Guardant Health Inc v. Foundation Medicine Inc

I. Executive Summary and Procedural Information

• Parties & Counsel:
  • Plaintiff: Guardant Health, Inc. (Delaware)
  • Defendant: Foundation Medicine, Inc. (Delaware)
  • Plaintiff’s Counsel: Farnan LLP; Weil, Gotshal & Manges LLP
• Case Identification: 1:17-cv-01616, D. Del., 11/09/2017
• Venue Allegations: Venue is asserted on the basis that Defendant is a Delaware corporation and transacts business in the district, including selling the accused product.
• Core Dispute: Plaintiff alleges that Defendant’s FoundationACT liquid biopsy test infringes a patent related to methods for detecting rare genetic mutations in cell-free DNA by using molecular barcodes to reduce sequencing errors.
• Technical Context: The lawsuit concerns the field of non-invasive cancer diagnostics, where "liquid biopsies" analyze circulating tumor DNA (ctDNA) in a patient's blood to identify cancer-related genetic mutations without requiring a tissue biopsy.
• Key Procedural History: The complaint references a scientific poster presented by Defendant in February 2017, which Plaintiff uses as evidence describing the allegedly infringing methodology.

Case Timeline

Date	Event
2012-09-04	’731 Patent Priority Date
2016-Mid	FoundationACT Test Commercial Launch
2017-02-01	Foundation Medicine Poster Presentation (approximate date)
2017-03-21	’731 Patent Issue Date
2017-11-09	Complaint Filing Date

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

U.S. Patent No. 9,598,731 - "Systems and Methods to Detect Rare Mutations and Copy Number Variation"

• Patent Identification: U.S. Patent No. 9,598,731, “Systems and Methods to Detect Rare Mutations and Copy Number Variation,” issued March 21, 2017.

The Invention Explained

• Problem Addressed: The patent describes the challenge of detecting rare genetic alterations, such as those from cancer, in cell-free DNA (cfDNA) samples. A key problem is that the low frequency of these true mutations can be masked by errors introduced during the DNA amplification and sequencing processes, making it difficult to distinguish signal from noise (’731 Patent, col. 31:40-54).
• The Patented Solution: The invention proposes a method to improve the accuracy of sequencing by attaching unique or non-unique molecular barcodes to individual DNA fragments before they are amplified. After sequencing, the resulting reads are grouped into "families" based on their shared barcode and other sequence characteristics. By comparing the reads within each family, a "consensus sequence" is generated for the original DNA fragment, which effectively filters out random PCR and sequencing errors (’731 Patent, Abstract; col. 5:2-6). This error-correction method is designed to enable the high-sensitivity detection of very low-frequency mutations (’731 Patent, col. 19:48-55).
• Technical Importance: This approach of creating consensus sequences from barcoded DNA families addresses a fundamental signal-to-noise problem in liquid biopsy, potentially allowing for earlier and more reliable cancer detection from blood samples than was previously feasible (’731 Patent, col. 1:24-34).

Key Claims at a Glance

• The complaint asserts infringement of at least claims 1-3, 6-9, 12, and 16-17 (Compl. ¶21). Independent claim 1 is presented as representative (Compl. ¶16).
• Independent Claim 1: This method claim recites the essential steps for quantifying single nucleotide variant (SNV) tumor markers:
  • (a) providing at least 10 ng of cell-free DNA;
  • (b) attaching tags with barcodes to the cfDNA to create "non-uniquely tagged parent polynucleotides";
  • (c) amplifying these tagged parent polynucleotides;
  • (d) sequencing the amplified products to get reads containing a barcode and a cfDNA-derived sequence;
  • (e) grouping the sequence reads into "families" based on the barcode and sequence information (e.g., start/end sequences);
  • (f) comparing reads within each family to determine "consensus sequences";
  • (g) providing reference sequences from a human genome with known tumor markers;
  • (h) identifying consensus sequences that map to the tumor marker loci; and
  • (i) calculating the number of consensus sequences with the SNV to quantify the tumor markers.

III. The Accused Instrumentality

Product Identification

• The accused instrumentality is the FoundationACT® liquid biopsy test, a service performed and sold by Defendant Foundation Medicine, Inc. (Compl. ¶6, ¶14).

Functionality and Market Context

• The FoundationACT test is described as an assay that analyzes ctDNA from a patient's blood to identify "clinically relevant genomic alterations" for cancer patients when a tissue biopsy is not feasible (Compl. ¶14).
• The complaint presents a figure from a scientific poster (Exhibit 3) detailing the alleged methodology of the FoundationACT test. This figure outlines a four-part workflow: (A) Tumor Sample (Blood), (B) Sequencing Library Preparation, (C) Analysis Pipeline, and (D) Clinical Report (Compl. ¶15). The "Sequencing Library Preparation" step explicitly notes the use of "Molecular and sample barcodes" and "Hybridization capture," followed by sequencing on an Illumina platform (Compl. ¶15, Exhibit 3). The "Analysis methods" section describes "Error correction to <0.05%" and "Variant calling" down to very low variant allele frequencies (VAF) (Compl. ¶15, Exhibit 3).
• The complaint alleges this test provides oncologists with comprehensive molecular information to match patients with targeted therapies and clinical trials (Compl. ¶14).

IV. Analysis of Infringement Allegations

U.S. Patent No. 9,598,731 Infringement Allegations

Claim Element (from Independent Claim 1)	Alleged Infringing Functionality	Complaint Citation	Patent Citation
(a) providing at least 10 ng of cell-free DNA obtained from a bodily sample of the subject;	The FoundationACT test obtains more than 10 ng of cfDNA from a patient blood draw, with the referenced poster specifying "≥50 ng extracted ctDNA".	¶17; ¶15, Ex. 3	col. 61:12-14
(b) attaching tags comprising barcodes...to generate non-uniquely tagged parent polynucleotides...	Tags with barcodes are attached to the cfDNA fragments; the poster explicitly lists "Molecular and sample barcodes" as part of the library preparation.	¶17; ¶15, Ex. 3	col. 61:15-21
(c) amplifying the non-uniquely tagged parent polynucleotides...	The tagged DNA sample undergoes PCR amplification.	¶17	col. 61:22-24
(d) sequencing the amplified non-uniquely tagged progeny polynucleotides...	The amplified DNA is sequenced on an Illumina platform, producing reads that include a barcode and a sequence from the cfDNA.	¶17	col. 61:25-29
(e) grouping the plurality of sequence reads...into families based on i) the barcode sequence and ii) at least one of: sequence information at a beginning...end...and length...	Sequence reads are grouped into families using the barcode and other sequence information, which allows for collection of information from the same original DNA molecule.	¶17	col. 61:30-41
(f) comparing the sequence reads grouped within each family to each other to determine consensus sequences for each family...	Reads in each family are compared to generate a "consensus sequence" that provides a more accurate determination of the original molecule's sequence.	¶17	col. 61:42-48
(h) identifying consensus sequences that map to a given locus...	The generated consensus sequences are mapped to a reference genome to identify sequences corresponding to regions associated with cancer tumor markers.	¶17	col. 61:54-56
(i) calculating a number of consensus sequences that map to the given locus that include the single nucleotide variant thereby quantifying...tumor markers...	The number of tumor markers present in the original sample are quantified.	¶17	col. 61:57-62

• Identified Points of Contention:
  • Scope Questions: A central dispute may concern whether Defendant's "Error correction" and "Variant calling" processes, as described in its poster, fall within the scope of the patent's specific method of "grouping... into families" (limitation 1e) and "determin[ing] consensus sequences" (limitation 1f). The complaint makes a narrative allegation but does not map specific features from the poster to these claim elements.
  • Technical Questions: What evidence will demonstrate that the "Molecular and sample barcodes" used by Defendant function to enable the family-based error correction described in the patent, as opposed to merely serving as sample identifiers for multiplexing? The infringement case will depend on showing that Defendant’s process technically aligns with the patent's specific error-reduction mechanism, not just that it uses barcodes in a general sense.

V. Key Claim Terms for Construction

• The Term: "families"

• Context and Importance: This term is the lynchpin of the invention's error-correction methodology. Claim 1(e) defines a "family" as a group of sequence reads sharing a barcode and at least one other characteristic (start/end sequence or length). Proving infringement will require showing that Defendant's bioinformatics pipeline groups reads in this specific manner. Practitioners may focus on this term because the defendant will likely argue its "Error correction" algorithm does not create "families" as defined in the patent, but rather uses a different statistical method.

• Intrinsic Evidence for Interpretation:

  • Evidence for a Broader Interpretation: The specification describes the general concept as grouping reads that "are descendants of the same original parent molecule" which could be argued to encompass any method that achieves this result, regardless of the specific grouping criteria (’731 Patent, col. 44:4-6).
  • Evidence for a Narrower Interpretation: The claim itself provides a specific, multi-part definition: grouping must be based on "(i) the barcode sequence and (ii) at least one of: sequence information at a beginning..., sequence information at an end..., and length of the sequence read" (’731 Patent, col. 61:30-41). This explicit definition could be used to argue for a narrow construction that requires proof of these specific grouping criteria.

• The Term: "consensus sequences"

• Context and Importance: This term is directly tied to "families" and describes the outcome of the error-correction step. The dispute will likely center on whether Defendant's process actually "determine[s]" a "consensus sequence" for each family, or if its "Error correction to <0.05%" (Compl. ¶15, Ex. 3) represents a distinct, non-infringing technique.

• Intrinsic Evidence for Interpretation:

  • Evidence for a Broader Interpretation: The specification states that a consensus sequence "represents a more accurate determination of the sequence of the molecule in question," which could support a reading that covers any process that improves sequence accuracy by comparing related reads (’731 Patent, col. 5:4-6).
  • Evidence for a Narrower Interpretation: The claim requires "comparing the sequence reads grouped within each family to each other to determine consensus sequences" (’731 Patent, col. 61:42-45). This could be interpreted to require a direct, intra-family comparison step, which Defendant might argue its global error correction model does not perform. Figure 9 depicts a distinct step of collapsing reads into a consensus sequence, which could reinforce a narrower, step-wise interpretation (’731 Patent, Fig. 9, step 810).

VI. Other Allegations

The complaint does not contain specific factual allegations to support indirect or willful infringement. The prayer for relief includes a request for a determination that the case is "exceptional" under 35 U.S.C. § 285, which could lead to an award of attorneys' fees, but the basis for this request is not detailed in the factual allegations (Compl. p. 8, ¶D).

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

The resolution of this case will likely depend on the answers to two central questions:

1. A core issue will be one of definitional scope and technical equivalence: Does Foundation Medicine’s "Error correction" algorithm, as practiced in the FoundationACT test, perform the specific steps of "grouping" reads into "families" and "determin[ing] consensus sequences" as those terms are defined in Claim 1 of the ’731 patent, or does it employ a technically distinct method to achieve a similar outcome?
2. A key evidentiary question will be what proof Plaintiff can obtain in discovery to demonstrate the inner workings of the FoundationACT bioinformatics pipeline. The case will turn on whether the evidence shows a functional match to the patent's claimed method for error reduction, or a fundamental mismatch in technical operation.