DCT

1:24-cv-00687

Guardant Health Inc v. Tempus Ai Inc

Log In

I. Executive Summary and Procedural Information

  • Parties & Counsel:
  • Case Identification: 1:24-cv-00687, D. Del., 11/04/2024
  • Venue Allegations: Venue is alleged to be proper in the District of Delaware because Defendant is a Delaware corporation.
  • Core Dispute: Plaintiff alleges that Defendant’s liquid biopsy panels, used for cancer diagnostics, infringe five U.S. patents related to methods for detecting rare genetic variants and isolating cell-free DNA.
  • Technical Context: The technology at issue is liquid biopsy, which analyzes cell-free DNA (cfDNA) from a patient's blood to detect and monitor cancer, offering a non-invasive alternative to traditional tissue biopsies.
  • Key Procedural History: The complaint references prior patentability challenges. For U.S. Patent No. 9,902,992, the complaint notes that the USPTO repeatedly found the core innovation nonobvious over prior art during inter partes review proceedings. For a limitation similar to one in U.S. Patent No. 11,149,306, the complaint cites a Federal Circuit decision that vacated a finding of obviousness against a related Guardant patent.

Case Timeline

Date Event
2012-09-04 Priority Date for ’992, ’810, and ’916 Patents
2013-12-28 Priority Date for ’306 Patent
2018-02-27 Issue Date for U.S. Patent No. 9,902,992
2018-09-14 Tempus announces release of Tempus xF liquid biopsy test
2019-01-31 Priority Date for ’693 Patent
2019-12-10 Issue Date for U.S. Patent No. 10,501,810
2020-10-06 Issue Date for U.S. Patent No. 10,793,916
2021-10-19 Issue Date for U.S. Patent No. 11,149,306
2022-06-03 Tempus announces launch of Tempus xF+ test
2023-05-09 Issue Date for U.S. Patent No. 11,643,693
2023-11-03 Tempus announces launch of Tempus xM Monitor test
2024-01-18 Tempus announces launch of Tempus xM MRD test
2024-11-04 Complaint Filing Date

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

U.S. Patent No. 11,149,306 - "Methods and systems for detecting genetic variants"

The Invention Explained

  • Problem Addressed: The patent's background section, as described in the complaint, notes that prior art methods for detecting cfDNA could not infer the counts of molecules that were converted for sequencing but were not ultimately sequenced, an inability that "can dramatically and adversely affect the sensitivity that can be achieved" (Compl. ¶38; ’306 Patent, col. 1:59-67).
  • The Patented Solution: The invention addresses this deficiency by describing a technique for tagging both strands of double-stranded DNA and then estimating the number of "unseen molecules" by analyzing the number of "pairs" (where both complementary strands were sequenced) and "singlets" (where only one strand was sequenced) detected in a specific genomic region (’306 Patent, col. 2:1-18; Compl. ¶39-40). This method uses a specific number of molecular barcodes to enable the tracking and counting process (Compl. ¶39).
  • Technical Importance: This approach provides a new and unconventional way of sequencing and analyzing DNA samples that aims to improve sensitivity for detecting rare genetic mutations (Compl. ¶40).

Key Claims at a Glance

  • The complaint asserts infringement of at least Claim 1 and recites Claim 17 (Compl. ¶42, ¶47, ¶56).
  • Independent Claim 1 of the ’306 Patent includes the following essential elements:
    • Providing a population of cell-free DNA (cfDNA) molecules with first and second complementary strands.
    • Tagging the cfDNA with duplex tags comprising molecular barcodes attached to both ends of the molecules, using between 2 and 100,000*z different barcode combinations, where z is a mean expected number of duplicate molecules.
    • Amplifying the tagged polynucleotides.
    • Sequencing at least a subset of the amplified polynucleotides to produce sequence reads.
    • Reducing or tracking redundancy by using the molecular barcodes to determine distinct cfDNA molecules based on (i) "paired reads" (from a first tagged strand and a second tagged complementary strand) or (ii) "unpaired reads" (from a first tagged strand with no corresponding second strand).
  • The complaint notes that the asserted claims include dependent claims 2-16 and 29 (Compl. ¶43, ¶45).

U.S. Patent No. 9,902,992 - "Systems and methods to detect rare mutations and copy number variation"

The Invention Explained

  • Problem Addressed: The patent addresses the challenge that detecting and analyzing cfDNA was known to be difficult because it is highly fragmented and present in only minute quantities in clinical samples, creating a "need in the art for improved methods" (’992 Patent, col. 1:55-57; Compl. ¶64).
  • The Patented Solution: The invention claims a method that creates a high-efficiency conversion of cfDNA into non-uniquely tagged polynucleotides (’992 Patent, Abstract; Compl. ¶65). This is achieved by specific process parameters: attaching tags with barcodes to at least 20% of the cfDNA molecules by ligating adaptors to both ends of the molecules, where the ligation uses more than a 10x molar excess of adaptors compared to the cfDNA molecules (’992 Patent, cl. 1; Compl. ¶67).
  • Technical Importance: This method provided a novel technological solution to the problem of efficiently preparing minute quantities of fragmented cfDNA for genetic analysis (Compl. ¶64).

Key Claims at a Glance

  • The complaint asserts infringement of at least Claim 1 (Compl. ¶72, ¶74).
  • Independent Claim 1 of the ’992 Patent includes the following essential elements:
    • Providing cfDNA molecules from a subject.
    • Attaching tags with barcodes to at least 20% of the cfDNA molecules by ligating adaptors to both ends, using more than 10x molar excess of the adaptors.
    • Amplifying the tagged polynucleotides.
    • Sequencing the amplified polynucleotides to produce sequence reads comprising a barcode and a sequence derived from cfDNA.
    • Mapping the sequence reads to a reference genome.
    • Grouping the mapped reads into families based on their barcode sequences.
    • Collapsing sequence reads in each family to yield a base call for that family at a genetic locus.
    • Detecting genetic aberrations from the collapsed base calls.
  • The complaint notes that the asserted claims include dependent claims 2-33 (Compl. ¶67).

U.S. Patent No. 10,501,810 - "Systems and methods to detect rare mutations and copy number variation"

  • Patent Identification: U.S. Patent No. 10,501,810, "Systems and methods to detect rare mutations and copy number variation," issued December 10, 2019.
  • Technology Synopsis: The ’810 Patent is directed to methods for detecting genetic variation in cfDNA with high sensitivity by using a specific, limited number of molecular barcodes (5 to 100) that are non-unique to the much larger number of cfDNA molecules in a sample (Compl. ¶82, ¶86). The invention improves on prior art by, among other things, obtaining a specific amount of cfDNA (10 to 100 ng) and ligating adapters with a specific set of molecular barcodes of a particular length (5-20 nucleotides) (Compl. ¶84, ¶85).
  • Asserted Claims: At least Claim 1 (Compl. ¶91, ¶93).
  • Accused Features: The complaint alleges infringement by the Accused Tests, which are described as using barcodes (UMIs) to group sequence reads into families based on both the barcode and the alignment start/end positions, which are then collapsed into consensus sequences (Compl. ¶32). This process is alleged to practice the elements of Claim 1 of the ’810 Patent (Compl. ¶91).

U.S. Patent No. 10,793,916 - "Methods and systems for detecting genetic variants"

  • Patent Identification: U.S. Patent No. 10,793,916, "Methods and systems for detecting genetic variants," issued October 6, 2020.
  • Technology Synopsis: The ’916 Patent, which shares a common specification with the ’810 Patent, is directed to techniques for detecting polymorphic forms, particularly microsatellite changes, in cfDNA from a cancer patient (Compl. ¶98). The invention improves on prior art by using non-unique molecular barcodes (up to 1,000,000) to tag parent cfDNA molecules and infer quantitative measures of microsatellite changes (Compl. ¶106, ¶108).
  • Asserted Claims: At least Claim 13 (Compl. ¶113, ¶115).
  • Accused Features: The infringement allegation is directed at the Accused xF Tests (Compl. ¶113). The complaint states that the Tempus xF and xF+ tests report on Microsatellite Instability High (MSI-H) status (Compl. ¶28, ¶29).

U.S. Patent No. 11,643,693 - "Compositions and methods for isolating cell-free DNA"

  • Patent Identification: U.S. Patent No. 11,643,693, "Compositions and methods for isolating cell-free DNA," issued May 9, 2023.
  • Technology Synopsis: The ’693 Patent addresses the challenge of isolating useful cfDNA fractions for liquid biopsy procedures (Compl. ¶123). The invention is an improved method to "isolate cell-free DNA so as to capture two sets of target regions—a sequence-variable target region set and an epigenetic target region set—wherein the capture yield of the sequence-variable target region set is greater than the capture yield of the epigenetic target region set" (’693 Patent, col. 1:49-54; Compl. ¶124). This differential capture allows the sequence-variable regions, which may contain mutations, to be sequenced to a greater depth than the epigenetic regions (Compl. ¶129).
  • Asserted Claims: At least Claim 14 (Compl. ¶132, ¶134).
  • Accused Features: The infringement allegation is directed at the Accused xM Tests (Compl. ¶132). The complaint states that the Tempus xM MRD test "delivers a binary MRD assessment based on both methylation and genomic variant MRD classifiers" and analyzes consensus sequences to detect methylation profiles (Compl. ¶34).

III. The Accused Instrumentality

Product Identification

The accused instrumentalities are liquid biopsy tests performed by Defendant Tempus, identified as the Tempus xF, Tempus xF+, and Tempus xM Monitor (collectively, the "Accused xF Tests"), and the Tempus xM MRD (the "Accused xM Tests") (Compl. ¶11).

Functionality and Market Context

The complaint alleges the Accused Tests function by obtaining cfDNA from subjects and ligating adapters containing unique molecular identifiers (UMIs), or barcodes, to the cfDNA fragments (Compl. ¶31). These tagged fragments are then amplified and sequenced to generate sequence reads, which are mapped to a reference sequence (Compl. ¶32). Reads are grouped into "families" based on their barcodes and genomic alignment, and these families are then "collapsed into consensus sequences" (Compl. ¶32). These consensus sequences are subsequently analyzed to determine the presence of genetic variants, such as single nucleotide variations (SNVs), copy number variations (CNVs), and microsatellite instabilities (MSIs) (Compl. ¶33). The Accused xM Tests are further alleged to analyze these sequences to detect methylation profiles (Compl. ¶34).

The complaint characterizes the accused products as "copy-cat" tests that have achieved market success through the unauthorized use of Plaintiff's patented technology (Compl. ¶10, ¶14).

No probative visual evidence provided in complaint.

IV. Analysis of Infringement Allegations

The complaint references preliminary claim chart exhibits for each patent-in-suit but does not attach them (Compl. ¶56, ¶74, ¶93, ¶115, ¶134). The following summarizes the narrative infringement theory presented in the complaint.

’306 Patent Infringement Allegations

  • Narrative Theory: The complaint alleges that Tempus's Accused Tests practice the method of at least Claim 1 (Compl. ¶54, ¶56). The core of the theory is that Tempus's process of generating consensus sequences from families of reads inherently involves "reducing or tracking redundancy" and determining distinct cfDNA molecules (Compl. ¶32-33, ¶42). This process necessarily relies on identifying molecules where both DNA strands were successfully sequenced ("paired reads") and distinguishing them from instances where only one strand was sequenced ("unpaired reads"), thereby meeting the central limitation of the claim (Compl. ¶46, ¶47).
  • Identified Points of Contention:
    • Scope Questions: A central question for claim construction may be whether Tempus’s process of generating consensus sequences from barcode-defined families meets the claim limitation of "determining distinct cfDNA molecules...based on...unpaired reads." The dispute may turn on whether this requires a specific analytical step of counting "singlets" to estimate unseen molecules, as described in the patent's specification, or if it can be read more broadly to cover any process that computationally distinguishes reads originating from single strands versus both strands of an original DNA duplex.
    • Technical Questions: What evidence does the complaint provide that the accused process for generating a consensus sequence performs the specific function of accounting for "unpaired reads" (i.e., a tagged strand with no sequenced complementary strand), as required by Claim 1, as opposed to a more general process of error correction or read filtering?

’992 Patent Infringement Allegations

  • Narrative Theory: The complaint alleges that Tempus's Accused Tests practice the method of at least Claim 1 (Compl. ¶72, ¶74). The theory relies on allegations, made "on information and belief," that Tempus's process meets the key quantitative limitations of the claim: ligating adapters using "more than a 10x molar excess" and achieving a tagging efficiency of "at least 20% of the cfDNA molecules" (Compl. ¶31, ¶66). The complaint further alleges that Tempus's described workflow of amplifying, sequencing, mapping, grouping reads into families by barcode, collapsing, and detecting aberrations meets the remaining method steps of the claim (Compl. ¶32-33, ¶66).
  • Identified Points of Contention:
    • Evidentiary Questions: The infringement allegation hinges on specific quantitative parameters ("more than 10x molar excess" and "at least 20%"). As these facts are pleaded on "information and belief," a key point of contention will be what evidence emerges during discovery to substantiate these claims about the precise operating parameters of Tempus's laboratory processes.
    • Scope Questions: Does the claim term "grouping the sequence reads...into families based at least on barcode sequences" read on the accused process, which is described as grouping by barcodes and by the alignment of the sequence reads to a reference sequence (Compl. ¶32, ¶66)? The analysis may question whether the additional basis for grouping (alignment) places the accused method outside the literal scope of the claim language.

V. Key Claim Terms for Construction

For the ’306 Patent

  • The Term: "unpaired reads corresponding to sequence reads generated from a first tagged strand having no second tagged complementary strand" (from Claim 1).
  • Context and Importance: This term is critical because it defines a key element of the patented solution for improving sensitivity. Practitioners may focus on this term because the infringement analysis will depend on whether Tempus's method of generating consensus sequences is found to perform a determination based on these specific types of "unpaired reads," or if it merely filters out reads that lack a partner without performing the inventive function.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The patent specification describes the overall goal as providing "a new and unconventional way of sequencing and analyzing DNA samples not found in the prior art" and estimating "unseen molecules" (’306 Patent, col. 2:1-18; Compl. ¶40). This could support a functional interpretation covering any process that accounts for single-strand-derived data to improve accuracy.
    • Evidence for a Narrower Interpretation: The claim language is highly specific, requiring a determination "based on... unpaired reads." The detailed description explains this concept in the context of counting "singlets (i.e., molecules where only one strand was identified)" to "estimat[e] the number of unseen molecules" (’306 Patent, col. 2:1-18). This may support a narrower construction requiring a specific calculation or estimation step, not just a general data processing step.

For the ’992 Patent

  • The Term: "attaching tags...to tag at least 20% of the cfDNA molecules" (from Claim 1).
  • Context and Importance: This quantitative threshold is a cornerstone of the patent's asserted improvement over the prior art and is central to the infringement allegation. Practitioners may focus on this term because the dispute will likely involve extensive discovery into Tempus's lab procedures to determine if this efficiency is met, and the definition of what constitutes a "tagged" molecule will be dispositive.
  • Intrinsic Evidence for Interpretation:
    • Evidence for a Broader Interpretation: The patent discusses solving the general problem of inefficient conversion of cfDNA into sequenceable libraries (’992 Patent, col. 1:55-57). This could support a reading where any cfDNA molecule successfully ligated to an adapter, making it potentially sequenceable, counts toward the 20% threshold.
    • Evidence for a Narrower Interpretation: The claim element requires "ligating adaptors...to both ends of the cfDNA molecules." This language may support a narrower construction requiring that for a cfDNA molecule to be counted as "tagged" for the 20% efficiency calculation, adapters must be successfully ligated to both of its ends.

VI. Other Allegations

Willful Infringement

The complaint does not include a formal count for willful infringement. However, it alleges facts that may support such a claim later in the litigation. The complaint cites Tempus's S-1 filing, which states Tempus has "monitored and continue[s] to monitor" patent litigation involving Guardant, as evidence of pre-suit knowledge of Guardant's intellectual property (Compl. ¶12, ¶13). The prayer for relief seeks a finding that this is an "exceptional case" under 35 U.S.C. § 285, which is the statutory basis for awarding attorney's fees, often in cases of willful or egregious infringement (Compl. p. 38, ¶g).

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

This case appears poised to turn on three central, open questions for the court:

  • An evidentiary question of process specifics: Can Plaintiff Guardant, through discovery, produce evidence to confirm its "information and belief" allegations that Defendant Tempus's laboratory processes meet the specific quantitative thresholds recited in the ’992 and ’810 Patents, such as the ">10x molar excess of adaptors" and "at least 20%" tagging efficiency?
  • A question of definitional scope: Can the '306 Patent’s claim to determine distinct molecules "based on...unpaired reads" be construed to cover a general bioinformatics process of generating consensus sequences, or does it require a more specific analytical step of counting "singlets" to estimate unsequenced molecules as described in the patent’s specification?
  • A question of technical operation: Does the combined genomic and methylation analysis in Tempus’s xM MRD test practice the '693 Patent's method of capturing two distinct target sets (sequence-variable and epigenetic) with different relative yields for the purpose of enabling differential sequencing depths, or is there a fundamental mismatch in the underlying technical method of sample preparation and analysis?