antidepressants: 5% vs 5%, p = 1.00; antipsychotics: 2% vs 1%, p = 1.00; anxiolytics: 0.4% vs 0.6%, p = 1.00, respectively). However, mutation carriers who had suffered a cardiac event reported more often taking daily antidepressant drug at the end of the follow-up (12% vs 3%, p = 0.02).

Characteristics of the 39 patients with ICD, pacemaker or LCSD are detailed in Additional file 2: Table S4. The incidence rate of cardiac events showed reduction after ICD implantation: 152.4 and 56.8 per 1000 person-years before and after implantation, respectively. Similarly, cardiac events decreased after pacemaker implantation: 82.3 and 0 per 1000 person-years. ICD was implanted more frequently to non-FF than FF mutation carriers (7% vs 1%, p < 0.001). Common ICD implantation indications were ACA, or LQTS-related syncope during β-blocker medication. An appropriate ICD shock therapy occurred in seven (32%) and an inappropriate shock in three (14%) patients. Six patients suffered a complication in the ICD or pacemaker system (incidence rate 26.8 per 1000 person-years).

The patients with SCD, ACA, or ICD shock therapy are presented in Table 5. Common factors predisposing to SCD were QT-prolonging medication (n = 4; citalopram, thioridazine or tizanidine), and absence of β-blocker therapy (n = 3). Only one β-blocker medication -compliant patient without predisposing factors suffered a SCD. Of the eight patients who suffered ACA, two were using a QT-prolonging drug (terfenadine or amiodarone), and none were on β-blocker medication at the time of the event.

Similar to previous studies, female gender, LQT2 genotype, cardiac events before the age of 18, and prolonged QTc duration were found to increase the risk of cardiac events. [7, 8, 10, 11, 17‐19, 22] In the present study, mutation carriers with a normal QTc duration (< 440 ms) had a 4.2-fold risk compared to non-carrier relatives, whereas in an earlier study the corresponding risk was 10-fold. [4] A potential reason for the difference is the end point of ACA or SCD in the previous study, as opposed to LQTS-related syncope, ACA, ICD shock, or SCD in our study. An earlier study demonstrated that cardiac events among genotype-negative family members are mostly attributed to nonfatal syncopal episodes. [23] Although mutation carriers with a normal QTc had a higher risk than non-carrier relatives, normal QTc associated with a good prognosis even in patients left untreated with β-blockers: none of the previously asymptomatic non-proband mutation carriers suffered a cardiac event during the prospective follow-up. This is of note as cascade screening of family members reveals a growing number of asymptomatic mutation carriers with a normal or only slightly prolonged QTc.

In the current study, missense KCNQ1 mutations located in the pore-loop region associated with a higher rate of cardiac events than non-pore-loop mutations, which was not seen in a previous study. [24] However, in the present study most of the pore-loop mutation carriers had the highly malign KCNQ1 D317N mutation. Previously, KCNQ1 cytoplasmic loop (c-loop) mutations have been associated with a higher risk. [11] In our study, only six patients were carriers of a c-loop mutation precluding exact comparison of c-loop and non-c-loop mutations. In the present study, KCNH2 mutations situated in the pore-loop region were not associated with an increased risk as seen in previous studies. [10, 22] On the other hand, only 15 subjects had a pore-loop mutation in our analysis.

All four founder mutations had a significant QT-prolonging effect and associated with increased risk of cardiac events. Similarly to our recent study of pediatric LQT1 and LQT2 population, the KCNH2 FF mutations led to a milder phenotype than non-FF KCNH2 mutations. [14] This might be related to the fact that both KCNH2 FF mutations lead to a functional channel with increased deactivation rate, [21, 25] whereas many non-FF KCNH2 mutations have more dramatic effects on channel function. However, the risk between KCNQ1 FF and non-FF mutations did not differ from each other after excluding the KCNQ1 D317N mutation.

The KCNQ1 D317N mutation appeared exceptionally malign, and previously it has been shown to associate with diminished chronotropic response and exaggerated QTc prolongation after exercise. [26] This mutation is located in the pore-loop region and leads to complete loss of channel function with a dominant negative effect on the wild type channel protein. [27] On the other hand, the phenotype associated with the KCNH2 c.453delC mutation turned out to be reasonably mild, in harmony with an earlier study. [28] This N-terminal mutation leads to a premature termination codon, which likely targets the mutated mRNA to non-sense mediated mRNA decay without any dominant negative effect.

Only 29% of the patients in the present study used β-blocker medication compared to 45–62% in earlier studies. [10, 11, 19, 22] However, the previous studies included adolescents, who are more frequently treated with β-blockers. [29] In accordance with previous studies, non-compliance to β-blocker therapy increased the risk of cardiac events. [14, 30] Nevertheless, also non-compliant patients demonstrated a decrease in the incidence rate of cardiac events after initiating medication, which suggests a protective role for β-blockers, even when present in suboptimal therapeutic concentrations. However, considerable proportion of the SCD and ACA cases associated with insufficient β-blocker medication underlining the importance of uninterrupted use of prescribed β-blockers.

According to recent ESC guidelines, β-blocker treatment should be initiated if QTc is prolonged, and it may be useful even with normal QTc duration. [3, 4] The results of the current study indicate that asymptomatic adult LQT1 and LQT2 males with FF mutation and QTc duration < 500 ms have a very low risk, and suggest that the avoidance of risk factors may be a sufficient measure. Also, β-blocker medication might not be mandatory for primary prevention in adult KCNH2 c.453delC mutation carriers and female FF mutation carriers with QTc < 500 ms. For the remaining patients β-blocker treatment is recommended.

In the current study, the use of antidepressant drugs at the end of the follow-up was more common in symptomatic patients. Analysis of a possible causal connection was not feasible due to limited data on the length of the antidepressant medication. However, it is possible that these patients were treated with antidepressants already at the time of the cardiac event. Furthermore, four of the seven SCD cases and two of the eight ACA cases involved treatment with a potentially QT-prolonging drug. Therefore, the present and a previous study [30] emphasize avoidance of QT-prolonging drugs in prevention of potentially life-threatening cardiac events.

The incidence rate of cardiac events showed reduction after ICD or pacemaker implantation. A potential explanation arises from bradycardia pacing which has been previously demonstrated to reduce the risk of ICD shocks in high-risk patients. [5] However, in the current study, the initiation of β-blocker therapy was coincidental with device implantation in 18 of the 37 cases. Therefore, the reduction in cardiac events might be attributable to β-blocker treatment.

There are a number of noteworthy limitations in our study. First, only 55% of the patients initially surveyed responded to our inquiry leading to a possibility of selection bias. Second, comparison of patients with different mutation categories occasionally resulted in relatively small patient subgroups. Third, initiation of β-blocker medication was not standardized across participants leading to a concern about confounding by indication. Fourth, we have performed rather many statistical comparisons which may increase the number of false positive findings.