On the assessment of the added value of new predictive biomarkers

The Impact of Edema on MRI Radiomics for the Prediction of Lung Metastasis in Soft Tissue Sarcoma

INTRODUCTION

This study aimed to evaluate whether radiomic features extracted solely from the edema of soft tissue sarcomas (STS) could predict the occurrence of lung metastasis in comparison with features extracted solely from the tumoral mass.

MATERIALS AND METHODS

We retrospectively analyzed magnetic resonance imaging (MRI) scans of 32 STSs, including 14 with lung metastasis and 18 without. A segmentation of the tumor mass and edema was assessed for each MRI examination. A total of 107 radiomic features were extracted for each mass segmentation and 107 radiomic features for each edema segmentation. A two-step feature selection process was applied. Two predictive features for the development of lung metastasis were selected from the mass-related features, as well as two predictive features from the edema-related features. Two Random Forest models were created based on these selected features; 100 random subsampling runs were performed. Key performance metrics, including accuracy and area under the ROC curve (AUC), were calculated, and the resulting accuracies were compared.

RESULTS

The model based on mass-related features achieved a median accuracy of 0.83 and a median AUC of 0.88, while the model based on edema-related features achieved a median accuracy of 0.75 and a median AUC of 0.79. A statistical analysis comparing the accuracies of the two models revealed no significant difference.

CONCLUSION

Both models showed promise in predicting the occurrence of lung metastasis in soft tissue sarcomas. These findings suggest that radiomic analysis of edema features can provide valuable insights into the prediction of lung metastasis in soft tissue sarcomas.

Height Loss in Old Age and Fracture Risk Among Men in Late Life: A Prospective Cohort Study

To assess the association of height loss in old age with subsequent risk of hip and any clinical fracture in men late in life while accounting for the competing risk of mortality, we used data from 3491 community-dwelling men (mean age 79.2 years). Height loss between baseline and follow-up (mean 7.0 years between examinations) was categorized as <1 cm (referent group), ≥1 to <2 cm, ≥2 to <3 cm, and ≥3 cm. Men were contacted every 4 months after the follow-up examination to ask about fractures (confirmed by radiographic reports) and ascertain vital status (deaths verified by death certificates). Competing risk methods were used to estimate absolute probabilities of fracture outcomes by height loss category and calculate adjusted risks of fracture outcomes by height loss. During an average of 7.8 years, 158 (4.5%) men experienced a hip fracture and 1414 (40.5%) died before experiencing this event. The absolute 10-year probability of fracture events accounting for the competing risk of death increased with greater height loss. For example, the hip fracture probability was 2.7% (95% confidence interval [CI] 1.9-3.8%) among men with height loss <1 cm increasing to 11.6% (95% CI 8.0-16.0%) among men with height loss ≥3 cm. After adjustment for demographics, fall history, multimorbidity, baseline height, weight change, and femoral neck bone mineral density and considering competing mortality risk, men with height loss ≥3 cm versus <1 cm had a nearly twofold (subdistribution hazard ratio [HR] = 1.94, 95% CI 1.06-3.55) higher risk of hip fracture and a 1.4-fold (subdistribution HR = 1.42, 95% CI 1.05-1.91) increased risk of any clinical fracture. Height loss ≥3 cm in men during old age was associated with higher subsequent risk of clinical fractures, especially hip fractures, even after accounting for the competing risk of death and traditional skeletal and non-skeletal risk factors. © 2021 American Society for Bone and Mineral Research (ASBMR).

Joint Associations of Prevalent Radiographic Vertebral Fracture and Abdominal Aortic Calcification With Incident Hip, Major Osteoporotic, and Clinical Vertebral Fractures

Prevalent vertebral fractures (PVFx) and abdominal aortic calcification (AAC) are both associated with incident fractures and can be ascertained on the same lateral spine images, but their joint association with incident fractures is unclear. Our objective was to estimate the individual and joint associations of PVFx and AAC with incident major osteoporotic, hip, and clinical vertebral fractures in 5365 older men enrolled in the Osteoporotic Fractures in Men (MrOS) Study, using Cox proportional hazards and Fine and Gray subdistribution hazards models to account for competing mortality. PVFx (Genant SQ grade 2 or 3) and 24-point AAC score were ascertained on baseline lateral spine radiographs. Self-reports of incident fractures were solicited every 4 months and confirmed by review of clinical radiographic reports. Compared with men without PVFx and AAC-24 score 0 or 1, the subdistribution hazard ratio (SHR) for incident major osteoporotic fracture was 1.38 (95% confidence interval [CI] 1.13-1.69) among men with AAC-24 score ≥2 alone, 1.71 (95% CI 1.37-2.14) for men with PVFx alone, and 2.35 (95% CI 1.75-3.16) for men with both risk factors, after accounting for conventional risk factors and competing mortality. Wald statistics showed improved prediction model performance by including both risk factors compared with including only AAC (chi-square = 17.3, p < .001) or including only PVFx (chi-square = 8.5, p = .036). Older men with both PVFx and a high level of AAC are at higher risk of incident major osteoporotic fracture than men with either risk factor alone. Assessing prevalent radiographic vertebral fracture and AAC on the same lateral spine images may improve prediction of older men who will have an incident major osteoporotic fracture, even after accounting for traditional fracture risk factors and competing mortality. © 2021 American Society for Bone and Mineral Research (ASBMR).

Texture analysis using machine learning-based 3-T magnetic resonance imaging for predicting recurrence in breast cancer patients treated with neoadjuvant chemotherapy

OBJECTIVES

To determine whether texture analysis for magnetic resonance imaging (MRI) can predict recurrence in patients with breast cancer treated with neoadjuvant chemotherapy (NAC).

METHODS

This retrospective study included 130 women who received NAC and underwent subsequent surgery for breast cancer between January 2012 and August 2017. We assessed common features, including standard morphologic MRI features and clinicopathologic features. We used a  commercial software and analyzed texture features from pretreatment and midtreatment MRI. A random forest (RF) method was performed to build a model for predicting recurrence. The diagnostic performance of this model for predicting recurrence was assessed and compared with those of five other machine learning classifiers using the Wald test.

RESULTS

Of the 130 women, 21 (16.2%) developed recurrence at a median follow-up of 35.4 months. The RF classifier with common features including clinicopathologic and morphologic MRI features showed the lowest diagnostic performance (area under the receiver operating characteristic curve [AUC], 0.83). The texture analysis with the RF method showed the highest diagnostic performances for pretreatment T2-weighted images and midtreatment DWI and ADC maps showed better diagnostic performance than that of an analysis of common features (AUC, 0.94 vs. 0.83, p < 0.05). The RF model based on all sequences showed a better diagnostic performance for predicting recurrence than did the five other machine learning classifiers.

CONCLUSIONS

Texture analysis using an RF model for pretreatment and midtreatment MRI may provide valuable prognostic information for predicting recurrence in patients with breast cancer treated with NAC and surgery.

KEY POINTS

• RF model-based texture analysis showed a superior diagnostic performance than traditional MRI and clinicopathologic features (AUC, 0.94 vs.0.83, p < 0.05) for predicting recurrence in breast cancer after NAC. • Texture analysis using RF classifier showed the highest diagnostic performances (AUC, 0.94) for pretreatment T2-weighted images and midtreatment DWI and ADC maps. • RF model showed a better diagnostic performance for predicting recurrence than did the five other machine learning classifiers.

Beta-catenin activation and immunotherapy resistance in hepatocellular carcinoma: mechanisms and biomarkers

Mutations involving CTNNB1, the gene encoding beta-catenin, and other molecular alterations that affect the Wnt/beta-catenin signaling pathway are exceptionally common in hepatocellular carcinoma. Several of these alterations have also been associated with scarcity of immune cells in the tumor microenvironment and poor clinical response to immune checkpoint inhibitor therapy. In light of these associations, tumor biomarkers of beta-catenin status could have the potential to serve as clinical predictors of immunotherapy outcome. This editorial review article summarizes recent pre-clinical and clinical research pertaining to associations between beta-catenin activation and diminished anti-tumor immunity. Potential non-invasive biomarkers that may provide a window into this oncogenic mechanism of immune evasion are also presented and discussed.

A Decision Aide for the Risk Stratification of GU Cancer Patients at Risk of SARS-CoV-2 Infection, COVID-19 Related Hospitalization, Intubation, and Mortality

Treatment decisions for both early and advanced genitourinary (GU) malignancies take into account the risk of dying from the malignancy as well as the risk of death due to other causes such as other co-morbidities. COVID-19 is a new additional and immediate risk to a patient’s morbidity and mortality and there is a need for an accurate assessment as to the potential impact on of this syndrome on GU cancer patients. The aim of this work was to develop a risk tool to identify GU cancer patients at risk of diagnosis, hospitalization, intubation, and mortality from COVID-19. A retrospective case showed a series of GU cancer patients screened for COVID-19 across the Mount Sinai Health System (MSHS). Four hundred eighty-four had a GU malignancy and 149 tested positive for SARS-CoV-2. Demographic and clinical variables of >38,000 patients were available in the institutional database and were utilized to develop decision aides to predict a positive SARS-CoV-2 test, as well as COVID-19-related hospitalization, intubation, and death. A risk tool was developed using a combination of machine learning methods and utilized BMI, temperature, heart rate, respiratory rate, blood pressure, and oxygen saturation. The risk tool for predicting a diagnosis of SARS-CoV-2 had an AUC of 0.83, predicting hospitalization for management of COVID-19 had an AUC of 0.95, predicting patients requiring intubation had an AUC of 0.97, and for predicting COVID-19-related death, the risk tool had an AUC of 0.79. The models had an acceptable calibration and provided a superior net benefit over other common strategies across the entire range of threshold probabilities.

Microtransplantation in older patients with AML: A pilot study of safety, efficacy and immunologic effects

Older AML patients have low remission rates and poor survival outcomes with standard chemotherapy. Microtransplantation (MST) refers to infusion of allogeneic hematopoietic stem cells without substantial engraftment. MST has been shown to improve clinical outcomes compared with chemotherapy alone. This is the first trial reporting on broad correlative studies to define immunologic mechanisms of action of MST in older AML patients. Older patients with newly diagnosed AML were eligible for enrollment, receiving induction chemotherapy with cytarabine (100 mg/m2) on days 1-7 and idarubicin (12 mg/m2) on days 1-3 (7 + 3). MST was administered 24 hours later. Patients with complete response (CR) were eligible for consolidation with high dose cytarabine (HiDAC) and a second cycle of MST. Responses were evaluated according to standard criteria per NCCN. Immune correlative studies were performed. Sixteen patients were enrolled and received 7 + 3 and MST (median age 73 years). Nine (56%) had high-risk and seven (44%) had standard-risk cytogenetics. Ten episodes of CRS were observed. No cases of GVHD or treatment-related mortality were reported. Event-free survival (EFS) was 50% at 6 months and 19% at 1 year. Overall survival (OS) was 63% at 6 months and 44% at 1 year. Donor microchimerism was not detected. Longitudinal changes were noted in NGS, TCR sequencing, and cytokine assays. Addition of MST to induction and consolidation chemotherapy was well tolerated in older AML patients. Inferior survival outcomes in our study may be attributed to a higher proportion of very elderly patients with high-risk features. Potential immunologic mechanisms of activity of MST include attenuation of inflammatory cytokines and emergence of tumor-specific T cell clones.

Texture Analysis with 3.0-T MRI for Association of Response to Neoadjuvant Chemotherapy in Breast Cancer

Background Previous studies have suggested that texture analysis is a promising tool in the diagnosis, characterization, and assessment of treatment response in various cancer types. Therefore, application of texture analysis may be helpful for early prediction of pathologic response in breast cancer. Purpose To investigate whether texture analysis of features from MRI is associated with pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) in breast cancer. Materials and Methods This retrospective study included 136 women (mean age, 47.9 years; range, 31-70 years) who underwent NAC and subsequent surgery for breast cancer between January 2012 and August 2017. Patients were monitored with 3.0-T MRI before (pretreatment) and after (midtreatment) three or four cycles of NAC. Texture analysis was performed at pre- and midtreatment T2-weighted MRI, contrast material-enhanced T1-weighted MRI, diffusion-weighted MRI, and apparent diffusion coefficient (ADC) mapping by using commercial software. A random forest method was applied to build a predictive model for classifying those with pCR with use of texture parameters. Diagnostic performance for predicting pCR was assessed and compared with that of six other machine learning classifiers (adaptive boosting, decision tree, k-nearest neighbor, linear support vector machine, naive Bayes, and linear discriminant analysis) by using the Wald test and DeLong method. Results Forty of the 136 patients (29%) achieved pCR after NAC. In the prediction of pCR, the random forest classifier showed the lowest diagnostic performance with pretreatment ADC (area under the receiver operating characteristic curve [AUC], 0.53; 95% confidence interval: 0.44, 0.61) and the highest diagnostic performance with midtreatment contrast-enhanced T1-weighted MRI (AUC, 0.82; 95% confidence interval: 0.74, 0.88) among pre- and midtreatment T2-weighted MRI, contrast-enhanced T1-weighted MRI, diffusion-weighted MRI, and ADC mapping. Conclusion Texture parameters using a random forest method of contrast-enhanced T1-weighted MRI at midtreatment of neoadjuvant chemotherapy were valuable and associated with pathologic complete response in breast cancer. © RSNA, 2019 Online supplemental material is available for this article.

Quantifying performance of a diagnostic test as the expected information for discrimination: Relation to the C-statistic

Although the C-statistic is widely used for evaluating the performance of diagnostic tests, its limitations for evaluating the predictive performance of biomarker panels have been widely discussed. The increment in C obtained by adding a new biomarker to a predictive model has no direct interpretation, and the relevance of the C-statistic to risk stratification is not obvious. This paper proposes that the C-statistic should be replaced by the expected information for discriminating between cases and non-cases (expected weight of evidence, denoted as Λ), and that the strength of evidence favouring one model over another should be evaluated by cross-validation as the difference in test log-likelihoods. Contributions of independent variables to predictive performance are additive on the scale of Λ. Where the effective number of independent predictors is large, the value of Λ is sufficient to characterize fully how the predictor will stratify risk in a population with given prior probability of disease, and the C-statistic can be interpreted as a mapping of Λ to the interval from 0.5 to 1. Even where this asymptotic relationship does not hold, there is a one-to-one mapping between the distributions in cases and non-cases of the weight of evidence favouring case over non-case status, and the quantiles of these distributions can be used to calculate how the predictor will stratify risk. This proposed approach to reporting predictive performance is demonstrated by analysis of a dataset on the contribution of microbiome profile to diagnosis of colorectal cancer.

MRI-Based Methods for Spinal Cord Atrophy Evaluation: A Comparison of Cervical Cord Cross-Sectional Area, Cervical Cord Volume, and Full Spinal Cord Volume in Patients with Aquaporin-4 Antibody Seropositive Neuromyelitis Optica Spectrum Disorders

BACKGROUND AND PURPOSE

Measures for spinal cord atrophy have become increasingly important as imaging biomarkers in the assessment of neuroinflammatory diseases, especially in neuromyelitis optica spectrum disorders. The most commonly used method, mean upper cervical cord area, is relatively easy to measure and can be performed on brain MRIs that capture cervical myelon. Measures of spinal cord volume (eg, cervical cord volume or total cord volume) require longer scanning and more complex analysis but are potentially better suited as spinal cord atrophy measures. This study investigated spinal cord atrophy measures in a cohort of healthy subjects and patients with aquaporin-4 antibody seropositive neuromyelitis optica spectrum disorders and evaluated the discriminatory performance of mean upper cervical cord cross-sectional area compared with cervical cord volume and total cord volume.

MATERIALS AND METHODS

Mean upper cervical cord area, cervical cord volume, and total cord volume were measured using 3T MRIs from healthy subjects (n = 19) and patients with neuromyelitis optica spectrum disorders (n = 30). Group comparison and receiver operating characteristic analyses between healthy controls and patients with neuromyelitis optica spectrum disorders were performed.

RESULTS

Mean upper cervical cord area, cervical cord volume, and total cord volume measures showed similar and highly significant group differences between healthy control subjects and patients with neuromyelitis optica spectrum disorders (P < .01 for all). All 3 measures showed similar receiver operating characteristic-area under the curve values (mean upper cervical cord area = 0.70, cervical cord volume = 0.75, total cord volume = 0.77) with no significant difference between them. No associations among mean upper cervical cord cross-sectional area, cervical cord volume, or total cord volume with disability measures were found.

CONCLUSIONS

All 3 measures showed similar discriminatory power between healthy control and neuromyelitis optica spectrum disorders groups. Mean upper cervical cord area is easier to obtain compared with cervical cord volume and total cord volume and can be regarded as an efficient representative measure of spinal cord atrophy in the neuromyelitis optica spectrum disorders context.

Diagnostic performance enhancement of pancreatic cancer using proteomic multimarker panel

Due to its high mortality rate and asymptomatic nature, early detection rates of pancreatic ductal adenocarcinoma (PDAC) remain poor. We measured 1000 biomarker candidates in 134 clinical plasma samples by multiple reaction monitoring-mass spectrometry (MRM-MS). Differentially abundant proteins were assembled into a multimarker panel from a training set (n=684) and validated in independent set (n=318) from five centers. The level of panel proteins was also confirmed by immunoassays. The panel including leucine-rich alpha-2 glycoprotein (LRG1), transthyretin (TTR), and CA19-9 had a sensitivity of 82.5% and a specificity of 92.1%. The triple-marker panel exceeded the diagnostic performance of CA19-9 by more than 10% (AUC_CA19-9 = 0.826, AUC_panel= 0.931, P < 0.01) in all PDAC samples and by more than 30% (AUC_CA19-9 = 0.520, AUC_panel = 0.830, P < 0.001) in patients with normal range of CA19-9 (<37U/mL). Further, it differentiated PDAC from benign pancreatic disease (AUC_CA19-9 = 0.812, AUC_panel = 0.892, P < 0.01) and other cancers (AUC_CA19-9 = 0.796, AUC_panel = 0.899, P < 0.001). Overall, the multimarker panel that we have developed and validated in large-scale samples by MRM-MS and immunoassay has clinical applicability in the early detection of PDAC.

Semiquantitative Imaging Biomarkers of Knee Osteoarthritis Progression: Data From the Foundation for the National Institutes of Health Osteoarthritis Biomarkers Consortium

OBJECTIVE

To determine the association between changes in semiquantitative magnetic resonance imaging (MRI) biomarkers over 24 months and radiographic and pain progression over 48 months in knees with mild-to-moderate osteoarthritis (OA).

METHODS

We undertook a nested case-control study as part of the Foundation for the National Institutes of Health Biomarkers Consortium Project. We used multivariable logistic regression models to examine the association between change over 24 months in semiquantitative MRI markers and radiographic and pain progression in knee OA. MRIs were read according to the MRI OA Knee Score system. We focused on changes in cartilage, osteophytes, meniscus, bone marrow lesions, Hoffa-synovitis, and effusion-synovitis.

RESULTS

The most parsimonious model included changes in cartilage thickness and surface area, effusion-synovitis, Hoffa-synovitis, and meniscal morphology (C statistic 0.740). Compared with no worsening, worsening in cartilage thickness in ≥3 subregions was associated with 2.8-fold (95% confidence interval [95% CI] 1.3-5.9) greater odds of being a case, and worsening in cartilage surface area in ≥3 subregions was associated with 2.4-fold (95% CI 1.3-4.4) greater odds of being a case. Worsening of meniscal morphology in any region was associated with 2.2-fold (95% CI 1.3-3.8) greater odds of being a case. Worsening effusion-synovitis and Hoffa-synovitis were also associated with a greater odds of being a case (odds ratios 2.7 and 2.0, respectively).

CONCLUSION

Twenty-four-month changes in cartilage thickness, cartilage surface area, effusion-synovitis, Hoffa-synovitis, and meniscal morphology were independently associated with OA progression, suggesting that these factors may serve as efficacy biomarkers in clinical trials of disease-modifying interventions for knee OA.

Simulation of complex data structures for planning of studies with focus on biomarker comparison

BACKGROUND

There are a growing number of observational studies that do not only focus on single biomarkers for predicting an outcome event, but address questions in a multivariable setting. For example, when quantifying the added value of new biomarkers in addition to established risk factors, the aim might be to rank several new markers with respect to their prediction performance. This makes it important to consider the marker correlation structure for planning such a study. Because of the complexity, a simulation approach may be required to adequately assess sample size or other aspects, such as the choice of a performance measure.

METHODS

In a simulation study based on real data, we investigated how to generate covariates with realistic distributions and what generating model should be used for the outcome, aiming to determine the least amount of information and complexity needed to obtain realistic results. As a basis for the simulation a large epidemiological cohort study, the Gutenberg Health Study was used. The added value of markers was quantified and ranked in subsampling data sets of this population data, and simulation approaches were judged by the quality of the ranking. One of the evaluated approaches, the random forest, requires original data at the individual level. Therefore, also the effect of the size of a pilot study for random forest based simulation was investigated.

RESULTS

We found that simple logistic regression models failed to adequately generate realistic data, even with extensions such as interaction terms or non-linear effects. The random forest approach was seen to be more appropriate for simulation of complex data structures. Pilot studies starting at about 250 observations were seen to provide a reasonable level of information for this approach.

CONCLUSIONS

We advise to avoid oversimplified regression models for simulation, in particular when focusing on multivariable research questions. More generally, a simulation should be based on real data for adequately reflecting complex observational data structures, such as found in epidemiological cohort studies.

The use of angiogenic biomarkers in maternal blood to identify which SGA fetuses will require a preterm delivery and mothers who will develop pre-eclampsia

OBJECTIVE

To determine (1) whether maternal plasma concentrations of angiogenic and anti-angiogenic factors can predict which mothers diagnosed with "suspected small for gestational age fetuses (sSGA)" will develop pre-eclampsia (PE) or require an indicated early preterm delivery (≤ 34 weeks of gestation); and (2) whether risk assessment performance is improved using these proteins in addition to clinical factors and Doppler parameters.

METHODS

This prospective cohort study included women with singleton pregnancies diagnosed with sSGA (estimated fetal weight <10th percentile) between 24 and 34 weeks of gestation (n = 314). Plasma concentrations of soluble vascular endothelial growth factor receptor-1 (sVEGFR-1), soluble endoglin (sEng) and placental growth factor (PlGF) were determined in maternal blood obtained at the time of diagnosis. Doppler velocimetry of the umbilical (Umb) and uterine (UT) arteries was performed. The outcomes were (1) subsequent development of PE; and (2) indicated preterm delivery at ≤ 34 weeks of gestation (excluding deliveries as a result of spontaneous preterm labor, preterm pre-labor rupture of membranes or chorioamnionitis).

RESULTS

(1) The prevalence of PE and indicated preterm delivery was 9.2% (n = 29/314) and 7.3% (n = 23/314), respectively; (2) the area under the receiver operating characteristic curve (AUC) for the identification of patients who developed PE and/or required indicated preterm delivery was greater than 80% for the UT artery pulsatility index (PI) z-score and each biochemical marker (including their ratios) except sVEGFR-1 MoM; (3) using cutoffs at a false positive rate of 15%, women with abnormal plasma concentrations of angiogenic/anti-angiogenic factors were 7-13 times more likely to develop PE, and 12-22 times more likely to require preterm delivery than those with normal plasma MoM concentrations of these factors; (4) sEng, PlGF, PIGF/sEng and PIGF/sVEGFR-1 ratios MoM, each contributed significant information about the risk of PE beyond that provided by clinical factors and/or Doppler parameters: women who had low MoM values for these biomarkers were at 5-9 times greater risk of developing PE than women who had normal values, adjusting for clinical factors and Doppler parameters (adjusted odds ratio for PlGF: 9.1, PlGF/sEng: 5.6); (5) the concentrations of sVEGFR-1 and PlGF/sVEGFR-1 ratio MoM, each contributed significant information about the risk of indicated preterm delivery beyond that provided by clinical factors and/or Doppler parameters: women who had abnormal values were at 8-9 times greater risk for indicated preterm delivery, adjusting for clinical factors and Doppler parameters; and (6) for a two-stage risk assessment (Umb artery Doppler followed by Ut artery Doppler plus biochemical markers), among women who had normal Umb artery Doppler velocimetry (n = 279), 21 (7.5%) developed PE and 11 (52%) of these women were identified by an abnormal UT artery Doppler mean PI z-score (>2SD): a combination of PlGF/sEng ratio MoM concentration and abnormal UT artery Doppler velocimetry increased the sensitivity of abnormal UT artery Doppler velocimetry to 76% (16/21) at a fixed false-positive rate of 10% (p = 0.06).

CONCLUSION

Angiogenic and anti-angiogenic factors measured in maternal blood between 24 and 34 weeks of gestation can identify the majority of mothers diagnosed with "suspected SGA" who subsequently developed PE or those who later required preterm delivery ≤ 34 weeks of gestation. Moreover, incorporation of these biochemical markers significantly improves risk assessment performance for these outcomes beyond that of clinical factors and uterine and umbilical artery Doppler velocimetry.

A simulation based method for assessing the statistical significance of logistic regression models after common variable selection procedures

Classification models can demonstrate apparent prediction accuracy even when there is no underlying relationship between the predictors and the response. Variable selection procedures can lead to false positive variable selections and overestimation of true model performance. A simulation study was conducted using logistic regression with forward stepwise, best subsets, and LASSO variable selection methods with varying total sample sizes (20, 50, 100, 200) and numbers of random noise predictor variables (3, 5, 10, 15, 20, 50). Using our critical values can help reduce needless follow-up on variables having no true association with the outcome.

Association of Aging-Related Endophenotypes With Mortality in 2 Cohort Studies: the Long Life Family Study and the Health, Aging and Body Composition Study

One method by which to identify fundamental biological processes that may contribute to age-related disease and disability, instead of disease-specific processes, is to construct endophenotypes comprising linear combinations of physiological measures. Applying factor analyses methods to phenotypic data (2006-2009) on 28 traits representing 5 domains (cognitive, cardiovascular, metabolic, physical, and pulmonary) from 4,472 US and Danish individuals in 574 pedigrees from the Long Life Family Study (United States and Denmark), we constructed endophenotypes and assessed their relationship with mortality. The most dominant endophenotype primarily reflected the physical activity and pulmonary domains, was heritable, was significantly associated with mortality, and attenuated the association of age with mortality by 24.1%. Using data (1997-1998) on 1,794 Health, Aging and Body Composition Study participants from Memphis, Tennessee, and Pittsburgh, Pennsylvania, we obtained strikingly similar endophenotypes and relationships to mortality. We also reproduced the endophenotype constructs, especially the dominant physical activity and pulmonary endophenotype, within demographic subpopulations of these 2 cohorts. Thus, this endophenotype construct may represent an underlying phenotype related to aging. Additional genetic studies of this endophenotype may help identify genetic variants or networks that contribute to the aging process.

Developing a utility decision framework to evaluate predictive models in breast cancer risk estimation

Combining imaging and genetic information to predict disease presence and progression is being codified into an emerging discipline called "radiogenomics." Optimal evaluation methodologies for radiogenomics have not been well established. We aim to develop a decision framework based on utility analysis to assess predictive models for breast cancer diagnosis. We garnered Gail risk factors, single nucleotide polymorphisms (SNPs), and mammographic features from a retrospective case-control study. We constructed three logistic regression models built on different sets of predictive features: (1) Gail, (2) Gail + Mammo, and (3) Gail + Mammo + SNP. Then we generated receiver operating characteristic (ROC) curves for three models. After we assigned utility values for each category of outcomes (true negatives, false positives, false negatives, and true positives), we pursued optimal operating points on ROC curves to achieve maximum expected utility of breast cancer diagnosis. We performed McNemar's test based on threshold levels at optimal operating points, and found that SNPs and mammographic features played a significant role in breast cancer risk estimation. Our study comprising utility analysis and McNemar's test provides a decision framework to evaluate predictive models in breast cancer risk estimation.

Comparing two correlated C indices with right-censored survival outcome: a one-shot nonparametric approach

The area under the receiver operating characteristic curve is often used as a summary index of the diagnostic ability in evaluating biomarkers when the clinical outcome (truth) is binary. When the clinical outcome is right-censored survival time, the C index, motivated as an extension of area under the receiver operating characteristic curve, has been proposed by Harrell as a measure of concordance between a predictive biomarker and the right-censored survival outcome. In this work, we investigate methods for statistical comparison of two diagnostic or predictive systems, of which they could either be two biomarkers or two fixed algorithms, in terms of their C indices. We adopt a U-statistics-based C estimator that is asymptotically normal and develop a nonparametric analytical approach to estimate the variance of the C estimator and the covariance of two C estimators. A z-score test is then constructed to compare the two C indices. We validate our one-shot nonparametric method via simulation studies in terms of the type I error rate and power. We also compare our one-shot method with resampling methods including the jackknife and the bootstrap. Simulation results show that the proposed one-shot method provides almost unbiased variance estimations and has satisfactory type I error control and power. Finally, we illustrate the use of the proposed method with an example from the Framingham Heart Study.

Salivary biomarkers for detection of oral squamous cell carcinoma - current state and recent advances

Oral squamous cell carcinoma (OSCC) is the most common malignant neoplasm of the oral cavity. Detection of OSCC is currently based on thorough clinical oral examination combined with biopsy for histological analysis. Most cases of OSCC are not detected until the cancer has developed into advanced stages; thus, a reliable early stage diagnostic marker is needed. This literature review presents an overview of the status of current advances in salivary diagnostics for OSCC. Though many protein and mRNA salivary biomarkers have been identified that can detect OSCC with high sensitivity and specificity, the most discernable findings occur with the use of multiple markers. Studies that incorporate proteomic, transcriptomic, and potentially additional "omics", including methylomics, need to be initiated to bring technology to clinical applications and allow the best use of saliva in diagnosing OSCC.