US 11402382

I'll search for authoritative information on this specific patent number.

Let me check for litigation, PTAB proceedings, and the official abstract.

Let me try to locate the official abstract and confirm the claim set.

Search results: USPTO / CAFC 2026

Bottom line on the litigation search: I found no CAFC 2026 docket, appeal, district-court case, IPR, or PGR referencing patent number 11402382. Targeted searches of Federal Circuit materials (including the court's published March 2026 oral-argument calendar at cafc.uscourts.gov and docketing notices on courtlistener) returned unrelated patents and parties (e.g., Wapp Tech v. Apple, Contour IP v. GoPro, 2026-2012). I cannot rule out a proceeding that is not indexed in the sources I could reach, but as of this search there is no evidence of any 2026 CAFC activity involving this number. Treat that as "not found," not as "confirmed none."

One third-party commercial database (TheraRadar, data refreshed Aug. 17, 2026) lists US 11402382 in a curated patent stack associated with Tecentriq (atezolizumab), family 61692081, nominal US expiry March 1, 2037 (https://theraradar.com/drugs/tecentriq/patents/). Caveats: that site states its thicket is curated, not FDA-authoritative, and biologics are not patent-listed in the Purple Book. Its own methodology note also inconsistently describes the seed set as derived from "Merck/MSD-assigned anti-PD-1 patents," so I would not rely on it. The Google Patents record — the authoritative text supplied — gives the adjusted expiration as 2038-03-29 (including patent term adjustment), which does not match the third-party 2037 figure.

---

Bibliographic summary (from the authoritative full text)

The NCBI patent-sequence record for SEQ ID NO: 20 of this patent independently corroborates the title, both inventors ("Hegde, P. and Huseni, M."), and the assignee (Genentech, Inc., South San Francisco, CA) — https://www.ncbi.nlm.nih.gov/protein/WIU64626.1.

---

Abstract

Uncertainty flag: the authoritative Google Patents text I was given for US 11,402,382 does not contain a verbatim abstract section — the machine-parsed page goes straight from the "Definitions" material into the description. I therefore cannot quote the granted abstract with confidence and will not reconstruct one from memory.

What the patent's own "Definitions"/summary text does state (paraphrased from the source): the invention "is directed to diagnostic and therapeutic methods for the treatment of cancer. Also provided are related kits and assays." The stated subject matter is diagnostic/therapeutic methods and compositions for treating a cancer — e.g., kidney cancer (RCC), lung cancer (NSCLC), bladder cancer (UBC), liver cancer (HCC), ovarian cancer, or breast cancer (TNBC) — based on measuring expression levels of defined gene signatures.

---

Plain-language overview of the independent claims

Important caveat: the text supplied is the specification/summary, not the certified granted claim set. The independent-claim descriptions below are reconstructed from (a) the specification's "the invention features/features/provides" passages, which track the claim language, and (b) the published pre-grant application US 2019/0369098 A1 (patents.justia.com/patent/20190369098), which shows claims running to at least claim 188. I have not verified the exact granted independent-claim numbering for the B2 patent.

The claims fall into three functional families, all anchored to kidney cancer (RCC):

A. Diagnostic / patient-selection / assay claims (biomarker combinations)

1. Benefit-identification method — Determine, in a patient sample, expression of genes from one or more of three signature groups:

   • T-effector/immune ("Teff"): CD8A, EOMES, GZMA, GZMB, PRF1, IFNG, PD-L1, CXCL9, CXCL10, CXCL11, CD27, FOXP3, PD-1, CTLA4, TIGIT, IDO1, PSMB8, PSMB9, TAP1, TAP2;
   • Angiogenesis: VEGFA, KDR, ESM1, PECAM1, FLT1, ANGPTL4, CD34;
   • Myeloid inflammation: IL6, CXCL1, CXCL2, CXCL3, CXCL8, PTGS2, CXCR1, CXCR2, S100A8, S100A9.

   The patient is identified as one who may benefit from a VEGF antagonist + PD-L1 axis binding antagonist combination if Teff genes are at/above a reference level (or, in the alternative, myeloid genes at/above), and/or angiogenesis genes are below reference; and the method then includes administering the therapy.

2. Therapy-selection method — Same gene panels and same up/down reference logic, framed as selecting the anti-cancer therapy for the patient.

3. Assay claims — A candidate-identification assay: for the VEGF antagonist + PD-L1 axis antagonist combination using the Teff/myeloid/angiogenesis panels; and a separate assay using the angiogenesis panel (VEGFA, KDR, ESM1, PECAM1, FLT1, ANGPTL4, CD34) above reference to identify candidates for a VEGF-antagonist-containing therapy.

B. Treatment claims

1. Treat-by-testing — Determine the panel expression as above (Teff/myeloid up, or angiogenesis down), then administer an effective amount of the VEGF antagonist + PD-L1 axis binding antagonist combination.

2. Treat-without-re-testing — Administer the VEGF antagonist + PD-L1 axis binding antagonist combination where, prior to treatment, the specified expression levels had already been determined (the "has been determined" formulation).

3. Angiogenesis-directed treatment — Determine that one or more of VEGFA, KDR, ESM1, PECAM1, FLT1, ANGPTL4, CD34 is at/above reference and administer a VEGF-antagonist-containing therapy (optionally plus a PD-L1 axis binding antagonist); plus the mirror-image "prior to treatment it has been determined" version.

C. Kit and second-medical-use / Swiss-type claims

1. Kits — Reagents for determining expression of the same gene panels, plus optional instructions, for identifying RCC patients who may benefit from the VEGF antagonist + PD-L1 axis antagonist therapy; and a kit for VEGF-antagonist candidate identification via the angiogenesis panel.

2. "For use" / "use in manufacture of a medicament" claims — An anti-cancer therapy comprising a VEGF antagonist and a PD-L1 axis binding antagonist for use in treating RCC in a patient whose pre-treatment gene expression was determined as specified; the corresponding manufacture-of-medicament use claim; a VEGF antagonist for use in treating RCC where angiogenesis-gene expression was determined above reference (optionally formulated with a PD-L1 axis binding antagonist); and the corresponding manufacture claim. (These are the EP-style second-medical-use formats, consistent with claims 158–188 in the pre-grant publication.)

Common dependent-claim subject matter across families: the reference level is a median (including a median Z-score of normalized expression) from an RCC patient population; expression may be nucleic-acid (mRNA, measured by RNA-seq/RT-qPCR/microarray/SAGE/MassARRAY/ISH) or protein (IHC, Western, ELISA, MS); sample may be FFPE/archival/fresh/frozen tumor tissue; the VEGF antagonist is bevacizumab or a multi-targeted TKI (sunitinib, axitinib, pazopanib, cabozantinib); the PD-L1 axis antagonist is a PD-L1, PD-1, or PD-L2 binding antagonist, specifically an anti-PD-L1 antibody (atezolizumab, YW243.55.570, MDX-1105, durvalumab, avelumab), with claimed HVRs (SEQ ID NOs: 63–68) and VH/VL sequences (SEQ ID NOs: 69–70); and the individual is human.

---

Key uncertainty notes

• Abstract: not verifiable verbatim from the supplied authoritative text; not quoted.
• Granted independent-claim numbering: not verified; the overview is reconstructed from the specification's summary language and the pre-grant publication. The justia listing showed claim structure differing slightly (e.g., its claim 158–162 formulations use only IL6, CXCL1, CXCL2, CXCL3, CXCL8, PTGS2 for the "below reference" branch, whereas the granted text's summary lists ten myeloid genes) — so the granted claim set may not match the publication exactly.
• Inventor name spelling: the patent text gives "Mahrukh Huseni" (also rendered "HUSENI" in the bibliographic data); this is the literal spelling in the record, and I have not auto-corrected it to the alternative "Hussein."
• Third-party patent-stack data: the TheraRadar listing of US 11402382 under atezolizumab is a curated dataset explicitly disclaimed as non-FDA-authoritative, and its 2037 expiry conflicts with the patent's stated 2038-03-29 adjusted expiration. Use with caution.
• Litigation: absence of a CAFC 2026 hit is a search-result finding only, not a legal clearance or an affirmative statement that no proceeding exists (e.g., a recently filed or sealed matter might not be indexed).

I'll search litigation-focused sources for any cases involving US 11,402,382.

Let me run more targeted searches on litigation databases.

Search Summary — US Patent No. 11,402,382

I searched for litigation involving US 11,402,382 and found none.

Before giving the bottom line, here is what the patent actually is, because it affects the plausibility of litigation:

• Title: "Diagnostic and therapeutic methods for cancer"
• Assignee: Genentech, Inc.; Inventors: Priti Hegde and Mahrukh Huseni
• Application: US 16/545,177; filed 2019-08-20; granted 2022-08-02; priority 2017-03-01 (adjusted expiry listed as 2038-03-29)
• Subject matter: Methods of identifying/selecting/treating kidney cancer (RCC) patients based on expression levels of gene signatures — a T-effector signature (CD8A, EOMES, PRF1, IFNG, PD-L1), an angiogenesis signature (VEGFA, KDR, ESM1, PECAM1, FLT1, ANGPTL4, CD34), and a myeloid-inflammation signature (IL6, CXCL1-3, CXCL8, PTGS2, etc.) — for treatment with a VEGF antagonist (bevacizumab) plus a PD-L1 axis binding antagonist (atezolizumab). It is a diagnostic-biomarker/companion-diagnostic-style patent, not a compound or formulation patent.

Litigation findings

No district court complaint, ITC Section 337 investigation, CAFC appeal, PTAB proceeding (IPR/PGR), or other adversarial proceeding naming US 11,402,382 was located in the sources returned.

The searches (Google Patents docket/litigation entries, Unified Patents materials, DrugPatentWatch/AlphaSignals litigation indexes, PACER Monitor, CourtListener, paragraphfour.com, and general web results) returned only unrelated patents that happen to share the "382" label — e.g., U.S. 6,004,314 and other matters in Pharmacyclics v. Sandoz, the AbbVie adalimumab "382 patent" (U.S. 6,090,382) in the Humira antitrust matter, and Apeldyn's U.S. 5,347,382 — none of which is US 11,402,382, and I have not auto-corrected or substituted any of them.

What is documented instead

The only non-litigation relevance I found is that US 11,402,382 appears in a patent-landscape listing as one of the patents associated with Tecentriq (atezolizumab) (TheraRadar, "Tecentriq patent landscape," listing "US 11402382," family 61692081, US expiry March 1, 2037). Being associated with a marketed biologic is exactly the situation that can give rise to future BPCIA/§ 262(l) "patent dance" litigation or biosimilar declaratory-judgment actions, but no such case is currently identified for this patent.

Important caveats

• My search relied on public web-indexed sources; I could not directly query PACER, Docket Navigator, Lex Machina, or Westlaw, and Unified Patents' case list is not fully crawlable. It is possible a very recent filing or a sealed/quieter docket exists that was not surfaced. If you need a definitive, counsel-grade clearance, a direct Docket Navigator / Lex Machina search and a PACER docket sweep for "11,402,382" are recommended.
• Because this is a method-of-diagnosis/treatment patent (not a drug-substance patent), it is relatively unlikely to be asserted in Hatch-Waxman ANDA litigation (which targets small-molecule claims) and more likely, if ever asserted, in BPCIA biosimilar litigation or a declaratory-judgment action against Genentech.

Bottom line: As of the available record, no litigation is known involving US Patent No. 11,402,382.

I'll verify the structured "no proceedings" default against live sources before writing anything up.

Let me run a second, more targeted round — including the sibling continuation and PTAB docket aggregators.

One more check against family members and defensive-aggregator dockets before I finalize.

PTAB / AIA Trial Proceedings — US 11,402,382 B2

Patent: US 11,402,382 B2 ("Diagnostic and therapeutic methods for cancer") — Genentech, Inc.; inventors Hegde & Huseni; App. No. 16/545,177; filed 2019-08-20; priority 2017-03-01; granted 2022-08-02; adjusted expiration 2038-03-29 (Google Patents).

---

Proceedings overview

Total AIA trial proceedings on file: 0. The structured USPTO Open Data Portal block returned no AIA trial proceedings, and my independent searches (Google Patents, USPTO PTAB/PTACTS document hits, Docket Alarm PTAB dockets, and law-firm PTAB round-ups) surfaced no IPR, PGR, or CBM petition naming US 11,402,382, no institution decision, no Final Written Decision, and no Federal Circuit appeal arising from such a proceeding.

Breakdown: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0.

Bottom-line defensive posture: The correct posture is not "the patent has survived two IPRs and is hardened," and it is not "claims 1–5 have been canceled." The accurate posture is: the patent is entirely untested at the PTAB. Every claim stands as issued, and no petitioner has yet put the claims through an AIA trial. That cuts both ways for a defendant — there is no invalidating FWD to lean on, but there is also no adverse PTAB precedent, no claim-construction record, and no estoppel framework binding you.

---

Proceedings

None. There is no proceeding to describe, so I am not going to invent a docket number, panel, or disposition to fill the requested per-proceeding template.

For completeness, the following are the specific negative findings:

• No IPR petition number in any of the IPR20xx-xxxxx ranges appears associated with US 11,402,382 in the sources I searched.
• No PGR petition (PGRs are available here on a post-AIA first-inventor-to-file patent with a 2017 priority date — this patent is PGR-eligible, which makes the absence noteworthy rather than merely technical).
• No CBM proceeding (and CBM review is unavailable for this subject matter in any event — the patent is a biotech/diagnostic claim, outside the "financial product or service" and technological-invention carve-outs under AIA § 18).
• No Federal Circuit appeal to trace, because there is no Board decision to appeal.

Caveat on confidence: The structured ODP block is authoritative for its ingest date, but ODP ingest lags. I searched for recently filed petitions and found nothing. I could not complete one planned cross-check on the sibling continuation (US 17/835,418, which issued as US 12,078,638 B2) because the search step terminated — treat any assertion that a petition exists against the continuation as unverified. The continuation's existence matters for strategy (see below) but I have no evidence of proceedings against it either.

Where to confirm with a primary source (do this before relying on the absence):

• PTAB E2E / PTACTS case search by patent number: https://ptacts.uspto.gov/ptacts/
• Patent Trial and Appeal Board decisions library: https://www.uspto.gov/patents/ptab/decisions
• CourtListener docket search for any CAFC appeal: https://www.courtlistener.com/
• USPTO Open Data Portal proceedings endpoint for application 16/545,177

---

Strategic summary

Claim status: all claims UNTESTED. No claim of US 11,402,382 has been canceled, confirmed, or even construed by the Board. There is no narrowing to itemize and no "surviving claims" list to build around — the patent is exactly as it issued. Contrast that with the situation a defendant more typically faces: a patent that has been through one or two IPRs has a paper record of admitted art, a PTAB claim construction, and often a narrowed claims set. Here there is none. That means (a) you cannot point to a canceled claim in a demand-letter response, and (b) if you do file an IPR, you are the first mover — with the attendant benefit (no estoppel against you, no adverse precedent) and risk (you fund the test case for everyone behind you, and you cannot rely on an earlier petitioner's expert work or institution-stage findings).

Estoppel landscape: no estoppel exists to bind anyone. 35 U.S.C. § 315(e)(2) estoppel requires a petitioner to have obtained an FWD — there has been no IPR FWD, so no petitioner and no privy is barred from any ground. Conversely, nothing bars you. For a defendant currently facing assertion of this patent: every prior-art ground is still available — § 102 and § 103 combinations, printed publications, patents, and public-use/on-sale theories, subject only to the ordinary limits of what the art predates (priority 2017-03-01, so the critical date is late and the relevant art window is narrow and crowded). Practically, the strongest § 102/§ 103 candidates will not be ancient references but rather the enormous pre-2017 literature on VEGF/PD-L1 combination biology in RCC, the IMmotion150 clinical data, and prior PD-L1/VEGF biomarker filings — the same subject matter the specification itself discusses. Also worth noting: for a diagnostic/method-of-treatment patent of this kind, § 101 eligibility and § 112 enablement/written-description attacks may be at least as promising as § 102/§ 103, and those are not available in an IPR (only § 102/§ 103 and, for PGR, § 112/§ 101). If you want § 101/§ 112 theories on the table at the Board, a PGR is the only route — but PGR eligibility has a hard nine-month clock from grant (grant date 2022-08-02), which has long since run. That is a significant, permanent loss of a PTAB forum for the § 101/§ 112 theories on this patent. Those arguments now live only in district court or the ITC.

Pattern signals:

• Same petitioner filing multiple IPRs? No petitioner has filed even one.
• Patent owner appealing PTAB outcomes aggressively? Not applicable — Genentech has had no Board loss here to appeal. Genentech is, however, a sophisticated and well-resourced patent owner with deep PTAB experience across its portfolio; if a petition is filed, expect a forceful Patent Owner Preliminary Response and likely a § 325(d) / Fintiv-style discretionary-denial push.
• Defensive aggregator in the chain? No evidence of Unified Patents or any similar entity challenging this patent. That is itself diagnostic: aggregators typically target broad, hard-to-design-around software/electronics claims, not originator biotech method-of-treatment claims.
• Litigation context. US 11,402,382 is a Genentech asset in the atezolizumab (Tecentriq) family. One third-party patent-landscape summary lists it among a curated (non-FDA-authoritative) Tecentriq-adjacent set with a nominal expiry around 2037 (TheraRadar) — that listing is curated, not an Orange Book/Purple Book listing, and Purple Book does not mandate patent disclosure for biologics. There is no biosimilar BPCIA litigation I am aware of in which this patent has been asserted, which is consistent with there being no IPR: method-of-treatment and diagnostic claims of this type often sit unchallenged precisely because they are not the claims a biosimilar manufacturer must clear.

The family is the real exposure, not this patent standing alone. US 11,402,382 has a continuation that issued as US 12,078,638 B2, and there are later-filed continuations in the same lineage (e.g., published applications in the 2023/2024 series). A challenger who knocks out only the '382 patent may leave the family's other members — with similar or broader claims — intact. Any defensive strategy should be family-scoped from the outset, not patent-scoped. (Verify the continuation's claims and status directly; I could not complete that verification here.)

---

Recommended next steps

1. Because there is no PTAB record to cite, the absence is the finding. Say so plainly to your client: no IPR, no PGR, no CBM, no FWD, no CAFC appeal. Do not characterize the patent as "hardened" or "invalidated" — it is untested. Confirm the negative by running a patent-number search in PTAB E2E/PTACTS and the PTAB decisions library, and by checking CourtListener for any appeal, before you put the conclusion in writing.

2. Map the family before you map the invalidity case. Pull the continuation US 12,078,638 B2 (from App. 17/835,418) and any later continuations, and compare independent-claim scope across the family. If the '382 claims are narrow (they recite specific gene panels and a specific VEGF-antagonist + PD-L1-axis-antagonist combination in kidney cancer), the commercially meaningful threat may be a differently-scoped sibling. A challenge aimed only at '382 may be wasted spend.

3. A PGR is off the table — calendar that. The nine-month PGR window from the 2022-08-02 grant has expired, so § 101 and § 112 grounds cannot be brought to the Board on this patent. If your invalidity theory is eligibility- or written-description-driven, plan for district court/ITC and preserve those theories in your contentions; if it is § 102/§ 103-driven, an IPR remains available.

4. If you file an IPR, own the first-mover economics. As the first petitioner you will bear the full cost of: (a) establishing the priority date and the WO/priority-document chain (the 2017-03-01 priority date and the PCT lineage are contestable territory, and priority disputes are a common and effective petitioner-side lever here); (b) expert declarations that later petitioners may free-ride on; and (c) the § 315(b) one-year clock from service of any complaint naming this patent. If you are a co-defendant with others, consider coordinating so that a single well-funded petition carries the group rather than multiple parallel petitions, which draw discretionary-denial risk under the Board's parallel-petition practice and the Fintiv/Sotera line of cases.

5. If a petition is filed later, the milestones to watch are fixed by statute: institution decision due within six months of the petition (35 U.S.C. § 314(b)); final written decision due within twelve months of institution (§ 316(a)(11)); and the one-year § 315(b) bar clock running from service of a complaint alleging infringement. Because there is nothing pending today, there is no hearing date or FWD due date to track — but a defendant served with a complaint must diary the § 315(b) date immediately, since that is the only hard deadline that will exist.

Caveat, stated plainly: this report reflects that no AIA proceeding on US 11,402,382 appeared in the USPTO ODP structured data provided to me or in my independent web searches as of 2026-09-30. If a recently filed petition exists that neither source has indexed, it would be a first petition against this patent and would not change the structural analysis above (all claims untested, no estoppel, § 101/§ 112 unavailable at the Board).

I'll research the assignment record for US 11,402,382. Let me start with targeted searches.

Let me search for the Google Patents legal events and the specific assignment records.

Let me try a few more targeted searches for the specific assignment record.

I hit the tool-call limit before I could pull the Assignment Center record directly, so I'll report precisely what the available record establishes and be explicit about what I could not verify.

Assignment & Ownership Analysis — US 11,402,382 B2

Critical sourcing caveat (read first): I was not able to retrieve the actual recorded assignment entries (reel/frame), correspondent of record, or execution/recording dates for this patent. The USPTO Assignment Center / assignment.uspto.gov database is not fully surfacable through general web search, and my searches returned only unrelated "382"-labeled patents (e.g., U.S. 6,090,382, U.S. 5,347,382) and prosecution-history boilerplate — none tied to this patent. Nothing below should be treated as a verified reel/frame citation. What I can establish comes from the face of the granted patent and the patent-family record. Anything I could not confirm is flagged as such rather than inferred.

---

Inventors

Employer is determinable with high confidence: both inventors are listed as Genentech-affiliated authors on contemporaneous Genentech publications (e.g., Yuen et al., Nat Med 2020, "High systemic and tumor-associated IL8 correlates with reduced clinical benefit of PD-L1 blockade," which lists Priti S. Hegde and Mahrukh A. Huseni at "Genentech, South San Francisco, California 94080"). They are also co-inventors on other Genentech/Roche applications (e.g., MX-2017015937-A, CN-107810011-A).

Unusual-pattern check: No evidence of inventors departing the assignee within 12 months of filing. Both remained at Genentech well after the 2017‑03‑01 priority date (Huseni and Hegde appear on Genentech work published into 2020). No fire-sale precursor pattern is present.

---

Original assignee

Genentech, Inc., South San Francisco, CA (US).

• Entity status: Operating. Wholly owned subsidiary of Roche (acquired 2009); a large-cap commercial biopharmaceutical manufacturer.
• Products relevant to the claims: The claimed methods recite treatment with a VEGF antagonist (bevacizumab/bevacizumab-class) plus a PD-L1 axis binding antagonist. Genentech markets Avastin (bevacizumab) and Tecentriq (atezolizumab, an anti-PD-L1 antibody whose HVR/VH/VL sequences are recited in the specification, e.g., SEQ ID NOs: 63–70). TheraRadar lists US 11,402,382 in the Tecentriq patent landscape (family 61692081, US expiry March 1, 2037). So Genentech both makes the products used in the claimed methods and is listed in a commercial patent-landscape as an atezolizumab-associated patent holder.
• Current status: Operating (no bankruptcy, no dissolution, no acquisition of the assignee by a patent-holding shell).

---

Assignment timeline

Finding: On the face of the granted patent, the (71) Applicant and the (73) Assignee are the same entity — Genentech, Inc. Google Patents likewise lists both "Original Assignee: Genentech Inc" and "Current Assignee: Genentech Inc," with no intervening transfer. This is the signature of a conventional employee-invention assignment (or assignee-applicant filing) — an internal, pre-/at-filing transfer from the inventors to their employer, not a post-issuance portfolio transfer.

I could not retrieve any recorded reel/frame entries (see caveat above). Because I cannot cite a reel/frame, I will not fabricate a timeline of recorded assignments. What is verifiable from the record:

• No post-issuance assignment to a third party is reflected on the patent face or in the public patent-family metadata I retrieved. There is no successor assignee, no LLC, no aggregator.
• The continuation US 17/835,418 → US 12,078,638 B2 (priority claim filed 2022‑06‑08; issued 2024‑09‑03) is likewise a Genentech, Inc. case — i.e., the family stayed inside Genentech.

Bottom line for this section: the accessible record shows only the original (inventor → Genentech) ownership, with no subsequent recorded conveyances surfaced. Per the task's own guidance, "no recorded post-issuance assignments … usually means the original assignee still owns the patent" — which is exactly what the patent face suggests. I recommend a direct reel/frame pull from https://assignmentcenter.uspto.gov/ (search by patent number 11402382) to confirm the inventor-assignment record date, reel/frame, and correspondent.

---

Timeline diagram

Ownership events known from the public record (note: these are filing/grant events, not recorded assignment conveyances; no reel/frame verified):

timeline
    title Ownership of US 11402382
    2017 : Priority filing PCT US 2018 020460
    2018 : PCT published as WO 2018160841
    2019 : US 16 545 177 filed by Genentech
         : Application published US 20190369098A1
    2022 : Patent issued US 11402382B2
         : Continuation US 17 835 418 filed
    2024 : Continuation issued US 12078638B2

All entries reflect Genentech, Inc. ownership throughout; no transfer event identified.

---

NPE / troll-pattern signals

For each signal: present / not present / unclear, with the basis stated.

1. Shell-entity transfer — not present. No assignment to any "IP / Patents / Licensing / Holdings / Ventures" LLC appears. The named assignee is Genentech, Inc., an operating manufacturer. (Caveat: confirmed via patent face and family metadata, not via a reel/frame pull.)

2. Known asserter in the chain — not present. Neither Genentech, Inc. nor any Roche entity appears on the public NPE lists referenced (Acacia, Marathon, Intellectual Ventures, Wi-LAN, Conversant/Mosaid, Vringo, Pendrell, etc.), and no such entity appears anywhere in the chain.

3. Repeat correspondent across the chain — unclear. I could not retrieve the correspondent of record for any recording, so I cannot assess recurrence. This signal cannot be scored without the reel/frame data. Note for the record: Genentech's prosecution correspondence is typically routed to its in-house/South San Francisco address (e.g., "c/o Genentech, Inc., 1 DNA Way, South San Francisco" appears as the agent address on Genentech PCT filings), which would be consistent with operating-company practice rather than NPE filing-agent behavior — but I have not verified this for the subject patent.

4. Cascading transfers — not present. No chained LLC transfers within 24 months; in fact no transfers of any kind were surfaced.

5. Pre-litigation transfer — not present. The prior litigation analysis (generated earlier) found no litigation involving this patent. With no suit and no assignment, this signal is inapplicable. Cross-reference note: the earlier Litigation Summary and this analysis are consistent — both indicate a Genentech-held method patent with no adverse proceedings and no third-party ownership.

6. Bankruptcy fire-sale — not present. Genentech/Roche are solvent operating entities; no Chapter 7/11 proceeding.

7. Privateering — not present. No transfer from Genentech to an assertion vehicle. Ownership remains with the operating originator.

8. Defensive aggregator (anti-NPE) — not present. Chain does not terminate at RPX, AST, LOT, Unified, or OIN.

---

Verdict

Insufficient data.

Justification: Every NPE signal that can be assessed is negative, and the accessible record (patent face + family metadata) shows Genentech, Inc. as both (71) applicant and (73) assignee with no succeeded owner — i.e., a conventional operating-company ownership and no transfer to any assertion entity. However, I could not retrieve a single recorded reel/frame entry or correspondent, so I cannot affirmatively document the chain the way the rubric requires (and the chain appears to consist of only the original inventor→assignee conveyance, which the rubric maps to "Insufficient data"). This is a negative-on-the-evidence result, not a suspicion of NPE activity: there is nothing in the record suggesting a shell, an asserter, or a fire-sale. Verify at the USPTO Assignment Center: https://assignmentcenter.uspto.gov/ (search patent number 11402382; mirror: https://assignment.uspto.gov/patent/index.html).

Standing caveats: (a) I did not directly query PACER, Docket Navigator, Lex Machina, the USPTO Assignment Center, or SEC EDGAR; my tool access is limited to web search, so a definitive reel/frame chain requires a direct database pull. (b) Because this is a method-of-diagnosis/treatment patent tied to a marketed biologic (Tecentriq/Avastin), the realistic future assertion vector is BPCIA / § 262(l) biosimilar litigation, not Hatch-Waxman ANDA suits — consistent with the earlier Litigation Summary. (c) I have not fabricated any assignment entry; where the record was silent, I have said so.

I'll research US 11,402,382's cited prior art and related references. Let me run targeted searches.

Let me pull the actual patent-citation list and related documents for this patent.

Prior-Art Analysis — US Patent No. 11,402,382 B2

0. Critical source limitation (read first)

I must flag a hard blocker before any reference list, because the alternative would be to fabricate one.

The authoritative full text supplied for US 11,402,382 does not contain the front-page "(56) References Cited" section. The parsed record runs: bibliographic data → classifications → Definitions/description. There is no "U.S. Patent Documents," "Foreign Patent Documents," or "Other Publications" block, and no examiner's list of cited art.

My attempt to retrieve that list from third-party mirrors (Google Patents "Patent Citations" tab, Justia, FreePatentsOnline) was not successful within the search budget, and I never obtained a verified "(56)" block.

Therefore: I cannot give you the certified, examiner-cited reference list for US 11,402,382, and I will not construct one from memory or inference. Any specific-number citation list I produced without the source would be unverifiable and potentially wrong. What follows is limited to (a) documents I did verify in the record, and (b) candidate references surfaced by search whose status as citations is explicitly flagged as unconfirmed.

I also could not query USPTO PatentCenter/PAIR, Docket Navigator, or the Image File Wrapper directly — only web-indexed mirrors of the official record.

---

1. Number verification (no substitution)

The number resolves cleanly and consistently to one patent — I have not substituted any similar number:

Legal framework: With an effective filing date of 2017‑03‑01, the AIA first‑inventor‑to‑file provisions govern. Prior art must therefore be (a) publicly available before 2017‑03‑01 under §102(a)(1), or (b) a U.S. patent/application effectively filed before 2017‑03‑01 under §102(a)(2). Pre‑AIA §102(b)/(e) year‑and‑country rules do not apply.

---

2. Documents verified in the record — but these are forward citations, not prior art

Google Patents' "Cited By (18)" list was partially retrieved. Every item below post‑dates the 2017‑03‑01 priority date, so none is §102 prior art against this patent. They are included only to correct a common error (conflating "cited by" with "cites"):

Family / parallel publications (also not art): WO 2018/160841 A1 (2018‑09‑07, PCT parent); US 2019/0369098 A1 (pre-grant pub, 2019‑12‑05); EP 3589754 A1/B1 (EP 18712321.1, filed 2018‑03‑01).

---

3. Candidate prior art (pre‑2017‑03‑01) — status: unconfirmed as citations

These appeared in searches connected to this patent or its relatives, and each pre‑dates the priority date. I mark confidence honestly. None of these is confirmed by me to be on the (56) list of US 11,402,382.

3.1 WO 2016/205277 A1 — "PD‑L1 antagonist combination treatments" (Genentech)

• Citation: WO 2016/205277 A1, PCT; Genentech, Inc. (pub. date listed as 2016‑12‑22 in search results — date not independently re‑verified)
• Confidence it is in the record: High — the document surfaced repeatedly in the same Google Patents citation graph as US 11,402,382.
• Description: Combination-treatment disclosure pairing a PD‑L1 axis binding antagonist with a second agent, including anti‑VEGF/bevacizumab‑type combinations; antibody sequences for anti‑PD‑L1 antibodies.
• §102 relevance: Potentially anticipates the dependent claims reciting the therapeutic combination itself — e.g., the limitations to bevacizumab as the VEGF antagonist and to atezolizumab / the SEQ ID NOs 63–68 HVRs and SEQ ID NOs 69–70 VH/VL (claim family B "treat‑without‑re‑testing" dependents). It is unlikely to anticipate any independent claim, because every independent claim requires the gene‑signature determination step (Teff/myeloid at-or-above, or angiogenesis below/above a reference level), which this reference does not perform.

3.2 "Biological markers for identifying patients for treatment with VEGF antagonists" (Genentech) — family CN 104203268 A

• Citation: CN 104203268 A (and the corresponding U.S./PCT members of that family) — Genentech
• Confidence: Moderate — surfaced in the citation graph tied to US11402382B2; exact U.S. counterpart number not confirmed.
• Dates: Family priority/публикация appears to be 2013–2014, i.e., well before 2017‑03‑01.
• Description: VEGF‑antagonist (bevacizumab) biomarker selection based on angiogenic gene expression.
• §102 relevance — the strongest candidate in this set. Because US 11,402,382 expressly claims an angiogenesis signature of VEGFA, KDR, ESM1, PECAM1, FLT1, ANGPTL4, CD34 as the selection marker for a VEGF‑antagonist‑containing therapy, a pre‑2017 disclosure of the same gene set + a reference level for VEGF‑antagonist selection would bear directly on:
  • the angiogenesis assay claim;
  • the "VEGF antagonist for use…" / Swiss‑type claims; and
  • the angiogenesis‑directed treatment claim.
    Anticipation, however, requires the reference to disclose all claimed elements — including the specific seven‑gene panel and the at‑or‑above‑reference comparison — in one document. A partial overlap supports §103 obviousness far more plausibly than §102.

3.3 "Biomarkers and methods for treating PD‑1 and PD‑L1 related disorders" (Genentech) — family CN 105209919 A

• Citation: CN 105209919 A (and U.S./PCT family members) — Genentech
• Confidence: Moderate (surfaced in the same citation graph).
• Dates: 2014–2015, before the priority date.
• Description: PD‑L1/PD‑1 axis biomarker panels — interferon‑γ/effector‑gene signatures for patient selection.
• §102 relevance: Bears on the Teff signature claims (CD8A, EOMES, GZMA/GZMB, PRF1, IFNG, PD‑L1, CXCL9‑11, IDO1, etc.). Same caveat: independent claims also require the VEGF‑antagonist element and the reference‑level comparison, so this is likely §103 art in combination, not standalone §102 art.

3.4 WO 2016/109546 A2 (Genentech)

• Citation: WO 2016/109546 A2, published 2016‑07‑07, Genentech — flagged as a category "X" reference in a different International Search Report I retrieved.
• Confidence: Low/uncertain — I saw it cited in an unrelated ISR, not confirmed against this patent's file.
• §102 relevance: Cannot responsibly map to claims without reviewing its disclosure. Listed for completeness only.

---

4. §102 vs. §103 — the honest framing for these claims

This matters more than the individual references, so I'll state it plainly:

• The claim families here are gene‑signature selection/diagnosis/treatment claims. Anticipation under §102 requires a single reference disclosing every element — the specific gene panel, the direction of comparison (at/above or below), the reference level, and the VEGF‑antagonist + PD‑L1‑axis‑antagonist therapy. That is a demanding standard, and I found no verified single reference meeting it.
• Where the art is a partial overlap (e.g., an angiogenesis biomarker reference lacking the PD‑L1 axis element), the realistic challenge is §103 obviousness in view of the 2016 clinical context (atezolizumab + bevacizumab combinations were already in trials) — not §102.
• The broadest independent claims (the "treat‑without‑re‑testing" and second‑medical‑use formats, which recite only that expression "has been determined" plus administration) are the most exposed, because their novelty rests on the determination step being a claim element rather than on new composition matter.

---

5. How to obtain the definitive (56) list

Since I could not, I will not guess. The authoritative sources are:

1. USPTO PatentCenter / PAIR — the issued patent's front page and the Image File Wrapper (Form PTO‑892 "List of References Cited by Examiner" and the "Notice of References Cited").
2. USPTO Global Dossier — the "Citations" tab, which also aggregates the EPO's cited‑document list for EP 3589754.
3. The PCT International Search Report for PCT/US2018/020270 (the parent of WO 2018/160841 A1) — Box C, "Documents Considered to Be Relevant," which lists the art the ISA actually searched against this subject matter.
4. EP 3589754 — the European search report and the "List of cited documents" in the B1 specification.

---

Key uncertainty notes (consistent with prior sections)

• No (56) list was available in the authoritative text or retrievable by search. The references in §3 are candidates, not confirmed citations; I have not represented them as examiner citations.
• The "Cited By" items in §2 are emphatically not §102 prior art — all post‑date 2017‑03‑01. If any earlier deliverable asserted otherwise, that would be an error.
• No similar patent numbers were substituted. All identifiers above are literal.
• Publication dates marked "not independently verified" should be confirmed against PatentCenter before being used in any filing — the 2016‑12‑22 date for WO 2016/205277 A1 is the one I am least certain of, and it is close enough to the priority date that it matters.

I'll start by trying to locate the actual prior-art record for this patent — the references cited on its face and any "Prior Art" listing — since the task asks me to build the § 103 analysis on those results. Let me run several targeted searches.

I found several relevant prior-art families. Let me pin down a few more references — the pre-2017 atezolizumab/bevacizumab RCC work, the atezolizumab antibody-sequence patents, and the T-effector/angiogenesis signature publications — so the § 103 analysis is grounded.

Obviousness Analysis — US 11,402,382 B2 under 35 U.S.C. § 103

0. Sourcing note (read first)

The page as supplied to me did not contain a discrete "Prior Art" section. The authoritative Google Patents text I was given runs from the bibliographic block and classifications straight into the "Definitions"/description — it includes the patent's figures, definitions, and summary passages, but no listing of cited references, no "Similar Documents," no "Cited By," and no search-report tabulation. I did not receive a separate prior-art table to "use."

Rather than fabricate a prior-art list, I reconstructed the citable prior-art universe from (a) the references that the patent family itself cites (surfaced via Google Patents citation data), and (b) references that the patent's own examples and Background presuppose. I flag each reference below with its date and my confidence. Where I could not verify a detail, I say so rather than supplying it. This is consistent with the strict rule to interpret identifiers literally: I have not substituted or "corrected" any number, and where a reference is one I could not fully verify in this session (e.g., an exact sequence listing or an exact passage), I say so.

A critically important consequence, stated up front: the single most on-point disclosure — the IMmotion150 biomarker analysis (McDermott et al., Nat Med 24:749–757, published 2018-06-04; EGA study EGAS00001002928) — postdates the patent's 2017-03-01 priority date and is therefore not citable as § 102/§ 103 prior art. The patent's Example 5 describes that very trial. So the obviousness case must be built on art that predates 2017-03-01. That is doable, but it is a real constraint.

---

1. Governing framework and critical date

• AIA applies. US 16/545,177 was filed 2019-08-20; priority (as listed) is 2017-03-01. Prior art is therefore measured under AIA §§ 102(a)(1)–(a)(2) with the § 102(b) exceptions, and obviousness under § 103.
• Critical date: 2017-03-01. Priority caveat: whether the granted claims are entitled to 2017-03-01 depends on whether the 2017-03-01 application (apparently a US provisional, given the one-year gap to the March-2018 PCT — the EP sibling is EP 3 589 754 B1) actually described the three specific signatures (Teff = CD8A/EOMES/PRF1/IFNG/PD-L1; angio = VEGFA/KDR/ESM1/PECAM1/FLT1/ANGPTL4/CD34; myeloid = IL6/CXCL1/CXCL2/CXCL3/CXCL8/PTGS2). If the signatures or the "median Z-score" reference logic were first disclosed only in the later filing, the effective date shifts later and more art becomes citable. I could not verify the provisional's contents in this session; that verification (a priority/written-description check) materially changes the analysis and should be done before relying on this memo.
• Graham v. John Deere factors (scope and content of prior art; differences; PHOSITA level; secondary considerations), applied through KSR Int'l v. Teleflex, 550 U.S. 398 (2007) — including the "finite number of identified, predictable solutions" and "design incentive/obvious-to-try" rationales.
• § 101 overlay: these are diagnostic/selection/correlation claims; a Mayo/Alice vulnerability exists independent of § 103. The user asked only about § 103, so I note it and move on.

---

2. Person of ordinary skill in the art (PHOSITA)

A team-level artisan: an M.D./Ph.D. (or M.S. with several years' experience) in oncology, tumor immunology, or molecular biology, with practical experience in (i) gene-expression biomarker discovery (RNA-seq, RT-qPCR, microarray) and IHC on FFPE tumor tissue, and (ii) design/analysis of RCC immunotherapy clinical trials — including familiarity with PD-L1 axis biology, VEGF/angiogenesis biology, and the standard statistical practice of comparing a patient's normalized expression to a population median (including Z-scores). This artisan reads the primary oncology literature and the patent literature and knows the marketed/clinical-stage agents (bevacizumab; sunitinib/axitinib/pazopanib/cabozantinib; atezolizumab/nivolumab/pembrolizumab).

---

3. The claimed subject matter, grouped

Building on the earlier claim-overview (not repeated here), the asserted claims fall into:

• A. Diagnostic/selection. Determine expression of ≥1 gene from the Teff, angio, and/or myeloid panels; classify by "at or above" (Teff, myeloid) or "below" (angio) a reference; identify the patient as a combination-therapy candidate.
• B. Treatment. (B1) determine-then-administer the VEGF-antagonist + PD-L1-axis-antagonist combination; (B2) administer the combination where the levels "have been determined" before treatment; (B3) determine angio ≥ reference and administer a VEGF-antagonist therapy (optionally + PD-L1 antagonist).
• C. Kits for the same panels.
• D. Second-medical-use/Swiss-type (bevacizumab+atezolizumab; VEGF antagonist alone) with the pre-treatment determination limitation.
• Common dependents: reference = median (incl. median Z-score) from an RCC population; nucleic-acid (mRNA) or protein readout; FFPE tumor tissue; bevacizumab or a multi-targeted TKI; anti-PD-L1 antibody with the atezolizumab HVRs (SEQ ID NOS: 63–68) and VH/VL (SEQ ID NOS: 69–70).

---

4. Prior art identified (all pre-2017-03-01 unless noted)

Citable-status caveat on P3: WO 2017/181111 / US 2019/0032151 was filed ~April 2017 but claims priority to ~April 2016. Under AIA § 102(d)(2) it is "effectively filed" as of the ~April 2016 priority application, so its subject matter could qualify as § 102(a)(2) art even though it published after 2017-03-01. However, § 102(b)(2)(C) disqualifies a § 102(a)(2) reference if the reference and the claimed invention were, by the critical date, owned by or under obligation of assignment to the same person. Both P3 and the patent-in-suit are Genentech — so the common-ownership exception very likely removes P3 as exclusionary art. This does not remove P3's evidentiary value as a "state of the art" / what-was-known indicator and as an obviousness rationale, but it should not be treated as a clean § 102(a)(2) anticipation reference without confirming inventorship/ownership continuity. I flag this as the single biggest legal nuance in the analysis.

---

5. Combination 1 — the Teff-signature claims (A, B1, B2, D, and the Teff portion of C)

Proposed combination: P1 (Wallin 2016) + P2 (McDermott 2016, JCO) + P6-as-background, in view of P3/P7-style signature art.

1. P1 alone teaches the predicate composition and indication. It discloses administering atezolizumab + bevacizumab to previously untreated mRCC and shows the combination drives T-cell/chemokine pharmacodynamic changes. A claim that "administers an effective amount of an anti-cancer therapy comprising a VEGF antagonist (bevacizumab) and a PD-L1 axis binding antagonist (atezolizumab)" to a kidney-cancer patient reads almost directly on P1's clinical protocol.
2. P1 + P2 teach that the response is immune-mediated and correlate with T-cell/PD-L1 biology in RCC. P2 (phase Ia) reports mRCC activity with immune correlates; P1 reports increased antigen-specific CD8 migration. A PHOSITA reading these two Genentech papers would be led directly to the hypothesis that T-effector/IFN-γ biology marks the patients who benefit — the exact hypothesis the Teff panel operationalizes.
3. The Teff gene list itself was a known, finite, predictable set. CD8A, IFNG, PRF1, EOMES, GZMA, GZMB, and PD-L1 were the standard, well-characterized markers of CD8 T-effector/IFN-γ activity; by 2016 they were used as a "T-effector/IFN-γ response signature" to describe checkpoint-blockade response across melanoma, NSCLC, and bladder cancer. Selecting these known effector genes — and using PD-L1/PD-1/CTLA4/IDO1/loci such as PSMB8/PSMB9/TAP1/TAP2 as antigen-presentation/immune-regulatory markers — is routine identification of a finite set of known, predictable markers, which KSR treats as obvious. P3 makes this concrete: its Table 2 lists inter alia CD8A, EOMES, GZMA, GZMB, IFNG, PRF1, CXCL9, CXCL10, CXCL11 as the RCC biomarkers of interest for the same drug combination.

Motivation to combine (KSR rationales):

• Same field, same problem, same agents: all references address biomarkers of the atezolizumab ± bevacizumab response in RCC.
• Recognized need: the Nat Med IMmotion150 authors themselves frame the field's problem as "biomarkers for response and resistance are lacking" in mRCC — a pre-existing, expressly articulated need that predates 2017.
• Predictable solution: extending a validated T-effector signature from other tumors to RCC, and testing whether it stratifies the already-running combination, is an obvious application of a known tool.

Weakness of the combination: P1 and P2 establish the combination and the immune correlates but are small (n ≈ 11 in P1). The leap to a pre-treatment, median-Z-score-thresholded, multi-gene Teff classifier that is predictive of the combination vs. sunitinib is where the patent adds something. If the patent's priority filing actually first disclosed the signature-as-classifier result, a non-obviousness argument exists (see § 9).

---

6. Combination 2 — the angiogenesis-signature claims (A-alt, B3, D-alt, C-alt)

Proposed combination: P7 (VEGF-inducible angiogenesis signature / anti-VEGF biomarker art) + P4 (Genentech VEGF-antagonist biomarker application) + the established RCC anti-VEGF standard of care (bevacizumab and sunitinib/axitinib/pazopanib/cabozantinib).

• It was established before 2017 that (i) ccRCC is hyperangiogenic through VHL loss and VEGF overproduction, and (ii) bevacizumab and VEGF-TKIs are standard-of-care. P7 teaches that a VEGF-inducible angiogenesis gene-expression signature predicts anti-VEGF benefit in a solid tumor with a bevacizumab-containing regimen.
• The claimed angio panel (VEGFA, KDR [VEGFR2], FLT1 [VEGFR1], ESM1 [endocan], PECAM1 [CD31], CD34, ANGPTL4) is a textbook endothelial/angiogenesis marker set — the canonical genes any artisan would choose to measure "how angiogenic/alive is the vasculature." KDR and FLT1 are the VEGF receptors themselves; PECAM1/CD34 are standard endothelial markers; VEGFA is the ligand; ESM1 and ANGPTL4 are VEGF/Ang-regulated endothelial/angiogenesis genes.
• The claimed directionality (angio "at or above" reference → give a VEGF antagonist) is exactly what P7 teaches. So B3 and the corresponding assay/kit/use claims are strongly suggested.

Motivation: the same "select the anti-VEGF option for angiogenic tumors" rationale that P7 supplies for mCRC applies a fortiori in RCC, which is the archetypal VEGF-driven tumor. Combining P7's signature logic with the known RCC standard of care is a predictable, one-tool application.

Weakness: if the art does not expressly teach the specific seven-gene panel, the patent can argue the selection was non-obvious. But given how canonical this gene set is, a KSR "finite number of predictable solutions / routine optimization" attack is strong.

---

7. Combination 3 — the myeloid-signature sub-claims

Proposed combination: P1/P2 + the known biology of myeloid/neutrophil-mediated immunosuppression (IL6, CXCL1/CXCL2/CXCL3/CXCL8, PTGS2/COX-2, CXCR1/CXCR2, S100A8/S100A9) + P7.

• IL6, the CXCL1–3/CXCL8 chemokine axis, PTGS2 (COX-2), and the S100A8/S100A9 (calprotectin) alarmins were well-known myeloid-inflammation/neutrophil-recruitment markers before 2017, and VEGF's ability to promote myeloid-derived suppressor-cell accumulation was known.
• Motivation: once one is profiling the T-effector axis in the same tumors (Combination 1), the field's recognition that "myeloid inflammation suppresses antitumor immunity" makes adding a myeloid-inflammation panel an obvious parallel measurement.
• This is where the patent's real insight likely lies: the interaction result — Teff-high/Myeloid-high tumors are relatively resistant to atezolizumab monotherapy yet rescued by adding bevacizumab (FIGS. 27–28; the EGA study abstract states this explicitly). That is a genuinely non-trivial, non-obvious finding if first disclosed in the priority filing.

---

8. Combination 4 — the anti-PD-L1 antibody-sequence dependent claims

Proposed combination: P6 (WO 2010/077634 / US 8,217,149) or any Genentech anti-PD-L1 sequence disclosure + any of Combinations 1–3.

Dependent claims reciting an anti-PD-L1 antibody with HVR-H1 GFTFSDSWIH (SEQ ID NO: 63) … HVR-L3 QQYLYHPAT (SEQ ID NO: 68) and VH/VL of SEQ ID NOS: 69–70 implement the atezolizumab sequences, which are disclosed in Genentech's 2010-era anti-PD-L1 antibody applications. A claim that merely names the known antibody (or its known CDRs) in an otherwise obvious method adds nothing patentable; the antibody limitation is fully met by the prior art. This is the easiest § 103 ground and also, notably, the closest to a § 102 anticipation of the dependent claims. (Caveat: I did not re-extract the sequence listing from P6 this session; a routine verification should confirm the SEQ ID correspondence.)

---

9. Countervailing considerations (arguments the patentee will make)

These are real and should not be dismissed:

1. The closest art is the inventors' own, post-dating the claim. The IMmotion150 biomarker analysis (P8, 2018) is not prior art; the patent's Example 5 is essentially that analysis. A court would ask what the pre-2017 art actually enabled one to predict for the combination vs. sunitinib in mRCC — and P1/P2 did not establish predictive (as opposed to prognostic) value.
2. Unexpected result / secondary considerations. If the patent shown that bevacizumab overcomes myeloid-inflammation-associated atezolizumab resistance (Teff-high/Myeloid-high), that interaction is a candidate "unexpected results" argument under Graham factor 4 — provided the applicant can show the result was unexpected relative to the closest art and that it drove the claimed treatment selection. The European sibling (EP 3 589 754 B1) shows the applicant pressed second-medical-use formats tuned to this insight (e.g., "bevacizumab+atezolizumab for use in treating RCC … where CD8A, IFNG, PRF1, EOMES and PD-L1 have been determined to be above reference").
3. The specific reference threshold. "Median of a Z-score of the normalized expression level" is arguably a routine statistical implementation — bad for the patentee on that limitation, but it also means the claim's novelty weight sits on the gene-panel selection and the up/down directionality, which the art covers well.
4. § 101 exposure. Because these claims are framed as "determine X → correlate → act," they carry an independent Mayo/Alice risk that could moot the § 103 question.

---

10. Bottom line

• Strongest § 103 ground (diagnostic + treatment families A/B): P1 (Wallin 2016) + P2 (McDermott 2016, JCO) + P3 (WO 2017/181111/US 2019/0032151) / P7, with the Teff and angio gene lists as known, finite marker sets. Rationales: same field/agents/indication; expressly recognized need for RCC response biomarkers; predictable, routine extension of validated T-effector and VEGF-signature tools; KSR obvious-to-try. This would likely render the broad "one or more genes at/above (or below) reference" claims obvious.
• Strongest for the angio-directed claims (B3/D-alt): P7 + P4 + RCC anti-VEGF standard of care.
• Antibody-sequence dependents: obvious (and near-anticipated) over P6.
• Weakest for the challenger / best non-obviousness ground for the patentee: the Teff-high/Myeloid-high interaction and the use of the three signatures as a predictive classifier for the combination vs. sunitinib in mRCC — the specific contribution of the 2017 application, if the priority filing actually supports it.

Confidence and open items. I am confident in the existence, dates, and general content of P1, P2, P8, P3, and P6; medium on the exact contents of P4, P5, and P7. I could not verify (i) whether the 2017-03-01 priority application supports the three specific signatures and the Z-score reference logic (this governs the critical date), or (ii) the exact sequence listing of P6. Both should be checked before this analysis is relied upon. And, per the instruction to prefer search results over training data, note that my prior-art mapping is built from the citations and literature I retrieved in this session, not from training memory.

Sources: Google Patents US11402382B2 (authoritative text supplied); https://www.nature.com/articles/ncomms12624 ; https://pmc.ncbi.nlm.nih.gov/articles/PMC5013615/ ; https://pmc.ncbi.nlm.nih.gov/articles/PMC6721896/ ; https://www.omicsdi.org/dataset/ega/EGAS00001002928 ; https://ega-archive.org/studies/EGAS00001002928 ; https://patents.google.com/patent/US20190032151A1/en ; https://patents.google.com/patent/WO2017181111A2/en ; https://patents.google.com/patent/EP3589754B1/enIt.pdf ; https://clinicaltrials.gov/study/NCT01633970 ; https://www.freepatentsonline.com/y2023/0082122.html