Using patient self-reports to study heterogeneity of treatment effects in major depressive disorder.

BACKGROUNDS: Clinicians need guidance to address the heterogeneity of treatment responses of patients with major depressive disorder (MDD). While prediction schemes based on symptom clustering and biomarkers have so far not yielded results of sufficient strength to inform clinical decision-making, prediction schemes based on big data predictive analytic models might be more practically useful. METHOD: We review evidence suggesting that prediction equations based on symptoms and other easily-assessed clinical features found in previous research to predict MDD treatment outcomes might provide a foundation for developing predictive analytic clinical decision support models that could help clinicians select optimal (personalised) MDD treatments. These methods could also be useful in targeting patient subsamples for more expensive biomarker assessments. RESULTS: Approximately two dozen baseline variables obtained from medical records or patient reports have been found repeatedly in MDD treatment trials to predict overall treatment outcomes (i.e., intervention v. control) or differential treatment outcomes (i.e., intervention A v. intervention B). Similar evidence has been found in observational studies of MDD persistence-severity. However, no treatment studies have yet attempted to develop treatment outcome equations using the full set of these predictors. Promising preliminary empirical results coupled with recent developments in statistical methodology suggest that models could be developed to provide useful clinical decision support in personalised treatment selection. These tools could also provide a strong foundation to increase statistical power in focused studies of biomarkers and MDD heterogeneity of treatment response in subsequent controlled trials. CONCLUSIONS: Coordinated efforts are needed to develop a protocol for systematically collecting information about established predictors of heterogeneity of MDD treatment response in large observational treatment studies, applying and refining these models in subsequent pragmatic trials, carrying out pooled secondary analyses to extract the maximum amount of information from these coordinated studies, and using this information to focus future discovery efforts in the segment of the patient population in which continued uncertainty about treatment response exists.

Testing a machine-learning algorithm to predict the persistence and severity of major depressive disorder from baseline self-reports.

Heterogeneity of major depressive disorder (MDD) illness course complicates clinical decision-making. Although efforts to use symptom profiles or biomarkers to develop clinically useful prognostic subtypes have had limited success, a recent report showed that machine-learning (ML) models developed from self-reports about incident episode characteristics and comorbidities among respondents with lifetime MDD in the World Health Organization World Mental Health (WMH) Surveys predicted MDD persistence, chronicity and severity with good accuracy. We report results of model validation in an independent prospective national household sample of 1056 respondents with lifetime MDD at baseline. The WMH ML models were applied to these baseline data to generate predicted outcome scores that were compared with observed scores assessed 10-12 years after baseline. ML model prediction accuracy was also compared with that of conventional logistic regression models. Area under the receiver operating characteristic curve based on ML (0.63 for high chronicity and 0.71-0.76 for the other prospective outcomes) was consistently higher than for the logistic models (0.62-0.70) despite the latter models including more predictors. A total of 34.6-38.1% of respondents with subsequent high persistence chronicity and 40.8-55.8% with the severity indicators were in the top 20% of the baseline ML-predicted risk distribution, while only 0.9% of respondents with subsequent hospitalizations and 1.5% with suicide attempts were in the lowest 20% of the ML-predicted risk distribution. These results confirm that clinically useful MDD risk-stratification models can be generated from baseline patient self-reports and that ML methods improve on conventional methods in developing such models.

Anxious and non-anxious major depressive disorder in the World Health Organization World Mental Health Surveys.

BACKGROUND: To examine cross-national patterns and correlates of lifetime and 12-month comorbid DSM-IV anxiety disorders among people with lifetime and 12-month DSM-IV major depressive disorder (MDD). METHOD: Nationally or regionally representative epidemiological interviews were administered to 74 045 adults in 27 surveys across 24 countries in the WHO World Mental Health (WMH) Surveys. DSM-IV MDD, a wide range of comorbid DSM-IV anxiety disorders, and a number of correlates were assessed with the WHO Composite International Diagnostic Interview (CIDI). RESULTS: 45.7% of respondents with lifetime MDD (32.0-46.5% inter-quartile range (IQR) across surveys) had one of more lifetime anxiety disorders. A slightly higher proportion of respondents with 12-month MDD had lifetime anxiety disorders (51.7%, 37.8-54.0% IQR) and only slightly lower proportions of respondents with 12-month MDD had 12-month anxiety disorders (41.6%, 29.9-47.2% IQR). Two-thirds (68%) of respondents with lifetime comorbid anxiety disorders and MDD reported an earlier age-of-onset (AOO) of their first anxiety disorder than their MDD, while 13.5% reported an earlier AOO of MDD and the remaining 18.5% reported the same AOO of both disorders. Women and previously married people had consistently elevated rates of lifetime and 12-month MDD as well as comorbid anxiety disorders. Consistently higher proportions of respondents with 12-month anxious than non-anxious MDD reported severe role impairment (64.4 v. 46.0%; χ 2 1 = 187.0, p < 0.001) and suicide ideation (19.5 v. 8.9%; χ 2 1 = 71.6, p < 0.001). Significantly more respondents with 12-month anxious than non-anxious MDD received treatment for their depression in the 12 months before interview, but this difference was more pronounced in high-income countries (68.8 v. 45.4%; χ 2 1 = 108.8, p < 0.001) than low/middle-income countries (30.3 v. 20.6%; χ 2 1 = 11.7, p < 0.001). CONCLUSIONS: Patterns and correlates of comorbid DSM-IV anxiety disorders among people with DSM-IV MDD are similar across WMH countries. The narrow IQR of the proportion of respondents with temporally prior AOO of anxiety disorders than comorbid MDD (69.6-74.7%) is especially noteworthy. However, the fact that these proportions are not higher among respondents with 12-month than lifetime comorbidity means that temporal priority between lifetime anxiety disorders and MDD is not related to MDD persistence among people with anxious MDD. This, in turn, raises complex questions about the relative importance of temporally primary anxiety disorders as risk markers v. causal risk factors for subsequent MDD onset and persistence, including the possibility that anxiety disorders might primarily be risk markers for MDD onset and causal risk factors for MDD persistence.

The effects of co-morbidity in defining major depression subtypes associated with long-term course and severity.

BACKGROUND: Although variation in the long-term course of major depressive disorder (MDD) is not strongly predicted by existing symptom subtype distinctions, recent research suggests that prediction can be improved by using machine learning methods. However, it is not known whether these distinctions can be refined by added information about co-morbid conditions. The current report presents results on this question. METHOD: Data came from 8261 respondents with lifetime DSM-IV MDD in the World Health Organization (WHO) World Mental Health (WMH) Surveys. Outcomes included four retrospectively reported measures of persistence/severity of course (years in episode; years in chronic episodes; hospitalization for MDD; disability due to MDD). Machine learning methods (regression tree analysis; lasso, ridge and elastic net penalized regression) followed by k-means cluster analysis were used to augment previously detected subtypes with information about prior co-morbidity to predict these outcomes. RESULTS: Predicted values were strongly correlated across outcomes. Cluster analysis of predicted values found three clusters with consistently high, intermediate or low values. The high-risk cluster (32.4% of cases) accounted for 56.6-72.9% of high persistence, high chronicity, hospitalization and disability. This high-risk cluster had both higher sensitivity and likelihood ratio positive (LR+; relative proportions of cases in the high-risk cluster versus other clusters having the adverse outcomes) than in a parallel analysis that excluded measures of co-morbidity as predictors. CONCLUSIONS: Although the results using the retrospective data reported here suggest that useful MDD subtyping distinctions can be made with machine learning and clustering across multiple indicators of illness persistence/severity, replication with prospective data is needed to confirm this preliminary conclusion.

Statistical epistasis and progressive brain change in schizophrenia: an approach for examining the relationships between multiple genes.

Although schizophrenia is generally considered to occur as a consequence of multiple genes that interact with one another, very few methods have been developed to model epistasis. Phenotype definition has also been a major challenge for research on the genetics of schizophrenia. In this report, we use novel statistical techniques to address the high dimensionality of genomic data, and we apply a refinement in phenotype definition by basing it on the occurrence of brain changes during the early course of the illness, as measured by repeated magnetic resonance scans (i.e., an 'intermediate phenotype.') The method combines a machine-learning algorithm, the ensemble method using stochastic gradient boosting, with traditional general linear model statistics. We began with 14 genes that are relevant to schizophrenia, based on association studies or their role in neurodevelopment, and then used statistical techniques to reduce them to five genes and 17 single nucleotide polymorphisms (SNPs) that had a significant statistical interaction: five for PDE4B, four for RELN, four for ERBB4, three for DISC1 and one for NRG1. Five of the SNPs involved in these interactions replicate previous research in that, these five SNPs have previously been identified as schizophrenia vulnerability markers or implicate cognitive processes relevant to schizophrenia. This ability to replicate previous work suggests that our method has potential for detecting a meaningful epistatic relationship among the genes that influence brain abnormalities in schizophrenia.

Development of real-time PCR assays for detection of the Streptococcus milleri group from cystic fibrosis clinical specimens by targeting the cpn60 and 16S rRNA genes.

Cystic fibrosis (CF) is a multiorgan disease, with the majority of mortalities resulting from pulmonary failure due to repeated pulmonary exacerbations. Recently, members of the Streptococcus anginosus group (S. anginosus, S. constellatus, and S. intermedius), herein referred to as the "Streptococcus milleri group" (SMG) have been implicated as important etiological pathogens contributing to pulmonary exacerbations in CF patients. This is partly due to better microbiological detection of the SMG species through the development of a novel specific medium termed "McKay agar." McKay agar demonstrated that SMG has been an underreported respiratory pathogen contributing to lung exacerbations. Our aim was to develop a real-time PCR assay to expedite the detection of SMG within diagnostic samples. The cpn60 gene was chosen as a target, with all three members amplified using a single hybridization probe set. SMG strain analysis showed that speciation based on melting curve analysis allowed for the majority of the S. constellatus (96%), S. intermedius (94%), and S. anginosus (60%) strains to be correctly identified. To increase specificity for S. anginosus, two 16S rRNA real-time PCR assays were developed targeting the 16S rRNA gene. The 16s_SA assay is specific for S. anginosus (100%), while the 16s_SCI assay is specific for S. constellatus and S. intermedius (100%). These assays can detect <10 genome equivalents in pure culture and >10(4) genome equivalents in sputum samples, making this a great tool for assessment of the presence of SMG in complex polymicrobial samples. Novel molecular methods were developed providing detection ability for SMG, an emerging opportunistic pathogen.

Comparison of the QTDT analysis for IgE in the CSGA data set.

Over the past few years at least 13 transmission/disequilibrium test (TDT)-based tests have been developed for quantitative (Q) traits for the assessment of association or linkage in the presence of the other. A total of six of these QTDT methods were used to analyze log10IgE in the Collaborative Study on the Genetics of Asthma data set. Only moderate agreement was found between the tests. The results of the QTDT analyses were only slightly affected by the use of gender and age as covariates. Results from analysis of IgE and log10IgE were inconsistent. Our conclusion is that there is only modest agreement among the QTDT methods examined, covariates should be used even if they have a small effect, and that data should be normalized before analysis.

Reduced fetal movements and maternal medication - new pregnancy risk factors for neurodevelopmental disability in childhood.

A case-control study was undertaken of 471 children on the Nottingham Special Needs Register (SNR) who were born in one of the two maternity units in the city between 1987 and 1993 (inclusive). Controls were selected as the next infant born at the same hospital following each index case. The aim of the study was to identify risk factors on the Nottingham Obstetric Database for a baby subsequently appearing on the SNR. Disability was analysed by both ICD-9 coding and functional assessment. Factors which independently and significantly predicted a child's likelihood of being on the SNR were breech presentation (adjusted odds ratio (OR) = 4.0), congenital abnormality (OR=4.9), intrapartum fetal distress (OR=1.7), fetal growth restriction (OR=2.0), socioeconomic deprivation (OR=1.8), prematurity (OR=2.2), reduced fetal movements (OR=2.5) and medication in pregnancy (OR=10.4). To our knowledge the last two factors have not previously been reported as risk predictors for neurodevelopmental disability.

Using recursive partitioning for exploration and follow-up of linkage and association analyses.

We first conducted a genome-wide screen for association with discordant sibships using the multi allelic and diallelic SDT. Markers at D4S1628, D8S1109, D9S66 and D7S1797 showed multi-allelic association. Deleterious diallelic association was found for markers at D1S1613, D1S534, D3S2459, D7S1817, and D9S131. Protective association was found at markers D8S1109, D8S1136, and D9S66. We then incorporated these findings with previous linkage findings in the exploration of oligogenes and epistasis using recursive partitioning. We conclude that recursive partitioning can be a useful adjunct to traditional linkage and association analyses in the exploration of these effects.

Alpha-2 macroglobulin is genetically associated with Alzheimer disease.

Alpha-2-macroglobulin (alpha-2M; encoded by the gene A2M) is a serum pan-protease inhibitor that has been implicated in Alzheimer disease (AD) based on its ability to mediate the clearance and degradation of A beta, the major component of beta-amyloid deposits. Analysis of a deletion in the A2M gene at the 5' splice site of 'exon II' of the bait region (exon 18) revealed that inheritance of the deletion (A2M-2) confers increased risk for AD (Mantel-Haenzel odds ratio=3.56, P=0.001). The sibship disequilibrium test (SDT) also revealed a significant association between A2M and AD (P=0.00009). These values were comparable to those obtained for the APOE-epsilon4 allele in the same sample, but in contrast to APOE-epsilon4, A2M-2 did not affect age of onset. The observed association of A2M with AD did not appear to account for the previously published linkage of AD to chromosome 12, which we were unable to confirm in this sample. A2M, LRP1 (encoding the alpha-2M receptor) and the genes for two other LRP ligands, APOE and APP (encoding the amyloid beta-protein precursor), have now all been genetically linked to AD, suggesting that these proteins may participate in a common neuropathogenic pathway leading to AD.

The effect of different sampling intervals on the measurement of intrapartum fetal heart rate variability.

OBJECTIVE: To test the hypothesis that increasing the sampling interval affects the intrapartum fetal heart rate (FHR) variability measurement. METHODS: Fetal electrocardiograms were obtained from women in labor. Using the peak of the fetal R wave, the R-R interval and FHR were calculated on a beat-to-beat basis. Retrospectively, the original data were repartitioned using different intervals (2-900 seconds) to generate a window of measurement (epoch). The mean value for each epoch and the last FHR in that epoch (epochal value) were compared with published animal and human data. Errors were quantified by comparing the epochal and mean values for each epoch. Fetal heart rate variability between epochs and within each epoch was compared. RESULTS: Fetal heart rate and R-R interval were measured in 146 cases. The FHR had a normal distribution (mean 140.1 beats per minute, +/- standard deviation [SD] 15.6, skew -0.07), but its inverse, the R-R interval, was not normally distributed (mean 432 milliseconds, +/- SD 52.4, skew 1.78). Using a single value for an epoch duration of 2 seconds resulted in an error that was similar to the within-epoch variability (+/- SD of 2.2 beats per minute difference between mean and epochal value compared to +/- SD of 2 beats per minute within epoch) but which increased with epoch duration. CONCLUSION: An epoch duration of 2 seconds and a single sampled value within this period may be appropriate for measurement of both medium and long-term variability in any computerized intrapartum FHR interpretation system. Fetal heart rate (not R-R interval, because of its normal distribution) should be used to design such a computerized system.

The effects of parity on birthweight using successive pregnancies.

OBJECTIVE: Accurate assessment of the difference in birthweight between first and second live-births to the same woman having excluded the effects of physiological factors known to affect birthweight. DESIGN: Retrospective longitudinal observational study. SETTING: Three large obstetric units in the East Midlands of the United Kingdom. SUBJECTS AND METHODS: Women in whom data were recorded for their first two pregnancies on the UK East Midlands Obstetric database which resulted in the delivery of a liveborn, singleton and congenitally normal baby. Six thousand five hundred and thirty such cases were identified, of which 3457 had complete datasets and delivered both babies at term (259 to 300 days). An analysis was performed of changes between the paired pregnancies of physiological factors known to affect birthweight. Regression analyses were used to enable prediction of the second birthweight with the knowledge of the first birthweight. RESULTS: The mean crude birthweight difference between first and second pregnancies was an increase of 138 g. Significant differences between the paired pregnancies were found in maternal booking visit weight, blood pressure, maternal age and gestation at delivery. Independent factors affecting difference in birthweight were gestation at delivery, maternal booking weight and baby's sex. Regression towards the mean was demonstrated which meant that a woman delivering a first baby weighing more than 3720 g could expect a lighter baby for her second delivery provided that all other factors remained constant. CONCLUSIONS: In general terms a woman is more likely to deliver a heavier baby in her second pregnancy than in her first pregnancy. However, maternal physiological factors differ in the two pregnancies and these differences have additional effects on birthweight. The effects of both these observations are tempered by regression towards the mean which has a profound influence in predicting the likely change in birthweight between first and second pregnancies. Clinical decisions should not be based on the assumption that a second baby will inevitably be heavier than the first baby.

The effect of social deprivation on birthweight, excluding physiological and pathological effects.

OBJECTIVE: To study the effect of social deprivation on birthweight, excluding the effect of known physiological factors and exploring the effect of possible pathological factors. DESIGN: Retrospective analysis of computerised obstetric database. SETTING: Two teaching hospitals and an associated district general hospital which provided a defined catchment area in the East Midlands. SUBJECTS: The final analysis included 7493 women with complete datasets and gestations of between 259 and 300 days at delivery, dated by ultrasound scan. MAIN OUTCOME MEASURES: Smoking habit, alcohol consumption, weight gain during pregnancy, systolic and diastolic blood pressures at booking, bleeding during pregnancy and Jarman score; also, the effect of these variables on birthweight, adjusted for the effects of physiological factors using the individualised birthweight ratio. RESULTS: Smoking during pregnancy reduced birthweight but the effect is not linear, becoming less marked as the number of cigarettes smoked increases. Alcohol intake, diastolic and systolic blood pressures at the booking visit and vaginal bleeding during early pregnancy were not significantly related to birthweight. Pregnancy weight gain was significantly positively related to birthweight especially in the normal weight range (60-99 kg). A multivariate analysis including physiological and pathological factors found increasing Jarman score to be negatively related to birthweight. CONCLUSIONS: In this central British population social deprivation is correlated negatively with birthweight: the most socially deprived mothers have the smallest babies. This association cannot be explained in terms of physiological differences in the population nor in a higher prevalence of known pathological factors.

Paternal influences on birthweight.

OBJECTIVE: To assess the influence of paternal size on birthweight after suitable control for maternal and fetal factors. DESIGN: Prospective observational study. SETTING: Delivery suite, City Hospital, Nottingham. SUBJECTS: 571 husbands/partners of unselected women delivering August 1992 to February 1993. MAIN OUTCOME METHODS: Individualised birthweight ratio and thereby an adjusted birthweight for a typical mother. The results of a multiple regression analysis with the individualised birthweight ratio as the dependent variable. RESULTS: When considered in isolation both paternal height and weight are significantly positively associated with crude and adjusted birthweight (p < 0.01, analysis of variance). Due to correlations of paternal size with maternal size and smoking habit, only paternal height is significant in the multiple regression analysis (p = 0.01). CONCLUSION: If the partner of an average woman is short (mean-2s.d.) then the baby will be 183 g lighter than if he is tall (mean + 2s.d.). This effect of paternal height on birthweight must be genetic and therefore should be taken into account when defining intra-uterine growth retardation and macrosomia.

The individualised birthweight ratio: a new method of identifying intrauterine growth retardation.

OBJECTIVE: To assess the effectiveness of the newly developed individualised birthweight ratio in identifying growth retarded infants. DESIGN: Prospective observational study. SETTING: Obstetric unit, City Hospital Nottingham. SUBJECTS: Two thousand eight hundred and thirty-five women delivered between December 1991 and July 1992 and the infants of 616 of these selected by virtue of their birthweight for gestation and individualised birthweight ratio centile positions. MAIN OUTCOME MEASURES: Skinfold thickness and ponderal index measurements, occurrence of abnormal fetal heart rate patterns, operative delivery due to fetal distress and the need for neonatal resuscitation. RESULTS: Using an individualised birthweight ratio less than the 10th centile as a cut-off results in 25% of those less than the 10th centile of birthweight for gestation being reclassified as normally grown. A slightly larger group are reclassified as small; significantly more of these infants have abnormal ponderal indices and skinfold thicknesses, suffer abnormal fetal heart rate patterns, operative delivery for fetal distress and need neonatal resuscitation than do those who are reclassified as normally grown. CONCLUSION: The individualised birthweight ratio combines the simplicity of birthweight measurement with the accuracy of clinical measurements in the identification of the growth retarded baby.

Relative macrosomia identified by the individualised birthweight ratio (IBR). A better method of identifying the at risk fetus.

OBJECTIVE: To assess the effectiveness of a newly developed individualised birthweight ratio (IBR), which corrects for physiological birthweight determinants, in identifying infants at risk from the complications of macrosomia. DESIGN: Prospective observational study. SETTING: Obstetric unit, Nottingham City Hospital. SUBJECTS: 2835 women delivered between December 1991 and July 1992 and the infants of 624 of these, selected by virtue of their birthweight for gestation and IBR centile positions. MAIN OUTCOME MEASURES: Skinfold thickness and ponderal index measurements, operative delivery, shoulder dystocia, fetal trauma, impaired glucose tolerance. RESULTS: Using an IBR above the 90th centile as a cut off results in 2.4% of infants being reclassified as normally grown and 3.1% are reclassified as large. The IBR does not result in the identification of any more infants with abnormal ponderal indices or skinfold thicknesses than birthweight for gestation. It does, however, identify more of the infants at risk of operative delivery, shoulder dystocia, fetal trauma and impaired glucose tolerance. CONCLUSION: The IBR significantly improves upon birthweight for gestation in identifying infants who suffer from the complications of relative macrosomia.

The individualised birthweight ratio: a more logical outcome measure of pregnancy than birthweight alone.

OBJECTIVE: To provide a new outcome measure for pregnancy specifically related to the individual. DESIGN: Computer analysis of physiological factors affecting birthweight. SETTING: Two provincial teaching hospitals (University and City Hospitals, Nottingham) and an associated district general hospital (Derby City Hospital) serving a defined catchment area in the East Midlands. SUBJECTS: All women delivering in the above hospitals since the start of computerised obstetric records: 31,561 women with gestational age verified by early pregnancy ultrasound scan data. MAIN OUTCOME MEASURES: Calculation of the predicted birthweight taking into account maternal and fetal physiological factors. Derivation of the individualised birthweight ratio (actual birthweight divided by predicted birthweight expressed as a percentage) for each individual baby. RESULTS: The individualised birthweight ratio redefines as normally grown 41% of babies below the 10th centile of crude birthweight for gestation. Other babies previously regarded as normal are redefined as growth retarded. At the upper end of the distribution 46% of those above the 90th centile of birthweight for gestation are redefined as normally grown. CONCLUSIONS: The predicted birthweight can be calculated for an individual pregnancy at a given gestation. The standardised comparison between this predicted birthweight and the actual birthweight is a more logical reflection of the normality of intrauterine growth and therefore more logical as an outcome measure for pregnancy than crude birthweight for gestation.

Stimulation of hepatic microsomal beta-glucuronidase by calcium.

Hydrolysis of 3-methylumbelliferyl glucuronide by liver microsomal beta-glucuronidase is enhanced about 2-fold by micromolar concentrations of Ca2+; half-maximal stimulation occurs with 0.35 microM Ca2+. Dissociation of the enzyme from microsomal membranes by various treatments increases basal beta-glucuronidase activity and markedly decreases the sensitivity of the enzyme to Ca2+. Under similar conditions, the soluble lysosomal form of the enzyme is insensitive to Ca2+. Ca2+ stimulation was unaltered by addition of calmodulin inhibitors or exogenous calmodulin. Thus, interaction of cytosolic Ca2+ with membrane bound beta-glucuronidase may modulate glucuronidation in intact hepatocytes via a novel, calmodulin-independent mechanism.