Patent Application
20220364185
This invention relates to sample analysis techniques, and more particularly to methods and systems for detecting and analyzing nucleic acids from cancer cells, e.g., in blood samples to determine which therapies would be most effective in a specific patient.
The ability to detect the presence of rare circulating tumor cells (CTCs), exosomes, and cell-free nucleic acids, such as cell-free deoxyribonucleic acid (DNA) or ribonucleic acid (RNA), using a simple blood test, or “liquid biopsy,” has the potential to greatly enhance the monitoring of cancers, providing instant sampling of tumor cell numbers, genetic composition, and drug response parameters, without requiring invasive tumor biopsies. Thus, the detection of CTCs, exosomes, and cell-free DNA or RNA for early cancer detection has the potential to revolutionize the treatment of cancer, enabling the diagnosis of invasive cancer at a stage before it has metastasized, when curative treatment is expected.
However, CTCs, exosomes, and cell-free nucleic acids are very rare and/or small and/or are easily degraded, and thus identifying, visualizing, measuring, and scoring these rare components admixed with normal blood components remains a significant challenge, even after partial purification with known microfluidic devices or similar technologies. For example, per milliliter of whole blood, there are only 1-10 CTCs amongst more than 5 billion red blood cells (RBCS) and more than 5 million white blood cells (WBCs)(Plaks et al. “Cancer Circulating Tumor Cells,” Science, 341:1186; 2013).
While exosomes are not that rare, they are only about 30-100 nm in diameter, making them difficult to isolate and detect in blood samples. Due to the complexity of body fluids such as blood, physical separation of exosomes from cells and similar-sized particles is challenging. Isolation of exosomes using differential ultracentrifugation and micro-filtration or a gradient can improve purity. Single step isolation of extracellular vesicles by size-exclusion chromatography has been demonstrated to provide greater efficiency for recovering intact vesicles over centrifugation, although a size-based technique alone will generally not be able to distinguish exosomes from other vesicle types. To isolate a pure population of exosomes a combination of techniques is necessary, based on both physical (e.g. size, density) and biochemical parameters (e.g. presence/absence of certain proteins involved in their biogenesis). A key challenge to isolating tumor-derived exosomes is to differentiate them from exosomes produced by normal tissues.
When employing cell-free RNA, it is important to minimize release of cellular RNA following blood draw, because cell-free RNA is present at low quantities in the blood. Thus, blood samples require special handling and/or systems to avoid degradation or contamination with nucleic acids from cells, and to stabilize the cell-free RNA.
In addition, antibody staining of tumor cells is highly variable, due to high heterogeneity among cancer cells, even within an individual patient, as well as the poor physical condition of many tumor cells that circulate in the bloodstream, many of which have begun to undergo programmed cell death or anoikis. In addition, accurate scoring of antibody-stained tumor cells requires differentiation from large numbers of contaminating white blood cells, some of which bind to antibody reagents non-specifically. As such, only a subset of candidate tumor cells can be robustly identified by antibody staining, and as many as half of patients tested have no detectable cells, despite having widely metastatic cancer.
The present disclosure relates to methods and uses to obtain the highest possible sensitivity of data relating to tumor-specific RNA, e.g., from rare CTCs, exosomes, and/or cell-free RNA, in standard blood samples to predict which cancer therapies may be most effective to treat a specific detected cancer in a given patient. In particular, the new methods do not need the CTCs and/or exosomes to be completely isolated from contaminating WBCs, and instead can reliably detect as few as one CTC or exosome in products containing, e.g., up to 10,000 WBCs or more. The new assay methods combine (1) an isolation system that can consistently obtain intact CTCs and exosomes with high quality RNA from blood with (2) a droplet-based digital polymerase chain reaction (PCR) assay focused on RNA markers of specific cancer lineages for each tumor type that are absent in blood of healthy patients. The new methods can be used to determine which therapeutic agents have the highest potential to effectively treat the specific cancer type found in each patient.
In general, the disclosure relates to methods for predicting the efficacy of specific therapeutic regimens, e.g., therapeutic agents, to treat specific cancers in a given subject or patient with ultra-high sensitivity and specificity. The new methods comprise or consist of obtaining tumor-specific RNA from a blood sample and determining which of a series of lineage markers are expressed in the RNA in the blood sample, wherein an expression level of or more specific lineage markers is predictive of progression-free survival and overall survival for a specific anti-cancer treatment regimen. For example, in some implementations, the methods can include or consist of isolating circulating tumor cells (CTCs) from a blood sample from the subject; converting CTC-derived RNA into cDNA; encapsulating the cDNA into individual droplets; amplifying the cDNA in each droplet in the presence of a reporter group configured to bind specifically to cDNA from CTCs and not to cDNA from other cells in the blood; and determining which of a series of lineage markers are expressed in the CTCs in the blood sample, wherein an expression level of a specific one or more lineage markers is predictive of progression-free survival, time to progression, overall survival, or other clinically relevant endpoints for a specific anti-cancer treatment regimen.
In some implementations, the potential efficacy of a specific anti-cancer treatment regimen for a specific cancer in the subject is determined by comparing the expression levels of one or more of the subject's specific lineage markers to a reference standard established for the specific anti-cancer treatment regimen for the specific cancer to determine whether the subject will be treated effectively with the specific anti-cancer treatment regimen. For example, in some implementations, the subject may have prostate cancer and if the subject's specific lineage markers assayed before treatment is begun include an elevated level of FOLH1 (PSMA) and HOXB13 above a background noise level as determined by evaluation of healthy donors without cancer, then the methods described herein predict that the patient will not improve if treated only with abiraterone (e.g., ZYTIGA®). In some implementations, such a subject is further prescribed a combination therapy of abiraterone and another anti-prostate cancer therapy.
In other implementations, the subject may have hormone receptor-positive (“HR+”) breast cancer and if the subject's specific lineage markers assayed at three to four weeks after treatment with a drug targeting the estrogen-signaling pathway include an elevated level of one or more, e.g., one, two, three, four, five, or all six, of PIP, SERPINA3, AGR2, SCGB2A1, EFHD1, and WFDC2 genes above a background noise level determined by evaluation of healthy donors without cancer, then the methods described herein predict that the patient will not improve if treated only with a drug that targets the estrogen-signaling pathway. For example the drugs may be, e.g., ER inhibitors (e.g., tamoxifen), selective ER. degraders (e.g., fulvestrant), and aromatase inhibitors (AI), which block the production of estrogen (e.g., anastrozole, letrozole, and exemestane). The results of the method may cause a healthcare provider to further prescribe for the subject a combination therapy of a drug targeting the estrogen-signaling pathway and another anti-breast cancer therapy.
In various implementations, the methods can include the use of microfluidic isolation of circulating tumor cells (CTCs), or exosomes or cell-free RNA, and digital detection of RNA derived from these components. In sonie embodiments, the RNA can be converted into cDNA and encapsulated into individual droplets for amplification in the presence of reporter groups that are configured to bind specifically to cDNA from CTCs (or other tumor RNA) and not to cDNA from other noncancerous cells.
The methods described herein can further include reducing a volume of the product before isolating RNA and/or removing contaminants from the cDNA-containing solution before encapsulating the cDNA molecules.
In various implementations of the new methods, generating cDNA molecules from the isolated RNA can include conducting reverse transcription (RT) polymerase chain reaction (PCR) of the isolated RNA molecules and/or amplifying cDNA molecules within each of the droplets can include conducting PCR in each droplet. In o the new methods, encapsulating individual cDNA molecules and PCR reagents in individual droplets can include forming at least 1000 droplets of a non-aqueous liquid, such as one or more fluorocarbons, hydrofluorocarbons, mineral oils, silicone oils, and hydrocarbon oils and/or one or more surfactants. Each droplet can contain, on average, one target cDNA molecule obtained from a CTC. In some embodiments, the reporter groups can be or include a fluorescent label.
In various implementations, the methods described herein include using probes and primers in amplifying the cDNA molecules within each of the droplets that correspond to one or more genes selected from the list of cancer-selective genes in Table 1 herein. For example, the selected genes can include prostate cancer-selective genes, e.g., any one or more of AGR2, FOLH1, HOXB13, SCHLAP1, AMACR, AR variants, including AR-V7, UGT2B15, STEAP2, and TMPRSS2:ERG (as can be easily determined from Table 1). In another example, any one or more of ALDH1A3, CDH11, EGFR, FAT1, MET, PKP3, RND3, S100A2, and STEAP2 are selective for pancreatic cancer. Similar lists can be generated for the other types of cancers listed in Table 3.
In other examples, the selected genes include any one or more of the breast cancer-selective genes listed in Table 3, in other examples, the selected genes include genes selective for one or more of lying, liver, prostate, pancreatic, and melanoma cancer. For example, a multiplexed assay can include 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 or even all of the selected genes that are listed in Table 3 as being selective for a particular type of cancer, e.g., breast cancer, lung cancer, prostate cancer, pancreatic cancer, liver cancer, and melanoma. Typically, a group of primers and probes for 1, 2, 3. 4, 5, 6, 7, 8, 9. 10, 11, or 12 or more cancer-selective genes from Table 1 are used for a particular type of cancer. Other specific combinations of selected genes (markers for those genes) are described in the Examples below.
In the methods described herein, the CTCs can arise from metastatic or primary/localized cancers.
The disclosure also provides uses of the probes and primers related to one or more selected cancer genes listed in Table 3 for amplifying and detecting cDNA molecules obtained from circulating tumor cells (CTCs) in a blood sample, and for o determining which of a series of lineage markers are expressed in the CTCs in the blood sample, wherein an expression level of a specific one or more lineage markers is predictive of progression-free survival and overall survival for a specific anti-cancer treatment regimen.
As used herein, the phrase “circulating tumor cells” (CTCs) refers to cancer cells derived from solid tumors (non-hematogenous cancers) that are present in very rare numbers in the blood stream of patients (e.g,, about 1 CTC in about 10,000,000 WBCs in whole blood). CTCs can arise from both metastatic as well as primary/localized cancers.
As used herein, a “product” means a group of isolated rare cells and other contaminating blood cells, e.g., red blood cells, white blood cells (e.g., leukocytes), e.g., in some sort of liquid, e.g., a buffer, such as a pluronic buffer, that arise from processing in the methods described herein, e.g., using the systems described herein. A typical product may contain only about one to ten CTCs admixed with 500 to 2,500 or more WBCs, e.g., one to ten CTCs in a mixture of 1000 to 2000 WBCs. However, the limit of detection of the present methods can be about 1 CTC in 10,000 WBC. Thus, while the present methods can achieve a level of purity of about 1 CTC in 500 WBCs, the present methods do not require highly purified CTCs, as is required in some known methods of CTC analysis.
The polymerase chain reaction (PCR) is a process of amplification of known DNA fragments by serial annealing and re-annealing of small oligonucleotide primers, resulting in a detectable molecular signal.
Reverse Transcription (RT)-PCR refers to the use of reverse transcription to generate a complementary c-DNA molecule from an RNA template, thereby enabling the DNA polymerase chain reaction to operate on RNA. An important aspect of the new methods disclosed herein is the availability of high quality RNA from whole cell CTCs that are not lysed or treated in such a way that might destroy or degrade the RNA, or from exosomes or cell-free RNA.
As used herein, “positive droplets” are lipid-encapsulated molecules inwhich a PCR reaction performed with tagged primers allows visualization of the PCR amplified product. Thus, a droplet that contained a single template cDNA molecule of a particular targeted gene can become visible using fluorescence microscopy, while an “empty” or “negative” droplet is one that contains no targeted cDNA.
The new methods and systems provide numerous advantages and benefits. For example, the current methods and systems provide results that are far more accurate and robust than either of the prior known systems when used alone. By breaking down the signal from a single CTC or exosome into hundreds or thousands of brightly fluorescent droplets, each derived from a single cDNA molecule, the new digital-CTC assays enable dramatic signal amplification. Given the strict criteria in selecting and optimizing the biomarker genes described herein, the background signal from normal blood cells is negligible in d-CTC. Thus, d-CTC enables greatly amplified signal from patients with advanced cancer (nearly 100% of patients with prostate, lung, breast, and liver cancers). Not only is the fraction of patients with a positive score significantly increased, but also the high level of signal enables dynamic measurements as tumor load declines following cancer therapy, and enables accurate prediction of clinical outcomes of specific therapies even before the therapies are started.
In sum, this novel microfluidics provides a streamlined, ultrahigh-throughput, rapid (e.g., 3 hours per run), and extremely high sensitivity method of enriching, detecting, and analyzing CTCs in patient blood samples. The platform provides rich, clinically actionable information, including the prediction of clinical outcomes of specific cancer-directed therapies.
Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, suitable methods and materials are described below. All publications, patent applications, patents, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting.
Other features and advantages of the invention will be apparent from the following detailed description, and from the claims.
FIGS. 11D1 and 11D-2 are a pair of Kaplan-Meier curves depicting OS (left) and TTP (right) in HR+ patients receiving endocrine therapy based on presence of ESR1 mutations. p-values based on log rank tests.
The present disclosure relates to methods and systems to obtain information from RNA from cancer cells, e.g., CTCs, in blood samples, exosoines from cancer cells in blood samples, or cell-free RNA from cancer cells, to help predict whether a given anti-cancer regimen will work effectively to treat a specific type of cancer in a given patient, These methods and systems combine the power of isolation techniques such as ultrahigh-throughput microfluidic techniques, for example, negative depletion techniques, e.g., those using negative depletion of hematopoietic cells to isolate untagged CTCs in a blood sample, with analysis techniques, such as droplet-based clicital polymerase chain reaction (PCR) assays focused on RNA markers of specific cancer lineages. The specific assay methods, but not the new predictive analysis methods described herein, are described in further detail in PCT WO2016154600, which is incorporated herein by reference in its entirety.
The new methods include steps carried out by comparing the expression levels of various markers to reference standards, and by comparing these expression levels in patients who are destined to respond to specific cancer therapy or likely to have an early progression of their cancer. Such measurements can be informative at o pretreatment baseline or they may emerge through serial blood monitoring once treatment is initiated. The value of these measurements lies in the information provided with respect to specific treatment choices. As multiple treatment options are available for patients with a variety of different cancers, information that helps individualize and guide the rational selection of therapy based on molecular markers becomes critical for effective cancer therapy.
As a specific example discussed in more detail below, patients with prostate cancer that have an elevated level of FOLH1 (PSMA) and/or HOXB13 before any therapy is started (e.g., above 2.5 transcripts per mL or other predetermined threshold) will not do well if treated. only with ahiraterone (e.g., ZYTIGA®). Such patients should be considered for alternative non-hormonal therapies (e.g., taxane chemotherapy or radio-isotope therapy), PARP inhibitors, or novel experimental therapies currently being developed, or combinations of existing therapies that are being tested in patients at high risk of recurrence.
In addition, other CTC isolation technologies than are described herein can also be used in the new methods as long as they provide partially purification of cells (e.g., filtration, positive tumor cell selection), although the quality of the RNA and hence the sensitivity of the assay will be inferior to the microfluidic technologies. Similarly, other digital PCR technologies applied to RNA are capable of detecting lineage-specific primers, although the sensitivity of the droplet-based assay is likely to be the highest.
The isolation techniques are used to enrich CTCs from a blood sample, e.g., using ultrahigh-throughput microfluidic such as the so-called “CTC-iChip” described in, for example, international PCT Application WO 2015/058206 and in Ozkumur et al., “Inertial Focusing for Tumor Antigen-Dependent and -Independent Sorting of Rare Circulating Tumor Cells,” Sci, Transl. Med., 5:179ra47 (2013). The CTC-iChip uses a CTC antigen-independent approach in which WBCs in the blood sample are labeled with magnetic beads, and the sample is then processed through two enrichment stages. The first stage uses deterministic lateral displacement to remove small and flexible cells/particles (RBCs, platelets, unbound magnetic beads, and plasma) while retaining larger cells (CTCs and WBCs). The second stage moves all cells into a narrow fluid stream using inertial focusing and then uses a magnetic field to pull bead-labeled WBCs out of the focused stream, leaving highly enriched CTCs. The CTC-iChip product from 10 ml of whole blood typically contains <500,000 RBCs, <5,000 WBCs, and a variable number of CTCs.
Some analysis techniques further enrich and analyze the isolated CTCs, e.g., as obtained from the CTC-iChip, e.g., using droplet microfluidics. Some basic information on droplet microfluidics is described generally in Jeremy et al., “Ultrahigh-Throughput Screening in Drop-Based Microfluidics for Directed Evolution,” Proc. Natl, Acad. Sci. USA, 107:4004 (2010).
As used herein, the droplet microfluidic techniques can, in certain implementations, include encapsulation of single cells, RT-PCR reagents, and lysis buffer into droplets of typically non-aqueous liquids (e.g., fluorocarbons, hydrofluorocarbons, mineral oil, silicone oil, and hydrocarbon oil; surfactants can also be include in the non-aqueous liquid, e.g., Span80, Monolein/oleic acid, Tween20/80,
SDS, n-butanol, ABIL EM90, and phospholipids), in the size range of, e.g., about 0.5 pL, to 15 nL, in volume and, e.g., 10 to 300 μn, e,g., 20 to 100 μm, e.g., 30 to 50 μm, e.g., 35 μm in diameter. As used in the new methods described in the present disclosure, these techniques further include amplification of cancer-specific transcripts within the droplets to produce a fluorescent signal, and sorting of amplification-positive drops. This approach results in isolation of pure CTCs that can be sequenced and analyzed for determining the potential efficacy of a specific anti-cancer therapy in a specific patient.
Due to the high heterogeneity of CTCs, it is useful to use multiplexed amplification to detect as many CTCs as possible. Thus, instead of using one pair of primers in the PCR mixture, one can increase the probability of detecting and sorting
CTCs using a combination of tumor specific primers. For additional information on the use of PCR for sorting cancer cells, see, e.g., Eastburn et al., “Identification and genetic analysis of cancer cells with PCR-activated cell sorting,” Nucleic Acids Research, 2014, Vol, 42, No, 16 e128.
In the new assay methods, CTCs are lysed to release RNA molecules, which are representative of the genes expressed in a cancer cell. Most are “lineage” specific, rather than cancer specific, for example any prostate cell (whether cancerous or not) expresses these markers. However, normal blood cells do not, and the fact that the signal is derived from a cell circulating in the bloodstream defines it as an abnormal signal, By converting the RNA to cDNA, one can amplify this lineage signal using PCR, Droplet digital PCR, which is extraordinarily sensitive, is used to allow converting the signal from a single cancer cell (i.e., one signal in an imaging assay) into thousands of positive immunofluorescent droplets. The combination of multiple, highly curated gene transcripts ensures high sensitivity and specificity for cancer, and also allows for functional insights (as in the status of hormone responsive pathways in prostate and breast cancers).
As noted, the new assay methods focus on the detection and analysis of high quality RNA rather than DNA. While there has been considerable work on DNA mutation detection in plasma and in CTCs, the present methods rely on RNA markers for the following reasons:
1. DNA mutations are not tumor specific, and the discovery that a healthy individual has some unidentified cancer cells in the blood is a very difficult clinical situation. In contrast, by selecting tumor-specific RNAs (e.g., prostate vs lung), the new methods can identify the source of cancer cells in the blood.
2. DNA mutations are very heterogeneous and besides a few recurrent mutations shared by many cancers, most blood-based mutation detection strategies require pre-existing knowledge of the mutations present in the primary tumor (i.e., not appropriate for screening for unknown cancers). In contrast, all tumor cells derived from specific organs express common lineage markers at the RNA level. Thus, a single cocktail of markers is used in the new methods for each individual type of cancer.
3. Low levels of CTCs are shed by invasive cancers before metastases are established (i.e., it is not too late for blood-based detection), but the presence of tumor cells in the blood connotes vascular invasion (i.e., invasive rather than indolent cancer). That is not the case for plasma DNA or plasma protein markers, which are leaked from dying cells in the primary tumor, and do not necessarily indicate vascular invasion. For example, serum PSA protein in the blood is shed by both benign prostate cells as well as primary prostate cancers. On the other hand, CTCs expressing PSA are shed only by invasive prostate cancers.
4. The analysis of RNA using the novel digital scoring technologies described herein is extraordinarily sensitive. However, free RNA is degraded in the bloodstream, and the use of isolation systems as described herein, such as microfluidic negative depletion systems (e.g., the CTC-Chip system) is unique in that the untagged tumor cells have high quality RNA that is extractable.
The choice of cDNA as a target molecule over DNA was made to not only to boost the signal originating from each tumor cell, but also to specifically target only tumor cell transcripts to the exclusion of white blood cell (WBC) transcripts. The boost in signal is a significant advantage, as it avoids the need for the isolation of CTCs to very high levels of purity. That is, it enables robust and repeatable results with products that contain one or more “isolated” CTCs that are still surrounded by hundreds or thousands of contaminating WBCs, e.g., leukocytes, in the same product. Nevertheless, the strategy of targeting cDNA made from RNA as used in the new methods allows the new assay methods to be exquisitely tailored for maximum specificity with minimal levels of CTC purity compared to prior approaches.
The CTC-iChip technology is highly efficient at isolating non-hematopoietic cells by microfluidic depletion of antibody tagged leukocytes. This feature of the CTC-iChip provides intact tumor-derived RNA (at levels far above those obtained using other technologies), and it is independent of tumor cell surface epitopes (which are highly heterogeneous among cancers and among epithelial vs mesenchymal cell subtypes within an individual cancer). Furthermore, even pre-apoptotic cancer cells whose antibody staining and selection is suboptimal for imaging analysis can provide a source of tumor-specific RNA that can be scored using the methods described herein. For all these reasons, an isolation technology or system that provides high quality RNA from intact CTCs with at least some reduction in the WBCs found in the sample along with the rare CTCs, such as a microfluidic negative depletion system, e.g., the CTC-iChip, is an important first step isolation before the tumor-specific digital readout is applied to the product.
The droplet-based digital detection of extremely rare molecules within a heterogeneous mixture was originally developed for PCR amplification of individual DNA molecules that are below detection levels when present within a heterogeneous mixture, but which are readily identified when sequestered within a lipid droplet before being subjected to PCR. The basic technology for droplet-based digital PCR (“Droplet Digital PCR (ddPCR)”) has been commercialized by RainDance and Bio-Rad, which provide equipment for lipid encapsulation of target molecules followed by PCR analysis. Important scientific advances that made this possible include work in the laboratory of David Weitz at Harvard and Bert Vogelstein at Johns Hopkins. For example, see U.S. Pat. Nos. 6,767,512; 7,074,367; 8,535, 889; 8,841,071; 9,074,242; and U.S. Published Application No. 2014/0303005. See also U.S. Pat. No. 9,068,181.
However, droplet digital PCR itself is not biologically significant unless coupled to a biological source of material, which is key to the new methods described herein. For instance, detection of lineage-specific RNAs (the central focus of the detection strategy described herein) does not distinguish between normal prostate epithelial cells and cancerous prostate cells. As such, detection of prostate-derived transcripts in the blood is not meaningful: they are present within debris from normal prostate cells or exosomes. it is only when coupled with the isolation of whole CTCs (i.e., intact CTCs in the blood) that the ddPCR assay achieves both extraordinary sensitivity and specificity. Hence, these two technologies are ideally suited for each other, because the isolation systems provide high quality RNA, and the droplet-based digital PCR assays are focused on RNA markers in the new methods.
One additional aspect is important to the overall success of the new assay methods. As noted, the new assay methods described herein use cDNA made from total RNA, but key to this use is the identification of appropriate biomarkers that are tumor lineage-specific for each type of cancer, yet are so unique as to be completely absent in normal blood cells (even with ddPCR sensitivity). The selection, testing, and validation of the multiple target RNA biomarkers for each type of cancer described herein enable the success of the new assay methods.
The new assay methods start with the isolation of partially pure CTCs using an isolation system, such as a microfluidic negative depletion system, up to and including the analysis of data from a droplet digital PCR instrument. There are ten main assay steps, some of which are optional, though generally provide better results:
1. isolating from the blood sample a product including CTCs and other cells present in blood; e.g. from a patient or a subject;
2. reducing a volume of the rare cell-containing product (optional); 3. isolating ribonucleic acid (RNA) molecules from the product, e.g., by cell lysis, and generating cDNA molecules in solution from the isolated RNA; e.g., by RT-PCR of RNA released from cells contained in the product;
4. cleanup of cDNA synthesized during the RT-PCR step (optional);
5. pre-amplifying the cDNA using gene-specific targeted preamplification probes, e.g., using the Fluidigm BioMark™ Nested PCR approach, or non-specific whole-transcriptome amplification, e.g., using the Clontech SMARTer™ approach (optional);
6. encapsulating cDNA molecules in individual droplets, e.g., along with PCR reagents;
7. amplifying cDNA molecules within each of the droplets in the presence of reporter groups configured to bind specifically to cDNA from CTCs and not to cDNA from other cells, e.g., using PCR;
8. detecting droplets that contain the reporter groups (e.g., “positive” droplets) as an indicator of the presence of cDNA molecules from CTCs the droplets;
9. analyzing CTCs in the detected droplets, e.g., to determine the presence of a particular disease in a patient or subject; and
10. detecting the expression of specific cancer-specific or lineage-specific genes in the cancer cells, e.g., CTCs, above the low background levels as determined by healthy donor controls (set at a level of 2.5), to determine whether a specific anti-cancer regimen is expected to be effective for that specific patient's specific tumor.
The background levels of these cancer-specific or lineage-specific genes are determined by measuring their expression in CTCs (or exosomes) in the blood of many patients without cancer (age-matched to those patients with cancer for a given type of cancer). The predictive value of these cancer-specific or lineage-specific gene markers are then evaluated by monitoring their expression prior to initiation of and during treatment with a specific cancer therapy in many patients over time, e.g., 6 to 12 months, 15 months, 18 months, 21 months, 24 months or more, and determining each patient's progression-free survival and overall survival statistics over each time period. These data are then used to prepare reference standards for each gene and each anti-cancer treatment regimen against which new patient samples can be compared to determine whether a proposed anti-cancer treatment regimen is likely to be effective in a specific patient, and if so, how effective compared to another potential treatment regimen.
For example, in patients with metastatic castration-resistant prostate cancer, the present inventors have discovered that those patients not having detectable expression of the genes HOXB13 and FOLH1 (PSMA), in their CTCs, as measured by the digital CTC quantitation assay, e.g., a level lower than 2.5 transcripts per mL of blood, will have a better overall survival and progression-free survival when treated with anti-androgen therapy than patients who have a high expression level of these two genes in their CTCs, e.g., a level higher than 2.5 transcripts per mL of blood. The expression of these prostate lineage markers is also detectable in exosomes and other tumor-derived RNA in the blood of patients with prostate cancer.
As described in further detail below, one of the important features of the new d-CTC assay methods is the careful selection of a number of target gene biomarkers (and corresponding primers) that deliver excellent sensitivity, while simultaneously maintaining nearly perfect specificity. A unique list of target gene biomarkers described herein (Table 3, below) was determined using bioinformatics analyses of publicly available datasets and proprietary RNA-Seq CTC data. Great care was taken to select markers that are not expressed in any subpopulations of leukocytes, but are expressed at a high enough frequency and intensity in CTCs to provide a reliable signal in a reasonably wide array of different and distinct patients, A specific set of markers was selected for each cancer type (e.g., prostate cancer, breast cancer, melanoma, lung cancer, pancreatic cancer, among others) and it is specific ones or sets of these markers that are predictive of the potential efficacy of various anti-cancer therapies.
The digital measurement of CTC-derived mRNAs provides not only a level of overall tumor burden for these specific cancers, which is an indicator of cancer activity and response or non-response to particular therapies, but it also provides specific information related to the genes being tested. For example, HOXB13 and FOLH1 in prostate cancers are markers of abnormal androgen signaling, a key characteristic of prostate cancers that are resistant to anti-androgenic therapies. Similarly, in breast cancer, response to hormonal therapies is dependent on the activity of the estrogen receptor pathway, which can be measured within CTCs or exosomes using RNA transcripts. In patients undergoing immunotherapy for cancer, such as melanoma, the presence of differentiation markers within CTCs or related blood vesicles can also indicate the expression of unique sets of genes that activate the immune system, and hence predict response or non-response to immunological treatments.
The separate steps of the assay methods will now be described in more detail.
1. CTC Isolation
Patient blood is run through the CTC-iChip, e.g., version 1.3 M or 1.4.5 T and a sample is collected in a 15 mL conical tube on ice. CTC-iChips were designed and fabricated as previously described (Ozkumur et al., “Inertial Focusing for Tumor Antigen-Dependent and -Independent Sorting of Rare Circulating Tumor Cells,” Science Translational Medicine, 5(179):179ra47 (DOI: 10.1126/scitranslmed.3005616) (2013)).
The blood samples (˜20 mls per cancer patient) are collected in EDTA tubes using approved protocols. These samples are then incubated with biotinylated antibodies against CD45 (R&D Systems) and CD66b (AbD Serotec, biotinylated in house) and followed by incubation with Dynabeads® MyOne® Streptavidin T1(Invitrogen) to achieve magnetic labeling of white blood cells (Ozkumur et al., 2013).
The sample is then processed through the CTC-iChip, which separates the blood components (red and white blood cells and platelets) as well as unconjugated beads away from the CTCs. The CTCs are collected in solution while the red blood cells, platelets, unconjugated beads and the tagged white blood cells are collected in a waste chamber. The process is automated and 10 ml of blood is processed in 1 hour.
2. Volume Reduction and Storage of the Rare Cell;-Containing Product
To fully lyse all cells isolated in the product, it is preferable to reduce the product volume from a typical starting point of several milliliters to a final volume of about 100 μl. This can be achieved, for example, by centrifuging the product, and resuspending in pluronic buffer in preparation for cell lysis and generation of cDNA. At this point samples can be processed for long-term storage by adding RNAlater™ (ThermoFisher), followed by flash-freezing in liquid nitrogen and storage at −80C.
3. Isolating RNA and Generation of cDNA from Cells in the Product
The RNA isolation step is important to the process to fully release all RNA molecules from cells in preparation for RT-PCR. A one-step, in-tube reaction can be used to minimize the risk of cell and. RNA loss likely to be incurred during standard transfer steps. For example, one can use the lnvitrogen SuperScript First-Strand Synthesis Supermix® for qRT-PCR kit, by adding the RT-PCR mastermix directly to the pelleted product, pipetting to lyse fully, and performing the reaction according to the kit protocol targeting a 1:1 RNA:cDNA ratio. Once cDNA has been synthesized, RNase H is applied to the reaction to remove any remaining RNA. Alternatively, if one wants to perform whole transcriptome pre-amplification of the sample in a later step, cDNA can be synthesized using the SMARTer™ Ultra. Low Input RNA Kit protocol, which uses proprietary oligonucleotides and reverse transcriptase enzyme.
4. Cleanup of cDNA Synthesized During RT-PCR
Another useful, yet optional, step in the process involves the removal of lysis reagents from the cDNA-containing solution. The presence of harsh detergents can lead to the destabilization of the droplets used in the ddPCR method, once the cDNA-containing s olution is transferred to the ddPCR instrument. Detergent removal can be accomplished, e.g., through the use of Solid Phase Reversible Immobilization (SPRI). This technique uses coated magnetic beads to first bind cDNA of a specific size range, then allows removal of detergent-containing supernatant, and finally elution of pure cDNA for input into the ddPCR instrument. In addition to the cleanup of the RT-PCR, the SPRI process also accomplishes a size selection of cDNA, which reduces the number of non-target cDNA molecules that enter the ddPCR phase of the process, which in turn reduces background and noise.
5. Pre-Amplification
Pre-amplification of the cDNA is an optional step that increases the number of template molecules that can be detected in the droplet PCR step thus improving signal-to-noise ratio and boosting the confidence in a positive read-out. It can be a very powerful approach for the detection of markers that are expressed at low levels in CTCs, and for analyzing samples that contain very small numbers of possibly apoptotic CTCs, such as in the context of early detection of pre-metastatic disease. These two approaches have been modified to be applied in the workflow of d-CTC assay. Specific Targeted Amplification (STA), based on the Fluidigm BioMark™ Nested PCR protocol, relies on the use of primers specifically designed to amplify the region targeted by the probes used in the droplet PCR step (see Table 2). These primers were carefully designed and tested in conjuncture with their respective fluorescent probes to ensure efficient and specific amplification without increase in noise in healthy controls. Alternatively, whole transcriptome amplification, based on the SMARTer™ Ultra. Low input RNA Kit protocol, relies on the amplification of every transcript in the product, including both those found in WBCs and those found in CTCs, using random primers.
6. Encapsulation of cDNA plus PCR Reartents in Droplets
Once cDNA has been synthesized and purified of contaminating detergents, the entire aggregate of cDNA molecules in solution plus qPCR reagents is divided into many tiny compartmentalized reactions, for example, by a droplet making instrument, e.g., a droplet generator such as the Biorad Automated Droplet Generator, which generates 20,000 droplets per sample. Each reaction consists of an extremely small droplet of non-aqueous fluid, e.g., oil (PCR stable, e.g., proprietary formulation from vendor), which contains Taqman-type PCR reagents with gene-specific primers and an oligonucleotide probe, and a small amount of sample. Once droplet generation is complete, the sample consists of an emulsion containing a vast number of individual PCR-ready reactions.
For this step, one can use the PCR probes and related primers for any one or two or more different target genes listed in Table 1 below for overall determination of tumor load, e.g., to determine tumor progression or response to therapy, in single or multiplex reactions. Thus, although in some cases a single set of PCR primers and probes for a particular gene from Table 1 can be included in each droplet, it is also possible to multiplex PCR primers and probes for two or more different genes in each droplet using different fluorescent probes for each primer/probe set, to maximize the detection of tumor cells, given the heterogeneity of gene expression in CTCs. It is also possible to multiplex PCR primers and probes for multiple genes targeting different cancer types in each droplet, thus enabling the broad yet specific detection of multiple tumor types in a single assay.
7. PCR of Droplet Encapsulated cDNA Molecules
Standard PCR cycling is performed on the entire emulsion sample using qPCR cycling conditions. The reaction is carried to 45 cycles to ensure that the vast majority of individual droplet-PCR volumes are brought to endpoint. This is important because, although the reaction is performed with Taqman-type qPCR reagents and cycled under qPCR conditions, the fluorescent intensity of the sample will not be measured during the PCR cycling, but rather in the next step.
8. Detection of Positive Droplets
Since each individual partitioned PCR is brought fully to endpoint before any measurement of fluorescence is performed, each individual droplet will be either a fully fluorescent droplet or will contain virtually no fluorescence at all. This enables the simple enumeration of all positive (fluorescent) and negative (non-fluorescent) droplets.
9. Analysis
Because the upstream RT-PCR targeted a 1:1 RNA:cDNA ratio, each positive droplet should represent a. single originating RNA transcript, This interpretation depends on the number of individual droplets far exceeding the number of target cDNA molecules. In the new process, at one extreme we consider the possibility of a single CTC being isolated and lysed, releasing some number of RNA transcripts that are then reverse-transcribed 1:1 into cDNA, partitioned, PCR-amplified, and enumerated.
We estimate that in the case of a moderately expressed gene, such as the KLK3 gene in prostate cancer cells, each cell contains approximately 80-120 copies of KLK3 mRNA. The Biorad QX200 ddPCR System generates 20,000 droplets, which ensures that for small numbers of isolated CTCs and moderately-expressed target genes there will never be more than one target cDNA molecule per droplet. On the other hand, in cases where the numbers of CTCs reach dozens or hundreds, for moderately-expressing genes there will likely be multiple copies of target cDNA per droplet. In such cases, approximate numbers of originating transcript can be estimated using Poisson statistics.
10. Detecting and Determining Anti-Cancer Regimen Efficacy
The last step includes detecting the expression of particular cancer-specific or lineage-specific genes in the cancer cells, e.g., CTCs, above the low background levels as determined by healthy donor controls (e.g., set at a level of 2.5 transcripts per mL blood), to determine whether a specific anti-cancer regimen is expected to be effective for that patient's specific tumor.
The background levels of these cancer-specific or lineage-specific genes are determined by measuring their expression in CTCs for exosomes) in the blood of many patients without cancer (age-matched to those patients with cancer for a given type of cancer). The predictive value of these cancer-specific or lineage-specific gene markers are then evaluated by monitoring their expression prior to initiation of and during treatment with a specific cancer therapy in many patients over time, e.g., 6 to 12 months, 15 months, 18 months, 21 months, 24 months or more, and determining each patient's progression-free survival and overall survival statistics over each time period. These data are then used to prepare reference standards for each gene and each anti-cancer treatment regimen against which new patient samples can be compared to determine whether a proposed anti-cancer treatment regimen is likely to be effective in a specific patient, and if so, how effective compared to another potential treatment regimen.
For example, in patients with metastatic castration-resistant prostate cancer (“CRPC”), those patients not highly expressing the genes HOXB13 and FOLH1 (PSMA), e.g., a level lower than 2.5 transcripts per mL blood, will have a better overall survival and progression-free survival when treated with anti-androgen therapy than patients who have a high expression level of these two genes, e.g., a level higher than 2.5 transcripts per mL blood.
In particular, by combining microfluidic enrichment of viable CTCs with digital quantitation of CTC-derived RNA, the new methods described herein provide a highly sensitive and specific assay for serial non-invasive sampling of prostate cancer. This approach overcomes a major limitation of CTC analyses to date, namely the microscopy-based quantitation of multiple immunofluorescence-conjugated antibody stains within mixed cell populations, with its associated requirement for calibration and thresholding of multiple fluorescence parameters, followed by manual verification of individual images. The extraordinary high sensitivity and specificity of sequence-based approaches, which are readily multiplexed to interrogate multiple markers simultaneously, provide greatly improved signal over traditional cell imaging methods. In a pilot cohort of men on first line therapy for early recurrence of prostate cancer, we demonstrated the potential utility of quantitative CTC measurements of both normal prostatic transcripts and aberrant RNA products in informing therapeutic choices.
Conceptually, the application of a digital RNA-based PCR output to microfluidic CTC-enriched cell populations presents a number of important advantages. The use of purified whole CTCs in the bloodstream as the source of RNA ensures that the measured signal is derived from invasive cancer cells, as opposed to normal tissues, and hence it enables the use of RNA-based markers that are not unique to cancer. Recurrent cancer-specific markers are rare in prostate cancer, which has limited. the application of mutation-based plasma DNA sequencing. In addition to lineage-based RNA markers, the role of aberrant androgen receptor (“AR”) splice forms in acquired resistance to hormonal therapy necessitates blood-based RNA measurement. In this context, the microfluidic depletion of normal hematopoietic cells from blood specimens is particularly effective in preserving RNA integrity within CTCs, which are not subject to antibody-manipulation or fixation and thus provide excellent signal for digital PCR quantitation. Along with microfluidic CTC isolation, digital scoring of CTC signal for both prostate lineage transcripts and prostate cancer-specific transcripts can be readily automated for high-throughput analyses, making it a realistic tool for clinical applications.
The recent development of multiple potent treatment modalities for metastatic prostate cancer brings with it the need to identify predictive makers of response. To date, the most significant markers have focused on the demonstration of continued activity of the androgen receptor, which is targeted by many therapeutic modalities. Molecular imaging-based strategies to measure androgen signaling have been demonstrated in some cases, but the availability of blood-based sampling would greatly enhance the utility of such monitoring. We have previously reported that scoring of CTCs for expression of the androgen-driven protein PSA versus the androgen-repressed protein PSMA can be translated into an androgen receptor-induced gene (“AR-on”) versus an androgen-repressed gene (“AR-off”) CTC immunofluorescence-based signature.
In treatment-naive patients, virtually all CTCs have AR-on signal, which converts to AR-off following initiation of Androgen Deprivation Therapy (“ADT”). Patients with CRPC, however, most frequently show simultaneous expression of AR-on and AR-off protein signatures, consistent with aberrant AR signaling. In this context, the predictive value of CTC-derived expression of the non-AR target genes HOXB13 and FOLH1 (PSMA) is consistent with altered AR signaling. Germline mutations in HOXB13 have been correlated with increased susceptibility to prostate cancer, and the gene encodes a transcriptional coactivator of AR, which is a known marker of less differentiated prostate cancer, which has also been linked to hormonal therapy resistance in ER-positive breast cancer. FOLH1 is a well-established marker for prostate lineage, normally suppressed by androgen signaling, but co-expressed with PSA in CRPC. Thus, overexpression of these markers within prostate CTCs identify cancers in which altered AR signaling pathways have significant roles in malignant proliferation, lessening the effectiveness of the androgen synthesis inhibitor abiraterone.
AR-V7 has recently emerged as a readily measurable surrogate for acquired androgen pathway independence, predicting resistance to third or fourth line abiraterone or enzalutamide therapy. Discordant results as to the predictive value of AR-V7 measurements most likely result from different CTC or exosome-based detection assays, as well as their application in patients at different stages of treatment and disease progression. For example, in a large retrospective clinical trial, AR-V7 was detectable in CTCs from only 3% of patients prior to fourth line therapy.
The application of a high sensitivity digital CTC assay and the serial sampling of patients before and during therapy provide a novel perspective on the significance of AR-V7 positivity. First, we note that detection of this splice variant in untreated patients at the time of first disease recurrence does not by itself indicate resistance to abiraterone; however, the persistence or emergence of AR-V7 in the setting of drug treatment is highly predictive of adverse outcome. In this context, it is likely that drug sensitive tumor cells are suppressed and AR-V7 directly measures the emergence of drug resistant tumor populations.
Second, the observation that downstream indicators of altered AR signaling (HOXB13 and FOLH1 ) are more commonly elevated than AR-V7 and are more predictive of adverse outcome when measured in pretreatment CTC specimens suggests that AR-V7 is one of a number of mechanisms that limit the efficacy of AR targeted therapies. The recent application of combined paclitaxel and leuprolide therapy in the initial treatment of high-risk prostate cancer shows the utility of risk ao stratification as described herein to enable individualized therapies in advanced disease.
The same techniques can be used to determine the expected efficacy of different therapies used for other types of cancers including melanoma and breast cancer. For example, as shown in Table 1 for melanoma, the following examples of treatments, treatment categories, and drugs can be tested for expected efficacy in specific patients using the assays and methods described herein. Similarly. Table 2 shows drugs and combinations of drugs used to treat breast cancer, which can be tested for efficacy in specific patients using the assays and methods described herein.
As discussed above, the identification of gene transcripts that are highly specific for cancer cells within the context of surrounding normal blood cells is central to the new methods. While many genes are known to be more highly expressed in cancer cells, the vast majority of these genes also typically have at least limited expression in normal tissues, including blood. Given the extraordinary sensitivity required for this assay, complete absence of signal in normal blood cells is essential for high confidence identification of tumor cells in the bloodstream.
Candidate tumor-specific transcripts used to detect CTCs in blood are first selected by analyzing publicly available gene expression data sets derived from breast, prostate, lung, pancreas, and liver cancers and melanoma, as well as our lab-generated single cell RNA-Seq data from CTCs isolated from breast, prostate and pancreatic cancer patients and mouse models of these cancers. Transcripts whose is expression is restricted to tumors and absent or undetectable in blood components are chosen for further downstream analysis. Demonstrating and validating total absence of expression (with the highest level of sensitivity, i.e., Digital PCR assays) in normal blood cells is important. In general, only ˜10% of candidate genes predicted based on computational models or RNA Seq data are truly negative in human blood samples.
In particular, candidate tumor-specific mRNA transcripts for the detection of CTCs were initially identified through the analysis of gene expression data sets (microarray and RNA-Seg) derived previously for human breast, prostate, lung, pancreas, hepatocellular, and melanoma cancers. Specific publicaily available data. sets used for this analysis include The Cancer Genome Atlas (TCCA) (The Cancer Genome Atlas, available online at tcga-data.nci.nih.gov/tega/tegaHome2.jsp) and the Cancer Cell Line Encyclopedia (CCLE) (available online at broadinstitute.org/ccle/home; see also, Barretina et al., The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity, Nature 483:603-607 (2012)). In addition, single-cell RNA-seq gene expression data from CTCs isolated from human patients with breast, prostate, and pancreatic cancers were analyzed (GEO accession numbers GSE51827, GSE60407, and GSE67980) (Aceto et al., Circulating tumor cell clusters are oligoclonal precursors of breast cancer metastasis, Cell, 158:1110-1122 (2014); Ting et al., Single-Cell RNA Sequencing Identifies Extracellular Matrix Gene Expression by Pancreatic Circulating Tumor Cells, Cell Rep, 8:1905-1918 (2014); and Miyamoto et al., RNA-Seq of single prostate CTCs implicates noncanonical Wnt signaling in antiandrogen resistance, Science 349:1351-1356 (2015). Tumor specific transcripts identified through these databases were then compared to human leukocyte RNA-Seq gene expression data (GEO accession numbers GSE30811, GSE24759, GSE51808, GSE48060, GSE54514, and GSE67980). Transcripts that displayed significant differential expression, with high expression in tumors and low or undetectable expression in leukocytes, were then selected for further downstream analysis. Moreover, a literature search was performed to select additional candidate tumor-specific transcripts. Between 50 and 100 candidate genes were selected for each type of human cancer.
For each candidate gene within each specific cancer type, two to four sets of PCR primers were designed to span regions across the target transcript. Primers are synthesized by IDT (Integrated DNA Technologies), probes are labeled with FAM or HEX, ZEN, and IABkFQ to create a probe targeting the middle of the amplicon. Unique features of our PCR primer design methodology necessary for the successful application of digital PCR-based mRNA transcript detection in human CTCs include the following: 1) the specific targeting of the 3′ end of each mRNA transcript, given the proclivity of cellular mRNA transcripts to degrade from the 5′-end, particularly in unfixed, fragile cells such as CTCs; 2) the design of primers to generate amplicons that span introns in order to exclude the unintentional amplification of contaminating genomic DNA, for example from excess contaminating leukocytes in the enriched CTC mixture; and 3) the design of primers to inclusively amplify multiple splice variants of a given gene, given the uncertainty in some cases regarding the clinical relevance of specific splice variants.
The specificity of the primers was first tested by qRT-PCR using cDNA derived from cancer cell lines (representing breast, prostate, lung, pancreas, and liver cancers and melanoma). For each type of human cancer, 2 to 5 established cancer cell lines were cultured and used for initial testing to evaluate PCR primer performance and assess for expression of the target transcript in the specified cancer. To provide an initial test of specificity; the same primers were used to evaluate expression of the target transcript in leukocytes from healthy individuals who do not have a diagnosis of cancer. Leukocytes from a minimum of five different healthy individuals were tested in this phase of testing (mixture of male and female individuals—this was dependent on the type of cancer; i.e. candidate prostate cancer and breast cancer genes required the use of male or female healthy donors only, respectively).
Leukocytes from healthy individuals were isolated from whole blood using Cell Preparation Tubes with Sodium Heparin (CPT) (Becton, Dickinson, and Co., N.J.) following product insert instructions. RNA extraction and first-strand cDNA synthesis was performed for cancer cell lines and isolated leukocytes using standard methods. The specificity of expression of each gene (using 2 to 4 distinct sets of primers for each gene) was tested using qRT-PCR (cell line cDNA as positive controls, leukocyte cDNA from healthy donors as negative controls, and water as an additional negative control). Transcripts present in cancer cell lines, but absent in leukocytes based on qRT-PCR testing were then selected for further validation by droplet digital PCR. The selection criteria to pass this stage of testing were highly stringent, and required qRT-PCR signal to be present in at least one cancer cell line and absent in all healthy donor leukocyte samples tested.
Target transcripts and specific primer pairs that passed the qRT-PCR stage of testing were further validated using droplet digital PCR. For this stage of testing, the CTC-iChip (see, e.g., Ozkumur et al., “Inertial focusing for tumor antigen-dependent and -independent sorting of rare circulating tumor cells,” Sci Transl Med, 5, 179ra147 (2013) was used to process whole blood samples donated by healthy individuals. The CTC-iChip performs negative depletion of red blood cells, platelets, and leukocytes from whole blood, and generates a sample product that is enriched for cells in the blood that do not express leukocyte markers, including CTCs (which should not be present in healthy individuals). For each blood sample, the product from the CTC-iChip was supplemented with an RNA stabilization solution (RNAlater®, Life Technologies) and processed for RNA extraction and cDNA synthesis using standard methods. Droplet digital PCR (Biorad, Calif.) was then used to quantitate the number of transcripts present in each sample based on the specific primer pairs being tested. Samples assessed by droplet digital PCR during this phase of testing included cDNA from cancer cell lines, leukocyte cDNA from healthy donors processed through the CTC-iChip (at least four healthy individuals per primer pair being tested), and water as a negative control.
Criteria for passing droplet digital PCR testing were stringent, and included: 1) the presence of transcript signal in cancer cell lines (at least one cell line with >10 positive droplets); 2) excellent signal-to-noise ratio represented by separation of signal between positive and negative (empty) droplets; 3) minimal or absent droplet signal in healthy donors (<3 droplets per healthy donor); and 4) absent droplet signal in water (0 positive droplets).
Primers that amplified transcripts specifically in cell lines and not in leukocytes in the above droplet digital PCR. testing were then subjected to detailed testing of sensitivity of signal. Using single cell micromanipulation, precise numbers of cancer cells (1, 5, 10, 25, and 50 cells) were spiked into whole blood donated by healthy individuals, and then processed through the CTC-iChip. Each sample was then processed as above for testing with droplet digital PCR, and evaluated for sensitivity to ensure the signal was sufficient for the desired clinical application.
The above stringent procedure of evaluating candidate genes and primers using qRT-PCR and droplet digital PCR resulted in a final primer list consisting of approximately 10% of the initial list of 50-100 candidate genes for each type of cancer (total of approximately 400 initial candidate genes). These primers are then further evaluated. for signal in patient CTCs using blood samples donated by cancer patients undergoing cancer treatment at the MGH Cancer Center, collected under an IRB-approved clinical protocol. Key to this portion of the evaluation is a comparison with blood collected from healthy individuals without a diagnosis of cancer. The following Table 3 lists the primers and probes for that have been developed thus far using these methods for the specific detection of CTCs from patients with prostate, breast, hepatocellular, pancreatic, lung, and melanoma cancers using droplet digital PCR.
While a single gene for each cancer type could be used, the presence of multiple genes within each panel is useful both for sensitivity (CTCs are heterogeneous even within individual patients in their expression patterns) and specificity (detection of multiple gene signals confers added confidence that this o represents a true cancer cell signature).
The gene list provided below in Table 3 includes transcripts that are unique to specific types of cancer (e.g., highly specific markers of prostate or breast or liver cancers), as well as genes that are shared by several cancer types, e.g., all epithelial cancer types (and thus may serve as pan-cancer markers), and genes that are induced in certain conditions (e.g., active androgen signaling in prostate cancer or active estrogen signaling in breast cancer). Thus, each type of cancer was assigned a specific panel of genes that is designed for optimal sensitivity, specificity, and clinically actionable information for the given cancer type.
In addition, primers described in Table 4 are designed to pre-amplify sonie of the genes listed in Table 3, while maintaining their high specificity. If STA is a method of choice, these nested primers become additional components of each cancer panel.
The following Table 3 provides a list of names of genes (with (Genbank ID) and Sequence Identification numbers (SEQ ID NO)), along with cancer types for which they are selective (Br: breast, Lu: lung, Li: liver, Pr: prostate, Panc: pancreatic, Mel: melanoma). In addition, optimized primer sets are listed for each gene (primers 1 and 2), along with the composition of the fluorescent primer probes (e.g., 6-FAM™ (blue fluorescent label) or HEX™ (green fluorescent label) for tagged probes, and ZEN-31 ABkFQ quencher) for optimal visualization of the digital PCR product.
The following Table 4 lists nested primers designed to specifically pre-amplify the regions targeted by primers listed in Table 3.
To improve the detection of tumor-specific mRNA from minimal amounts of RNA derived from CTCs, we established a multiplex assay capable of testing many different gene transcripts from a minute amount of CTC-Chip product. This combines the higher sensitivity/specificity of using multiple independent genes, with to the fact that the amount of input template is limited (and hence should not be diluted into multiple reactions). Our assay includes 4 genes per reaction, with each gene being resolved uniquely in 2-dimensional space by selecting different ratios of -fluorescent conjugated primers. Thus, in a single reaction, we can independently measure 4 gene transcripts without having to dilute the template, For different cancers, we have gone as far as up to 4 different reactions (i.e., up to 20 different gene transcripts), and with application of nested RT-PCR digital assays, there is no limit to the number of reactions that can be performed.
This multiplex strategy achieves the ideal balance between analyzing multiple transcripts and hence ensuring against heterogeneous variation in cancer cell expression patterns), but not diluting the input material by performing multiple independent PCR reactions. Depending on tumor types and the number of genes required for optimal signal, we have developed assays ranging from 2-4 multiplex reactions (each multiplex reaction testing for 4-genes). Thus, without undue dilution of input template, we can interrogate the product of a single CTC for expression of o anywhere from 8 to 16 different genes. It is important to the assay to be able to add the signal from all of these genes (i.e. cumulative signal), while also having individual gene results (to optimize signal/noise at the individual gene level, and also gather information from specific signaling pathways that each gene interrogates—for example androgen signaling in prostate CTCs).
To display the results of the multiplex reaction in a single view (and hence differentiate amplification of each gene is isolation), we varied the concentrations of the two fluorescent probes (FAM (blue) and HEX (green)). By doing this, each individual gene amplification reaction has a unique combination of FAM/HEX signal that reflects the composition of the gene-specific primers, and hence identifies the gene-specific PCR product. In 2-dimensional space, we can illustrate the signal position of 4 different gene amplification products produced from a single multiplex reaction. As applied to digital PCR using droplets to encapsulate each PCR reaction, this method separates the targets into individual clusters by modifying the binary signal amplitude of positive droplets, which are displayed quantitatively. As predicted, this method allows both cumulative scoring of total signal for multiple genes (e.g., 16 markers in a total of 4 reactions), while also retaining the ability to quantify the signal from each individual gene target.
Virtually all patients with metastatic prostate cancer experience an initial clinical response following androgen deprivation therapy (ADT). As tumors develop castration-resistance, half of patients have a sustained second remission following treatment with the potent androgen synthesis inhibitor abiraterone (e.g., ZYTIGA®), while others have only a short response and would hence benefit from alternative or combination therapies. To test whether CTC-derived signatures provide predictive markers of response to anti-androgen therapies after ADT, we prospectively evaluated 25 patients with metastatic CRPC who were initiating abiraterone therapy in the first-line setting.
Remarkably, an elevated CTC-Score at pretreatment baseline was predictive of early progression, an effect that was driven by expression of FOLH1 (PSMA) and HOXB13 within CTCs. Both of these markers have been associated with aberrant androgen receptor (“AR”) signaling, and HOXB13 has been associated with more aggressive, hormone refractory, prostate cancer. The correlation between HOXB13 and FOLH1 CTC-derived signal was evident for radiographic progression-free survival (HOXB13, P=0.015; FOLH1, P=0.015), as well as overall survival (HOXB13, P=0.017; FOLH1, P=0.017). In contrast, the pretreatment serum PSA protein level is correlated with reduced overall survival, but it is not indicative of radiographic progression-free survival or PSA progression, and it is no longer correlated with outcome following initiation of treatment.
The early detection of epithelial cancers at a time when they can be surgically resected or irradiated provides the best chance of cure, and the administration of adjuvant chemotherapy in the setting of minimal cancer dissemination is far more effective in achieving cure than the treatment of established metastatic disease. Just as important as early detection is to select a proper anti-cancer therapy. The new methods described herein use the d-CTC assay methods to not only provide early detection of a specific type of cancer, but in combination with the appropriate reference standards, can be used to determine and compare the predicted efficacy of different therapeutic regimens in a specific patient depending on his or her tumor gene expression profile.
The d-CTC assays described herein can he used for both initial screening and to determine the best therapeutic regimen. The use of the new d-CTC assays described herein, in which each CTC (no matter how intact or pre-apoptotic) can give rise to hundreds of molecular signals, dramatically enhances the ability to detect and monitor CTCs in patients with known cancer, and to quantitatively monitor and analyze their response to therapeutic interventions. Beyond scoring for cell numbers through molecular markers, specific interrogation of mutations or cancer-associated rearrangements (e.g., EML4-ALK in lung cancer) can he achieved with comparable sensitivity.
As discussed in the examples below, the new methods described herein are illustrated in prostate cancer, where the analysis demonstrated that an elevated CTC-Score at pretreatment baseline was predictive of early progression, an effect that was driven by expression of FOLHI (PSMA) and HOXB13 within CTCs.
The invention is further described in the following examples, which do not limit the scope of the invention described in the claims.
The following materials and methods were used in the Examples set for the below.
This example provides a general digital CTC assay protocol that can be used for the methods described herein. Different aspects of this general protocol were used in Example 2 below.
1. Patient blood is run through I-Chip, version 1.3 M or 1.4.5 T. Sample is collected in a 15 mL conical tube on ice.
2. Sample is spun down at 4C. Supernatant is decanted and. SUPERase™ In (DTT independent RNAse inhibitor)+RNALater® Stabilization Solution (prevents RNA degradation by inhibiting RNAses) is added to the pellet. Sample is flash frozen and placed at −80 until further processing. Samples are stable at −80.
3. There are two different processing protocols for RNA purification to cDNA synthesis that were used in the examples described below.
Approach 1
Approach 2
4. cDNA (synthesized from Approach 1 or 2) can be processed in two different ways:
5. cDNA (from step 4a or 4b), Biorad. Supermix™ for probes, primer or primers (for gene of interest; up to 4 different primers (FAM and HEX) can be multiplexed) were added in a total volume of 22 μl.
6. Droplets were generated (˜15,000-18,000 droplets per well).
7. Droplet Sample were put in a PCR machine. The PCR conditions were different than Biorad recommendations. We used a step-down rather than a slow ramp to ensure that all droplets reach the same temperature. This is different than what both RainDance and Biorad uses. Better results (i.e., more signal and more separation between positive and negative droplets) can be obtained with the step-down rather than the gradient.
8. After the PCR, positive droplets were counted in a ddPCR machine.
9. Data is collected and analyzed using TIBCO® Spotfire® analysis software.
The reagents, reagent concentrations, and reaction volumes are provided below:
Reagents:
Testing Relevant Cell Lines
Per Single Reaction:
A master-mix containing ddPCR supermix, cDNA, and water were aliquoted into so wells and 1.1 μl of each the primer was added to each well and mixed well.
Patient Samples
Per Single Reaction for Individual Genes
Per Single Multiplexed Reaction for Multiple Genes
When testing multiple patients against a gene-specific primer or multiplexing primers against multiple genes, a master-mix, which includes the ddPCR supermix and primers, was aliquoted into wells followed by addition of patient cDNA to each well and mixed well.
Patients with a diagnosis of prostate cancer provided informed consent to one of two Institutional Review Board approved protocols, DF/HCC 05-300 or DF/HCC 13-209. Patients donated 20 mL of blood for CTC analysis, including patients with metastatic prostate cancer and patients with localized prostate cancer. Formalin-fixed, paraffin-embedded primary tumor tissues from patients were sectioned, and subjected to RNA extraction, prior to processing for droplet digital PCR (see below).
CTCs were isolated from fresh whole blood following leukocyte depletion using the microfluidic CTC-iChip as previously described. To maximize recovery of intact CTCs with high quality RNA, blood samples were processed within 4 hours of being collected from the patient. The total time for CTC isolation after receipt of fresh blood samples in the lab was approximately 2.5 hours. Briefly, whole blood samples were spiked with biotinylated antibodies against CD45 (R&D Systems, clone 2D1) CD66b (AbD Serotec, clone 80H3), and CDI6 (Janssen Diagnostics), followed by incubation with Dynabeads MyOne Streptavidin T1 (Invitrogen) to achieve magnetic labeling and depletion of white blood cells. After processing of whole blood with the CTC-iChip and collecting the enriched CTC product on ice, cells were centrifuged at 4750 rpm and flash frozen in liquid nitrogen in the presence of RNAlater® (Ambion) to preserve RNA integrity.
CTC samples were subjected to RNA extraction using the RNeasy Plus Micro Kit (Qiagen), followed by reverse transcription using SuperScript III First-Strand Synthesis System (Life Technologies). cDNA and primers/probes were combined with ddPCR Supermix for Probes (Bio-Rad) in a 96-well plate and loaded onto Bio-Rad's automated droplet generator. Droplets were amplified using a modified 45-cycle PCR with a 70°C. step-down in between the denaturation and annealing steps. Following thermal cycling, amplified droplets were detected via fluorescence with the QX200 Droplet Reader System (Bio-Rad).
A list of potential gene candidates was generated using publically available databases as well as single cell RNA-sect data. A two-step approach using both RT-PCR and ddPCR was developed to validate these genes. In the first step, cDNA prepared from healthy donor leukocytes and prostate cell lines (LNCaP, PC: 3, VCaP) was tested against primers using the ABI 7500 and Bio-Rad CFX96 Real-Time PCR Systems, Leukocytes were isolated from male healthy donors using BD Vacutainer® CPT™ Cell Preparation Tubes. Total RNA was extracted from isolated leukocytes and prostate cancer cell lines using RNeasy Micro Kit (Qiagen) and SOOng reverse transcribed with SuperScript III First-Strand. Synthesis System (Life Technologies). 1 ng of total cDNA was used per RT-PCR reaction. Genes expressed in cell lines and absent in healthy donor leukocytes by RT-PCR were further validated in a second step using ddPCR. cDNA prepared from CTC-iChip products of healthy donor males and patients was tested against genes using the ddPCR platform. Differential expression between healthy donors and patients determined by droplet count was used to select genes for the assay (Table 5).
To test the limit of detection for the ddPCR assay a series of cell spiking experiments were performed using the CTC-iChip. Single LNCaP cells were manipulated using 10 urn Eppendorf TransferMan® NK2 transfer tips into Kolliphor P188 buffer and spiked into healthy donor male blood. The spiked samples were prepped for processing as described above and run through the CTC-iChip. RNA and cDNA were isolated and prepped from the CTC-iChip products and run on ddPCR using Reactions 1 and 2.
Given the limitations inherent in fluorescence-based imaging and scoring of CTCs admixed with contaminating blood cells, we tested whether RNA-based digital PCR quantitation could provide a higher throughput, more sensitive and more specific readout. Microfluidic (CTC-iChip) depletion of hematopoietic cells from blood samples achieves 104 to 105 purification of CTCs, with approximately 500 WBCs remaining per 1 ml of processed whole blood. The high quality of RNA within the purified CTCs allows the application of new and highly robust digital droplet-PCR technologies, in which rare cDNA templates are encapsulated within lipid droplets, followed by PCR amplification and fluorescence scoring of positive droplets. The combination of microfluidic whole cell isolation of CTCs from blood and RNA-based digital PCR of CTC-derived transcripts (d-CTC assay) allows the use of prostate tissue lineage-specific mRNAs as highly specific markers to monitor metastatic prostate cancer (
To test the application of this strategy for prostate CTC detection, we first identified a panel of prostate-specific transcripts whose expression is virtually absent in normal hematopoietic cells, even following high sensitivity droplet-PCR amplification. We selected multiple markers, both to address the known heterogeneity of prostate cancer cells, as well as to allow interrogation of cellular signaling pathways, including androgen receptor activity. We derived an initial set of 40 candidate genes, both from RNA sequencing of single prostate CTCs (Miyamoto, et al. Science 2015; 349: 1351-6.), as well as from publicly available expression databases.
Twenty-nine transcripts were identified as having high levels of expression in prostate tissue and/or prostate cancer, but without detectable RNA reads in normal blood cells contaminating the microfluidic CTC-iChip product (
To validate the assay, we first micro-manipulated individual cells from the prostate cancer line LNCaP, and introduced these into 2.5 mL, of whole blood from healthy donors, followed by processing through the CTC-iChip and droplet digital PCR quantitation, Introduction of a single LNCaP cell into a control blood sample generated 150 positive droplets (SD=65.3), with a progressive increase in signal as 3, 5, 10, and 50 cells were spiked into the blood samples (5562±1853 droplets for 50 prostate cell input) (
We tested the d-CTC detection strategy in 12 patients with metastatic prostate cancer, compared with 8 patients with localized prostate cancer, 34 male healthy blood donors (19 >50 years old; 15 <50 years old), and 5 female controls. The so observed signal across all 8 markers is shown in
To compare the digital CTC assay with more traditional immunofluorescence-based detection of CTCs, pre-treatment blood samples were obtained from 25 patients with mCRPC enrolled on a prospective clinical trial of abiraterone in the first-line setting. Each blood sample was processed through the CTC-iChip and the output was equally divided between immunofluorescence-based microscopy scoring versus d-CTC assay. As expected, concordance between microscopic scoring and digital readouts was evident in samples with high numbers of CTCs, but the d-CTC assay was far more sensitive in identifying cases below microscopic detection, even using sophisticated multispectral fluorescence-based imaging. (R2=0.01; P=0.6;
Example 4—Detection of AR-V7 and TMPRSS2-ERG Prostate Cancer Specific Transcripts in CTCs
While recurrent missense mutations are rare in prostate cancer, two specific RNA fusion transcripts are characteristic of this tumor type. To complement the quantitation of prostate lineage-based transcripts in CTCs, we developed droplet PCR assays for both the TMPRSS2-ERG fusion transcript, which is present in 50% of cases, and the AR-V7 RNA splice variant, which constitutes a marker of resistance to anti-androgen therapy. Both tests were highly specific and sensitive when applied to prostate cell lines spiked into control blood specimens, followed by CTC-iChip purification (
Example 5—Prospective Serial Monitoring of Patients on First-Line Abiraterone Therapy
Virtually all patients with metastatic prostate cancer experience an initial clinical response following androgen deprivation therapy (ADT). As tumors develop castration-resistance, half of patients have a sustained second remission following treatment with the potent androgen synthesis inhibitor abiraterone, while others have only a short response and would hence benefit from alternative or combination therapies. To test whether CTC-derived signatures provide predictive markers of response to anti-androgen therapies after ADT, we prospectively evaluated 25 patients with metastatic CRPC who were initiating abiraterone therapy in the first-line setting.
We first applied the prostate lineage CTC-Score at the baseline pretreatment time point (CID1), at on-treatment time points of 2 weeks (CD15), 4 weeks (C2D1), 12 weeks (C4D1), and at the time of disease progression and discontinuation of therapy (
AR-V7 expression has been detected in patients with metastatic CRPC in the second line or greater setting, and it has been shown to predict acquired resistance to abiraterone when administered to patients with such advanced disease. Using the digital CTC assay, AR-V7 was detectable in 4/20 patients tested at the pretreatment baseline time point. In this first line setting, quantitative detection of AR-V7 was not predictive of radiographic progression or overall survival. However, serial monitoring of these patients indicated that the predictive value of AR-V7 increased during the first three courses of therapy, achieving a high predictive value for radiographic progression (P=0.026), and overall survival (P<0.001) using the 3-month time point.
This observation is consistent with the initiation of anti-androgen therapy suppressing the proliferation of susceptible tumor cells, with the emergence of AR-'V7-driven resistant disease in patients destined for early relapse. In contrast to AR-V7, the TMPRSS2-ERG translocation was not enriched as a function of anti-androgen therapy and it was not correlated with acquired resistance.
The AR-V7 splice variant measures one of several mechanisms linked to anti-androgen resistance, whereas expression of HOXB13 and FOLH1 are downstream indicators of aberrant androgen signaling. We therefore compared the predictive value of these orthogonal markers, either alone or in combination, in pretreatment CTCs drawn from patients in our prospective first line abiraterone cohort (
Example 6—Persistent Estrogen Receptor Signaling in CTCs Identifies Metastatic Breast Cancer Patients Who will be Resistant to Hormonal Therapy
In patients with hormone receptor-positive (“HR+”) disease, persistent expression of a six-gene resistance signature (“RS”) associated with estrogen signaling correlates with adverse outcomes, including shorter time to progression (TTP) and poor overall survival (OS) (p=0.02 (OS), 32 0.003 (TTP)) when treated with drugs that target the estrogen-signaling pathway such as ER inhibitors (e.g., tamoxifen), selective ER degraders (“SERDs” such as fulvestrant), and aromatase inhibitors (AI), which block the production of estrogen (e.g., anastrozole, letrozole, and exemestane), e.g., in combination with CDK4/6 inhibitors.
Only half of the patients with a high RS score harbor ESR1 mutations, suggesting the involvement of additional mechanisms for drug-refractory estrogen signaling. Thus, digital RNA scoring of CTCs enables early monitoring of treatment response and provides the potential for a noninvasive measurement of drug effect on intracellular ER signaling pathways.
Patients were consented through an Institutional Review Board approved protocol for CTC collection (DFHCC 05-300. For the initial clinical benchmarking o of the assay. 10-20 ml of peripheral blood (17 ml average) was collected from a total of 78 unique patients, representing 85 samples. These include pretreatment samples from 23 Stage I, 24 Stage 11 and 8 Stage III unique patients, and 30 on-treatment samples from 23 unique Stage IV patients. 33 samples from female healthy donors (HD) were obtained from the blood bank (9 ml average volume).
To determine if CTC monitoring through the breast CTC-ddPCR assay is predictive of treatment outcome and overall survival, we prospectively collected pretreatment and 3-4 weeks on-treatment draws from metastatic breast cancer patients initiating a new therapy (TRACK cohort). At least one sample was collected from 52 patients; 50% of the patients received some form of endocrine therapy, 10% received chemotherapy, 13% received anti-HER2 therapy while the rest were on a therapy that does not fall into any of these categories. To validate the assay detection characteristics established in the initial phase of assay development on the TRACK cohort, we also collected samples from 10 healthy women with negative breast biopsies after suspicious mammogram findings.
The CTC-iChip technology for enrichment of CTC from whole blood, through the negative selection of RBC, WBC, and platelets, has been described above. In short, 8-20 ml of whole blood was incubated with biotinylated antibodies against the WBC markers CD45 (R&D Systems, clone 2D1), CD66b (AbD Serotec, clone 80H3), and CD16 (Janssen Diagnostics). Dynabeads MyOne Streptavidin T1 (Invitrogen) were then added to tag the WBC. The blood was subsequently fed through the CTC-iChip, where RBC and platelets were removed through size-based separation, while WBCs were depleted magnetically. The CTC-enriched product was centrifuged, preserved in RNA-later (Ambion) and flash-frozen for long-term storage.
17 markers for breast CTCs (AGR2, CXCL13, CXCL14, EFHD1, FAT1, FAT2, MGP, MUC16, PGR, PIP, PRAME, SCGB2A1, SERPINA3, SFRP1, SFRP2, TMPRSS4, WFDC2) were selected through literature search and mining in-house and publically available datasets including GTeX® and Oncomine® for markers expressed in breast cancer but not in whole blood. The specific genes and IDT probes used in the finalize breast cancer assay are listed in Table 6.
For the initial in vitro testing the panel, and to determine the linearity of the signal, we micro-manipulated increasing numbers of BRX-142 cells into 4 ml of HD blood and ran the samples through the CTC-iChip as described above. RNA was extracted using RNeasy® Micro Kit (Qiagen) and a quarter of it was then used for cDNA synthesis and amplification using the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech). To establish the clinical specificity and sensitivity of the assay, CTC-iChip products from healthy donors and patients were similarly processed. ddPCR analysis was performed using predesigned Taqman-based qPCR assays (Invitrogen) and the ddPCR Supermix for Probes (No dUTP)(Biorad), on the Biorad ddPCR system. For markers detected with multiple probes, the average transcript number was used.
To normalize for differences in blood volumes among samples, all raw data were corrected for the blood-volume equivalent used in each ddPCR reaction. To further normalize the signal, the median and the doubled standard deviation of the expression of each marker within the 33 test healthy donors was established. The product of the two values was then subtracted from every patient and healthy donor sample analyzed in this study. The total CTC score was calculated by summing the normalized expression of all markers in a sample without additional weighing and reported as transcripts/ml of blood-volume equivalent used.
Probes specific for the L536R, Y537C, Y537N, Y537S and D538G ESR1 mutations have been previously published. Their amplification efficiency, as well as that of their respective wild-type probes, was tested on synthetic sequences (data not shown). We established the ability of Y5375 to detect mutations present in cDNA from CTC-enriched IFIL product by micro-manipulating increasing number of BRx-68 cells in healthy donor blood, and then processing it as described above. 18-cycle WTA was performed using ⅓ of the extracted RNA with the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech) following manufacturer protocols; 1 μl of undiluted WTA product was used per reaction. Patient samples were treated in identical manner; probe specificity was established at 100% after testing at least 5 halthy donor samples per probe. The cut-off for the presence of ESR1 mutation was established at >3 positive droplets.
Receiver-operator curve analysis was performed to establish the specificity and sensitivity of each marker and the total CTC score for different cancer stages in our initial test cohort. The analysis was performed in R using the ROCR package. The specific script is available upon request. Wilcoxon tests were performed to establish significance of the AUC. The specificity and sensitivity in Stage IV cancer were validated using a new set of healthy donors and the pretreatment samples from the TRACK cohort.
To determine the pre-treatment division point of high/low RS score, resampling using leave-one-out jack-knife was applied to the algorithm of Contal-O'Quigley to produce a division point that maximizes the difference in overall survival between the two resulting sub-groups. Comparisons of clinical variables between resulting groups are based on Fisher's exact tests for categorical characteristics and exact Wilcoxon rank-sum tests for continuous characteristics.
Survival analyses based on changes in CTC scores during treatment and on RS scores were analyzed using log-rank test, as subgroups within those comparisons had no events, preventing the use of cox model statistics. Unsupervised clustering of pretreatment and 3-4 week on-treatment samples was performed using Ward's minimum variance method.
To develop an RNA expression signature to detect breast cancer cells within the background of contaminating normal blood cells, we first analyzed RNA-Sett and microarray gene expression data sets derived from normal breast tissue, breast cancer and whole blood as described above. We ultimately selected 17 markers whose expression is virtually absent in blood cells, but strongly expressed in breast-derived tissues. The markers include breast lineage-specific transcripts (PGR, SCGB2A1, PIP) and transcripts highly expressed in breast cancer (MGP, EFHD1), as well as genes implicated in endocrine signaling (SERPINA3, WFDC2), endocrine drug resistance (AGR2), cancer growth and metastasis (MUC16, TMPRSS4), cellular signaling (FAT1, FAT2, SFRP1, SFRP2), epithelial-derived cytokines (CXCL13, CXCI14), and oncofetal antigens (PRAME).
Single cell RNA sequencing revealed high, but variable, expression of the 17 markers in 15 individual CTCs isolated as single cells from the blood of women with breast cancer; 5 similarly analyzed single WBCs had negligible expression of these genes. Unlike traditional mutational signatures designed to distinguish between breast cancer and normal breast tissue, the CTC RNA signature panel was intended to inform on the tissue of origin of non-hematopoi etic cells within a blood sample, with the potential to provide actionable clinical information for the diagnosis and monitoring of breast cancer.
To interrogate the entire panel of biomarkers for subsets that may be correlated with endocrine-refractory disease we performed unsupervised clustering of the breast assay components at 3-4 weeks after start of endocrine treatment in the subset of patients with I IR disease, reasoning that treatment-induced expression changes may distinguish responding from non-responding patients. Indeed, we identified 6 genes (PIP, SERPINA3, AGR2, SCGB2A1, EFHD1 and WFDC2) within a Resistance Signature (RS), whose expression was associated with rapid disease progression (within 120 days) and poor survival (p=0.0031 and p=0.0175 respectively, Fisher's exact test) (
Remarkably, all 6 RS transcripts are significantly enriched in ER+tumors compared to ER-tumors in the TCGA database, suggesting that their expression may be related to estrogen signaling. Indeed, a metascore based on the mean expression of the RS genes shows a highly significant correlation with the Hallmark Estrogen Receptor (Late) gene signature from the Molecular Signatures Database across multiple publically available gene expression datasets (R=0.70; p=1.7e-70). The RS gene metascore is also correlated with multiple other MSigDB sets related to estrogen signaling and endocrine resistance, resulting in median correlation coefficients of 0.54 and 0.51 respectively (
Activating mutations in the ESR1 gene encoding ER have been reported in breast cancers with acquired resistance to hormonal therapy and are thought to mediate persistent, ligand-independent ER signaling. In our 3-4 week HR+ patient cohort, 2/20 women had been diagnosed as having an ESR1 mutation based on tumor re-biopsy during the course of clinical care. See Table 7 below, which shows ESR1 mutations detected by CTC-ddPCR and SNapShot genotyping in HR+ patients on endocrine treatment and TNBC patients (negative control).
However many of the patients ( 8/20) had only undergone genotyping of their primary tumors, while others had had no genotyping performed ( 5/20). To noninvasively ascertain ESRI mutation status in all patients, we established a specific digital PCR mutation assay using CTC-derived RNA template, with probes specific for L536R, Y537C, Y537, Y537s and D538G, which together account for the majority of ESRImutations (20, 22). The sensitivity and accuracy of the assay was confirmed by spiking single cells carrying the Y537S mutation into blood samples, followed by microfluidic CTC isolation and ddPCR performed on whole transcriptome-amplified cDNA from the product.
Using this CTC-based assay, additional 3 patients within our HR+ cohort were found to harbor ESRI mutations, resulting in a total mutation frequency of 5/20 (25%), a prevalence that is consistent with previous studies of heavily treated metastatic HR+ breast cancer (20) (Table 7). Interestingly, the cases with ESRI mutations were overlapping but not identical with those expressing the RS gene signature (
Of the 5 cases with an ESR1 mutation, 3 had the RS gene signature reflecting persistent ER signaling, and of 6 women with the RS signature. 3 had an ESRI mutation. In our HR+ cohort, high RS score at 3-4 weeks after initiation of endocrine therapy, was highly prognostic of both poor survival and faster time to progression (
The 17 genes that constitute the breast CTC signature were selected to include multiple tissue-derived and cancer-related transcripts with absent expression in blood cells that contaminate the enriched CTC product. As such, the 6 genes included in the RS sub-signature do not represent canonical ER targets, but their expression is, nonetheless, highly correlated with both ER signaling and resistance to endocrine therapy. Their persistent expression within CTCs after treatment initiation identifies women with greatly reduced response to hormonal therapy and shortened overall survival on treatment. The fact that this CTC signature emerges 3-4 weeks after start of hormonal therapy suggests that it may reflect drug-mediated effects on tumor cells. Initiation of novel ER-targeting therapy should suppress ER signaling in susceptible tumor cells, whereas persistent pathway activity would remain evident in cancer cells in which the drug fails to hit its intended target.
In addition, in this study, SSRI mutations were noted at the expected frequency, but they were less predictive of adverse clinical outcome than persistent ER signaling as measured by the CTC expression signature.
It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.
This application is a Divisional of application Ser. No. 16/344,557, filed on Apr. 24, 2019, which is a 371 U.S. National Phase Application of PCT/US2017/058855, filed on Oct. 27, 2017, which claims priority from U.S. Provisional Application Ser. No. 62/413,952, filed on Oct. 27, 2016, which is incorporated herein by reference in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 62413952 | Oct 2016 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 16344557 | Apr 2019 | US |
| Child | 17826834 | US |
