Nonresponse Models Research Articles

Abstract 5086: Functional profiling in lung tumor explants show contributions of the tumor microenvironment to response of Dato-DXdex vivo

Abstract Background: Antibody drug conjugates (ADCs) have demonstrated significant antitumor activity against multiple treatment refractory cancers. While Trastuzumab deruxtecan, a HER2 directed ADC, has shown impressive clinical activity in HER2-expressing cancers, the ability to predict responses of ADCs by IHC has been limited. Dato-DXd is a TROP2 directed ADC that is currently being studied in a number of registrational phase 3 trial incld. TROPION-Lung01. In the phase 1b TROPION-Pan tumor 01 trial (TP01) response to Dato-DXd did not correlate to TROP2 IHC expression. Here we investigate molecular determinants of sensitivity to Dato-DXd using short-term microfluidic culture of patient derived organotypic tumor spheroids (PDOTS). Methods: Surgical NSCLC cases collected from Brigham and Women’s Hospital and St. Elizabeth’s Medical Center under an IRB approved protocol were studied. PDOTS were generated as previously described. Ex vivo response was assessed by live/dead imaging, TROP2 antigen density on EPCAM+ cells by flow cytometry (FCM) and by immunofluorescence (IF). T cell and myeloid cell populations were analyzed by FCM of CD45+ cells isolated during tumor preparation. For a subset of samples single cell RNA sequencing (scRNAseq) was performed. Results:Eleven NSCLC explants were studied. Two specimens were excluded due to excess of fibrotic material. Response to Dato-DXd was measured by change in raw live cell area compared to control samples. Median response to Dato-DXd was -48.5% (ranging from +33.9% to -77.8%). Using a cut-off of -40%, four cases were characterized as non-responders (NR) and 5 as responders (R). Tumor TROP2 expression by IF ranged from 0 to 3+ with no relationship between TROP2 expression and response being observed. TROP2 staining positivity by IF didn’t correlate to Dato-DXd sensitivity ex vivo. TROP2 FCM, on the other hand discriminated between responsive and non-responsive models in 100% of the cases, using a cutoff of MFI >13000. Furthermore, we observed an increase of tumor CD8 abundance in Dato-DXd R compared to NR (median %CD8 in R 10.3% (range 5.1%-62.5%); median %CD8 of live cells in NR 2.3% (range 1.2%-3.7%) p-value 0.029). Conversely, we also observed a trend of higher granulocytic myeloid-derived suppressor cells abundance in NR compared to R (NR median %gMDSC of CD15+ cells 84.1% (range 73.4%-90.2%) ; R median %gMDSC 61.7% (range 54.3%-90.6%); p-value 0.069). In 2 cases with high and low TROP2 MFI available scRNAseq data showed TROP2 expression exclusively in tumor cells, with further analysis ongoing. Conclusion: By functionally profiling NSCLC PDOTS that retain immune cells, we correlated multiple parameters with ex vivo response to Dato-DXd. Our results have important implications for precision deployment of ADCs and suggest that target tumor antigen density in conjunction with addition immune cell features may refine patient selection strategy. Citation Format: Satoru Yasuda, Elena Ivanova, Ha V. Vo, Sophie Kivlehan, Iliana Gjeci, Minh Ha, Patrick Lizotte, Raphael Bueno, David A. Barbie, Cloud P. Paweletz. Functional profiling in lung tumor explants show contributions of the tumor microenvironment to response of Dato-DXdex vivo [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5086.

Closed-form estimates for missing counts in multidimensional incomplete tables

A useful technique for analyzing incomplete tables is to model the missing data mechanisms of the variables using log-linear models. In this paper, we use log-linear parametrization and propose estimation methods for arbitrary three-way and n-dimensional incomplete tables. All possible cases in which data on one or more of the variables may be missing are considered. We provide simple closed form estimates of expected cell counts and parameters for the various missing data models. We also obtain explicit boundary estimates under nonignorable nonresponse models. Finally, a real-life dataset is analyzed to illustrate our results for modelling and estimation in multidimensional incomplete tables.

Hierarchical Bayesian non-response models for error rates in forensic black-box studies.

Forensic science plays a critical role in the United States criminal legal system. Historically, however, most feature-based fields of forensic science, including firearms examination and latent print analysis, have not been shown to be scientifically valid. Recently, black-box studies have been proposed as a means of assessing whether these feature-based disciplines are valid, at least in terms of accuracy, reproducibility and repeatability. In these studies, forensic examiners frequently either do not respond to every test item or select an answer equivalent to 'don't know'. Current black-box studies do not account for these high levels of missingness in statistical analyses. Unfortunately, the authors of black-box studies typically do not share the data necessary to meaningfully adjust estimates for the high proportion of missing responses. Borrowing from work in the context of small area estimation, we propose the use of hierarchical Bayesian models that do not require auxiliary data to adjust for non-response. Using these models, we offer the first formal exploration of the impact that missingness is playing in error rate estimations reported in black-box studies. We show that error rates currently reported as low as 0.4% could actually be at least 8.4% in models accounting for non-response where inconclusive decisions are counted as correct, and over 28% when inconclusives are counted as missing responses. These proposed models are not the answer to the missingness problem in black-box studies. But with the release of auxiliary information, they can be the foundation for new methodologies to adjust for missingness in error rate estimations. This article is part of the theme issue 'Bayesian inference: challenges, perspectives, and prospects'.

BRCA1-methylated triple negative breast cancers previously exposed to neoadjuvant chemotherapy form RAD51 foci and respond poorly to olaparib.

About 15% of Triple-Negative-Breast-Cancer (TNBC) present silencing of the BRCA1 promoter methylation and are assumed to be Homologous Recombination Deficient (HRD). BRCA1-methylated (BRCA1-Me) TNBC could, thus, be eligible to treatment based on PARP-inhibitors or Platinum salts. However, their actual HRD status is discussed, as these tumors are suspected to develop resistance after chemotherapy exposure. We interrogated the sensitivity to olaparib vs. carboplatin of 8 TNBC Patient-Derived Xenografts (PDX) models. Four PDX corresponded to BRCA1-Me, of which 3 were previously exposed to NeoAdjuvant-Chemotherapy (NACT). The remaining PDX models corresponded to two BRCA1-mutated (BRCA1-Mut) and two BRCA1-wild type PDX that were respectively included as positive and negative controls. The HRD status of our PDX models was assessed using both genomic signatures and the functional BRCA1 and RAD51 nuclear foci formation assay. To assess HR restoration associated with olaparib resistance, we studied pairs of BRCA1 deficient cell lines and their resistant subclones. The 3 BRCA1-Me PDX that had been exposed to NACT responded poorly to olaparib, likewise BRCA1-WT PDX. Contrastingly, 3 treatment-naïve BRCA1-deficient PDX (1 BRCA1-Me and 2 BRCA1-mutated) responded to olaparib. Noticeably, the three olaparib-responsive PDX scored negative for BRCA1- and RAD51-foci, whereas all non-responsive PDX models, including the 3 NACT-exposed BRCA1-Me PDX, scored positive for RAD51-foci. This suggested HRD in olaparib responsive PDX, while non-responsive models were HR proficient. These results were consistent with observations in cell lines showing a significant increase of RAD51-foci in olaparib-resistant subclones compared with sensitive parental cells, suggesting HR restoration in these models. Our results thus support the notion that the actual HRD status of BRCA1-Me TNBC, especially if previously exposed to chemotherapy, may be questioned and should be verified using the BRCA1- and RAD51-foci assay.

Abstract 526: Novel mechanisms of Neuropilin-1 inhibition result in improved tumor growth inhibition in vivo

Abstract Background Cygnal Therapeutics, pioneer in exoneuronal biology, has identified the neuronal protein, Neuropilin-1 (NRP1) as a critical regulator of tumor growth and anti-tumor immunity. NRP1 is a co-receptor that complexes with diverse ligands and their cognate receptors. As such, it plays a role in multiple different biological processes, including axon guidance and angiogenesis. NRP1 contains two CUB domains (a1 and a2) involved in binding the ligand Semaphorin3A (SEMA3A), two Factor V/VIII domains (b1 and b2) involved in VEGF ligand binding and one MAM domain (c domain). While functional antibodies with anti-tumor activity have been generated against the SEMA3A and VEGF binding domains, targeting other domains of NRP1 has not been well-studied. For example, the c-domain of NRP1 has been implicated in the dimerization of NRP1, a prerequisite for functionality. Thus c-domain blockers might inhibit multiple signaling pathways required for tumor growth. Alternatively, binders targeting the interface of two different domains might allow for simultaneous blocking of multiple ligands. We therefore hypothesized that these novel binders to alternative regions of NRP1 such as the b1/b2 interface and the c-domain would have improved and significant anti-tumor activity. Methods Antibodies with diverse binding characteristics were tested for in vivo anti-tumor activity in multiple syngeneic models, including anti-PD1 non-responsive models. Ability of these of the anti-NRP1 antibodies to elicit an adaptive immune response was evaluated. Results Novel binders to alternative regions of NRP1 such as domain interfaces and the c-domain resulted in significant tumor growth inhibition in multiple syngeneic models. Additionally, the NRP1 antagonists induced a robust CD8 T cell response, indicating immune-stimulatory role of these antibodies. Conclusions Cygnal has developed fully human monoclonal antibodies against previously untargeted domains of NRP1. These novel domain binders show remarkable inhibition of tumor growth in multiple syngeneic models and offer a novel means of therapeutic intervention in patients. Citation Format: Shalini Sethumadhavan, Eric Zhu, Aaron Fulgham, Mandana Abbassi, Ameya Apte, Tiffany Liao, Chengfeng Merriman, Katherine Molloy, Caitlin Stein, Amber Hanna, James Geoghegan, Bianka Prinz, Hongyue Dai, Jonathan Hurov, Pearl Huang, Daniel Blom. Novel mechanisms of Neuropilin-1 inhibition result in improved tumor growth inhibition in vivo [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 526.

An item response tree model with not-all-distinct end nodes for non-response modelling.

The non-response model in Knott etal. (1991, Statistician, 40, 217) can be represented as a tree model with one branch for response/non-response and another branch for correct/incorrect response, and each branch probability is characterized by an item response theory model. In the model, it is assumed that there is only one source of non-responses. However, in questionnaires or educational tests, non-responses might come from different sources, such as test speededness, inability to answer, lack of motivation, and sensitive questions. To better accommodate such more realistic underlying mechanisms, we propose a a tree model with four end nodes, not all distinct, for non-response modelling. The Laplace-approximated maximum likelihood estimation for the proposed model is suggested. The validation of the proposed estimation procedure and the advantage of the proposed model over traditional methods are demonstrated in simulations. For illustration, the methodologies are applied to data from the 2012 Programme for International Student Assessment (PISA). The analysis shows that the proposed tree model has a better fit to PISA data than other existing models, providing a useful tool to distinguish the sources of non-responses.

Leveraging Auxiliary Information on Marginal Distributions in Nonignorable Models for Item and Unit Nonresponse.

Often, government agencies and survey organizations know the population counts or percentages for some of the variables in a survey. These may be available from auxiliary sources, for example, administrative databases or other high quality surveys. We present and illustrate a model-based framework for leveraging such auxiliary marginal information when handling unit and item nonresponse. We show how one can use the margins to specify different missingness mechanisms for each type of nonresponse. We use the framework to impute missing values in voter turnout in a subset of data from the U.S. Current Population Survey (CPS). In doing so, we examine the sensitivity of results to different assumptions about the unit and item nonresponse.

Targeting Key Survey Variables at the Unit Nonresponse Treatment Stage

Abstract In the presence of nonresponse, unadjusted estimators are vulnerable to nonresponse bias when the characteristics of the respondents differ from those of the nonrespondents. To reduce the bias, it is common practice to postulate a nonresponse model linking the response indicators and a set of fully observed variables. Estimated response probabilities are obtained by fitting the selected model, which are then used to adjust the base weights. The resulting estimator, referred to as the propensity score-adjusted estimator, is consistent provided the nonresponse model is correctly specified. In this article, we propose a weighting procedure that may improve the efficiency of propensity score estimators for survey variables identified as key variables by making a more extensive use of the auxiliary information available at the nonresponse treatment stage. Results from a simulation study suggest that the proposed procedure performs well in terms of efficiency when the data are missing at random and also achieves an efficient bias reduction when the data are not missing at random. We further apply our proposed methods to 2017–2018 National Health Nutrition and Examination Survey.

Nurse Effects on Non-response in Survey-Based Biomeasures

ABSTRACT Collecting biological data in representative surveys is becoming more common due to their potential to inform research and policy. Nevertheless, using nurses to collect these data can lead to unintended effects. In this paper, we investigate how nurses influence the non-response process by looking at five waves of data coming from two surveys in the UK: Understanding Society and the English Longitudinal Study of Ageing. We find that nurses explain between 5 and 14% of the variance in non-response to biological data collection. We also find that older nurses are more successful in obtaining cooperation and consent to biological data collection and nurses with more survey experience are more likely to successfully collect blood samples. Finally, we show that including nurse characteristics in non-response weighting models leads to modest changes in population estimates of biological markers.

Abstract 5857: Molecular characterization of response to interferon alpha in sensitive and resistant bladder cancer models

Abstract Background: Over 70% of patients newly diagnosed with bladder cancer will present with non-muscle invasive disease (NMIBC). The standard of care treatment to prevent recurrence and progression is tumor resection followed by Bacillus Calmette-Guerin (BCG). While this treatment regimen is effective for many patients, a significant fraction become resistant over time. Adenoviral-delivered IFNα gene therapy is emerging as a safe and efficacious treatment for BCG-refractory NMIBC. IFNα is a type 1 interferon that mediates complex tumor-immune interactions, and also has direct cytotoxic effect on tumors. While the mechanisms of IFNα have been extensively studied, it is unclear which of these mechanisms are most relevant for driving therapeutic response. It is also poorly understood why some tumors are sensitive to the direct cytotoxic effects of IFNα and others are resistant. Methods: Here, we investigate the molecular mechanisms of tumor-intrinsic sensitivity and resistance to IFNα. We characterized response of a panel of bladder cancer cell lines to treatment with IFNα, and performed concomitant gene expression profiling to identify molecular features and pathways associated with response and sensitivity. These features were then compared across responsive and non-responsive cancer cell line models from other tissues of origin including skin, pancreas, liver and lung. Results/Conclusions: Several bladder cancer models exhibited a dose-dependent cytotoxic response to IFNα treatments, while others were resistant to these effects. Genomic and mRNA features associated with IFNα sensitivity/resistance were identified. Pan-cancer cell line analysis identified conserved and putative tissue-specific differences in IFNα downstream signaling genes in resistant vs sensitive cell lines. These features are being evaluated and will inform future investigation of the mechanism of IFNα anti-tumor activity in patients. Citation Format: Robin Osterhout, Jennifer Green, Scott McDonnell, Amy Klova, Adeela Kamal. Molecular characterization of response to interferon alpha in sensitive and resistant bladder cancer models [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5857.

Abstract A120: Intratumoral dendritic cell dynamics in responsive and nonresponsive syngeneic murine tumor models

Abstract Conventional dendritic cells (cDC) play a vital role in T-cell-mediated antitumor immunity by transporting and cross-presenting tumor antigens to CD8 T-cells in draining lymph nodes (dLN) and tumor tissue. DC maturation and antigen uptake takes place in the tumor, which can be heavily affected by the suppressive tumor microenvironment. Intratumoral DCs are a scarce population and their phenotypes and functions have not been fully understood. Here we thoroughly characterized cDC phenotypes and dynamics in a variety of commonly used syngeneic murine tumor models, both at baseline and following anti-PD-L1 (aPDL1) treatment to investigate the correlation and potential contribution of DCs to response. Surprisingly, we observed a lower density of intratumoral DCs in responsive tumor models when compared to nonresponsive ones and their abundance was further reduced by aPDL1 treatment in an IFNg-dependent manner. Their PDL1 expression levels, albeit lower than tumor macrophages, were positively correlated with response. These results demonstrate an inverse correlation between intratumoral DCs and aPDL1-mediated antitumor immunity across different syngeneic murine tumor models, and ongoing studies are exploring the fates of these intratumoral DCs with a focus on their causal relation with the efficacy of immunotherapy. Citation Format: Huizhong Xiong, Stephanie Mittman, Ryan Rodriguez, Marina Moskalenko, Patricia Pacheco Sanchez, Yagai Yang, Rafael Cubas. Intratumoral dendritic cell dynamics in responsive and nonresponsive syngeneic murine tumor models [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr A120.

On log-linear modeling for an incomplete two-way contingency table with one variable subject to nonresponse

In this paper we address two issues in the use of the nonresponse log-linear models for the analysis of an incomplete two-way contingency table with one variable subject to nonresponse, the occurrence of nonresponse boundary solutions and the assessment of the missing data mechanism. To this end, we employ a set of response odds from the fully classified counts and nonresponse odds from partially classified counts. We first investigate the role of the set of odds in identifying the occurrence of boundary solutions and assessing the nonresponse log-linear models suitable for the data, and then propose a data analytic guideline for the analysis of an incomplete two-way contingency table. We also examine the theoretical properties of the set of odds when nonresponse boundary solutions occur.

Abstract A31: Reprogramming of tumor-associated macrophages by anti-Program Death Ligand 1 (PDL1)

Abstract Tumor-associated macrophages play a vital role in shaping tumor environment and contribute to the cancer-immune set point. Immunotherapy such as aPD1/aPDL1 has demonstrated remarkable therapeutic efficacy on a variety of cancers through reinvigorating CD8+ T cells. However, its impact on macrophages largely remains unclear. Here by using mouse MC38 cancer model, we show that aPDL1 re-programs macrophages from M2-like to M1-like: it decreases Arginase-I (ARG1), an M2-like marker, and increases iNOS and MHCII, both M1-like markers, on macrophages. Next, we investigate the mechanism behind the reprogramming. We identify that IFNg from CD8+ T cells and NK cells is boosted by aPDL1 and is required for macrophage re-polarization. IFNg blockade abrogates this process. Finally, we examine additional cancer models and observe similar phenotypes in another aPDL1-responsive model, but not in a third, non-responsive model. We conclude that tumor associated macrophage is re-programmed from M2-like to M1-like following aPDL1 treatment, and that it is mediated by enhanced IFNg in the tumor environment. Our findings suggest macrophage reprogramming as a novel mechanism and an independent avenue to promote antitumor activity via modulation of the cancer-immune set point. Citation Format: Huizhong Xiong, Marina Moskalenko, Ryan Rodriguez, Stephanie Mittman, Michelle Yang, Jeong Kim, Rafael Cubas Barcelli. Reprogramming of tumor-associated macrophages by anti-Program Death Ligand 1 (PDL1) [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2017 Oct 1-4; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2018;6(9 Suppl):Abstract nr A31.

Latent Variable Modelling with Non-Ignorable Item Non-Response: Multigroup Response Propensity Models for Cross-National Analysis

SummaryWhen missing data are produced by a non-ignorable non-response mechanism, analysis of the observed data should include a model for the probabilities of responding. We propose such models for non-response in survey questions which are treated as measures of latent constructs and analysed by using latent variable models. The non-response models that we describe include additional latent variables (latent response propensities) which determine the response probabilities. We argue that this model should be specified as flexibly as possible, and we propose models where the response propensity is a categorical variable (a latent response class). This can be combined with any latent variable model for the survey items, and an association between the latent variables measured by the items and the latent response propensities then implies a model with non-ignorable non-response. We consider in particular such models for the analysis of data from cross-national surveys, where the non-response model may also vary across the countries. The models are applied to data on welfare attitudes in 29 countries in the European Social Survey.

Abstract B035: Evaluation of a novel monoclonal antibody targeting ASC amino acid transporter 2 using patient-derived xenograft mouse models of gastric cancer

Abstract Introduction: Gastric cancer is a leading cause of global cancer mortality with an overall 5-year survival rate of approximately 20% and is particularly prevalent in many Asian countries. ASC amino acid transporter 2 (ASCT2), also known as SLC1A5, is a Na+-dependent glutamine/neutral amino acid transporter. ASCT2 acts as a high-affinity transporter of L-glutamine (Gln) and has been reported to be upregulated in a variety of cancerous tissues, including stomach. Several studies have shown that glutamine is a major nutrient contributing to cancer cell growth, so the glutamine metabolism pathway is an attractive target for gastric cancer treatment. In this study, we evaluated the efficacy of a novel anti-ASCT2 humanized monoclonal antibody, KM8094, as a therapeutic agent against gastric cancer and explored predictive biomarker candidates using patient-derived xenograft (PDX) mouse models. Materials and Methods: The tumor tissues from which the PDX models were generated were provided by Dr. Chan Shing Leng (NUS, Singapore). For the glutamine uptake assay, PDX-tumor cells were incubated with 3H-glutamine in the presence of KM8094. Intracellular 3H-glutamine was measured by liquid scintillation spectrophotometry. In vitro ADCC activity against human gastric cancer cells (SNU-16) was evaluated using human peripheral blood mononuclear cells (PBMC) from either gastric cancer patients or healthy controls as effector cells. In vivo antitumor activity of KM8094 was examined using PDX models. Mice were treated with KM8094 at 10 mg/kg i.v. or phosphate-buffered saline once weekly for 2-3 weeks. Finally, omics analyses of PDX tumor tissues were performed via a gene expression array analysis, DNA methylation array analysis, and metabolomics analysis. Results: All five gastric cancer PDX models histologically expressed ASCT2. KM8094 significantly inhibited the 3H-glutamine uptake in vitro and showed in vivo antitumor efficacy in the gastric cancer PDX models. In addition, KM8094 showed in vitro ADCC activity against gastric cancer cells using PBMCs from gastric cancer patients. These results indicated that KM8094 is a potential new therapeutic agent for gastric cancer treatment. Finally, we explored predictive biomarker candidates of KM8094 by a multi-omics analysis on the PDX models, and a few candidates were selected by genomic analyses. In addition, a metabolomics analysis revealed clear differences in the intracellular energy and redox status between responsive and nonresponsive PDX models. Conclusions: These results demonstrated the therapeutic potential of KM8094 for the treatment of gastric cancer and provide some insight into predictive biomarker candidates for antitumor efficacy. Citation Format: Noriyuki Kasai, Aya Sasakawa, Kenta Hosomi, Tze Wei Poh, Bernadette Lynn Chua, Wei Peng Yong, Jimmy So, Shing Leng Chan, Richie Soong, Koji Kono, Toshihiko Ishii, Kazuya Yamano. Evaluation of a novel monoclonal antibody targeting ASC amino acid transporter 2 using patient-derived xenograft mouse models of gastric cancer [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr B035.

Discrete Choice Models for Nonmonotone Nonignorable Missing Data: Identification and Inference.

Nonmonotone missing data arise routinely in empirical studies of social and health sciences, and when ignored, can induce selection bias and loss of efficiency. In practice, it is common to account for nonresponse under a missing-at-random assumption which although convenient, is rarely appropriate when nonresponse is nonmonotone. Likelihood and Bayesian missing data methodologies often require specification of a parametric model for the full data law, thus a priori ruling out any prospect for semiparametric inference. In this paper, we propose an all-purpose approach which delivers semiparametric inferences when missing data are nonmonotone and not at random. The approach is based on a discrete choice model (DCM) as a means to generate a large class of nonmonotone nonresponse mechanisms that are nonignorable. Sufficient conditions for nonparametric identification are given, and a general framework for fully parametric and semiparametric inference under an arbitrary DCM is proposed. Special consideration is given to the case of logit discrete choice nonresponse model (LDCM) for which we describe generalizations of inverse-probability weighting, pattern-mixture estimation, doubly robust estimation and multiply robust estimation.

Multiply robust imputation procedures for zero-inflated distributions in surveys.

Item nonresponse in surveys is usually treated by some form of single imputation. In practice, the survey variable subject to missing values may exhibit a large number of zero-valued observations. In this paper, we propose multiply robust imputation procedures for treating this type of variable. Our procedures may be based on multiple imputation models and/or multiple nonresponse models. An imputation procedure is said to be multiply robust if the resulting estimator is consistent when all models but one are misspecified. The variance of the imputed estimators is estimated through a generalized jackknife variance estimation procedure. Results from a simulation study suggest that the proposed procedures perform well in terms of bias, efficiency and coverage rate.

Estimation of population mean in two– stage sampling under a deterministic response mechanism in the presence of non-response

In the present paper, we have considered the problem of estimation of population mean in the presence of non-response under two-stage sampling. Two different models of non-response with deterministic response mechanism have been discussed in the paper. The estimators under two non-response models have been developed by using Hansen and Hurwitz (1946) technique. The expressions for the variances and estimates of variance of these estimators have been derived. The optimum values of sample sizes have been obtained by considering a suitable cost function for a fixed variance. A limited simulation study has been carried out to examine the magnitude of percent relative loss (% RL) in standard error due to non-response. An empirical study with the real populations has also been carried out to assess the % RL in standard error due to non-response.

Abstract 1657: Evaluating anti-tumor activity of the human anti-PD-1 antibody pembrolizumab using humanized mouse models

Abstract Targeting immune checkpoints has proven to be an effective strategy for reactivating the immune system to elicit potent anti-tumor activity. For example, the anti-PD-1 antibody pembrolizumab has shown robust and durable clinical activity in numerous malignancies including melanoma and non-small cell lung cancer. Unfortunately, a subset of patients do not benefit from pembrolizumab treatment. Ongoing molecular analyses of patient samples are establishing gene signatures which aim to more accurately predict which patients will benefit from immunotherapy. However, there are currently no preclinical in vivo platforms that comprehensively translate these clinical findings. Therefore, we have explored the use of mice engineered with a human immune system to examine the anti-tumor activity of pembrolizumab against a panel of human tumor xenografts derived from both cultured cell lines as well as patient tissue. Tumor models were selected based on parameters including PD-L1 expression and mutational load in order to closely mimic responder patients in the ongoing pembrolizumab clinical trials. Our data confirm that pembrolizumab can induce significant anti-tumor efficacy in humanized mice inoculated with human tumor models derived from both cancer cell lines and patient tumors. In responsive models, pembrolizumab caused significant tumor growth inhibition and, in some instances, tumor regressions. FACS analyses confirmed effective targeting of PD-1+ T cell populations. Anti-tumor efficacy correlated with increased effector T cell to regulatory T cell ratios as well as an elevated T cell infiltrate presence. Nonresponsive models lacked robust immune infiltration detectable by FACS or IHC, consistent with clinical data. Furthermore, presence of cytolytic activity at baseline and T-cell inflamed environment were evaluated using transcriptome analysis and subsequently linked to pembrolizumab response and to somatic neoantigen burden of the tumors. Together these studies validate the utility of humanized mice in the preclinical development of novel cancer immunotherapies and may also be beneficial in better defining the patient responder hypothesis and its prognostic value. Additionally, humanized mice may help identify optimal combination treatment regimens to provide more meaningful clinical benefit to patients who do not respond to single-agent therapy with pembrolizumab. Citation Format: Douglas E. Linn, Razvan Cristescu, Kallol Ray, Shuli Zhang, Sripriya Dhandapani, Sarav Kaliyaperumal, Jennifer H. Yearley, Brian J. Long. Evaluating anti-tumor activity of the human anti-PD-1 antibody pembrolizumab using humanized mouse models [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 1657. doi:10.1158/1538-7445.AM2017-1657

Comparing Weighting Methods for Nonresponse in the HKPSSD Survey and the BCSPS

Using data collected from the Hong Kong Panel Study of Social Dynamics (HKPSSD) Survey and the Beijing College Students Panel Survey (BCSPS), the current study examined four commonly used weighting adjustments: the logistic regression model, the Response Propensity Stratification (RPS) method, the Generalized Exponential Model (GEM), and the Random Forests Model (RFM) for longitudinal nonresponse. Our results indicated that the logistic regression model, the RPS method, and the GEM produced comparable results, with the GEM and the RPS methods slightly outperforming in many instances, while the results obtained by the RFM were not as reliable as those generated by the other methods. Moreover, high bias and MSE changes were associated with variables or categories that had low occurrences for all the methods considered. In particular, for categorical variables, there were considerable variations on bias across categories while the overall difference compared to the baseline was not significant. In addition, relative bias and MSE changes were correlated to the specification of the nonresponse model, the baseline weights, as well as the intervals between the baseline, and the wave for weighting adjustment.