Genetic screening in acquired long QT syndrome? CAUTION: proceed carefully

This editorial refers to ‘The genetics underlying acquired long QT syndrome: impact for genetic screening’†, by H. Itoh et al., on page 1456. The QT interval on the electrocardiogram reflects the duration of ventricular repolarization which results from the concerted opening and closing of ion channels in the cell membrane of ventricular myocytes. The long QT syndrome (LQTS) is a potential lethal entity because a prolonged QT interval is associated with the development of a polymorphic ventricular tachycardia, termed Torsades de Pointes (TdP), which can degenerate to ventricular fibrillation and sudden death.1 Several clinically easily identifiable factors are recognized to prolong the QT interval, including bradycardia, hypokalaemia, hypomagnesaemia, and particularly a large variety of drugs such as antiarrhythmics, antibiotics, antihistamines, and antidepressants. A prolonged QT interval in the presence of such risk factors is called acquired LQTS (aLQTS), and the principal underlying mechanism is believed to be mediated by a reduction of IKr, one the most essential potassium currents in the heart.2 Unfortunately, the response of the QT interval to acquired risk factors is often unpredictable, even in individuals with similar clinical characteristics, ranging from no or minimal change in QT interval to severe QT prolongation and TdP.3 In clinical practice, this has led to the unforeseen death of subjects using therapeutic dosages of a certain drug, and, as a result, withdrawal of some promising drugs from the market. Therefore, although difficult, it is of utmost clinical importance to identify those subjects at risk of developing aLQTS. In the mid-1990s, the discovery that mutations in genes coding for cardiac ion channels and their regulatory proteins can cause the congenital form of LQTS (cLQTS) shed new light on the pathophysiological mechanism of aLQTS. Mutations were shown to cause cLQTS by impairing the expression or function of ion channels important for the ventricular repolarization process.1 This created great interest among clinicians, geneticists, and pharmaceutical companies to search for genetic variants that may increase the risk of aLQTS. Indeed, mutations previously linked to cLQTS were also found in a small subset of patients with aLQTS, and, in addition, several common genetic variants (which have a higher allelic frequency in the general population than mutations) were associated with an increased risk of aLQTS.4,5 These findings were compatible with the concepts of ‘repolarization reserve’ and ‘second hit’, proposed at that time to imply the possibility that a combination of genetic and acquired factors may be required to impair repolarization to an extent large enough to cause QT prolongation (Figure 1).6,7 The molecular mechanism underlying congenital (cLQTS) and acquired long QT syndrome (aLQTS). The lower panel shows the distribution of QTc duration in health (green line), aLQTS (orange), and cLQTS (red line). The circles display the genetic architecture of the QTc duration, consisting of rare variants (mutations) with a large effect on QTc (red circles), common variants with an intermediate effect on QTc (orange circles), and very common variants with a small effect on QTc. The middle panels show that in aLQTS relatively fewer pathogenic mutations (e.g. mutations in non-pore regions of cardiac ion channels) or multiple common variants may cause mild QT prolongation at baseline and facilitate further QT prolongation upon exposure to acquired triggers. In comparison, in cLQTS, a disease-causing mutation in a transmembrane region of a cardiac ion channel may lead to clinically manifest QT prolongation at baseline, which is further prolonged upon exposure to acquired triggers. In this line of research, in this issue of the journal, a large international team headed by Dr Minoru Horie from Japan assessed the prevalence of mutations in five major cLQTS genes among 188 probands with aLQTS, most of them being female (75%) and Japanese (78%), with drugs as the most prevalent trigger (47%).8 They found possibly pathogenic mutations in 53 aLQTS cases (28%). Next, they compared their data with those from 2379 genotyped members of cLQTS families. Their main findings are that (i) aLQTS mutation carriers had longer mean heart rate-corrected QT intervals (QTc) in the absence of the triggering conditions (i.e. control QTc) than aLQTS patients without a relevant variant; (ii) control QTc in aLQTS mutation carriers was comparable with baseline QTc in cLQTS cases; (iii) aLQTS patients without a mutation had longer control QTc compared with healthy family members of cLQTS cases; and (iv) the distribution of genetic subtypes differed between aLQTS and cLQTS, with relatively more KCNH2 mutations in aLQTS cases and more KCNQ1 mutations in cLQTS cases. The latter is consistent with an earlier report that acute IKr blockade with erythromycin can unmask abnormal repolarization in cLQTS patients with a KCNH2 mutation but not a KCNQ1 mutation.9 Interestingly, the authors aim to distinguish true aLQTS from unmasked cLQTS in their aLQTS cohort by setting the border for QTc at 460 ms for women and 450 ms for men, assuming that patients have true aLQTS if their control QTc is <460 ms in females and <450 ms in males. Finally, by using a multivariate logistic regression model, they identified age <40 years at the time of exposure to acquired triggers, control QTc >440 ms for women and men, and a history of symptoms at the time of aLQTS diagnosis as independent predictors for carrying a mutation. By using these three factors as dichotomized variables, with the presence of each variable conferring 1 point (i.e. age <40 years; QTc >440 ms; history of cardiac events), the authors provide a clinical score to predict mutation carriership. In their cohort, of the aLQTS patients carrying a mutation, 88% had a score of 2 or 3 points, while only 3% of patients with a score of 0 or 1 point had a mutation. This justified the authors recommending genetic screening in aLQTS patients with a score of 2 or 3 to enable ‘cascade screening’ in their family members and thereby prevent adverse events. The study of Itoh et al. shows us, as is the case with so many other genetic studies, how complex the underlying molecular mechanism of both aLQTS and cLQTS can be. After the discovery of rare variants as disease-causing mutations in cLQTS, we soon realized that a mutation often fails to predict outcome, since many mutation carriers remained asymptomatic (incomplete penetrance), while some carriers only showed QT prolongation without cardiac events, while others experienced lethal arrhythmias at a young age (variable expressivity).10–12 In the last two decades, several factors have been identified as modifiers of the phenotypic manifestation of a disease-causing mutation, ranging from the acquired triggers previously related to aLQTS to common variants in genes involved in the cardiac electricity, including KCNQ1, KCNH2, KCNE1, and NOS1AP.13 We have also learned that not all mutations act in the same manner, with some being benign, some being disease-causing, and some being pathogenic only in the presence of a second trigger, and that this may, at least partially, depend on the type and location of the mutation involved. Due to these uncertainties regarding the final clinical outcome of a mutation, it is of utmost importance that, before drawing definite conclusions regarding the causative link between a mutation and a phenotype, we obtain detailed knowledge about the frequency of a variant in the general population, verify its association in independent cohorts, evaluate its co-segregation in families, and obtain experimental evidence for a mechanistic connection between the variant and the phenotype. Complying with these rules is important to avoid designation of some genetic variants falsely as mutations and thereby of some healthy individuals as mutation carriers. Since in the pathogenesis of aLQTS a growing role is being attributed to genetic factors, the same rules should also apply for aLQTS. In that case, in their study Horie and colleagues leave us with some unresolved matters. For example, the authors show that control QTc (i.e. after withdrawal of the acquired trigger) is comparable between aLQTS mutation carriers and cLQTS cases, suggesting that these aLQTS cases are in fact unmasked cLQTS patients. However, it remains unknown why they distinguish true aLQTS from unmasked cLQTS using a QTc cut-off value of 460 ms in women and 450 ms in men. In addition, it is not fully clear why the authors do not use these same cut-off values in their clinical prediction tool but instead use a cut-off value of 440 ms for both men and women. Vice versa, why do they not use a QTc cut-off value of 440 ms to distinguish true aLQTS from unmasked cLQTS, or 430 ms, which has been previously reported in this journal to be the optimal cut-off value to distinguish cLQTS family members with the familial mutation from those without?14 Anyway, using a cut-off value for QTc is tricky when one realizes that QTc in cLQTS patients can vary from normal (<440 ms) to severely prolonged (>500 ms) values.10–12,14 Another issue that deserves more clarity is why aLQTS patients with no mutation have a longer control QTc compared with healthy individuals. This may be, at least partially, because the authors limited their genetic search to 5 out of the 16 genes that so far have been linked to cLQTS,1 and did not also search for the presence of common variants associated with increased risk of aLQTS, e.g. variants in NOS1AP or genes encoding drug-metabolizing cytochrome P450 enzymes.15 Last, but not least, proposing a clinical tool to predict the genetic status in aLQTS is not without risk. It is understandable that such a tool is very attractive to identify patients at risk of aLQTS. However, the genetics of LQTS seems to have become far too complex to be predicted by such a simple tool. For example, consider a 35-year-old woman with depression, a QTc of 445 ms (prolonging to 460 ms with an antidepressant), and no history of cardiac symptoms; according to the proposed tool, this patient has a score of 2 and genetic testing is recommended. Imagine that an as yet unknown variant in KCNQ1 is identified. What should we do next? Should we withhold from this patient the antidepressant drug that mildly prolongs her QT interval but improves her depressive symptoms and perform, as recommended by Horie and colleagues, a ‘cascade screening’ in her asymptomatic family members? Alternatively, let us consider a 45-year-old man with atypical pneumonia. His QTc is 400 ms (prolonging to 440 ms after administration of erythromycin), he has experienced an unexplained syncope at the age of 40 years, and he has a first-degree relative who has experienced a syncope at the age of 14 years immediately after the ringing of an alarm clock. According to the proposed tool, this patient has a score of 1 and genetic testing is not recommended. These two examples illustrate that for a genetic entity such as LQTS, clinicians do not necessarily need prediction tools to treat their patients and prevent adverse events, but rather increasing knowledge of the mechanism of LQTS, help in recognizing the presence of potential risk factors in a timely fashion and choosing the most appropriate management strategy, and assistance in discussing benefits and drawbacks of genetic testing on an individual basis when such testing is indicated. Thus, in an era when the field of human genetics is continuously expanding and making us realize that we do not yet fully understand the genetic backgrounds of entities such as LQTS, we should be cautious in our interpretation of the genetic data obtained and in our recommendations to perform genetic screening. Instead, we should establish discussion groups such as Genetic Teams in which clinicians, geneticists, genetic counsellors, basic scientists, and pharmacologists take into consideration the phenotype, age, gender, family and drug history, co-morbidities, patient preferences, and therapeutic alternatives before proceeding to genetic testing. Thereafter, such Genetic Teams would be of great value to explore the functional effects of a genetic variant and its implications for risk stratification and therapy in a carrier and her/his family members. We acknowledge the support from the Netherlands CardioVascular Research Initiative, the Dutch Heart Foundation, Dutch Federation of University Medical Centres, the Netherlands Organisation for Health Research and Development and the Royal Netherlands Academy of Sciences (PREDICT; AAMW), and the Netherlands Heart Foundation (td/dekk/2388 2013T042; ASA). Conflict of interest: none declared.