DCT

1:21-cv-01126

TwinStrand Biosciences Inc v. Guardant Health Inc

Log In

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 1:21-cv-01126, D. Del., 08/03/2021
  • Venue Allegations: Venue is alleged to be proper in the District of Delaware because Defendant Guardant Health, Inc. is a Delaware corporation.
  • Core Dispute: Plaintiffs allege that Defendant’s genetic-sequencing services and related products, marketed as "Digital Sequencing Technology," infringe four patents related to methods for increasing the accuracy of DNA sequencing.
  • Technical Context: The technology, known as Duplex Sequencing, aims to reduce the high error rate of standard Next Generation Sequencing (NGS), enabling the reliable detection of rare genetic mutations for applications such as non-invasive cancer screening ("liquid biopsies").
  • Key Procedural History: The complaint alleges a history of pre-suit interactions, including that Defendant made unsuccessful attempts to license the asserted patent family. It further alleges that Defendant repeatedly encountered the asserted patents and related applications during the prosecution of its own patents in both the U.S. Patent and Trademark Office and in European opposition proceedings, as well as in inter partes review proceedings.

Case Timeline

Date Event
2012-03-20 Earliest Priority Date for Asserted Patents
2014-01-01 Accused Products Launch (Starting in 2014)
2019-05-14 U.S. Patent No. 10,287,631 Issues
2020-06-23 U.S. Patent No. 10,689,699 Issues
2020-08-25 U.S. Patent No. 10,752,951 Issues
2020-09-01 U.S. Patent No. 10,760,127 Issues
2021-08-03 Complaint Filing Date

II. Technology and Patent(s)-in-Suit Analysis

U.S. Patent No. 10,287,631

  • Patent Identification: U.S. Patent No. 10,287,631, "Methods of Lowering the Error Rate of Massively Parallel DNA Sequencing Using Duplex Consensus Sequencing," issued May 14, 2019.
  • The Invention Explained:
    • Problem Addressed: The patent’s background section describes that standard "massively parallel" DNA sequencing technologies have a significant error rate of approximately 1%, creating a level of "artifactual heterogeneity" that obscures the detection of true, rare genetic variants that exist at frequencies below this threshold (e.g., circulating tumor DNA) (’631 Patent, col. 2:1-6).
    • The Patented Solution: The invention, termed Duplex Consensus Sequencing (DCS), solves this problem by using the redundant information stored in double-stranded DNA. The method involves uniquely tagging both complementary strands of an original DNA molecule with adapter molecules. After amplification and sequencing, the reads derived from one original strand are compared to the reads derived from its complementary partner strand. Because true mutations are present on both strands, whereas sequencing or amplification errors are typically confined to only one strand, this comparison allows for the computational removal of artifacts to generate a highly accurate "consensus sequence" (’631 Patent, Abstract; col. 4:19-24).
    • Technical Importance: This error-correction methodology enables the detection of genetic mutations at frequencies far below the error rate of conventional sequencing, a critical capability for early, non-invasive cancer detection and monitoring (Compl. ¶13).
  • Key Claims at a Glance:
    • The complaint asserts independent claim 1 (Compl. ¶37).
    • Essential elements of Claim 1 include:
      • Ligating a double-stranded target nucleic acid molecule to an adapter molecule, wherein the adapter comprises both (a) a single molecule identifier (SMI) sequence that uniquely labels the molecule and (b) a strand-distinguishing nucleotide sequence that creates a region of non-complementarity.
      • Amplifying each strand of the resulting complex.
      • Sequencing the amplified products (amplicons) to generate first and second strand sequence reads.
      • Grouping the sequence reads into a family based on the shared SMI sequence.
      • Separating the reads into first and second strand sets based on the region of non-complementarity.
      • Confirming the presence of reads from both the first and second strands.
      • Comparing the first and second strand sequence reads.
      • Generating a "high accuracy consensus sequence read" that includes only the nucleotide positions where the compared reads are complementary.

U.S. Patent No. 10,689,699

  • Patent Identification: U.S. Patent No. 10,689,699, "Methods of Lowering the Error Rate of Massively Parallel DNA Sequencing Using Duplex Consensus Sequencing," issued June 23, 2020.
  • The Invention Explained:
    • Problem Addressed: The patent addresses the same fundamental problem as the ’631 Patent: the need to overcome the inherent error rate of high-throughput sequencing to accurately detect low-frequency mutations in heterogeneous samples, such as circulating DNA obtained from a bodily sample (’699 Patent, col. 1:47-2:6).
    • The Patented Solution: The invention provides a method for analyzing circulating DNA by first converting the DNA molecules into a population of "non-uniquely tagged parent polynucleotides," where each tagged molecule is nonetheless "substantially unique." After amplification and sequencing, the resulting reads are grouped into "families" that share the same identifier sequence and the same start and stop positions. The reads within each family, which all derive from the same original parent molecule, are then computationally "collapsed" to yield a single, high-accuracy base call for that genetic position, thereby filtering out stochastic errors (’699 Patent, Abstract; col. 20:1-21).
    • Technical Importance: The method provides a framework for processing sequence data from circulating DNA to achieve high-fidelity mutation detection, which is crucial for liquid biopsy applications (Compl. ¶13).
  • Key Claims at a Glance:
    • The complaint asserts independent claim 1 (Compl. ¶58).
    • Essential elements of Claim 1 include:
      • Providing a population of circulating DNA molecules from a subject.
      • Converting the DNA into "non-uniquely tagged parent polynucleotides," where each tagged molecule comprises a DNA sequence and an identifier sequence (barcode) and is "substantially unique" within the population.
      • Amplifying the population of tagged polynucleotides.
      • Sequencing at least a portion of the amplified products to generate sequence reads.
      • Grouping the reads into "families," where each family comprises reads with the same identifier sequence and the same start and stop positions.
      • Collapsing the sequence reads in each family to yield a base call for one or more genetic loci.

Multi-Patent Capsule: U.S. Patent No. 10,752,951

  • Patent Identification: U.S. Patent No. 10,752,951, "Methods of Lowering the Error Rate of Massively Parallel DNA Sequencing Using Duplex Consensus Sequencing," issued August 25, 2020 (Compl. ¶11).
  • Technology Synopsis: This patent describes a method for increasing sequencing accuracy by tagging double-stranded DNA molecules with "duplex tags." After sequencing, the resulting reads are bioinformatically sorted into "paired" reads, which derive from both complementary strands of an original molecule, and "unpaired" reads, which derive from only one. The method comprises quantifying these distinct categories of reads and their read depths, providing a basis for statistical analysis to distinguish true mutations from artifacts (’951 Patent, Abstract; Compl. ¶77).
  • Asserted Claims: Claim 1 is asserted (Compl. ¶77).
  • Accused Features: The complaint alleges infringement through Guardant's process of tagging cell-free DNA, sequencing it, and bioinformatically sorting and quantifying reads corresponding to single- and double-stranded original molecules (Compl. ¶¶ 79-85).

Multi-Patent Capsule: U.S. Patent No. 10,760,127

  • Patent Identification: U.S. Patent No. 10,760,127, "Methods of Lowering the Error Rate of Massively Parallel DNA Sequencing Using Duplex Consensus Sequencing," issued September 1, 2020 (Compl. ¶12).
  • Technology Synopsis: This patent claims a method of sequencing that involves attaching barcoded adapters to create "partially-complementary, asymmetrical" DNA complexes. A key step of the claimed method is, after amplification and sequencing, "confirming the presence of at least one sequence read derived from each of the original first and second strands." Only after this confirmation are the reads from the two strands compared to a reference sequence to identify a variation, ensuring that the analysis is based on information from the complete original DNA duplex (’127 Patent, Abstract; Compl. ¶96).
  • Asserted Claims: Claim 1 is asserted (Compl. ¶96).
  • Accused Features: Infringement is alleged based on Guardant's process of attaching barcoded adapters, amplifying, sequencing, and then confirming the presence of reads from both original DNA strands before analyzing them for sequence variations (Compl. ¶¶ 98-104).

III. The Accused Instrumentality

  • Product Identification: The accused instrumentalities include Guardant's diagnostic kits and services, which are based on its "Digital Sequencing Technology" (Compl. ¶27). The specific "Accused Products" named are the Guardant360 LDT, Guardant360 CDx, GuardantOMNI, Guardant Reveal, Guardant LUNAR-2, and others. The "Accused Services" include data platforms such as Guardant Connect and Guardant Inform (Compl. ¶¶ 28, 30).
  • Functionality and Market Context: The Accused Products are kits used to collect tissue or blood samples, which are then returned to Guardant's laboratory for genetic sequencing (Compl. ¶28). The core accused functionality is a method described by Guardant's then-Chief Medical Officer as attaching a "digital bar code" to each of the two strands of a double-stranded DNA molecule. After sequencing, if the two strands do not match, Guardant's process will "correct it" (Compl. ¶29). This high-accuracy sequencing is marketed for non-invasive cancer detection and monitoring via liquid biopsy (Compl. ¶¶ 22, 30).

IV. Analysis of Infringement Allegations

'631 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
ligating each of the double-stranded target nucleic acid molecules to at least one adapter molecule... Repairing DNA ends and using "blunt-end ligation" to attach library adapters to fragments of cell-free DNA. ¶40 col. 3:12-14
(a) a degenerate or semi-degenerate single molecule identifier (SMI) sequence that...uniquely labels each ligated double-stranded target nucleic acid molecule... Attaching "library adapters containing inline barcodes" to the ends of cell-free DNA fragments to distinguish individual input molecules. ¶41 col. 3:1-4
(b) a strand-distinguishing nucleotide sequence that...provides a region of non-complementarity between a first strand...and a second strand... Separately encoding each single-stranded half of an original double-stranded DNA molecule with distinct tags, allowing software to compare the two complementary strands. ¶42 col. 38:43-49
amplifying each strand of the adapter-target nucleic acid complex... Performing PCR amplification on the cfDNA fragments after they have been ligated to barcoded adapters. ¶43 col. 4:50-54
separating the first and second strand sequence reads into a set of first strand sequence reads and a set of second strand sequence reads based on the region of non-complementarity... Individually tagging each strand of a double-stranded cfDNA molecule, which allows custom software to compare the two complementary strands. ¶46 col. 4:62-67
generating a high accuracy consensus sequence read...that includes only the nucleotide positions where the compared first and second strand sequence reads are complementary. Using sequencing reads to reconstruct individual cfDNA molecules with "high-fidelity using proprietary double-stranded consensus sequence representation," where errors are reduced by combining barcoding with bioinformatics filtering. ¶51 col. 4:19-24

'699 Patent Infringement Allegations

Claim Element (from Independent Claim 1) Alleged Infringing Functionality Complaint Citation Patent Citation
b) converting the population of circulating DNA molecules into a population of non-uniquely tagged parent polynucleotides...such that each...is substantially unique... Converting circulating cfDNA fragments into digital sequence libraries by ligating "non-unique oligonucleotide heptamer barcodes" and then bioinformatically reconstructing "[i]ndividual unique input molecules." ¶¶ 61, 63 col. 20:1-8
e) grouping the sequence reads into families, each...comprising...the same identifier sequence and having the same start and stop positions... Using "inferred molecular barcodes and read start/stop positions" to "build[] double-stranded consensus representations of original unique cfDNA molecules." ¶¶ 66-68 col. 20:11-17
f) collapsing sequence reads in each family to yield a base call for each family corresponding to one or more genetic loci. Reconstructing individual input molecules using barcode and sequence data to "suppress analytic error modes" and merging reads to create "accurate, molecule-based variant calls." ¶70 col. 20:18-21
  • Identified Points of Contention:
    • Scope Questions: For the ’631 Patent, a central question may be the construction of "uniquely labels." The complaint alleges infringement while simultaneously describing Guardant’s process as using "non-unique" barcodes (Compl. ¶42). The analysis may turn on whether the combination of a non-unique barcode and other information (like fragment ends) meets the "uniquely labels" limitation as defined in the patent. For the ’699 Patent, a similar question arises from the claim language requiring conversion into "non-uniquely tagged parent polynucleotides" that are nevertheless "substantially unique."
    • Technical Questions: A key evidentiary question for the ’631 Patent will be whether Guardant's accused process creates a molecular structure with a "region of non-complementarity" that allows for strand distinction, as claimed. The complaint includes a chart comparing standard NGS to the patented Duplex Sequencing method, visually demonstrating how the technology removes background "noise" to reveal a "true mutation" (Compl. ¶21, p. 6). This visual may be used to argue that Guardant's method achieves the same technical outcome, but the dispute will likely focus on whether it does so by practicing the specific claimed steps.

V. Key Claim Terms for Construction

  • The Term: "uniquely labels" (’631 Patent, Claim 1)

    • Context and Importance: This term is critical because the complaint's description of the accused technology appears to create tension with the claim language. Plaintiffs allege Guardant uses "non-unique oligonucleotide heptamer barcodes" (Compl. ¶42), raising the question of how this can satisfy a "uniquely labels" requirement. The viability of the infringement case for the ’631 patent may depend on whether this term is construed to mean the barcode alone must be unique, or if the barcode in combination with other features (like DNA fragment endpoints) can create the required uniqueness.
    • Intrinsic Evidence for Interpretation:
      • Evidence for a Broader Interpretation: The claim itself states the SMI sequence acts "alone or in combination with the target nucleic acid fragment ends" to uniquely label the molecule (’631 Patent, col. 38:1-4). This language may support an argument that the SMI sequence itself need not be absolutely unique.
      • Evidence for a Narrower Interpretation: The patent repeatedly refers to a "single molecule identifier (SMI)" and the goal of tagging "individual duplex DNA molecules" (’631 Patent, col. 16:11-15). This phrasing may support an argument that the inventive concept is a one-to-one mapping between a tag and a molecule, requiring the tag itself to be the source of uniqueness.

  • The Term: "substantially unique" (’699 Patent, Claim 1)

    • Context and Importance: This term appears alongside the limitation of using "non-uniquely tagged" molecules. Practitioners may focus on this term because its construction will determine the threshold for what constitutes a sufficiently distinct molecular identifier in the context of the claimed method. The complaint alleges Guardant meets this by using "nonrandom" adapters and reconstructing "individual unique input molecules" (Compl. ¶63).
    • Intrinsic Evidence for Interpretation:
      • Evidence for a Broader Interpretation: The term "substantially" is a term of degree, suggesting that absolute uniqueness is not required, so long as molecules can be distinguished for the purpose of forming accurate consensus reads from their progeny.
      • Evidence for a Narrower Interpretation: The patent’s overall goal is to trace sequence reads back to a single original molecule to eliminate errors (’699 Patent, Abstract). A defendant may argue that "substantially unique" requires a level of uniqueness high enough to prevent mis-grouping of reads from different original molecules, thereby setting a high bar for infringement.

VI. Other Allegations

  • Indirect Infringement: The complaint focuses on direct infringement by Guardant for "performing the methods" of the asserted patents and does not plead separate counts for indirect infringement (Compl. ¶¶ 38, 59).
  • Willful Infringement: The complaint alleges willful infringement for all asserted patents (Compl. ¶¶ 54, 73, 92, 107). The basis for willfulness is alleged pre-suit knowledge of the asserted patents and the underlying technology. Specific allegations include that Guardant: (1) engaged in unsuccessful licensing negotiations for the patent family (Compl. ¶31); (2) repeatedly cited the asserted patents and related applications during the prosecution of its own patent portfolio (Compl. ¶32); and (3) was aware of the patents through their citation in third-party challenges to Guardant's own patents in IPR and European opposition proceedings (Compl. ¶¶ 33-34).

VII. Analyst’s Conclusion: Key Questions for the Case

  • A core issue will be one of definitional scope: can the term "uniquely labels" in the ’631 patent be construed to cover a barcoding system that, according to the complaint, uses "non-unique" components which, in combination with other molecular data, may achieve a unique result? A similar question of scope will apply to the interpretation of "substantially unique" in the ’699 patent.
  • A second central question will be one of factual correspondence: does Guardant's bioinformatics pipeline, which allegedly "corrects" mismatches between sequenced strands, perform the specific claimed steps of first separating reads based on a "region of non-complementarity" and then generating a consensus sequence that includes only the positions where the strands are complementary, as required by claim 1 of the ’631 patent?
  • Finally, given the extensive history of alleged pre-suit knowledge detailed in the complaint, a key focus of the litigation will likely be on willfulness: does the evidence of prior licensing discussions, patent prosecution history, and related patent challenges establish that Guardant acted despite an objectively high likelihood that its actions constituted infringement of a valid patent?