Understanding the dynamic tumor immune microenvironment (TIME) is important in guiding immunotherapy. We have previously validated signatures predictive of checkpoint inhibitor efficacy which distinguish immunomodulatory, mesenchymal stem-like, and mesenchymal phenotypes. Here we use twenty tumor types (7162 samples) to identify potentially conserved immune biology within these TIME spaces, genes co-expressed across distinct cell types involved these immune processes, and the association of these signatures with ICI response. One signature, which contained multiple B-cell markers, was associated with immunotherapy efficacy in three cohorts, including IMvigor210. This signature of potentially conserved B-cell biology in co-infiltrated immune cell ecosystems had a more consistent association with outcome than comparable single cell type models and likely reflects a complex immunological response involving multilayered relationships between distinct immune effector cell types. These signatures were most highly expressed in tumors with prominent immune cell invasion, however there was consistent identification of infiltrate presence in relatively immune restricted cases. This suggests that these immune population signatures may identify conserved immune cell type co-infiltrate physiology of the TIME that best captures immune physiology with potential clinical utility.

Supplementary Figure 1. Quality control of NeoTRIP RNA-Seq data. Boxplots showing the distribution of read depth of targeted coding regions by arm and response status. Dots represent the read depth of single samples. The horizontal grey dashed line represents the cut-off of 40 million paired-end reads.

A 36-year-old man was evaluated in the infectious diseases clinic of the hospital because of fevers and throat discomfort that had begun 18 days earlier. An extensive workup was unrevealing. A diagnosis was made.

Supplementary Table 1. List of 27 genes of the DetermaIO assay.

Supplementary Table 5. Multivariable regression of pCR according to NGS-based IO score and TNBC subtypes in the RNA-Seq cohort. For TNBC subtypes LAR was taken as reference group.

Supplementary Table 2. Patients’ characteristics of the NeoTRIP trial in the RNA-Seq and RT-qPCR cohorts.

Supplementary Figure 7. Association of TNBC subtypes and IO classifications. A) Frequency of RT-qPCR DetermaIO class by TNBC subtype; B) Frequency of RNAseq IO classes by TNBC subtype. C) Multivariable logistic regression model with TNBC subtypes and RT-qPCR IO score for association with pCR by treatment arm. Odds ratio (OR) and 95% confidence intervals are shown. D) Multivariable logistic regression model with TNBC subtypes and RNA-Seq IO score for association with pCR by treatment arm. Odds ratio (OR) and 95% confidence intervals are shown.

Supplementary Figure 2. Comparison of RT-qPCR DetermaIO and RNA-Seq IO scores. A) Scatter plot showing the relation between continuous IO scores. Concordance and correlation coefficient = 0.94; 95% CI = 0.92 − 0.95; B) Contingency plot of binary IO classifications. Cohen’s kappa = 0.84; 95% CI = 0.77–0.91; p < 0.0001.

PURPOSE:We assessed the 27-gene RT-qPCR-based DetermaIO assay and the same score calculated from RNA sequencing (RNA-seq) data as predictors of sensitivity to immune checkpoint therapy in the neoTRIPaPDL1 randomized trial that compared neoadjuvant carboplatin/nab-paclitaxel chemotherapy (CT) plus atezolizumab with CT alone in stage II/III triple-negative breast cancer. We also assessed the predictive function of the immuno-oncology (IO) score in expression data of patients treated with pembrolizumab plus paclitaxel (N = 29) or CT alone (N = 56) in the I-SPY2 trial. EXPERIMENTAL DESIGN:RNA-seq data were obtained from pretreatment core biopsies from 242 (93.8%) of the 258 patients in the per-protocol-population. The DetermaIO RT-qPCR test, performed in the CAP/CLIA-accredited laboratory of Oncocyte Corp., was available for 220 patients (85.3%). A previously established threshold was used to assign DetermaIO-positive versus DetermaIO-negative status. Publicly available microarray data were used from I-SPY2. RESULTS:IO scores calculated from RNA-seq and RT-qPCR data were highly concordant. In neoTRIPaPDL1, DetermaIO-positive cancers (N = 92, 41.8%) had pathologic complete response (pCR) rates of 69.8% and 46.9% in the CT + atezolizumab and CT arms, respectively. In DetermaIO-negative cases, pCR rates were similar in both arms (44.6% vs. 49.2%; interaction test P = 0.04). PDL1 protein expression and stromal tumor-infiltrating lymphocyte count were not predictive of differential benefit from atezolizumab. In I-SPY2, IO-positive cancers (45.9%) had pCR rates of 85.7% and 16%, with and without immunotherapy, respectively. In IO-negative cancers, pCR rates were 46.7% versus 16.1%. CONCLUSIONS:DetermaIO identified patients who benefited from neoadjuvant immunotherapy resulting in improved pCR rate, independently of PDL1.

Supplementary Data S1. FPKM RNA-Seq data for the genes included in the IO score signature.

Supplementary Table 4. Univariate regression of pCR according to TNBC subtypes in the RNA-Seq cohort. One-vs-all approach was applied.

Supplementary Figure 6. Comparison of TNBC subtype classification systems. A) Sankey plot showing the overlap between the TNBC subtype classification according to Ring and colleagues and the original TNBCtypes from Lehmann and colleagues B) pCR rate by Lehmann TNBCtypes and treatment arm.

Supplementary Figure 3. Continuous baseline RT-qPCR DetermaIO score. Distribution of continuous IO score by treatment arm and pCR status. P-values by logistic regression.

Supplementary Figure Legends from Drug Efflux by Breast Cancer Resistance Protein Is a Mechanism of Resistance to the Benzimidazole Insulin-Like Growth Factor Receptor/Insulin Receptor Inhibitor, BMS-536924

Multiple targeted therapeutics have been approved by the FDA for mUC, including immune checkpoint inhibitors (ICIs) and more recently targeted agents for both FGFR and Nectin-4. FGFR3-aberrant and Nectin-4 expressing cells have been associated with an immunosuppressed phenotype. Given that less than half of all patients respond to these agents as monotherapies and less than 20% are eligible to receive salvage therapy, effective personalized treatment plans are critical. Typical biomarkers for ICIs such as PD-L1 and TMB have not been definitive in mUC, yet a biomarker-driven optimization of first-line therapy and subsequent sequencing have the potential to achieve higher and more durable response rates. The IO score is a 27-gene tumor immune microenvironment (TIME) classifier that has been associated with the clinical benefits of ICIs in multiple cancer types, including mUC. This study demonstrates that the IO score was associated with both progression-free survival (PFS) and overall survival (OS) in a real-world cohort of mUC patients treated with ICIs. Furthermore, the IO score was independent of and provided information incremental to TMB. Interestingly, the IO score predicted benefit in patients with high FGFR expression, despite conflicting data regarding response rates among the FGFR aberrant population. Taken together, these results demonstrate that the IO score assessment of the TIME is associated with a clinical benefit from ICI therapy and that this novel biomarker may inform therapeutic sequencing decisions in mUC, potentially improving outcomes for this notoriously difficult-to-treat disease.

Background: miRNA have been shown to be central communicators between the immune system and cancer. They are attractive as candidate blood-based cancer detection targets because they are secreted in high copy number from cancer cells, are bound to circulating proteins and have secondary structures that prolong their half-life in blood. miRNAs that are thought to communicate between cancer associated fibroblasts and immune cells are of special interest for their potential involvement in transitions in the TIME from "hot" - an inflammatory state with a greater likelihood of sensitivity to immune checkpoint inhibitors - to "cold" - tumors with an immunosuppressive state or immune inert state or vice versa. We have previously described a gene expression signature that distinguishes immunomodulatory (IM, inflammatory cells), mesenchymal (M, EMT differentiated cells), and mesenchymal stem-like (MSL, cancer associated fibroblasts) features of the TIME, and when converted to a binary classifier, has been shown to be correlated with response to immune checkpoint inhibitors (ICI). Here we use TCGA microRNA data to identify a microRNA correlate of this classifier. Methods: miRNA and RNA expression were collected from TCGA across five tissue types. Each tumor case was assigned a TIME phenotype as previously described [1]. Lung, breast, colon, and bladder data were used to identify miRNA’s whose expression pattern significantly correlated with at least one of three immune phenotypes identified by the previously defined algorithm. Data from the IntAct database were used to identify putative gene targets of the candidate miRNA. Results: Of the 151 miRNAs of interest, 147 known interactions were found in the IntAct database. Of these 147 interactions, 107 were between miRNA and genes of different TIME (or immune) phenotypes while 89 were interactions between miRNAs and genes implicated in regulating immune hot and cold phenotypes. Conclusions: We identified miRNA whose expression patterns correlates with a classifier of the TIME that has been shown to identify likely responders to immune checkpoint inhibitors. This candidate gene list was significantly enriched for both miRNA targets of known immune mediators and or targets implicated in modulating the TIME. Next steps are to further narrow the list to those detectable in blood in cancer, and to train a blood-based diagnostic that might be able to predict response to ICI therapy for those patients where tissue is not available. References: [1] Seitz, R.S., Hurwitz, M.E., Nielsen, T.J. et al. Translation of the 27-gene immuno-oncology test (IO score) to predict outcomes in immune checkpoint inhibitor treated metastatic urothelial cancer patients. J Transl Med 20, 370 (2022). https://doi.org/10.1186/s12967-022-03563-9 Citation Format: Catherine T. Cronister, Robert S. Seitz, Brian Z. Ring, Douglas T. Ross, Brock Schweitzer. The role of microRNAs in the tumor immune microenvironment. [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 4118.

<p>Figures S1-S3, Tables 1-2, further data</p>

Introduction: In the setting of metastatic pancreatic adenocarcinoma (mPDAC), lower baseline plasma KRAS mutation levels have been associated with improved survival. While tissue-agnostic, plasma-based copy number instability (CNI) has been demonstrated as an early indicator of response to immunotherapy for some solid tumors, it has not been assessed for patients with mPDAC, nor in combination with KRAS mutations for patients receiving standard of care chemo/radiotherapy. Here we evaluate the combination of mutant KRAS (mKRAS) and CNI detection in plasma as a predictor of overall and progression-free survival (OS/PFS) in mPDAC patients who received standard of care therapy. Methods: Cell-free DNA was extracted from plasma and libraries prepared at baseline (Week 0) and weeks 8, 16 and 24 on therapy, and analyzed by next-generation sequencing (CNI) and droplet digital PCR (mKRAS). Descriptive statistics were computed for variables including CNI (score is a measure of circulating tumor DNA) and mKRAS variant allele fraction. Detection was defined as above the limit of detection (mKRAS=0.13%) and above the 95th percentile of the value in normal individuals (CNI=24). Therapy response was assessed by OS and PFS. Results: 196 plasma samples from 64 mPDAC patients were analyzed. When dichotomized as detectable vs undetectable, CNI alone was significantly associated with OS at all on-therapy timepoints but not baseline, whereas mKRAS was significantly associated with OS for all 4 timepoints (Table 1). Detection of both CNI and mKRAS in combination was strongly associated with worse OS at all timepoints, yielding the highest HR. Similar results were obtained when mKRAS and CNI were dichotomized at their respective median values or with PFS as the clinical endpoint. Conclusions: Combined CNI and mKRAS detection at baseline and on-therapy may provide a strong and early indication of worse prognosis for patients with mPDAC. Table 1. Association of CNI and mKRAS with Overall Survival (HazardRatio [95% CI], log-rank p-value) Timepoint CNI mKRAS CNI and KRAS Baseline/Week 0 1.54 [0.89-2.68], 0.1 2.05 [1.12-3.78], 0.02 2.50 [1.46-4.28], 0.0006 Week 8 1.78 [0.99-3.18], 0.05 2.21 [1.19-4.08], 0.01 9.81 [3.40-28.28], <0.0001 Week 16 1.91 [1.03-3.53], 0.04 3.26 [1.60-6.62], 0.0006 11.11 [4.28-28.83], <0.0001 Week 24 2.55 [1.28-5.09], 0.006 4.55 [2.03-10.23], <0.0001 6.42 [2.61-15.84], <0.0001 Citation Format: Samuele Cannas, Jacob E. Till, Kristine Kim, Michael J. LaRiviere, Charles M. Vollmer, Jennifer R. Eads, Thomas B. Karasic, Peter J. O'Dwyer, Charles J. Schneider, Ursina R. Teitelbaum, Kim A. Reiss Binder, Mark H. O'Hara, Douglas T. Ross, Kim McGregor, Kirsten Bornemann-Kolatzki, Ekkehard Schütz, Julia Beck, Erica L. Carpenter. Liquid biopsy signature combining copy number instability and mutant KRAS detection is associated with survival for patients with metastatic pancreatic 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 1043.

<p>Figures S1-S3, Tables 1-2, further data</p>

Supplementary Table S1 from Novel 5′ Untranslated Region Variants of <i>BCRP</i> mRNA Are Differentially Expressed in Drug-Selected Cancer Cells and in Normal Human Tissues: Implications for Drug Resistance, Tissue-Specific Expression, and Alternative Promoter Usage