The use of an RNA-based approach to determine quantitative immune profiles of solid tumor tissues and leverage clinical cohorts for immune-oncology biomarker discovery is described through a molecular and informatics protocols.
Immunotherapies show promise in the treatment of oncology patients, but complex heterogeneity of the tumor microenvironment makes predicting treatment response challenging. The ability to resolve the relative populations of immune cells present in and around the tumor tissue has been shown to be clinically-relevant to understanding response, but is limited by traditional techniques such as flow cytometry and immunohistochemistry (IHC), due the large amount of tissue required, lack of accurate cell type markers, and many technical and logistical hurdles. One assay (e.g., the ImmunoPrism Immune Profiling Assay) overcomes these challenges by accommodating both small amounts of RNA and highly degraded RNA, common features of RNA extracted from clinically archived solid tumor tissue. The assay is accessed via a reagent kit and cloud-based informatics that provides an end-to-end quantitative, high-throughput immuno-profiling solution for Illumina sequencing platforms. Researchers start with as few as two sections of formalin-fixed paraffin-embedded (FFPE) tissue or 20-40 ng of total RNA (depending on sample quality), and the protocol generates an immune profile report quantifying eight immune cell types and ten immune escape genes, capturing a complete view of the tumor microenvironment. No additional bioinformatic analysis is required to make use of the resulting data. With the appropriate sample cohorts, the protocol may also be used to identify statistically significant biomarkers within a patient population of interest.
Quantification of tumor-infiltrating lymphocytes (TILs) and other immune-related molecules in formalin-fixed and paraffin embedded (FFPE) solid tumor human tissue samples has demonstrated value in clinical research1,2,3. Common techniques such as flow cytometry and single-cell ribonucleic acid (RNA) sequencing are useful for fresh tissue and blood4, but are unsuitable for analysis of FFPE materials due to the inability to create viable cell suspensions. Current methods that have been used to quantify these cells in FFPE tissue suffer from major challenges. Immunohistochemistry (IHC) and other similar imaging workflows require specific antibodies to detect cell-surface proteins, which can be difficult to standardize across laboratories to enable reproducible quantification5. Platforms such as the nCounter system rely on the expression of single genes to define key immune cells6, limiting sensitivity and specificity of detection. More generic RNA sequencing methods, coupled with standalone software tools, are available but require significant optimization and validation prior to use7,8,9,10,11,12. Recent advances in combining laser capture microdissection (LCM) with RNA sequencing for FFPE tissue has shown promise; however, a more high-throughput, turnkey solution is required for translational studies aimed at identifying robust biomarkers13,14. Methods to generate multidimensional biomarkers, such as Predictive Immune Modeling, that define patient cohorts including therapy responders, cancer subtypes, or survival outcomes with high predictive accuracy and statistical significance are becoming increasingly important in the age of precision medicine and immunotherapy15,16.
To address this need, an immune profiling assay was developed to enable sensitive and specific quantification of immune cells in solid tumor FFPE tissue using standardized RNA-sequencing reagents and cloud-based informatics. In addition to accommodating degraded RNA from FFPE tissue, the protocol is able to accommodate RNA derived from limiting tissue samples such as core needle biopsies, needle aspirates, and micro- or macro-dissected tissue. RNA data from each sample is compared to a database of gene expression models of immune cells, called immune Health Expression Models, to quantify immune cells as a percentage of total cells present in the sample. Briefly, these models were built using machine-learning methods to identify unique multigenic expression patterns from whole-transcriptome data generated from purified immune cell populations (isolated using canonical cell-surface markers)17,18. The multidimensional Health Expression Models underlying the technology enables the assay to quantify each immune cell as a percent of the total cells present in the heterogenous mixture. This enables the researcher to generate inter- and intra-sample immune cell comparisons, which have been shown to have clinical value19,20. Other applications include quantification of immune response pre- and post-treatment, as described in the representative results. The assay reports on multiple features of immune contexture of the tumor and tumor microenvironment including the absolute percentages of eight immune cell types (derived from gene expression models): CD4+ T cells, CD8+ T cells, CD56+ Natural Killer cells, CD19+ B cells, CD14+ monocytes, Tregs, M1 macrophages, and M2 macrophages. In addition, the assay reports the expression (in transcripts per million, or TPM) of ten immune escape genes: PD-1, PD-L1, CTLA4, OX40, TIM-3, BTLA, ICOS, CD47, IDO1, and ARG1.
The reagent kit is used to make high quality libraries ready for sequencing on an Illumina platform following a hybrid capture-based library preparation method, as shown in Figure 1. If a researcher does not have an Illumina sequencing platform in their laboratory, they may submit their samples to a core laboratory for sequencing. Once generated, sequencing data is uploaded to the Prism Portal for automated analysis, and a comprehensive, quantitative profile for each individual sample, in the form of the Immune Report (Figure 2A), is returned to the user. Users may also define sample groupings in the Prism Portal to generate a Biomarker Report (Figure 2B), highlighting statistically significant biomarkers that distinguish two patient cohorts. Importantly, the data generated by the reagent kit is for research use only and may not be used for diagnostic purposes.
Figure 1: Overview of Workflow. In this protocol, RNA is first converted to cDNA. Sequencing adaptors are ligated, and adaptor-ligated cDNA is amplified and barcoded by PCR to create a pre-capture library. Biotinylated probes are then hybridized to specific cDNA targets which are then captured using streptavidin beads. Unbound, non-targeted cDNA is removed by washing. A final PCR enrichment yields a post-capture library ready for sequencing. *Total RNA must be from human samples; may be intact or degraded (FFPE) RNA. Please click here to view a larger version of this figure.
Figure 2: Representative Immune Reports. The workflow generates two reports, an individual immune report (A) for each sample processed, and a biomarker report (B) for defined patient cohorts. Please click here to view a larger version of this figure.
The protocol requires approximately 16 h of preparation time (from total RNA to libraries ready for sequencing); however, there are a number of optional stopping points, as noted in the protocol. The assay makes use of the rich, dynamic nature of transcriptomics to move beyond legacy single-analyte biomarkers to multidimensional gene expression models, thereby enabling comprehensive biological characterization of tissue samples with standardized reagents and easy-to-use software tools. It empowers researchers to utilize a contemporary technology in their own laboratory, by leveraging machine-learning and a database of Health Expression Models to derive more accurate, quantitative immune profiles of precious clinical samples, and discover multidimensional RNA biomarkers with full statistical analysis.
The protocol requires 20 ng intact or 40 ng highly degraded (FFPE) RNA. The RNA sample should be free of DNA, salts (e.g., Mg2+, or guanidinium salts), divalent cation chelating agents (e.g., EDTA, EGTA, citrate), or organics (e.g., phenol and ethanol). It is not recommended to proceed with RNA samples that have a DV200 <20%. Use of the in-kit control RNA is strongly recommended as these controls provide a means to evaluate performance throughout the entire protocol, from library preparation to analysis.
The protocol is designed to be performed using 0.2 mL PCR strip tubes. If preferred, the protocol can also be performed using the wells in a 96-well PCR plate. Simply use the wells of a 96-well PCR plate in place of all references to PCR tubes or strip tubes. Use PCR plates with clear wells only, as it is critical to visually confirm complete resuspension of beads during bead purifications and wash steps.
Throughout the protocol, keep reagents frozen or on ice unless otherwise specified. Do not use reagents until they are completely thawed. Be sure to thoroughly mix all reagents before use.
Keep enzymes at -20 °C until ready to use and return to -20 °C promptly after use. Use only molecular-grade nuclease-free water; it is not recommended to use DEPC-treated water. When pipetting to mix, gently aspirate and dispense at least 50% of the total volume until the solutions are well mixed. Pipette mix all master mixes containing enzymes. Using vortex to mix the enzymes could lead to denaturation and compromise their performance. During bead purifications, use freshly made 80% ethanol solutions from molecular grade ethanol. Using ethanol solutions that are not fresh may result in lower yields. Avoid over drying the beads, as this can reduce elution efficiency (beads look cracked if over dried).
As described in Step 10, unique index primers are added to each reaction. Based on the sequences of these indices, for low-level multiplexing, certain index combinations are optimal. The sequences of these indices are required for demultiplexing the data post-sequencing. The sequences and recommended multiplexing combinations are provided in Supplemental Table 3. In this same step, it is important to note that the number of recommended PCR cycles varies depending on the quality of RNA used, and, some optimization may be required to prevent PCR over-amplification. For the ImmunoPrism Intact Control RNA and other high-quality RNA, start optimization with 10 PCR cycles. For the ImmunoPrism FFPE Control RNA and other highly degraded/FFPE RNA, start optimization with 15 PCR cycles. Producing a test library using RNA representative of the material to be analyzed in order to optimize PCR cycles is recommended. The minimum number of PCR cycles that consistently yield sufficient pre-capture library yields (>200 ng) should be used. A secondary peak around 1000 bp on the Bioanalyzer trace is indicative of over-amplification (Figure 4). Over-amplification should be minimized, but the presence of a small secondary peak will not interfere with assay results.
To minimize sample loss and avoid switching tubes, Step 13 may be performed in PCR tubes, strip tubes, or a 96-well PCR plate instead of 1.5 mL microtubes, if your vacuum concentrator allows. The rotor can be removed on many concentrators. This enables the strip tubes or plates to fit in the vacuum. The vacuum concentration can then be run using the aqueous desiccation setting with no centrifugation. Consult the manual for your vacuum concentrator for instructions. If the samples are dried down in strip tubes or a 96-well plate, the hybridization step can be performed in the same vessel.
During Step 17, be sure to vortex every 10-12 min to increase the bead capture efficiency. Carefully hold the caps of the warm strip tubes when mixing to prevent tubes from opening.
The washes described in Step 18 are critical to avoid high nonspecific contamination and must be followed closely. Be sure to completely resuspend the beads at each wash, completely remove the wash buffers, and during the Wash Buffer 2 wash, transfer the samples to a fresh strip tube (Step 18.6.5). Ensure that the streptavidin beads are completely resuspended and remain in suspension during the entire incubation. Splashing on the tube caps will not negatively impact the capture. During the room temperature washes, a microplate vortex mixer may be used to vortex the samples for the entirety of the two-minute incubation period for easier resuspension. Do not let the streptavidin beads dry out. If needed, extend incubations in the buffers to avoid drying the beads. If using more than one strip tube, work with one strip tube at a time for each wash while the other strip tubes sit in the thermocycler. This can help avoid over drying the beads or rushing, resulting in poor resuspension or other sub-optimal techniques. For first time users, it is not recommended to process more than 8 library reactions at a time.
Current immune profiling techniques deliver a continuum of information – from thousands of data points that require significant interpretation (RNA sequencing) to an individual, discrete data point (single-plex IHC). The protocol described here represents an approach that is somewhere in the middle, with a focused scope enabling high sensitivity, but capturing only a subset of clinically relevant transcriptomic data. Due to the nature of bulk RNA extraction, this protocol does not provide information about the spatial relationships between immune cells and the tumor microenvironment, however, results may be complemented with imaging technologies to add this information. There are a myriad of applications for the data generated by this protocol, as there is much to be learned about biology of cancer as a disease, and the therapies being developed to treat it. As shown in the representative results, the individual immune report is useful for understanding how a patient's immune profile may change in response to events such as disease progression or treatment. While the results presented here provide some example use cases, other applications including investigating the mechanism of action of a therapy and identifying putative biomarkers of clinical outcomes such as progression free and overall survival are also practical. When using this protocol for biomarker discovery applications, it is important to practice good study design to ensure homogenous populations are analyzed, sufficient samples are included for statistical power, and sources of bias are considered. Due to the focused, streamlined nature of the assay, it is feasible to imagine a path towards clinical validation and downstream application of these biomarkers once discovered.
The authors have nothing to disclose.
The authors wish to acknowledge TriStar Technology Group for providing the biological specimens for the representative results, as well as the entire molecular, analysis, product, and commercial teams at Cofactor Genomics for their technical expertise and support.
0.2 mL PCR 8 tube strip | USA Scientific | 1402-2700 | USA Scientific 0.2 mL PCR 8-tube strip |
200 Proof Ethanol | MilliporeSigma | EX0276-1 | Prepare 80% by mixing with nuclease-free water on the day of the experiment |
96-well thermal cyclers | BioRad | 1861096 | |
Solid-phase Reversible Immobilization (SPRI) Beads | Beckman-Coulter | A63882 | Agencourt AMPure XP – PCR Purification beads |
Digital electrophoresis chips and kit | Agilent Technologies | 5067-4626 | Agilent High Sensitivity DNA chips and kit |
Digital electrophoresis system | Agilent Technologies | G2939AA | Agilent 2100 Electrophoresis Bioanalyzer |
Streptavidin Beads | ThermoFisher Scientific | 65306 | Dynabeads M-270 Streptavidin |
ImmunoPrism Kit – 24 reaction | Cofactor Genomics | CFGK-302 | Cofactor ImmunoPrism Immune Profiling Kit – 24 reactions |
Human Cot-1 DNA | ThermoFisher Scientific | 15279011 | Invitrogen brand |
Magnetic separation rack | Alpaqua/Invitrogen | A001322/12331D | 96-well Magnetic Ring Stand |
Microcentrifuge | Eppendorf | 22620701 | |
Microcentrifuge tubes | USA Scientific | 1415-2600 | USA Scientific 1.5 mL low-adhesion microcentrifuge tube |
NextSeq550 | Illumina | SY-415-1002 | Any Illumina sequencer may be used for this protocol |
Nuclease-free water | ThermoFisher Scientific | AM9937 | |
Prism Extraction Kit | Cofactor Genomics | CFGK-401 | Cofactor Prism FFPE Extraction Kit – 24 samples |
Purified RNA | – | – | Purified from human tissue samples |
Fluorometer | ThermoFisher Scientific | Q33226 | Qubit 4 System |
Fluorometric Assay Tubes | Axygen | PCR-05-C | 0.5mL Thin Wall PCR Tubes with Flat Caps |
High Sensitivity Fluorometric Reagent Kit | Life Technologies | Q32854 | Qubit dsDNA HS Assay Kit |
Vacuum concentrator | Eppendorf | 22820001 | VacufugePlus |
Vortex mixer | VWR | 10153-838 | |
Water bath or heating block | VWR/USA Scientific | NA/2510-1102 | VWR water bath/USA Scientific heating block |