Cognitive Symptoms in Cross-Sectional Parkinson Disease Cohort Evaluated by Human-in-the-Loop Machine Learning and Natural Language Processing.

Background and Objectives: Cognitive impairment is experienced by up to 80% of people with Parkinson disease (PD). Little is known regarding the subjective experience and frequency of bothersome cognitive problems across the range of disease duration as expressed directly in patients' own words. We describe the types and frequency of bothersome cognitive symptoms reported verbatim by patients with PD. Methods: Through the online Fox Insight study and the Parkinson Disease Patient Report of Problems, we asked patients with PD to self-report by keyboard entry up to five most bothersome problems and how these problems affect their functioning. Human-in-the-loop curation, natural language processing, and machine learning were used to categorize responses into 8 cognitive symptoms: memory, concentration/attention, cognitive slowing, language/word finding, mental alertness/awareness, visuospatial abilities, executive abilities/working memory, and cognitive impairment not otherwise specified. Associations between cognitive symptoms and demographic and disease-related variables were examined in our cross-sectional cohort using multivariate logistic regression. Results: Among 25,192 participants (55% men) of median age 67 years and 3 years since diagnosis (YSD), 8,001 (32%) reported a cognitive symptom at baseline. The 3 most frequently reported symptoms were memory (13%), language/word finding (12%), and concentration/attention (9%). Depression was significantly associated with bothersome cognitive problems in all domains except visuospatial abilities. Predictors of reporting any cognitive symptom in PD were depression (adjusted OR 1.5), increasing MDS-UPDRS Part II score (OR 1.4 per 10-point increment), higher education (OR 1.2 per year), and YSD 1, 2, 6-7, and 8-9 vs 0 YSD. Among individuals with at least one cognitive symptom, posterior cortical-related cognitive symptoms (i.e., visuospatial, memory, and language) were reported by 17% (n = 4325), frontostriatal-related symptoms (i.e., executive abilities, concentration/attention) by 7% (n = 1,827), and both by 14.2% (n = 1,020). Odds of reporting posterior cortical symptoms vs frontostriatal symptoms increased with age and MDS-UPDRS part II score, but not depression. Discussion: Nearly one-third of participants with PD, even early in the disease course, report cognitive symptoms as among their most bothersome problems. Online verbatim reporting analyzed by human-in-the-loop curation, natural language processing, and machine learning is feasible on a large scale and allows a detailed examination of the nature and distribution of cognitive symptoms in PD.

Longitudinal Cohort Study of Verbatim-Reported Postural Instability Symptoms as Outcomes for Online Parkinson's Disease Trials.

BACKGROUND: The Parkinson's Disease Patient Report of Problems (PD-PROP) captures the problems and functional impact that patients report verbatim. Online research participation and advances in language analysis have enabled longitudinal collection and classification of symptoms as trial outcomes. OBJECTIVE: Analyze verbatim reports longitudinally to examine postural-instability symptoms as 1) precursors of subsequent falling and 2) newly occurring symptoms that could serve as outcome measures in randomized controlled trials. METHODS: Problems reported by >25,000 PD patients in their own words were collected online in the Fox Insight observational study and classified into symptoms by natural language processing, clinical curation, and machine learning. Symptoms of gait, balance, falling, and freezing and associated reports of having fallen in the last month were analyzed over three years of longitudinal observation by a Cox regression model in a cohort of 8,287 participants. New onset of gait, balance, falling, and freezing symptoms was analyzed by Kaplan-Meier survival techniques in 4,119 participants who had not previously reported these symptoms. RESULTS: Classified verbatim symptoms of postural instability were significant precursors of subsequent falling among participants who were older, female, and had longer PD duration. New onset of symptoms steadily increased and informed sample size estimates for clinical trials to reduce the onset of these symptoms. CONCLUSION: The tools to analyze symptoms reported by PD patients in their own words and capacity to enroll large numbers of research participants online support the feasibility and statistical power for conducting randomized clinical trials to detect effects of therapeutic interventions.

Predictive Value of Verbatim Parkinson's Disease Patient-Reported Symptoms of Postural Instability and Falling.

BACKGROUND: Postural instability is an intractable sign of Parkinson's disease, associated with poor disease prognosis, fall risk, and decreased quality of life. OBJECTIVE: 1) Characterize verbatim reports of postural instability and associated symptoms (gait disorder, balance, falling, freezing, and posture), 2) compare reports with responses to three pre-specified questions from Part II of the Movement Disorder Society Unified Parkinson Disease Rating Scale (MDS-UPDRS), and 3) examine postural instability symptoms and MDS-UPDRS responses as predictors of future falls. METHODS: Fox Insight research participants reported their problems attributed to PD in their own words using the Parkinson Disease Patient Reports of Problems (PD-PROP). Natural language processing, clinical curation, and data mining techniques were applied to classify text into problem domains and clinically-curated symptoms. Baseline postural instability symptoms were mapped to MDS-UPDRS questions 2.11-2.13. T-tests and chi-square tests were used to compare postural instability reporters and non-reporters, and Cochran-Armitage trend tests were used to evaluate associations between PD-PROP and MDS-UPDRS responses; survival methods were utilized to evaluate the predictive utility of PD-PROP and MDS-UPDRS responses in time-to-fall analyses. RESULTS: Of participants within 10 years of PD diagnosis, 9,692 (56.0%) reported postural instability symptoms referable to gait unsteadiness, balance, falling, freezing, or posture at baseline. Postural instability symptoms were significantly associated with patient-reported measures from the MDS-UPDRS questions. Balance problems reported on PD-PROP and MDS-UPDRS 2.11-2.13 measures were predictive of future falls. CONCLUSION: Verbatim-reported problems captured by the PD-PROP and categorized by natural language processing and clinical curation and MDS-UPDRS responses predicted falls. The PD-PROP output was more granular than, and as informative as, the categorical responses.

Informativeness of Early Huntington Disease Signs about Gene Status.

BACKGROUND: The cohort-level risk of Huntington disease (HD) is related to the age and symptom level of the cohort, but this relationship has not been made precise. OBJECTIVE: To predict the evolving likelihood of carrying the Huntington disease (HD) gene for at-risk adults using age and sign level. METHODS: Using data from adults with early signs and symptoms of HD linked to information on genetic status, we use Bayes' theorem to calculate the probability that an undiagnosed individual of a certain age and sign level has an expanded CAG repeat. RESULTS: Both age and sign levels have substantial influence on the likelihood of HD onset, and the probability of eventual diagnosis changes as those at risk age and exhibit (or fail to exhibit) symptoms. For example, our data suggest that in a cohort of individuals age 26 with a Unified Huntington's Disease Rating Scale (UHDRS) motor score of 7-10 70% of them will carry the HD mutation. For individuals age 56, the same motor score suggests only a 40% chance of carrying the mutation. Early motor signs of HD, overall and the chorea subscore, were highly predictive of disease onset at any age. However, body mass index (BMI) and cognitive performance scores were not as highly predictive. CONCLUSIONS: These results suggest that if researchers or clinicians are looking for early clues of HD, it may be more foretelling to look at motor rather than cognitive signs. Application of similar approaches could be used with other adult-onset genetic conditions.

A comparison of treatment thresholds in two large Parkinson's disease clinical trial cohorts.

Deprenyl and Tocopherol Antioxidative Therapy of Parkinsonism (DATATOP) and Parkinson Research Examination of CEP-1347 Trial (PRECEPT) were two clinical trials of potential disease-modifying agents for Parkinson's disease that used the time to reaching disability sufficient to require dopaminergic therapy as the primary endpoint. To compare the thresholds for initiating dopaminergic treatment for Parkinson's disease between the two studies, conducted fifteen years apart. Baseline and 12-month endpoint characteristics for subjects in the placebo arms of the two studies were compared. DATATOP placebo subjects had slightly higher total Unified Parkinson's Disease Rating Scale (UPDRS) scores at baseline than PRECEPT placebo subjects (26.1 vs. 23.6, P = 0.03). Time to endpoint was not significantly different. Mean total UPDRS scores at endpoint among those subjects reaching endpoint by 12 months were 48.4 in DATATOP and 37.5 in PRECEPT (P < 0.0001). Baseline disease severity and time to disability requiring dopaminergic therapy were similar in the DATATOP and PRECEPT trials. The threshold for starting dopaminergic treatment was lower in PRECEPT than in the earlier DATATOP study. This may relate to changes in philosophies with respect to starting treatment for Parkinson's disease, but the factors underlying this change remain to be elucidated.

Cumulative burden of atherosclerotic risk genotypes and the age at onset of a first myocardial infarction: a case-only carriership approach.

BACKGROUND: Previously identified atherosclerotic genetic factors have been studied mostly in case-control studies and in nonuniform ethnic populations, whereas data on the cumulative contribution of genetic factors to an earlier onset of a first myocardial infarction (MI) are limited. We hypothesized that several genetic atherosclerotic single nucleotide polymorphisms (SNPs) may exert an additive effect on the earlier occurrence of coronary atherothrombotic disease after adjustment for clinical factors. METHODS: Eighteen atherosclerotic high-risk SNPs were selected based upon meta-analyses of 614 published reports, and were incorporated into a carriership model. Multivariate regression analysis was used to identify the independent contribution of selected genotypes to the age at onset of a first MI in a cohort of 814 white (n = 622) and nonwhite (n = 192) patients enrolled in the Thrombogenic Factors and Coronary Events Study. RESULTS: The analysis demonstrated that selected genotypes were significantly associated with an earlier occurrence of a first MI among white patients (an average of 0.6 year reduction per carried genotype; P = 0.027), whereas the contribution of genotypes to MI onset among nonwhite patients was not significant (an average of 0.7 year increase per carried genotype; P = 0.16), with a significant ethnic x genotype interaction effect (P = 0.02). CONCLUSIONS: Our findings suggest that currently identified atherosclerotic genetic factors confer an independent additive contribution to the earlier onset of coronary atherothrombotic disease among white patients. The lack of a significant association between these genotypes and outcome in other ethnic groups suggests that cardiovascular genetic risk should be studied directly in these populations.

Polymorphism in the cholesteryl ester transfer protein gene and the risk of early onset myocardial infarction among cigarette smokers.

BACKGROUND: Cigarette smoking and the common B1 allele of the TaqIB polymorphism have both been reported to be associated with increased cholesteryl ester transfer protein (CETP) activity and altered lipoprotein levels. Thus, it is possible that the combined presence of these two respective environmental and genetic factors may enhance cardiovascular risk. We hypothesized that susceptibility to early onset myocardial infarction (MI) among cigarette smokers may be related to the presence of TaqIB polymorphism in the CETP gene. METHODS: The age at onset of a first MI among current (n = 199), past (n = 345), and never (n = 270) smokers was related to the presence of the TaqIB1 and B2 alleles in a cohort of 814 first MI patients. RESULTS: Multivariate regression analysis demonstrated that cigarette smoking was associated with a significant increase in the risk for early onset MI only among carriers of the TaqIB1 allele: current smokers with the B1B1 and B1B2 genotypes displayed a respective 9.4 (P < 0.001) and 8.4 (P < 0.001) year reduction in the age at onset of a first MI compared with never smokers, and past smokers with these genotypes exhibited a respective 3.8 (P = 0.003) and 3.7 (P = 0.01) year reduction. By contrast, current and past smoking was not associated with a significant increase in the risk for early onset MI among B2B2 homozygotes (3.0 [P = 0.28] and 0.2 [P = 0.93] year reduction, respectively). The smoking x genotype interaction was statistically significant (P = 0.04). CONCLUSIONS: The current findings suggest that genetic factors may modify susceptibility to early onset MI among cigarette smokers.

Polymorphisms in the paraoxonase and endothelial nitric oxide synthase genes and the risk of early-onset myocardial infarction.

In young patients, the accumulative burden of traditional cardiovascular risk factors may not be as significant as in an older population. Genetic risk factors were suggested to have a role in the early development of myocardial infarction (MI). However, data about the association between polymorphisms in heart disease-related genes and the early onset of a first MI are limited. In the present study, age at onset of a first MI was related to individual single-nucleotide polymorphisms in each of 18 prespecified candidate genes in a cohort of 814 patients enrolled in the Thrombogenic Factors and Recurrent Coronary Events (THROMBO) Study. Multivariate regression analysis showed in patients who had the high-risk genotypes of paraoxonase 1 (PON1) Q192R and endothelial nitric oxide synthase (eNOS) E298D that ages at onset of a first MI were 1.8 (p = 0.02) and 3.5 years (p = 0.02) earlier than in noncarriers of the genotypes, respectively. Consistently, high-risk genotypes of the PON1 Q192R and eNOS E298D polymorphisms were significantly associated with onset of a first MI at age <50 years (adjusted odds ratio 1.70, p = 0.005, adjusted odds ratio 2.15, p = 0.01, respectively). In conclusion, our findings suggest that high-risk genotypes of the PON1 Q192R and eNOS E298D polymorphisms are independently associated with a significantly earlier occurrence of coronary events.

Polymorphism in the angiotensinogen gene, hypertension, and ethnic differences in the risk of recurrent coronary events.

The M235T mutation of the human angiotensinogen gene has been shown to be associated with elevated circulating angiotensinogen concentrations and essential hypertension. The frequencies of the 235T allele are significantly different in black and white subjects. We analyzed the independent contribution of the angiotensinogen M235T mutation to the development of recurrent coronary events (coronary-related death, nonfatal myocardial infarction, or unstable angina) in a cohort of 916 black (n=145) and white (n=771) postmyocardial infarction patients who were prospectively studied during an average follow-up of 28 months. The frequency of the 235T allele was significantly higher among black (82%) than among white (44%) patients (P<0.001). There was no evidence for Hardy-Weinberg disequilibrium. During follow-up, 41 cardiac events (28%) occurred in blacks and 197 (26%) in whites (P=0.49). Multivariate Cox proportional hazards regression analysis demonstrated that 235T homozygosity was independently associated with increased risk of coronary events among black (hazard ratio: 2.37; P=0.04) but not white (hazard ratio: 0.93; P=0.68) patients, with a significant ethnic-related interaction effect (P for the difference=0.04). Among hypertensive black patients, the TT genotype was associated with a 3.3-fold (P=0.02) increase in the risk of coronary events. Our findings suggest that homozygosity for the 235T mutation in the angiotensinogen gene is an independent risk factor for coronary events in black postmyocardial infarction patients. The presence of hypertension significantly augments the risk associated with this genetic mutation.

Atherosclerotic risk genotypes and recurrent coronary events after myocardial infarction.

The association of a group of prespecified atherosclerotic risk genotypes with recurrent coronary events (coronary-related death, nonfatal myocardial infarction, or unstable angina) was investigated in a cohort of 1,008 patients after infarction during an average follow-up of 28 months. We used a carrier-ship approach with time-dependent survivorship analysis to evaluate the average risk of each carried genotype. Contrary to expectation, the hazard ratio for recurrent coronary events per carried versus noncarried genotype was 0.89 (95% confidence interval 0.80 to 0.99, p = 0.03) after adjustment for relevant genetic, clinical, and environmental covariates. This hazard ratio, derived from the 7 prespecified genotypes, indicated an average 11% reduction in the risk of recurrent coronary events per carried versus noncarried genotype. At 1 year after hospital discharge, the cumulative probability of recurrent coronary events was 26% in those who carried < or =1, 20% for those with 2 to 4, and 13% for those with > or =5 of these genotypes (p = 0.02). This unexpected risk reversal is a likely consequence of changes in the mix of risk factors in pre- and postinfarction populations. In conclusion, this under appreciated, population-based, risk-reversal phenomenon may explain the inconsistent associations of genetic risk factors with outcome events in previous reports involving coronary populations with different risk attributes.

Percutaneous angioplasty and stenting of the superficial femoral artery.

OBJECTIVES: The objectives of this study were to examine factors predictive of success or failure after percutaneous angioplasty (PTA) and stenting (S) of the superficial femoral artery (SFA) and to compare the results of PTA/S with a contemporary group of patients treated with femoropopliteal bypass. METHODS: A database of patients undergoing PTA and/or S of the SFA between 1986 and 2004 was maintained. Intention-to-treat analysis was performed. Patients underwent duplex scanning follow-up at 1, 3, and every 6 months after the intervention. Angiograms were reviewed in all cases to assess lesion characteristics and preprocedure and postprocedure runoff. Results were standardized to current TransAtlantic Inter-Society Consensus (TASC) and Society for Vascular Surgery (SVS) criteria. Kaplan-Meier survival analyses were performed to assess time-dependent outcomes. Cox proportional hazard analyses were performed to assess factors associated with patient survival and treatment efficacy. RESULTS: Three hundred eighty total limbs underwent PTA/S in 329 patients (67% male, 33% female; average age, 65 years). Mean follow-up was 1.8 years from the date of initial intervention. Indications for intervention were claudication in 66%, rest pain in 16%, and tissue loss in 18%. Runoff at the tibial level was 2.1 +/- 0.8 patent vessels. Mean SVS ischemia grade was 3.1 (range, 1 to 5). TASC lesion grades were A (48%), B (18%), C (22%), and D (12%). Angioplasty alone was used in 63% of cases. Primary treatment failure (inability to cross lesion) was seen in 7% of patients. There was one periprocedural death. Primary patency rates were 86% at 3 months, 80% at 6 months, 75% at 12 months, 66% at 24 months, 60% at 36 months, 58% at 48 months, and 52% at 60 months. Assisted primary patency rates were slightly higher ( P = not significant). By Cox proportional hazards analysis, patency of PTA/S was associated with higher preoperative ankle/brachial index ( P = .016) and the performance of angioplasty only ( P = .011). Failed or occluded PTA/S was associated with TASC C ( P < .0001) and TASC D lesions ( P < .0001). Patient death was associated with the presence of congestive heart failure ( P = .003). Subgroup analysis revealed that primary patency rates are highly dependent on lesion type (A > B > C > D, P < .0001). PTA/S patency for TASC A and B lesions compared favorably to prosthetic and venous femoropopliteal bypass. Surgical bypass was superior to PTA/S for TASC C and D lesions. CONCLUSIONS: PTA and stenting of the SFA can be performed safely with excellent procedural success rates. Improved patency of these interventions was seen with increased ankle/brachial index and the performance of angioplasty only. Worse patency was seen with TASC C and TASC D lesions. Patency rates were strongly dependent on lesion type, and the results of angioplasty and stenting compared favorably with surgical bypass for TASC A and B lesions.