Obviousness

Obviousness Analysis of US Patent 10793916 under 35 U.S.C. § 103

This analysis addresses the obviousness of US Patent 10793916, "Systems and methods to detect rare mutations and copy number variation," considering prior art available before its priority date of September 4, 2012. A Person Having Ordinary Skill in the Art (PHOSITA) in this field would likely possess a strong background in molecular biology, genomics, next-generation sequencing (NGS) technologies, and bioinformatics.

The previous PTAB analysis for IPR2025-01435 resulted in a discretionary denial of institution, which means the merits of the obviousness arguments were not fully adjudicated. Therefore, this analysis provides an independent assessment based on publicly available prior art and the detailed definitions within the patent.

Core Inventive Concepts of US10793916

US10793916 generally describes methods and systems for:

1. Detecting Copy Number Variation (CNV): Involving sequencing extracellular polynucleotides (cfDNA) from bodily samples, filtering, mapping, quantifying reads in predefined regions, normalizing, and comparing to controls to determine CNV.
2. Detecting Rare Mutations: Similar to CNV detection, but focusing on identifying sequence variants, calculating variant ratios, normalizing, and determining rare variants. This often incorporates molecular barcoding (tags/UMIs) and collapsing reads to consensus sequences for error reduction.
3. Characterizing Heterogeneity: Generating genetic profiles from cfDNA combining CNV and rare mutation analyses.
4. Molecular Barcoding/Unique Molecular Identifiers (UMIs): Attaching unique or non-unique barcodes to polynucleotides, often before amplification, to aid in distinguishing true variants from PCR and sequencing errors by forming consensus sequences.
5. Bioinformatics Analysis: Advanced computational methods for filtering, mapping, normalizing, and detecting variants, including statistical and probabilistic models to infer unique molecules and correct for biases.
6. Application in Disease Monitoring: Specifically in cancer and fetal abnormalities, involving serial monitoring, therapy selection, and localization via imaging.

Prior Art Landscape (Pre-September 4, 2012)

Before the priority date, several key technologies and concepts were established and widely known to a PHOSITA:

• Next-Generation Sequencing (NGS): NGS had revolutionized genomic research, with technologies steadily improving and applications increasing exponentially. The basic workflow of NGS library preparation, including fragmentation, end repair, adapter ligation, and optional PCR amplification, was well-documented.
• Cell-Free DNA (cfDNA) Analysis: The presence of cfDNA in blood was known since 1948, and its elevation in pathological conditions like cancer was recognized. By the early 2010s, cfDNA was considered a promising source for studying cancer development and progression, and for non-invasive cancer screening or monitoring. Studies had already identified tumor-specific mutations in cfDNA using PCR-based methods.
• Detection of Rare Variants and CNVs in cfDNA: The concept of detecting genetic mutations, copy number alterations, and methylation changes in cfDNA using NGS was recognized, although challenges in sensitivity for low-level abnormalities were acknowledged. Methods like multiplex ligation-dependent probe amplification (MLPA) were used to identify CNVs in circulating, cell-free DNA. Whole-genome sequencing to examine plasma cfDNA CNVs in colorectal cancer patients was also performed.
• Molecular Barcoding/Unique Molecular Identifiers (UMIs) and Consensus Sequencing: The use of unique molecular identifiers (UMIs) or molecular barcodes (MBCs) to uniquely tag each molecule in a sample library, to provide error correction, and increase accuracy during sequencing was a recognized technique. UMIs allowed for distinguishing PCR duplicates from unique molecules and for filtering out PCR and sequencing errors, leading to more accurate variant detection, especially for low-frequency variants. Consensus sequencing, where multiple copies of a DNA template are sequenced and then computationally processed to obtain a consensus sequence, was also known to dramatically improve error rates. This method was used to identify many artifactual variations as technical errors.
• Digital PCR (dPCR): dPCR was a known technology for precise quantitation of individual DNA copies and highly sensitive detection of rare sequences, including mutations, with sensitivities as low as 0.001%. This technique partitioned samples to improve the signal-to-noise ratio for low-abundance targets.
• Bioinformatics for NGS Data Analysis: Bioinformatics approaches were critical for analyzing cfDNA sequencing data to detect genetic mutations, copy number alterations, and methylation changes. This included filtering low-quality reads and secondary alignments. Statistical and probabilistic methods were also employed, such as Poisson statistics in dPCR for absolute quantification.

Obviousness Combinations and Rationale

A PHOSITA, at the time of the invention, would have been motivated to combine these existing technologies to address the challenges of detecting rare mutations and CNVs in low-abundance cfDNA, particularly in clinical applications like cancer monitoring.

1. Combination for Detecting Rare Mutations in cfDNA with Error Correction:

• References: General knowledge of NGS, cfDNA analysis in cancer, and molecular barcoding/UMI and consensus sequencing for error correction.
• Motivation: The primary challenge in detecting rare mutations in cfDNA is the low fraction of tumor-derived DNA and the inherent error rates of NGS technologies. A PHOSITA would recognize that sequencing cfDNA (known to carry tumor characteristics) and then applying known error-correction techniques would be highly desirable. Molecular barcoding, which involves tagging individual DNA molecules before amplification and then computationally collapsing reads with the same barcode into a consensus sequence, was explicitly known to "reduce the rate of false-positive variant calls and increase sensitivity of variant detection" for "rare and low frequency somatic variants present in DNA samples such as cfDNA isolated from plasma". The ability to "confidently identify PCR duplicates" and "filter out PCR errors" was a clear benefit. This directly addresses the patent's teaching of "collapsing the set of sequencing reads to generate a set of consensus sequences, each consensus sequence corresponding to a unique polynucleotide among the set of tagged parent polynucleotides" to "reduce noise and/or distortion" and detect rare mutations (as described in the patent's definitions).
• Result: The combination would lead to a method for detecting rare mutations in cfDNA with improved sensitivity and accuracy, which is a central theme of US10793916. The patent's steps of sequencing cfDNA, filtering reads, mapping to a reference, identifying variants, calculating ratios, normalizing, and using barcodes to generate consensus sequences and filter out reads (as detailed in the patent definitions) directly map to the capabilities offered by combining these prior art elements.

2. Combination for Detecting Copy Number Variations (CNVs) in cfDNA:

• References: General knowledge of NGS, cfDNA analysis for cancer, and methods for CNV detection in cfDNA. Bioinformatics tools for analyzing cfDNA sequencing data, including copy number alteration detection, were also recognized.
• Motivation: Detecting CNVs in cfDNA was a known area of research for cancer diagnosis and monitoring. A PHOSITA would be motivated to apply NGS to cfDNA samples to detect CNVs, leveraging existing bioinformatics methods for quantifying mapped reads in predefined genomic regions and normalizing these counts to identify gains or losses. For example, the use of whole-genome sequencing of cfDNA to systematically examine plasma cfDNA CNVs in colorectal cancer patients was reported, involving mapping reads and identifying CNVs. The idea of normalizing read counts across regions and comparing to a control (even implicitly via comparison to a normal genome or population data) is a fundamental aspect of CNV detection by read depth.
• Result: This combination would render obvious the method for detecting CNV comprising sequencing cfDNA, mapping reads, quantifying in predefined regions, normalizing read counts, and comparing to a control sample, as described in US10793916's definitions.

3. Integration of Bioinformatics for Enhanced Detection:

• References: Bioinformatics for cfDNA analysis, statistical and probabilistic models in nucleic acid quantification (e.g., Poisson statistics in dPCR), and error correction methods for NGS data, including consensus sequencing and filtering.
• Motivation: A PHOSITA would understand the necessity of robust bioinformatics to handle the challenges of NGS data, especially from low-input cfDNA, which includes high error rates and biases. The use of statistical or probabilistic models for inferring true molecular counts and correcting for amplification bias, as taught in dPCR literature, would be readily applicable to NGS data. Furthermore, methods for filtering low-quality reads and correcting for biases (e.g., GC bias, window-averaged coverage) were standard practices in NGS bioinformatics. The patent's explicit mention of "hidden markov, dynamic programming, support vector machine, Bayesian network, trellis decoding, Viterbi decoding, expectation maximization, Kalman filtering, or neural network methodologies" for normalization and detection (from its definitions) points to well-known computational techniques that a PHOSITA would consider for complex data analysis problems.
• Result: The sophisticated bioinformatics pipelines described in US10793916 for filtering, normalizing, and correcting mapped reads, and for inferring molecular information with reduced noise and distortion, would have been obvious adaptations of existing computational methods to the specific challenges of cfDNA sequencing.

4. Applications for Disease Monitoring and Therapy Guidance:

• References: cfDNA analysis for cancer detection, diagnosis, prognosis, and monitoring. Digital PCR for rare mutant detection in cancer.
• Motivation: The clinical utility of cfDNA analysis for cancer monitoring, including tracking tumor evolution, detecting residual disease, and assessing treatment response, was recognized prior to the patent's priority date. It was known that ctDNA dynamics reflect tumor responses and progression. Therefore, applying the improved rare mutation and CNV detection methods to serial cfDNA samples for monitoring disease progression, selecting or modifying therapy, and prognosing conditions would be a natural and obvious extension for a PHOSITA seeking to translate these technologies into clinical practice. The ability to identify variants in cfDNA and then potentially use imaging to localize the abnormality was also a logical clinical follow-up.
• Result: The applications described in US10793916 for monitoring disease progression, guiding therapy, and correlating genetic findings with imaging or tissue biopsies would be obvious clinical implementations of the underlying detection technologies.

Conclusion of Obviousness

Considering the state of the art before September 4, 2012, many of the core methodologies described in US10793916, such as using NGS to detect rare mutations and copy number variations in cell-free DNA, employing molecular barcodes/UMIs and consensus sequencing for error reduction, and applying advanced bioinformatics for data analysis and disease monitoring, would have been obvious to a PHOSITA. The motivation to combine these known elements stemmed from the recognized challenges and opportunities in non-invasive genetic testing, particularly in oncology and prenatal diagnostics, which demanded high sensitivity and accuracy for low-abundance analytes like cfDNA. The PHOSITA would have sought to improve the accuracy and sensitivity of cfDNA sequencing for clinical applications by integrating available techniques for error correction and robust data analysis. While the patent may describe specific implementations or refinements, the overarching methods, when viewed through the lens of the extensive prior art in NGS, cfDNA, molecular barcoding, and bioinformatics, appear to be readily derivable combinations.