Understanding suboptimal insulin use in type 1 and 2 diabetes: a cross-sectional survey of healthcare providers who treat people with diabetes.

BACKGROUND: The purpose of this study was to understand the healthcare provider (HCP) perspective on the extent of suboptimal insulin dosing in people with diabetes (PwD), as well as specific challenges and solutions to insulin management. METHODS: An online survey of general practitioners and specialists (N = 640) who treat PwD in Germany, Spain, the United Kingdom, and the United States was conducted. Responses regarding HCP background and their patients, HCP perceptions of suboptimal insulin use, and challenges associated with optimal insulin use were collected. Categorical summary statistics were presented. RESULTS: Overall, for type 1 diabetes (T1D) and type 2 diabetes (T2D), most physicians indicated < 30% of PwD missed or skipped a bolus insulin dose in the last 30 days (T1D: 83.0%; T2D: 74.1%). The top 3 reasons (other than skipping a meal) HCPs believed caused the PwD to miss or skip insulin doses included they "forgot," (bolus: 75.0%; basal: 67.5%) "were too busy/distracted," (bolus: 58.8%; basal: 48.3%), and "were out of their normal routine" (bolus: 57.8%; basal: 48.6%). HCPs reported similar reasons that they believed caused PwD to mistime insulin doses. Digital technology and improved HCP-PwD communication were potential solutions identified by HCPs to optimize insulin dosing in PwD. CONCLUSIONS: Other studies have shown that PwD frequently experience suboptimal insulin dosing. Conversely, results from this study showed that HCPs believe suboptimal insulin dosing among PwD is limited in frequency. While no direct comparisons were made in this study, this apparent discrepancy could lead to difficulties in HCPs giving PwD the best advice on optimal insulin management. Approaches such as improving the objectivity of dose measurements for both PwD and HCPs may improve associated communications and help reduce suboptimal insulin dosing, thus enhancing treatment outcomes.

Fasting and postprandial plasma glucose contributions to hemoglobin A1c and time in range in people with diabetes on multiple daily injection insulin therapy: Results from the PRONTO-T1D and PRONTO-T2D clinical trials.

AIMS: To investigate contributions of changes in fasting plasma glucose (FPG) and postprandial glucose (PPG) to changes in hemoglobin A1c (HbA1c) and time-in-range (TIR, 70-180 mg/dL) in people with type 1 diabetes (T1D) and type 2 diabetes (T2D) treated with multiple daily injections (MDI) of insulin lispro (rapid/ultra-rapid formulations). METHODS: Multivariate regression models were used to quantify the contributions of FPG and PPG reductions to change in HbA1c and TIR using data from the PRONTO-T1D (N = 1222) and PRONTO-T2D (N = 673) clinical trials. TIR was derived from 10-point self-monitored blood glucose (SMBG) profiles overall, as well as from continuous glucose monitoring (CGM) in the PRONTO-T1D CGM substudy (n = 269/1222). RESULTS: A 1 mmol/L FPG reduction corresponded with a 0.09-0.12 % (95 % confidence interval [CI] 0.06-0.15 %) HbA1c reduction in PRONTO-T1D and 0.17-0.26 % (95 % CI 0.13-0.30 %) in PRONTO-T2D (both p < 0.0001). A 1 mmol/L PPG reduction corresponded with a 0.05-0.09 % (95 % CI 0.01-0.12 %) HbA1c reduction in PRONTO-T1D (p < 0.001) and 0.10-0.15 % (95 % CI 0.05-0.19 %) in PRONTO-T2D (p < 0.0001). Reductions in FPG and PPG were significantly associated with increased TIR whether derived from SMBG (7.87-12.95 % [95 % CI 6.81-14.23 %]; all p < 0.0001) or CGM (3.35-4.18 % [95 % CI 2.11-5.39 %]; all p < 0.05). CONCLUSIONS: FPG and PPG significantly impact HbA1c and TIR. Balanced management of both FPG and PPG is important to achieve glycemic goals for people with diabetes on MDI insulin therapy. CLINICAL TRIALS REGISTRATION: PRONTO-T1D ClinicalTrials.gov identifier: NCT03214367; PRONTO-T2D ClinicalTrials.gov Identifier: NCT03214380.

Clinical, Economic, and Patient-Reported Benefits of Connected Insulin Pen Systems: A Systematic Literature Review.

INTRODUCTION: The objective of this systematic literature review was to evaluate the available literature concerning the clinical, economic, and patient-reported benefits of insulin pen platforms, including connected insulin pens/caps/sleeves and insulin platforms, as well as mobile apps capable of receiving near real-time insulin dosing information. METHODS: Medline and Embase databases and the Cochrane Library were searched for published literature between January 2015 and May 2021, and manual searches for conference abstracts from 2018 to May 2021 were performed. These searches were supplemented by internet searches for relevant literature and clinical trials. Study selection involved the population, intervention, comparator, outcomes, time frame, and study design outline. Included studies investigated connected insulin systems or connected caps/sleeves enabling pens to be connected, or apps able to connect to these systems, in individuals of all ages with type 1 or type 2 diabetes mellitus. RESULTS: Searches identified a total of 26 publications (mostly observational studies and conference abstracts) for inclusion, representing ten unique, predominantly small studies. Evidence in this field is still in its early stages, and only two randomized controlled trials met our inclusion criteria. Available results showed that connected insulin pens and their systems potentially helped reduce suboptimal insulin use and may therefore improve glycemic control. Satisfaction of people with diabetes with the technologies used was high, and economic benefits were noted. Features of effective connected insulin pen devices include simplicity of use and data upload/sharing, useful "point-of-care" alerts, and simple and understandable data presentation to facilitate more effective consultations. CONCLUSIONS: Connected insulin pen systems could be increasingly considered as part of routine clinical care for insulin-treated persons with diabetes who must manage the complexity of their daily insulin routine. Future research focusing on the way data obtained from these devices can be most effectively used alongside other information is urgently needed.

Digital health tools, like text message reminders and mobile apps, are being used more often to help people with diabetes improve their health in a way that works for them. For people who take insulin to treat their diabetes, what has been missing is a way to track insulin doses alongside other diabetes information in an app. Connected insulin pens, also called smart pens, are able to do this. In this article we have looked at the evidence available on the benefits of connected insulin pens. We found that while information on connected insulin pens is limited at the moment, what there is shows that using a connected insulin pen can help people remember to take their insulin and give themselves the right dose and that those who have used a connected insulin pen or related technology are happy with it. Useful features of connected insulin pens include being easy to use, having an alert function, and being able to share the insulin information with the user's doctor. Connected insulin pens may also reduce diabetes-related costs. Connected insulin pens are likely to become more common for people with diabetes who take insulin, but there is a need for more research on how best to use them to improve the treatment of people with diabetes.

Adherence and Persistence to Insulin Therapy in People with Diabetes: Impact of Connected Insulin Pen Delivery Ecosystem.

Diabetes is an increasing public health problem, and insulin is the mainstay for treatment of type 1 diabetes. In type 2 diabetes treatment, insulin therapy is used after oral or other injectable agents become inadequate to achieve glycemic control. Despite the advances in insulin therapy, management of diabetes remains challenging. Numerous studies have reported low adherence and persistence to insulin therapy, which acts as a barrier to successful glycemic control and diabetes management. The aim of this targeted review article is to provide an overview of adherence and persistence to insulin therapy in people with diabetes and to discuss the impact of the emergence of a new connected ecosystem of increasingly sophisticated insulin pens, glucose monitoring systems, telemedicine, and mHealth on diabetes management. With the emergence of a connected diabetes ecosystem, we have entered an era of advanced personalized insulin delivery, which will have the potential to enhance diabetes self-management and clinical management. Early systems promise to unlock the potential to address missed or late bolus insulin delivery, which should help to address non-adherence and non-persistence. Over time, improvements in this ecosystem have the potential to combine insulin data with previously missing contextualized patient data, including meal, glucose, and activity data to support personalized clinical decisions and ultimately revolutionize insulin therapy.

Individualized HbA1c target selection and achievement in the Multinational Observational Study Assessing Insulin Use (MOSA1c) type 2 diabetes study.

AIM: To identify which individual-, physician-, and the healthcare system-related factors can predict individualized hemoglobin A1c (HbA1c) targets and the likelihood of reaching those targets after initial insulin therapy over a two-year follow-up period. METHODS: Real-world data, including baseline characteristics of people with type 2 diabetes mellitus (T2DM), psychosocial data, and diabetes medication use, collected from the Multinational Observational Study Assessing Insulin Use (MOSA1c) study in 18 countries were analyzed. RESULTS: Overall, 225 of 1194 people with T2DM (18.8%) who received initial insulin therapy for ≥3 months reached HbA1c targets at two-year follow-up; most were likely to be White (64.9%) and perceptions of their relationship with physicians were less positive than those who did not reach HbA1c targets. Higher baseline HbA1c (>8%) was the strongest predictor of being assigned an HbA1c target >7% (odds ratio [OR] 6.06, 95% confidence interval [CI] 3.97, 9.26). A smaller difference between baseline and target HbA1c levels was the strongest predictor of reaching an HbA1c target at two-year follow-up (large vs small difference, OR 0.28, 95% CI 0.17, 0.47). CONCLUSIONS: Several factors were significantly associated with establishing individualized HbA1c targets and reaching these targets. A small proportion of people with T2DM on insulin therapy reached their HbA1c target. Personalized management of glycemic targets necessitates the adoption of multi-factorial strategies, as several factors could influence an individual's glycemic outcome. CLINICALTRIALS. GOV IDENTIFIER: NCT01400971.

Missed and Mistimed Insulin Doses in People with Diabetes: A Systematic Literature Review.

Background: Development of coordinated management approaches is important to facilitate self-care in people with diabetes (PwD). Gaining a better understanding of suboptimal insulin use is key in this endeavor. This review aimed, for the first time, to systematically identify and narratively summarize real-world evidence on the extent of suboptimal insulin use (missed and mistimed insulin) in PwD. Methods: A systematic literature search of MEDLINE, EMBASE, and the Cochrane Database of Systematic Reviews identified studies reporting on missed and mistimed insulin dosing. Results: From 3305 studies, 37 publications reporting on 30 unique studies that involved 58,617 PwD were included. Studies were conducted across 12 different countries and most employed cross-sectional surveys. Observations regarding missed and mistimed insulin doses were reported in 25 and 10 studies, respectively. PwD reported missing insulin doses, but rates varied due to differences in reporting methods, participant populations, and insulin regimens. The association between missed dosing and glycemic control was evaluated in ten studies in which the authors reported lower glycated hemoglobin (HbA1c) levels in PwD who did not omit insulin. The proportion of PwD reporting mistiming of insulin was in the range of 20-45%, depending on the study; this was associated with higher rates of hypoglycemia and higher HbA1c as reported by study authors. Reasons for suboptimal insulin use were multifactorial, occurring due to disrupted daily routines, social situations, and hypoglycemia avoidance. Conclusions: This review suggests that suboptimal insulin use is widespread and that PwD using insulin may still be struggling with disease management. There is an unmet need for better integrated support in managing the complexities of insulin therapy and for the development of systems (e.g. digital solutions) that empower people to take control of insulin-treated diabetes.

Fasting and postprandial plasma glucose contribution to glycated haemoglobin and time in range in people with type 2 diabetes on basal and bolus insulin therapy: Results from a pooled analysis of insulin lispro clinical trials.

AIMS: To investigate the interrelations between glycaemic metrics of fasting plasma glucose (FPG), postprandial glucose (PPG), glycated haemoglobin (HbA1c), and percentage of time in target range 3.9 to 10.0 mmol/L (%TIR) in patients on insulin therapy. MATERIALS AND METHODS: A pooled analysis was conducted using datasets extracted from an integrated database of insulin lispro clinical trials (Eli Lilly and Company). Studies in patients with type 2 diabetes on basal-bolus or basal-plus insulin therapy, and with ≥7-point self-monitored blood glucose profiles were included in the analysis. A multivariate regression model was used to quantify the contribution of FPG and PPG change to the change in HbA1c and %TIR. In addition, a linear regression model was used to describe the relationship between %TIR and HbA1c. RESULTS: Five studies encompassing 1572 patients met the criteria for inclusion. On average, a 1-mmol/L change in FPG was associated with 2.7 mmol/mol (0.25%) change in HbA1c (range 2.0 to 2.8 mmol/mol [0.18%-0.26%]; all P <0.0001), and a 1-mmol/L change in PPG with 1.8 mmol/mol (0.16%) change in HbA1c (range 1.2 to 2.1 mmol/mol [0.11%-0.19%]; all P <0.01). Furthermore, a 1-mmol/L reduction in FPG and PPG was associated with an increase in TIR of 6.5% (range 5.8%-9.2%) and 5.3% (range 4.1%-8.7%), respectively, all P <0.0001. A decrease in HbA1c of 10.9 mmol/mol (1%) corresponded with an increase in TIR of 8.3%, on average. CONCLUSIONS: In patients with type 2 diabetes on basal-bolus or basal-plus insulin therapy, management of both FPG and PPG is important for achievement of HbA1c and TIR goals.

Making sense of glucose metrics in diabetes: linkage between postprandial glucose (PPG), time in range (TIR) & hemoglobin A1c (A1C).

While A1C is the standard diagnostic test for evaluating long-term glucose management, additional glucose data, either from fingerstick blood glucose testing, or more recently, continuous glucose monitoring (CGM), is necessary for safe and effective management of diabetes, especially for individuals treated with insulin. CGM technology and retrospective pattern-based management using various CGM reports have the potential to improve glycemic management beyond what is possible with fingerstick blood glucose monitoring. CGM software can provide valuable retrospective data on Time-in-Ranges (above, below, within) metrics, the Ambulatory Glucose Profile (AGP), overlay reports, and daily views for persons with diabetes and their healthcare providers. This data can aid in glycemic pattern identification and evaluation of the impact of lifestyle factors on these patterns. Time-in-Ranges data provide an easy-to-define metric that can facilitate goal setting discussions between clinicians and persons with diabetes to improve glycemic management and can empower persons with diabetes in self-management between clinic consultation visits. Here we discuss multiple real-life scenarios from a primary care clinic for the application of CGM in persons with diabetes. Optimizing the use of the reports generated by CGM software, with attention to time in range, time below range, and postprandial glucose-induced time above range, can improve the safety and efficacy of ongoing glucose management.

Patient perspectives on the use of half-unit insulin pens by people with type 1 diabetes: a cross-sectional observational study.

BACKGROUND: Half-unit pens offer the ability to dose insulin more precisely. Information about half-unit pen use and evidence of their benefits and drawbacks is limited. This study aims to characterize people with type 1 diabetes (T1D) who have used (current/former = EVER) vs. those who have never used half-unit pens (NEVER users) and to understand their perspective. METHODS: An observational cross-sectional online survey was administered through T1D Exchange's online patient community, myGlu.org, to understand the use of half-unit insulin pens. RESULTS: The 278 adult participants (156 EVER, 122 NEVER) had a mean age of 41.8 ± 12.7 years, body mass index of 26.0 ± 3.8 kg/m2, glycated hemoglobin of 6.3% ± 1.0%, and 55% were male. EVER users had T1D for a shorter duration than NEVER users (p < .001). EVER users were less likely to use continuous subcutaneous insulin infusion (p < .001) and more likely to start correcting high blood glucose at a lower level (p < .001) and low blood glucose at a higher level (p < .001). The highest ranked benefits of half-unit pen reported by its current users (N = 131) included prevention of hyperglycemia (40.5%), less anxiety or worry (23.7%), and prevention of hypoglycemia (16.8%). CONCLUSIONS: Half-unit insulin pen is perceived as an insulin device that may help people with T1D to avoid hypo- and hyperglycemic events and decrease their level of disease worry and anxiety. This study highlights the need for patients and health care providers to understand the benefits of half-unit pens while considering options for individualized diabetes management.

Circulating Markers Reflect Both Anti- and Pro-Atherogenic Drug Effects in ApoE-Deficient Mice.

BACKGROUND: Current drug therapy of atherosclerosis is focused on treatment of major risk factors, e.g. hypercholesterolemia while in the future direct disease modification might provide additional benefits. However, development of medicines targeting vascular wall disease is complicated by the lack of reliable biomarkers. In this study, we took a novel approach to identify circulating biomarkers indicative of drug efficacy by reducing the complexity of the in vivo system to the level where neither disease progression nor drug treatment was associated with the changes in plasma cholesterol. RESULTS: ApoE-/- mice were treated with an ACE inhibitor ramipril and HMG-CoA reductase inhibitor simvastatin. Ramipril significantly reduced the size of atherosclerotic plaques in brachiocephalic arteries, however simvastatin paradoxically stimulated atherogenesis. Both effects occurred without changes in plasma cholesterol. Blood and vascular samples were obtained from the same animals. In the whole blood RNA samples, expression of MMP9, CD14 and IL-1RN reflected pro-and anti-atherogenic drug effects. In the plasma, several proteins, e.g. IL-1beta, IL-18 and MMP9 followed similar trends while protein readout was less sensitive than RNA analysis. CONCLUSION: In this study, we have identified inflammation-related whole blood RNA and plasma protein markers reflecting anti-atherogenic effects of ramipril and pro-atherogenic effects of simwastatin in a mouse model of atherosclerosis. This opens an opportunity for early, non-invasive detection of direct drug effects on atherosclerotic plaques in complex in vivo systems.

Analysis of pathway activity in primary tumors and NCI60 cell lines using gene expression profiling data.

To determine cancer pathway activities in nine types of primary tumors and NCI60 cell lines, we applied an in silica approach by examining gene signatures reflective of consequent pathway activation using gene expression data. Supervised learning approaches predicted that the Ras pathway is active in approximately 70% of lung adenocarcinomas but inactive in most squamous cell carcinomas, pulmonary carcinoids, and small cell lung carcinomas. In contrast, the TGF-beta, TNF-alpha, Src, Myc, E2F3, and beta-catenin pathways are inactive in lung adenocarcinomas. We predicted an active Ras, Myc, Src, and/or E2F3 pathway in significant percentages of breast cancer, colorectal carcinoma, and gliomas. Our results also suggest that Ras may be the most prevailing oncogenic pathway. Additionally, many NCI60 cell lines exhibited a gene signature indicative of an active Ras, Myc, and/or Src, but not E2F3, beta-catenin, TNF-alpha, or TGF-beta pathway. To our knowledge, this is the first comprehensive survey of cancer pathway activities in nine major tumor types and the most widely used NCI60 cell lines. The "gene expression pathway signatures" we have defined could facilitate the understanding of molecular mechanisms in cancer development and provide guidance to the selection of appropriate cell lines for cancer research and pharmaceutical compound screening.

Retrospective analysis of protein kinase C-beta (PKC-beta) expression in lymphoid malignancies and its association with survival in diffuse large B-cell lymphomas.

BACKGROUND: Both mechanistic features and recent correlative findings suggest a potential role for protein kinase C-beta (PKC-beta) in tumor pathogenesis, particularly in B-cell malignancies. To evaluate the role of this gene in lymphoid malignancies, we analyzed global gene expression data to quantify PKC-beta expression across diagnostic groups and, when possible, determined correlations between PKC-beta expression and survival. RESULTS: Our analysis showed that the level of PKC-beta expression was highest in chronic lymphocytic leukemia and follicular lymphoma. Within diffuse large-B cell lymphoma (DLBCL), PKC-beta expression was significantly higher in activated B-cell- like subtype than germinal center B-cell- like subtype (P < 0.0001). Elevated PKC-beta appeared to be associated with worse survival in both of these subtypes. When analyzed within clinically defined risk groups established by the International Prognostic Index (IPI), PKC-beta expression was lowest in patients with low IPI scores (0-1). Within intermediate- and high-risk IPI groups, elevated PKC-beta expression was associated with worse survival, suggesting that PKC-beta may expand the prognostic value of the IPI. Results of global gene expression analyses of DLBCL samples corroborate previous observations that anti-apoptosis, cell proliferation, and B-cell proliferation signaling pathways are functionally related to PKC-beta. CONCLUSION: We present a first detailed pharmacogenomics report comparing PKC-beta mRNA expression across different lymphoid malignancies and evaluating it as an outcome predictor. Our findings suggest that DLBCL patients with elevated PKC-beta have a worse prognosis, indicating that further evaluation of PKC-beta as a chemotherapeutic target for lymphoid malignancies is warranted.

A flexible integration and visualisation system for biomarker discovery.

Biological data have accumulated at an unprecedented pace as a result of improvements in molecular technologies. However, the translation of data into information, and subsequently into knowledge, requires the intricate interplay of data access, visualisation and interpretation. Biological data are complex and are organised either hierarchically or non-hierarchically. For non-hierarchically organised data, it is difficult to view relationships among biological facts. In addition, it is difficult to make changes in underlying data storage without affecting the visualisation interface. Here, we demonstrate a platform where non-hierarchically organised data can be visualised through the application of a customised hierarchy incorporating medical subject headings (MeSH) classifications. This platform gives users flexibility in updating and manipulation. It can also facilitate fresh scientific insight by highlighting biological impacts across different hierarchical branches. An example of the integration of biomarker information from the curated Proteome database using MeSH and the StarTree visualisation tool is presented.

Too much data, but little inter-changeability: a lesson learned from mining public data on tissue specificity of gene expression.

BACKGROUND: The tissue expression pattern of a gene often provides an important clue to its potential role in a biological process. A vast amount of gene expression data have been and are being accumulated in public repository through different technology platforms. However, exploitations of these rich data sources remain limited in part due to issues of technology standardization. Our objective is to test the data comparability between SAGE and microarray technologies, through examining the expression pattern of genes under normal physiological states across variety of tissues. RESULTS: There are 42-54% of genes showing significant correlations in tissue expression patterns between SAGE and GeneChip, with 30-40% of genes whose expression patterns are positively correlated and 10-15% of genes whose expression patterns are negatively correlated at a statistically significant level (p = 0.05). Our analysis suggests that the discrepancy on the expression patterns derived from technology platforms is not likely from the heterogeneity of tissues used in these technologies, or other spurious correlations resulting from microarray probe design, abundance of genes, or gene function. The discrepancy can be partially explained by errors in the original assignment of SAGE tags to genes due to the evolution of sequence databases. In addition, sequence analysis has indicated that many SAGE tags and Affymetrix array probe sets are mapped to different splice variants or different sequence regions although they represent the same gene, which also contributes to the observed discrepancies between SAGE and array expression data. CONCLUSION: To our knowledge, this is the first report attempting to mine gene expression patterns across tissues using public data from different technology platforms. Unlike previous similar studies that only demonstrated the discrepancies between the two gene expression platforms, we carried out in-depth analysis to further investigate the cause for such discrepancies. Our study shows that the exploitation of rich public expression resource requires extensive knowledge about the technologies, and experiment. Informatic methodologies for better interoperability among platforms still remain a gap. One of the areas that can be improved practically is the accurate sequence mapping of SAGE tags and array probes to full-length genes.

Comparative analysis and integrative classification of NCI60 cell lines and primary tumors using gene expression profiling data.

BACKGROUND: NCI60 cell lines are derived from cancers of 9 tissue origins and have been invaluable in vitro models for cancer research and anti-cancer drug screen. Although extensive studies have been carried out to assess the molecular features of NCI60 cell lines related to cancer and their sensitivities to more than 100,000 chemical compounds, it remains unclear if and how well these cell lines represent or model their tumor tissues of origin. Identification and confirmation of correct origins of NCI60 cell lines are critical to their usage as model systems and to translate in vitro studies into clinical potentials. Here we report a direct comparison between NCI60 cell lines and primary tumors by analyzing global gene expression profiles. RESULTS: Comparative analysis suggested that 51 of 59 cell lines we analyzed represent their presumed tumors of origin. Taking advantage of available clinical information of primary tumor samples used to generate gene expression profiling data, we further classified those cell lines with the correct origins into different subtypes of cancer or different stages in cancer development. For example, 6 of 7 non-small cell lung cancer cell lines were classified as lung adenocarcinomas and all of them were classified into late stages in tumor progression. CONCLUSION: Taken together, we developed and applied a novel approach for systematic comparative analysis and integrative classification of NCI60 cell lines and primary tumors. Our results could provide guidance to the selection of appropriate cell lines for cancer research and pharmaceutical compound screenings. Moreover, this gene expression profile based approach can be generally applied to evaluate experimental model systems such as cell lines and animal models for human diseases.

Evaluating methods for classifying expression data.

An attractive application of expression technologies is to predict drug efficacy or safety using expression data of biomarkers. To evaluate the performance of various classification methods for building predictive models, we applied these methods on six expression datasets. These datasets were from studies using microarray technologies and had either two or more classes. From each of the original datasets, two subsets were generated to simulate two scenarios in biomarker applications. First, a 50-gene subset was used to simulate a candidate gene approach when it might not be practical to measure a large number of genes/biomarkers. Next, a 2000-gene subset was used to simulate a whole genome approach. We evaluated the relative performance of several classification methods by using leave-one-out cross-validation and bootstrap cross-validation. Although all methods perform well in both subsets for a relative easy dataset with two classes, differences in performance do exist among methods for other datasets. Overall, partial least squares discriminant analysis (PLS-DA) and support vector machines (SVM) outperform all other methods. We suggest a practical approach to take advantage of multiple methods in biomarker applications.