Binary Biomarker Research Articles

Addressing heterogeneous sensitivity in biomarker screening with application in NanoString nCounter data

Biomarkers are measurable indicators of biological processes and have wide biomedical applications including disease screening and prognosis prediction. Candidate biomarkers can be screened in high-throughput settings, which allow simultaneous measurements of a large number of molecules. For binary biomarkers, the ability to detect a molecule may be hindered by the presence of background noise and the variable signal strength, which lower the sensitivity to a different extent for different target molecules in a sample-specific manner. This heterogeneity in detection sensitivity is often overlooked and leads to an inflated false positive rate. We propose a novel sensitivity adjusted likelihood-ratio test (SALT), which properly controls the false positives and is more powerful than the unadjusted approach. We show that sample-and-feature-specific detection sensitivity can be well estimated from NanoString nCounter data, and using the estimated sensitivity in SALT results in improved biomarker screening.

Pilot implementation of the revised criteria for staging of Alzheimer's disease by the Alzheimer's Association Workgroup in a tertiary memory clinic.

We aimed to evaluate the feasibility of the 2024 Alzheimer's Association Workgroup's integrated clinical-biological staging scheme in outpatient settings within a tertiary memory clinic. The 2018 syndromal cognitive staging system, coupled with a binary biomarker classification, was implemented for 236 outpatients with cognitive concerns. The 2024 numeric clinical staging framework, incorporating biomarker staging, was specifically applied to 154 individuals within the Alzheimer's disease (AD) continuum. The 2024 staging scheme accurately classified 95.5% AD. Among these, 56.5% exhibited concordant clinical and biological stages (canonical), 34.7% demonstrated more advanced clinical stages than biologically expected (susceptible), and 8.8% displayed the inverse pattern (resilient). The susceptible group was characterized by a higher burden of neurodegeneration and inflammation than anticipated from tau, whereas the resilient group showed the opposite. The 2024 staging scheme is generally feasible. A discrepancy between clinical and biological stages is relatively frequent among symptomatic patients with AD. The 2024 AA staging scheme is generally feasible in a tertiary memory clinic. A discrepancy between clinical and biological stages is relatively frequent in AD. The mismatch may be influenced by a non-specific pathological process involved in AD. Individual profiles like aging and lifestyles may contribute to such a mismatch. Matched and mismatched cases converge toward similar clinical outcomes.

A prognostic and predictive computational pathology immune signature for ductal carcinoma in situ: retrospective results from a cohort within the UK/ANZ DCIS trial

The density of tumour-infiltrating lymphocytes (TILs) could be prognostic in ductal carcinoma in situ (DCIS). However, manual TIL quantification is time-consuming and suffers from interobserver and intraobserver variability. In this study, we developed a TIL-based computational pathology biomarker and evaluated its association with the risk of recurrence and benefit of adjuvant treatment in a clinical trial cohort. In this retrospective cohort study, a computational pathology pipeline was developed to generate a TIL-based biomarker (CPath TIL categories). Subsequently, the signature underwent a masked independent validation on H&E-stained whole-section images of 755 patients with DCIS from the UK/ANZ DCIS randomised controlled trial. Specifically, continuous biomarker CPath TIL score was calculated as the average TIL density in the DCIS microenvironment and dichotomised into binary biomarker CPath TIL categories (CPath TIL-high vs CPath TIL-low) using the median value as a cutoff. The primary outcome was ipsilateral breast event (IBE; either recurrence of DCIS [DCIS-IBE] or invasive progression [I-IBE]). The Cox proportional hazards model was used to estimate the hazard ratio (HR). CPath TIL-score was evaluable in 718 (95%) of 755 patients (151 IBEs). Patients with CPath TIL-high DCIS had a greater risk of IBE than those with CPath TIL-low DCIS (HR 2·10 [95% CI 1·39-3·18]; p=0·0004). The risk of I-IBE was greater in patients with CPath TIL-high DCIS than those with CPath TIL-low DCIS (3·09 [1·56-6·14]; p=0·0013), and the risk of DCIS-IBE was non-significantly higher in those with CPath TIL-high DCIS (1·61 [0·95-2·72]; p=0·077). A significant interaction (pinteraction=0·025) between CPath TIL categories and radiotherapy was observed with a greater magnitude of radiotherapy benefit in preventing IBE in CPath TIL-high DCIS (0·32 [0·19-0·54]) than CPath TIL-low DCIS (0·40 [0·20-0·81]). High TIL density is associated with higher recurrence risk-particularly of invasive recurrence-and greater radiotherapy benefit in patients with DCIS. Our TIL-based computational pathology signature has a prognostic and predictive role in DCIS. National Cancer Institute under award number U01CA269181, Cancer Research UK (C569/A12061; C569/A16891), and the Breast Cancer Research Foundation, New York (NY, USA).

Comprehensive modelling of prostate cancer progression: integrating continuous and binary biomarkers with event time data and missing covariates

Longitudinal studies involve recording observations at scheduled visits or time points for individuals until a predetermined event, like reaching satisfactory tumor shrinkage in cancer studies. Furthermore, dropout in longitudinal studies leads to incomplete data, which significantly increases the risk of bias. An amended joint shared-random effects model (SREM) is proposed for mixed continuous and binary longitudinal measurements and a time-to-event (TTE) outcome, incorporating missing covariates. In the proposed model, a conditional model is applied for the mixed continuous and binary longitudinal outcomes; a mixed effect model is considered for the continuous longitudinal outcome. For the binary longitudinal outcome, given the continuous longitudinal outcome, a logistic mixed effect model is considered. These models share common random effects with the model for the event time outcome. The model formulation is based on Bayesian statistical thinking via Markov Chain Monte Carlo (MCMC). The proposed joint modelling is applied to contribute to the understanding of the progression of prostate cancer (PCa) by considering a generalized linear mixed effects model for time-varying covariates that incorporate ignorable missingness. The association between prostate-specific antigen (PSA) with alkaline phosphatase (ALP) and tumor status has been studied with mixed conclusions.

Abstract 967: Real-world transcriptomic biomarkers as replacement for immunohistochemistry and FISH studies in breast cancer

Abstract Background: Human epidermal growth factor receptor-2 (HER2) and hormone receptors are typically used as binary biomarkers for selecting breast cancer therapy. There is a need to explore the clinical relevance of these biomarkers as continuous variables. This is particularly relevant for the new class of antibody-drug conjugates (ADC), in which a relatively low HER2 expression level is adequate for targeting tumor cells. We explored the potential of RNA profiling, determined by next generation sequencing (NGS), to provide more flexible clinical biomarkers as compared with immunohistochemistry (IHC) or fluorescent in situ hybridization (FISH). Methods: Clinical and laboratory data from 57 breast cancers collected by the COTA real-world data company were used to study biomarker levels as detected by routine clinical transcriptomic tests. HER2 (ERBB2), estrogen receptor alpha (ESR1), and androgen receptor (AR) mRNA levels were compared with reported HER2 and estrogen receptor (ER) IHC and FISH results. Results: RNA levels accurately reflected and predicted HER2 amplification (see table below). Importantly, RNA data showed significant variation and overlap in the levels of ERBB2 mRNA between cases scored by IHC as zero, 1+, and 2+. This variation correlated with progression-free survival (PFS). Similarly, the ESR1 RNA levels accurately reflected ER status and demonstrated significant variation between positive cases. RNA data also showed significant variations in the AR levels. Patients wssith high AR mRNA levels had significantly better PFS (P=0.05). Patients expressing high ER and AR levels had significantly better PFS than those expressing low ESR1 and AR levels (P=0.03). Conclusions: These findings suggest that RNA analysis using NGS is an alternative to IHC and FISH. RNA provides continuous data that can determine cut-off points predicting response to therapy and should be explored in predicting ADC response. ERBB2 mRNA levels (FPKM) in various HER2 IHC groupss IHC Score Valid N Mean Median Minimum Maximum Quartile Range Std.Dev. Zero 19 231 243 63 537 170 110 Zero vs one 3 151 155 45 253 208 104 Zero to one vs one 18 397 302 172 845 365 222 One Vs two 13 495 423 159 1094 273 284 Two vs three 4 7523 7649 937 13859 7183 5310 Citation Format: Maher Albitar, Andre Goy, Andrew Pecora, Deena Graham, Donna Donna McNamara, Ahmad Ahmad Charifa, Andrew IP, Wanlong Ma, Stanley Waintraub. Real-world transcriptomic biomarkers as replacement for immunohistochemistry and FISH studies in breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 967.

Efficient feature extraction from highly sparse binary genotype data for cancer prognosis prediction using an auto-encoder.

Genomics involving tens of thousands of genes is a complex system determining phenotype. An interesting and vital issue is how to integrate highly sparse genetic genomics data with a mass of minor effects into a prediction model for improving prediction power. We find that the deep learning method can work well to extract features by transforming highly sparse dichotomous data to lower-dimensional continuous data in a non-linear way. This may provide benefits in risk prediction-associated genotype data. We developed a multi-stage strategy to extract information from highly sparse binary genotype data and applied it for cancer prognosis. Specifically, we first reduced the size of binary biomarkers via a univariable regression model to a moderate size. Then, a trainable auto-encoder was used to learn compact features from the reduced data. Next, we performed a LASSO problem process to select the optimal combination of extracted features. Lastly, we applied such feature combination to real cancer prognostic models and evaluated the raw predictive effect of the models. The results indicated that these compressed transformation features could better improve the model's original predictive performance and might avoid an overfitting problem. This idea may be enlightening for everyone involved in cancer research, risk reduction, treatment, and patient care via integrating genomics data.

Plasma pTau181/Aβ42 identifies cognitive change earlier than CSF pTau181/Ab42

AbstractBackgroundPlasma biomarkers now show an accuracy in detecting Amyloid Beta (Aβ) similar to AD biomarkers derived from cerebral spinal fluid (CSF). However, the ability of plasma AD biomarkers, alone or in combination, to predict cognitive decline has not yet been compared to that of CSF AD biomarkers.MethodPlasma biomarker data from 233 participants’ first visit in the Australian Imaging, Biomarkers and Lifestyle study (AIBL) was submitted to linear mixed effects models (LME) to quantify the relationship with change in cognition (measured using the AIBL PACC) and in clinical disease stage (CDR SoB) in both PET Aβ‐ (Centiloid value <20CL) and Aβ+ (Centiloid value ≥20CL) participant subgroups. Separate models were used to assess CSF (Elecsys) and plasma (ADx NeuroSciences) data for Aβ42, pTau181 and the pTau181/Aβ42 ratio. Biomarker values were classified into low vs high levels based on ROC‐derived thresholds optimizing separation of PET Aβ status (low vs high at 20 CL). Changes in cognitive and clinical symptoms were then compared between the low/high plasma biomarker groups.ResultIn Aβ‐ participants, no significant interactions between binary biomarker classification and time were observed for AIBL PACC or CDR SoB, for either CSF or plasma biomarkers. In the Aβ+ participants, interactions between the binary plasma biomarker classification and change in cognition were greater in magnitude that those detected for CSF biomarker classification. For plasma, abnormally high values of both pTau181 and the pTau181/Aβ42 ratio predicted a significant increase over time in CDR SoB (Figure 1H & 1L) and a significant decrease over time in the AIBL PACC score (Figure 1F & 1J), compared the group with low values on the same biomarkers. In cognitively unimpaired Aβ‐ participants, the AIBL PACC score declined in those with abnormally high values of the pTau181 and the pTau181/Aβ42 ratio (Figure 1F & 1J).ConclusionAssays to measure pTau181 and Aβ42 in the plasma possess an accuracy equivalent to those derived from CSF. In particular, abnormally high levels of plasma pTau181 or the ratio of pTau181 to Aβ42 ratio provide a strong prediction of early cognitive changes, even in those with normal PET Aβ status.

Propensity score-based adjustment for covariate effects on classification accuracy of bio-marker using ROC curve

The potential performance of bio-marker in classifying diseased from healthy population may be affected by baseline covariates (X) that are associated with both the bio-marker (Y) and the disease status (D). Some existing approaches can be able to adjust for the effect of a single covariate at a time. However, several potential covariates can be available in practice for which simultaneous adjustment in the ROC curve is essential. This study proposed a propensity score (PS) based adjustment for the effects of several covariates in the ROC curve. The PS is first derived from a linear transformation of several covariates and the PS-adjusted (and PS-specific) ROC curve was then estimated using the existing non-parametric induced ROC regression framework. The method is illustrated for both continuous and binary bio-markers. The simulation study suggests that the PS-based adjustment performed well by providing a consistent estimate of the true ROC curve and showing robustness to the mis-specification of the propensity score model as well as to a non-linear function of covariates. Further, an application of the method is provided to evaluate the effectiveness of the body-mass-index in classifying patients with hypertension or diabetes after adjusting for the potential covariates such as age, sex, education, socio-economic status.

Optimal Combinations of AT(N) Biomarkers to Determine Longitudinal Cognition in the Alzheimer's Disease.

Background: National Institute on Aging—Alzheimer's Association (NIA-AA) proposed the AT(N) system based on β-amyloid deposition, pathologic tau, and neurodegeneration, which considered the definition of Alzheimer's disease (AD) as a biological construct. However, the associations between different AT(N) combinations and cognitive progression have been poorly explored systematically. The aim of this study is to compare different AT(N) combinations using recognized biomarkers within the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort.Methods: A total of 341 participants were classified into cognitively unimpaired (CU; n = 200) and cognitively impaired (CI; n = 141) groups according to the clinical manifestations and neuropsychological tests. Cerebrospinal fluid (CSF) Aβ42 and amyloid-PET ([18F]flutemetamol) were used as biomarkers for A; CSF phosphorylated tau (p-tau) and tau-PET ([18F]flortaucipir) were used as biomarkers for T; CSF total tau (t-tau), hippocampal volume, temporal cortical thickness, [18F]fluorodeoxyglucose (FDG) PET, and plasma neurofilament light (NfL) were used as biomarkers for (N). Binary biomarkers were obtained from the Youden index and publicly available cutoffs. Prevalence of AT(N) categories was compared between different biomarkers within the group using related independent sample non-parametric test. The relationship between AT(N) combinations and 12-year longitudinal cognition was assessed using linear mixed-effects modeling.Results: Among the CU participants, A–T–(N)– was most common. More T+ were detected using p-tau than tau PET (p < 0.05), and more (N)+ were observed using fluid biomarkers (p < 0.001). A+T+(N)+ was more common in the CI group. Tau PET combined with cortical thickness best predicted cognitive changes in the CI group and MRI predicted changes in the CU group.Conclusions: These findings suggest that optimal AT(N) combinations to determine longitudinal cognition differ by cognitive status. Different biomarkers within a specific component for defining AT(N) cannot be used identically. Furthermore, different strategies for discontinuous biomarkers will be an important area for future studies.

Oh brother, where art tau? Amyloid, neurodegeneration, and cognitive decline without elevated tau

Mild cognitive impairment (MCI) can be an early manifestation of Alzheimer’s disease (AD) pathology, other pathologic entities [e.g., cerebrovascular disease, Lewy body disease, LATE (limbic-predominant age-related TDP-43 encephalopathy)], or mixed pathologies, with concomitant AD- and non-AD pathology being particularly common, albeit difficult to identify, in living MCI patients. The National Institute on Aging and Alzheimer’s Association (NIA-AA) A/T/(N) [β-Amyloid/Tau/(Neurodegeneration)] AD research framework, which classifies research participants according to three binary biomarkers [β-amyloid (A+/A-), tau (T+/T-), and neurodegeneration (N+/N-)], provides an indirect means of identifying such cases. Individuals with A+T-(N+) MCI are thought to have both AD pathologic change, given the presence of β-amyloid, and non-AD pathophysiology, given neurodegeneration without tau, because in typical AD it is tau accumulation that is most tightly linked to neuronal injury and cognitive decline. Thus, in A+T-(N+) MCI (hereafter referred to as “mismatch MCI” for the tau-neurodegeneration mismatch), non-AD pathology is hypothesized to drive neurodegeneration and symptoms, because β-amyloid, in the absence of tau, likely reflects a preclinical stage of AD. We compared a group of individuals with mismatch MCI to groups with A+T+(N+) MCI (or “prodromal AD”) and A-T-(N+) MCI (or “neurodegeneration-only MCI”) on cross-sectional and longitudinal cognition and neuroimaging characteristics. β-amyloid and tau status were determined by CSF assays, while neurodegeneration status was based on hippocampal volume on MRI. Overall, mismatch MCI was less “AD-like” than prodromal AD and generally, with some exceptions, more closely resembled the neurodegeneration-only group. At baseline, mismatch MCI had less episodic memory loss compared to prodromal AD. Longitudinally, mismatch MCI declined more slowly than prodromal AD across all included cognitive domains, while mismatch MCI and neurodegeneration-only MCI declined at comparable rates. Prodromal AD had smaller baseline posterior hippocampal volume than mismatch MCI, and whole brain analyses demonstrated cortical thinning that was widespread in prodromal AD but largely restricted to the medial temporal lobes (MTLs) for the mismatch and neurodegeneration-only MCI groups. Longitudinally, mismatch MCI had slower rates of volume loss than prodromal AD throughout the MTLs. Differences in cross-sectional and longitudinal cognitive and neuroimaging measures between mismatch MCI and prodromal AD may reflect disparate underlying pathologic processes, with the mismatch group potentially being driven by non-AD pathologies on a background of largely preclinical AD. These findings suggest that β-amyloid status alone in MCI may not reveal the underlying driver of symptoms with important implications for enrollment in clinical trials and prognosis.

Biomarker analysis from the KEYNOTE-199 trial of pembrolizumab in patients (pts) with docetaxel-refractory metastatic castration-resistant prostate cancer (mCRPC).

5526 Background: In the phase II KEYNOTE-199 study (NCT02787005), pembrolizumab monotherapy demonstrated antitumor activity in pts with docetaxel-refractory mCRPC (n = 258). Here we evaluated the association between prespecified molecular biomarkers and clinical outcomes. Methods: Cohorts 1 (C1) and 2 (C2) enrolled pts with RECIST-measurable PD-L1–positive (combined positive score [CPS] ≥1 using immunohistochemistry) and PD-L1–negative (CPS &lt;1) disease, respectively. C3 enrolled pts with nonmeasurable, bone-predominant disease, irrespective of PD-L1 status. Biomarkers evaluated in this analysis were tumor mutational burden ([TMB; mutations/exome] n = 155), PD-L1 CPS (n = 255), tumor microenvironment–based 18-gene RNA expression profile ([GEP] n = 196), and microsatellite instability ([MSI] as determined by Promega PCR analysis; n = 147). Outcomes evaluated for C1 and C2 (n = 200) were ORR, disease control rate (DCR), and radiographic PFS (rPFS) per blinded, independent central review per PCWG-modified RECIST v1.1. Outcomes evaluated for C1-C3 (n = 258) were prostate-specific antigen (PSA) response, time to PSA progression, and OS. Significance of continuous biomarkers (CPS; TMB; GEP) was prespecified at 0.05 for one-sided P values from logistic (ORR; DCR; PSA response) and Cox proportional hazard regression (rPFS; OS; PSA progression) adjusted for Eastern Cooperative Oncology Group performance status. Binary biomarkers (MSI) were analyzed using Fisher’s exact test (ORR; DCR; PSA response). Clinical data cutoff date: Jun 24, 2019. Results: Median TMB was 53.0 (interquartile range [IQR], 40.5 to 78.0), median CPS was 1 (IQR, 0 to 5), and median GEP was –0.64 (IQR, –0.88 to –0.46); 6 pts (2.3%) had MSI-high tumors. In C1-C3, TMB was associated with PSA response (one-sided nominal P = 0.0016) and time to PSA progression (one-sided nominal P = 0.00092). In C1-C3, PD-L1 CPS was associated with PSA response (one-sided nominal P = 0.046) and time to PSA progression (one-sided nominal P = 0.021). In C1-C3, GEP was not significantly associated with response. In C1-C3, MSI was associated with PSA response (one-sided nominal P = 0.019). Conclusions: In this biomarker analysis from KEYNOTE-199 C1-C3, TMB and PD-L1 CPS were associated with better PSA response; however, small pt numbers limit definitive conclusions on ORR, DCR, and OS. Further evaluation of molecular biomarkers in pts with mCRPC treated with pembrolizumab is warranted. Clinical trial information: NCT02787005 .

Real Time Intraoperative Functional Brain Mapping Based on RGB Imaging

Intraoperative optical imaging is a localization technique for the functional areas of the human brain cortex during neurosurgical procedures. However, it still lacks robustness to be used as a clinical standard. In particular new biomarkers of brain functionality with improved sensitivity and specificity are needed. We present a method for the real time identification of the activated cortical areas based on the analysis of the cortical hemodynamic using a RGB camera and a white light source. We measure the quantitative oxy and deoxy-hemoglobin concentration changes in the human brain cortex with the modified Beer-Lambert law and Monte Carlo simulations. A functional model has been implemented to define in real time a binary biomarker of the cortical activation following neuronal activation by physiological stimuli. The results show a good correlation between the computed activation maps and the brain areas localized by electrical brain stimulation. We demonstrate that a RGB camera combined with a quantitative modeling of brain hemodynamics biomarkers can evaluate in real time the functional areas during neurosurgery and serve as a tool of choice to complement electrical brain stimulation.

Combining binary and continuous biomarkers by maximizing the area under the receiver operating characteristic curve

In any clinical case, a decision is made with the maximum possible accuracy. To achieve such accuracy, in the presence of multiple diagnostic tests or biomarkers, biomarker combinations aim to achieve maximum accuracy. As existing biomarker combination methods combine only continuous biomarkers, therefore in this study biomarker combination for binary biomarkers was created by suggesting an approach using Youden’s J statistic for combining binary biomarkers. The proposed approach will facilitate binary and continuous biomarker combinations. A simulation study was conducted to compare the performance of our proposed combination approach according to different sample sizes. Both in the analysis of real data and the simulation studies for different samples, the proposed approach has been shown to yield favorable results and higher area under the curve.

Comparison of crossover and parallel-group designs for the identification of a binary predictive biomarker of the treatment effect.

Pros and cons of crossover design are well known for estimating the treatment effect compared to parallel-group design, but remain unclear for identifying and estimating an interaction between a potential biomarker and the treatment effect. Such 'predictive' biomarkers, or 'effect modifiers', help to predict the response to specific treatments. The purpose of this report was to better characterize the advantages and disadvantages of crossover versus parallel-group design to identify predictive biomarkers. The treatment effect, the effect of a binary biomarker and their interaction were modelled using a linear model. The intra-subject correlation in the crossover design was taken into account through an intra-class correlation coefficient. The variance-covariance matrix of the parameters was derived and compared. For both trial designs, the variance of the parameter estimating an interaction between the treatment effect and a potential predictive biomarker corresponds to the variance of the parameter estimating the treatment effect, multiplied by the inverse of the frequency of the candidate biomarker. The ratio of the variance of the interaction parameter in the crossover to the variance estimated in the parallel-group design depends on the complement of the intra-class correlation coefficient. When planning a clinical trial including a search for candidate biomarker, the frequency of the candidate biomarker helps design the sample size, and the intra-subject correlation of the outcome should be taken into account for choosing between parallel-group and crossover designs.

An approach to use of an adaptive procedure to clinical trials for molecularly heterogenous subject selection at interim

For clinical trials, molecular heterogeneity has played a more important role recently. Many novel clinical trial designs prospectively incorporate molecular information to evaluation of treatment effects. In this paper, an adaptive procedure incorporating a non-pre-specified genomic biomarker is employed in the interim of a conventional trial. A non-pre-specified binary genomic biomarker, which is predictive of treatment effect, is used to classify study patients into two mutually exclusive subgroups at the interim review. According to the observations at the interim stage, adaptations such as adjusting sample size or shifting eligibility of study patients are then made in case of different scenarios.

Power estimation in biomarker studies where events are already observed.

Background The clinical utility of a new biomarker should ideally be established in a prospective randomized clinical trial. However, such trials are not always practical. As such, it is common for investigators to identify promising biomarkers using archived specimens and clinical data collected from previously completed therapeutic trials. Simon et al. defined such biomarker studies as prospective-retrospective studies and proposed specific conditions to satisfy for such evaluations to be more than hypothesis generating. One condition they proposed is that archived tissues must be available on a sufficiently large number of patients from the pivotal trials to ensure adequately powered analyses. Purpose The purpose of this article is to provide a new perspective on how to estimate power for assessing the prognostic and predictive values of a single binary biomarker in prospective-retrospective biomarker studies. Computer programs are provided to facilitate the use of these methods in practice. Methods The proposed methods utilize additional information that becomes available during the course of the treatment trial including sample size, accrual time, additional follow-up time, and the observed number of events at time of biomarker analysis. These methods involve solving for the exponential hazard rates that give rise to the event numbers that are consistent with those observed while satisfying other design parameter constraints. Conclusion Simon et al. proposed a new paradigm for biomarker design, conduct, analysis, and evaluation in prospective-retrospective studies that offers a more efficient alternative than fully prospective biomarker studies. As a general rule, they suggest that samples from at least two-thirds of the patients be available for analysis. In this article, I expand on this issue and provide a methodological tool useful for estimating study power in prospective-retrospective biomarker studies. It is my hope that these incremental efforts to improve the quality and statistical rigor in biomarker studies will hasten the introduction of useful tumor biomarkers into clinical practice.

Abstract CT122: BCL2 expression is a potential predictive biomarker of response to venetoclax in combination with bortezomib and dexamethasone in patients with relapsed/refractory multiple myeloma

Abstract The anti-apoptotic proteins BCL-2 and MCL-1 have been shown to promote multiple myeloma (MM) cell survival. Venetoclax (VEN) is a potent, selective, and orally bioavailable small-molecule inhibitor of BCL-2. Bortezomib (BTZ) is a proteasome inhibitor that can inhibit MCL-1 activity by increasing the MCL-1 antagonist, NOXA. Results presented herein describe correlative biomarker analyses in the ongoing phase 1b study of VEN in combination with BTZ and dexamethasone in patients with relapsed/refractory (R/R) MM (NCT01794507). As of 19 Aug 2016, 66 patients were enrolled on study. Baseline bone marrow aspirate samples were available from 52 patients, of which 45 were evaluable for BCL-2 family gene expression by droplet digital PCR in CD138-selected tumor cells. Correlation between BCL2 (BCL-2), BCL2L1 (BCL-XL) and MCL1 (MCL-1) mRNA expression (log2-transformed copies/ul normalized to housekeeping gene) and preliminary efficacy [overall response rate (ORR), time to disease progression (TTP) and duration of response (DoR)] were examined by Log-rank and Wilcoxon tests for binary biomarkers, and by risk ratio from Cox proportional hazard model for continuous biomarkers. The ORR was 68% (44/65) for all evaluable patients and 89% (31/35) in patients who had 1-3 prior therapies (31/35). A broad range of BCL2, BCL2L1 and MCL1 expression was observed, however higher BCL2 levels were detected in patients who achieved a partial response (PR) or better (median: 3.01 vs 0.87, p&amp;lt;0.01). Additionally, higher BCL2 levels were observed in patients who had 1-3 prior lines of therapy compared to 4 or more lines of therapy (median: 3.03 vs 0.94, p&amp;lt;0.01). In contrast, no association was observed between BCL2L1 or MCL1 gene expression and response or number of prior therapies. Bootstrapping and aggregating thresholds from trees was used to estimate a threshold value for BCL2 expression that would provide optimum selection of patients likely to have a response. Overall, seventeen of 18 patients with high BCL2 expression (≥3.0) achieved at least a PR (ORR 94%), with 12 patients (66%) achieving VGPR or better. Sixteen of 27 patients with low BCL2 expression achieved at least a PR (ORR 59%), with 6 patients (22%) achieving a VGPR or better. Median TTP (11.6 vs 5.7 months) and DoR (10.2 vs 7 months) were longer for patients with high versus low BCL2 expression. Responses in high BCL2 expressors were independent of cytogenetic status as determined by interphase FISH analysis, including t(11;14), t(4;14), del(13q) and del(17p). Targeting BCL-2 and MCL-1 with the combination of VEN, BTZ and dexamethasone provides a unique approach for MM treatment. Efficacy results in tumors expressing high BCL2 levels, including 94% ORR, provide supportive evidence for the evaluation of this combination regimen in the ongoing phase 3 study (NCT02755597) in R/R MM. Citation Format: Jeremy Ross, Brenda Chyla, Rasna Goswami, Lisa Roberts-Rapp, Yan Sun, Yanwen Jiang, Elizabeth Punnoose, Maria Verdugo, Anahita Bhathena, Paulo Maciag. BCL2 expression is a potential predictive biomarker of response to venetoclax in combination with bortezomib and dexamethasone in patients with relapsed/refractory multiple myeloma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr CT122. doi:10.1158/1538-7445.AM2017-CT122

Thresholding of a Continuous Companion Diagnostic Test Confident of Efficacy in Targeted Population

ABSTRACTIn personalized medicine, continuous biomarker values are often dichotomized to classify patients into target and nontarget populations. For example, baseline hemoglobin A1c (A1C) and Positive and Negative Syndrome Scale (PANSS) measurements are potential biomarkers in diabetes and psychiatry, respectively. For Alzheimer's Disease (AD), currently Mild versus Moderate classification is used as a potential binary biomarker for treatment effect prediction. However, Normal, Mild, Moderate, and Severe staging of AD is a discretization of the mini-mental state examination (MMSE), which is measured on a scale of 0–30. Therefore, MMSE is potentially a continuously valued biomarker for AD treatment. For such scenarios, we provide simultaneous confidence intervals for efficacy corresponding to all candidate thresholds. Our method allows for rigorous and flexible decision making, taking into consideration medical impact based on both the size of the target population and efficacy in the target population.

Comparison of different clinical development plans for confirmatory subpopulation selection

Given ever increasing costs to develop a new drug and intense competition, adaptive enrichment designs are an attractive option for a development program that allows selecting a potential subgroup defined by a binary biomarker. Such designs explicitly factor in the possibility that the new drug might differentially benefit distinct biomarker subgroups. We have compared three clinical development plans for a time-to-event endpoint, such as overall survival, that all lead to a decision in a pivotal trial either in all comers only, in allcomers and biomarker positive, in the biomarker positive only, or to declare the drug futile. The decision about which hypothesis to test at the final analysis is made based on a fast time-to-event endpoint, such as progression-free survival, at an interim analysis. We quantify the time gain when using an adaptive enrichment Phase II/III design versus alternative development approaches and we outline what type of biomarker needs to be available prior to Phase II in each scenario. We conclude with a discussion of further features of each of the considered development plans.

Biomarker and Ex Vivo Sensitivity Analysis of Venetoclax (ABT-199/GDC-0199) in Combination with Bortezomib and Dexamethasone in Relapsed/Refractory Multiple Myeloma

Introduction: Multiple myeloma (MM) tumor cell survival is largely dependent on a complex network of interactions between pro-survival (BCL-2, BCL-XL, MCL-1) and pro-death (BIM, NOXA, BID, BAD, BAK, BAX) BCL-2 family members. Venetoclax (VEN) is a potent, selective, orally bioavailable small-molecule BCL-2 inhibitor. Bortezomib (BTZ) is a proteasome inhibitor that can impede MCL-1 activity by increasing the MCL-1 antagonist NOXA. The addition of VEN has been shown to enhance the activity of BTZ in MM cell line and xenograft models. Chromosomal abnormalities (immunoglobulin heavy chain translocations, amplification, deletions) can also influence sensitivity of MM cells to treatment. Results presented herein describe retrospective biomarker analysis in a subset of dose-escalation patients (pts) from an ongoing phase 1b study of VEN in combination with BTZ and dexamethasone in relapsed/refractory (R/R) MM.Methods: Exploratory objectives included evaluation of chromosomal abnormalities, BCL-2 family member gene expression levels, and ex vivosensitivity of pt MM tumor cells to VEN and/or BTZ. Interphase FISH analysis was performed on CD138-positively selected bone marrow mononuclear cells (BMMC) using probes for t(11;14), t(4;14), deletion 17p (17p-), deletion 13q (13q-), and amplification of chromosomes 5, 9, or 15 (denoted as hyperdiploid [HY]). Absolute quantitation of BCL-2 family members (BCL-2, BCL-XL, MCL-1, BCL-2A1, BIM, BID, BAD, BAX, and NOXA) was performed on CD138-selected BMMCs collected at baseline using TaqMan probe-based droplet digital polymerase chain reaction. Ex vivosensitivity of BMMCs collected at baseline to increasing doses of VEN and/or BTZ was determined by flow cytometry using a combined analysis for the loss of CD138 surface expression and the alteration of cellular morphology (lower FCS). Correlations between biomarkers and median best percentage change in M-protein were examined by Spearman test. Correlations between biomarkers and median time on study (mToS) were examined by log-rank and Wilcoxon tests for binary biomarkers, and by risk ratio from a Cox proportional hazard model for continuous biomarkers. No correction for multiple testing was performed.Results: Forty-one pts were enrolled as of 06/15/2015. The biomarker study subgroups were similar to the overall pt population with respect to demographics. Of the 37/41 (90%) pts assessed by FISH, 5 were t(11;14), 3 were t(4;14), 20 were 13q-, 10 were 17p-, and 15 were HY. In pts with HY, median best percentage change in M-protein was -79.4% compared to +101.65% in non-HY, while mToS was 309 days compared to 29 days in non-HY (p <0.0001), consistent with literature regarding a positive prognostic effect in MM. No difference in efficacy was observed in other FISH subtypes. BCL-2 family gene expression levels were determined in 25/41 (61%) pts. BCL-2, but not BCL-XL or MCL-1 gene expression, correlated with M-protein response (Spearman -0.5116, p=0.0149) and with a trend for increased mToS (HR=0.42 [95% CI 0.13-1.0]). Responders (≥ partial response, n=13; assessed by International Myeloma Working Group criteria 2011) had higher BCL-2 levels than subjects with stable disease/progressive disease (n=12; p=0.015). Pts with high BCL-XL or high MCL-1 were found in both the responder and nonresponder populations. Of the 14 pt samples collected for ex vivo sensitivity, 8 had sufficient CD138-positive cells for VEN ± BTZ dose curve analysis. Four of 8 (50%) were sensitive to VEN alone (median lethal dose [LD50] ≤300 nM) while 4/8 (50%) were sensitive to BTZ alone (LD50 ≤5 nM). Seven of 8 (88%) were sensitive to the combination (LD50 ≤5 nM BTZ/300 nM VEN).Conclusions: VEN in combination with BTZ and dexamethasone is active in R/R MM. CD138-enriched BMMCs were obtained in sufficient numbers to enable downstream FISH analysis, BCL-2 family gene expression profiling, and ex vivo sensitivity evaluation. The exploratory biomarker analysis showed that higher BCL-2 gene expression levels at baseline significantly correlated with clinical responses to therapy, with a trend toward increased mToS. Additionally, MM cells treated ex vivo demonstrated greater sensitivity to the combination of VEN/BTZ than to the 2 single agents alone. Further investigation of this dataset continues, including analysis of on-treatment tumor samples in a subset of pts. DisclosuresRoss:AbbVie: Employment, Equity Ownership. Off Label Use: Venetoclax is an investigational drug that is not yet approved in this indication.. Chyla:AbbVie: Employment, Equity Ownership. Roberts-Rapp:AbbVie: Employment, Equity Ownership. Goswami:AbbVie: Employment, Equity Ownership. Yan:AbbVie: Employment, Equity Ownership. Huang:AbbVie: Employment, Equity Ownership. Bhathena:AbbVie: Employment, Equity Ownership. Enschede:AbbVie: Employment, Equity Ownership. Maciag:AbbVie: Employment, Equity Ownership. Humerickhouse:AbbVie: Employment, Equity Ownership. McKeegan:AbbVie: Employment, Equity Ownership.