1:24-cv-00687
Guardant Health Inc v. Tempus Ai Inc
I. Executive Summary and Procedural Information
- Parties & Counsel:
- Plaintiff: Guardant Health, Inc. (Delaware)
- Defendant: Tempus AI, Inc. (Delaware)
- Plaintiff’s Counsel: Potter Anderson & Corroon LLP
- Case Identification: 1:24-cv-00687, D. Del., 09/11/2026
- Venue Allegations: Venue is alleged to be proper as Defendant Tempus AI, Inc. was incorporated in Delaware, has not contested venue in this action, and has previously consented to venue in the District for the purposes of this action.
- Core Dispute: Plaintiff alleges that Defendant’s liquid biopsy cancer diagnostic tests infringe four U.S. patents related to methods for detecting genetic variants in cell-free DNA.
- Technical Context: The technology involves liquid biopsy, a non-invasive method for cancer detection and monitoring that analyzes fragments of cell-free DNA (cfDNA) circulating in a patient's bloodstream.
- Key Procedural History: The complaint alleges that Defendant had pre-suit knowledge of the patents-in-suit through various channels, including Defendant's monitoring of Plaintiff's prior litigation where the ’306 Patent was asserted, and through Defendant's own patent prosecution activities where its counsel cited the ’992, ’916, and ’306 patents in Information Disclosure Statements submitted to the USPTO. The complaint also references USPTO and PTAB proceedings that allegedly confirmed the novelty and non-obviousness of certain claim limitations in the ’306 and ’992 patents over prior art.
Case Timeline
| Date | Event |
|---|---|
| 2012-09-04 | Priority Date for ’992 and ’916 Patents |
| 2013-12-28 | Priority Date for ’306 Patent |
| 2018-02-27 | ’992 Patent Issued |
| 2018-09-14 | Tempus xF Launched |
| 2019-01-31 | Priority Date for ’693 Patent |
| 2020-09-17 | Tempus prosecution counsel cites ’992 Patent in an IDS |
| 2020-10-06 | ’916 Patent Issued |
| 2021-08-03 | TwinStrand Biosciences, Inc. v. Guardant files complaint |
| 2021-10-19 | ’306 Patent Issued |
| 2022-01-12 | Guardant asserts ’306 Patent in counterclaim against TwinStrand |
| 2022-06-03 | Tempus xF+ Launched |
| 2023-05-09 | ’693 Patent Issued |
| 2023-11-03 | Tempus xM Monitor Launched |
| 2024-01-18 | Tempus xM MRD Launched |
| 2024-05-20 | Tempus S-1 filing mentions monitoring Guardant litigation |
| 2024-06-11 | Guardant files initial complaint against Tempus |
| 2026-09-11 | Second Amended Complaint Filed |
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 11,149,306 - "Methods and systems for detecting genetic variants"
- Issued: October 19, 2021 (the "'306 Patent")
The Invention Explained
- Problem Addressed: The patent's background section describes that prior art methods for analyzing cfDNA were unable to account for molecules that were converted for sequencing but not actually sequenced, an inability which "can dramatically and adversely affect the sensitivity that can be achieved" Compl. ¶56 ’306 Patent, col. 1:59-67
- The Patented Solution: The invention proposes a method of "duplex tagging" where both complementary strands of a double-stranded cfDNA molecule are tagged with molecular barcodes Compl. ¶60 By analyzing the resulting sequence data to identify "paired reads" (where both original strands are detected) and "unpaired reads" (where only one is detected), the method claims to estimate the number of "unseen" molecules and thereby improve the accuracy and sensitivity of detecting genetic variants Compl. ¶¶57-58 ’306 Patent, col. 2:1-18
- Technical Importance: This approach provided a new technique for reducing redundancy and improving error correction in cfDNA sequencing, enhancing the ability to detect rare genetic variants from liquid biopsies Compl. ¶59
Key Claims at a Glance
- The complaint asserts infringement of at least Claim 1 of the ’306 Patent Compl. ¶74 Independent claims 1 and 17 are recited as exemplary Compl. ¶¶60, 65
- The essential elements of independent Claim 1 include:
- Providing a population of cfDNA molecules with complementary strands.
- Tagging the cfDNA molecules with "duplex tags" attached to both ends, using a specific number (n) of different barcode combinations, where n is a function of the expected number of duplicate molecules (z).
- Amplifying the tagged polynucleotides.
- Sequencing a subset of the amplified polynucleotides.
- Reducing or tracking redundancy by using the barcodes to determine distinct cfDNA molecules based on detecting either "paired reads" or "unpaired reads" Compl. ¶60 ’306 Patent, cl. 1
- The complaint notes that dependent claims 2-16 and 29 are also recited, and reserves the right to assert other claims Compl. ¶¶63-64
U.S. Patent No. 9,902,992 - "Systems and methods to detect rare mutations and copy number variation"
- Issued: February 27, 2018 (the "'992 Patent")
The Invention Explained
- Problem Addressed: Detecting and analyzing cfDNA is challenging because it is highly fragmented and present in minute quantities in clinical samples, creating a "need in the art for improved methods and systems" (Compl. ¶83; U.S. Patent No. 10,501,810, col. 1:51-54).
- The Patented Solution: The patent discloses a method for high-efficiency conversion of cfDNA into a sequenceable library. The method involves attaching barcode tags to at least 20% of the cfDNA molecules by ligating adaptors to both ends, wherein the ligation uses a "more than 10x molar excess" of adaptors compared to the cfDNA molecules Compl. ¶86 ’992 Patent, cl. 1 The resulting tagged molecules are amplified, sequenced, and bioinformatically processed to detect genetic aberrations.
- Technical Importance: This high-efficiency conversion method allows for more reliable detection of genetic aberrations from the very small amounts of cfDNA found in bodily fluids Compl. ¶84
Key Claims at a Glance
- The complaint asserts infringement of at least Claim 1 of the ’992 Patent Compl. ¶93
- The essential elements of independent Claim 1 include:
- Providing cfDNA molecules from a bodily sample.
- Attaching tags with barcodes to at least 20% of the cfDNA molecules, using more than 10x molar excess of adaptors.
- Amplifying the tagged polynucleotides.
- Sequencing the amplified progeny polynucleotides.
- Mapping the sequence reads to a human genome.
- Grouping the sequence reads into families based on barcode sequences.
- Collapsing sequence reads in each family to yield a base call.
- Detecting a plurality of genetic aberrations (e.g., single base substitution, CNV, indel, or gene fusion) Compl. ¶85 ’992 Patent, cl. 1
- The complaint references dependent claims 2-33 as improving on prior art Compl. ¶86
U.S. Patent No. 10,793,916 - "Methods and systems for detecting genetic variants"
- Issued: October 6, 2020 (the "'916 Patent") Compl. ¶23
Technology Synopsis
The patent addresses the challenge of detecting genetic variations, particularly microsatellite changes, in highly fragmented and low-concentration cfDNA samples Compl. ¶102 The disclosed solution uses efficient conversion of individual polynucleotides into sequence-ready tagged molecules and high-yield generation of consensus sequences to increase the probability that rare polynucleotides are represented and accurately sequenced (Compl. ¶102; U.S. Patent No. 10,793,916, col. 27:47-59).
Key Claims at a Glance
- Asserted Claims: Independent claims 1 and 13 Compl. ¶¶105-106
- Accused Features: The Accused xF Tests are alleged to use non-unique tagging with a specific number of molecular barcodes to detect microsatellite genetic variation Compl. ¶¶108, 115
U.S. Patent No. 11,643,693 - "Compositions and methods for isolating cell-free DNA"
- Issued: May 9, 2023 (the "'693 Patent") Compl. ¶24
Technology Synopsis
The patent addresses the need for improved methods of isolating cfDNA for liquid biopsy procedures Compl. ¶126 The invention provides a method to isolate cfDNA to capture two sets of target regions—a "sequence-variable target region set" and an "epigenetic target region set"—wherein the capture yield for the sequence-variable set is greater than the epigenetic set, enabling differential sequencing depths for the two types of analyses Compl. ¶127
Key Claims at a Glance
- Asserted Claims: Independent claim 14 Compl. ¶128
- Accused Features: The Accused xM Tests are alleged to use target-specific probes configured to capture cfDNA from sequence-variable and epigenetic regions at different yields, followed by sequencing the sequence-variable regions to a greater depth, to determine the likelihood of cancer Compl. ¶¶129-131, 134
III. The Accused Instrumentality
Product Identification
- The accused instrumentalities are Defendant's liquid biopsy panels, identified as Tempus xF, Tempus xF+, and Tempus xM Monitor (collectively, the "Accused xF Tests"), and Tempus xM MRD (the "Accused xM Tests") Compl. ¶11
Functionality and Market Context
- The Accused Tests are liquid biopsy panels that analyze cfDNA from patient blood specimens to detect genetic mutations and other cancer-related biomarkers Compl. ¶¶28-29, 34 The complaint alleges a specific technical workflow: Tempus obtains cfDNA samples, ligates unique molecular identifiers (UMIs) to the ends of each fragment, amplifies and sequences the ligated DNA, groups the resulting reads into families based on the UMIs, and collapses the families into consensus sequences to determine the presence of genetic variants Compl. ¶¶31-33 The Accused xM Tests are further alleged to analyze these consensus sequences to detect methylation profiles (Compl. ¶34).
No probative visual evidence provided in complaint.
IV. Analysis of Infringement Allegations
'306 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| (a) providing a population of cell-free deoxyribonucleic acid (cfDNA) molecules having first and second complementary strands; | Tempus obtains samples of cell-free DNA from subjects, which are double-stranded. | ¶31 | col. 17:6-7 |
| (b) tagging a plurality of the cfDNA molecules in the population with duplex tags comprising molecular barcodes to produce tagged parent polynucleotides, wherein the duplex tags are attached to both ends of a molecule ... wherein the plurality of the cfDNA molecules are tagged with n different combinations of molecular barcodes, wherein n is at least 2 and no more than 100,000*z...; | Tempus ligates unique molecular identifiers (UMIs), which function as barcodes, to the ends of each cfDNA fragment. The complaint alleges Tempus uses 64 different barcodes, which falls within the claimed range for n. | ¶31; ¶48 | col. 17:8-24 |
| (c) amplifying a plurality of the tagged parent polynucleotides to produce amplified progeny polynucleotides; | Tempus amplifies the ligated cell-free DNA. | ¶32 | col. 17:25-27 |
| (d) sequencing at least a subset of the amplified progeny polynucleotides to produce a set of sequence reads; and | Tempus sequences the amplified cell-free DNA to generate sequence reads. | ¶32 | col. 17:28-30 |
| (e) reducing or tracking redundancy ... to determine distinct cfDNA molecules... based on (i) paired reads ... or (ii) unpaired reads... | Tempus groups sequence reads into families based on barcodes and collapses them into consensus sequences. The complaint alleges this process determines paired and unpaired molecules. | ¶32; ¶64 | col. 18:1-2 |
- Identified Points of Contention:
- Scope Question: A potential point of contention is whether Tempus's use of 64 different barcodes satisfies the limitation "tagged with n different combinations of molecular barcodes, wherein n is at least 2 and no more than 100,000*z," where z is the "mean of an expected number of duplicate molecules" Compl. ¶60 The infringement analysis may turn on the factual determination of the value of z in Tempus's process and whether 64 falls within the claimed range.
- Technical Question: The complaint alleges that Tempus's process of grouping reads into families and creating consensus sequences meets the claim limitation of determining distinct molecules based on "paired reads" and "unpaired reads" (Compl. ¶64). A key technical question will be whether the bioinformatic process used by Tempus performs the specific functions of identifying and counting paired versus unpaired strands as described in the patent, or if there is a functional difference.
'992 Patent Infringement Allegations
| Claim Element (from Independent Claim 1) | Alleged Infringing Functionality | Complaint Citation | Patent Citation |
|---|---|---|---|
| b) attaching tags comprising barcodes... to tag at least 20% of the cfDNA molecules, which attaching comprises ligating adaptors comprising the barcodes to both ends of the cfDNA molecules, wherein ligating comprises using more than 10x molar excess of the adaptors as compared to the cfDNA molecules... | Tempus is alleged to ligate UMIs to cfDNA fragments using a quantity of adapters "more than ten times the number of moles of cell-free DNA" and with a ligation efficiency of "20 percent or more." | ¶31; ¶47 | col. 24:1-12 |
| f) grouping the sequence reads mapped in e) into families based at least on barcode sequences of the sequence reads... | Tempus groups its sequence reads into families based on the UMIs (barcodes) and the alignment of the reads to a reference sequence. | ¶32 | col. 24:49-54 |
| g) ...collapsing sequence reads in each family to yield a base call for each family at the genetic locus; and | Tempus collapses the families of reads into consensus sequences, which represent the sequence of the original cfDNA fragment. | ¶32 | col. 24:55-58 |
| h) detecting, at one or more genetic loci, a plurality of genetic aberrations... | Tempus uses the consensus sequences to determine the presence of genetic variants such as CNVs, SNVs, and fusions. | ¶33 | col. 24:59-67 |
- Identified Points of Contention:
- Evidentiary Question: The infringement allegation hinges on two specific quantitative thresholds: using "more than 10x molar excess of the adaptors" and achieving a tagging efficiency of "at least 20%" (Compl. ¶¶31, 86). A central dispute will likely be evidentiary, focused on whether Plaintiff can prove that Defendant's laboratory processes actually meet these precise numerical limitations.
- Scope Question: The claim requires "detecting... a plurality of genetic aberrations" comprising "two or more different members" from the group of SNV, CNV, indel, and gene fusion Compl. ¶85 The analysis may raise the question of whether every performance of the Accused Tests necessarily detects two or more different types of aberrations as required, or if some tests might only detect, for example, multiple SNVs without detecting any other aberration type.
V. Key Claim Terms for Construction
- Term ('306 Patent):
"paired reads" / "unpaired reads" - Context and Importance: These terms are central to Claim 1's method of "reducing or tracking redundancy" and determining distinct cfDNA molecules Compl. ¶60 The definition of what constitutes a "paired read" (detection of both complementary strands) versus an "unpaired read" (detection of only one) is critical to establishing infringement, as it defines the core logic of the claimed invention.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification describes the concept generally as tracking redundancy by "tagging and counting both halves of double-stranded DNA" (’306 Patent, col. 2:1-2). This could be argued to encompass any bioinformatic method that distinguishes between molecules where both strands were sequenced versus only one, regardless of the specific algorithm.
- Evidence for a Narrower Interpretation: The detailed description provides a specific example of estimating unseen molecules based on the number of "pairs (i.e., molecules where both strands were identified) and singlets (i.e., molecules where only one strand was identified)" (’306 Patent, col. 2:6-10). A party might argue this ties the terms to a specific statistical estimation method, potentially narrowing their scope to exclude processes that do not perform such an estimation.
- Term ('992 Patent):
"more than 10x molar excess of the adaptors as compared to the cfDNA molecules" - Context and Importance: This quantitative limitation in Claim 1 is a cornerstone of the infringement allegation and was a key point of distinction over prior art during prosecution, according to the complaint (Compl. ¶87). Practitioners may focus on this term because its construction will dictate the type of evidence required to prove or disprove infringement of this crucial step.
- Intrinsic Evidence for Interpretation:
- Evidence for a Broader Interpretation: The specification of a related patent (U.S. Patent No. 10,501,810) discloses a wide range of high molar ratios, such as "10:1, 25:1, 50:1... 1000:1 or more" (U.S. Patent No. 10,501,810, col. 39:5-13). This language could support an interpretation that any ratio exceeding 10:1, as measured by standard laboratory techniques, meets the limitation.
- Evidence for a Narrower Interpretation: The specific examples in the patent's detailed description may describe particular concentrations and reaction conditions used to achieve the claimed molar excess. A party could argue that the term should be limited to the context of those specific embodiments, potentially requiring a showing that the accused process uses not just the ratio, but similar underlying conditions.
VI. Other Allegations
- Indirect Infringement: The complaint focuses on allegations of direct infringement by Tempus, stating that Tempus performs the accused methods at its facilities in the United States (Compl. ¶¶31, 34). No specific facts are alleged to support a standalone claim for indirect infringement.
- Willful Infringement: The complaint alleges that Tempus's infringement has been and continues to be willful, based on pre-suit knowledge of the patents-in-suit Compl. ¶¶77, 96 The allegations of knowledge are supported by:
- Tempus's public S-1 filing, in which it stated it "monitored and continue[s] to monitor" patent litigation involving Guardant Compl. ¶12 Compl. ¶42
- Tempus's alleged awareness of litigation between Guardant and TwinStrand, where the ’306 Patent was asserted Compl. ¶¶36-37
- Information Disclosure Statements filed by Tempus's own patent prosecution counsel that cited the ’992, ’916, and ’306 patents between 2020 and 2022 Compl. ¶¶38-40
- Allegations of Tempus deliberately copying Guardant's commercially successful Guardant360, Guardant Response, and Guardant Reveal products Compl. ¶¶46-51
VII. Analyst’s Conclusion: Key Questions for the Case
- A core issue will be one of evidentiary proof for quantitative limitations: can Guardant produce sufficient evidence to demonstrate that Tempus's internal laboratory processes meet the precise numerical thresholds recited in the claims, such as the "more than 10x molar excess" of adaptors and "at least 20%" tagging efficiency required by the ’992 patent?
- A key legal question will be one of claim scope and functional equivalence: does Tempus's bioinformatic process of grouping reads into families and collapsing them into consensus sequences perform the specific function of determining distinct cfDNA molecules based on "paired reads" and "unpaired reads" as required by the ’306 patent, or is there a fundamental mismatch in technical operation that places the accused method outside the claim's scope?
- A central theme of the case will likely be willfulness and intent: given the complaint's detailed allegations of pre-suit knowledge through monitoring litigation and citing the patents-in-suit during its own patent prosecution, a critical question for the court will be whether Tempus's conduct constituted an "unjustifiably high risk of infringement" that was consciously disregarded, potentially exposing it to enhanced damages.