Patient stratification and treatment effects in diseases with disturbed cardiac function

Sammanfattning: Background: Cardiovascular disease characteristics are often measured using a combination of different measurement modalities. By combining information from these modalities using statistical modelling we can gain additional knowledge about these diseases. Patients with hereditary transthyretin amyloidosis (ATTRv amyloidosis) often present with heart rate disturbances and ventricular hypertrophy. Although ATTRv amyloidosis can lead to a thick ventricular myocardium and heart failure this can also be due to other pathologies e.g. hypertrophic cardiomyopathy (HCM), which may lead to misdiagnosis. Children with hereditary long QT syndrome (LQTS) carry the risk of lethal cardiac events linked to increased sympathetic activity. If the regulatory pattern, measured by heart rate variability (HRV), could differentiate between LQTS genotypes and healthy children during childhood, it may then establish a role for HRV in evaluating such patients. Statins is the most common pharmaceutical treatment for atherosclerosis. However, the effect of statins on coronary calcification has shown mixed results between studies.Aim: The overall aim of this thesis was to use statistically rigorous methodology to explore stratification of subgroups of patients and estimate treatment effects in diseases with cardiac involvement. The specific aims were: to evaluate discriminating features between ATTRv amyloidosis and HCM patients in different clinical examinations of their hearts; to evaluate the longitudinal growth pattern in HRV in children with LQTS; and, to estimate the short- and long-term treatment effects of statin treatment on coronary calcification in patients with coronary arteriosclerosis. Methods: In Study I we used classification and regression trees in order to create a clinical interpretable decision tree in order to differentiate between 35 ATTRv amyloidosis patients and 37 HCM patients based on features derived from echocardiography and ECG. In Study II a retrospective sample of 38 ATTRv amyloidosis patients, 41 HCM patients and 62 healthy controls were analyzed using k-means clustering and Random forest models in order to investigate similarities and differences in echocardiographic features and HRV. In Study III we calculated gender and treatment dependent age trends for spectral HRV features from longitudinal measurements from 70 children with LQTS (58 LQT1, 12 LQT2) and 65 healthy controls, using generalized additive mixed effects models. In Study IV data from two large clinical trials, in total 1585 patients with mild cardiac symptoms, was used to estimate the short- and long-term treatment effects of statin treatment on coronary calcification.Results: In study I we found that a decision tree using both ECG and echocardiographic measurements provided the best separation between HCM and ATTRv amyloidosis. In study II we found that HRV could be used for discriminating between these diseases and that abnormalities in HRV are related on ATTRv fibril type but uncommon in HCM. In study III we found that the age trend in HRV in LQTS-patients showed no clear differences between controls and patients different LQTS genotypes. In study IV we found that high dose statin treatment resulted in a dose dependent increase in calcium score at both short- and long-term follow up. At the long-term follow up, we found no evidence to suggest that the increased calcium score had resulted in increased cardiac events.Conclusions: The results from our studies could be used for differentiating HCM and ATTRv amyloidosis patients. The heterogeneity within these patient groups could further be characterized using features from echocardiography and HRV. Our data do not support evidence for a different age trend in HRV parameters in children with LQTS compared to controls of similar age. We also found that statin treatment increased calcium score on both short-and long-term follow up. Taken together the results and methods from this thesis may be used for future support in clinical decisions regarding patient stratification and knowledge of treatment effects.