The Texas Immuno-Oncology Biorepository, a statewide biospecimen collection and clinical informatics system to enable longitudinal tumor and immune profiling

Abstract

A detailed understanding of the molecular and immunological changes that occur longitudinally across tumors exposed to immune checkpoint inhibitors is a significant knowledge gap in oncology. To address this unmet need, we created a statewide biospecimen collection and clinical informatics system to enable longitudinal tumor and immune profiling and to enhance translational research. The Texas Immuno-Oncology Biorepository (TIOB) consents patients to collect, process, store, and analyze serial biospecimens of tissue, blood, urine, and stool from a diverse population of over 100,000 cancer patients treated each year across the Baylor Scott & White Health system. Here we sought to demonstrate that these samples were fit for purpose with regard to downstream multi-omic assays. Plasma, urine, peripheral blood mononuclear cells, and stool samples from 11 enrolled patients were collected from various cancer types. RNA isolated from extracellular vesicles derived from plasma and urine was sufficient for transcriptomics. Peripheral blood mononuclear cells demonstrated excellent yield and viability. Ten of 11 stool samples produced RNA quality to enable microbiome characterization. Sample acquisition and processing methods are known to impact sample quality and performance. We demonstrate that consistent acquisition methodology, sample preparation, and sample storage employed by the TIOB can produce high-quality specimens, suited for employment in a wide array of multi-omic platforms, enabling comprehensive immune and molecular profiling.

Recent advances in immunotherapeutics have revolutionized medical oncology, with immune checkpoint inhibitors (ICIs) now being standard of care across numerous cancer types both in metastatic disease and in the operable neoadjuvant or adjuvant setting.¹ A meta-analysis of ∼160 studies demonstrated a combined overall response rate of 20.21%, but ICI combination studies to date have failed to increase efficacy.² A detailed understanding of the immunological changes that occur across multiple tumor types both in early and late-stage disease when exposed to ICIs over a long period of time is a significant knowledge gap in oncology and is a barrier to improved clinical trial design utilizing novel agents.

Tumor response and resistance have focused on drug-target expression levels (PD-1 and/or PD-L1), tumor immune microenvironment, and tumor mutation burden. The tumor immune microenvironment (hot vs. cold tumors) has been examined for response/resistance associations with limited understanding of how to improve outcomes in immune excluded or desert tumors.³ Although tumor mutation burden correlates with ICI response in certain tumors, examples of low tumor mutation burden and low PD-L1 cases responding to ICIs are common.⁴ Thus, there is an unmet clinical need to understand the mechanism(s) of response/resistance to ICIs, necessitating the longitudinal collection of high-quality biospecimens of patients treated with ICIs, alone and in combination with other systemic agents.

To address these unmet needs, we sought to create a statewide biospecimen collection and clinical informatics system to enable tumor and immune profiling in a longitudinal manner over many years in an effort to enhance translational research. The Texas Immuno-Oncology Biorepository (TIOB) purposely consents patients to collect, catalog, process, store, and analyze well-defined and well-annotated serial biospecimens of tissue, blood, urine, and stool from a diverse population of over 100,000 cancer patients from different racial, social, and ethnic backgrounds who are treated across Baylor Scott & White Health cancer centers each year (Figure 1).

Figure 1.

The Texas Immuno-Oncology Biorepository (TIOB), a statewide oncology and clinical informatics system to enable translational research.

Potential types of studies enabled by this resource include but are not limited to (a) morphological studies including immunohistochemistry; (b) genomic and molecular analysis, including circulating, cell-free DNA, RNA, as well as extracellular vesicles (EV) and their cargo; (c) proteomic analysis involving protein isolation, splicing neoforms, and neoantigens; (d) interrogation of circulating immune cells as well as the underlying tumor immune microenvironment; and (e) single-cell and spatial transcriptomics (Figure 2). The correlation of these studies with clinical, treatment outcome, and demographic information will offer opportunities to improve our understanding of the complex biological interactions that define cancer.

Figure 2.

Longitudinal tumor and immune profiling with multi-omic platforms in real-world patients participating in the Texas Immuno-Oncology Biorepository.

METHODS

The TIOB is aiming to partner with Baylor Scott & White cancer patients across the third largest Commission on Cancer network of accredited cancer hospitals in the United States, with 13 participating cancer centers covering a geographical footprint ranging from Dallas to Fort Worth in North Texas through central Texas into Waco, Temple, and Austin. The 11 patients that participated in this study were consented onto an institutional review board–approved protocol at Baylor Scott & White Health (IRB # 019-350). All patients are at least 18 years of age with a confirmed malignancy and are candidates for or are due to start an ICI in the near future. Patients are excluded if they are unable or unwilling to donate blood or unable or unwilling to provide informed consent for collection of fresh or archived tumor tissue or have already started an immunotherapy regimen before being consented for the study.

Specimen collection and processing

Patients are followed longitudinally with specimen and ongoing data collection occurring at regular intervals (pretreatment or baseline and approximately every 3 months for immunotherapy subjects in the metastatic setting and 3 to 6 months for surgical subjects) to coincide, when possible, with routine follow-up visits of subjects’ cancer therapy (next cycle of treatment or restaging computed tomography scans) or regular clinical visits. The collections may continue for as long as patients remain on treatment or surveillance or until they withdraw consent. A variety of biological specimens, including bodily fluids or excess tissue left over after complete pathologic evaluations, are prospectively accessioned for the biorepository. Standard operating procedures (SOPs) have been developed to ensure that all specimens that are collected and stored are of known quality to ensure robust and comparable immune profiling.

For blood samples of up to 50 mL, sample processing includes standard methodologies for isolation and cryopreservation of viable peripheral blood mononuclear cells (PBMC), serum, and plasma (see www.protocols.io). PBMC isolation is performed manually from CPT whole blood tubes, and automated isolation techniques such as RoboSep are being implemented.

Urine is collected at the stated intervals. Up to 30 mL of urine is mixed with EDTA as part of standard collection, whereas a second urine vial is left untreated. Cell pellets from the urine after processing are also stored.

Collection of stool specimen (at home or the clinic) is performed 72 hours prior to initiation of immunotherapy and thereafter every 3 to 6 months for the patients’ duration with cancer. Collection materials are provided in a kit given to patients by the key study personnel or sent to them by mail. Stool collection (Zymo Feces Catcher, Zymo Research Corp., Irvine, CA) and sample preparation (Zymo DNA/RNA Shield Fecal Collection Tube, Zymo Research Corp.) are provided as part of the kit. Patients bring their samples to a designated location, or alternatively, they can mail their sample to the study investigators using a prepaid mailer (see www.protocols.io).

Tissue samples from TIOB participants who undergo standard of care surgery may be collected. Tissue is released to the TIOB only when it is no longer required for diagnosis or further patient care. The tissue is collected and then snap frozen, processed for spatial transcriptomics via controlled freezing in OCT and/or processed for formalin-fixed paraffin embedding. Representative quality control samples are taken to allow assessment of quality variables such as percent tumor, tumor type, percent necrosis, and degree of inflammation for each frozen specimen collected.

Individual TIOB participant biospecimens are stored according to SOPs for each sample type in ultra-low temperature freezers (approximately −80°C) or in liquid nitrogen vapor (approximately −196°C).

Specimen/clinical database

The TIOB is building a scalable, flexible platform to meet current and evolving needs. This infrastructure involves multiple components including biospecimen management and tracking, clinical data management, patient-reported data, incident reporting, and results from sample analysis. Infrastructure for the TIOB is being developed following the software development lifecycle with explicit steps for requirements gathering, design, testing and validation, and maintenance. The TIOB utilizes a scalable cloud-based platform to integrate the different components, all linked through distinct study IDs. Biospecimen inventory and tracking is performed on the Freezerworks Summit platform. The software is accessed in a secure, role-specific, tiered-permission access information technology environment that assigns a unique sample identifier for each specimen or derivative. The software supports individual specimen check-in, check-out, and real-time inventory for each sample, in each location, in each freezer. The software also tracks specimen IDs, aliases, limited demographics, aliquots, and derivatives and relevant preanalytic variables for individual projects for each specimen and supports the production and tracking of individualized project-specific sample collection kits. Clinical data and patient-reported data are collected and managed using REDCap electronic data capture tools hosted at Baylor Scott & White Health.^5,6

The biorepository has a written overall quality management program to ensure that the processing, annotation, storage, and distribution of specimens occurs at consistently high levels of quality for biological specimens and to ensure overall regulatory compliance and a safe working environment. The TIOB quality program includes the following:

• All biorepository personnel have continual access to all policies and procedures.

• The collection, processing, annotation, storage, and distribution of tissues are tracked for chain of custody, handling, and adherence to SOPs to ensure high-quality samples and comparability of research results.

• All equipment is verified and validated for use and subject to routine and scheduled maintenance and continuous temperature monitoring for freezers.

• Information technology includes limited access, secure systems, with audit trails. Freezerworks is validated prior to each use and upgrade. Periodic inventory checks verify that TIOB samples and their derivatives in the Freezerworks database are verifiably present and in the correct location within the storage freezers.

• Standardized preanalytic variables, based on international consensus standards (Standard PREanalyic Code; SPREC), are recorded for biological specimens so that end users can assess their quality to assess whether they are “fit for use” for their individual project.

• For tissue, quality control testing is performed routinely for tissue specimens and periodically for DNA/RNA quality. Additional quality testing for suitability for immunohistochemistry or in situ hybridization can be performed upon request.

• Representative samples of PBMCs are periodically analyzed for viability. In addition, the biorepository participates in an International Society of Biological and Environmental Repositories–endorsed proficiency testing program for viable PBMC isolation from the Integrated Biobank of Luxembourg with “very satisfactory” results for PBMC viability and functional interferon-gamma enzyme-linked immunosorbent spot assay.

• Researcher feedback about sample quality is solicited following tissue distribution and is used to reexamine quality control procedures.

• Deviations from SOPs, insufficient samples, unexpected results, equipment failures, shipping issues, or anything that could affect sample quality are routinely tracked and reviewed.

• There is planning for risk mitigation/disaster response, including backup power and contingencies for relocation.

SOPs and annotation

The SOPs for PBMC isolation, buffy coat, and cell free plasma from whole blood are listed on www.protocols.io, as are the urine, stool, and tissue collection and processing SOPs.

A data management plan documents data handling processes and data quality assurance for clinical data needed to annotate the samples. Clinical data quality is maintained by multiple levels of data quality assurance. First, electronic data capture forms are designed with data quality and end users in mind. Where possible, data validation checks are built in to minimize data entry errors. Next, data is abstracted from the electronic health record by individuals who have undergone training and passed a data abstraction test. Thorough documentation on methods of abstraction has been created for every data field. After data entry, designated data leads review critical data fields for accuracy. Additional data review occurs regularly on a percentage of entered forms to check for errors. Data is extracted from the REDCap system and additional data quality checks are performed. These data quality checks will continue to evolve to ensure that the data meets quality standards.

Biospecimen processing toward fit for purpose

For analysis of plasma samples for EV RNA quantity/quality, blood collected in EDTA tubes is spun 10 min at 1600 g at room temp (centrifuge brake “off”), with the plasma collected by pipette from “above the buffy coat,” spun a second time (1600 g × 10 min), pipetted as cell-free plasma, and promptly frozen.

For EVs, concentrated cell-free plasma or urine (12 mL spun to 0.5 mL Amicon Ultra-15 Centrifugal Filter Unit, 100 KDa cutoff; MilliporeSigma, Burlington, MA) is passed through a size exclusion column (qEVoriginal 70 nm; IZON Science, Portland, OR). After the void volume plus 3 mL, four 0.5 mL fractions containing the EVs are collected and pooled; an aliquot is processed for RNA quality and yield.

RNA from plasma-derived EVs is isolated using the Qiagen AllPrep DNA/RNA Mini Kit. Specifically, 125 µL of purified EVs suspended in PBS is combined with 475 µL of Buffer RLT Plus (Qiagen AllPrep DNA/RNA Mini Kit, Qiagen, Germantown, MD) and homogenized using the Qiashredder (Qiagen). The lysate is transferred to the Qiagen AllPrep DNA spin column to remove the DNA. The flow-through is added to the AllPrep RNA purification column and isolation is conducted as written in the AllPrep DNA/RNA Mini Handbook. RNA is quantified using the Quant-It RiboGreen RNA Reagent and Kit in 96-well format with 5 µL of RNA in 200 µL reaction volume (Thermo Fisher Scientific, Waltham, MA). Quantity is further assessed using the Nanodrop spectrophotometer.

RNA from urine EVs is isolated using the Qiagen AllPrep DNA/RNA Mini Kit. Specifically, 125 µL of purified EVs suspended in PBS is combined with 475 µL of Buffer RLT Plus (Qiagen AllPrep DNA/RNA Mini Kit) and homogenized using the Qiashredder (Qiagen). The lysate is transferred to the Qiagen AllPrep DNA spin column to remove the DNA. The flow-through is added to the AllPrep RNA purification column and isolation is conducted as written in the AllPrep DNA/RNA Mini Handbook. RNA is quantified using the Quant-It RiboGreen RNA Reagent and Kit in 96-well format with 5 µL of RNA in 200 µL reaction volume (Thermo Fisher). Quantity is also assessed using the Nanodrop spectrophotometer.

For microbiome assessment, DNA is extracted from stool using the MagMax Microbiome Ultra Extraction Kit (Thermo Fisher), with prior bead beating on a TissueLyser (Qiagen), using the KingFisher Magnetic Extraction Instrument (Thermo Fisher). Purified DNA is separated on 1% agarose gel and quantified via densitometry and spectrophotometry using a Qubit fluorometer assay (Thermo Fisher). DNA is further quantitated by the BactQuant assay.

A Trypan blue viability assay is used for PBMCs. Frozen PBMC cells are thawed quickly in a 37°C water bath by gently swirling the vial. Cells (20 µL aliquots) are removed and diluted with 20 µL prewarmed DMEM:10% FBS media. A total of 40 µL of Trypan Blue stain (0.4%) is added to the cells, and 10 µL is loaded onto a chamber slide for counting using the Countess Automated Cell Counter System (Thermo Fisher).

RESULTS

To demonstrate fit for purpose, samples from 11 enrolled patients (7 women, 4 men) were employed (Table 1). Five of the seven women had a diagnosis of breast cancer, while three of the four men had lung cancer. TIOB is actively enrolling patients with multiple tumor types (Figure 2), but the first two cohorts that opened were breast and lung cancer, and hence they make up the majority of tumor types for this analysis. The average age at diagnosis was 61 years old. The majority of the patients (n = 7) were exposed to pembrolizumab, either alone (6/7) or in combination (1/7). All 11 patients were able to provide blood and urine, while 10 provided stool.

Table 1.

Clinical characteristics of the first 11 patients in the Texas Immuno-Oncology Biorepository

Pt	Sex	Samples	Tumor type	Age at diagnosis (yr)	Stage	Histology	Immuno-oncology	Imaging available
1	M	B, U, S	Lung	55	IV	Squamous cell	Pembro	Yes
2	F	B, U, S	Lung	66	IV	Adenocarcinoma	Pembro	Yes
3	M	B, U	Lung	67	IV	Poorly differentiated with giant cell	Pembro	Yes
4	M	B, U, S	Lung	62	IV	Squamous cell	Pembro	Yes
5	F	B, U, S	Lung	57	IV	Adenocarcinoma	Pembro	Yes
6	M	B, U, S	Renal	62	IV	High-grade, poorly differentiated	Pembro	Yes
7	F	B, U, S	Breast	51	IIA	IDC	None	Yes
8	F	B, U, S	Breast	72	IIB	ILC	None	Yes
9	F	B, U, S	Breast	47	IIA	IDC	Pembro/Carboplatin/Taxotere	Yes
10	F	B, U, S	Breast	60	DCIS	IDC	None	Yes
11	F	B, U, S	Breast	76	IIA	IDC	None	Yes

B indicates blood; DCIS, ductal carcinoma in situ; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; Pembro, pembrolizumab; S, stool; U, urine.

The TIOB has collected, and will continue to collect, longitudinal plasma samples from patients receiving ICIs, but a critical next step to ensure translational actionability is to ensure that the plasma specimens are fit for downstream multi-omic analyses. Table 2 demonstrates that total RNA (and RNA concentration) from plasma-derived EVs was appropriate for downstream next-generation sequencing in 9 out of the 11 TIOB patients recruited. Much like plasma, urine represents a noninvasive sample that can contain molecular information that could inform therapeutic response/resistance. Here, we examined total RNA levels and RNA concentrations from urine-derived EVs. Table 2 shows that all 11 patients had RNA levels that can be employed in downstream next-generation sequencing. Three specimens yielded >100 ng total EV RNA, which may be due to urinary tract infection independent of EV cargo.

Table 2.

Analysis of RNA isolate from extracellular vesicles

Patient	Urine RNA		Plasma RNA
	Total (ng)	ng/mL	Total (ng)	ng/mL
1	36	3.5636	45	2.0506
2	60	2.5538	9	1.8026
3	81	2.1062	60	1.8012
4	138	2.074	51	1.8228
5	27	2.247	45	4.6967
6	72	3.9402	Undetected	0.8722
7	66	5.3665	20	0.996
8	144	1.2546	Undetected	0.8778
9	75	5.1623	15	1.4534
10	540	3.8482	9	1.1195
11	36	1.6773	21	2.8506

The TIOB demonstrated an ability to collect stool samples from patients on ICI regimens. Table 3 demonstrates that 10 of the 11 samples produced an RNA concentration (average 16.35 ng/µL) and bacterial load (average 204,872,996 copies/µL) consistent with downstream analyses that can accurately inform on microbial communities.

Table 3.

RNA concentration and bacterial load quantification in stool

Patient	EV RNA (ng/µL)	BactQuant (copies/µL)
1	24.3	289101600**
2	17.2	213177075**
3	8.27	101741650**
4	34.1	485697250**
5	13.8	133306188**
6	8.05	103682400**
7	N/A	N/A
8	11.7	187649663**
9	18.1	196083550**
10	14.4	172702488**
11	13.6	165588100**

EV indicates extracellular vesicles.

As the expansion of ICIs continues across tumor types, the examination of circulating immune cell repertoires may hold critical measures of response/resistance. In this study, we showed that 10 out of 11 TIOB patients produced cell counts (average 11.02 × 10⁶ cells/mL) consistent with employment in downstream studies (Table 4). In the one instance of very low PBMC viability (TIOB-02-21-0003), review of that biospecimen’s handling uncovered protracted storage at room temperature, which most likely led to this specimen fail. Remediation of sources of deviations from proscribed specimen handling and processing is a feature of TIOB’s continuous improvement practices.

Table 4.

Viability analysis of peripheral blood mononuclear cells

Patient	Total cell count (cells/mL)	Viability (%)
1	10.7 × 106	89.5
2	Not collected	84.5
3	11.0 × 106	88.5
4	8.9 × 106	83.5
5	7.0 × 106	61.0
6	10.6 × 106	62.0
7	5.0 × 106	59.5
8	18 × 106	56.5
9	10.4 × 106	71.5
10	8.4 × 106	1.0
11	10.6 × 106	70.0

DISCUSSION

Given the fact that biomarkers that will define response/resistance to immunotherapeutic agents are prone to variability in the face of sample collection, processing, and storage, only concerted efforts to longitudinally collect relevant patient samples are likely to uncover clinically meaningful results. Assay harmonization in an effort to reduce variability and to ensure reproducibility is a key objective of the TIOB. In this study, we demonstrate the clinical utility of the TIOB, a repository built upon collecting longitudinal samples from cancer patients treated with immunotherapeutic agents across a unified statewide network. A common institutional review board, shared SOPs, and a Clinical Laboratory Improvement Amendments–approved pathology department enabled collection of high-quality biospecimens. Further, a collaboration with the Translational Genomics Research Institute (TGen) highlighted the utility of the specimens and the clear evidence that these samples were sufficient for state-of-the-art, multi-omic downstream analyses.

Most studies describing the mechanisms of response/resistance to ICIs have been performed in vitro or with a patient’s primary tumor. These have focused on PD-1/PD-L1 protein levels, immune cell infiltration status, or tumor mutation burden.⁷ More recently, tumor-specific alterations have also been identified that impact ICI response. As more molecular alterations are identified, the need for longitudinal sampling will become more critical and is expected to provide new mechanisms for resistance that can not only be monitored, but also seed the discovery of new therapeutic avenues.

The data provided in this study show fit-for-purpose metrics across analytes known to impact therapeutic response and resistance, including response to immunotherapeutics.^7,8 EVs were successfully collected and contained high-quality RNA cargo. EV cargo has been correlated with ICI response/resistance across a number of tumor types, including gastric,⁹ melanoma,¹⁰ and lung cancer.¹¹ We also successfully collected viable PBMCs from the patients. Alterations in immune cell ratios are known to impact ICI response. For example, neutrophil-to-lymphocyte ratio has been correlated with immunotherapy response.¹² Finally, we collected high-quality microbiome data from the stool of the TIOB patients. The gut microbial community makeup has demonstrated impact on the response to ICIs in vivo.⁸ It is reasonable to assume that these analytes can provide high-quality genomic and proteomic data, which can be mined for associations with ICI response and resistance.

It is well accepted that sample acquisition and processing methods impact sample quality and performance in downstream analyses. PD-1 protein level measures via immunohistochemistry have been plagued by variability driven by sample quality and differing antibodies.¹³ Further, these protein measures at a single time point associate poorly with ICI response in multiple tumor types.^14,15 The consistent acquisition methodology, sample preparation, and sample storage utilized across the TIOB sites have produced high-quality specimens, suited for employment in an array of multi-omics platforms.

One limitation of the current study is the small sample size. While the 11 patients sampled provided an indication of assay performance, increased sample sizes will be necessary. Additionally, no longitudinal samples were included in this analysis. Demonstration of consistent analyte performance across time points from the same patient will be assessed in future TIOB studies. Further, we did not take the analytes through to next-generation sequencing and proteomic assays. While TGen has significant expertise in these assays and is confident that the analyte concentrations and qualities would perform well in next-generation sequencing studies, these assays should be performed in subsequent studies. Finally, although the TIOB routinely employs protocols that enable cell-free DNA analyses, this assay was not performed on this first set of samples, but will be explored in future studies.

Approved ICIs and novel immunotherapeutic agents will continue to play a critical role in the treatment of diverse cancer types, and it is important to understand not only the alterations that predict initial response and resistance, but also those dynamic alterations that govern continued response. Thus, longitudinal sampling across a substantial number of patients will be necessary. The vision of the TIOB is to prospectively collect these longitudinal samples, provide them for multi-omic analyses, and discover those molecular/immunological alterations that govern response/resistance to current and future immunotherapeutic agents.

FUNDING

Funding was provided by the W. W. Caruth, Jr. Foundation.

DISCLOSURE STATEMENT

Dr. Kelly has received institutional research support from Bristol Myers Squibb and Eli Lilly and has served on advisory boards for BMS, Merck, EMD Serono, Astra Zeneca, Daiichi Sankyo, Eisai, Astellas, Ipsen, Novartis, Takeda, Novocure, and Eli Lilly. The other authors have no conflicts to disclose.